Home

TSP 5.0 Reference Manual - National Bureau of Economic Research

1. 1 HIST isis 183 185 Hypothesis testing 23 IDENT 188 IF 190 IN191 INPUIT tib 192 INS T itte 194 Interactive commands 27 INTERVAL eene 200 K 0 27 204 KEEP Piet at Pee 210 KERNEL 3 irte 212 L tede e 214 LENGTH tetti 218 EIME dne 219 Linear Estimation 20 DIST uicti 225 229 LOAD ub 233 LOCAL 2 5 V do 234 Login file ee 14 235 aie 242 MATRIX 2 20 020 251 Matrix Operations Commands 19 MFORM 22 0 000 254 Missing values 13 ML iiie ite e 259 MMAKE eese 265 MODEL 268 Model Simulation Commands 24 MOD orra 270 N NAME it 273 Names in 4 22 2 22 0 00000000004 274 278 Nonlinear Estimation 21 22 0000000 286 NOPRIN 287 NOREPE 288 289 290 Numbers composing 5 Obsolete Commands 28 OESQ tatis 291 OPTIONS 16 298
2. 0 A1 X1 A2 X2 A3 X1 X1 A4 X1 X2 A5 X2 X2 is calculated if there are sufficient degrees of freedom TR y 5 for this example BPHET is the same as WHITEHT except the user specifies a presumably more general list of variables in the E E regression with the BPLIST option Note that the ARCH command with the GT option can also be used to estimate such general heteroskedastic regression models LMHET is the same as WHITEHT and BPHET where the squared residuals are regressed on a constant term and the squared fitted values RESET is Ramsey s RESET test where the residuals are regressed on the original right hand side variables and powers of the fitted values The default order 2 is basically a check for missing quadratic terms and interactions for the right hand side variables It may also be significant if a quadratic functional form happens to fit outliers in the data JB is a powerful joint Lagrange Multiplier test of the residuals skewness and kurtosis It is asymptotically distributed as a chi squared with two degrees of freedom under the null of normality Small sample critical values are obs 5 10 20 3 26 2 13 30 3 71 2 49 40 3 99 2 70 50 4 26 2 90 75 4 27 3 09 100 4 29 3 14 125 4 34 3 31 150 4 39 3 43 200 4 43 3 48 250 4 51 3 54 300 4 60 3 68 400 4 74 3 76 500 4 82 3 91 800 5 46 4 32 inf 5 99 4 61 382 REGOPT SWILK is a normality test based on normal order statistics which
3. 302 306 QUIT iiti cades 307 OUTPUT 309 P deua 311 PANEL 41 194 312 345 430 PARAM etes 322 PDE vin ei eaa 324 PLOT i innu 328 331 PLOTS 334 YLDFAC 337 PRIN itam 341 PRINT eine 344 PROBIT ei 345 PHOOGC eie 351 Q 22 QUIT ees 353 R 354 READ iie 362 RECOVER 372 REGOPT eine 373 RED 2 uni 41 312 345 386 REPE unies 387 388 RETRY niania neia 389 390 5 391 393 SAVE ate 397 SELECT 398 SET iai 399 SHOW ee 401 SIME ii 403 s ee et 408 SMPLIF 410 SOLVE 412 SORT cedes 417 STOP ie 419 420 Index SUPRESix nS 421 SUR a taii 422 424 SYSTEM unu 426 T 0 0 09 428 Text 5 6 429 Time Series Identification 25 DITE citt 433 TOBETz tire s 434 TREND 438 TSP 2 458 TOTAIS 440 U UNIT scient periti deceit 441 UNMAKE i t
4. Suppose that we want to extract a few parameters and their associated VCOV matrix from a system with a large number of parameters in arbitrary order SUPRES COEF LSQ SILENT EQ1 EQ50 estimation with a large number of equations SUPRES DOT B1 B5 FRML EQ construct FRML 1 B1 1 etc ENDDOT ANALYZ EQB1 EQB5 print results for 5 of the parameters only RENAME VCOVA VCVB1 5 for use later RENAME COEFA CB1_B5 Here is an example using random draws to compute asymmetric confidence intervals FRML 1 Y A B X LSQ 1 FRML EQS SUM A B FRML EQR RATIO A B ANAL YZ NDRAW 500 EQS EQR In this example the scalars SUM and RATIO will be stored and eight statistics on the 500 computations of the two functions will be printed and stored in the 2 x 8 matrix MSD An example of series output and random draws PROBIT DC x 39 Commands FRML P CNORM A B X ANALYZ NDRAW 200 NAMES A B In this example the series P P_SE P_T P_LB P_UB P_MEAN P_MIN P_MAX P_NG will be stored and printed References Bishop Y M M S E Fienberg and P W Holland Discrete Multivariate Analysis Theory and Practice MIT Press Cambridge MA 1975 pp 486 502 Gallant A Ronald and Dale Jorgenson Statistical Inference for a System of Simultaneous Non linear Implicit Equations in the Context of Instrumental Variable Estimation Journal of Econometrics 11 1979
5. Usage There are four basic estimators available in LSQ single or multi equation least squares and single or multi equation instrumental variables They are all iterative methods which minimize a distance function of the general form f y X b S 8 H H H H f y X b where f y X b is the stacked vector of residuals from the nonlinear model S is the current estimate of the residual covariance matrix being used as a weighting matrix and H is a matrix of instruments The form of f y X b is specified by the user as a FRML which be either unnormalized in the form f y x b with no sign or normalized the usual form of y z f x b The latter form will cause equation by equation statistics for the estimated model to be printed To obtain a particular estimator various assumptions are made about the exact form of this distance function These assumptions are described below Nonlinear single equation least squares In this case there are no instruments H is identity and S is assumed to be unity This makes the objective function the sum of squared residuals of the model minimizing this function is the same as obtaining maximum likelihood estimates of the parameters of the model under the assumption of normality of the disturbances 242 LSQ The form of the LSQ statement for estimating this model is LSQ followed by options in parentheses and then the name of the equation Any of the standard NONLINEAR options
6. Examples SET VALUE X 1 1 SET SE S MATRIX 2 3 A B 2 LOG LABW FREQ A SMPL 1983 1990 GENR PFOR 100 DO I 1984 TO 1990 SET I1 l 1 SET PFOR I PFOR I1 EXP 1 0 DELTA ENDD The last example above shows how SET can be used to compute a series in which each observation is a dynamic function of a previous observation This can be done more efficiently however by replacing the last DO loop with a dynamic GENR SMPL 84 90 PFOR PFOR 1 EXP 1 0 DELTA 400 SHOW SHOW Output Options Examples SHOW displays information about specific symbols or classes of symbols It can also be used to display the internal limits on data in TSP SHOW DATE DOC lt list of symbols gt SMPL FREQ ALL EQUATION LIST MATRIX MODEL PROC SCALAR SERIES Usage SHOW may be used at any time during the interactive session to obtain information about symbols and how they have been stored Symbol names and class names may be freely mixed as arguments to the SHOW command The classes are EQUATION LIST MATRIX MODEL PROC SCALAR and SERIES Providing a class name will list all symbols belonging to that class SMPL and FREQ will display the current setting of each and ALL will display all symbols most recent entries first Unique abbreviations of class names are allowed SHOW LIST provides a list of all lt listnames gt provide information about all members of the particular lis
7. Options 447 Commands NHORIZ number of time periods for the impulse response function default 10 Specify NHORIZ 0 or SILENT to suppress the impulse response output NLAGS number of lags of the dependent variables to include on the right hand side of the equations default zero SBIC or AIC can be used to choose the number of lags minimize SBIC see the REGOPT command SHOCK ALL or CHOL or STDDEV or UNIT or matrix name specifies the type of shock for the impulse response function CHOL the default is a Choleski factorization matrix square root using the current ordering of the dependent variables A shock to the first factor affects the first variable initially while a shock to the second factor affects the first two variables etc ALL computes Choleski factorizations for all n orderings of the variables since different orderings can produce markedly different results this option is useful for making sure the results are robust to ordering You can also supply your own square matrix factorization STDDEV and UNIT specify residual standard deviation or unit shocks to single variables variance decompositions are not computed for these shocks SILENT NOSILENT suppresses all printed output This is useful for running regressions for which you only want selected output which can be obtained from the 9 variables which will be stored TERSE NOTERSE causes minimal output results only to be printed Example
8. SAVE creates a binary file which contains TSP variable names and types alternating with the actual data for each variable The RESTORE command can read this file and recreate the TSP variables as they existed when the SAVE command was executed Variables whose names begin with are not saved it is assumed they are intermediate results from TSP procedures like RES which is usually created by an estimation procedure Examples SAVE creates TSPSAV SAV If TSPSAV SAV already exists it is destroyed SAVE FOO creates FOO SAV 397 Commands SELECT Options Examples SELECT selects a subsample of observations from the last SMPL statement SELECT SILENT PRINT lt logical expression gt Usage SELECT is exactly the same as SMPLIF except it operates on the last SMPL statement instead of the last SMPLIF or SELECT statement This means that consecutive SELECT statements are independent they do not create a nested sequence of subsamples This is more convenient than saving and restoring the previous SMPL manually Output Some pairs of sample observations are printed unless the SILENT option or SUPRES SMPL has been used Options PRINT NOPRINT prints the full set of sample pairs resulting from SELECT Normally only one line of output is printed SILENT NOSILENT prints output at all Same as SUPRES SMPL before SELECT Examples SMPL 1 10 TREND T SELECT T gt 5 MSD SELECT T gt 2 amp T lt 9 M
9. 105 106 DBDEL 107 108 DBPRINT 109 DEBUG iine 110 DELETE ies tein nied 111 DELETE interactive 112 DIFFER ikisen iias 113 DIR Liang tds 115 Display Commands 15 DIVIND essen 116 DQ i sss das 119 DOO biete 121 DOT eee 122 DROP aues 125 DUMMY idus 127 E EDIT t 129 ELSE i 132 133 134 ENDDOT eee 135 0000 136 ENTER 137 138 Estimation Commands 25 3 EXEC rim evi epee A ee 141 EXE itte 142 F Al xo 41 194 242 345 430 143 niit 144 FIND te tin outre n 150 EORGST RU ES 151 FORM a 155 FORMAL 160 Formula Manipulation Commands ME 21 FREQ totu 163 ERME sees 165 Functions in TSP 10 G GENRA cuneis 167 GMM Eie 170 GOTO ih omae 175 GRAPH 176 177 Graphics for DOS Win version GRAPH sas 177 antt 331 Graphics for Givewin TSP GRAPH onsec 177 HIST dtes 185 PEQT 4 iiia de 331 Graphics for MAC version GRAPH uie 177 PLQT iii iiam 331 456 H PAE PcG totae trot 181 Help
10. COND NCHOICE lt number gt NREC lt series name gt SUFFIX lt list of names nonlinear options lt dependent variable gt lt conditional variables gt lt multinomial alternative variables gt Usage There are three types of logit model those where the regressors are the same across all choices for each observations i e they are characteristics of the chooser those where the regressors are characteristics of the specific choice and mixed models which have regressors of both kinds In the first case multinomial logit a separate coefficient for each regressor is estimated for all but one of the choices In the second case conditional logit the regressors change across the choices and a single coefficient is estimated for each set of regressors 1 Binary or multinomial logit like OLSQ or PROBIT LOGIT lt dependent variable gt lt multinomial variables chooser characteristics gt LOGIT Y C X1 X2 XK Y can be 0 1 or 1 2 or any integral values If Y takes on more than 2 values the model is multinomial logit The names of the coefficients are determined by appending the values of Y for each choice to the names of the explanatory variables The coefficients are normalized by setting the coefficients for the lowest choice to 0 If Y is 0 1 the coefficients C1 X11 X21 XK1 would be estimated with CO X10 X20 XKO normalized to zero If Y is 1 2 3 the coefficients C2 X12 X22 XK2 and C3 X13
11. OLSQ DW C U DP FRML EQ DW A B U G DP RHO PARAM A B G RHO LSQ EQ ENDDOT If you wanted to work with data for the same four countries on prices and quantities of commodities for example food housing energy etc you could use the double dot construction DOT FOOD HOUS OIL DOT 1 4 GENR S P Q ENDDOT PRINT S 1 S 4 ENDDOT This example generates 12 series with the names SFOOD1 SFOOD2 SFOOD3 SFOOD4 SHOUS1 SHOUS4 and SOIL1 SOIL4 in that order Each series is equal to the product of the corresponding price and quantity series In the outer DOT loop a table of the series for all the countries of each commodity is printed note the use of the imbedded dot 123 Commands Here is another example with numbered sectors DOT VALUE J 0 9 SELECT COUNTRYzJ OLSQ FISH FITz FIT SET B QCOEF 2 ENDDOT SELECT 1 PRINT COUNTRY FIT PRINT BO B9 This example regresses the same dependent variable FISH on C and X in separate samples for each of 10 countries same as PANEL BYID ID COUNTRY FISH X The fitted values are saved and printed together Here is another examples showing the use of numeric character strings and double dots DOT 017710711 DOT FOOD HOUSE OIL GENR S Q ENDDOT ENDDOT In this case the variables S01FOOD S10FOOD SO1HOUSE etc are created using a single price index PFOOD PHOUSE POIL for each set of series QO1FOOD Q
12. PRINT CONS GNP METHOD MLGRID RSTEP 0 05 CONS C GNP The next three estimations are exactly equivalent and demonstrate the FAIR option with instrumental variables SMPL 11 50 INST C G TIME LM CONS GNP NOFAIR INST C G TIME LM GNP 1 CONS 1 CONS GNP FORM NAR 1 PARAM VARPREF B EQAR1 CONS C GNP Drop first observation to compare with AR1 OBJFN GLS results SMPL 12 50 LSQ INST C G TIME LM GNP 1 CONS 1 EQAR1 Lagged dependent variable default OBJUFN GLS since EXACTML has small sample bias AH1 CONS C GNP CONS 1 Time series cross section with 10 years of data and 3 cross section units and fixed effects SMPL 1 10 FREQ PANEL T 10 FEI SALES C ADV GNP 1 REI SALES C ADV POP GNP References 46 AR1 Baltagi B H and Q Li A Transformation That Will Circumvent the Problem of Autocorrelation in an Error Component Model Journal of Econometrics 48 1991 pp 385 393 Beach Charles M and MacKinnon James G A Maximum Likelihood Procedure for Regression with Autocorrelated Errors Econometrica 46 1978 pp 51 58 Buse A Efficient Estimation of a Structural Equation with First Order Autocorrelation Journal of Quantitative Economics 5 January 1989 pp 59 72 Cochrane D and Orcutt G H Application of Least Squares Regression to Relationships Containing Autocorrelated Error Terms JASA 44 1949 p
13. PRINT SILENT STEP lt squeezing method SQZTOL lt squeezing tolerance SYMMETRIC TERSE TOL lt parameter convergence tolerance TOL G lt gradient CRIT convergence tolerance TOLS lt squeezed parameter convergence tolerance VERBOSE Usage Include these options among any other special options which you supply within parentheses after the name of the command which invokes the estimation procedure FIML ENDOG nonlinear options list of eq names Method The method used for nonlinear estimation is generally a standard gradient method explained in more detail in Chapter 10 of the User s Guide Briefly at each iteration a new parameter vector is computed by moving in the direction specified by the gradient of the likelihood uphill weighting this gradient by an approximation to the matrix of second derivatives at that point in order to adjust for the curvature Convergence is declared when the changes in the parameters are all small where small is defined by the TOL option Options DROPMISS NODROPMISS specifies whether observations with missing values in any variables are to b dropped This can be useful if the equation being estimated varies according to the presence of good data but use with caution 278 Nonlinear Options EPSMIN minimum parameter change for numeric derivatives default is 0001 This is also used to control the numeric stepsize when computing and HCOV U If you
14. 11 Reproduce exact P value for Durbin Watson statistic this can be done automatically using REGOPT SMPL 1 10 data from Judge et al 1988 example DW 1 037 P value 0286 READY 1 42747 56432132534 547 5856 REGOPT DWPVAL EXACT OLSQ YC X1 MMAKE X RNMS MAT XPXI X X TREND OBS SELECT OBS gt 1 DC 0 DX1 X1 1 1 MMAKE DX DC DX1 MMAKE BVEC 2 1 MFORM BAND NROW NOB DDP BVEC MAT DMDP DDP DX XPXI DX Eigenvalues of DMD D D DX X X DX same as nonzero eigenvalues of MA because A D D MAT ED EIGVAL DMDP CDF WTDCHI EIG ED DW References Cheung Yin Wong and Lai Kon S Lag Order and Critical Values of the Augmented Dickey Fuller Test Journal of Business and Economic Statistics July 1995 pp 277 280 ACM Collected Algorithms New York 1980 Brown Barry W DCDFLIB http odin mdacc tmc edu downloaded v1 1 4 1998 79 Commands DiDinato A R and Morris Alfred H Jr Computation of the Incomplete Gamma Function Ratios and Their Inverse ACM Transactions on Mathematical Software 12 1986 pp 377 393 DiDinato A R and Morris Alfred H Jr Algorithm 708 Significant Digit Computation of the Incomplete Beta Function Ratios ACM Transactions on Mathematical Software 18 1993 pp 360 373 Engle R F and Granger C W J Co integration and Error Correction Representation Estimation and Testing Econometrica 55 1987 pp 25
15. 2 EIGVEC CORR 2 Note that PRIN change signs to make top row positive MAT FACTLD EVEC DIAG SQRT EIGVAL MFORM TYPE TRI NROW 3 ONE 1 2 Fraction of variance explained sum 3 MAT FRACVAR ONE EIGVAL 3 PRINT EIGVAL FRACVAR FACTLD MMAKE XM X Y Z 2 Assumes X Y Z are already standardized MAT PCOM XM FACTLD DIAG EIGVAL Options FRAC the fraction of the variance of the input variables which you wish to explain with the principal components the prefix to be given to the names of the principal components the components will be called prefix1 prefix2 and so forth You may use legal TSP name as the name for the principal components but the names generated by adding the numbers must also be legal TSP names i e of the appropriate length 342 PRIN the maximum number of components to be determined The actual number will be the minimum of the number requested the number of variables and the number needed to explain FRAC of the variance PRINT NOPRINT tells whether the results of PRIN are to be printed or just stored in data storage The default values of the PRIN options are NAME P NCOM number of variables FRAC 1 This set of options is non limiting that is the maximum number of components possible will always be constructed Example PRIN NAME PC NCOM 3 FRAC 95 I TIME CONS GOVEXP EXPORTS a specifies that three principal components a
16. CNORM Xb NORM Xb CNORM Xb NORM Xb CNORM Xb DLCNORM Xb E lt X D 1 DLCNORM Xb 0 where NORM is the normal density CNORM is the cumulative normal and DLCNORM is the derivative of the log cumulative normal with respect to its argument Before estimation PROBIT checks for univariate complete and quasi complete separation of the data and flags this condition The model is not identified in this case because one or more of the independent variables perfectly predict the dependent variable for some of the observations and therefore their coefficients would slowly iterate to plus or minus infinity if estimation was allowed to proceed The scaled R squared is a measure of goodness of fit relative to a model with just a constant term it replaced the Kullback Leibler R squared beginning with TSP 4 5 since it has somewhat better properties for discrete dependent variable problems See the Estrella 1998 article The Probit random effects model estimated is the following Vie I X Ba E gt 0 N 0 01 and N 0 0 o 1 NE xem 348 PROBIT This normalization means that the slope estimates are normalized the same way as the results from the usual Probit command The parameter RHO is estimated and corresponds to the share of the variance that is within individual The likelihood function involves computing a multivariate integral and this is done with Hermite quadrature u
17. Commands PLOTS Options Examples References PLOTS turns on the option which produces plots of actual and fitted values and residuals following estimation The default is not to produce plots It can also be used to turn on plots of CUSUM and CUSUMQ which are used to look for structural change in a regression PLOTS PREVIEW ALL CUSUM CUSUMSQ Usage Include a PLOTS statement in your program before the first regression OLSQ AR1 INST or LSQ for which you wish to see residual plots PLOTS remains in force until a NOPLOT statement is encountered Note that even though residual plots are not printed residuals and fitted values will still be stored in data storage To suppress this feature also use the OPTIONS NORESID statement The regression diagnostics are based on the maximum values of the CUSUMSs relative to their bounds or mean They provide a compact alternative to the plot for example if a P value is 05 the CUSUM crosses a bound CSMAX max 9479 CUSUM t CSUB5 t CSQMAX max GCUSUMSQY t CSQMEAN t The P value CSMAX is a function of CSMAX given in Brown et al 1975 CSQMAX is a function of CSQMAX and the degrees of freedom described in Durbin 1969 This P value and the critical value for the CUSUMSQ plot are computed from the algorithms given in Edgerton and Wells 1994 The critical values are based on the asymptotic approximation when the number of observations is greater than 6
18. Commands The FEI and REI options compute estimates for models with fixed and random effects for individuals respectively FREQ PANEL must be in effect For fixed effects a very efficient algorithm is used so large unbalanced panels can easily be handled The FEPRINT option prints a table of the effects their standard errors and t statistics Individuals that have dependent variable values that are all zero or all one are allowed although their data are not informative for the slopes The fixed effects for such individuals will be either a very large negative number in the case of zero or a very large positive number in the case of one These values yield the correct probability for these observations zero or one Note that this estimator has a finite T bias so the number of time periods per individual should not be too small The random effects model is estimated by maximum likelihood see the method section below for details Output The output of PROBIT begins with an equation title and the name of the dependent variable Starting values and diagnostic output from the iterations will be printed Final convergence status is printed This is followed by the mean of the dependent variable number of positive observations sum of squared residuals R squared and a table of right hand side variable names estimated coefficients standard errors and associated t statistics PROBIT also stores some of these results in data storage f
19. GENR DATE 1 SMPL 1120 GENR DATEz2 will be a series DATE which is one for the first 10 observations and 2 for the second ten The result of this sequence of statements NOREPL SMPL 1 20 GENR DATE 1 SMPL 11 20 GENR DATEz2 is a series DATE which is missing for observations 1 to 10 and equal to two for observations 11 to 20 387 Commands RESTORE Examples RESTORE reads TSP variables from a file which has been created with the SAVE command RESTORE or RESTORE filename string Usage RESTORE reads a file named TSPSAV SAV by default If a filename string is supplied the filetype SAV is appended if it is not present The RESTORE command is useful for restarting an interactive session which was stopped after issuing a SAVE command If a SMPL is already present in the current session the SMPL in the save file is not restored Any variables present in the current session with names equal to variables in the save file will be replaced by the save variables Output The current SMPL is printed if it has been restored Examples RESTORE reads TSPSAV SAV RESTORE FOO reads FOO SAV 388 RETRY Interactive RETRY Interactive RETRY combines the functions provided by the EDIT and EXEC commands When editing is complete execution is automatic RETRY line numbers Usage RETRY is identical to EDIT with two exceptions execution of the modified command is automatic after RETRY and e
20. GENR GNPN GNP DELTA R Argument Contents GENR GNPN MEM CUN a 129 Commands The editing commands and their arguments are as follows EXIT Editing completed takes no arguments A simple carriage return will also be interpreted as EXIT DELETE arg n Delete the nth occurrence of arg in the command REPLACE arg1 arg2n Replace the nth occurence of arg1 with arg2 in the command INSERT arg arg2n Insert arg1 after the nth occurence of arg2 in the command If there is only one argument provided it is inserted at the end of the command NOTE n is always assumed to be 1 if absent Restrictions 1 Subscripts you will not be able to edit successfully double subscripts or subscripts with dates 2 Parentheses the editor will not successfully find specific parentheses in commands which contain lags leads or subscripts prior to the parenthesis you are specifying Examples One type of modification you may wish to make is to change the list of options on a command without having to retype the whole thing particularly if it is lengthy As a simple example here is how you might change a static forecast into a dynamic one as well as request printing and plotting 5 FORCST STATIC DEPVAR z I IFIT 130 EDIT Interactive lt no output since the print option is off gt 6 EDIT 5 gt gt INS PRINT gt gt REP STATIC DYNAMIC gt gt EX 5 FORCST PRINT DYNAMIC DEPVAR I
21. Output Options Example References SOLVE solves linear and nonlinear simultaneous equation models using the Gauss Seidel method or the Fletcher Powell method for minimization The model is solved period by period if a dynamic simulation the default is specified the solved values for lagged endogenous variables are fed forward to later periods SOLVE is suitable for large mostly linear loosely structured models Since the algorithms are designed to operate on the model in blocks or groups of equations you can obtain cost savings by using SOLVE instead of SIML when your model is large but not very interrelated for example it includes several different sectors The Gauss Seidel algorithm which is fundamentally a recursive loop through the equations is the least powerful of the algorithms available in TSP for model simulation The Fletcher Powell algorithm which solves simultaneous blocks by minimizing the sum of squared residuals from each equation is somewhat more powerful but not as good as the Newton s method implementation in SIML since it does not use an analytic Jacobian SOLVE CONV2 lt secondary convergence criterion gt DEBUG DYNAM or STATIC KILL MAXPRT lt iterations to be printed METHOD GAUSS or FLPOW or JACOBI PRNDAT PRNRES PRNSIM TAG lt tagname gt or NONE nonlinear options lt name of collected model gt Usage To simulate a model with SOLVE first specify all the equations of the model in normal
22. X23 XK3 would be estimated with C1 X11 X21 XK1 normalized to zero LOGIT NCHOICE 2 Y C X1 X2 XK 235 Commands Including the NCHOICE option causes TSP to check the range of Y to make sure there are only 2 choices The model estimated has K 1 coefficients and K 1 variables The multinomial logit procedure checks for univariate complete and quasi complete separation which prevents identification of the coefficients 2 Conditional logit LOGIT COND NCHOICEzn dependent variable conditional variables choice characteristics For example LOGIT COND NCHOICE 2 Y XZ looks for the variables X1 X2 Z1 Z2 corresponding to the 2 choices The coefficients X and Z would be estimated C is not allowed as a conditional variable since it does not vary across choices it is not identified For a choice specific set of dummies use C as a multinomial variable in mixed logit In this case the example shown becomes LOGIT COND NCHOICE 2 Y XZ C The CASE option allows you to use data organized with one choice per observation rather than with one case per observation For example LOGIT CASE V Y XZ where the variable V is a case ID which is equal for adjacent observations which belong to the same case In this case the variable names X and Z are used directly not X1 X2 etc There need not be an equal number of observations per case Only the first Y for each case is examined for a valid choice
23. if it was a long session this could be costly and you may not need all results duplicated It is highly recommended that you REVIEW the session after it is recovered then EXEC ranges of lines that will restore what you need to proceed During your interactive session TSP is maintaining the file INDX TMP in your current directory which contains all the commands you have entered so far in a special format indexed keyed access This file is referenced any time you REVIEW EDIT EXEC etc Upon normal exit from the program this file is used to create a sequential file BKUP TSP containing the commands and INDX TMP is deleted If the program terminates abnormally INDX TMP will still exist and BKUP TSP will not Every time you start up interactive TSP recovery is automatically offered if an INDX TMP file exists in the current directory If you use this command to recover some other file it MUST be a file that had been created by TSP originally as an INDX TMP file RECOVER follows the same conventions as INPUT and OUTPUT for accepting filenames Although you may specify a filename on the command line you will probably want to be prompted for it since it is likely that you will be recovering a file from a different directory or with a different extension RECOVER produces no printed output other than the information that the session has been recovered 372 REGOPT REGOPT Output Options Examples References REGOPT controls
24. list of series Usage Follow HIST with the names of one or more series for which you would like to see a frequency distribution The default options for output yield a histogram with ten equally spaced bins or cells running from the minimum value of the series to the maximum value The bars for each cell have a width of two lines on the printed page and are based on the left hand axis of the graph Output If the PRINT option is on a plot of the histogram for each series is produced The following is stored in data storage variable type length description HIST matrix nbins series matrix with observation counts for each series HISTVAL matrix nbins series matrix with bin lower bounds for each histogram Options BOT NOBOT causes the printed histogram to be based on the left side of the page The default is the middle of the page DISCRETE NODISCRE specifies whether the series are discrete or continuous If the series are discrete there will be one cell for each unique value limited by NBINS upper bound on the last cell The default is the maximum value of the series 183 Commands lower bound on the first cell The default is the minimum value of the series NBINS the number of bins or cells The default is 10 for NODISCRETE and 20 for DISCRETE PERCENT NOPERCEN causes the percent in each cell rather than the absolute number to be printed the graph looks the same but the labels on t
25. 1 X 20 LIST STATES 1 50 MSD ALL prints simple statistics for all data series LIST INSTVAR C TIME LMG LIST ENDOGVAR GNP CONS LIST ALLVARS INSTVAR ENDOGVAR 226 LIST The list ALLVARS consists of the series C TIME LM G GNP CONS I R and LP The following example shows the power of implicit lists when combined with DOT loops to eliminate repetitive typing PRINT PAT72 PAT76 RND72 RND76 PARAM A72 A76 DELT72 DELT76 BETA DOT 72 76 FRML PAT EXP A BETA RND DELT72 RND72 DELT73 RND73 DELT74 RND74 DELT75 RND75 DELT76 RND76 PARAM A 1 0 DELT ENDDOT LSQ NOPRINT STEP BARDB EQ72 EQ76 In the above example the series equations and parameters for a panel data model 5 years of data on each of several hundred units can all be referred to by their LIST names to save the repetitive typing of each year s variables Obviously LISTs defining more than one variable cannot be used within equations because they do not reduce to algebraic expressions but it is useful in some applications to use EQSUB in a DOT loop to reproduce similar equations Suppose you do not know the number of variables in a LIST until runtime you can use the options to construct a variable length list in this case BEGYR 72 ENDYR 76 LIST PREFIX EQ FIRST BEGYR LAST ENDYR EQS creates a list called EQS consisting of EQ72 EQ73 EQ74 EQ75 and EQ76 More simple examples LIST FIRST 5 L
26. ALL NOALL computes the median first and third quartiles and the interquartile range in addition to the normal statistics The median etc are computed using any weight that has been supplied 271 Commands BYVAR NOBYVAR treats missing values for each series separately so that the maximum possible number of observations for each series is used NOBMSD will be stored in this case Normally if any series has missing values for any observation that observation is dropped for all series CORR tells MSD to compute and print the correlation matrix of the variables COVA tells MSD to compute and print a covariance matrix MOMENT tells MSD to compute and print an uncentered moment matrix also This matrix is divided by the number of observations with positive weights to scale it conveniently PAIRWISE NOPAIRWISE treats missing values for each pair of series separately from other series It applies to CORR COVA and MOMENT matrices NOBCOVA will be stored PRINT NOPRINT specifies whether the results of the procedure are to be printed or just stored in data storage SILENT NOSILENT specifies that all printed output is to be suppressed TERSE NOTERSE specifies that only the means standard deviations minima and maxima are computed and printed The sum variance skewness and kurtosis are suppressed WEIGHT the name of series which will be used to weight the observations The data are multiplied by the square roots of the weight
27. Econometrics 4 May 1976 pp 115 145 Divisia F Economique rationnelle Gaston Doin Paris 1928 Divisia F L indice monetaire et la theorie de la monnaie Revue d Economie Politique 39 1925 pp 842 861 980 1008 1121 1151 118 DO DO Examples DO specifies the beginning of a loop or grouped set of statements The loop or group of statements must be terminated by an ENDDO statement DO or DO index name gt start value end value BY lt increment gt or DO index name gt start value end value increment Usage The first form of the DO statement without arguments is primarily used to specify the beginning of a block of statements which form a THEN or ELSE clause after an IF statement The other form of the DO statement specifies a conventional loop as in many programming languages TSP executes the statements between the DO and ENDDO statement repetitively as many times as specified by the information given on the DO statement The index or counter variable is set equal to the start value the first time through and is changed each time through by the increment until the end value has been reached or exceeded This test is done at the end of the loop so the program always goes through once DO loops can be nested with other DO loops or with DOT loops The start value end value and increment may be any real numbers positive or negative unlike some earlier
28. Finally the ESACF correlations their p values and a table of Indicators is printed If the PRINT option is on a table of AR coefficient estimates is printed The following matrices are stored variable type length description AC matrix NLAG diff autocorrelations PAC matrix NLAGP diff partial autocorrelations IAC matrix NLAGP NLAGP inverse autocorrelations if requested ESACF matrix NAR NMA ESACF correlations ESACF matrix NAR NMA p values for ESACF correlations PHI matrix NAR NAR 1 2 AR coefficient estimates from 34 NMA ESACF ESACFI matrix NAR NMA ESACF Indicators Options Note that for all the Box Jenkins procedures BJIDENT BJEST and BJFRCST TSP remembers the options from the previous Box Jenkins command so that you only need to specify the ones you want to change BARTLETT NOBART specifies that the Bartlett estimate using lower order autocorrelations is to be used for the variance of the ESACF option NOBART will simply use 1 T p q 69 Commands ESACF NOESACF computes the extended sample ACF of Tsay and Tiao 1984 This can be useful for identifying stationary and nonstationary ARMA models The upper left vertex of a triangle of zeroes in the Indicator matrix identifies the order of the ARMA model The zeroes correspond to nonsignificant autocorrelations See the examples IAC NOIAC specifies whether the inverse autocorrelations are to be computed and printed NAR maximum order of AR for ESACF
29. If the DYNAM option is off the output will be identical to that of a single SMPL over the entire period except for the indicated gaps With RHO not zero or the DYNAM option forecasting will start anew with each pair of SMPL numbers that is it will start with new initial conditions and thus the results will be different than a single forecast over the entire sample Output When the PRINT option is off no output is printed by FORCST and only the single forecasted series is stored in data storage When the PRINT option is on the procedure prints a title the vector of coefficients used in the forecast the serial correlation parameter and whether the forecast is static or dynamic A plot of the forecasted series is printed which has the observation s name down the left hand side and the values of the series printed on the right hand side The series is also stored Options COEF the name of a vector containing the coefficient estimates to be used in the forecast This could be the vector of coefficients stored under the name COEF after a previous estimation The order of the coefficients in this vector should match the order of the names of the right hand side variables in the forecast model DEPVAR the name of the dependent variable in the estimation model This option is necessary to obtain correct dynamic forecasts when the FORCST procedure is not executed immediately following the estimation procedure 152 FORCST DYNA
30. OLSQ WEIGHT POP YOUNG C RSALE URBAN CATHOLIC MARRIED This is also how the command will now look if you REVIEW it since it has been modified and replaced both in TSPs internal storage and in the backup file Another use for ADD might be in producing plots The following will produce two plots with the same option settings but two series are added to the second plot PLOT MIN 500 MAX 1500 LINES 1000 GNP GNPS ADD CONS C CONSS D 34 ANALYZ ANALYZ Output Options Examples References ANALYZ computes the values and estimated covariance matrix for a set of nonlinear functions of the parameters estimated by the most recent OLSQ LIML LSQ FIML PROBIT etc procedure It also computes the Wald test for the hypothesis that the set of functions are jointly zero If the functions are linear after an OLSQ command the F test of the restrictions and implied restricted original coefficients will be printed ANALYZ can also be used to compute values and standard errors for function of parameters and series in this case the result will be two series one containing the values corresponding to each observation and the other the standard errors The method used linearizes the nonlinear functions around the estimated parameter values and then uses the standard formulas for the variance and covariance of linear functions of random variables See the references for further discussion of this delta method TSP obtains analytic de
31. Output Options Examples Reference FORCST allows you to use the results of any linear equation estimation routine in TSP to compute predicted values of the dependent variable over observations which may be the same or different from those used for estimation You may also do a forecast using a different set of exogenous variables The model for the forecast is either that specified on the previous linear index estimation procedure OLSQ INST PROBIT TOBIT ORDPROB POISSON NEGBIN ARCH or AR1 or you may supply the model yourself using various options To compute predicted values after a nonlinear estimation procedure LSQ or FIML use the GENR SIML or SOLVE commands Use BJFRCST if your model was estimated by Box Jenkins techniques in BJEST FORCST COEFz vector name gt DEPVAR lt var name DYNAM or STATIC PRINT RHO lt scalar gt lt predvarname gt lt list of indep variables gt Usage The simplest form of FORCST follows the estimation command which specifies the model to be used for prediction no other estimation commands can intervene You should change the sample with a SMPL statement between the estimation and the forecast if you want to forecast for a different time period The statement in this case is FORCST lt to be given to predicted variable gt Include a PRINT option with the command to have the results printed and plotted The other way to use the FORCST command does not require
32. PARAMs If any constraint is violated set LOGL to MISS before exiting from the PROC OPTIONS DOUBLE is advised if you want to use double precision to form intermediate results such as residuals The main disadvantage of using the PROC instead of the FRML method is that analytic derivatives are not available However numeric derivatives the default HITER F and GRAD C2 will often be quite adequate A slight disadvantage is that you have to explicitly list the PARAMs to be estimated in the command line HCOV U numeric second derivatives is the default method of computing the standard errors for MLPROC For iteration HITER F is the default but HITER U can be chosen as an option Good Applications for the PROC method 1 Time series models like ARMA and GARCH where the equations are recursive depend on residuals or variance from the previous time period s Models which can be evaluated by the KALMAN command also fit into this category ML thus allows estimation of the hyperparameters 260 ML Multi equation models like FIML These involve Jacobians matrix inverses and determinants which would have to be written into the log likelihood equation by hand very difficult for more than about 4 equations unless the Jacobian is sparse Models which require several diverse commands to evaluate such as multivariate normal integrals via simulation or other functions that are not built in to TSP Another example in this
33. QRNMSA QCOEFA etc Note that REGOPT NOPRINT COEF is also needed to suppress printing of the table of coefficients VCOVz specifies the name of a variance covariance matrix of the input parameters whose names are given by NAMES The use of these two options enables one to do an ANALYZ on matrices other than the VCOV matrix from a standard estimation procedure The default is VCOV Examples Obtain long run coefficients for models with lagged dependent variables FRML LR1 ALPHALR ALPHA 1 LAMBDA FRML LR2 PHILR PHI 1 PSI ANALYZ LR1 LR2 See the EQSUB command for an example of using ANALYZ with EQSUB to evaluate and obtain standard errors for restricted parameters in a translog system The next example shows how to calculate an elasticity and its standard errors when the elasticity changes over the sample FRML EQ1 LQ1 A1 B1 LP1 B2 LP2 B12 LP1 LP2 B13 LP1 LP3 B23 LP2 LP3 FRML EL1 ELD1 B1 B12 LP2 B13 LP3 d LQ1 d LP1 SMPL 48 95 LSQ 1 38 ANALYZ Obtain print and plot elasticity for each year between 1948 and 1995 SMPL 48 95 ANALYZ EL1 PLOT ELD1 Compute the average elasticity and its average s e MSD ELD1 ELD1 SE Here is an example of using ANALYZ after OLSQ It computes a chi squared test of the hypothesis that the sum of the two coefficients is zero this test statistic equals the standard F statistic OLSQYCX1X2 FRML SUM X1 X2 ANALYZ SUM
34. Results are still stored in AC PAC etc 70 BJIDENT Example This example computes the auto sales example from Nelson s book BJIDENT IAC NDIFF 1 NSDIFF 1 NSPAN 12 NLAG 48 NLAGP 20 AUTOSALE The following example shows the ESACF output for Box Jenkins Series C BJIDENT ESACF NAR 5 NMA 8 CHEM The result of the above command is the following matrix MA O14 o 0n Oo oO o OND O o O W WO 4 c C0 0 Rc A triangle of zeroes with upper left vertex at 2 0 is seen this indicates an ARMA 2 0 model References Box George P and Gwilym M Jenkins Time Series Analysis Forecasting and Control Holden Day New York 1976 Ljung G M and Box George On a measure of lack of fit in time series models Biometrika 66 1978 pp 297 303 Nelson Charles Applied Time Series Analysis for Managerial Forecasting Holden Day New York 1973 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Co 1976 Chapters 13 and 14 Tsay and Tiao Consistent Estimates of Autoregressive Parameters and Extended Sample Autocorrelation Function for Stationary and Nonstationary ARMA Models JASA March 1984 pp 84 96 71 Commands CAPITL Options Examples CAPITL computes a capital stock series from a gross investment seri
35. SMPL vector length 2 402 SIML SIML Output Options Examples References SIML solves linear and nonlinear simultaneous equation models using Newton s method with an analytic Jacobian The model is solved period by period if a dynamic simulation the default is specified the solved values for lagged endogenous variables are fed forward to later periods SIML is a more powerful and more working space intensive alternative to the SOLVE procedure Use SIML if your model is highly nonlinear and difficult to solve with the conventional Gauss Seidel recursive algorithms SIML is also recommended for any small model less than about 25 equations because of its ease of use requiring less setup cost than SOLVE If the model is linear SIML will solve it in one iteration per time period SIML DEBUG DYNAM or STATIC ENDOGz list of endogenous variables METHOD NEWTON or GAUSSN PRNDAT PRNRES PRNSIM SILENT TAG lt tagname gt or NONE nonlinear options list of equation names Usage To simulate a model with the options set at default values specify SIML with the ENDOG option to give the list of variables to be solved for and then a list of equations which are to be solved These equations are specified earlier with FRML or IDENT statements There is no difference between the two types of equations in SIML for either one the model solution tries to make the error as small as possible Equations for SIML
36. SYSTEM This flaw can actually be a blessing in disguise after giving the SYSTEM command it is easy to get distracted by other tasks on the computer and forget that you are still running TSP Sometimes it takes something like the SET DEFAULT command why doesn t this thing WORK to remind you of unfinished business with TSP At any rate typing CONTINUE will quickly tell you if this is the source of your problems 427 Commands TERMINAL Interactive Redirect output that has previously been routed to an output file with the OUTPUT command to the terminal or screen TERMINAL Usage The TERMINAL command takes no arguments and performs a very simple operation subsequent output is directed to the terminal An important result of the TERMINAL command it that the previous file being used for output is closed The output may not be examined after using the SYSTEM command until it has been closed An output file may always be reopened and output appended to it with another OUTPUT command and the same filename 428 THEN THEN Example THEN is part of the compound statement IF THEN ELSE It comes before the statement or group of statements surrounded by DO ENDDO which to be executed if the result of the expression on the IF statement is true THEN Usage THEN has no arguments it is required as the next statement immediately following an IF statement The statement immediately f
37. This feature enables you to give operating system commands so you can EDIT the contents of your file or call your favorite editor to display it for you Typing CONTINUE will return you to your interactive TSP session without any loss of continuity 309 Commands Note In order to see the contents of your currently open output file you must close it before giving the SYSTEM command Otherwise your output will appear to be missing from the file The TERMINAL command or opening a new OUTPUT file will close the current file you may reopen the original file when you return from SYSTEM and output will continue to be appended to it Examples 13 Sending regression output to a file 14 OLSQ YC XZ 15 OUTPUT YXZ 16 EXEC 14 17 TERM In addition to any output files you have stored results in you may wish to document your session before you quit 74 OUTPUT AUG2385 752 75 Interactive TSP session on Aug 23 1985 RSS 75 75 comments about results files used 752 75 REVIEW photo of session 76 76 display of symbols created during session sorted 76 into classes 76 76 SHOW SERIES EQUATION MATRIX PROC 77 EXIT These commands and comments will be written to the disk file as well as the resulting output You may choose to document just significant points by REVIEWing specified ranges or EXECing important results The comment delimiter may be used freely to make output f
38. VAR NLAGSz5 1 2 4 C X1 X2X3 is equivalent to the following regressions OLSQ Y1 1 1 1 5 2 1 2 5 Y3 1 Y3 5 4 1 4 5 C X1 X3 OLSQ Y2 1 1 1 5 Y2 1 Y2 5 Y3 1 Y3 5 Y4 1 Y4 5 C X1 X3 OLSQ 1 1 1 5 2 1 2 5 Y3 1 Y3 5 Y4 1 Y4 5 C X1 X3 OLSQ Y4 1 1 1 5 Y2 1 Y2 5 Y3 1 Y3 5 Y4 1 Y4 5 C X1 X3 See the TSP User s Guide for more examples Reference Judge George G Helmut Lutkepohl et al Introduction to the Theory and Practice of Econometrics Second Edition Wiley 1988 Chapter 18 pp 751 781 448 WRITE WRITE Options Examples WRITE is used to write variables to the screen output file or an external file The output may be labelled free format or a format of your specification WRITE can create Lotus Excel and binary files PRINT is synonymous with WRITE WRITE FILEz filename string FORMAT BINARY or DATABANK or EXCEL or FREE or LABELS or LOTUS or 4 or RB8 or format text string FULL lt 1 0 unit number list of variables Usage WRITE is the inverse of the READ statement series or other variables which are read with a particular READ statement may be written by a WRITE statement of the same form When the list of variables contains only series WRITE writes one record for each observation in the current sample unless the format statement specifies more than one record this recor
39. WNAME OWN CONSEQ INVEQ INTRSTEQ PRICEQ You can get the same three stage least squares estimates without the intermediate two stage least squares printout by using this command 3SLS INST C LM G TIME CONSEQ INVEQ INTRSTEQ PRICEQ See the description of the LIST command for an example of using cross equation restrictions References Amemiya Takeshi The Nonlinear Two Stage Least Squares Estimator Journal of Econometrics July 1974 pp 105 110 Amemiya Takeshi The Maximum Likelihood and the Nonlinear Three Stage Least Squares Estimator in the General Nonlinear Simultaneous Equation Model Econometrica May 1977 pp 955 966 249 Commands Berndt E K B H Hall R E Hall and J A Hausman Estimation and Inference in Nonlinear Structural Models Annals of Economic and Social Measurement October 1974 pp 653 665 Chamberlain Gary Multivariate Regression Models for Panel Data Journal of Econometrics 18 1982 pp 5 46 Jorgenson Dale W and Jean Jacques Laffont Efficient Estimation of Nonlinear Simultaneous Equations with Additive Disturbances Annals of Economic and Social Measurement October 1974 pp 615 640 Judge et al The Theory and Practice of Econometrics 1980 John Wiley and Sons New York Chapter 7 Maddala G S Econometrics 1982 McGraw Hill Book Co New York pp 174 175 470 492 Theil Henri Principles of Econometrics John Wiley and Sons New York 1971
40. X MAT TOTAL TOTAL SUM X ENDDO SET CORRECT 5000494 15 PRINT TOTAL CORRECT References 360 RANDOM Efron Bradley Bootstrap Methods Another Look at the Jackknife Annals of Statistics 7 1979 pp 1 26 Efron Bradley The Bootstrap the Jackknife and Other Resampling Plans Philadelphia SIAM 1982 Efron Bradley and G Gong A Leisurely Look at the Bootstrap Jackknife and Cross validation American Statistican February 1983 37 1 pp 36 48 Fishman George S and Louis R Moore A Statistical Evaluation of Multiplicative Congruential Random Number Generators with Modulus 231 1 JASA 77 1982 pp 129 136 L Ecuyer Pierre Good Parameter Sets for Combined Multiple Recursive Random Number Generators Operations Research 47 1999 http www iro umontreal ca lecuyer papers html L Ecuyer Pierre Random Numbers for Simulation Communications of the ACM October 1990 pp 85 97 Schaffer Henry E Algorithm 369 Collected Algorithms from ACM Volume ACM New York 1980 361 Commands READ Options Examples READ is used to read series matrices or scalars into data storage Normally data will be read from an external file but small quantities of data can be included in a data section at the bottom of your program file when running TSP in batch mode READ can be also be used to read spreadsheets or stata data files If you plan to repeatedly use a very large d
41. and a diagonal matrix D such that X LDL and the diagonals of D are functions of the characteristic roots of X YLDFAC symmetric matrix diagonal matrix triangular matrix Usage Three required arguments to YLDFAC are name given to the matrix to be factored must be symmetric or an error message will be printed name given to the diagonal matrix and name given to the upper triangular matrix Most symmetric matrices can be factored in this way The elements of D are functions of the characteristic roots of the input matrix If all the diagonal elements of D are positive the input matrix is positive definite If all of them are nonnegative the input matrix is positive semi definite If there are some positive elements of D and some negative the input matrix is indefinite Zero elements on the diagonal of D imply that the input matrix is singular or near singular The diagonal elements of L are normalized to 1 Output YLDFAC produces no printed output Two matrices are stored in data storage Method A modified Choleski method of factorization is used where the diagonal element is extracted from the square root matrix as the factorization is performed The underlying method is described in the Faddeev reference Example YLDFAC A DIAG UPPER MAT ANEW z UPPER DIAG UPPER generates a matrix ANEW which is identical to the original matrix A Note that because the triangular matrix is stored as an upper triangle it must
42. f h h ifaz0 f h logh ifa 0 NAR NGT h a Dae L Aht DAI K l An immediate issue is the identification of the order of the ARCH or GARCH process Bollerslev suggests obtaining squared residuals from OLS and using standard Box Jenkins techniques BJIDENT in TSP on these squared residuals It is also possible to estimate several ARCH models and use likelihood ratio tests to determine the proper specification 49 Commands All models are estimated by maximum likelihood normally with analytic first and second derivatives Presample values of h and the disturbance epsilon are initialized by the methods specified in the E2INIT options ARCH will not work if there are gaps in the SMPL instead it is possible to use dummy variables as right hand side variables and in GT to exclude observations from the fit Default starting values for gamma are obtained from the OLS slope coefficients while 0 and presample h t if HINITZESTALL is used are started from the OLS ML estimate of the variance of the disturbance Other parameters are initialized to 0 These defaults apply to all models except when NAR 0 and NMASO in which case 0 0 b 1 1 are used to prevent false convergence to the OLS saddlepoint solution The defaults may be overridden if an START vector is provided by the user with a value for each parameter see the final example below Several constraints are imposed on the A
43. gt 1 fit less closely to the data more smooth Output KERNEL produces no printed output A series called DENSITY is stored when there is one argument and FIT is stored when there are two arguments Options BANDWIDTH specifies the absolute value of the bandwidth 212 KERNEL RELBAND specifies the bandwidth relative to h the default bandwidth IQR NOIQR specifies whether the interquartile range is to be used to compute the bandwidth Method Given the observed data series i 1 N the Kernel estimator f x of the density of x may be obtained using the following equation 12 X X f x YK Nh 53 h where K is the kernel function and h is a band width or smoothing parameter TSP uses the Normal or Gaussian kernel and a method based on a Fast Fourier Transform to evaluate this density The Kernel regression of y conditional on x is computed using the following equation X X K f y n zkh zA Neither estimator is very sensitive to the choice of kernel function but both are sensitive to the choice of band width h The options allow the user to control the bandwidth either in absolute size or in size relative to the variance or interquartile range if IQR is used of the series The default value of h is given by h h 0 9N where 0 is the standard deviation of the x series Silverman 1986 shows that this choice has good mean squared error propertie
44. imply larger file sizes and printing times FILE the name of a file to which the graphics image is to be written This file can be printed later for example if you are running under DOS and your printer device is LPT1 print the plot with the command 332 PLOT graphics version copy b file LPT1 HEIGHT letter height in inches The default is 25 Values in the range 0 1 are valid HIRES NOHIRES controls how graphs are printed in batch mode when PREVIEW is not being used Normally NOHIRES graphs are printed in character mode to the batch output file When the HIRES option is used the patched DEVICE and FILE will be used usually this will send a page to LPT1 for each graph LANDSCAP PORTRAIT specifies the orientation of the plot On the Mac specify this option in the dialog box MAC only WIDTH NOWIDTH specifies whether varying width sizes are to be used to distinguish the lines corresponding to different series on the graph Examples The following example plots the graph shown in the User s Guide FREQ Q SMPL 53 1 67 4 LOAD expend approp Original Almon data from Maddala p 370 2072 1660 2077 1926 2078 2181 2043 1897 2062 1695 2067 1705 1964 1731 1981 2151 9715 5412 5637 5465 5383 5550 5467 5465 PLOT PORT PREV DEV LJ3 FILEZ ALMON PLT TITLE ALMON DATA EXPEND APPROP Here is an example of setting the plot options without actually plotting anything PLOT DEV LJ3 HEIGHT 2 333
45. nonlinear with heteroskedasticity and autocorrelation robust standard errors same as FRML but for an identity no implied disturbance for FIML minimum distance estimation of single or multiple equation linear or nonlinear equations general cross equation restrictions additive error terms see SUR 3515 also Maximum Likelihood estimation log likelihood specified in FRML Maximum Likelihood estimation log likelihood specified ina PROC describes iteration methods and options used by ARCH BJEST LSQ FIML LOGIT ML MLPROC PROBIT TOBIT SIML SOLVE defines scalars as estimable parameters can supply starting values Seeming Unrelated Regressions LSQ without instruments Three Stage Least Squares LSQ with instruments 21 Command summary QDV Qualitative Dependent Variable Commands INTERVAL LOGIT NEGBIN NONLINEAR options ORDPROB POISSON PROBIT SAMPSEL TOBIT 22 estimates the Interval model ordered Probit with known limits estimates binary multinomial conditional and mixed Logit models estimates the Negative Binomial regression model for count data describes iteration methods used by ARCH BJEST estimates the Ordered Probit model estimates the Poisson model for count data estimates the Probit model 0 1 with normal error term estimates a two equation Sample Selection model estimates the Tobit model 0 positive with normal error term Hypothesis Testing Commands Hypothe
46. of Actual on Predicted U66 1 Theil s U Inequality coef Changes U66 U66P 1 Theil s U Ineq coef Percent changes U66P FBIAS 1 Fraction of MSE due to Bias 31 Commands FDVAR 1 Fraction of MSE due to different Variation FDCOV 1 Fraction of MSE due to difference Covariation FDB1 1 Alt Decomp Frac due to Diff of BETA from 1 FRES 1 Alt Decomp Frac due to Residual variance Note U is defined differently in the 1961 and 1966 references The 1966 definition is used in TSP Versions 4 0 and 4 1 under this definition U can be greater than one In TSP Version 4 2 and above both versions of U are printed This output is followed by a time series residual plot of the two series if the PLOTS option is on see the OPTIONS command If the RESID option is on the residual series will be stored under the name RES whether or not the PLOTS option is on Options SILENT NOSILENT suppresses all printed output TERSE NOTERSE prints a reduced output Example ACTFIT R RS References Theil Henri Economic Forecasts and Policy North Holland Publishing Company 1961 Theil Henri Applied Economic Forecasting North Holland Publishing Company 1966 32 ADD interactive ADD interactive Examples ADD adds list of variables to the previous statement and re executes it It is the opposite of DROP ADD lt list of variables gt ADD offers a convenient means of adding variables to a regression and
47. written to the file in error if any 2 User controlled access to more than 12 files in a given run is possible with CLOSE Since the number of simultaneously open files is limited on most operating systems often it is less than 12 TSP will close the most recently opened file and issue a warning message when access to a new file would result in too many simultaneously open files If this arbitrary choice of the file to close causes problems with your program use the CLOSE statement to reduce the number of simultaneously open files 3 If results from a repeated iterative estimation process are to be saved repeatedly in a file the CLOSE command could be used to cause repeated creation of the file instead of appending the new results each time to the file It would be slightly easier to use the OUT command for this type of problem OUT databank KEEP variables OUT 82 CLOSE 4 If important data has been written to a file and it is likely that later commands may cause TSP to abort or power failures may occur with the computer the file may be closed to guarantee that the data is completely written For whatever reason after the READ or WRITE statement issue the CLOSE command specifying the file either with a filename string or with a unit number in the options with the parentheses Example READ FILE FOO DAT X Y Z CLOSE FILE FOO DAT 83 Commands COINT Output Options Examples References COINT per
48. 97 98 99 100 101 102 103 104 105 106 107 108 Table of Contents ENDPROC 136 ENTER Interactive 137 EQSUB 138 EXEC Interactive 141 EXIT Interactive 142 FETCH 143 FIML 144 FIND Interactive 150 FORCST 151 FORM 155 FORMAT 160 FREQ 163 FRML 165 GENR 167 GMM 170 GOTO 175 GRAPH 176 GRAPH graphics version 177 HELP 181 HIST 183 HIST graphics version 185 IDENT 188 IF 190 IN Databank 191 INPUT 192 INST 194 INTERVAL 200 KALMAN 204 KEEP Databank 210 KERNEL 212 LAD 214 LENGTH 218 LIML 219 LIST 225 LMS 229 LOAD 233 LOCAL 234 LOGIT 235 LSQ 242 Table of Contents 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 MATRIX MFORM ML MMAKE MODEL MSD NAME NEGBIN Nonlinear Options NOPLOT NOPRINT NOREPL NORMAL NOSUPRES OLSQ OPTIONS ORDPROB ORTHON OUT Databank OUTPUT Interactive PAGE PANEL PARAM PDL PLOT PLOT graphics version PLOTS POISSON PRIN PRINT PROBIT PROC QUIT Interactive RANDOM READ RECOVE
49. An Expository Note and Comment The American Statistician June 1972 pp 32 35 Hall Robert E Polynomial Distributed Lags Econometrics Working Paper No 7 Department of Economics MIT July 1967 326 PDL Shiller Robert A Distributed Lag Estimator Derived from Smoothness Priors Econometrica 41 1973 pp 775 787 327 Commands PLOT Output Options Examples PLOT produces a plot of one or more series versus the observation number usually in units of time The series are plotted on the horizontal axis and time on the vertical axis The user has a good deal of freedom in formatting this plot with options For DOS Windows unix or MAC PC with the graphics version of TSP see the entry for PLOT graphics version PLOT BAND STANDARD or series name BMEAN BMID BOX HEADER ID INTEGER LINES list of values MAX lt y axis maximum MEAN MIN lt y axis minimum ORIGIN RESTORE VALUES series name plotting character series name gt plotting character gt Usage PLOT is followed by a series name the character to use in plotting the series possibly a second series name and a second character and so on Up to nine series may be plotted The characters may be anything except Parameters that control the appearance of the plot may be specified in an options list in parentheses following the word PLOT These parameters all have default values so you do not need to specify th
50. BETWEEN BYID FEPRINT HCOMEGA BLOCK or DIAGONAL 0 or 1 ID lt id series MEAN PRINT REG REI REIT ROBUST SILENT T lt number of time periods TERSE TIMEz time series TOTAL VARCOMP VBET lt between variance VSMALL VWITH lt within variance WITHIN Nonlinear options dependent variable list of independent variables Usage The basic PANEL statement is like the OLSQ statement first list the dependent variable and then the independent variables C is optional an intercept term is central to these models and will be added if it is not present You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The observations over which the models are computed are determined by the current sample PANEL treats missing values lags and leads correctly That is lags and leads are applied only within an individual Your data must be set up with all the time periods for each individual together Additionally you must specify when the observations for one individual end and data for the next individual begins The default method is to provide a series named ID which takes on different values for each individual If your data are balanced the same number of time periods for every individual the Tz option can be used If the data are not in this o
51. COEF HALTON specifies that a shuffled Halton sequence is used for the random draws when the NDRAW option is given This provides more uniform coverage of the range of values so it may yield more accurate integration for a given number of draws NAMES specifies an optional list of parameter names which are the labels for an associated covariance matrix supplied by the VCOV option The default is RNMS 37 Commands NDRAWzn computes asymmetric confidence intervals for nonlinear functions by drawing n simulated parameter vectors These functions can vary over time as well This is an alternative to the default delta method which uses derivatives and is exact for linear functions The percentiles 2 5 and 97 5 are computed to construct a two tailed confidence interval at the 95 significance level A matrix named MSD with columns SE LB2 5 UB97 5 MEAN MIN MAX NUM GOOD is stored NUM_GOOD is the number of nonmissing results computed Numeric errors such as division by zero result in missing values When ANALYZ is used with series and NDRAW the results are stored in series whose names are the name of the parameter computed followed by SE T LB UB MEAN MIN MAX See the Examples for an illustration PRINT NOPRINT tells whether or not the ANALYZ input is to be printed Under the default NOPRINT only the results are printed SILENT NOSILENT specifies that no output is to be produced The results are stored under the names
52. Commands ZERO 0 KALMAN Y1 2 C1 X1 ZERO ZERO ZERO ZERO C2 X2 Harvey s Example 2 p 116 117 signal plus noise or Cooley Prescott READ Y 4 4 4 0 3 5 4 6 SET Q 4 SET A0 4 SET P0 12 KALMAN VT Q BPRIOR A0 VBPRIOR P0 Y C Harvey s Exercise 1 p 119 stochastically convergent READ Y 4 4 4 0 3 5 4 6 4 Y 4 SET RHO 5 SET 4 SET 0 2 SET 3 KALMAN BT RHO VT Q BP A0 VBP P0 Y4 C Bootstrapping a variance for the transition equation SMPL 1 100 KALMAN NOETRANS Y C X1 X2 UNMAKE STATE B1 B3 SMPL 4 100 DOT 1 3 D B B 1 ENDDOT COVA D1 D3 VTOS COVA S2 KALMAN VTRANS VTOS Y C X1 X2 Hyperparameter estimation using ML PROC This example estimates the variances of the transition matrix Q using the ML PROC 2 5 Q PARAM 011 1 022 2 011 022 evaluates log likelihood for ML PROC PROC KFQ IF Q11 lt 0 OR Q22 lt 0 THEN Check constraints SET LOGL MISS Are variances gt 0 ELSE DO yes evaluate SET Q 1 1 Q11 SET Q 2 2 Q22 SUPRES COEF KALMAN SILENT VTRANS Q Y C X NOSUPRES COEF ENDDO ENDPROC 208 KALMAN References Cooley T F and Edward Prescott Varying Parameter Regression A Theory and Some Applications Annals of Economic and Social Measurement 2 1973 pp 463 474 Cooper J Philip Time Varying Regression Coefficients A Mixe
53. Commission Discussion Paper No 373 Chicago 1954 Rao P and Griliches Z Small Sample Properties of Several Two Stage Regression Methods in the Context of Auto Correlated Errors JASA 64 1969 pp 253 27 48 ARCH ARCH Output Options Examples References ARCH estimates regression models with AutoRegressive Conditional Heteroskedasticity originated by Robert Engle It will estimate any model from linear regression to GARCH M ARCH models allow the residuals to have a variable variance but still have zero conditional mean over the sample This contrasts with AR 1 models or general transfer function models where the residuals do not have zero conditional mean ARCH models are often used to model exchange rate fluctuations and stock market returns ARCH 2 GT lt list of weighting series gt HEXP lt value of lambda gt HINIT method MEAN NAR lt number of AR terms for regular ARCH gt NMA lt number of MA terms for GARCH gt RELAX ZERO nonlinear options lt dependent variable gt lt list of independent variables gt Usage The ARCH command is just like the OLSQ command except for the options which model the heteroskedasticity of the residuals The default model estimated if no options are included is GARCH 1 1 Method The generalized form of ARCH GARCH M estimated by TSP is given by the following equations see McCurdy and Morgan 8 h g amp 0
54. Composing Text Strings in TSP A text string must be enclosed by matching pair of quotes or Quotes are allowed in a string when they are of a different type from the enclosing quotes i e Can t or sometimes interpreted as Can t and sometimes Composing TSP Commands Composing TSP Commands Every statement begins with a command name Exceptions X Y implicit GENR 1 implicit SET 100 lt statement gt statement label for GOTO The command name may be abbreviated as long as it is uniquely identified Many statements can have options specified in parentheses after the command name Option names may be abbreviated like command names There are three kinds of options 1 Boolean options either on or off On is specified by the name of the option as in PRINT and off is specified by the option name with NO in front of it as in NOPRINT 2 Options of the form option name option value The value may be the name of a variable a numerical value or just a keyword depending on the context 3 Options which give lists of variables and are of the form option name list of variables Note that the parentheses are required unless the list contains only one name or the list is a listname A few commands can be followed by an algebraic formula GENR Most commands are followed by one or more series names separated by commas or spaces These series names may include lags An implicit list
55. DOC Options Examples DOC creates and maintains documentation for variables Documentation can then be displayed with the SHOW command DOC ADD REPLACE variable description of variable Usage List the name of a variable it doesn t have to exist yet and give a description for it in quotes There is no restriction on the length of the description and you can separate lines in the description with backslashes Note that the semicolon character cannot be part of the description string due to TSP s rules regarding strings see BASIC RULES Number 2 TSP automatically maintains a date field for the variable if it has a description and the DATE option in SHOW and DBLIST is used to display this SHOW and DBLIST will display as much of the description as they can fit on one line following the other information on the variable SHOW DATE DOC will print the description on separate lines following the regular information The DBLIST command also has the DATE and DOC options The documentation is stored in TLB databank files and DBCOPY can be used to move it to other computers Output No output is produced until the SHOW or DBLIST commands are used Options ADD NOADD specifies that the description should be added appended to any existing description REPLACE NOREPLAC causes the description to replace any existing description Examples DOC CONS72 Consumption in 1972 dollars Source ERP 1990 This is equiv
56. Default is 20 NDIFF the degree of differencing to be applied to the series The default is zero no differencing BJIDENT will calculate statistics for all the differences of the series up to and including the NDIFFth order NLAG the number of autocorrelations to be computed The default is 20 NLAGP the number of partial autocorrelations to be computed The default is 10 maximum order of MA for ESACF Default is 10 NSDIFF the degree of seasonal differencing to be applied to the series The default is zero no differencing As in the case of ordinary differencing BJIDENT will calculate statistics for all the differences of the series up to and including the NSDIFFth order NSPAN the span number of periods of the seasonal cycle i e for quarterly data NSPAN should be 4 The default is the current frequency that is 1 for annual 4 for quarterly 12 for monthly PLOT NOPLOT specifies whether all the differenced series are to be plotted PLOTAC NOPLOTAC specifies whether the autocorrelations and partial autocorrelations are to be plotted PLTRAW NOPLTRAW specifies whether the original raw series is to be plotted PREVIEW NOPREVIEW TSP Givewin only specifies that the raw and differenced series are to be displayed in a high resolution graphics window if the PLOT option is on PRINT NOPRINT specifies whether the AR coefficients for the ESACF option are to be printed SILENT NOSILENT Turns off all output
57. ENDDO ENDDO is used to close DO loops ENDDO Usage ENDDO takes no arguments Every DO statement must have an associated ENDDO statement somewhere following it in the program The ENDDO statement always applies to the last DO which was encountered so that DO loops may be nested to any level ENDD or END DO are synonyms for ENDDO 134 ENDDOT ENDDOT ENDDOT is used to close DOT loops ENDDOT Usage ENDDOT takes no arguments Every DOT statement must have an associated ENDDOT statement somewhere following in the program The ENDDOT statement always applies to the last DOT which was encountered so that DOT loops may be nested to any level END DOT is a synonym for ENDDOT 135 Commands ENDPROC ENDPROC is used to end user PROCs ENDPROC lt name of PROC gt Usage ENDPROC takes one optional argument the name of the PROC to which it belongs Every PROC statement must have an associated ENDPROC statement somewhere following in the program PROCs may be nested to any level but if there is an ENDPROC statement missing for one of them a fatal error will be trapped ENDP or END PROC are synonyms for ENDPROC 136 ENTER Interactive ENTER Interactive Example ENTER allows you to input data from the terminal in an interactive session It will prompt you for observation as the data is entered Use the READ command in batch mode ENTER lt list of series names gt Usage ENTER must have at lea
58. Economic Forecasts Chapter 6 McGraw Hill Book Company New York 1976 154 FORM FORM Options Examples FORM makes a TSP equation FRML from the results of a linear estimation procedure or from a list of names such as regression variables The FRML can then be used in an estimation or simulation model it can have either names or constants as its coefficients The NAR option makes it easy to estimate linear regression equations with AR p errors just use the resulting FRML in LSQ for direct nonlinear estimation FORM COEFPR lt coefficient prefix NAR lt number of AR terms PARAM PRINT RESIDUAL RHOPREFz rho prefix SUM VALUE VARPR lt coefficient prefix equation name list of gt or FORM VALUE lt list of equation names gt Usage The first argument to FORM is required it is the name you wish to give to the equation If there is only one argument FORM must follow the linear estimation command OLSQ INST LIML VAR or AR1 which created the results you wish to save as an equation In this case the equation created by FORM is the same equation that would be used by FORCST to generate a single equation forecast If you GENRed this equation you would obtain a static forecast just as if you had run the FORCST procedure but with a slight loss of precision The variable names are retrieved from RNMS and LHV while the coefficient values are taken from COEF and RHO If there a
59. IFIT In this case execution of the modified command would not take place until specifically requested Another common modification is to fix a typo If you have made a typing error in line 10 for instance and wish to correct it the sequence might look like this 10 INT DP DP1 LGNP TIME C INVR C G LM TIME lt error message because procedure INST is misspelled gt 11 EDIT 10 gt gt REPLACE INT INST 1 gt gt 10 INST DP DP1 LGNP TIME C INVR C G LM TIME 12 EXEC 10 10 INST DP DP1 LGNP TIME C INVR C G LM TIME lt output from the INST command gt There are a number of ways that the previous example could be simplified to reduce typing First of all commands may be abbreviated Note that INT was not a valid abbreviation for INST see Basic Rules Second n 1 is the default on the REPLACE command as is carriage return for the EXIT command so they may be omitted Also since line 10 is the previous command it may be omitted from the EDIT command Lastly statements 11 and 12 may be combined by substituting the RETRY command for EDIT Assuming the same error on line 10 the correction would now look like this 11 RET gt gt RINT INST gt gt 10 INST DP DP1 LGNP TIME C INVR C G LM TIME lt output from the INST command gt 131 Commands ELSE Example ELSE signals that the statement or DO group of statements immediately following are to be executed if the last IF clause had a false result
60. Manual If the equations are unnormalized only the standard error sum of squared residuals and Durbin Watson are printed Method 246 LSQ The method used by LSQ is a generalized Gauss Newton method The Gauss Newton method is Newton s method applied to a sum of squares problem where advantage is taken of the fact that the squared residuals are very small near the minimum of the objective function This enables the Hessian of the objective function to be well approximated by the outer product of the gradient of the equations of the model Generalized refers to the fact that the objective function contains a fixed weighting matrix also rather than being a simple sum of squares This implementation of the Gauss Newton method in TSP uses analytic first derivatives of the model which implies that the estimating equations must be differentiable in the parameters TSP defines the derivatives of discontinuous functions like SIGN to be zero so this will always be true The method is one of the simplest and fastest for well behaved equations where the starting values of the parameters are reasonably good When the equation is highly nonlinear or the parameters are far away from the answers this method often has numerical difficulties since it is fundamentally based on the local properties of the function These problems are usually indicated by numerical error messages from TSP the program tries to continue executing for a while but if t
61. Note that you may have to explicitly CLOSE files before you can manipulate them with SYSTEM commands In interactive mode SYSTEM usually takes no arguments and simply produces the message Enter system commands Type EXIT or CONTINUE to resume TSP session or some other system prompt when you have the system prompt you may create or modify files send MAIL to a friend or most anything you usually do You may keep entering commands as long as you like EXIT will resume your interactive session where you left off with no loss of continuity There are many uses for this feature one of the most apparent being the ability to examine output files created during your session without halting the program Please note that in order to use it in this way the file must be closed first with either the TERM or OUTPUT command Technical note for VAX VMS and some other operating systems You may find that a few VMS commands you use don t seem to work one example is the SET DEFAULT command to change the current directory The reason is that although it appears that you are issuing commands directly to VMS this is not actually so You are still running TSP TSP prints the and reads the command you type the command is then passed to VMS as a spawned subprocess There are things that the subprocess will not be allowed to do such as modify things about the environment within which the parent process TSP is executing 426
62. Quarterly Monthly turns residual plots off OLSQ INST AR1 LSQ prevents splicing of series generates missing values instead general option setting CRT HARDCOPY LIMPRN etc turns residual plots off OLSQ INST AR1 LSQ allows updating of series during GENR the default restricts the set of observations to those meeting a condition set the sample of observations to be processed same as SELECT but restricts starting from current sample Moving Data to from Files Commands Moving Data to from Files Commands CLOSE closes an external input or output file DBCOMP compresses a databank Databank DBCOPY copies databank for moving to another computer Databank DBDEL deletes variables from a databank Databank DBLIST lists all variable names in a databank Databank DBPRINT prints all series in a databank Databank FETCH reads a microTSP format databank FORMAT option used in READ and WRITE with numbers IN causes automatic searching of databanks listed Databank KEEP stores TSP variables on specified OUT files Databank LOAD reads variables from a file or from program same as READ NOPRINT suppresses echoing of commands in a LOAD section OUT causes automatic databank storage in files listed Databank PRINT same as WRITE READ reads variables from a file or from program RECOVER recovers lost program from INDX TMP file Interactive RESTORE reads variables from a SAVE file into the program
63. RANK MINTIES SCORE RM yields RM 1 224 418 STOP STOP Examples STOP causes the TSP program to stop If any variables are marked for storage on output databanks they are written to the databank before the program stops STOP Usage Often older TSP programs include a STOP statement at the end of the program section before the END statement that separates the TSP program section from the data section This STOP statement is no longer required since the END statement itself implies a STOP However if you want to stop somewhere else in your TSP program you can do this by using a STOP statement at any time This can be convenient if you encounter an error and wish to abort the program If you put a STOP statement at the beginning of your TSP program TSP will check your whole program for syntax and then stop as soon as it reaches execution This can be useful for debugging long programs Output STOP produces no printed output If output databanks have been used variables are stored on them before stopping the program Examples Here is an example of using STOP to check a TSP program for syntax STOP Abort execution before doing anything SMPL 1 1000 long involved TSP program including complex equations etc END 419 Commands STORE Example Reference STORE writes microTSP format or EViews format databank files STORE list of series STORE drive letter series name gt PC
64. SIGNIF may also cause single column output QLAGS maximum number of autocorrelations for Ljung Box Q statistics Portmanteau test of residual autocorrelation The default is zero RESETORD order of Ramsey s RESET test The default is 2 SHORTLAB NOSHORTL indicates whether short or long labels are used when printing all diagnostics STAR1 upper bound on p value for printing at least one star when STARS option is on The default is 05 There can be up to 5 pairs of STAR1 STAR2 values which can apply to different sets of diagnostics This option only applies to the diagnostics listed for the REGOPT command STAR2 upper bound on p value for printing two stars when STARS option is on The default is 01 This option only applies to the diagnostics listed for the REGOPT command 377 Commands STARS NOSTARS indicates whether stars should be printed indicating significance of diagnostics STARS implies PVCALC except for regression coefficients T Examples REGOPT STARS LMLAGS 5 QLAGS 5 BPLIST C X X2 ALL turns on all possible diagnostic output including VCOV matrix and residual plots REGOPT restores the default settings REGOPT NOCALC AUTO stops calculation of all the autocorrelation diagnostics useful for pure cross sectional datasets REGOPT NOPRINT RSQ FST suppresses printing of the R squared and F statistics This is the same as the old TSP command SUPRES RSQ FST REGOPT STARS
65. STAR1 10 STAR2 05 T REGOPT STARS STAR1 05 STAR2 02 AUTO uses one set of significance levels for the t statistics and another for the autocorrelation diagnostics Output Summary table of diagnostics OLSQ output Name value Name p value Group Name Description None LHV Dependent variable name SMPL Current sample NOB Number of observations COEF Regression coefficients SES Standard errors T t statistics VCOV Variance covariance matrix VCOR Correlation version of VCOV NCOEF Number of coefficients 378 REGOUT AUTO HET None NCID YMEAN SDEV SSR 52 5 RSQ ARSQ DW DH DHALT LMARx QSTATx WNLAR ARCH RECRES CUSUM CUSUMSQ CSMAX CSQMAX CHOW CHOWHET LRHET WHITEHT BPHET LMHET FST RESETx JB SWILK AIC SBIC REGOPT Number of identified coefficients rank of VCOV Mean of dependent variable Standard deviation of dependent variable Sum of squared residuals Estimated variance of residuals SSR NOB NCID Standard error of residuals SQRT S2 R squared squared correlation between actual and fitted Adjusted R squared adjusted for number of RHS variables Durbin Watson statistic Durbin s h statistic for single lagged dependent Var Durbin s h alternative for any lagged dependent Breusch Godfrey LM test for autocorrelation of order x Ljung Box Q statistic for autocorrelation of order X Wald test for nonlinear AR1 restriction vs Y 1 X 1 Te
66. THETA 2 0 20 DELTA 1 0 82 Note that NBACK is specified as 15 since the backcasted residuals fall to exactly zero after NSPAN NSMA NMA periods in a pure moving average model The next example estimates a third order autoregressive process with one parameter fixed 61 Commands BJEST GNP START PHI 2 0 5 PHI 3 0 1 FIX PHI 1 0 9 The model being estimated is GNP t 0 9 GNP t 1 2 GNP t 2 3 GNP t 3 a t Exact ML estimation QSTART 1 1 1 BJEST NAR 2 NMA 1 NDIFF 1 EXACTML Y References Box George P and Gwilym M Jenkins Times Series Analysis Forecasting and Control Holden Day New York 1976 Ljung G M and Box George P On a measure of lack of fit in times series models Biometrika 66 1978 pp 297 303 M lard G Algorithm AS 197 A Fast Algorithm for the Exact Likelihood of Autoregressive moving Average Models Applied Statistics 1984 p 104 109 code available on StatLib Nelson Charles Applied Times Series Analysis for Managerial Forecasting Holden Day New York 1973 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Co New York 1976 Chapter 15 Statlib http lib stat cmu edu apstat 62 BJFRCST BJFRCST Output Options Examples References BJFRCST calculates the forecasts of a time series that are implied by an ARIMA model see the description of BJEST for more information of ARIMA models
67. The forecasts and other statistics are calculated using the equations in Chapter 5 and Part V program 4 of Box and Jenkins BJFRCST CONBOUNDz probability for forecasting bounds CONSTANT EXP NAR lt number of AR parameters gt NBACK lt number of back forecasting residuals gt NDIFF lt degree of ordinary differencing gt NHORIZ lt length of forecasting horizon NLAG lt number of lagged observations to plot NMAz number of parameters NSAR lt number of seasonal AR parameters gt NSDIFF lt degree of seasonal differencing gt NSMA lt number of seasonal MA parameters gt NSPAN lt seasonal frequency ORGBEG lt first forecasting origin gt ORGENDz final forecasting origin PLOT PREVIEW PRINT RETRIEVE SILENT series name S standard error of disturbance parameter name parameter value Usage The entry for the BJEST procedure describes the ARIMA model and its notation in some detail To obtain forecasts for such a model use the BJFRCST command followed by the options you want in parentheses and then the name of the series If the BJFRCST command immediately follows the BJEST command for the desired model no parameter values need to be specified BUFRCST will automatically use the final estimates generated by BJEST Alternatively any or all of the model parameters can be specified by listing the parameters and their values following the series name as in BJEST If no value is s
68. The full syntax of the IF THEN ELSE sequence is IF expression THEN statement or block of statements ELSE statement or block of statements ELSE Usage ELSE has no arguments It is optional following an IF statement If several ELSE statements appear in sequence the first refers to the most recent IF clause the second to the next most recent and so forth If more than one statement or an INPUT command is to be executed when the IF clause is false enclose all the statements in a DO ENDDO group Example IF SSR LIMIT THEN SET RESULT SSR ELSE DO GENR Y Y INCR OLSQ Y C POP TIME SET RESULT SSR ENDDO In the above example the statements in the DO group are executed once and only once when the original value of SSR is less than or equal to the value of LIMIT 132 END END Example END terminates both the TSP program section and the TSP data section END Usage END has no arguments END is also used as a delimiter In the data section itis used to mark the end of the current section of data and to force a return to execution of the TSP program If there is more than one LOAD statement in the program section each one will terminate at successive END statements in the data section Example USER LOAD M execution with data from load section STOP END errs load section with data END 133 Commands
69. These P values are fine for testing at the 596 and 10 levels but they are not accurate for testing at other levels The P value for the case with a constant and a trend is only good for testing at the 5 level 86 COINT The regressions for the Dickey Fuller tests are quite simple See the example below which reproduces the Dickey Fuller tests in the Examples section below SMPL 10 70 DY LRGNP LRGNP 1 Sample for comparing AIC is the same for all lags MAXLAG 1 observations are dropped SMPL 20 70 TREND T DO LAG 1 10 SET MLAG LAG OLSQ LRGNP LRGNP 1 C T DY 1 DY MLAG SET alpha COEF 1 SET tauDF alpha 1 95 5 1 CDF DICKEYF tauDF ENDDO If you are computing this test by hand it is easier to use OLSQ DY LRGNP 1 C T DY 1 DY MLAG CDF DICKEYF T 1 The Phillips Perron test is done with the same Dickey Fuller regression variables using no augmenting lags This test is given in Davidson and MacKinnon equations 20 17 and 20 18 see also the warnings there about the possibly poor finite sample behavior of this test These tests can be computed for 1 to 10 lags by using the following TSP commands following the Dickey Fuller example above OLSQ silent LRGNP C T SET ssr SSR to ensure the same sample for each test SMPL 10 70 TREND T OLSQ LRGNP LRGNP 1 C T Y RES SET alpha COEF 1 SET s2 252 SET n NOB FRML EQPP Y z YO PARAM YO DO LAGz1 1
70. against the first The graph is labeled and surrounded by a box If more than one point is plotted in the same place on the graph the number of such points will be printed if it is less than or equal to 9 If there are more than 10 to 35 observations at one point A to Z will be plotted If there are more than 35 observations at one point an X will be plotted and the number of such points will be listed after the graph sorted by the X values Any observations with missing values will not be plotted and a warning message will be printed The size of the graph is controlled by the line printer width option LIMPRN and the page length option LINLIM These options are automatically set at the beginning of your TSP run but they may be changed by using the OPTIONS command If you have already issued an OPTIONS CRT command they will be set at 80 characters wide by 24 characters high GRAPH does not store any data in data storage Options Examples GRAPH TIME COSINE GRAPH Y X 176 GRAPH graphics version GRAPH graphics version Output Options Examples GRAPH plots one series versus another using a scale determined by the range of each series The plot may be printed as well as displayed if a hardcopy device such as a Laserjet or dot matrix printer is available This section describes the graphics version of GRAPH which is available only for TSP Givewin Macintosh TSP unix and DOS Win TSP For other versions see the non graph
71. and fail to end it properly before returning to interactive mode e g 1 COLLECT Enter commands to be collected 95 Commands 2 gt PROC X 3 gt other statements 4 gt ENDPROC lt proper end to structure 5 gt EXIT Collect mode is always entered from and control always returned to the interactive mode 96 COMPRESS COMPRESS Options COMPRESS removes unused TSP variables and frees up wasted working space COMPRESS PRINT Usage If you are trying to fit a particularly large TSP program onto a computer with limited working space this command may be useful It deletes internal variables such as old program lines and GENR formulas variables whose names begin with such as RES and variables which have been marked with the DELETE command It also frees up the space associated with these variables and the space used by old copies of variables which have grown in size This command is not usually necessary since TSP does an automatic compress operation when it runs out of working space or when too many TSP variables have been defined However in some cases this automatic compress will not work For instance no compression takes place inside a PROC which has arguments and internal program lines are deleted only up to the first DO loop first PROC or current DOT loop Automatic compression also may not clear enough space when creating a large matrix In these cases use the COMPRESS command first before ru
72. are fewer dots in a name than loops the innermost loop is used first The DOT procedure cannot be used in a LOAD section however it can be used when reading data from files Legal dotted variable names are A VAR D and ALB Illegal dotted variable names are 1 2 numbers and EQ logical operator Options 122 DOT special character to be used conjunction with in nested DOT loops in order to specify explicitly which DOT statement is used to expand the period It is normally used only to make a single refer to the outer DOT loop when inside an inner loop INDEX scalar variable name to hold the values 1 2 3 during the DOT loop This is like having a DOT loop and a DO loop at the same time VALUE scalar variable name to hold the values of numeric DOT sectors Examples Suppose that you have data on the rate of change of wages DW the unemployment rate U and prices P for each of four countries 1 United States 2 England 3 Sweden and 4 Germany When you create and load the data give the series names like DW1 DW2 DW3 DW4 U1 U2 etc Then you can easily run the same set of TSP statements on all four countries by means of a DOT loop For example to normalize the series to the same year compute the rate of change in prices and run a regression of DW on the unemployment and rate of change of prices use the following DOT 1 4 NORMAL P 72 100 GENR DP LOG P P 1
73. as in the other linear and nonlinear regression procedures in TSP If you wish to estimate a nonstandard ordered probit model e g adjusted for heteroskedasticity or with a nonlinear regression function use the ML command See the website for examples of how to do this Options LOWERc scalar or series containing lower bounds required 202 INTERVAL UPPERcz scalar or series containing upper bounds required The usual nonlinear options are available see the NONLINEAR section of this manual Example A simple example showing how to estimate a binary Probit model using PROBIT and INTERVAL with scalars as the lower and upper bounds for the dependent variable PROBITD C X1 X8 Probit estimation D 0 or 1 Q 2 D 1 redefine dep variable to be 1 0 INTERVAL LOWER 0 UPPER 0 Q C X1 X8 A more complex example where there are 4 categories lt 40 40 to 50 50 to 60 and gt 60 showing how to code the lower and upper bounds and the dependent variables YCAT takes on the values 1 to 4 corresponding to the four categories yrec 35 ycat 1 45 ycat 2 55 ycat 3 65 ycat 4 ylo 40 ycat 1 40 ycat 2 50 ycat 3 60 ycat 4 yhi 40 ycat 1 50 ycat 2 60 ycat 3 60 ycat 4 interval lowerzylo upperzyhi c x1 x8 Note that by coding the upper and lower limits to be equal for YCAT 1 and YCAT 4 we have specified that they represent a single bound upper in the case of YCAT 1 and lower in th
74. be described as conditional ML conditional on the initial residual EXACTML is the usual default but if there is a lagged dependent variable on the right hand side GLS becomes the default because EXACTML has a small sample bias in this case GLS uses an initial grid search to locate starting values and potential multiple local optima It is well known that multiple local optima can occur for GLS especially when there are lagged dependent variables Multiple optima are noted in the output if they are detected AR1 then iterates efficiently to locate an accurate global optimum EXACTML normally skips the grid search because no cases of multiple local optima are known when the Jacobian is included METHOD MLGRID will turn on this grid search REI NOREI specifies that AR 1 model with panel random effects is to be estimated by means of maximum likelihood GLS is not available for this model RMINz specifies the minimum value of the serial correlation parameter rho for the initial grid search when OBJFN GLS or METHOD MLGRID are used The default value is 0 9 specifies the maximum value of rho for the grid search methods The default value is 1 05 for OBJFN GLS or 95 for METHOD MLGRID RSTEP specifies the increment to be used the grid search over rho The default value is 0 1 until rho 8 Then the values 85 9 95 are used plus 9999 1 0001 and 1 05 when OBJFN GLS These last 3 values help to detect op
75. can be used Nonlinear two stage least squares for this estimator H is the matrix of instrumental variables formed from the variables in the INST option and Sis again assumed to be unity The estimator is described in Amemiya 1974 If the model is linear conventional two stage least squares or instrumental variable estimates result Nonlinear multivariate regression this case there are no instruments H is the identity matrix and S is either estimated or fixed This estimator can be computed with two completely different objective functions The default in TSP is to compute maximum likelihood estimates if the LSQ command is specified with no instruments and more than one equation These estimates are obtained by concentrating variance parameters out of the multivariate likelihood and then maximizing the negative of the log determinant of the residual covariance matrix They are efficient if the disturbances are multivariate normal and identically distributed Using the option MAXITWZO it is possible to obtain minimum distance estimates of a nonlinear multivariate regression model For these estimates the objective function is the distance function given above with the instrument matrix H equal to identity The S matrix is given by the WNAME option it can be identity which is similar to estimating each equation separately except that cross equation constraints will be enforced and the parameter standard errors will be wrong unless
76. capital stock at the beginning of the computation is assumed to be zero To alter these assumptions see the options below CAPITL requires that there be no gaps in the current SMPL Output CAPITL produces no printed output The capital stock series is stored in data storage 72 CAPITL Options 5 an observation identifier for the benchmark observation This identifier should be contained in the current SMPL If the frequency is quarterly and the SMPL is 47 4 80 4 for example the benchmark observation could be 47 4 56 1 80 4 etc BENCHVAL the value of the capital stock series at the benchmark observation CAPITL will compute the capital stock both forwards and backwards from this observation END NOEND computes the end of period capital stock see the previous formulas Examples SMPL 1 74 CAPITL BENCHVAL 145 4 BENCHOBS 4 INV 04 KSTOCK In this example the gross investment series is INV CAPITL computes capital KSTOCK The benchmark applies to the 4th observation and has the value 145 4 The rate of depreciation is 04 CAPITL BENCHOBS 1 BENCHVAL X 1 END X 0 0 XACCUM This example simply sums the series X and stores the result in XACCUM Note that since the formula gives the end of period capital stock the last observation of XACCUM contains the sum of all the observations on X 73 Commands CDF Output Options Examples References CDF calculates and prints tail probabilities P
77. compute such squared residuals if such observations are missing some observations from the current sample are used GT a list of weighting series g k t The estimated coefficients for these series are labeled as PHI_series1 PHI series2 etc in the output Note that the constant C should not be in the GT list because it is already included in h t via a 0 If GT is used without NAR or NMA the model is called OLS W or OLS M by TSP Use the RELAX option to relax the constraint on phi HEXP value of the exponent of the conditional variance h t in the regression equation The default value of HEXP is 0 5 which means that the disturbance depends on the standard deviation of its distribution HINIT ESTALL or OLS or SSR or STEADY or value specifies the initialization of the presample values of ht ESTALL estimates them as nuisance parameters labelled H 0 H 1 etc this was the default in TSP Version 4 3 and earlier OLS holds them fixed at the initial ML estimate of the residual variance from OLS The default SSR sets them equal to SSR T where SSR is the sum of squared residuals from the current parameter values see Fiorentini et al 1996 STEADY sets them equal to their steady state value hy NAR NMA PD j Note that the denominator must be strictly positive and that the expectation of phi k g k t is assumed to be zero The user may also specify an arbitrary fixed value 51 Commands MEAN NOMEAN controls
78. current SMPL vector If there are different equations depending on different cases write the equation as the sum of the individual equations with dummy variables multiplying each equation to select the appropriate one for any given observation Often the case will be determined by a dependent discrete choice variable and observations are usually i i d but the likelihood function could be made different for different parts of the SMPL by using more general time dependent dummy variables Of course the log likelihood must be additively separable over the sample for this method to work if it is not use the PROC method described below Use a PARAM statement to specify which of the variables in the LOGL equation are to be estimated and supply their starting values if desired Follow this by an ML command with any of the standard NONLINEAR options and the name of the equation which specifies the likelihood function ML will maximize this function with respect to the parameters using a standard gradient method the exact form of the Hessian approximation used as a weighting matrix depends on the HITER option The default is to use the BHHH method a method of scoring but with the sample covariance of the gradient of the likelihood used in place of its expectation Good Applications for the FRML method 1 Truncation models involving CNORM such as two limit Tobit 2 Nonlinear equations for PROBIT TOBIT LOGIT etc This includes para
79. data by using an OUT statement followed by a KEEP statement explicitly naming everything you want stored at the very end of your run Set MAXERR to zero and TSP aborts before storing the data if it encounters any errors Supplying the keyword ALL forces all variables to be stored in any open databanks Note that PROCs cannot be stored in a databank Output KEEP produces no printed output The variables named are placed in data storage with flags so they will be stored in the appropriate databank at the end of the TSP run Examples OUT PDATA FRML 1 Y1 A1 B11 LNP1 B12 LNP2 A1 G11 LNP1 G12 LNP2 FRML 2 Y2 A2 B12 LNP1 B22 LNP2 A1 G12 LNP1 G22 LNP2 PARAM 1 A2 B11 B12 B22 G11 G12 022 LSQ EQ1 EQ2 KEEP EQ1 EQ2 A1 A2 B11 B12 B22 G11 G12 G22 210 KEEP Databank This example specifies and estimates a two equation nonlinear least squares model and saves the equations and parameter estimates on the databank PDATA Note that the LSQ statement causes the current parameter values to be stored automatically when it finishes so that it is redundant to specify them on the KEEP statement However the equations EQ1 and EQ2 will not be stored unless you name them on a KEEP statement DOT US UK SWD OUT DB KEEP GNP CONS P DP DW U ENDDOT This example shows how data can be placed in different databanks in one run Identical databanks have been created to hold the series of each of four countries separa
80. displayed on screen as it executes or to have it sent to another file Either way the commands read and executed become part of your TSP session and may be REVIEWed EDITed EXECed etc If you send the output to a disk file you will be prompted for an output filename This name may also be up to 128 characters long If no extension is given out will be assumed If you do not provide a filename it will default to the same name as the input file except with the out extension If another file already has this name it will be overwritten unless your system allows multiple versions of the same file If you want the command stream to be read and stored but NOT executed use an EXIT command at the end of the input file instead of END This is the same as using the EXIT command in collect mode You may then REVIEW the commands read and EXEC selected sections 192 INPUT INPUT commands can be nested they may appear in files used for input login tsp is a special INPUT file it is read automatically at the start of interactive sessions and batch jobs this is useful for setting default options Examples 1 INPUT ILLUS Do you want the output printed at the terminal y n yJN Enter name of TSP output file This example reads the illustrative example from ILLUS TSP in the current directory and places the output in ILLUS OUT When execution is finished the prompt will reappear on screen the exact line number will depend on how m
81. equations and the default is a matrix of ones all instruments used for all equations NMA number of autocorrelation terms AR and or MA to be used in computing COVOC Some forecasting type models imply a given NMA value but other models have no natural choice See the Andrews reference for automatic bandwidth selection procedures When there are missing values in the series NMA does not include terms which cross the gaps in the data This is useful in panel data estimation OPTCOV NOOPTCOV specifies whether or not the COVOC matrix is optimal Under the default NOOPTCOV the VCOV matrix is computed using the sandwich formula of Hansen s 1982 Theorem 3 1 p 1042 This is appropriate for example if the user has supplied a COVOC matrix but has not scaled it properly Note in this improperly scaled COVOC case the GMMOVID statistic will be invalid When OPTCOV is in effect formula 10 of Hansen s Theorem 3 2 p 1048 is used for the VCOV matrix When the user has not supplied a COVOC matrix the OPTCOV and NOOPTCOV options produce almost exactly the same results The only difference is due to the difference between the COVOC matrix that was used for iterations and the COVOC matrix evaluated at the final parameter residual values This difference is usually small Examples GMM 5 21 210 2 EQ1 EQ2 173 Commands To exclude 72 as instrument for EQ1 Z1 as instrument for EQ2 READ NROW 3 NCOL 2 SE
82. estimation LOGIT NCHOIC 2 WORK C SCHOOL EXPER RACE UNMAKE COEF C1 C4 MMAKE START C1 C4 0 LOGIT WORK C SCHOOL EXPER RACE MSTAT This example makes a matrix of regression output to be written to disk 266 MMAKE LOGIT WORK C SCHOOL EXPER RACE MSTAT MMAKE REGTAB COEF SES T WRITE FILE REGTAB ASC FORMAT 3F10 5 REGTAB The final example creates the partitioned matrix A where D E and F are matrices with the same number of rows D and G have the same number of columns etc MMAKE DEFDEF GHI GHI MMAKE VERT A DEF GHI 267 Commands MODEL Output Options Examples References MODEL determines the order in which the equations of a model should be solved and saves this order under a collected model name It must be used before a SOLVE command invokes the model simulation procedure MODEL DONGALLO FILEz filename PRINT SILENT equation list lt endogenous variable list ordered model name gt Usage MODEL takes as its arguments the name of a list of equations in the model and produces a collected and ordered model which is stored under the name supplied by the user Each of the endogenous variables in the list must appear on the left hand side of one and only one of the equations For compatibility with older versions of TSP you may supply the endogenous variable list but for the current version this is optional Output MODEL prints information about the orde
83. fixed effects TAI vector vars t statistics on fixed effects FEI vector vars p values for t statistics FEs FEI VCOV matrix vars vars Variance covariance of estimated coefficients AI series obs Fixed effect for each obs in series form FEI RES series obs Fitted residuals from model FIT series obs Fitted values of dependent variable If the regression includes PDL variables SLAG MLAG and LAGF will also be stored see OLSQ for details Method AR1 uses an initial grid search to local possible multiple local optima when OBJFN GLS and then iterates efficiently to a global optimum with second derivatives The likelihood function and treatment of the initial observation are described completely in Davidson and MacKinnon 1993 When OBJFN EXACTML the default AR simply maximizes the likelihood function for disturbances that follow a stationary autoregressive process with respect to the serial correlation rho and the coefficients of the independent variables For panel data AR1 with fixed FEI or random REI effects is similar to the corresponding PANEL regressions but with an added AR 1 component The random effects estimator follows Baltagi and Li 1991 It uses analytic second derivatives to obtain quadratic convergence and accurate t statistics for all parameters including RHO and the intraclass correlation coefficient which can be negative After the fixed ef
84. fixed number of categories such as Ordered Probit and Multinomial Logit Options See the NONLINEAR section of this manual for the usual nonlinear options Example Ordered Probit regression of patents on lags of log R amp D science sector dummy and firm size ORDPROB PATENTS LRND LRND 1 LRND 2 DSCI SIZE References Cameron A Colin and Pravin K Trivedi Regression Analysis of Count Data Cambridge University Press New York 1998 pp 87 88 Estrella Arturo A New Measure of Fit for Equations with Dichotomous Dependent Variables Journal of Business and Economic Statistics April 1998 pp 198 205 Maddala G S Limited dependent and Qualitative Variables Econometrics Cambridge University Press New York 1983 pp 46 49 305 Commands ORTHON Example ORTHON orthonormalizes arbitrary matrix and saves the orthonormalizing transformation The columns of the resulting matrix span the same space as the columns of the original matrix but are orthonormal orthogonal and scaled so that their Euclidean norm is one ORTHON input matrix triangular matrix orthonormalized matrix Usage The input matrix X is a general NROW by NCOL matrix ORTHON obtains a triangular matrix S of order NROW such that X X S S It uses the inverse of S to transform X by postmultiplying it S inverse and the orthonormalized X are returned in the second and third arguments to the procedure ORTHON transforms a da
85. for the amount of data available SMPL 1 9 MFORM TYPE SYM NROW 3 XSYMzX yields XSYM 20 50 60 30 60 90 MFORM TYPEZTRIANG NCOLz3 X yields X MFORM 0 0 90 MFORM TYPE DIAG NCOL 3 X yields X 10 0 o o 50 10 o 0 90 MFORM TYPE DIAG NROW 9 X or MAT X DIAG X yields X The next example shows how you can make a diagonal matrix from a column vector if the input series or matrix is too short to make a diagonal matrix by selecting diagonal elements the whole vector becomes the diagonal MFORM TYPE DIAG NCOL 3 X 2 or MAT X 2 IDENT 3 yields X A band matrix MMAKE BVEC 2 1 READ NROW 2 TYPE SYM CORNER 257 Commands 11 21 22 MFORM BAND NROWZz5 B5 BVEC CORNER yields B5 258 ML ML Output Options Examples References ML is a general purpose maximum likelihood estimation procedure It can be used to estimate the parameters of any identified model for which you can write down the logarithm of the likelihood in a TSP equation FRML or evaluate the log likelihood in a procedure PROC ML nonlinear options lt log likelihood equation name gt or ML nonlinear options lt procedure name gt lt list of parameters gt Usage FRML method Usually the simplest approach is to write the log likelihood equation in a FRML with LOGL as the dependent variable Note that this equation is for each observation in the
86. from a series or vector or to change the dimensions or type of an existing matrix The matrix may be transposed as it is formed Matrices can be renamed or copied without reformatting with the RENAME or COPY commands MFORM NROW lt of rows in matrix NCOL lt of columns in matrix TRANS TYPE GENERAL or SYMMETRIC or TRIANG or DIAG variable name gt or new matrix old variable or old variable new matrix or new matrix scalar or MFORM BAND NROWznrows new matrix band vector lt corner matrix or MFORM BLOCK lt new matrix gt lt list of matrices gt Usage If there is only one argument either an existing matrix or series is transformed in place or a new matrix is initialized to zero The options specify the characteristics of the new matrix the number of rows and columns the type and whether it is to be transposed as it is formed When there is more than one argument the old variable may be a series a matrix a vector or a scalar The new variable will be a matrix of the type specified on the command If no type is specified a general matrix is created If the input variable is a series it is retrieved under control of SMPL and only those observations in the current sample are placed in the matrix If the input series or matrix is longer than NROW NCOL it is truncated except in the case of the DIAG type see the examples If it is shorter an error message is given
87. h i is zero for some i and e i is nonzero HCTYPE 1 is used Both HCTYPE 2 and HCTYPE 3 have good finite sample properties See Davidson and MacKinnon pp 552 556 for details HI NOHI specifies whether the diagonal of the hat matrix is stored in the series HI The hat matrix is defined as H X X X X This is useful for detecting influential observations data errors outliers etc For example SELECT HI gt 2 NCOEF NOB identifies the influential observations See the Belsley Kuh and Welsch or Krasker Kuh and Welsch references NORM UNNORM tells whether the weights are to be normalized so that they sum to the number of observations This has no effect on the coefficient estimates and most of the statistics but it makes the magnitude of the unweighted and weighted data the same on average which may help in interpreting the results The coefficient standard errors and t statistics are affected NORM has no effect if the WEIGHT option has not been specified ROBUSTSE NOROBUST causes the variance of the coefficient estimates the standard errors and associated t statistics to be computed using the formulas suggested by White among others These estimates of the variance are consistent even when the disturbances are not homoskedastic and when their variances are correlated with the independent variables in the model They are not consistent when the disturbances are not independent however See the Davidson and Mac
88. has good power in small samples Since it involves sorting the residuals it may be quite slow in large samples The test and its P value are computed using Royston 1995 with code from Statlib AIC Akaike Information Criterion and or SBIC Schwarz Bayesian Information Criterion can be minimized to select regressors in a model such as choosing the length of a distributed lag SBIC has optimal properties see Geweke 1981 In general these can be defined as AIC LOGL NCID 2 SBIC LOGL NCID LOG NOB 2 LOGL will include the sum of log weights if the OLSQ WTYPE HET WEIGHT x option is used The alternative is the default WTYPE REPEAT Distributions used for P values Note in all cases k is the number of identified coefficients in the model including the intercept Test Null Alternative Distribution Degrees of Statistic Freedom DW No autocorrelation Positive ratio of Qform autocorrelation usually DH No autocorrelation Normal DHALT No autocorrelation Normal LMARx No autocorrelation Autocorrelation of Chi squared pt k 1 order x QSTATX No autocorrelation Autocorrelation of Chi squared p order x WNLAR AR 1 disturbance Other dynamics Chi squared rhs vars ARCH Homoskedasticity 1 disturbance Chi squared 1 CSMAX Stable parameters Parameters change Durbin 1971 CSQMAX Stable parameters Parameters change Durbin 1969 CHOW Stable parameters Parameters differ F k nob 2k between tw
89. have parameters smaller than 00001 magnitude it will be helpful to use an EPSMIN with a value somewhat smaller than your smallest parameter Otherwise too large a stepsize is used and the parameters will appear to have zero standard errors GRADCHEC NOGRADCHEC evaluates and compares the analytic and numerical gradient for the current model at the starting values No actual estimation takes place Useful for checking derivatives of a new likelihood function The numeric gradient is evaluated in a time consuming but accurate way See the GRAD C4 option GRADIENT ANALYTIC or C2 or C4 or FORWARD specifies the method of calculating numeric first derivatives is the default when analytic first derivatives are available as is usually the case GRAD FORWARD calculates the numeric derivatives for a given parameter B as D F B EPS F B EPS 1 function evaluation per parameter GRAD C2 CENTRAL2 uses D F B EPS F B EPS 2 EPS 2 function evaluations per parameter GRAD C4 uses D F B 2 EPS 8 F B EPS 8 F B EPS F B 2 EPS 12 EPS 4 function evaluations per parameter In all cases EPS MAX ABS 001 B EPSMIN 279 Commands or N or G or F or D or Wor R or P or or U or C or BNW etc specifies the method for calculating the asymptotic covariance matrix of the parameter estimates and standard errors The default is usually N or B depending on the procedure Some pr
90. in interpreting the results The NORM option has no effect if the WEIGHT option has not been specified ROBUSTSE NOROBUST causes the variance of the coefficient estimates the standard errors and associated t statistics to be computed using the formulas suggested by White among others These estimates of the variance are consistent even when the disturbances are not homoskedastic although they must be independent and when their variances are correlated with the independent variables in the model See the references for the exact formulas When FEI is specified with ROBUST the standard errors for the fixed effects will still be conventional estimates SILENT NOSILENT suppresses all output The results are still stored TERSE NOTERSE suppresses printing of everything but the e P Z e objective function and the table of coefficients WEIGHT the name of a series which will be used to weight the observations The data and the instruments are multiplied by the square roots of the weighting series before the regression is computed so that the weighting series should be proportional to the inverses of the variances of the disturbances If the weight is zero for a particular observation that observation is not included in the computations nor is it counted in determining degrees of freedom This option is not available with FEI Examples This example estimates the consumption function for the illustrative model using the constant trend gov
91. in free format such as READ FILEZ FOO DAT XYZ 362 READ There are two special features for free format numbers The first is the use of the dot to specify a missing value and is similar to the SAS convention However note that in other places in TSP such as in formulas dot is treated as a DOT variable Use MISS or NA to represent a missing value in a formula see FRML The other special feature is a repeated number specified by an integer repeat count a star and the value to be repeated For example 3 0 is equivalent to 0 0 0 This is most often used for repeated zeroes in special matrices like band matrices and resembles the repeat count feature in the FORTRAN free format READ When the data is read in free format the number of items read must be equal to the number of series times the number of observations in the current SMPL TSP checks for this and prints a message when the check fails TSP will determine the length of the SMPL itself if the SETSMPL option is on To read matrices use a READ statement with options to define the matrix and the matrixs name for storage Follow the statement with the numbers which compose the matrix in free format a row at a time That is a 3 rows by 2 of columns matrix is read in the following order 1 1 1 2 2 1 2 2 3 1 3 2 A matrix of any type may be read by specifying all its elements but there are special forms for reading symmetr
92. in the current sample L MAXLAG order of VAR beyond 1 The trace tests are labelled HO 0 HO r lt 1 etc in the table of results Note that the trace test includes a finite sample correction mentioned in Gregory 1994 originally given in Bartlett 1941 These trace tests often have size distortions the null of no cointegration or fewer cointegrating vectors is rejected when it is actually true P values are interpolated from the Osterwald Lenum 1992 tables 0 1 1 and 2 with no constant constant or constant amp trend These P values are adequate for testing at the sizes given in the Osterwald Lenum tables 50 20 10 05 025 and 01 See Cushman et al 1995 for a detailed example of using Johansen tests in an applied setting They illustrate the importance of the finite sample degrees of freedom correction the size distortions of the P values lag length choice methods and hypothesis testing Options Unit Root Test Options ALL NOALL perform all available types of unit root tests WS DF and PP DF NODF perform augmented Dickey Fuller tau tests PP NOPP perform the Phillips Perron variation of the Dickey Fuller 2 test For the PP test the number of lags used is the order of the autocorrelation robust T2 long run variance estimate see the MAXLAG option WS NOWS perform augmented Weighted Symmetric tau tests This test seems to dominate the Dickey Fuller test and others in terms of power
93. in the standard K class formula to compute the coefficients The standard errors for the NOBEKKER option are computed from the K class inverse matrix times the sum of squared residuals divided by number of observations minus number of estimated coefficients Options BEKKER NOBEKKER specifies that Bekker standard errors are to be computed see Hansen Hausman and Newey 2004 These standard errors are better for small samples and or when there are large numbers of excluded instruments FEI NOFEI specifies whether a model with individual fixed effects is to be estimated FREQ PANEL must be in effect FEPRINT NOFEPRINT specifies whether the estimated effects and their standard errors are to be printed FULLER value used to weight the eigenvalue towards zero The formula used is K LAMBDA FULLER T NZ where K is the K class constant LAMBDA is the LIML eigenvalue T is the number of observations and NZ is the number of instruments FULLER 0 default is the standard LIML estimator which is median unbiased FULLER 1 yields a mean unbiased estimator FULLER values between 0 and 8 16 T NZ 2 dominate LIML in small sigma efficiency The LIML estimator modified in this way has smaller tails than the standard LIML estimator which gives it good small sample properties see the references for details INST list of instruments This list should include all the exogenous variables in the equation being estimated as well as the oth
94. independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the current sample NEGBIN will print a warning message and will drop those observations NEGBIN also checks that the observations on the dependent variable are integers and are not negative The list of independent variables on the NEGBIN command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See the PDL section for a description of how to specify such variables Output The output of NEGBIN begins with an equation title and frequency counts for the lowest 10 values of the dependent variable Starting values and diagnostic output from the iterations will be printed Final convergence status is printed 274 NEGBIN This is followed by the number of observations mean and standard deviation of the dependent variable sum of squared residuals correlation type R squared likelihood ratio test for zero slopes
95. is not to plot a band BMEAN NOBMEAN causes the band to be printed about the series mean BMID NOBMID causes the band to be printed about the midpoint of the plot draws a box around the plot HEADER NOHEADER causes the horizontal axis to be labelled at equispaced intervals ID NOID causes a vertical time axis to be labelled on the left hand side with the ID series INTEGER NOINTEGER causes the numeric labels on the horizontal axis to be rounded to the nearest integer value this improves readability of the plot LINES ist of up to 9 numeric values specifies points along the horizontal axis at which vertical lines will be drawn MAX the maximum value on the horizontal axis If not specified the maximum value of all the series to be plotted is used MEAN NOMEAN draws a vertical line from the mean of the series on the horizontal axis MIN the minimum value on the horizontal axis If not specified the minimum value of all the series to be plotted is used ORIGIN NOORIGIN causes a vertical line to be drawn starting at zero on the horizontal axis RESTORE NORESTORE causes the options to be set to their default values VALUES NOVALUES causes the value of each observation of the first series to be printed on the right hand side of the plot 329 Commands The list of options is obviously extensive and to make things easier for the user a set of default options has been chosen which produce a plot
96. it implicitly assumes that the x axis series will be sorted by the user before the procedure is executed PREVIEW NOPREVIE specifies whether the graph is to be shown on the screen before printing or saving The default is PREVIEW for interactive use and NOPREVIE for batch use SORT NOSORT controls the ordering of the data on the X axis so that a line connecting the points will not cross itself SYMBOL NOSYMBOL specifies that symbols are to be used for plotting When the LINE option is on NOSYMBOL is the default TITLE a string which will be printed across the top of the graph XMAX maximum value for the x axis This value must be greater than or equal to the maximum value of the x series XMIN minimum value for the x axis This value must be less than or equal to the minimum value of the x series 178 GRAPH graphics version YMAXz maximum value for the y axis This value must be greater than or equal to the maximum value of the y series YMINz minimum value for the y axis This value must be less than or equal to the minimum value of the y series For convenience the DEVICE FILE and HEIGHT options of GRAPH are retained for the next PLOT s or GRAPH s until they are overridden explicitly These options may also be set in a LOGIN TSP file with a GRAPH statement which does not specify any series to graph Givewin only SURFACE NOSURFACE specifies that a three dimensional surface plot is to be created This optio
97. model will be solved They do not have to be predefined unless you wish to supply starting values or you are doing a static simulation with lagged endogenous variables METHOD controls the action to be taken if the model becomes singular For METHOD NEWTON the default iteration stops at a singular point For METHOD GAUSSN a generalized inverse is used one or more variables are temporarily excluded from the model and are held constant for the iteration PRNRES NOPRNRES prints the residuals when solution is complete for each time period All residuals will be small enough to satisfy the convergence criterion The Jacobian for the first two time periods is also printed PRNDAT NOPRNDAT prints all the endogenous and exogenous variables at the beginning of the simulation PRNSIM NOPRNSIM prints a table containing the solved values of the endogenous variables 405 Commands SILENT NOSILENT suppresses all the output STATIC NOSTATIC specifies static simulation Actual values of lagged endogenous variables are used not earlier solved values tagname specifies that the solved values of all endogenous variables should be stored as series with names created by adding tagname to the variable names tagname should be a single character or perhaps two to avoid creating excessively long names Names larger than the allowed length of a TSP name will be truncated TAG NONE stores under the original endogenous names T
98. observations or fill in missing values as appropriate FREQ SMPL 4675 IN TSPDATA GENR CONSL1 CONS 1 OLSQ CONS GNP CONSL1 Other examples of legal IN statements IN BANK1 BANK2 BANKS IN cancel the previous databanks IN USDATA UKDATA DLDATA SWDATA 191 Commands INPUT Example INPUT reads a stream of TSP commands from an external disk file and executes them as a unit INPUT filename or INPUT filename string Usage With INPUT you can use a file of TSP commands you have already written The input file does not have to be a complete TSP program you may use disk files as a convenient way to store frequently used pieces of your programs such as user defined procedures PROCS or tables of data to load INPUT takes one filename only as an argument If it is not enclosed in quotes the filename must conform to restrictions placed on TSP variable names it must be limited to eight characters and the filename extension must be omitted tsp will be assumed With quotes the filename can include directory information and extensions and can be up to 128 characters long In interactive mode if the filename is absent you will be prompted for it In this case you may also specify a directory as well as an extension or disk unit but the whole name must be 128 characters or less Again if the extension is omitted tsp will be assumed You will also be prompted to have the file s output
99. of attractive appearance These options are PLOT LINES none BAND none NOORIGIN BOX NOMEAN ID NOINTEGER VALUES NOHEADER For convenience the options of PLOT which are set by you are retained in the next PLOT s until they are overridden either explicitly or by including the option RESTORE in the list RESTORE causes the options to be reset at their default values Examples PLOT GNP CONS X PLOT MIN 500 MAX 1500 LINES 1000 GNP G GNPS CONS CONSS D PLOT MIN 25 MAX 25 BMEAN HEADER VALUES BAND STANDARD INTEGER RESID 330 PLOT graphics version PLOT graphics version Output Options Examples PLOT produces a plot of one or more series versus the observation number usually in units of time The series are plotted on the vertical axis and time on the horizontal axis The plot may be printed as well as displayed if a hardcopy device such as a Laserjet or dot matrix printer is available This section describes the graphics version of PLOT which is available only for TSP Givewin DOS Win TSP unix and MAC TSP For other versions see the non graphics PLOT command PLOT 4 DASH DEVICE lt name of printer FiLE lt name of file gt HEIGHT lt height of characters gt HIRES LANDSCAP or PORTRAIT MIN lt y axis minimum MAX lt y axis maximum ORIGIN PREVIEW SYMBOL TITLE text string to be used as title WIDTH list of series names Usage PLOT is followed by a single seri
100. of instruments ITERU NOITERU specifies iteration on the COVU matrix provides the same function as the old MAXITW option MAXITW the number of iterations to be performed on the parameters of the residual covariance matrix estimate If MAXITW is zero the covariance matrix of the residuals is held fixed at the initial estimate which is specified by WNAME This option can be used to obtain estimates that are invariant to which equation is dropped in a shares model like translog HETERO NOHETERO causes heteroskedastic consistent standard errors to be used See the GMM NMA command for autocorrelation consistent standard errors Same as the old ROBUST option or HCOV R WNAME z the name of a matrix to be used as the starting value of the covariance matrix of the residuals WNAME OWN specifies that the initial covariance matrix of the residuals is to be obtained from the residuals corresponding to the initial parameter values If neither form of WNAME is used the initial covariance matrix is an identity matrix Nonlinear options control the iteration methods and printing They are explained in the NONLINEAR section of this manual Some of the common options are MAXIT MAXSQZ PRINT NOPRINT and SILENT NOSILENT The legal choices for HITERz are G Gauss the default and D Davidon Fletcher Powell HCOV G is the default method for calculating standard errors R Robust and D are the only other valid options although D is not recom
101. or more series for which you would like to see a frequency distribution The default options for output yield a histogram with ten equally spaced bins or cells running from the minimum value of the series to the maximum value Output Plots of the histogram for each series are produced each in a separate window The following is stored in data storage variable type length description HIST matrix nbins series matrix with observation counts for each series HISTVAL matrix nbins series matrix with bin lower bounds for each histogram Options CDF NOCDF includes a normal QQ plot below the histogram DENSITY NODENSITY superimposes a smooth density on the histogram DISCRETE NODISCRE specifies whether the series are discrete or continuous If the series are discrete there will be one cell for each unique value limited by NBINS Cells with zero counts will be omitted 185 Commands HIST NOHIST specifies whether to include a bar type histogram in the printout upper bound on the last cell The default is the maximum value of the series lower bound on the first cell The default is the minimum value of the series NBINS the number of bins or cells The default is 10 for NODISCRETE and 20 for DISCRETE NORMAL NONORMAL superimposes a normal density on the histogram PRINT NOPRINT tells whether the histogram is to be printed or just stored STANDARD NOSTANDARD standardizes the data before plotting TITL
102. pp 294 311 White Halbert Instrumental Variables Regression with Independent Observations Econometrica 50 March 1982 pp 483 500 Zellner Arnold An Efficient Method of Estimating Seemingly Unrelated Regressions and Tests of Aggregation Bias JASA 57 1962 pp 348 368 Zellner Arnold Estimators for Seemingly Unrelated Regression Equations Some Exact Finite Sample Result JASA 58 1963 pp 977 992 250 MATRIX MATRIX Examples MATRIX processes matrix algebra expressions Operations on matrices are specified in matrix equations preceded by the word MAT these equations are just like the variable transformations performed by GENR except for two things they do not operate under control of the current SMPL and the results are stored as a matrix The MAT procedure checks the matrices for conformability of the operations and gives an error message if the operation specified is not possible Often printing the matrices in question will reveal why the operation cannot be performed MATRIX lt matrix name gt lt matrix equation gt Usage All the ordinary operators and functions used in TSP equations can also be used in the MAT command They operate on an element by element basis and hence require conforming matrices if they are binary operators There is one important exception to this the multiply operator For simplicity this operator denotes the usual matrix multiplication and element by element mult
103. pp 531 533 Maddala G S Econometrics McGraw Hill Book Company New York 1977 Chapter 11 Appendix C Nelson C R and R Startz Some Further Results on the Exact Small Sample Properties of the Instrumental Variables Estimator Econometrica 58 1990 pp 967 976 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 Chapter 9 Appendix 9 4 223 Commands Rothenberg T J Approximating the Distributions of Econometric Estimators and Test Statistics Ch 15 in Z Griliches and M Intriligator eds Handbook of Econometrics Vol Il Amsterdam North Holland pp 881 935 Staiger D and J H Stock Instrumental Variables Regression with Weak Instruments Econometrica 65 1997 pp 557 586 Stock J H and Yogo Testing for Weak Instruments in Linear IV Regression NBER Technical Working Paper No 284 October 2002 Stewart G E Algorithm 384 Collected Algorithms from ACM Volume Il ACM New York Y Theil Henri Principles of Econometrics John Wiley amp Sons New York 1971 Chapter 10 224 LIST LIST Options Examples LIST gives a single name to a list of TSP variables for use later in the program wherever that list of variables is needed It provides a convenient way to handle long lists of variables repeatedly used After you define a list any time it appears in a command the contents of the li
104. regression coefficients and their standard errors is printed along with name of the dependent variable the sum of squared residuals standard error of the regression mean and standard deviation of the dependent variable R squared and adjusted R squared Other output varies by estimator If the data are unbalanced the Ahrens Pincus measure of the degree of unbalancedness is also printed this measure is one for balanced data values less than one provide an indication of how far the data is from balanced See the method section for the definition of this statistic and the reference for details on its interpretation MEAN prints a table of means for each individual MEAN obs vars is stored and excludes any constant term BYID prints an F test vs TOTAL labelled F stat for A B Ai Bi and an F test vs WITHIN labelled F stat for Ai B Ai Bi in the output of the respective estimators Only COEFI the individual coefficient estimates LOGLI and SSRI the individual sum or squared residuals are stored Use the PRINT option to print COEFI WITHIN prints an F test vs TOTAL labelled F stat for A B Ai B and stores FIXED effects vector VARCOMP prints the actual variance components the method used to compute them and the implied differencing factor THETA A Hausman specification test comparing VARCOMP null hypothesis and WITHIN is computed 313 Commands PANEL stores the standard regression results in data stora
105. reverse dynamic future observations are evaluated first The STATIC option can be used to prevent such dynamic evaluation Expressions like X 1 and X l are treated as lags leads if X is a series otherwise they are treated as subscripts like COEF 1 Expressions like X 76 2 a date and M 1 2 a matrix element are subscripts evaluate to scalars Output GENR produces no printed output except a note when a dynamic GENR is being performed It stores one new or replacement series in data storage Options SILENT NOSILENT suppresses the dynamic GENR message It does not suppress error or warning messages STATIC NOSTATIC prevents dynamic evaluation of equations with lagged or led dependent variables Examples GNPL1 GNP 1 GENR DP LOG PRICE PRICE 1 GENR WAVE GAMMA SIN TREND DUM X gt 0 This makes a dummy variable equal to one when X is greater than zero the logical expression has a value of one when true and equal to zero otherwise SCLEVEL 1 SC lt 6 2 SC gt 6 amp SC lt 9 3 SC 9 amp SC lt 12 4 SC gt 12 This makes SCLEVEL equal to 1 2 3 or 4 depending on which range the value of SC falls into this is essentially a recode operation GENR CONSEQ CONSFIT 168 GENR GENR IDENT12 The last two examples compute the series defined by equations the first computes a fitted consumption from a previously defined FRML or IDENT or estimated consum
106. s squared is the estimate of the residual variance X X is the moment matrix for the data in the regression and X 0 is the matrix of data for the prediction interval This PROC is invoked by the statements LIST VARLIST VAH1 VAR2 VAR3 Example with 3 variables VARFORC ERROR VARLIST The sample over which you wish the forecast and error must be specified before invoking VARFORC VARFORC expects that the estimation of the model for which you are doing the forecast has just been executed and that S2 and VCOV contain the estimated variances of the residuals and coefficients The series ERROR contains the prediction error for each observation on return This example also shows how to pass a list of variable names as an argument PROC VARFORC PREDERR XLIST MMAKE X XLIST PREDERR SER SQRT S2 VECH DIAG X GVCOV X ENDPROC Here is a user procedure to compute Theil s inequality coefficient U as a normalized measure of forecast error PROC THEILU ACT PRED U GENR RESID ACT PRED MAT SQRT RESID RESID ACT ACT ENDPROC 352 QUIT Interactive QUIT Interactive QUIT exits from interactive TSP without saving the backup file Otherwise it is equivalent to the END EXIT or STOP command gt QUIT 353 Commands RANDOM Options Examples References RANDOM creates pseudo random variables It can create random variables which follow the normal uniform Poisson Negative Binomial Laplace
107. scalar 1 Estimated concentration parameter mu squared CDF scalar 1 Cragg Donald F statistic for excluded instruments in RF COEF vector vars Coefficient estimates SES vector vars Standard errors T vector vars T statistics VCOV matrix vars vars Variance covariance of estimated coefficients RES series obs Residuals actual fitted values of the dependent variable FIT series obs Fitted values of the dependent variable AI series obs estimated fixed effects stored as a series COEFAI vector individuals estimated fixed effects SESAI vector individuals standard errors on fixed effects TAI vector individuals t statistics on fixed effects vector individuals p values associated with TAI If the regression includes PDL variables the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling Method The LIML eigenvalue is the minimum eigenvalue of the following matrix H H 221 Commands H is the residual covariance matrix of the endogenous and dependent variables regressed on all of the instruments H1 is the residual covariance matrix of the same variables regressed on just the included instruments This eigenvalue is calculated by CACM algorithm 384 The FULLER constant term if non zero is subtracted from this eigenvalue to yield K which is then used
108. size whichever is closer to the observed test statistic Such interpolated P values are fine for testing at the 01 05 or 10 sizes but would be highly speculative outside this range WTDCHI If w i are the eigenvalues supplied by the option EIGVAL c i are chi squared 1 variables and d is the test statistic supplied as an argument WTDCHI computes the following probability N 2 w i c i N Prob d Pr ob Y w i a i lt 0 wi i This is useful for computing P values for the Durbin Watson statistic other ratios of quadratic forms in normal variables and certain non nested tests for example Vuong 1989 suggests likelihood ratio tests for nonnested hypotheses The Pan method is used when the number of eigenvalues is less than 90 otherwise the Imhoff method is used If the absolute values of the smallest eigenvalues are less than 1D 12 they are not used otherwise duplicate eigenvalues are not checked for The inverse of this distribution is not implemented Options 75 Commands BIVNORM CHISQ DICKEYF F NORMAL T WTDCHI specifies the bivariate normal chi squared Dickey Fuller F standard normal student s t and weighted chi squared distributions respectively CONSTANT NOCONST specifies whether a constant term C was included in the regression for Dickey Fuller NOCONST is only valid for NVAR 1 DFz the degrees of freedom for the chi squared or student s t distribution or the numbe
109. speed and to insure a full period no repeats in 2 319 or 2 31 draws The multiplicative congruential method produces random numbers which are uniform on the 0 1 interval Normal and Poisson random variables are created from uniform random variables with ACM Algorithms 488 and 369 respectively Gamma random variables use ACM Algorithm 599 Negative binomial random variables are computed by drawing Gamma random variables to determine Y and then drawing Poisson random variables with mean Y All other random variables are derived from the uniform random variables using the inverse distribution function which usually involves an asymptotic expansion see the CDF references Options CAUCHY NOCAUCHY specifies that the random number generated is to follow the Cauchy distribution DF the degrees of freedom for student s t distribution see the t option 355 Commands DRAW the name of a series or matrix which will be sampled with probability one divided by the number of observations to generate the random numbers This series does not have to be sorted in any order If DRAW a matrix a multivariate set of random numbers is drawn The number of random variables is equal to the number of columns in the matrix Note that the matrix from which you are drawing does not have to have the number of rows equal to the number of observations This is very useful for simulation or bootstrap standard errors EDF Empiric
110. starting values for BJEST can be specified directly in the command not with a PARAM statement since the parameters have certain fixed predetermined names in the Box Jenkins notation There are several ways to specify starting values in BJEST The easiest way is to let TSP guess reasonable starting values for the parameters TSP will choose starting values for the non seasonal parameters only based on the autocorrelations of the time series You can suppress this feature with NOSTART If no other information is supplied see below TSP will use zero as the starting value for all the parameters In practice this generally leads to slower convergence of the parameter estimates You can also supply starting values in an START vector You may also specify the starting values for any or all of the parameters yourself For a model with more than one or two parameters it may be difficult to choose starting values since the individual parameters often do not possess a simple interpretation However you may wish to specify the starting values of at least some of the parameters based perhaps on previous estimates In this case any parameters for which no starting value is supplied will be given the default value TSP s guess if the START option is on or zero for NOSTART or the START value User supplied starting values are given after the name of the series on the command Follow the series name with the keyword START and then a series of pai
111. substituted into the FRMLs a FIML estimation of the resulting FRMLs will yield the same results as FIML on the original FRMLs and IDENTs IDENTs cannot be used to impose nonlinear or linear constraints on the parameters any constraints must be substituted directly into the structural equations An attempt to use IDENTs in this way would fail because of an unequal number of endogenous variables and equations and a singular Jacobian the gradient of the equations with respect to endogenous variables The parameters to be estimated must be defined previously and starting values assigned by a PARAM statement or statements or by being estimated in a previous nonlinear estimation 144 FIML If the equations are nonlinear in the endogenous variables the Jacobian will not be constant over time and it will be evaluated at each observation If the Jacobian is singular at the initial iteration estimation halts Usually this is caused by coefficients with zero starting values use 0 1 for these initial values instead If the Jacobian is singular during the iterations the parameter stepsize is automatically squeezed this is treated as a numerical error Output Normal FIML output begins with a listing of the options if the PRINT option is specified and the equations The model is checked for linearity in the parameters which simplifies the computations and for linearity in the variables which greatly simplifies the deri
112. tests if the excluded exogenous variables called Z2 have zero coefficients in the reduced form the concentration parameter is equal to CDF times the number of excluded exogenous variables These statistics are also valid for multiple RHS endogenous variables They can be used to assess whether the model has a weak instruments or many instruments problem For a single RHS endogenous variable the bias of 2SLS is proportional to rho CDF Nelson and Startz 1990 where rho is the correlation between the reduced form and structural equations so low values of the CDF imply a high bias For a single RHS endogenous variable values of CDF lower than 10 are considered to be problematic Staiger and Stock 1997 refined in Stock and Yogo 2004 Unrealistically low computed standard errors for 2SLS also occur in such a situation and the use of LIML FULLER 1 or plain LIML with the Bekker SEs instead of 2SLS is generally advised because these estimators have much lower bias and properly sized standard errors especially when the number of excluded instruments is large LIML FULLER 1 and LIML can still suffer from a weak instruments problem For a single RHS endogenous variable values of mu squared lower than 10 15 suggest a problem bias and or standard errors that are too small see Hansen Hausman and Newey 2004 For larger values of mu squared LIML FULLER 1 and LIML have low bias and properly sized standard errors LIML FULLER 1 has a sligh
113. the transformed residuals This is equal to the number of observations times the identity matrix if estimation is by ML VCOV matrix par par Estimated variance covariance of estimated parameters RES matrix obs eqs Matrix of residuals actual fitted values of the dependent variable FIT matrix obs eqs Matrix of fitted values of the dependent variables Options covariance matrix of the orthogonality conditions The default is to compute starting values with 3515 and form the covariance matrix from these COVOC OWN computes residuals from the current starting values and forms the covariance matrix from these COVU covariance matrix of residuals This is used for the initial 3515 estimates if the default LSQSTART option is in effect The default is the identity matrix This option is the same as the old WNAME option LSQ FEI NOFEI specifies whether a model with individual fixed effects is to be estimated FREQ PANEL must be in effect for the FEI option HETERO NOHETERO specifies conditional heteroskedasticity of the residuals and causes the COVOC matrix to include interaction terms of the residuals and the derivatives with respect to the parameters Specify this option to obtain the usual Hansen or Chamberlain estimator When HETERO is on GMM checks that the number of OC s is less that the number of observations so that COVOC will be positive definite if not an error message is printed INST l
114. the DIRECT method See the Bilias et al 2000 reference for details SILENT NOSILENT suppresses all printed output The results are stored TERSE NOTERSE suppresses all printed output except the table of coefficient estimates and the value of the likelihood function UPPER the value above which the dependent variable is not observed The default is no limit This option cannot be used at the same time as the LOWER option References Barrodale and F D K Roberts Algorithm 478 Collected Algorithms from ACM Volume Association for Computing Machinery New York NY 1980 Bilias Y S Chen and Z Ying Simple Resampling Methods for Censored Regression Quantiles Journal of Econometrics 99 2000 pp 373 386 Davidson Russell and James G MacKinnon Estimation and Inference in Econometrics Oxford University Press New York NY 1993 Chapter 16 Fitzenberger Bernd A Guide to Censored Quantile Regressions in G S Maddala and C R Rao eds Handbook of Statistics Volume 15 Robust Inference 1997 pp 405 437 Judge George R Carter Hill William E Griffiths Helmut Lutkepohl and Tsoung Chao Lee Introduction to the Theory and Practice of Econometrics John Wiley amp Sons New York Second edition 1988 Chapter 22 Koenker W W Bassett Regression Quantiles Econometrica 46 1978 pp 33 50 Machado J A F and J M C Santos Silva Glejser s Test Revisited Journ
115. the calculation and output of the regression diagnostics for OLSQ and some output of other commands It replaces the old SUPRES and NOSUPRES commands REGOPT BPLIST lt list of variables gt CALC CHOWDATE lt date for splitting sample DWPVALUE type LMLAGS lt of lags for LMAR test gt PRINT PVCALC PVPRINT QLAGS lt of Q statistics gt RESETORD value SHORTLAB STAR1 lt value for gt STAR2 lt value for gt STARS lt list of output names or keywords gt Usage OLSQ can produce a massive number of diagnostics REGOPT provides the user with extensive customization of this output so that irrelevant diagnostics do not crowd relevant ones or require extensive time to calculate The PV CALC and PV PRINT options are used along with a list of the diagnostic codes names that one wishes to control The keywords AUTO HET REGOUT and ALL may also be used to control groups of diagnostics instead of listing all the names Other options such as BPLIST and LMLAGS control individual diagnostics that have no clear default OPTIONS LIMCOL and SIGNIF also control the display Note that robust diagnostics are available with the HI option in OLSQ Output The following three examples of controlling regression output with REGOPT illustrate the range of output available The data for these examples is a regression squared on time options crt smpl 1 10 trend t 12 tt Example 1 default option olsq i2 c t defaul
116. the names of dummy arguments In programming terminology they are local variables When you enter a procedure the last SMPL processed is in force When you leave be careful to restore the SMPL if you have changed it during the procedure or you may get unpredictable results if you call the PROC from different parts of the program If the name of a LIST is used as an argument to a PROC the list is not expanded until it is used inside the PROC This allows the number of items in the list to vary between PROC uses The number of items in the list for a particular call can be determined by the LENGTH command Output PROC produces no output unless your procedure generates output Examples This simple example creates a dummy variable over a sample specified by START and STOP which is one every SKIPth observation and zero elsewhere This is an inefficient way to create a seasonal dummy or year dummies for panel data it is better to use a repeating TREND and the DUMMY command Note how the SMPL and FREQ are saved and restored on exit from the PROC 351 Commands PROC SKIPDUM START STOP SKIP VAR COPY SMPL SMPSAV COPY FREQ FRQSAV SMPL START STOP GENR VAR 0 DO I START TO STOP BY SKIP SET VAR I 1 0 ENDD FREQ FRQSAV SMPL SMPSAV ENDPROC The next example is a procedure to compute the prediction error for the classical linear regression model The formula for the error variance is s fIX 0C0X Xx where
117. the new series if the observations are mapped If no observations of the old series are mapped to a given element of the new series the element is given the value zero for both sum and average The length of the new series is equal to the maximum value in the map series SMPL NOSMPL applies only when the MAP option is used Otherwise the default is NOSMPL use all the data in the series Examples FREQ A CONVERT AVERAGE UNEMP CONVERT SUM SALES CONVERT LAST PCLOSE PRICE 100 CONVERT Assume that UNEMP and SALES are quarterly variables and PRICE is a monthly variable The statements above convert the unemployment rate by averaging the quarterly rates over the year but convert sales from quarterly to annual by adding them since they are a flow variable The end of year price PCLOSE is obtained by using the December observation of the monthly price variable FREQ SMPL 70 72 READ X 102040 FREQQ CONVERT DX X CONVERT SUM SX X CONVERT INTER LAST IX X results in SX IX 2 5 2 5 2 5 5 0 2 5 7 5 2 5 10 10 25 10 30 10 35 10 40 Example of the MAP option SMPL 15 TREND READ MAPS 01223 CONVERT MAP MAPS AVE CONVERT MAP MAPS SUM ST T which yields the following AT ST 2 2 3 5 7 5 5 101 Commands COPY Example Makes a copy of an old TSP variable series matrix constant etc with a new name COPY lt old TSP variable gt lt
118. the table of output results below regs MAXLAG MINLAG 2 if MINLAG lt MAXLAG regs 1 if MINLAG MAXLAG stats 3 2 trend_vars regs if PRINT is on 4 1 for PP 2 vars 3 3 for Johansen tests types number of types of unit root tests performed 2 for default 3 for ALL etc eg number of different cointegrating regressions for Engle Granger type tests vars for ALLORD or 1 by default Here are the results generally available after a COINT command Name Type Length Variable Description TABWS matrix stats regs table for augmented WS tau tests on a single variable TABDF matrix stats regs augmented Dickey Fuller tau tests TABPP matrix stats regs Phillips Perron Z tests UNIT matrix types vars summary table of unit root test statistics for optimal lags UNIT matrix types vars P values for optimal lags UNITLAG matrix types vars Optimal lag lengths chosen by RULE TABEG matrix types vars table for augmented Engle Granger tests CIVEG matrix types vars cointegrating vector normalized vector eg P values for optimal lags EGLAG vector eg Optimal lag lengths chosen by RULE TABJOH matrix stats regs table for Johansen tests 85 Commands CIVJOH matrix vars vars regs cointegrating vectors eigenvectors Method Unit root tests are based on the following regression equation y thnttrt V OV U H t Httyt nlt 1 n
119. the true residual variances are unity it can be supplied by you from a previous estimation or it can be computed from the parameters the WNAME OWN option The S matrix is always a symmetric matrix of the order of the number of equations To obtain conventional seemingly unrelated regression estimates of a nonlinear multivariate regression model use the SUR command which is a special form of the LSQ command This version of the procedure obtains single equation estimates of the parameters of the model uses these to form a consistent estimate of the residual covariance matrix and then minimizes the objective function shown above with respect to the parameters b If the model is linear this is a two stage procedure only two iterations The plain LSQ command will iterate simultaneously on the parameters and the residual covariance matrix In this case linear models may take more than one iteration to converge 243 Commands Nonlinear three stage least squares this estimator uses the distance function as shown with S equal to a consistent estimate of the residual covariance either supplied or computed and H equal to the Kronecker product of an identity matrix of the order of the number of equations and the matrix of instruments This means that all the instruments are used for all the equations Three stage least squares estimates can be obtained in two ways LSQ with the WNAME option the INST option and more than one equation n
120. to follow the negative binomial N p distribution which is excess waiting time to obtain N successes the number of trials minus N with success probability p for each trial N does not have to be an integer For this model the mean and variance of the data are given by N 1 p 1 E y E V a p y An alternative widely used specification is the following 1 5 1 5 V 5 y Fs Ely The two specifications are equivalent and imply the following identities p or G 620 PU T Use the MEAN and STDEV options to specify the parameters The MEAN must be non negative and STDEV must be larger than or equal to the square root of the mean POISSON NOPOISSON specifies that the random number generated is to follow the Poisson distribution This distribution has one parameter alpha which is both the expected value and the variance Supply this parameter using the option It must be a non negative number REPLACE NOREPLACE specifies that DRAWing from the empirical distribution function is to be done with replacement the default or no replacement 357 Commands SEEDIN value of random seed to start random generator replaces the current value of the random seed This must be an integer in the range 1 2 1 billion otherwise it is moved into this range Note that scalar values are stored in double precision which a
121. tspintl com support tsp ug_ online htm The User s Guide contains more discussion and examples of how to combine TSP statements and construct TSP programs The Help system can be useful in the following ways To read a basic introduction of what TSP and why it is useful see Introduction to TSP To find out which command to use see the functional index Commands by Function You know which command or procedure you want to use but are unsure of the options available You can look up the details of the command in the Index click on the Help Topics button above and then click the Index tab and check the Options To learn more about the methods used in any particular procedure see the details in References under the command To see how to use a command look at the Examples To check which results are stored after a procedure and how they are named consult the Output section of the command entry Basics gives complete definitions of TSP syntax the interpretation of special characters and the mathematical and statistical functions available If you want to find out something related to the use of TSP through the Looking Glass look it up in the TSP through the Looking Glass section Examples of TSP Programs describes where to find sample TSP programs You may also want to look at the TSP 5 0 read me Introduction Introduction to TSP TSP provides regression forecasting and other econometric tools on mainframe and pe
122. unless it is a scalar in which case it is duplicated throughout the new matrix If you specify a type such as triangular which is not consistent with the input matrix MFORM will change the input so that it is i e the elements below the diagonal will be zeroed You can use this feature for example to select the diagonal elements of a matrix and form a new matrix from them like a diag function UNMAKE can be used to form a series from the diagonal of a diagonal matrix Output 254 MFORM MFORM produces no printed output A single matrix is stored in data storage Options BAND NOBAND specifies that a symmetric band matrix is to be formed from a vector of band values and optionally a symmetric corner matrix for the upper left and lower right corners to override the band values BLOCK NOBLOCK specifies that a block diagonal matrix is to be formed from a list of symmetric or general matrics by placing them along the diagonal and putting zeros elsewhere This is useful for composing VCOV matrices from several independent estimations for minimum distance estimation with LSQ or for ANALYZ NROW the number of rows in the matrix to be formed This is required for a general matrix NCOL the number of columns in the matrix to be formed This is required for a general matrix Either NROW or NCOL must be specified for symmetric triangular or diagonal matrices unless the input variable is a matrix and you wish the n
123. version only where drive letter is A B C D etc Usage MicroTSP format databank files may be useful for transferring data between microTSP and TSP They are plain editable non binary files containing comments frequency starting and ending dates and data values one per line See the microTSP documentation for details They are not efficient in terms of disk space usage or the time required to read or write them However they are easy to edit for manual data revision TSP does not support the use of complex path names such as C foo ser on the PC if the data are in the same directory as your program or in the working directory for interactive use this should not be a problem To move regular TSP databanks between machines use the DBCOPY command However spreadsheet files are usually the easiest way to exchange small or medium sized datasets between different programs When STORE writes to an existing file it preserves any existing comments and always writes the current date Only time series may be stored matrices parameters etc can only be stored on regular TSP databanks The FETCH command reads files created by STORE Example STORE X Y writes the series X and Y to the files X DB and Y DB Reference Hall Robert E and Lilien David microTSP Version 6 5 User s Manual Quantitative Micro Software 1989 http www eviews com 420 SUPRES SUPRES Examples SUPRES suppresses the printing of output
124. years per individual balanced data and print individual means PANEL T 7 MEAN BYID LRNDL5 C PATENTS LRNDL4 Print VARCOMP output only using ID or FREQ PANEL to distinguish individuals PANEL NOTOT NOBET NOWITH LRNDL5 C PATENTS LRNDL4 319 Commands Estimate all models except BYID use large sample formulas for VARCOMP PANEL NOVSMALL LRNDL5 C PATENTS LRNDL4 Print individual means only PANEL MEAN NOREG LRNDL5 C PATENTS LRNDL4 References Ahn S C and P Schmidt Efficient Estimation of Panel Data Models with Exogenous and Lagged Dependent Regressors Journal of Econometrics 68 1995 5 27 Ahrens H and R Pincus On two measures of unbalancedness in a one way model and their relation to efficiency Biometric Journal 23 1981 pp 227 235 Baltagi Badi Econometric Analysis of Panel Data Wiley amp Sons New York 1995 first edition Bhargava A L Franzini and W Narendanathan Serial Correlation and the Fixed Effects Model Review of Economic Studies XLIX 1982 pp 533 549 Chamberlain Gary Multivariate Regression Models for Panel Data Journal of Econometrics 18 1982 pp 5 46 Chamberlain Gary Panel Data in Griliches and Intriligator eds Handbook of Econometrics Volume Il North Holland Publishing Co Amsterdam 1985 Davis Peter Estimating Multi Way Error Components Models with Unbalanced Data Structures Journal of Econometrics 106 July 2002 pp 67 9
125. 0 GMM INST C NMA LAG SILENT EQPP SET w2 COVOC SET z n alpha 1 n 2 w2 s2 2 ssr PRINT LAG z w2 ENDDO The regressions for the Engle Granger tests are just an extension of the Dickey Fuller test after an initial cointegrating regression 87 Commands TREND T OLSQ LRGNP LEMPLOY cointegrating regression E RES SMPL 10 70 DE E E 1 Estimation sample same for all lags MAXLAG 1 obs are dropped SMPL 20 70 DO LAG 1 10 SET MLAG LAG OLSQ E E 1 DE 1 DE MLAG SET alpha COEF 1 SET tauDF alpha 1 QSES 1 CDF DICKEYF NVAR 2 tauDF ENDDO The following equation defines the 2 7 order VAR Vector Auto Regression that is used in the Johansen trace test Y Y JH Y IL Y 15 4 CT BU where Y t is 1by Gand CT t are seasonal constants and trends II l 14 1L isaGbyG impact matrix of rank r 7 amp where and are by r matrices The Log Likelihood and AIC printed in the table are from the unrestricted version of this VAR The restricted version is estimated with a 2G equation VAR dY dY A dY A CT A E Y ua dY B dY _ B dy B CT B F 020 EF _ FF T L 1 T L 1 T L 1 Q S CHOL S that is QQ S 8 0 5 5 5 0 A eigenvalues of 5 sorted high to low 7 Q eigenvectors of S trace T L 1 L41 6 log 1 4 5 88 COINT where T number of observations
126. 0 or on iterating to obtain the exact P value calculation when the number of observations is less than 60 All these results are based on recursive residuals and they will not be calculated if the first K observations are not of full rank where K number of right hand side variables in the regression Recursive residuals can also be computed by going backwards through the sample and although this is not done at present in TSP it may be useful if the plots are being used to locate points of structural change Harvey 1990 contains some examples of interpreting the CUSUM and CUSUMSOQ plots Options 334 PLOTS PREVIEW NOPREVIEW turns graphics plots off or on for the following subsequent commands OLSQ INST 2SLS actual and fitted values residuals OLSQ CUSUM CUSUMSQ The default is PREVIEW in TSP Givewin and NOPREVIEW in other versions Examples Residual plots are shown only for the second AR regression NOPLOT OLSQ CONS GNP PLOTS CONS C GNP turn on the CUSUM plots for each regression until a NOPLOTS is given PLOTS CUSUM CUSUMSQ turn on all plots both CUSUMS plus residuals and fitted values PLOTS ALL turn on the regression diagnostics REGOPT PVPRINT CSMAX CSQMAX also include the diagnostics and P values REGOPT PVPRINT AUTO include both the diagnostics and all plots REGOPT PVPRINT ALL In all versions except TSP Givewin the CUSUM plots are done in l
127. 1 276 Imhoff P J Computing the Distribution of Quadratic Forms in Normal Variables Biometrika 48 1961 pp 419 426 Inverse Normal Computation Algorithm AS 241 Applied Statistics 37 1988 Royal Statistical Society Judge George G Hill R Carter Griffiths William E Lutkepohl Helmut and Lee Tsoung Chao Introduction to the Theory and Practice of Econometrics second edition Wiley New York 1988 pp 394 400 Mackinnon James G Critical Values for Cointegration Tests in Long Run Economic Relationships Readings in Cointegration eds R F Engle and C W J Granger New York Oxford University Press 1991 pp 266 276 Mackinnon James G Approximate Asymptotic Distribution Functions for Unit Root and Cointegration Tests Journal of Business and Economic Statistics April 1994 pp 167 176 Pan Jie Jian Distribution of Noncircular Correlation Coefficients Selected Transactions in Mathematical Statistics and Probability 1968 pp 281 291 StatLib http lib stat cmu edu apstat Vuong Quang H Likelihood Ratio Tests for Model Selection and Non Nested Hypotheses Econometrica 57 1989 pp 307 334 80 CLEAR interactive CLEAR interactive CLEAR terminates an interactive session and restarts TSP CLEAR Usage CLEAR is an alternative to QUIT Use CLEAR when you want to immediately restart TSP without saving any of your data or commands from the current session Use QUIT when you want t
128. 10FOOD etc Note the use of a single dot for the variable This example takes a list of variables called VARS and regresses each of them on the others in the list one by one Note the use of the INDEX and CHAR options LIST VARSXYZW DOT INDEX I CHAR VARS DOT INDEX J VARS IFI 5z J THEN OLSQ 6 ENDDOT ENDDOT 124 DROP Interactive DROP Interactive Examples DROP is used to drop a list of variables from the previous statement and re execute it It is the opposite of ADD DROP lt list of variables gt Usage DROP is a convenient way to drop variables from a regression and perform a second estimation without retyping the command It is not restricted to this usage and may be used anywhere this type of command modification is needed The command DROP var1 var2 and the sequence RETRY gt gt DELETE var1 gt gt DELETE var2 gt gt EXIT are identical in function since both permanently modify the previous command by deleting the first occurrences of vari and var2 The command is then automatically executed in both cases The only potential difference between these approaches besides the amount of typing is in the definition of previous RETRY with no line number argument assumes you want to modify the last line typed DROP will not accept a line number argument and always modifies the last line that is not itself a DROP or ADD command Because of the way previous is defined f
129. 285 Commands NOPLOT Examples NOPLOT turns off the PLOTS options so that actual and fitted values are not plotted after each regression NOPLOT is the default NOPLOT Usage NOPLOT takes no arguments it is needed only when a PLOTS statement has appeared earlier PLOTS NOPLOT are also available on the OPTIONS statement The PLOTS NOPLOT options apply to any procedures which produce residual plots these are the OLSQ INST AR1 LSQ and ACTFIT Examples This example suppresses the printing and plotting of residuals after a regression on large amounts of data and then uses ACTFIT with a different sample to plot a portion of them SMPL 1 896 NOPLOT OLSQ LOGP C LOGR 5 NPLNT72 DPATO LOGRY 1 SMPL 1 16 881 896 PLOTS ACTFIT ACT FIT SMPL 1 896 NOPLOT 286 NOPRINT NOPRINT Examples NOPRINT turns off the printing of input data in the load section It applies only to free format data input since data read under fixed format is not printed in any case NOPRINT Usage Include the NOPRINT statement at any point in your data section where you wish to turn off the printing of the input It remains in force until the end of the data section Do not put the statement between a LOAD statement and the data which goes with it these must be contiguous The PRINT NOPRINT option is also available on the LOAD statement itself Examples The following example of a LOAD section shows t
130. 4427191 12 2983739 000 Variance Covariance of estimated coefficients T 30 80000000 F 4 40000000 0 80000000 Correlation matrix of estimated coefficients T 1 0000000 T 0 88640526 1 0000000 ID ACTUAL FITTED RESIDUAL 0 0 1 1 0000 11 0000 2 12 0000 P 0 2 4 0000 0 0000 4 0000 0 3 9 0000 11 0000 2 0000 01 4 16 0000 22 0000 6 0000 0 5 25 0000 33 0000 ME 8 0000 0 6 36 0000 44 0000 ee 8 0000 0 pi 49 0000 55 0000 ey 6 0000 0 8 64 0000 66 0000 2 0000 01 9 81 0000 77 0000 TE 4 0000 OF 10 100 0000 88 0000 12 0000 tut 0 CUSUM PLOT ee x CUSUM PLOTTED WITH C UPPER BOUND 5 PLOTTED WITH U LOWER BOUND 5 PLOTTED WITH L MINIMUM MAXIMUM 8 04319191 10 72242260 0 3 L Ic 9 4 L Ic U 5 L 9 6 L 9 7 L 9 8 L 9 L Uc 10 L U c4 0 8 04319191 10 72242260 MINIMUM MAXIMUM CUSUMSQ PLOT TORR KK CUSUMSQ PLOTTED WITH C MEAN PLOTTED WITH M UPPER BOUND 5 PLOTTED WITH U LOWER BOUND 5 PLOTTED WITH L MINIMUM MAXIMUM 0 00000000 1 00000000 3 2 M U CL 4 12 M U CL 5 LC M U 6 LC M U 7 2 M U CL 8 Lc M 17 41 9 L M U 10 L 3 CMU 375 Commands list of scalar re
131. 46 LHV RNMS LOGL SBIC AIC SSR 95 YMEAN SDEV DW RSQ ARSQ FBEX COEF SES VCOV COVU IMPRES FEVD RES FIT Method list list scalar scalar scalar vector vector vector vector vector vector vector vector vector vector matrix matrix matrix matrix matrix matrix eqs eqs vars eqs eqs eqs eqs eqs eqs eqs eqs eqs vars eqs vars eqs vars 2 eqs eqs nhoriz eqs 2 nhoriz 1 eqs eqs obs eqs obs eqs VAR Name of the dependent variable list of names of right hand side variables Log of likelihood function Schwarz Bayesian Information Criterion Akaike Information Criterion Sum of squared residuals Standard error of the regression Mean of the dependent variable Standard deviation of the dependent variables Durbin Watson statistic R squared Adjusted R squared F statistics for block exogeneity Coefficient estimates Standard Errors Variance covariance of estimated coefficients Residual covariance matrix E E T K Impulse Response function Forecast error variance decomposition Residuals Fitted values of the dependent variable OLS is performed equation by equation This is efficient because the right hand side variables are identical for every equation The impulse response function is just a dynamic simulation with shocks in the first period for some variables and zeros for the other variables
132. 5 Farebrother R W Algorithm AS 256 Applied Statistics 39 1990 Pascal code posted on StatLib Hsiao Cheng Analysis of Panel Data Cambridge University Press Cambridge England 1986 Leamer Edward E Specification Searches Ad Hoc Inference with Nonexperimental Data Wiley New York 1978 p 114 Maddala G S Econometrics McGraw Hill New York 1977 pp 326 329 Maddala G S and M Nerlove Econometrica 1971 320 PANEL Nerlove Marc Likelihood Inference in Econometrics Academic Press New York 2000 StatLib http lib stat cmu edu apstat 321 Commands PARAM Example PARAM is used to define parameters for the nonlinear estimation procedures and to assign starting values to them To supply parameter starting values to PROBIT TOBIT SAMPSEL and LOGIT use the START vector see those procedures for further information PARAM lt parameter name gt lt value gt lt parameter name gt lt value gt Usage PARAM may be followed by as many argument pairs as desired limited only by TSP s argument limit Each pair is the name of the parameter followed by the value it is to be given The parameter names may be that of new or previously defined variables The value may be omitted in which case the variable is given the value zero if it is new or left unchanged if it has already been defined Note the keywords C and CONSTANT may not be used as parameters Procedures which estim
133. 6 show that HITER N results in fast convergence and HCOV W yields standard errors that are robust to non normal disturbances If no options are supplied a GARCH 1 1 model will be estimated Examples ARCH NAR 4 X ARCH X GARCH 1 1 the default ARCH NMA 1 NAR 3 MEAN Y C X GARCH M ARCH GT G1 G2 G3 MEAN Y C X OLS M GARCH 0 1 Here we try some alternative starting values to override the solution with 1 1 that can occur in these models The order of parameters in COEF and START is described below under Output the first parameters are the gammas and the last are 0 betat and 0 52 ARCH ARCH NMA 1 DF C DF2 MHOL COPY COEF START SET START 4 02 ALPHAO SET START 5 5 BETA1 ARCH NMA 1 DF C DF2 MHOL Other types of univariate and multivariate GARCH models can be estimated with ML and ML PROC Please see the 75 User s Guide and our web page http www tspintl com for examples References Bollerslev Tim Generalized Autoregressive Conditional Heteroscedasticity Journal of Econometrics 31 1986 pp 307 327 Engle Robert F Autoregressive Conditional Heteroscedasticity with Estimates of the Variance of U K Inflation Econometrica 50 1982 pp 987 1007 Engle Robert F David M Lilien and Russell P Robins Estimating Time Varying Risk Premia in the Term Structure The ARCH M Model Econometrica 55 1987 pp 3
134. 7 49 1 206 9 0 603 READ FILE SML WKS the series CJMTL and PMTL are defined FREQ Q SMPL 48 1 49 1 are set if there are nocurrent FREQ or SMPL READ FILE SML WKS TYPE GEN creates the 5x2 matrix LOTMAT with the values 2 of CJMTL PMTL in its columns READ FILE SML WKS PMTL only reads in PMTL CJMTL is ignored Here is the nm3 wk1 file as shown in Lotus using numeric dates for the following examples 04 30 57 23 22 34 55 10 9 05 31 57 23 6 351 11 0 06 30 57 23 9 35 8 11 2 07 31 57 24 0 11 5 2 Define the monthly series SF LA and SD from 57 4 to 57 7 If there is no current sample this is the new sample with FREQ The series LA will be given a missing value in its last observation READ FILE NM3 WK1 SF LA SD READ FILE NM3 WK1 SF An error message is printed because 3 series names are required 371 Commands RECOVER Interactive RECOVER s the command stream from a previously aborted TSP session RECOVER lt filename gt Usage With RECOVER the command stream from a previous session may be reinstated in the event it was terminated abnormally In most cases TSP will automatically recover an aborted session and this procedure is not necessary However if you have changed directories or renamed INDX TMP you will have to use this procedure to recover the session Note that the recovery process restores the commands entered but does not execute them
135. 8 THETA 1 0 2114 THETA 2 0 2612 DELTA 1 0 8471 The second example produces the same forecasts but the model parameters are provided by the immediately preceding BJEST procedure Notice that the options describing the model do not need to be repeated for the BJFRCST procedure Also the evel of the original series rather than its log are forecasted 66 BJFRCST BJEST NMA 2 NSMA 1 NDIFF 1 NSDIFF 1 NSPAN 12 LOGAUTO START THETA 1 0 12 THETA 2 0 20 DELTA 1 0 82 BJFRCST EXP PRINT NHORIZ 24 ORGBEG 264 LOGAUTO References Box George P and Gwilym M Jenkins Time Series Analysis Forecasting and Control Holden Day New York 1976 Nelson Charles Applied Time Series Analysis for Managerial Forecasting Holden Day New York 1973 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Co New York 1976 Chapter 15 67 Commands BJIDENT Output Options X Example References BJIDENT prints and plots descriptive statistics which are useful in identifying the process which generated a time series The process of time series identification is described in the two references TSP follows the notation of Box and Jenkins who developed this technique for analyzing time series BJEST is used for estimating the model you develop and BJFRCST for forecasting with it The decisions to be made in the process of identifying a time series process are 1 whether there is
136. 8036 SBPHET constant 0 18599239 SWILK constant 0 86938361 SSWILK constant 0 098324680 376 REGOPT Options BPLIST list of variables for the Breusch Pagan heteroscedasticity test CALC NOCALC indicates whether the listed diagnostics list of output names should or should not be calculated and stored under names CHOWDATE starting date of second period for Chow test The default is to split the sample exactly in half if the number of observations is odd the extra observation will be in the second period DWPVALUE APPROX or BOUNDS or EXACT specifies what method will be used for computing the P value for the Durbin Watson statistic The default depends on the current FREQ APPROX for FREQ N BOUNDS for other frequencies including Panel data LMLAGS maximum number of lagged residuals for Breusch Godfrey LM test of general autocorrelation AR or MA The default is zero PRINT NOPRINT indicates whether the diagnostics should be printed PRINT implies CALC PVCALC NOPVCALC indicates whether p values should be calculated and stored under names PVCALC implies CALC See Method for the distributions used to compute these P values in particular cases PVPRINT NOPVPRIN indicates whether p values should be printed PVPRINT implies PVCALC PRINT and CALC Using this option will sometimes cause regression output to be printed in one column instead of two unless SHORTLAB is used Other things like wide numbers OPTIONS NWIDTH
137. 91 407 Fiorentini Gabriele Calzolari Giorgio and Panattoni Lorenzo Analytic Derivatives and the Computation of GARCH Estimates Journal of Applied Econometrics 11 1996 pp 399 417 McCurdy Thomas H and Morgan leuan G Testing the Martingale Hypothesis in Deutsche Mark Futures with Models Specifying the Form of Heteroscedasticity Journal of Applied Econometrics 3 1988 pp 187 202 53 Commands ASMBUG Example ASMBUG turns the DEBUG switch on during the first phase of TSP execution when the input program is being processed When the debug switch is on TSP produces much more printed output than it usually does This output is normally not of interest to users but may be helpful to a TSP programmer or consultant ASMBUG Usage Include the ASMBUG statement directly before any command s you believe are being parsed incorrectly The DEBUG switch will remain on until a NOABUG statement is encountered While it is the internal representation of each statement and equation will be displayed as it is stored For DEBUG output during the execution phase of the program see the DEBUG statement Output ASMBUG produces a considerable amount of output although not quite as much as the DEBUG statement The breakdown of every statement into its constituent parts by INPT is printed if the statement is a GENR SET IF FRML or IDENT intermediate results from FRML and PARSE are also printed Finally the inte
138. AME is the name of the series To use another name as the prefix include the PREFIX option If no input series is supplied and FREQ Q or FREQ M is in effect Quarterly or Monthly dummies will be created with the names Q1 Q4 or M1 M12 If a list of series is supplied the dummies will be given the names in the list be sure there are as many names as there are values of the variable Since the number of dummies created is usually equal to the number of unique values taken by the input series care should be taken that the series used has a limited number of values When a continuous rather than discrete variable is used for input the number of dummies created could be equal to the number of observations and storage allocation problems are likely to be the result It is well known that a complete set of dummy variables will be collinear with the constant term intercept in a linear regression If you wish to use the dummies together with a constant in this way you can create a set with one of the variables deleted by using the EXCLUDE option In this case the number of variables created will be equal to the number of unique values taken by the input series less one Output The set of dummy variables is stored in data storage and the listname if one was specified is defined as the variable names for the set of dummies Options EXCLUDE NOEXCLUDE excludes the last dummy variable from the list This option is useful if the list wi
139. AST 1 DECR 254321 LIST DROP DECR 3 2 2541 LIST PREFIX B BS X Y 2 1 BY BZ 1 LIST SUFFIX 5 X5 LW ED EX X5is LW5 ED5 EX5 LIST SUFFIX US XLIST LW ED EX XLIST is LWUS EDUS EXUS LIST BSMID BS 2 Y Subscripted list PRINT BS 1 X 1 Y 1 Z 2 Lagged list LIST PRINT BS prints the names X Y Z 1 but not their values LIST DELETE BS removes the list definition for BS 227 Commands Programming trick in fact a list can be used to store any sequence of words or arguments Here are some examples of lists which do not contain variable names but which TSP will understand LIST STRING A title string TITLE STRING same as TITLE A title string LIST OPTS MEAN 5 RANDOM OPTS X same as RANDOM MEAN 5 X 228 LMS LMS Output Options Examples References LMS computes least median of squares regression This is a very robust procedure that is useful for outlier detection It is the highest possible breakdown estimator which means that up to 50 of the data can be replaced with bad numbers and it will still yield a consistent estimate Proper standard errors Such as asymptotically normal for LMS coefficients are not known at present LMS ALL LTS MOST PRINT SILENT SUBSETSz value TERSE lt dependent variable gt lt list of independent variables gt Usage To estimate by least median of squares in TSP use the LMS command just like the OLSQ command F
140. B 339 Commands This guarantees that predicted values of Y are never negative Because the Poisson model implies that the variance of the dependent variable is equal to the mean the effect of Poisson estimation is to downweight the large Y observations relative to ordinary regression If you use LSQ to run an unweighted nonlinear regression with the same exponential mean function you Will get a better fit to large Y values than with the Poisson model The ML command can also be used to estimate Poisson models including panel data models with fixed and random effects See our web page for the panel examples Options The usual nonlinear estimation options can be used See the NONLINEAR entry Examples Poisson regression of patents on lags of log R amp D a scientific sector dummy and firm size POISSON PATENTS C LRND LRND 1 LRND 2 DSCI SIZE References Cameron A Colin and Pravin K Trivedi Regression Analysis of Count Data Cambridge University Press New York 1998 Hausman Jerry A Bronwyn H Hall and Zvi Griliches Econometric Models for Count Data with an Application to the Patents R amp D Relationship Econometrica 52 1984 pp 908 938 Maddala G S Limited dependent and Qualitative Variables in Econometrics Cambridge University Press New York 1983 pp 51 54 340 PRIN PRIN Output Options Example References PRIN obtains the principal components of a group of series The number of s
141. DOUBLE time series parameter equation output of FRML command identity output of IDENT command model output of MODEL command text string for FILE and TITLE program variable a command DO IF or PROC information 424 SYMTAB general matrix symmetric matrix triangular matrix diagonal matrix variable name list 4 Length this is the length of the variable in single precision words The length includes two extra items for time series and matrices which hold dating and dimension information 5 LDOC length of documentation if any see the DOC command 6 DB a flag for storage on the current OUT databank s Example Placement of the SYMTAB command at the end of the run will cause all the variables of the run to be printed out NAME USER TSP program statements SYMTAB causes information on all variables used in the program to be displayed END TSP data section END 425 Commands SYSTEM SYSTEM provides interaction with the operating system without having to terminate the interactive session or batch job If you are using a multitasking windowing system such as Windows you can do the same thing just by switching between windows SYSTEM command string Usage In batch mode the SYSTEM command will execute the command you give in quotes This may be useful for deleting files or running a program that processes files you have created with your TSP job
142. DS method calculates the minimum and maximum possible P values for a given DW using the minimum and maximum possible sets of eigenvalues for and stored as DWL and DWU See Bhargava et al 1982 for more details on bounds DW is not computed for OLSQ with explicit lagged dependent variable s since it is biased DH and or DHALT are computed instead The optional AUTO and HET diagnostics are not calculated for regressions with weights instruments or perfect fits nor when there are any gaps in the SMPL to simplify the processing of lags Note that some of the later diagnostics grouped under AUTO are not strictly for autocorrelation but for heteroskedasticity or structural stability in datasets with a natural time ordering DH is not calculated when it involves taking the square root of a negative value DHALT can be used in all cases it uses the same regression as LMAR1 380 REGOPT LMARx prints a series of test statistics if LMLAGS is greater than 1 The sample size is adjusted downwards with each test and the reported statistic is p k 1 F asymptotically distributed as chi squared p where p is the number of lags QSTATx also prints a series of test statistics using QLAGS WNLAR is a Wald test for AR 1 residuals versus mis specified dynamics left out lagged dependent and independent variables If the original equation was Y A XB the regression Y 2 RHO Y 1 D X 1 is run and the restri
143. E string labels the plot Examples HIST X produces a plot with the vertical axis containing ten cells running from the minimum value of X to the maximum value of X and the horizontal axis showing the number of observations of X which take on values within each of the cells HIST 100 0 5 40 Y1 2 produces two histograms each with 40 cells and a width equal to 2 5 The fraction of observations of Y1 or Y2 which fall in each cell are shown Suppose the variable REASON takes on the values 0 1 2 and 3 with the following counts Value Number of observation S pug dr Mae c s MEE amd The command 186 HIST graphics version HIST DISCRETE REASON will produce a histogram with three cells containing the number of observations taking on the values of REASON 0 2 3 On the other hand the command HIST REASON will default to the INTEGER mode and produce a histogram with four cells containing the number of observations taking on the four values of REASON If SIZE is a variable containing the log sales or employment of a cross section of firms HIST DENSITY TITLE Size distribution produces a graph of the size distribution for the firms with a smooth approximation to the density superimposed 187 Commands IDENT Example IDENT defines identities for TSP These identities can be used to complete simultaneous equations model for full information maxim
144. E PPROD References Amemiya Takeshi The Maximum Likelihood and the Nonlinear Three Stage Least Squares Estimator in the General Nonlinear Simultaneous Equation Model Econometrica May 1977 pp 955 966 148 FIML Berndt E K B H Hall R E Hall and J A Hausman Estimation and Inference in Nonlinear Structural Models Annals of Economic and Social Measurement October 1974 pp 653 665 Calzolari Giorgio and Panattoni Lorenzo Alternate Estimators of FIML Covariance Matrix A Monte Carlo Study Econometrica 56 1988 pp 701 714 Jorgenson Dale W and Jean Jacques Laffont Efficient Estimation of Nonlinear Simultaneous Equations with Additive Disturbances Annals of Economic and Social Measurement October 1974 pp 615 640 149 Commands FIND Interactive Example FIND lists all lines entered in the session so far that begin with the specified TSP command FIND lt TSP gt Usage FIND is useful in conjunction with other interactive commands that use line number as arguments You may want to EDIT or EXEC a particular command but cannot remember its line number The line could be found with REVIEW but if it has been a long session this could take some hunting FIND will accept only one TSP command as an argument Example FIND OLSQ will list all OLSQ commands you have entered or read in during the session along with their line numbers 150 FORCST FORCST
145. E causes all subsequent series to be stored in double precision 15 16 digits vs the default single precision FAST NOFAST performs fast regression calculations without orthonormalization These are slightly less accurate but usually yield no differences in the first 5 or so digits Such calculations are also used in the iterations of LSQ 3SLS and GMM Used to speed up runs with Monte Carlo loops or more than 1000 observations HARDCOPY seis several output format options to values suitable for output routed to a printer These are LIMPRN 120 LINLIM 60 LEFTMG 20 INDENT 10 and DATE and are the default options for printer output 299 Commands INDENT number of spaces to indent printed output from the left margin The default value is 5 LEFTMG left margin for printed output The default is 20 the first column in which output will be printed is 21 LIMCOL column width for input number of columns read in each input line The default value is 500 this option is essentially obsolete LIMERR maximum number of errors allowed in this TSP run The default value is 25 LIMNUM maximum number of numerical warnings divide by zero log of zero or negative number and exponentiation of too large a number allowed in this run before each subsequent one is treated as an ERROR instead of as a WARNING The default value is 100000 LIMPRN printer line width the maximum number of printing positions on the printer including
146. EQ is equivalent to the following commands FRML CONSEQ CONS B0 BGNP GNP BGNP1 GNP 1 PARAM BO 123 BGNP 9 BGNP7 05 LIST RNMSF B0 BGNP BGNP1 4 Same as example 3 but the RESID option is used to make the equation unnormalized This is a form which is most useful for use with the ML command but could also be used with commands like LSQ GMM or FIML OLSQ CONS C GNP GNP 1 FORM VARPREF B RESID EC is equivalent to the following commands FRML EC CONS BO BGNP GNP BGNP1 GNP 1 LIST 9RNMSF B0 BGNP BGNP PARAM RNMSF UNMAKE COEF RNMSF 5 This example creates a similar equation and estimates it with LSQ FORM CE CONS C GNP GNP 1 LSQ CE is equivalent to Note the default coefficient names based on the equation name FRML CE CONS BCE1 GNP BCE2 GNP 1 PARAM BCEO BCE2 LSQ CE 6 This example illustrates including the lagged residual term from AR1 AR1 IMPT C GNP RELP FORM PARAM IE creates the equation FRML IE IMPT IEO IE1 GNP IE2 RELP RHOIE IMPT 1 IEO IE1 GNP 1 E2 GNP 2 PARAM 0 251 8369 IE1 2 7984 IE2 2636 RHOIE 8328 158 FORM 7 this example we estimate the same equation as in example 6 except with AR 4 errors This FRML could also be used in a system of equations with LSQ or FIML FORM NAR 4 IE IMPT C GNP RELP LSQ IE 8 FORM can be used after a VAR estimation VAR Y1 Y2 CT FORM
147. EQ1 EQ2 creates FRML 1 Y1 0 HY1 FRML 2 Y2 2 0 2 where the HY1 0 HY1 HY2 0 HY2 T parameters are created and set to estimated values from the VAR 9 Example of the SUM option FORM SUM 5 X1 X3 creates FRML EQ S X1 X2 X3 159 Commands FORMAT FORMAT is an option used with the READ and WRITE commands It supplies the format for reading and writing data within a TSP program This section describes how to construct a FORMAT string See READ and WRITE for a description of where to use it FORMAT FREE or BINARY or RB4 or RB8 or DATABANK or EXCEL or Usage LABELS or LOTUS or format text string FORMAT has several alternatives 1 160 FORMAT FREE specifies that the numbers are to be read in free format that is they are delimited by one or more blanks but may be of varying lengths and mixed formats FORMAT BINARY specifies that the data are to be read in binary machine format where each variable occupies a single precision floating point word Binary data may not be mixed with data in other formats nor can it be moved from one computer type to another FORMAT RBA Real Binary 4 byte is the same thing FORMAT RB8 is double precision binary 8 byte FORMAT DATABANK specifies that the file being used is a TSP databank This is an alternative to the IN or OUT KEEP statements FORMAT EXCEL reads an Excel spreadsheet file If the file
148. Estimator and two step Estimator Economics Letters 45 1994 pp 33 40 Nawata Kazumitsu Estimation of Sample Selection Models by the Maximum Likelihood Method Mathematics and Computers in Simulation 39 1995 pp 299 303 Nawata Kazumitsu and Nobuko Nagase Estimation of Sample Selection Bias Models Econometric Reviews 15 1996 pp 387 400 Olsen R J Distributional Tests for Selectivity Bias and a More Robust Likelihood Estimator International Economic Review 23 1982 pp 223 240 396 SAVE Interactive SAVE Interactive Examples SAVE writes all current user defined TSP variables into a file which may be restored later with the RESTORE command SAVE filename string Usage SAVE creates a file named TSPSAV SAV by default If a filename string is supplied the filetype SAV is appended if it is not present This file contains all user defined variables including series parameters constants matrices formulas and lists It also contains the current SMPL The SAVE command is useful for stopping an interactive session and restarting it at a later time It is also useful for preserving a session environment if it is likely that later commands or a power failure may cause TSP to abort In general it is better to store TSP variables in databanks or regular files since it is easier to determine the origins and contents of such files and these files are transportable to other computers
149. For example WRITE FORMAT SE beta G12 5 SEB Alternatives for character strings are WRITE FORMAT LABELS the TITLE command and the H Hollerith format type if you like to count characters Format types within the parentheses describe how long numbers are and how many decimal places they have in the data record The format types useful to you in a TSP program are usually X F E and G X specifies columns to be skipped on reading F the format of floating point numbers and E the format of floating point numbers in exponential scientific notation G is used for output and specifies the most suitable format F or E to be used Here is a very simple example using the format F5 2 to read 5 columns of data Col 12345 10000 The number read will be 100 00 since the format specified a 5 digit field with 2 digits after the decimal point Here is an example of data for the format F10 5 5X E10 3 F8 0 0 1 2 4 1 1234567890123456789012345678901234567890123456789 343 5 98765 01 200 161 Commands The first format specifies a field of length 10 with 5 digits to the right of the decimal point but since a decimal point was explicitly included in the data the number will be read as 343 5 from columns 1 to 10 Then 5X specifies that 5 columns 11 to 15 are to be skipped E10 3 reads a number in exponential format which is used when a number is too big or too small for normal notation Once again the 10 s
150. IME YEAR This is not considered sufficient for identifying individuals since the last time period for one individual may be less than the first time period of the next individual TOTAL NOTOTAL selects the total or pooled estimator a plain OLS regression on the whole sample VARCOMP NOVARCOMP selects the variance components or random effects estimator The method of selecting the variance components is controlled with the VBET VSMALL and VWITH options described below Unbalanced data are not a problem For variance components in the time dimension use the REIT option or sort your data by time period and use time as the ID VBET specifies the value of the between variance for VARCOMP VSMALL NOVSMALL selects the small sample variance components formulas for VARCOMP as opposed to the large sample formulas Small sample formulas are unbiased but can result in negative variances while large sample formulas are biased but always yield positive variances To supply your own variance values use VBET and VWITH VWITHz specifies the value of the within variance for VARCOMP WITHIN NOWITHIN selects the within or fixed effects estimator different intercepts for each individual Nonlinear options may be used for the REI and REIT estimators See NONLINEAR Examples Global FREQ PANEL command with ID variable to identify individuals FREQ PANEL ID FIRM DY Y Y 1 PANEL DY C X X 1 Estimate all models 7
151. Independent Observations Econometrica 50 March 1982 pp 483 500 199 Commands INTERVAL Output Options Example References INTERVAL estimates a model like the linear Ordered Probit model but where the limits are known Unlike Ordered Probit the limits may be different for different observations INTERVAL is also similar to two limit Tobit with the difference that when the dependent variable is between the upper and lower bounds only that fact is observed and not its actual value INTERVAL is useful when the dependent variable is in a known range but the actual value has been censored for confidentiality reasons INTERVAL LOWER lt lowerlimit gt UPPER lt upperlimit gt nonlinear options dependent variable list of independent variables Usage The basic INTERVAL statement is like the OLSQ statement first list the dependent variable and then the independent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The LOWER and UPPER options are required Normally these will be series with the lower and upper limits for each observation The dependent variable should be coded so th
152. Interactive SAVE saves all current variables on disk Interactive STORE writes a microTSP format databank WRITE writes variables to a file or to printout 17 Command summary Data Transformations Commands CAPITL CONVERT COPY DELETE DIVIND DUMMY GENR LENGTH NORMAL RANDOM RENAME SAMA SET SORT TREND 18 accumulates a capital stock from an investment series changes a series from higher to lower frequency stock or flow copies any variable deletes any variables computes Divisia price and quantity indices makes dummy variables from a series creates a series using an algebraic formula computes length of a TSP list useful in PROCs normalizes a series usually a price index via division random number generator normal univariate or multivariate uniform Poisson or empirical bootstrap renames a variable Seasonal adjustment using the moving average method modify a scalar or series matrix element with an algebraic formula sorting data create linear trend variable can be repeating like months Matrix Operations Commands Matrix Operations Commands MATRIX MFORM MMAKE ORTHON UNMAKE YLDFAG matrix algebra and transformations change dimensions or type of matrix create a matrix from several series or a vector from scalars orthonormalization create several series from a matrix or scalars from a vector LDL decomposition symmetric indefinite matrix 19 Command summary Li
153. Kinnon reference See the HCTYPE option for the exact formulas SILENT NOSILENT suppresses all output The results are stored TERSE NOTERSE suppresses all regression output except for the log likelinood and the table of coefficients and standard errors 295 Commands WEIGHT the name of a series used to weight the observations The data are multiplied by the square roots of the normalized weighting series before the regression is computed see NORM above The series will be proportional to the inverses of the variances of the residuals If the weight is zero for a particular observation that observation is not included in the computations nor is it counted in determining degrees of freedom WTYPE HET or REPEAT the weight type The default is REPEAT where the weight is a repeat count it multiplies the likelihood function directly This is used for grouped data and is consistent with the UNNORM option WTYPE HET means the weight is for heteroskedasticity only it enters the likelihood function only through the variance The only difference between these two options in the regression output is the value of the log likelihood function all the coefficients standard errors etc are identical With WTYPE HET the log likelihood includes the sum of the log weights the default WTYPEZREPEAT does not include this Examples This example estimates the consumption function for the illustrative model OLSQ CONS C GNP Using population
154. Kolmogorov Smirnov test Journal of Applied Probability 8 1971 pp 431 453 Durbin J and Watson G S Testing for Serial Correlation in Least Squares Regression Biometrika 1951 pp 160 165 Farebrother R W Algorithm AS 153 AS R52 Applied Statistics 33 1984 pp 363 366 Code posted on StatLib with corrections Harvey Andrew The Econometric Analysis of Time Series 2nd ed 1990 MIT Press Geweke John F and Richard Meese Estimating Regression Models of Finite but Unknown Order International Economic Review 22 1981 pp 55 70 Jarque Carlos M and Bera Anil K A Test for Normality of Observations and Regression Residuals International Statistical Review 55 1987 pp 163 172 384 REGOPT Jayatissa W A Tests of Equality Between Sets of Coefficients in Linear Regressions when Disturbance Variances are Unequal Econometrica 45 July 1977 pp 1291 1292 Maddala G S Introduction to Econometrics 1988 Macmillan Chapters 5 6 12 Royston Patrick Algorithm AS R94 Applied Statistics 44 1995 Savin N E and Kenneth J White Testing for Autocorrelation with Missing Observations Econometrica 46 1978 59 67 Shapiro S S and M B Wilk An Analysis of Variance Test for Normality Complete Samples Biometrika 52 1965 pp 591 611 Shapiro S S M B Wilk and H J Chen A Comparative Study of Various Tests of Normality JASA 63 1968 1343 1372 Th
155. L 11 01 2 C Z1 enter EQ1 22 enter EQ2 GMM INST C Z1 Z2 MASK SEL EQ1 EQ2 The same thing can be done using a different list for each equation LIST INST C Z1 LIST INST2 C Z2 GMM INST INST1 INST2 EQ1 EQ2 References Andrews Donald W K Heteroskedasticity and Autocorrelation Consistent Covariance Matrix Estimation Econometrica 59 3 1991 pp 817 858 Arellano M and S Bond Some Tests of Specification for Panel Data Monte Carlo Evidence and An Application to Employment Equations Review of Economic Studies 58 1991 pp 277 297 Chamberlain Gary 1982 Multivariate Regression Models for Panel Data Journal of Econometrics 18 5 45 Gallant A Ronald Nonlinear Statistical Models Wiley 1987 Hansen Lars Peter Large Sample Properties of Generalized Method of Moments Estimation Econometrica 50 July 1982 pp 1029 1054 Hansen Lars Peter and Singleton Kenneth J Generalized Instrumental Variables Estimation of Nonlinear Rational Expectations Models Econometrica 50 September 1982 pp 1269 1286 Newey Whitney K and West Kenneth D A Simple Positive Semi Definite Heteroskedasticity and Autocorrelation Consistent Covariance Matrix Econometrica 55 May 1987 pp 703 708 174 GOTO GOTO Example GOTO provides a method of transferring control within your TSP program that is of changing the order in which the statements are executed However modern program
156. L THEN USER 26 delays execution of commands such as DO loops Interactive clears up space occupied by deleted variables defines a numeric indexed DO loop defines a character indexed DOT loop part of IF THEN ELSE conditional structure end of program end of DO loop end of DOT loop end of PROC definition execute a range of command lines Interactive end of COLLECT loop or program Interactive starts execution at the statement label specified part of IF THEN ELSE conditional structure read commands from an external file Interactive defines a list of variables defines local variables in a PROC directs output to a file instead of screen Interactive defines a user procedure stops TSP without saving backup Interactive stops TSP temporary exit to VMS or DOS without losing TSP session Interactive restores output to screen after OUTPUT Interactive part of IF THEN ELSE conditional structure user programmable command Mainframe Interactive Editing Commands and or Data Commands Interactive Editing Commands and or Data Commands ADD adds arguments to previous command Interactive CLEAR clears TSP s memory data storage Interactive DELETE deletes lines during execution Interactive DROP deletes arguments from previous command Interactive EDIT edits a command Interactive ENTER enter data for a series Interactive FIND lists lines containing a specific TSP command Interactive RETRY edits pr
157. L XLS which consists of quarterly data on two series CJMTL and PMTL for 1948 1 to 1949 1 A B e Date CJMTL PMTL Mar 48 183 4 N A Jun 48 185 2 436 Sep 48 192 1 562 Dec 48 193 3 507 Mar 49 206 9 603 The above file could be read with the following command READ FILE SML XLS Series are read from individual columns in the file Series names are optionally supplied in the first row of the file aligned above the data columns Dates may be given in the first column Many different file configurations are possible and it is possible to read in some series while ignoring others Following are some simple guidelines for creating a spreadsheet file for TSP 364 READ 1 Put the column names in the first row They should be valid series names in TSP lowercase is fine it will be converted to uppercase but imbedded blanks and special characters are not allowed If the file has no names you can supply them when you read the file in TSP but this can be inconvenient If the file contains invalid names the data can be read by TSP as a matrix ignoring the current names and you can supply your own names inside of TSP TSP will not recognize names in lower rows or in sheets other than the first they will be treated as missing values and numbers below them will be read as data for the original column names 2 The second row must contain data 3 If you are reading time series the first column should con
158. LANDSCAP PORTRAIT specifies the orientation of the plot On the Mac specify this option in the dialog box SURFACE NOSURFACE specifies that a 3 dimensional surface is to be plotted TSP Givewin only There must be three arguments x 2 axis variables Once you see the plot you can use ALT gt ALT lt ALT and ALT to rotate the view MAC only WIDTH NOWIDTH specifies whether varying width sizes are to be used to distinguish the lines corresponding to different series on the graph Examples This graph displays Student s t distribution for 2 4 and 10 degrees of freedom a Cauchy distribution and the Laplace and Normal distributions chosen to have the same variance as the others GRAPH DEVICE LJ3 LINE XMIN 5 XMAX 5 TITLE Some Fat Tailed Distributions T T2 T4 T10 CAUCHY LAPLACE NORMAL The following example plots and stores on disk the graph shown in the user s guide SMPL 1 1000 RANDOM SEEDIN 49813 X E Y 1 X E OLSQ GRAPH DEV LJ3 PREVIEW FILE GRAPH PLT TITLE Actual and Fitted Values X Y FIT 180 HELP HELP Examples HELP provides basic syntax information on commands and it will also list command by various functional groups This is useful for checking the syntax on an unfamiliar command or for checking related commands but it does not replace the manuals or the HELP system if you are using the Windows version The syntax summary may also be printed if you supply the
159. M STATIC specifies whether the forecast is to be dynamic or static A static forecast uses historical Supplied by the user values for the lagged endogenous variable s throughout the forecast period while a dynamic forecast uses the lagged forecasted values whenever it can i e in the second observation for lag one third observation for lag two and so forth This also applies to the lagged endogenous variable which appears in the residual in an AR 1 forecast Obviously STATIC is the default for non AR1 forecasting when there is no lagged dependent variable PRINT NOPRINT specifies whether or not the forecast is to be printed and plotted If this option is on a title and a description of the forecasting model are also printed RHO the value of the serial correlation parameter if an AR1 forecast is being requested and the value from the immediately preceding estimation is not wanted Examples In the example below an OLSQ equation is used to extrapolate into a future time period values of the exogenous variables for that time period should already have been loaded The extrapolated values of CONS are stored under the name CONSP they are also printed and plotted because of the PRINT option SMPL 1 20 OLSQ CONS C GNP SMPL 21 30 FORCST PRINT CONSP The example below computes a series CONSP using the equation estimated by OLSQ but with a different series for GNP on the right hand side If an AR1 forecast of this type wer
160. OBCOVA matrix vars vars Number of non missing observations for each pair of variables for PAIRWISE MSD matrix vars 4to Combined table of num obs 13 means std dev s min max sums variances skewness kurtosis median Q1 Q3 IQR Method The mean minimum maximum variance and standard deviation are computed in the usual way The estimated variance and covariance are computed by small sample formulas division by N 1 instead of The formulas for the skewness and kurtosis are the following N M N 1 N 2 S N N 1 M 3 N 1 M 1 2 3 5 Skewness Excess Kurtosis where and M the centered third and fourth moments and S is the estimated standard deviation These statistics can be used to test for normality of the variables The skewness multiplied by the square root of N 6 and the kurtosis multiplied by the square root of N 24 both have a normal 0 1 distribution under the null when the mean and standard deviation have been estimated see Davidson and MacKinnon for a derivation The median is the value of the series at the N 1 2 observation after sorting from low to high The first and third quartiles are the values of the series at the N 1 4 and 3N 3 4 observations respectively and the interquartile range is the difference between these two values If these observation numbers are not integers the values are a weighted average of the bracketing observations Options
161. OC RETRY SET THEN 30 PDL LOCAL ADD DROP EDIT DO GOTO Commands ACTFIT Options Example References ACTFIT computes and prints a variety of goodness of fit statistics for the actual and predicted values of a series Theil references below suggests using these statistics for evaluating an estimated time series equation or forecast ACTFIT SILENT TERSE lt actual series name gt lt predicted series name gt After ACTFIT give the name of the actual data series followed by the name of the fitted or predicted series Output ACTFIT prints a title the names of the series being compared the time period sample over which they are compared and then a variety of computed statistics on the comparison These include the correlation of the two series the mean square error the mean absolute error Theil s inequality coefficient U changes and percent changes in U and a decomposition of the source of the discrepancies between the two series differences in the mean or differences in the variance The following scalar results are stored by ACTFIT variable length description R Correlation coefficient 1 R2 1 Correlation coefficient squared RMSE 1 Root mean square error MSE 1 Mean squared error MAE 1 Mean absolute error ME 1 Mean error RMSPE 1 Root Mean Squared Percent Error MSPE 1 Mean Squared Percent Error MAPE 1 Mean Absolute Percent Error MPE 1 Mean Percent Error BETA 1 Regression coef
162. ORM identifies the observation where the price index is normalized The index will have the value given by PVAL at this observation Note that there is no default for PNORM 116 DIVIND PVAL is the value to which the observation PNORM is to be normalized The default is 1 0 PRINT NOPRINT tells whether the derived Divisia index series are to be printed The default is no printing QNORM identifies the observation where the quantity index is normalized The index will have the value given by QVAL at this observation Note that there is no default for QNORM is the value to which the observation QNORM is to be normalized The default is 1 0 TYPE Q specifies that the Divisia quantity index is to be computed and the price index is to be obtained by dividing total expenditure by the quantity index TYPE P specifies that the Divisia price index is to be computed and the quantity index is to be obtained by dividing total expenditure by the price index specifies that both a Divisia price index and a Divisia quantity index are to be computed Note that the product of the two indices will not be exactly proportional to total expenditure WEIGHTZARITH specifies that the weights to be used in computing this period s rate of change of the index are the arithmetic averages of the shares in this period and the previous period WEIGHT GEOM specifies that the weights are the geometric averages of the shares in
163. PDL polynomial distributed lag variable specification may be used for a right hand side variable in any linear estimation procedure OLSQ INST LIML AR1 PROBIT and TOBIT It constrains the coefficients on the lags of that variable to lie on a polynomial of the degree specified by the user The Shiller lag available with OLSQ only uses an additional argument to relax this constraint somewhat PDL lag variables have the following form varname lt degree gt lt lags gt NONE or FAR or NEAR or BOTH Shiller lag variables add the XIPRIOR argument varname lt degree gt lt lags gt NONE or FAR or NEAR or BOTH XIPRIOR Usage You may include a PDL polynomial distributed lag specification anywhere in the list of right hand side variables in a linear estimation procedure There is no explicit limit on the number of right hand side variables which may contain a PDL specification The form of a PDL variable is the name of the variable whose lags you want in the model followed by parentheses containing three items the number of terms in the polynomial the degree plus one the number of lags of the variable to be included including the zeroth lag and what kind of end point constraint you place on the polynomial The polynomial distributed lag is a method for including a large number of lagged variables in a model while reducing the number of coefficients which have to be estimated by requiring the coefficients to lie on
164. PL 1 50 READ FILEZ STATES DAT FORMAT F2 0 F4 0 3F 4 1 F6 0 F7 0 5F4 1 F4 0 STATE X1 X12 where STATE DAT contains 01284952 326 4 1 9 4120 19055553 320 813 7 8 9 5 46677 024592 0 021 211 0 403 303168 176 819 815 5 8 29047 so forth After this data set has been read the series STATE and X1 through X12 have the following values STATE 1 2 X1 2849 4592 X2 52 3 0 0 X11 8 9 15 5 X12 5 4 8 2 X12 6677 9047 Reading a Stata dta file printing documentation for variables and checking the min max mean of the variables read all variables from file names are stored RNMS READ FILEz acq95 dta show documentation Stata labels for all variables SHOW DOC 9RNMS check min max mean for all variables MSD TERSE RNMS 369 Commands Examples of reading matrices READ NROW 4 NCOL 3 COEFMAT 0 32 0 5 1 3 0 30 0 4 1 35 0 25 0 61 1 1 0 28 0 55 1 23 READ NROW 2 TYPE SYM COVAR 4 64 2 35 1 READ NROW 3 TYPE TRIANG TMAT 1 23 456 READ NCOL 5 TYPE DIAG BAND 110 140 0 35 50 The matrices stored by these four examples are the following COEFMAT 13 138 723 READ Examples for spreadsheet files We will use this SML WKS file it is the Lotus version of the SML XLS file shown earlier in the following examples CJMTL PMTL 48 1 183 4 NA 48 2 185 2 0 436 48 3 192 1 0 562 48 4 193 3 0 50
165. R Interactive REGOPT RENAME REPL 251 254 259 265 268 270 273 274 278 286 287 288 289 290 291 298 302 306 307 309 311 312 322 324 328 331 334 337 341 344 345 351 353 354 362 372 373 386 387 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 4 Index Table of Contents RESTORE 388 RETRY Interactive 389 REVIEW Interactive 390 SAMA 391 SAMPSEL 393 SAVE Interactive 397 SELECT 398 SET 399 SHOW 401 SIML 403 SMPL 408 SMPLIF 410 SOLVE 412 SORT 417 STOP 419 STORE 420 SUPRES 421 SUR 422 SYMTAB 424 SYSTEM 426 TERMINAL Interactive 428 THEN 429 3515 430 TITLE 433 TOBIT 434 TREND 438 TSTATS 440 UNIT 441 UNMAKE 442 UPDATE Interactive 444 USER Mainframe 445 VAR 446 WRITE 449 YLDFAC 453 455 Introduction Welcome to the TSP 5 0 Help System The TSP Help system contains the complete reference manual for TSP providing a description of every command and command option organized alphabetically by command It is not intended as an introduction to the program nor as a tutorial for the inexperienced TSP user To learn how to use the program you should obtain a copy of the TSP 5 0 User s Guide available at http www
166. R the number of variables for an Engle Granger Dickey Fuller cointegration test The default is 1 plain unit root test and the maximum is 6 the correlation coefficient for the bivariate normal distribution 76 CDF TREND NOTREND specifies whether a trend term 1 2 was included the regression for Dickey Fuller TSQ NOTSQ specifies whether a squared trend term 1 4 9 was included in the regression for Dickey Fuller PRINT NOPRINT turns on printing of results PRINT is true by default if there is no output specified Examples 1 To compute the significance level of a Hausman test statistic with 5 degrees of freedom CDF CHISQ DF 5 HAUS produces the output CHISQ 5 Test Statistic 7 235999 Upper tail area 20367 2 Significance level of the test for AR 1 with lagged dependent variable s CDF DHALT or REGOPT PVPRINT DHALT before the regression is run 3 Two tailed critical value for the normal distribution CDF INV 05 produces the output NORMAL Critical value 1 959964 Two tailed area 05000 4 Several critical values for the normal distribution READ PX 1 05 01 CDF INV NORM PX CRIT PRINT PX CRIT 5 Fcritical values CDF INV F DF1 3 DF2 10 05 77 Commands produces the output 3 10 Critical value 3 70827 Upper tail area 0500 6 Bivariate normal CDF BIVNORM RHO 5 PRINT 1 2 PBIV produces the output BIVNORM Test Statis
167. R 1 model FRML E Y A B X RHO Y 1 1 DIFFER E B creates the following unnormalized FRML FRML DE1 X RHO X 1 4 Differentiate a Constant Elasticity of Substitution production function with respect to the two inputs K and L compute the two marginal product series using the derivative equations and print them FRML CES Y A EXP GAM T ALPHA L RHO BETA K RHO 1 RHO DIFFER CES LK GENR DCES1 LMP GENR DCES2 KMP PRINT LMP KMP 5 Differentiate the log likelihood for a PROBIT model with respect to its parameters and store the generated equations FRML PROBIT LOGL LOG D CNORM A B X 1 D 1 CNORM A B X DIFFER PRINT PREFIX LOGL PROBIT A B The results are FRML LOGL1 DLOGL1 D NORM A B X 1 D NORM A B X D CNORM A B X 1 D 1 CNORM A B X FRML LOGL2 DLOGL2 D NORM A B X X 1 D NORM A B X X D CNORM A B X 1 D 1 CNORM A B X This example illustrates the convenience of DIFFER when you use it to generate analytic derivatives of an objective function for use in another program or language such as Fortran 114 DIR Interactive DIR Interactive DIR examines a disk directory without interrupting the interactive TSP session DIR or filename Usage DIR may be used in three different forms The first DIR makes use of a wildcard specification and produces a list of all files in your current directory with the extens
168. RCH parameters to protect against numerical problems from negative zero or infinite variances 08 lt 1 amp B 1 if HINITZSTEADY a gt 0 20 presample h 20 if 8 is not estimated presample h gt 0 if 8 is estimated h gt 0 Use of the ZERO NOZERO option controls the technique for bounding see the description under Options below If a parameter is bounded on convergence its standard error is set to zero to make the other estimates conditional on it In this case it is wise to try some alternative non bounded starting values to check for an interior ML solution see the final example Output A title is printed based on the options indicating OLS OLS W OLS M ARCH ARCH M GARCH or GARCH M estimation Standard iteration output follows with starting values etc Then standard regression output is printed the only difference is the extra coefficients The order of the estimated coefficients is 50 ARCH B p presample h t h t is stored in HT Options E2INIT HINIT or INDATA or PREDATA specifies the initialization of the presample values of epsilon The default HINIT sets them equal to h t their unconditional expectation as given by the current HINIT option INDATA reserves the first NAR observations in the current sample to compute residuals and squares these this was the default in TSP Version 4 3 and earlier PREDATA attempts to use NAR observations prior to the current sample to
169. RINT or TERSE options are set The names of the equations and endogenous variables are printed the value of the objective function at the optimum and the corresponding estimate of the covariance of the structural disturbances If minimum distance estimation was used the trace of the weighted residual covariance matrix is the objective function the equation given above with H equal to the identity matrix Otherwise the objective function is the negative of the log of the likelihood function If instrumental variable estimation was used the objective function is labelled E PZ E and stored as PHI This is analogous to the sum of squared residuals in OLSQ it can be used to construct a pseudo F test of nested models Note that it is zero for exactly identified models if they have full rank For two stage least squares a test of overidentifying restrictions FOVERID is also printed when the number of instruments is greater than the number of parameters It is given by PHI S2 inst params Following this is a table of parameter estimates and asymptotic standard errors as well as their estimated variance covariance unless it has been suppressed For each equation LSQ prints a few goodness of fit statistics the sum of squared residuals standard error mean and standard deviation of the dependent variable and the Durbin Watson statistic The computation of these statistics is described in the regression output section of the User s
170. RM CONS INST I C P P 1 K 1 INVR C P 1 K 1 E 1 TM W2 G TX FORM INV INST W1 C E E 1 TM INVR C P 1 K 1 1 TM W2 G TX FORM WAGES IDENT WAGE W W1 W2 IDENT BALANCE E E CX l G TX W P IDENT PPROD P E TX W1 IDENT CAPSTK K K 1 I LIST KLEIN CONS WAGES BALANCE PPROD INV WAGE CAPSTK MODEL KLEIN KLEINC SOLVE TAG S TOL 0001 METHOD FLPOW KLEINC This model solves for CX consumption investment W1 wages in the private sector W total wage bill E production of the private sector P profits and K capital stock using TM time W2 government wage bill TX taxes and G government expenditures as exogenous variables At the end of this simulation the solved variables are stored under the names CXS IS etc References Gilli Manfred Causal Ordering and Beyond International Economic Review November 1992 pp 957 971 Gili Manfred Graph theory based tools in the practice of macroeconometric modeling in Methods and Applications of Economic Dynamics S K Kuipers L Schoonbeek and E Sterken eds North Holland Amsterdam Steward D V On an Approach to Techniques for the Analysis of the Structure of Large Systems of Equations SIAM Review Volume 4 pp 321 342 269 Commands MSD See Also CORR COVA Output Options Examples References MSD produces a table of means standard deviations minima maxima sums variances skewness and kurtosi
171. S INST Z W1 C E E 1 FORM WAGES IDENT WAGE W W1 W2 IDENT BALANCE E E CX I G TX W P IDENT PPROD P E TX W1 IDENT CAPSTK KzK 1 4 415 Commands LIST KENDOGIWEPCSW1K LIST KLEIN CONS WAGES BALANCE PPROD INV WAGE CAPSTK MODEL KLEIN KENDOG KLEINC SOLVE TAG S TOL 0001 METHOD FLPOW KLEINC References Fletcher R and M J D Powell A Rapidly Converging Descent Method for Minimization Comput J 6 1963 pp 163 168 Maddala G S Econometrics McGraw Hill Book Company New York 1977 pp 237 242 Ortega J M and W C Rheinboldt Iterative Solution of Nonlinear Equations in Several Variables Academic Press New York 1970 Chapter 7 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 Chapters 10 11 12 Theil Henri Principles of Econometrics Wiley New York 1971 432 439 416 SORT SORT Options Examples SORT sorts the observations of a series in increasing order Other series or all series currently defined may also be reordered in the same order as the first series SORT ALL REVERSE lt key series gt lt list of other series gt or SORT AVETIES MINTIES RANK REVERSE lt key series gt lt rank of series gt Usage The simplest form is SORT followed by the name of a series to be sorted A QuickSort is performed so any equal values may be reordered relative to e
172. SDT creates subsamples 6 10 and 3 8 The second SELECT statement is equivalent to SMPL 1 10 SMPLIF T gt 2 amp T lt 9 MSD T SELECT 1 can be used to return to the last SMPL statement 398 SET SET Examples SET performs computations on scalar variables and single elements of time series or matrices SET lt scalar gt lt algebraic expression gt or lt subscripted variable gt lt algebraic expression gt Usage SET consists of the name of a scalar variable or an element of a series or matrix followed by an equal sign and an arbitrary algebraic expression The expression must follow the usual TSP rules for formulas The formula is evaluated using the current values of all the variables and the result is stored in the variable on the left hand side of the equation If the left hand side is an element in a series and matrix only that particular element is changed the remainder of the variable is unchanged whether or not the REPL mode is on Legal forms of scalar variables in TSP are the following these forms can be used wherever scalars are allowed 1 Asimple variable name such as BETA or D1 This could be already defined by a CONST or PARAM but this is not required 2 subscripted series such as GNP I GNP 72 1 GNP 72 or YOUNG 40 If the frequency is NONE you can use a simple subscript in the same units as your sample If the frequency is QUARTERLY 4 or MONTHLY 12 use a v
173. SE suppresses the main tables only the summary tables are printed Note that JOH and EG NOALLORD have no summary tables so TERSE suppresses all their output SILENT NOSILENT suppresses all output This is useful for running tests for which you only want selected output which can be obtained from the variables that are stored see the table below Examples FREQ A SMPL 1909 1970 COINT LRGNP LEMPLOY performs 11 augmented WS tau and Dickey Fuller tau unit root tests with to 10 lags All tests are first done for LRGNP then repeated for LEMPLOY Eleven augmented Engle Granger tau tests are constructed with O to 10 lags with LRGNP as the dependent variable in the cointegrating regression Optimal lag lengths for all tests are determined using the AIC2 rule The test is recomputed for the optimal lag using the maximum available observations and this is stored in the final column of the table All tests use a constant and trend variable UNIT ALL LRGNP performs the same unit root tests for LRGNP as the above example Also performs the Phillips Perron z tests computed separately for 0 to 10 lags in the autocorrelation robust estimate COINT NOUNIT ALLORD MAXLAGz8 LRGNP LEMPLOY LCPI performs 27 augmented Engle Granger tau tests That is 9 tests with 0 to 8 lags with LRGNP as the dependent variable in the cointegrating regression Then repeat the tests using LEMPLOY and later LCPI as the dependent variable i
174. STATE matrix Evolving state vectors RESD matrix obs n Direct residuals if SMOOTH is on SMOOTH matrix obs m Smoothed state vectors Note that the first few rows in the residuals or state vectors may be zero if those observations were used to calculate priors Note also that recursive residuals for OLS regressions can be obtained using OLSQ with the following options set REGOPT CALC RECRES Method 205 Commands The Kalman filter recursively updates the estimate of beta t and its variance using the new information in y t and X t for each observation so it can be viewed as a Bayesian method However the user does not have to supply priors they are calculated automatically for regression type models from the initial observations of the sample See the Harvey reference for the actual updating formulas If the default prior is singular KALMAN adds one more observation to the calculation The smoothed state vectors if requested are estimates based on the full sample again see Harvey for the details The method of estimation for each time period is maximum likelihood conditional on the data observed to that point An orthonormalizing transformation of X is used to improve accuracy The variance sigma squared and the log likelihood are computed from the recursive residuals The recursive residuals are e in Harvey s notation so that E e t e t If you do not factor sigma squared out of Q a
175. SUBSETS number of random subsets to use The default is 500 to 3000 If the number of possible subsets is less than the number of random subsets all possible subsets will be evaluated systematically TERSE NOTERSE suppresses all printed output except the table of coefficient estimates and the value of the objective function Examples LMS 21 26 2 uses 3000 random subsets LMS MOST Y C Z1 Z6 probably all subsets 2000 subsets but not the same set as the default LMS SUBSET 2000 Y C Z1 Z6 References Rousseeuw P J Least Median of Squares Regression JASA 79 1984 pp 871 880 Rousseeuw P J and Leroy A M Robust Regression and Outlier Detection Wiley 1987 Rousseeuw P J and Wagner J Robust Regression with a distributed intercept using Least Median of Squares Computational Statistics and Data Analysis 17 1994 pp 66 68 http www agoras ua ac be Rousseeuw Progress 232 LOAD LOAD LOAD is a synonym for READ LOAD BYOBS BYVAR FILE filename string or filename FORMAT BINARY or DATABANK or EXCEL or FREE or LOTUS or RB4 or RB8 or format text string FULL NCOL lt number of columns NROW lt number of rows PRINT SETSMPL TYPE CONSTANT or DIAG or GENERAL or SYMMETRI or TRIANG UNIT I O unit number list of series or matrices 233 Commands LOCAL Example LOCAL specifies the variables which will be considered local to a TSP PROC Their values
176. T statistics for fixed effects for FEI If the regression includes a PDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling Method PROBIT uses analytic first and second derivatives to obtain maximum likelihood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly TSP uses zeros for starting parameter values unless START is used to override this see the NONLINEAR entry As in other regression procedures in TSP estimation is done using a generalized inverse in the case of multicollinearity of the independent variables The numerical implementation involves evaluating the normal density and cumulative normal distribution functions The cumulative normal distribution function is computed from an asymptotic expansion since it has no closed form See the reference under the CDF command for the actual method used to evaluate CNORM The ratio of the density to the distribution function is also known as the inverse Mills ratio This is used in the derivatives and with the MILLS option MILLS is actually the expectation of the structural residual where the model is given by 347 Commands Y X B N 0 1 D I y 0 MILLS is the value of the following two expressions depending on whether D 0 or 1 NORM Xb 1
177. TSP 5 0 Reference Manual Bronwyn H Hall and Clint Cummins international TSP International 2005 Copyright 2005 by TSP International First edition Version 4 0 published 1980 TSP is a software product of TSP International The information in this document is subject to change without notice TSP International assumes no responsibility for any errors that may appear in this document or in TSP The software described in this document is protected by copyright Copying of software for the use of anyone other than the original purchaser is a violation of federal law Time Series Processor and TSP are trademarks of TSP International ALL RIGHTS RESERVED Table Of Contents 1 Introduction OO OPO d Co Too 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 3 Commands 28 29 1 Welcome to the TSP 5 0 Help System 1 Introduction to TSP 2 Examples of TSP Programs 3 Composing Names in TSP 4 Composing Numbers in TSP 5 Composing Text Strings in TSP 6 Composing TSP Commands 7 Composing Algebraic Expressions in TSP 8 TSP Functions 10 Character Set for TSP 11 Missing Values in TSP Procedures 13 LOGIN TSP file 14 2 Command summary 15 Display Commands 15 Options Commands 16 Moving Data to from Files Commands 17 Data Transformations Commands 18 Matrix Operations Commands 19 Linear Estimat
178. Then comes the iteration output If PRINT has not been specified only convergence or non convergence messages are printed for each block of the model as they are solved along with the iteration count If the PRINT option is on a one line message for each iteration is printed showing the iteration number the block number the objective function the sum of squared residuals and the value of the stepsize for this iteration If PRNRES is on the residual error from each equation is printed for the first MAXPRT iterations on each block and also at convergence of the block This can help in identifying problem equations if your model is difficult to solve After solution of the model over the whole sample a message is printed if the variables are being saved in data storage Following this message a table of the results of the simulation is printed labelled by the observation ids This table can be suppressed by specifying the NOPRNSIM option IFCONV is stored as a series which contains ones if the simulation for the observation converged and zeroes otherwise This is useful to check the convergence The sum of IFCONV should be equal to NOB if all periods converged Method The model to be solved is broken into a series of blocks by the procedure MODEL These blocks are alternately recursive and simultaneous A recursive block is one which can be solved simply by specifying the value s of a set of previously determined endogenous variabl
179. a seasonal component 2 how much ordinary and seasonal differencing is required to make the series stationary 3 and what are the minimum orders of the autoregressive and moving average polynomials required to explain adequately the time series Subject to various option settings BUIDENT will present plots of autocorrelations and partial autocorrelations for various levels of differencing of the input series BJIDENT BARTLETT ESACF IAC NAR lt order of AR for ESACF gt NDIFF lt degree of differencing gt NLAG lt number of autocorrelations to be computed gt NLAGP lt number of partial autocorrelations to be computed NMAz order of MA for ESACF gt NSDIFF lt degree of seasonal differencing NSPAN lt span of seasonal PLOT PLOTAC PLTRAW NOPRINT PREVIEW SILENT lt list of series gt Usage BJIDENT followed by the name of a series is the simplest form of the command No differencing of the series will be done in this case and the output will consist of a plot of the series and a printout and plot of the autocorrelations of the first 20 lags of the series and the first 10 partial autocorrelations Autocorrelations are the correlation of the series with its own values lagged once lagged twice and so forth for 20 lags Partial autocorrelations are the correlations measured with the series residual after all the prior lags have been removed That is the second partial autocorrelation is the correlation of the series lag
180. a smooth polynomial in the lag The purpose of the constraints is to force the lag coefficients at either end of the lags over which you are estimating to go to zero that is the NEAR constraint forces the coefficient of the first lead to zero while the FAR constraint forces the coefficient of the lag one past the last included lag to zero The BOTH and NONE constraints are self explanatory 324 PDL The number of coefficients which are estimated for a PDL variable are the number of terms in the polynomial less the number of constraints This must be positive and less than or equal to the number of lags in the unconstrained model For a long totally unconstrained distributed lag an implicit list is best For example the following three statements are equivalent OLSQ CONS72 C GNP72 GNP72 1 GNP72 15 OLSQ CONS72 C GNP72 16 16 NONE OLSQ CONS72 C GNP72 GNP72 1 GNP72 2 GNP72 3 GNP72 4 GNP72 5 GNP72 6 GNP72 7 GNP72 8 GNP72 9 GNP72 10 GNP72 11 GNP72 12 GNP72 13 GNP72 14 GNP72 15 For Shiller lags the additional argument specifies a prior for the variance of the differenced coefficients smaller values imply coefficients that are smoother A value for the prior equal to zero is equivalent to simply using PDL A very large value will yield unconstrained lag coefficients Output The coefficients of PDL variables are estimated by forming linear combinations of the underlying lagged variables and
181. a table of parameter estimates and asymptotic standard errors as well as their estimated variance covariance matrix if it has been unsuppresseq 145 Commands For the default HCOV B option this is computed by summing the outer product of the gradient vector for all structural parameters and error covariance parameters over all observations inverting this matrix the BHHH matrix and taking the submatrix corresponding to the structural parameters This is a consistent estimate of the information matrix with good small sample properties Note that the BHHH matrix is not block diagonal between the structural and error covariance parameters at the maximum even though the second derivative matrix is block diagonal at the maximum so this procedure is necessary See Calzolari and Panattoni 1988 for details HITER U and HCOV U use numeric second derivatives for iteration and computing the variance estimate respectively using equations 25 3 23 and 25 3 26 in Abramovitz and Stegun 1972 Numeric second derivatives can provide a very close approximation to the true Hessian The drawback is that computing them is relatively slow requiring 2 K K function evaluations for a model with K parameters HITER U yields quadratic convergence during iterations which can be faster than HITER F BFGS method if the number of parameters is less than about 7 HCOV U provides standard errors which are more reliable than BFGS HCOV F often produces false ze
182. ach other To sort several series in the same order as the first just list them in the command To sort all currently defined series use the ALL option All series must have the same length To sort series individually use a DOT loop DOTX YZ SORT To sort in random order draw a random variable and sort based on its values To obtain the rank order of a series use the second form of the command The options AVETIES and MINTIES specify how tied series are to be treated Output There is no printed output but the series are stored after reordering their observations Options ALL NOALL causes all currently defined series to be sorted in the order defined by the named series 417 Commands AVETIES NOAVETIE specifies that the average rank is to be stored for tied series Used with RANK option MINTIES NOMINTIE specifies that the minimum rank is to be stored for tied series Used with RANK option RANK NORANK stores the ordinal rank of the first series in the second series the order of the first series is not changed REVERSE NOREVERS sorts in decreasing order Examples If X 20 40 30 50 10 and Y 12345 SORT RANK X yields RX 2 4 3 5 1 X is unchanged SORT X Y yields X 10 20 30 40 50 and Y 51324 SORT ALL X yields the same as the previous command SORT REVERSE X yields X 50 40 30 20 10 If SCORE 10 20 20 40 SORT RANK AVETIES SCORE RA yields RA 1 2 5 2 5 4 SORT
183. al Distribution Function Same thing as DRAW EXPON NOEXPON specifies that the random number generated is to follow the exponential distribution Aexp Ax 37 V x 2 2 Use the LAMBDAc option to specify the parameter lambda GAMMA NOGAMMA specifies that the random number generated is to follow the gamma distribution ia Use the MEAN and STDEV options to specify the parameters which must be non negative GENERATOR 1 or 2 Type of uniform random number generator Use GEN 1 to reproduce results from older versions of TSP to 4 4 this generator has period equal to 2 31 1 The new default generator has period 2 319 LAMBDA the exponential or double exponential parameter See the EXPON and LAPLACE options LAPLACE NOLAPLACE specifies that the random number generated is to follow the Laplace double exponential distribution f x Aexp 2 Ax 0 V x 1 2a7 356 RANDOM MEAN the expected value of the random variable or a series containing expected values Applies only to the normal gamma Poisson and negative binomial random variables The default value is zero you will want to change this for the gamma Poisson and negative binomial distributions When a series is supplied each random number drawn will come from a distribution with a different mean NEGBIN NONEGBIN specifies that the random number generated is
184. al of Econometrics 97 2000 189 202 217 Commands LENGTH Examples LENGTH determines the length of a list It is useful in PROCs which are passed lists as arguments since the length may be required in matrix dimensions or degrees of freedom calculations This command counts arguments it does not compute the length of a series or a matrix LENGTH lt list of variables or lists gt lt length of list gt Output The list s on the command line are expanded and the final total number of items is stored in the last argument Examples Count the number of parameters from an estimation LSQ EQ1 EQ5 LENGTH RNMS NOPAR The following commands cause LENA to be stored with a value of 3 and LENAB to be stored with a value of 5 LIST LA X YZ LENGTH LA LENA LIST BB B1 B2 LENGTH LA BB LENAB 218 LIML LIML Output Options Examples References LIML computes the limited information maximum likelihood estimator for a single equation linear structural model To estimate a single equation nonlinear model via LIML use FIML with unrestricted reduced form equations for the included endogenous variables those which appear on the right hand side of the primary equation LIML BEKKER FEI FEPRINT FULLERz scalar value INSTz list of instruments SILENT TERSE dependent variable list of rhs endogenous and exogenous variables Usage LIML s form is identical to the 2SLS INST command specify a list o
185. alent to DOC CONS72 Consumption in 1972 dollars DOC ADD CONS72 Source ERP 1990 121 Commands DOT Options Examples DOT is the first statement in a DOT loop which is like a regular DO loop except that the values of the index are a series of character strings names Each of these names is substituted in turn each time through the loop wherever the symbol dot appears in a variable name The dot may appear anywhere in the variable name DOT CHAR lt nesting level character gt INDEX lt variablename gt VALUE lt variablename gt list of sector names or strings Usage The primary use of DOT loops in TSP is the processing of multisectoral data where the same group of statements is to be repeated on each sector The names on the DOT command are the names of the sectors They may also be integer numbers which will be treated as the corresponding character string Use the form 01 702 etc if you wish to include leading zeroes in the numbers Each DOT loop must be terminated by an ENDDOT statement Any number of statements may appear between DOT and ENDDOT statements TSP will cycle through them as many times as there are sectors on the DOT statement DOT loops may be nested up to ten deep and more than one dot included in each variable name i e VAR The names substituted for the dots are taken in the order that the DOT statements appear one from each statement See the second example below If there
186. alid TSP date as the subscript A variable 4 characters or less can also be used on dated and undated series Since variables do not take date values except for annual frequency a variable subscript is always relative to the start of the current SMPL For example the following loop fills the series X throughout the current SMPL SMPL 48 1 86 2 DO l 1 NOB SET ENDDO 3 subscripted matrix A matrix may have a single numeric or variable subscript which is computed by the following formula 399 Commands subscript j 1 NROW where is the row index of the element and j is the column index Matrices may also be doubly subscripted but any variable subscripts must be 2 characters or less Here are some examples of legal matrix elements XL LJ XL 2 6 MAT 345 MAT I XL IL247 A K1 LL These are illegal MAT VARSUB XL L20 2 A K L 1 GNP I 1 The first and second are illegal because subscript names are limited to 4 or 2 characters The third and fourth are illegal because expressions are not allowed as subscripts SET is not recommended for creating a series or matrix READ GENR TREND DUMMY etc should be used to create series READ MFORM MMAKE or COPY should be used to create matrices SET can be used to update series and matrices or to retrieve particular elements from them Output SET produces no printed output A scalar is stored in data storage or a series or matrix is updated
187. also on data such as income ANALYZ will compute the standard errors for such a FRML using the covariance matrix of the estimated parameters and treating the data as fixed constants See the example below of computing an elasticity series Output If the PRINT option is on ANALYZ prints a title the names of the input parameters the equations in symbolic form a table of the derived functions and their standard errors and the chi squared value of a test that the functions are jointly zero This chi squared has degrees of freedom equal to the number of equations The P value significance level for the chi squared test is also printed If the print option is off the default only the derived functions and the chi squared test are printed ANALYZ also stores the calculated parameters and their variances in data storage as though they were estimation results whether or not the PRINT option is on The results are stored under the following names variable type length description RNMSA list eqs Names of derived parameters WALD scalar 1 Value of Wald test scalar 1 Number of derived parameters NCIDA scalar 1 Degrees of freedom scalar 1 P value significance of Wald test COEFA vector eqs Values of derived parameters SESA vector eqs Standard errors of derived parameter TA vector eqs T statistics asymptotically normal vector eqs p values corresponding to TA MSD matrix eqs 8 M
188. alues for T statistics Gradient of log likelihood at convergence Variance covariance of estimated coefficients Means of probability derivatives Residuals for non truncated observations Inverse Mills ratios If the regression includes a PDL variable the following will also be stored SLAG scalar MLAG scalar 1 Sum of the lag coefficients 1 Mean lag coefficient number of time periods LAGF vector lags Estimated lag coefficients after unscrambling 435 Commands Method TOBIT uses analytic first and second derivatives to obtain maximum likelinood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly Starting values for the parameters are obtained from a regression on the observations with positive values of the dependent variable See Greene 1981 p 508 formula 13 and footnote 5 for the details Alternative starting values may be supplied in START see NONLINEAR A globally concave parameterization of the likelihood function is used for iterations Multicollinearity of the independent variables is handled with generalized inverses as in all TSP regression procedures The numerical implementation involves evaluating the normal density and cumulative normal distribution functions The cumulative normal distribution function is computed from an asymptotic expansion since it has no closed form See the references under CDF for the actual method used to evaluate CNORM The ratio o
189. ame will give three stage least squares estimates using the S matrix you specify Alternatively if you use the 3515 form of the LSQ command with the INST option LSQ automatically computes consistent nonlinear two stage least squares estimates of the parameters uses them to form an estimate of the residual covariance matrix S and then computes three stage least squares estimates To use any of these estimators first specify the equations to be estimated using FRML statements and name the parameters and supply starting values with PARAM statements an alternative to this is the FORM PARAM command after a linear estimation procedure Any parameters which appear in more than one equation are assumed to be the same parameter and the equality constraint is automatically imposed LSQ always determines the linearity or nonlinearity of the model if the model is linear in the parameters it prints a message to that effect and uses just one iteration Output LSQ stores its results in data storage The estimated values of the parameters are stored under the parameter names The fitted values and residuals will only be stored if the RESID option is on the default In addition the following results are stored variable type length description LOGL scalar 1 Log of likelihood function if valid TR scalar 1 Trace of COVT if the minimum distance estimator is used PHI scalar 1 E PZ E the objective function for instrumental varia
190. an ARIMA model use the BJEST command followed by the options you want in parentheses and then the name of the series and specification of the starting values if you want to override the default starting values The general form of the model which is estimated is the following 1 A B G B a t constant w t is the input series after ordinary and seasonal differencing and a t is the underlying white noise process B is the backshift or lag operator Bw t w t 1 Lw t in the usual notation are all polynomials in B The order of these polynomials is specified by the options NSAR NAR NSMA and NMA respectively Gamma B and Delta B are the seasonal AR and MA polynomials and Phi B and Theta B are the ordinary AR and MA polynomials All the polynomials are of the form 55 Commands 2 B 1 B B B Note that the coefficients all minus signs front of them The BJEST procedure uses conditional sum of squares estimation to find the best values of the coefficients of these polynomials consistent with the unobserved variable at being independently identically distributed The options specify the exact model which is to be estimated and can also be used to control the iteration process As with the other iterative estimation techniques in TSP it is helpful to specify reasonable starting values for the parameters Unlike the other nonlinear TSP procedures the
191. ances was Laplace this is by analogy to least squares where the likelihood function and conventional standard error estimates assume that the true distribution is normal with a small sample correction to the standard errors The additional statistics are shown in the table below PHI contains the sum of the absolute values of the residuals This quantity divided by the number of observations and squared is an estimate of the variance of the disturbances and is proportional to the scaling factor used in computing the variances of the coefficient estimates The LM test for heteroskedasticity is a modified Glejser test due to Machado and Santos Silva 2000 This test is the result of regressing weighted absolute values of the residuals on the independent variables 214 LAD variable type length description PHI scalar 1 sum of abs value of residuals IFCONV scalar 1 1 if there were no simplex iteration problems 0 otherwise UNIQUE scalar 1 1 if the solution is unique 0 otherwise Method The LAD estimator minimizes the sum of the absolute values of the residuals with respect to the coefficient vector b N min y X b The estimates are computed using the Barrodale Roberts modified Simplex algorithm A property of the LAD estimator is that there are K residuals that are exactly zero for K right hand side variables this is analogous to the least squares property that there are only N K linearly independent residuals I
192. angular square root of this matrix which saves a step VMEAN mean vector for multivariate normal random variables the default is a vector of zeroes or for creating several series at the same time each with a different mean Examples These examples each generate a thousand random numbers and store them under the series name given SMPL 1 1000 RANDOM STDNORM Std normal random variable RANDOM UNIFORM FLAT Uniform 0 1 r v 358 RANDOM FLATAB FLAT B A A Uniform on the interval A B CDF CHISQ DF 3 INV FLAT CHI3 Chi square 3 from uniform as p value T1EV LOG LOG FLAT Type I Extreme Value from uniform RANDOM CAUCHY FAT RANDOM EXPON LAMBDA 2 Z RANDOM LAPLACE LAMBDA 0 5 DE RANDOM T DF 5 FATAIL RANDOM GAMMA MEAN 10 STDEV 2 GAMMA 10 RANDOM NEGBIN MEAN 10 STDEV 5 10 RANDOM MEAN 10 STDEV 3 1623 NORM10 RANDOM MEAN 10 POISSON 5510 The last two examples produce random numbers with the same mean and variance but different distributions A normal variable with mean 10 and standard deviation 3 1623 could also have been generated by the following NORM10 10 STDNORM 3 1623 The next example generates a bivariate normal random vector with correlation 0 5 LOAD TYPE SYM NROW 2 COVMAT 1 51 RANDOM VCOV COVMAT NU1 NU2 Now we use the estimated residuals from a regression to generate 10 samples with the same empirical distribu
193. any lines were read from the INPUT file REVIEW would display everything that had been read 193 Commands INST Output Options Examples References INST obtains single equation instrumental variable estimates INST is a synonym for 2SLS By choosing an appropriate list of instrumental variables INST will obtain conventional two stage least squares estimates Options allow you to obtain weighted estimates to correct for heteroskedasticity or to obtain standard errors which are robust in the presence of heteroskedasticity of the disturbances INST FEI FEPRINT or DIAGONAL INST lt list of instruments gt ROBUSTSE SILENT TERSE UNNORM WEIGHT lt variable name gt lt dependent variable name gt lt independent variable names gt Usage In the basic INST statement list the dependent variable first and then the independent variables which are in the equation Include an option INST containing a list of variables to be included as instruments in parentheses The list of instruments must include any exogenous variables in the equation in particular the constant C as well as any additional instruments you may wish to specify There must be at least as many instrumental variables as there are independent variables in the equation to meet the rank condition for identification Any observations with missing values will be dropped from the sample Two stage least squares is INST with all the exogenous v
194. ariables in the complete model listed as instruments and no other variables Valid estimation can be based on fewer instruments when a complete model involves a large number of exogenous variables or estimates can be made even when the rest of a simultaneous model is not fully specified In these cases the estimator is instrumental variables but not really two stage least squares If there are exactly as many instruments as independent variables specified the resulting estimator is classic instrumental variables that is b Z X Z y where Z is the matrix of instruments X the matrix of independent variables and y the dependent variable For the more general case see the formulas below 194 INST Instrumental variable estimation can also be done using the AR1 procedure for models with first order serial correlation and the LSQ procedure for nonlinear and multi equation models In these cases include the list of instruments in the INST option See those commands for further information The FEI options specifies that a model including individual fixed effects is to be estimated FREQ PANEL must be in effect when using this option The estimates are computed by removing individual means from all the variables which implies that the effects are treated as exogenous variables The WEIGHT option is not available with FEI The list of independent variables on the INST command may include PDL variables however you are respons
195. as weights the next example regresses the fraction of young people living alone on other demographic characteristics across states Since the regression is in terms of per capita figures the variance of the disturbances is proportional to the inverse of population OLSQ WEIGHT POP YOUNG C RSALE URBAN CATHOLIC Other examples of the OLSQ command OLSQ ROBUSTSE LOGP C LOGP 1 LOGR OLSQ TBILL C RATE 4 12 FAR References Belsley David A Kuh Edwin and Welsch Roy E Regression Diagnostics Identifying Influential Data and Sources of Collinearity John Wiley amp Sons New York 1980 pp 11 18 Davidson Russell and James G MacKinnon Estimation and Inference in Econometrics Oxford University Press New York 1993 pp 552 556 Durbin J Testing for Serial Correlation in Least Squares Regression When Some of the Regressors are Lagged Dependent Variables Econometrica 38 1970 p 410 421 296 OLSQ Durbin J and G S Watson Testing for Serial Correlation in Least Squares Regression Biometrika 38 1951 pp 159 177 Judge et al The Theory and Practice of Econometrics John Wiley amp Sons New York 1981 pp 11 18 126 144 Krasker William S Kuh Edwin and Welsch Roy E Estimation for Dirty Data and Flawed Models Handbook of Econometrics Volume Griliches and Intrilligator eds North Holland Publishing Co New York 1983 pp 660 664 Longley James W An Appraisal of Leas
196. at each observation in turn The vector MOVA t may be rewritten as a matrix where each row corresponds to a year and contains p elements The total number of non missing elements in MOVA is T p because p 2 observations are dropped at the beginning and end of the series The p seasonal factors are formed by averaging each column in MOVA SFAC 1 1 1 MOVA 2 1 MOVA 3 1 MOVA T 1 SFAC p 1 1 MOVA 1 p MOVA 2 p MOVA T 1 p They are normalized to average to unity either arithmetically or geometrically depending on the option specified The seasonally adjusted series is then computed by dividing the old series by the seasonal factors Options PRINT NOPRINT specifies that the ratio of the series to its moving average and the computed seasonal factors be printed ARITH NOARITH specifies that the arithmetic mean be used for normalization rather than a geometric mean Examples SAMA PRINT ARITH GNPQ GNPQA SAMA ARITH GNPQ GNPQA PRINT SFAC 2MOVA These two examples have the same effect The first prints the seasonal factors and the moving average ratio series The second suppresses the printing but then prints SFAC and MOVA which are the same as what would have been printed Reference Census Bureau Seasonal Analysis of Economic Time Series proceedings of the Conference on the Seasonal Analysis of Economic Time Series September 1976 392 SAMPSEL SAMPSEL Output Options Example Re
197. at it lies between its two bounds For example if category 3 means that the dependent variable Y is between 10 and 20 then Y should be coded so that 10 Y 20 The lower and upper limits for this observation will take the values 10 and 20 See the examples below The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the current sample INTERVAL will print a warning message and will drop those observations The list of independent variables on the INTERVAL command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See the PDL section for a description of how to specify such variables Output 200 INTERVAL The output of INTERVAL begins with an equation title and the usual starting values and diagnostic output from the iterations Final convergence status is printed After convergence the number of observations the value of the log likelihood and the Schwarz Bayes information criterion are printed This is followed by observation counts for lower upper and double bounded observations and the usual table of right hand side variable names estimated coefficients standard errors and associated t statis
198. ata storage variable type length description LOGL scalar 1 Log of likelihood function IFCONV scalar 1 Convergence status 1 success LR scalar 1 Likelihood ratio test for zero slopes SRSQ scalar 1 Scaled R squared for multinomial logit RSQ scalar 1 Squared correlation between Y and FIT for binary logit SSR scalar 1 Sum of squared residuals Y FIT for binary logit RNMS list params List of parameter names GRAD vector Gradient of likelihood function at maximum COEF vector zZparams Estimated values of parameters SES vector params Standard errors of estimated parameters T vector params T statistics vector p values for T statistics VCOV vector par par Estimated variance covariance of estimated parameters DPDX matrix vars Mean of probability derivatives for choices multinomial variables This matrix is invariant to the set of coefficients which are normalized to zero as are the 237 Commands differences between sets of coefficients DPDZ matrix vars Mean of probability derivatives for choices conditional variables vars condvars choices This matrix consists of NCHOICE submatrices of dimension NCOND x NCHOICE stacked vertically and it is block symmetric Not calculated if NCHOICE varies by case FIT matrix obs Matrix of fitted probabilities when or choices gt 2 and there is series or obs observation per case Length NOB
199. ataset the fastest way to access it is as a TSP databank see the OUT and IN commands for further details READ BYOBS BYVAR FILE filename string or filename FORMAT BINARY or DATABANK or EXCEL or FREE or LOTUS or RB8 or STATA or format string FULL NCOL lt number of columns gt NROW lt number of rows gt PRINT SETSMPL TYPE CONSTANT or DIAG or GENERAL or or TRIANG UNIT I O unit number list of series or matrices or constants or READ Usage If TSP encounters a simple READ statement with no arguments it transfers to the data section and begins reading data until it reaches an END statement or end of file at which point it returns to the line in the TSP program following the READ statement A NOPRINT command in the data section will stop the data values from echoing to the output A READ statement followed by a list of series names is the easiest way to read small quantities of data If no options are specified the data is assumed to follow the READ statement directly in free format each number separated from the others by one or more blanks Each group of data may be terminated by a semicolon although this is not required If there is more than one series to be read in the order of the data is the first observation of each of the series followed by the second observation of each of the series and so forth READ can also be used to read several variables from an external file
200. ate values for parameters defined by the PARAM statement are LSQ for nonlinear single and multi equation least squares including minimum distance estimators and FIML All other procedures treat parameters like constants scalar variables which have the arithmetic value they have been assigned either by a PARAM statement or by later estimation FORM PARAM can also be used to create parameters KKK PARAM ignores any or in the command This is useful for pasting back in starting values of the parameters from a previous estimation Output PARAM produces no printed output it stores the variables named in data storage with a type equal to parameter Example A common problem in nonlinear estimation is that one or more parameters may enter the model in a highly nonlinear fashion making it difficult to estimate unless you have good starting values In this example we estimate a subset of the parameters of a model conditional on the value of another parameter DELTA and then reestimate with all the parameters free FRML INVEQ I LAMBDA I 1 ALPHA GNP DELTA R PARAM LAMBDA ALPHA CONST DELTA 15 322 PARAM LSQ PARAM DELTA LSQ When the second LSQ is done the starting values for LAMBDA and ALPHA will be those determined by the first estimation while the starting value for DELTA will be 15 which it was assigned by the CONST statement 323 Commands PDL Examples References A
201. ation If any of the arguments are not in the equation a zero derivative will be stored no error will be trapped If the equation is unnormalized no left hand side variable the derivative equation will also be unnormalized Output Ordinarily DIFFER produces no printed output It stores the derivatives as equations in data storage If the PRINT option is specified DIFFER prints each equation in symbolic form with a title specifying what the derivative is Options DEPVARPR prefix for new dependent variable The default is D hsvar If the original equation has no dependent variable is unnormalized then the default is to create unnormalized derivative equations PRINT NOPRINT tells whether the resulting derivatives are to be printed PREFIX the name to be given to the derivative equations if the names etc are not wanted The equation names will consist of the prefix name followed by the argument number for that derivative Examples 1 Using the default options FRML EQ Y A B X G X 2 DIFFER EQ creates the following FRMLs 113 Commands FRML DEQ1 DY1 1 dY dA FRML DEQ2 DY2 X dY dB FRML DEQ3 DY3 B 2 G X dY dX FRML DEQ4 DY4 0 dY dQ 2 Using some prefix options for naming the results FRML EQ Y A G X 2 DIFFER PREFIX GYX DEPVAR MPX EQ X creates the following FRML FRML GYX1 MPX1 B 2 G X dY dX 3 Unnormalized equation residual from A
202. ation symmetric matrix Default identity matrix Specifies the noise to signal ratio if H identity matrix In Harvey s notation this is RQR XFIXEDz X matrix for measurement equation when it is fixed over time Examples One of the simplest Kalman filter models is equivalent to OLSQ using a transition equation of b t b t 1 2b This model can be estimated with the command KALMAN NOETRANS YC X which produces the same coefficient estimates as OLSQ Y C X but calculates them recursively along with the recursive residuals To estimate a Cooley Prescott adaptive regression model where b t follows random walk with a nondiagonal variance matrix KALMAN VTRANS NSRMAT Y C X1 X2 A stochastically convergent parameter model convergent towards zero in this case since the transition matrix has roots less than one MFORM TYPE DIAG NROW 3 TMAT 9 KALMAN BTRANS TMAT VIRANS NSRMAT Y X1 X2 Here is an example with two dependent variables note the two lists of exogenous variables which must be of the same length In this case both equations are forced to have the same two coefficients KALMAN Y1 2 C1 X1 C2 X2 The example below has two dependent variables but in this case the equations have separate coefficients note the use of zero variables This specification is still somewhat unrealistic because H is identity same variance and no correlation between errors in the equations 207
203. ation of the 3SLS command is the same as that of the LSQ command except that the INST list is required The variables in the INST list will be used to instrument all the equations so that the actual instrumental variable matrix has the form given by Jorgenson and Laffont 1975 rather than that given by Amemiya 1977 In a simultaneous equations model this means that a variable cannot be exogenous to one equation and endogenous to another See the GMM command if you wish to relax this restriction Method Three stage least squares estimates are obtained by first estimating a set of nonlinear or linear equations with cross equation constraints imposed but with a diagonal covariance matrix of the disturbances across equations This is the constrained two stage least squares estimator The parameter estimates thus obtained are used to form a consistent estimate of the covariance matrix of the disturbances which is then used as a weighting matrix when the model is reestimated to obtain new values of the parameters 430 3SLS The actual method of parameter estimation is the Gauss Newton method for nonlinear least squares described under LSQ If the model is linear in the parameters and endogenous variables only two iterations will be required one to obtain the covariance matrix estimate and one to obtain parameter estimates For further details on the properties of the linear three stage least squares estimator see the Theil text or Z
204. atrix of simulation results when NDRAW option is used VCOVA matrix eqs eqs Estimated variance covariance of derived 36 ANALYZ parameters Method Assume that a previous estimation in TSP has stored a vector of K parameter estimates b stored COEF and their variance covariance matrix Var b stored in VCOV Values and standard errors for the functions f b are desired To compute these ANALYZ obtains the first derivatives of f with respect to b analytically af 7 2b The functions f b and the matrix G are evaluated at the current values of b and any constants or data values which may appear in f b The variance covariance matrix for f b is then asymptotically or exactly if f b is linear in b defined as Var f b E Var b G This is known as the delta method For example if Mz1 and f b f1 2 b1 then G 2 with zeros elsewhere if K gt 1 and Var f1 4 Var b1 If the equations are linear and an OLSQ command was used for estimation ANALYZ prints the F statistic for the set of joint restrictions FST WALD NCIDA In addition ANALYZ computes and prints the implied restricted original coefficients and their standard errors These are stored under COEFC VCOVGC etc Options COEF vector containing the values of the parameters in the equations to be analyzed This vector should correspond to the parameters listed in the NAMES option and also to the supplied VCOV matrix The default is
205. average parameters in the model The default is zero NSPAN the span length of the seasonal cycle For example with quarterly data NSPAN should be 4 The default is the current frequency that is 1 for annual 4 for quarterly 12 for monthly If the current FREQ is 0 the default is the value from the last BJ command ORGBEG the first origin to use in forecasting The default is the final observation in the current sample This option should be specified as a TSP date e g 82 4 Note that this date must lie within the current SMPL ORGEND the final origin to use in forecasting The default is ORGBEG This option should also be specified as a TSP date Like ORGBEG this date must lie within the current SMPL PLOT NOPLOT specifies whether the series and the forecasts are plotted PREVIEW NOPREVIEW TSP Givewin only specifies that the forecasts are to be displayed in a high resolution graphics window PRINT NOPRINT specifies whether the results will be printed versus just stored in FIT RETRIEVE NORETR specifies whether BJFRCST should try to find parameter values from an immediately preceding BJEST procedure SILENT NOSILENT suppresses all the output SILENT is equivalent to NOPLOT NOPRINT Examples The first example produces one set of forecasts for the Nelson auto sales data The log of the original series is forecasted BJFRCST PRINT NMA 2 NSMA 1 NDIFF 1 NSDIFF 1 NSPAN 12 NHORIZ 24 ORGBEG 264 LOGAUTO S 0 11106
206. be transposed to obtain L References 453 Commands Almon Clopper Matrix Methods in Econometrics Addison Wesley Publishing Company Reading Mass 1967 pp 115 120 Faddeev V N Computational Methods of Linear Algebra trans C Benster Dover New York 1959 Rao C Radhakrishna Linear Statistical Inference and its Applications John Wiley and Sons New York 1965 pp 17 20 454 3 ISES em sd ia irons ee 430 A Rud 31 ADD eed 33 Algebraic Functions 8 ANALYZ 35 41 49 54 BIE ST ici 55 BJERGST dies ries 63 68 C 72 itae s 74 Character 5 11 CLEAR tono 81 GEOSE dienen 82 ihi t aet ets 84 COLLEGIT ahh 95 Commands composing 7 5 97 CONST sess esi 98 Control Flow Commands 26 CONVERT ccccccsecerssceeteseesseens 99 COPY sacas 102 CORRICOVA 103 Cross Reference Pointers 29 D Data Analysis Commands 20 Data to from Files Commands 17 105 106 107 108 109 160 362 Data Transformations Commands 18 104
207. bers maximal output except for DH and DHALT which require a lagged dependent variable olsq t2 ct Equation 3 Method of estimation Ordinary Least Squares Dependent variable T2 Current sample 1 to 10 Number of observations 10 Mean of dep var 38 5000000 Std dev of dep var 34 1735765 Sum of squared residuals 528 000000 Variance of residuals 66 0000000 Std error of regression 8 12403840 R squared 949764521 Adjusted R squared 943485086 LM het test 391604968 531 Durbin Watson 454545455 012 Breusch Godfrey LM AR MA1 850705917E 38 000 Breusch Godfrey LM AR MA2 850705917E 38 000 Ljung Box Q statisticl 3 33333333 068 Ljung Box Q statistic2 3 38842975 184 ARCH test 258229904 611 CuSum test 1 26364964 003 CuSumSq test 465909091 051 Chow test 53 5714286 000 Chow het rob test 53 5714286 000 LR het test w Chow 26 4920970 000 White het test 3 38983051 184 Breusch Pagan het test 1 74908036 186 374 REGOPT Jarque Bera test 1 01478803 602 Shapiro Wilk test 869383609 098 Ramsey s RESET2 850705917E 38 000 F zero slopes 151 250000 000 Schwarz 36 3245264 Akaike Information Crit 36 0219413 Log likelihood 34 0219413 Estimated Standard Variable Coefficient Error t statistic P value 22 0000000 5 54977477 3 96412484 1 0041 T 11 0000000 89
208. bject to overall TSP limits on space The third form of GENR is usually used to compute predicted values after a nonlinear estimation LSQ or FIML It consists of GENR followed by an equation name and then the name of the variable where you want to place the computed values of the equation If no such name appears GENR will put the series in data storage under the name given on the left hand side of the equation If the equation was implicit there is no left hand side variable you must supply a name for GENR to store the results If there are any missing values in the input series for GENR or arithmetic errors during computation missing values will be stored for the affected observations of the output series and warnings will be printed unless the missing data is part of the argument to a MISS function Missing values can be generated directly with the internal names MVAL MISS NA or MV 167 Commands If the series appearing on the left hand side of the equation also appears on the right with a lag or lags or leads it may be updated dynamically as it is computed A warning message dynamic GENR is printed TSP Versions prior to 4 1 did not update the right hand side lagged dependent variables dynamically Use the SILENT option or the SUPRES SMPL command to suppress this message This is much more efficient than SET with subscripts For example U RHO U 1 E creates an AR 1 variable U PDV REV PDV 1 1 R does a
209. ble estimation FOVERID scalar 1 test of overidentifying restrictions for 2SLS IFCONV scalar 1 Convergence status 1 success RNMS list params list of parameter names GRAD vector Gradient of objective function at the convergence 244 COEF SES T SSR YMEAN SDEV S DW RSQ COVU W COVT VCOV RES FIT vector vector vector vector vector vector vector vector vector matrix matrix matrix matrix matrix matrix params params params eqs eqs eqs eqs eqs eqs eqs eqs eqs eqs eqs eqs par par obs eqs obs eqs LSQ Vector of estimated values of the parameters Vector of standard errors of the estimated parameters Vector of corresponding t statistics Sum of squared residuals for each of the equations stored in a vector Means of the dependent variable for each of the equations Standard deviations of the dependent variable for each of the equations Standard errors for each of the equations Durbin Watson statistics for each equation R squared for each equation Residual covariance matrix The inverse square root of COVU the upper triangular weighting matrix Covariance matrix of the transformed weighted residuals This is equal to the number of observations times the identity matrix if estimation is by maximum likelihood Estimated variance covariance of estimated parameters Residuals actual fitted va
210. can be the exact same ones which you estimated using FIML or LSQ If you want to have linear equations in your model from OLSQ INST or AR1 estimation use FORM to make them after the estimation Note that there must be as many equations as endogenous variables so that the Jacobian of the model is a square matrix SIML solves the model specified by the equations over the current SMPL one period at a time The starting values for the variables are chosen as follows 1 Ifthe variables already exist the actual values for the current period are used as starting values unless they are missing 403 Commands 2 If the variables do not exist and this is the first period of the simulation the value one is used as a starting value 3 If the variables do not exist and this is not the first period of the simulation the values of the last period solution are used as starting values Output If no options are specified the normal output from SIML begins with a title and listing of options This is followed by a table of the data series if the PRNDAT option is on Next is the iteration output If PRINT has not been specified only one line per iteration is printed showing the starting value of the objective function the ending value and the value of the stepsize for this iteration Even this information is not printed if you have specified the SILENT option If PRINT has been specified considerably more output is produced showing
211. cept it selects observations from the previous SMPL statement instead of the current sample The first of these is by far the most common the simplest form of the SMPL statement just specifies one continuous group of observations For example to request that observations 1 through 10 of the data be used use the command SMPL 1 10 If you want to use more than one group of observations specify the groups in any order on the SMPL statement For example the following SMPL skips observation 11 SMPL 1 10 12 20 408 SMPL The observation identifiers on the SMPL statement can be any legal observation identifier Simple integers if the frequency is none Years if the frequency is annual If the year is less than 201 and greater than 0 1900 will automatically be added this can be reset with the BASEYEAR option See the OPTIONS command entry for details Years followed by a colon and the period if the frequency is monthly or quarterly SMPL can be changed as often as you like during a TSP program While a SMPL is in force no observations on series outside that SMPL will be stored or can be retrieved unless they are specified with lags or leads and the lagged led value is within the sample For example if the sample runs from 48 to 72 and the variable GNP 1 is specified the 1947 value of GNP will be used for the 1948 observation of GNP 1 Output Every time the SMPL is changed TSP prints out the current sample un
212. ch the data is to be written The quotes are required and should surround a Fortran FORMAT statement including the parentheses but excluding the word FORMAT If you are unfamiliar with the construction of a Fortran FORMAT statement see FORMAT FULL NOFULL specifies whether symmetric triangular and diagonal matrices are to be expanded before being written out This option is ignored when the FORMAT DATABANK or FORMAT LABELS are used UNIT an integer number usually between 1 and 4 or 8 and 99 which is the Fortran input output number of an external file from which the variables listed will be written This is rarely used Examples This example is the inverse of the example for the READ command WRITE FILEZ FOO DAT XYZ To look at some transformed series PRINT X LX DX The following example creates a spreadsheet file that can be read by the corresponding READ command examples FREQ Q SMPL 48 1 49 1 READ CJMTL 183 4 185 2 192 1 193 3 206 9 READ PMTL 436 562 507 603 WRITE FILEZ SML WKS CJMTL PMTL 451 Commands The next example creates a spreadsheet file with the same data columns but no dates or series names since a matrix is used MMAKE M CJMTL PMTL WRITE FILE SMM WKS The SMM WKS file that is created A B 1 1834 NA 2 185 2 436 3 1921 562 4 198 3 507 5 206 9 603 452 YLDFAC YLDFAC Example References YLDFAC factors a symmetric matrix X into a triangular matrix L
213. changed using the RESAMPLE option See Judge et al 1988 for details on the statistical properties of this method of estimation See Davidson and MacKinnon 1993 on testing for normality of the residuals in least squares The censored version of the estimator is computed using an algorithm due to Fitzenberger 1997 Options LOWERc the value below which the dependent variable is not observed The default is no limit NBOOT number of replications for bootstrap standard errors For the uncensored model the default is zero if there are 100 or more observations conventional standard errors under the assumption that the disturbances are Laplace Otherwise NBOOT 200 The coefficient values from the bootstrap are stored in a BOOT an NBOOT by NCOEF matrix for use in computing other statistics such as the 95 confidence interval QUANTILE quantile to fit The default is 0 5 the median 216 LAD RESAMPLE BILIAS DIRECT specifies the resampling method to be used for the bootstrap standard error estimates for the censored model DIRECT resamples from the original data and runs the censored regression estimator to compute the SEs BILIAS the default zeros the observations where the predicted dependent variable is censored then resamples from the partially zeroed observations and runs the uncensored LAD quantile regression to compute the bootstrap SEs This method is faster and avoids possible convergence or local optima problems with
214. class is a concentrated log likelihood function the FRML method can only handle the unconcentrated log likelihood which is usually more nonlinear and often harder to write Models with complicated constraints A good example would be ARCH models where the conditional variance must be positive for every observation Bad Applications for either method 1 Existing linear models in TSP PROBIT TOBIT LOGIT SAMPSEL The regular TSP commands are more efficient more resistant to numerical problems often have better starting values and provide model specific statistics See Timing example below To give an idea of how much this convenience costs in terms of CPU time here is a timing example run on the VAX 11 780 of a Probit on 385 observations 8 variables canned Probit procedure PROBIT command BHHH algorithm method of ML HITER B scoring Newton s method uses 2nd ML derivatives HCOV N CPU Procedure seconds The moral is that ML should not be used when you have a Fortran coded alternative estimation program but could be useful if you don t want to spend your time developing such a program Also in this case the method of scoring was somewhat faster than Newton s method although the latter is more powerful it takes fewer iterations Tips 261 Commands Output Write the equation carefully to avoid things like Log x lt 0 or Exp x gt 88 These are fatal error
215. closing ENDDO statement will be executed If there is an ELSE statement subsequently the statements following it will be executed in the same manner if the result of the IF was false 190 IN Databank IN Databank Examples IN specifies a list of external databank files to be searched automatically for variables not found in TSP s working data storage IN list of filenames or IN filename strings Usage Follow the word IN with the names of the TSP databanks to be searched for your variables Most systems use binary TLB files for databanks After the IN statement appears in your program TSP searches each of the files in the order in which you specified for any variables which are not already loaded The IN statement remains in effect until another IN statement is encountered If you wish to stop searching any files include an IN statement with no arguments to cancel the previous statement Up to 8 IN databanks can be active at one time IN sets the FREQ and SMPL from the first series in a databank if no SMPL is present Examples Suppose you have created a TSP databank called TSPDATA TLB on disk with the members GNP CONS etc using the OUT statement Then you can access this databank as shown below It is important that the frequency specified match the frequency of the series you want from the databank or a warning will be issued The sample does not necessarily have to be the same since TSP will remove
216. coefficient vector variance covariance matrix Usage The simplest form is TSTATS followed by the name of a vector containing the estimated values of a set of coefficients and the name of a symmetric matrix which contains the estimated covariance matrix of those coefficients If the vector of coefficients is N long the matrix must be of order N The NAMES option allows you to label the coefficients in the table conveniently Output A table of regression coefficients etc is printed unless it has been SUPRESed RNMS COEF SES and VCOV are stored Option NAMES list of coefficient names The default is just to number the coefficients 1 2 etc Example Suppose for example that we have manually created PDL variables for use in a nonlinear regression and then unscrambled the regression coefficients and their covariance matrix by using the PDL transformation matrix The estimated lag coefficients are called BETA and their covariance matrix estimate is VARB A table of estimates with the corresponding t statistics is printed by implementing the following command TSTATS NAMES BETA1 BETA7 BETA VARB 440 UNIT UNIT UNIT is a synonym for COINT which performs unit root and cointegration tests UNIT ALL NOCOINT CONST DF FINITE MAXLAG lt number of lags gt MINLAG lt number of lags gt PP RULE AIC2 SEAS SEAST SEASTSQ SILENT TREND TSQ UNIT WS lt list of variables gt lt list of sp
217. coefficients after unscrambling Method Like the binary Probit model the Ordered Probit model is based on an unobserved continuous dependent variable y The model is y XB e Instead of y we observe a category value Y where a larger category value implies a larger value of y In binary Probit the category values are 0 for y lt 0 and 1 for y gt 0 In Ordered Probit more than 2 category values are usually involved The category values need not be consecutive and the lowest category does not have to be 0 The boundary values between the different categories are estimated parameters MUs The lowest effective boundary value is normalized to 0 just as in binary Probit For example suppose there are 3 categories with category values 0 1 and 2 Y 0 if MUO lt XB e lt MU1 infinity and MU1 0 Y 1 if MU1 lt XB e lt 02 MU2 is an estimated parameter Y 2 if MU2 lt XB e lt MUS Note MUS infinity The MUs are always given names based on the category value for which they are the lower bound MU2 in the example above is the lower bound for category with value 2 X normally includes a constant term C which can be though of as a replacement for MUT in this case the other MUs can be interpreted as being measured relative to the value of C The estimated MU values are constrained to follow a strict ordering MUO lt lt MU2 etc Negative and non integer catego
218. cription for further information NOREPL REPL tells whether replacement mode is to be used in updating series See REPL for further information RESID NORESID tells whether residuals and fitted values are to be computed and stored after the estimation procedures LSQ FIML INST and AR1 SECONDS number of seconds All commands which take longer than this amount of time to execute display a message on the screen giving line number command name and elapsed time for execution The default is 10 seconds For more precise control of timing single or multiple commands use the DATE variable command If you supply a fractional part to the argument like OPTIONS SECONDS 2 1 within command profile timings will be given for regression commands ML GMM and MATRIX This is used to investigate which parts of commands are slow for use in improving speed SIGNIF number of significant digits to be printed in tables In general this is the number of digits printed to the right of the decimal point and the default value is 5 TOL tolerance for matrix inversion This parameter is used to decide when a matrix is singular The value of TOL is compared to the diagonals of the square root matrix of the matrix being inverted as it is formed and if the diagonal is smaller than TOL it is set to zero effectively dropping that row and column from the matrix before inversion The default value of TOL on IBM is 10 E 13 a conservative value this value cau
219. ction D B RHO is tested This is asymptotically distributed as chi squared with degrees of freedom equal to the number of non singular coefficients on the lagged Xs WNLAR is the same as COMFAC in AR1 OBJFN GLS ADF is no longer computed here See the COINT command ARCH is a regression of the squared residual on the lagged squared residual RECRES are recursive residuals calculated using a Kalman Filter see the KALMAN command You can display CUSUM and CUSUMQ plots by turning on the PLOTS option RECRES can also be used for the Von Neumann ratio test for autocorrelation CHOW is an F test for parameter stability The default is to split the sample into equal halves but the CHOWDATE option can be used to choose an unequal split If there are insufficient degrees of freedom in one of the halves the test is still valid but it is usually not very powerful The CHOWHET test is robust to simple heteroskedasticity and is the MAC2 test from Thursby 1992 Note that the Chow test does not have the assumed F distribution under heteroscedasticity LRHET is a likelihood ratio test for heteroscedasticity between the two periods in the same sample division as the Chow test SSR SSR LRHET T log 7 log I SSR k WHITEHT is a regression of the squared residual on cross products of the RHS variables If the model is Y BO B1 X1 B2 X2 381 Commands and the residuals are E the regression
220. d TOL lt tolerance for matrix inversion Usage Usually OPTIONS is the first statement in a TSP run before the NAME statement if there is one doing this sets the output format for the entire run for example the CRT option However an OPTIONS statement may be included anywhere in your TSP program except the load section to change certain global parameters If you use the same options repeatedly you may want to place them in a login tsp file Every time TSP starts it checks for a login tsp file and sets the options accordingly Normally TSP looks for login tsp in your working directory If it does not find one it looks in the directory in which you installed TSP for DOS and Windows in the folder in which you installed TSP for Macs and in the home directory on Unix Many options on the OPTIONS statement can also be set using individual commands for compatibility with older versions of TSP These commands include PLOTS NOPLOT REPL NOREPL DEBUG NODBUG MAXERR same as LIMERR and TOL Options APPEND NOAPPEND updates the OUT file in batch mode at each nonlinear iteration This is useful for monitoring the progress of a long estimation on a multitasking operating system 298 OPTIONS ARGSUB NOARGSUB controls substitution of actual arguments for formal arguments inside a PROC The default is ARGSUB NOARGSUB is useful if the PROC has LOCAL variables with the same names as global variables being passed as argument
221. d Estimation Approach and Operational Limitation of the General Markov Structure Annals of Economic and Social Measurement 2 1973 pp 525 530 Harvey Andrew C Time Series Models 1981 Philip Allen London pp 101 119 Harvey Andrew C Forecasting Structural Time Series Models and the Kalman Filter 1989 Cambridge University Press New York Kalman R E A New Approach to Linear Filtering and Prediction Problems Journal of Basic Engineering Transactions ASME Series D 82 1960 35 45 Maddala G S Econometrics 1977 McGraw Hill Book Co New York pp 396 400 209 Commands KEEP Databank Examples KEEP causes variables to be saved in the currently open TSP databank s KEEP lt list of variable names gt or KEEP ALL Usage KEEP is followed by a list of variable names to be saved in the databank s named in the last OUT statement executed It marks the variables for storage at that point in your program Variables are automatically saved when they are changed or created with GENR or SET statements by matrix computations or as the output of CAPITL SAMA etc but it may be convenient to specify explicitly the storing of such variables as equations Any TSP variable may be named in a KEEP statement including equations constants parameters matrices series or models If you want to avoid cluttering up your databank with intermediate results or when you have errors in your run store
222. d This can be useful if your disk storage is space constrained and you have a great deal of data 105 Commands DBCOPY Databank Options Example DBCOPY makes it possible to move a TSP databank from one type of computer to another The actual databank files are not compatible between different computer types DBCOPY DOC lt list of filenames gt Usage Follow the word DBCOPY with the filenames of the TSP databanks to be moved A file containing TSP commands and data is created for each databank When this file is moved to another computer and run with TSP the databank is created with all its original variables and values The filename with the TSP commands will be the same as the databank name except it will have filetype TSP instead of TLB The TSP file has record length of at most 80 so it can be easily moved to another computer There are no restrictions on the sizes of the databanks SMPLs and FREQs of the series or types of the TSP variables it handles CONST PARAM SERIES MATRIX FRML and IDENT The SMPLs and FREQs are determined by the series in the databank not by the current SMPL or FREQ Options DOC NODCC controls the listing of documentation created with the DOC command Specify NODOC if you will be using TSP version 4 1 or earlier Example Suppose that you have a databank called FOO TLB which contains two time series X and Y and a parameter B The command DBCOPY FOO would create t
223. d contains the value of all the series listed on the statement for that observation If only one series is listed WRITE writes at least one observation per record The first WRITE command to a file creates a new file or overwrites an existing file Subsequent WRITEs to the same text file FREE or formatted in the same batch run will append to the file if the file has remained open this is usually the case unless you have a lot of open files Subsequent writes to the same spreadsheet file EXCEL or LOTUS overwrite the existing file If the list of variables on the WRITE statement includes some which are not series the variables are written to the file one at a time with one record per variable Symmetric triangular and diagonal matrices are written in compressed storage mode unless the FULL option is specified You are responsible for making sure your format allows for enough data points if you use the FORMAT format string option Subscripted matrices are treated like scalars In labelled or FREE format series scalars and matrices missing values are shown as the character period and a warning is printed FREE format values are written to all significant digits and each observation starts on a new line Minimal spacing is used to produce as compact output as possible 449 Commands A single WRITE command will write either a single matrix or several series to a spreadsheet file If the file specified already exists it is o
224. described in Pantula et al 1994 If j is the number of lags which minimizes AIC Akaike Information Criterion then L MIN j 2 MAXLAG is used Note that if MAXLAG you will probably want to increase MAXLAG AIC2 apparently avoids size distortions for the WS and DF tests AIC2 is also used here for EG tests No direct rule is used for PP tests yet Instead the optimal lag from the DF test is also used for PP if the DF test is performed at the same time A plain AIC rule is used for JOH i e L j this is not a very good rule for JOH you may prefer to run the unconstrained VAR and test its residuals for serial correlation These rules are a topic of current research so as more useful rules are found they will be added as options For example other possible rules are 1 testing for remaining serial correlation in the residuals 2 testing the significance of F statistics for the last lag of differenced lagged variable s 3 SBIC 27 4 automatic bandwidth selection for PP not very encouraging in the current literature The current RULE AIC uses a fixed number of observations for comparing regressions with different numbers of lags Each regression is a column in the output table If MINLAG MAXLAG then the RULE is used to select an optimal number of lags j A final column in the table is created for this labelled Opt j If jis less than MAXLAG then the regression for this column is computed with the maximum availabl
225. ditions which are minimized in the metric of an estimate of their expected covariance This estimate is computed using 3SLS estimates of the parameters unless the NOLSQSTART option has been specified If the HETERO and NMA options are not used the estimation method coincides with conventional 3515 estimation The usual GMM estimator that allows for heteroskedasticity in addition to cross equation correlation is the default If you wish to use different instruments for each equation supply each set in a series of separate lists separated by to the INST option see the examples Alternatively you can use the MASK option to drop instruments selectively The GMM estimator prints the Sargan or J of overidentifying restrictions if the degrees of freedom are greater than zero the model is exactly identified if they are equal to zero If you want to nest overidentifying tests of a series of models be sure to specify the NOLSQSTART option so that the variance covariance matrix of the OC s will be held fixed across the tests otherwise the chi squared for the difference between two nested models may have the wrong sign Method 170 GMM See Hansen 1982 for most of the details If the equations are nonlinear the iteration method is the usual LSQ method with analytical derivatives a variant of the method of scoring Output The following results are printed and stored variable type PHI scalar GMMOVID scalar GMMOVID scala
226. ds FEI NOFEI specifies that models with additive individual fixed effects are to be estimated The panel structure must have been defined previously with the FREQ PANEL command The equations specified must be linear in the parameters this will be checked and variables Nonlinear options These options control the iteration methods and printing They are explained in the NONLINEAR section of this manual Some of the common options are MAXIT MAXSQZ PRINT NOPRINT SILENT NOSILENT The legal choices for HITER are G Gauss the default B BHHH and D DFP numeric derivatives HCOV B BHHH is the default method for calculating standard errors and D is legal when is used Examples To obtain full information maximum likelihood estimates of the illustrative model from the User s Manual use the following statement FIML ENDOG GNP CONS I R LP GNPID CONSEQ INVEQ INTRSTEQ PRICEQ To do Box Cox regression properly use FIML to include the Jacobian term in the likelihood note unnormalized equation FRML EQ1 Y LAM 1 LAM A B X PARAMABLAM1 FIML ENDOG Y EQ1 Klein I model see Calzolari and Panattoni for correct results FORM VARPREF C CONS CX C P P 1 W FORM VARPREFzI INV I C P P 1 K 1 FORM VARPREF W WAGES W1 C E E 1 TM IDENT WAGE W W1 W2 IDENT BALANCE CX l G TX W P IDENT PPROD E P TX W1 FIML ENDOG CX I W1 W P E CONS INV WAGES WAGE BALANC
227. e The basic PROBIT statement is like the OLSQ statement first list the dependent variable and then the independent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the current sample PROBIT will print a warning message and will drop those observations PROBIT also checks for complete or quasi complete sample separation by one of the right hand side variables such models are not identified The list of independent variables on the PROBIT command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See the PDL section for a description of how to specify such variables The dependent variable need not be a strictly zero one variable Positive values are treated as one and zero or negative values are treated as zero 345
228. e Ortega and Rheinboldt pp 217 220 PRNRES NOPRNRES prints the residuals when solution is complete for each time period All of the residuals will be small enough to satisfy the convergence criterion PRNDAT NOPRNDAT prints the starting values for the endogenous variables at each time period PRNSIM NOPRNSIM prints a table of the solved values of the endogenous variables at the completion of the simulation STATIC NOSTATIC specifies static simulation Actual values of lagged endogenous variables are used not earlier solved values tagname specifies that the solved values of all endogenous variables should be stored as series with names created by adding tagname to the variable names tagname should be a single character or perhaps two to avoid creating excessively long names Names larger than the allowed length of a TSP name will be truncated TAG NONE stores under the original endogenous names The default values of the options are DYNAM CONV2 001 MAXPRT 5 METHOD GAUSS and no TAG The print options are off except PRNSIM so a one line summary of each iteration and a table of results will be printed Example This example shows how to set up the well known Klein Model for simulation At the end of this simulation the solved variables are stored under the names CXS IS etc LIST Z C P 1 K 1 1 W2 TX 2SLS INST Z CX C P P 1 W FORM CONS 2SLS INST Z 1C P P 1 K 1 FORM INV 2SL
229. e case of YCAT 4 Reference Verbeek Marno A Guide to Modern Econometrics Wiley 2000 pp 189 193 203 Commands KALMAN Output Options Examples References KALMAN estimates linear models using the Kalman filter method It can handle fairly general State Space models but it is typically used to estimate regression type models where the coefficients follow a random process over time KALMAN BPRIORz prior vector 5 lt of coefficients in transition equation EMEAS ETRANS PRINT SILENT SMOOTH VBPRIORz variance of prior VVIEASz variance factor in measurement equation VTRANS lt variance factor transition equation XFIXEDz X matrix for measurement equation list of dependent variables list of independent variables gt Usage The Kalman filter model consists of two parts the state space form measurement equation y X amp t amp amp N 0 0 nA transition equation 7 7 0 Q mel MM md me menm initial conditions The matrices Q are assumed to be known they each default to the identity matrix if they are not supplied by the user in the KALMAN options list Note that they are not allowed to vary over time but this constraint can be easily relaxed by running KALMAN within a loop over the sample The NOEMEAS and NOETRANS options are used to zero the variances of the meas
230. e desired the serial correlation coefficient would be specified as a RHO option in the parentheses also SMPL 1 20 OLSQ CONS C GNP COPY COEF B FORCST COEF B CONSP C GNPNEW This example shows the use of the DYNAM option to obtain a dynamic forecast with a lagged endogenous variable on the right hand side of the equation I 1 SMPL 1 20 OLSQ I I 1 GNP 153 Commands SMPL 21 30 FORCST PRINT DYNAM IFIT This example shows both uses of the FORCST procedure the dynamic option was specified on the first statement so that an extrapolation of the estimating equation will be produced The statements which save R and B were required only for the second forecast which is over the same sample as the original estimation but uses a different right hand side variable GNPNEW SMPL 1 20 I I 1 GNP SET RHO B COEF SMPL 21 30 FORCST PRINT DYNAM IFIT Other TSP program statements may occur here SMPL 1 20 FORCST COEF B RHO R I2FIT 1 GNPNEW The example below shows how to compute a set of forecasts for a sales variable based on three slightly different GNP projections SMPL 65 1 82 4 AR1 SALES C SALES 1 GNP GNP 1 SMPL 83 1 86 4 GENR GNP1 GNPFCST GENR GNP2 1 1 GNPFCST GENR GNP3 0 9 GNPFCST DOT123 FORCST DYNAM PRINT SALESF C SALES 1 GNP GNP 1 ENDDOT Reference Pindyck Robert S and Daniel L Rubinfeld Econometric Models and
231. e end while an IDENT does not If the FRML is not normalized it is treated as being equal to the implied disturbance The distinction is useful only in FIML where identities may be necessary to complete the Jacobian to ensure that it is a square matrix An equation defined by FRML statement can contain numbers parameters constants and series The equation can always be computed at any point by use of GENR see GENR for the form of the statement When it is computed the parameters and constants are supplied with their current values before computation and the equation is computed for all the values of the series in the current sample Examples 165 Commands These are the equations which are used to estimate the illustrative model by three stage least squares in LSQ FRML CONSEQ CONS A B GNP FRML INVEQ LAMBDA I 1 ALPHA GNP DELTA R FRML INTRSTEQ R D FLOG GNP LP LM FRML PRICEQ LP LP 1 PSI LP 1 LP 2 PHI LOG GNP TREND TIME P0 In these equations the dependent variables are CONS R and LP The other series are GNP LM and TIME Note the use of lagged series in the equation also The other variables A B LAMBDA ALPHA DELTA D F PSI PHI TREND and PO are parameters and constants The first FRML could be written in unnormalized implicit form as its residual FRML CONSEI CONS A B GNP When the dependent variable is actually an expression the unnormalized form
232. e examples below VALUE NOVALUE is used to fix the values of any PARAMs in a list of existing FRMLs Then they can be printed to show the values with the other variables or stored in a databank for later use with SIML or SOLVE without having to worry about storing the associated PARAMs VALUE is the default if there is only one argument or if there are several arguments and they are all existing FRMLs This option was formerly called CONST VARPR a prefix for the coefficient names to be used in conjunction with the variable names The default is to use a COEFPREF instead Special coefficient names are used for C intercept lags and leads 0 zero is used in place of C positive numbers are appended for lags and L plus positive numbers are appended for leads Examples 1 This example causes an equation named CONSEQ to be printed and stored OLSQ CONS C GNP GNP 1 FORM PRINT CONSEQ prints FRML CONSEQ CONS 123 0 0 9 GNP 0 05 GNP 1 assuming that the coefficient estimates from the OLSQ were 123 9 05 2 This example saves the same equation as in example 1 with names instead of values OLSQ CONS C GNP GNP 1 FORM COEFPREF B CONSEQ is equivalent to the following commands FRML CONSEQ CONS B0 B1 GNP B2 GNP 1 PARAM BO 123 B1 9 B2 05 LIST RNMSF BO B2 157 Commands 3 This example uses a VARPREF instead of a COEFPREF 01 50 CONS GNP GNP 1 FORM VARPREF B CONS
233. e interactive session via keyboard or input file EXEC first line numbers lt last line numbers Note that in DOS Win TSP it is easier to use the up down arrow keys to select and rerun a single command Usage EXEC varies slightly depending upon the mode in which you are currently operating In COLLECT mode EXEC is used to execute the range of lines just collected and return control to interactive mode This is considered the standard exit from collect mode the alternative is to suppress execution with the EXIT command The whole range will be executed so line number arguments will be ignored if you supply them Lines in the range may be EDITed or DELETEd if necessary before EXECuting them In interactive mode up to two arguments may be supplied with the EXEC command If no argument is supplied the previous command is re executed If only the first line number is supplied a single line is executed and if two line numbers are given their inclusive range is executed In any case these lines may be commands that have been previously executed and edited or commands that were entered but suppressed with an EXIT command in collect mode or at the end of an INPUT file 141 Commands EXIT Interactive EXIT terminates the current operating mode of the program EXIT Usage If used in interactive mode the interactive session will be terminated returning control to the operating system this is the same as typing STOP o
234. e model estimated is X 8 4 PANEL computes means for each variable by individual These are used directly in the BETWEEN regression WITHIN subtracts the individual means from each variable and runs a regression on this transformed data any variables which are constant over time for every individual are not identified VARCOMP does a transformation similar to WITHIN 1 SQRT theta times the mean is subtracted from each variable including the constant term where theta is given by Var u Var u TVar a T does not have to be the same for each individual The small and large sample formulas used for the variance components are variance small sample large sample within SSRW NOB NX NI SSRW NOB total SSRT NOB NX 1 used between VTOT VWITH SSRT SSRW NOB If the small sample formula produces a non positive variance PANEL switches over to the large sample formulas automatically The large sample formulas are asymptotically correct if T is becomes large relative to NI not usually the case otherwise they will be biased Note that if thetaz1 this corresponds to a zero between variance and VARCOMP will produce the same estimates as TOTAL If theta 0 this corresponds to a zero within variance and VARCOMP will be the same as WITHIN For each F test described under Output a P value and an alternative critical value are printed The critical value has a size which becomes smaller as the number of obser
235. e observations so the test results may vary slightly from the original column for j SEAS NOSEAS include seasonal dummy variables such as Q1 Q3 This option implies the CONST option The SEAS option is available for FREQ Q 2 or higher The seasonal coefficients are only printed if PRINT is on SEAST NOSEAST include seasonal trend variables like Q1 TREND Q2 TREND Q3 TREND This option implies the TREND option SEASTSQ NOSEASTS include seasonal squared trend variables SEASTSQ implies TSQ These are fairly simplistic trend terms which may not be enough to adequately model a time series that has a change in its intercept and or trend at some point in the sample See Perron 1989 for more details The special exogenous trend variables arguments described above may provide a crude examination of more detailed trends If these variables are supplied all series are regressed on these trend variables and the residuals from this regression are used in all tests instead of the original values of the series No corrections to the P values of the tests are made however other than in the degrees of freedom for calculating the t statistics and s2 TREND NOTREND include a time trend in the tests 91 Commands TSQ NOTSOQ include a squared time trend in the tests Output options PRINT NOPRINT prints the options and adds the coefficients and t statistics of the augmenting lagged difference variables to the main tables TERSE NOTER
236. e on Cointegration with Applications to the Demand for Money Oxford Bulletin of Economics and Statistics 1990 p 169 210 Mackinnon James G Approximate Asymptotic Distribution Functions for Unit Root and Cointegration Tests Journal of Business and Economic Statistics April 1994 pp 167 176 Osterwald Lenum Michael Practitioners Corner A Note with Quantiles for the Asymptotic Distribution of the Maximum Likelihood Cointegration Rank Test Statistic Oxford Bulletin of Economics and Statistics 1992 p 461 471 Pantula Sastry G Graciela Gonzalez Farias and Wayne A Fuller A Comparison of Unit Root Test Criteria Journal of Business and Economic Statistics October 1994 pp 449 459 93 Commands Perron Pierre The Great Crash The Oil Price Shock and the Unit Root Hypothesis Econometrica November 1989 pp 1361 1401 Phillips P C B Time Series Regression with a Unit Root Econometrica 1987 pp 277 301 Phillips P C B and Pierre Perron Testing for a Unit Root in Time Series Regression Biometrika 1988 pp 335 346 94 COLLECT Interactive COLLECT Interactive COLLECT allows a group of TSP statements to be entered before execution of the sequence COLLECT Usage This is particularly useful for introducing flow of control structures not directly allowed in interactive mode such as PROCs IF THEN ELSE sequences or DO or DOT loops You will continue to be prompted for new lin
237. e under the same name in this case more than one series can be converted at a time The second form takes the old series on the right hand side of the equal sign converts it to the new FREQ and stores it under the new series name only one series may be converted in this way on each command Depending on the type of series you are converting you can specify various methods of aggregating or disaggregating the series if you do not say anything and you are converting to a lower frequency CONVERT will average all the observations within an interval to produce a value for that interval The default for converting to a higher frequency is to duplicate the value for all observations in the new interval unless the INTERPOL or SUM option is used Output CONVERT produces no printed output It stores one converted series in data storage Options AVERAGE forms the new series by averaging all the observations within a period This is the default 99 Commands FIRST forms the new series by choosing the first observation in the period MID forms the new series by choosing the middle observation in the period If the number of observations per period is even CONVERT uses the one before the halfway point LAST forms the new series by choosing the last observation in the period SUM forms the new series by summing all the observations in the period If converting from a lower frequency to a higher the new values are divided by the conv
238. ecial exogenous trend variables gt 441 Commands UNMAKE Examples UNMAKE takes a matrix and splits it column by column into a set of series The matrix must have a number of rows equal to the number of observations in the current sample and a number of columns equal to the number of series whose names are supplied Similarly UNMAKE will split a vector into a set of scalars if the number of scalars is equal to the length of the vector UNMAKE is the reverse of MMAKE which makes a matrix from series or a vector from scalars UNMAKE matrix list of series or UNMAKE lt vector gt lt list of scalars gt Usage UNMAKE s first argument is the name of the matrix to be broken up the second a list of the names of the series where the columns of the matrix will be put The number of series is limited only by the maximum size of the argument list usually about 100 or more and the space available in data storage for the new matrix The series named will be replaced if they already exist if the NOREPL option is being used observations outside the current sample will be deleted The new series have the frequency and starting observation of the current sample The matrix to be unmade may be of any type it will be expanded before UNMAKE is executed An exception to this rule is in the case of a diagonal matrix if the length of the current sample is equal to the length of the diagonal and only one series name is suppli
239. ed the diagonal of the matrix will be stored in this series UNMAKE can also be used to split a vector into scalars this is useful for rearranging coefficient vectors and setting up starting value vectors This is easier than specifying several SET statements or a tricky DOT loop Output UNMAKE produces no printed output A set of series or scalars are stored in data storage Examples 442 UNMAKE If the current sample is SMPL 1 5 and there is a 5 by 2 matrix X with the following elements N the command UNMAKE X X1 X2 results in the following two series being stored Any submatrix within a matrix can also be obtained by unmaking the matrix under one SMPL and remaking it under another For example given an 8 by 8 covariance matrix a new covariance matrix that contains only the elements corresponding to the 3rd 4th and 5th variables can be extracted by the following commands SMPL 18 UNMAKE VAR COL1 COL8 SMPL 35 MMAKE VAR35 COL3 COL5 Here is an example of UNMAKE with a vector OLSQ Y C X1 X6 UNMAKE COEF B0 B6 See also the example under MMAKE for manipulating a vector of starting values for parameters 443 Commands UPDATE Interactive Update allows you to specify portions of a series or list of series whose observations you wish to modify It is for interactive use only UPDATE lt list of series gt Usage UPDATE is a special form of the ENTER command The onl
240. ed at the initial estimate which is specified by WNAME This option can be used to obtain estimates that are invariant to which equation is dropped in a shares model like translog HETERO NOHETERO causes heteroskedastic consistent standard errors to be used See the GMM NMA command for autocorrelation consistent standard errors Same as the old ROBUST option or HCOV R WNAME z the name of a matrix to be used as the starting value of the covariance matrix of the residuals 431 Commands WNAME OWN specifies that the initial covariance matrix of the residuals is to be obtained from the residuals corresponding to the initial parameter values If neither form of WNAME is used the initial covariance matrix is an identity matrix Nonlinear options control the iteration methods and printing They are explained in the NONLINEAR section of this manual Some of the common options are MAXIT MAXSQZ PRINT NOPRINT SILENT NOSILENT The only legal choice for HITER is G Gauss HCOV G is the default method for calculating standard errors R Robust is the only other valid option Example Klein model FORM VARPREF C_ CONS CX C P P 1 W FORM VARPREF L_ INV I C P P 1 K 1 FORM VARPREF W WAGES W1 C E E 1 TM 3SLS INST C TM W2 G TX P 1 K 1 E 1 CONS INV WAGES References Amemiya Takeshi The Maximum Likelihood and the Nonlinear Three Stage Least Squares Estimator in the General Nonlinear Simultaneo
241. efault is on SETSMPL if no SMPL has been specified yet in the program and off otherwise This option does not apply to matrix reading TYPE GENERAL or SYMETRIC or TRIANG or DIAG or CONSTANT specifies the type of the matrix which is to be READed GENERAL the default may be used for any rectangular or square matrix SYMETRIC implies that the matrix is equal to its transpose only the lower triangle will be stored internally to save space TRIANG implies that the matrix is triangular has zeroes above the diagonal Although a lower triangular matrix is read its transpose is stored since the TSP matrix procedures expect upper triangular matrices DIAG means a matrix all of whose off diagonal elements are zero Only the diagonal is stored and it is expanded before use CONSTANT means a scalar or scalars are to be READed and none of the matrix options will apply If no type is specified a warning is printed and the matrix is assumed to be general 368 READ UNIT an integer number usually between 1 and 4 or 8 and 99 which is the Fortran input output unit number of an external file from which the variables listed will be READ Usually just FILE is used but UNIT could be used to avoid typing in a long filename for several READ commands from the same file Examples A simple READ of one series in free format SMPL19 READ IMPT 100 106 107 120 110 116 123 133 137 This example reads formatted data from the TSP input file SM
242. efficient estimates Standard errors of coefficient estimates T statistics on coefficients Values of the gradient at convergence Variance covariance of estimated coefficients Ljung Box modified Q statistics P values for Q statistics 1 if AR polynomial is stationary 1 if MA polynomial is invertible Fitted values of the dependent variable Residuals actual fitted values of the dependent variable Real parts of the AR roots Imaginary parts of the AR roots BJEST vector NAR Moduli of the AR roots MARTRE vector NMA Real parts of the MA roots MARTIM vector NMA Imaginary parts of the MA roots QMARTMO vector NMA Moduli of the MA roots SARRTRE vector NSAR Real parts of the seasonal AR roots SARRTIM vector NSAR Imaginary parts of the seasonal AR roots SARRTMO vector NSAR Moduli of the seasonal AR roots SMARTRE vector NSMA Real parts of the seasonal roots SMARTIM vector NSMA Imaginary parts of the seasonal MA roots SMARTMO vector NSMA Moduli of the seasonal MA roots Method The method used by BJEST for the default method NOEXACTML to estimate the parameters is essentially the one described by Box and Jenkins in their book It uses a conventional nonlinear least squares algorithm with numerical derivatives The major difference between the estimation of time series models and estimation in the traditional nonlinear least squares way relates to the use of back forecasted residuals The likelihood fu
243. eing changed inadvertently Output Normally FORM produces no printed output If the PRINT option is on the options and the output equation are printed A TSP equation is stored in data storage under the name supplied by the user If the PARAM option is on scalar PARAMeters will be stored and a list of the parameter names is stored under RNMSF Options COEFPR a prefix for the coefficient names The default is B plus the equation name Specify this carefully to avoid overwriting existing variables For example avoid creating AO A11 and then 11 13 NAR number of autocorrelation terms The default is zero unless the previous estimation was AR1 PARAM NOPARAM specifies whether the coefficients will be names or constant values PARAM is the default if there are three or more arguments or if COEFPREF or VARPREF have been specified PRINT NOPRINT specifies whether the options and output equation will be printed RESIDUAL NORESIDUAL specifies whether the FRML should be unnormalized have no dependent variable 156 FORM RHOPREF a prefix for the autocorrelation coefficient names The default is RHO plus the equation name if NAR 1 Otherwise it is PHI plus the equation name plus the order of the autoregressive lag SUM NOSUM specifies that the equation is to be formed as a sum of terms specified in a list This is useful when constructing a log likelihood for ML when you do not know how many terms you will need See th
244. eletes observations with missing values for one or more variables before estimation When the SMPL frequency is type PANEL AR1 can also obtain estimates for a panel data model with fixed FEI or random REI effects AR1 estimates can also handle plain time series which have irregular spacing gaps in the SMPL Output The AR1 procedure produces output that is similar to OLSQ and LSQ including the iteration log The equation title and the chosen objective function method of estimation are printed first If the PRINT option is on this is followed by the list of option values the starting values for all the coefficients iteration output for all coefficients and any grid values for rho and the objective function 41 Commands The usual regression output follows as described under the OLSQ command The regression statistics are computed from the fitted values and residuals described below If the objective function chosen was GLS a common factor test is included in the regression statistics This test is a likelihood ratio test of the restrictions implied by AR 1 compared to an unconstrained OLS model that includes the lagged dependent as well as the current and lagged right hand side variables The test is not well defined when the model is estimated by ML due to the special treatment of the first observation As in OLSQ and INST a table of coefficient estimates is printed RHO is always the last coefficient in the table
245. ellner and Theil 1962 For the nonlinear three stage least squares estimator see Amemiya 1977 and Jorgenson and Laffont 1975 The method of estimation in TSP is described more fully in Berndt Hall Hall and Hausman 1975 also available at this website Options COVU residual covariance matrix same as the old WNAME option below DEBUG NODEBUG specifies whether detailed computations of the model and its derivatives are to be printed out at every iteration This option produces extremely voluminous output and is not recommended for use except by systems programmers maintaining TSP FEI NOFEI specifies that models with additive individual fixed effects are to be estimated The panel structure must have been defined previously with the FREQ PANEL command The equations specified must be linear in the parameters this will be checked and variables Individual specific means will be removed from both variables and instruments INST ist of instrumental variables The list of instrumental variables supplied is used for all the equations See the INST section of this manual and the references for further information on the choice of instruments ITERU NOITERU specifies iteration on the COVU matrix provides the same function as the old MAXITW option MAXITW the number of iterations to be performed on the parameters of the residual covariance matrix estimate If MAXITW is zero the covariance matrix of the residuals is held fix
246. em if you just want a simple plot Any observations with missing data are excluded from the plot Output PLOT prints a title followed by the names of all the series being plotted and the characters used to plot them If there are lines drawn on the plot a message giving the locations of the lines is printed The plot itself is labelled at its four corners with the horizontal minima and maxima the axes are labelled at several points if the HEADER option was specified and the mean is marked with an M if a line at the mean was requested The ID series labels the left hand side of the vertical axis and the values of the first series are on the right hand side if the VALUES option was specified If more than one series is being plotted any points which are superimposed are plotted with the number of series which have that value instead of the plotting character The plotting characters of the duplicate series are shown on the right hand side of the plot 328 PLOT PLOT uses the LIMPRN option to decide how wide to make the plot so you have some control over the format by use of the OPTIONS command Options BAND STANDARD or seriesname specifies the name of a series which is used as the width of a band to be printed around the observations of the first series to be plotted usually this series is a set of computed standard errors The keyword STANDARD will cause the standard deviation of the series to be used as the band The default
247. en David microTSP Version 6 5 User s Manual Quantitative Micro Software 1989 143 Commands FIML Output Options Examples References FIML invokes the Full Information Maximum Likelihood procedure This procedure obtains maximum likelihood estimates of nonlinear simultaneous equations model The model should have N equations some of which may be identities in N endogenous variables and may be written in implicit form equations without left hand side variables FIML is an asymptotically efficient estimator for simultaneous models with normally distributed errors It is the only known efficient estimator for models that are nonlinear in their parameters For further details on this estimator and the method of estimation see the references and the TSP User s Guide See the LIML command for details on doing nonlinear Limited Information Maximum Likelihood single equation ML with the FIML command FIML ENDOG lt list of endogenous variables FEI nonlinear options list of equation names Usage FIML in its simplest form is invoked by listing the endogenous variables of the model in the options and following the options by the equation names FIML ENDOG Y1 Y2 YN EQ1 EQ2 EQN The equations must be previously defined by FRML and IDENT statements Those which are IDENTs are assumed to hold exactly in the data and will not contribute to the covariance matrix of the disturbances If the IDENTs can be
248. en New York 1990 336 POISSON POISSON Output Options Examples References POISSON obtains estimates of the Poisson model where the dependent variable takes on nonnegative integer count values and its expectation is an exponential linear function of the independent variables In the Poisson model the variance of the dependent variable equals its mean which is rarely the case in practice More general models where the variance is larger than the mean are the Negative Binomial types 1 and 2 See the NEGBIN command The Poisson command is POISSON nonlinear options lt dependent variable gt lt list of independent variables gt Usage The basic POISSON statement is like the OLSQ statement first list the dependent variable and then the independent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the current sample POISSON will print a warning message and will drop those observations POISSON also checks that the observations on the dependent va
249. ent list Usually 2000 or more and the space available in data storage for the new matrix Only the observations in the series which are within the current sample will be used so you can select data for the matrix very conveniently by changing the SMPL If you use matrices instead of series the SMPL is ignored The matrix made by MMAKE will always be a general matrix and when using series its dimensions are normally NROW NOB the number of observations by NCOL the number of input series To change its type use MFORM If the second and following arguments to MMAKE are scalars CONSTs PARAMs or numbers a vector will be created instead of a matrix This is useful for creating vectors like START or COEF from estimated parameters or scalars created by UNMAKE from previous COEF vectors Output 265 Commands MMAKE produces no printed output A single matrix or vector is stored in data storage Options VERT NOVERT stacks the input series or matrices vertically instead of horizontally Examples If the current sample is SMPL 1 4 and there are two data series X1 1 2 3 4 and X2 9 8 7 6 the following command X X1 X2 results in the matrix X WD whereas MMAKE VERT XV X1 X2 would create the 8 x 1 vector XV O90 0A 0l Here is an example of adding a coefficient to a LOGIT estimation while retaining the starting values for the other coefficients from the previous
250. ent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space but of course the number is limited by the number of data observations you have available The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the current sample TOBIT will print a warning message and will drop those observations The list of independent variables on the TOBIT command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when you are entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See PDL for a description of how to specify such variables The dependent variable need not be a strictly zero positive variable Negative values are treated as zero The standard Tobit model involves truncation of the dependent variable below zero Models with upper and or lower truncation can be estimated by using the UPPER and or LOWER option s See the Examples for more details Output 434 TOBIT The output of TOBIT begins with a
251. er detection Least Trimmed Squares LTS also has about the same properties The LMS estimator occasionally produces a non unique estimate of the coefficient vector b TSP reports the number of non unique subsets in this case An extremely rough estimate of the variance covariance of the estimated coefficients is computed with an OLS type formula V B d X X where the estimate of o squared is computed after deleting the largest residuals This estimate is not asymptotically normal and is likely to be an underestimate so it should not be used for serious hypothesis testing It all depends on how the outliers are generated by the underlying model The code used in TSP was adapted from Rousseeuw s Progress program obtainable from his web page referenced below Options ALL NOALL uses all possible observation subsets see Method even if there are over one million of them LTS NOLTS computes the Least Trimmed Squares estimates which minimize the sum of squared residuals from the smallest up to the median instead of LMS which minimizes just the squared median residual Usually the LTS and LMS estimates are fairly close to each other MOST NOMOST uses all possible subsets unless the number of subsets is one million or more in which case random subsets are used PRINT NOPRINT prints better subsets as progress towards a minimum is made and the final outliers SILENT NOSILENT suppresses all output 231 Commands
252. er exogenous variables in the model Do not forget to include a constant if there is one in the model Weights are not supported at present SILENT NOSILENT suppresses all output TERSE NOTERSE prints minimal output estimated coefficients and a summary statistic 222 LIML Examples This example estimates the consumption function for the illustrative model in the TSP User s Manual using the constant trend government expenditures G and the log of the money supply LM as instruments LIML INST C G TIME LM CONS C GNP Other examples LIML INST C LOGR LOGR 1 LOGR 2 LOGR 3 LOGP C LOGP 1 LOGR LIML FULLER 1 INST C LOGR LOGR 1 LOGR 2 LOGR 3 LOGP C LOGP 1 LOGR References Anderson T W Kunitomo Naoto and Morimune Kimio Comparing Single Equation Estimators in a Simultaneous Equation System Technical Report No 1 Econometric Workshop Stanford University January 1985 Cragg J G and S G Donald Testing Identifiability and Specification in Instrumental Variable Models Econometric Theory 9 1993 pp 222 240 Fuller Wayne A Some Properties of a Modification of the Limited Information Estimator Econometrica 45 939 953 Hansen C J A Hausman and W Newey Weak Instruments Many Instruments and Microeconometric Practice MIT Cambridge Mass working paper 2004 Judge et al The Theory and Practice of Econometrics John Wiley amp Sons New York 1981
253. er pages will use the new title until it is changed If you do not want to start a new page use the NOPAGE option This option makes use of the PAGE command unnecessary If OPTIONS CRT is in effect the title is centered and printed underlined with stars on the current page or screen Output Under OPTIONS HARDCOPY a new page is started with the new title at the top Under OPTIONS CRT the title is centered and printed underlined with equals signs Options PAGE NOPAGE tells whether a new page should be started when OPTIONS HARDCOPY is in effect When OPTIONS CRT the default is being used PAGE has no effect Examples TITLE Results for small firms SELECT SALES lt 10 DOTX YZ TITLE prints variable names from DOT as titles OLSQ CR LAD CR ENDDOT 433 Commands TOBIT Output Options Examples References TOBIT obtains estimates of the linear Tobit model where the dependent variable is either zero or positive The method used is maximum likelihood under the assumption of homoskedastic normal disturbances For non normal censored regression see LAD TOBIT LOWER lt lower limit MILLS lt name for output inverse Mills ratio UPPER lt upper limit WEIGHT lt weighting series nonlinear options dependent variable list of independent variables Usage The basic TOBIT statement is like the PROBIT or OLSQ statements first list the dependent variable and then the independ
254. ergence VCOV matrix vars Variance covariance of estimated vars coefficients FIT series obs Fitted values of dependent variable RES series obs Residuals actual fitted values of dependent variable If the regression includes a PDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling Method POISSON uses analytic first and second derivatives to obtain maximum likelihood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly Zeros are used for starting parameter values except for the constant term START can be used to provide different starting values see NONLINEAR in this help system As in other regression procedures in TSP estimation is done using a generalized inverse in the case of multicollinearity of the independent variables The overdispersion test is a Lagrange multiplier test based on regressing the difference between the estimated variance and the dependent variable on the fitted value The statistic is T the number of observations times the R squared from the following regression y By U See Cameron and Trivedi 1998 78 equation 3 39 The exponential mean function is used in the Poisson model That is if X are the independent variables and B are their coefficients E Y X exp X
255. ernment expenditures G and the log of the money supply LM as instruments INST INST C G TIME LM CONS C GNP 198 INST Using population as weights the following example regresses the fraction of young people living alone on various other demographic characteristics across states Population is proportional to the inverse of the variance of per capita figures INST WEIGHT POP INST C URBAN CATHOLIC SERVEMP SOUTH YOUNG C RSALE URBAN CATHOLIC Other examples of the INST 2SLS command INST ROBUSTSE INST C LOGR LOGR 1 LOGR 2 LOGR 3 LOGP C LOGP 1 LOGR 2SLS INST C LM 1 LM 3 TBILL C RATE 4 12 FAR Note that the constant C must always be named explicitly as an instrument if itis needed References Judge et al The Theory and Practice of Econometrics John Wiley amp Sons New York 1981 pp 531 533 Keane Michael P and David E Runkle On the Estimation of Panel Data Models with Serial Correlation When Instruments not strictly Exogenous Journal of Business and Economic Statistics 10 1992 pp 1 29 Maddala G S Econometrics McGraw Hill Book Company New York 1977 Chapter 11 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 Chapter 5 Theil Henri Principles of Econometrics John Wiley amp Sons New York 1971 Chapter 9 White Halbert Instrumental Variables Regression with
256. ersion ratio e g by four if converting from annual to quarterly Only one of the above options should be included INTERPOL NOINTERP specifies linear interpolation when converting to a higher frequency the default is to duplicate observations rather than interpolate INTERPOL is used in conjunction with one of the other options to determine the placement of the peak value series computes SUM default or AVERAGE from an old series stores it in a new series using a MAP of pointers This is helpful for aggregating grouped data such as industries states or individuals with panel data The rows of the map correspond to the rows in the old series The values in the map correspond to the rows of the new series Zero values mean the observation is not mapped to the new series The SMPL option is the default when MAP is used and it puts the map and old series under the control of the current SMPL while the new output series will be FREQ N starting at observation 1 If NOSMPL is used the traditional CONVERT method is used where the old and map series are used at their maximum defined lengths and the current FREQ SMPL are only used to determine the FREQ and starting point of the new series With NOSMPL the map and old series must be defined over exactly the same set of observations The map cannot contain any missing values or contain only zeroes The old series can contain missing values they will result in missing values in
257. es using a perpetual inventory and a constant rate of depreciation If is gross investment K is the capital stock and d is the rate of depreciation then CAPITL computes K t 1 d K t 1 1 1 for the NOEND option K t 1 d K t 1 I t for the END option It starts from a capital stock benchmark specified by an observation If the benchmark is in the middle of the sample CAPITL also applies the backward version of the formula K t K t 1 I t 1 d for the NOEND option K t K t 1 t 1 1 d for the END option to compute values of the capital stock in periods before the benchmark Note that the depreciation rate d is stated as a rate applicable to the frequency of the data For example for quarterly data the depreciation rate must be a quarterly rate one fourth of the annual rate It is possible to compute a depreciation series dK t and a net investment series I t dK t CAPITL may be used in any application where a moving average with geometrically declining weights needs to be calculated Further by setting the depreciation rate to zero it will simply cumulate a series CAPITL BENCHOBS obs id BENCHVAL scalar END investment series gt lt depreciation rate gt lt capital stock series gt Usage The only arguments required for the CAPITL statement are an investment series a depreciation rate and the name to be given to the derived capital stock series In this case the value of the
258. es and then computing each equation in the block in turn Each equation must depend only on endogenous variables which are input to the block or computed previously within the block 413 Commands Obviously a recursive block is easily solved on the conditional values of the input endogenous variables A simultaneous block on the other hand does not have a triangular Jacobian and thus requires either the inversion of the Jacobian or some sort of iterative technique The two methods available in SOLVE are the Gauss Seidel and the Fletcher Powell The first is simply a generalization of the method for computing recursive blocks the equations are computed in order each endogenous variable being evaluated in turn Then the new values of the endogenous variables are used to start the process over again until convergence no change in the variables is achieved This process works best on mildly simultaneous and fairly linear models it does not guarantee convergence The criterion function for SOLVE is the sum of squared deviations of the equations At the solution all deviations will be zero Away from the solution the deviations are computed by substituting the current values of the variables into the equations and evaluating them This sum of squared deviations is the objective function printed out by SOLVE at each iteration The Fletcher Powell algorithm solves simultaneous blocks by minimizing this criterion function with respect to the e
259. es in collect mode until you terminate the COLLECT mode Termination of the mode may be accomplished in two ways 1 The EXEC command requests automatic execution of the entire range of commands just collected Unlike its usage in interactive mode the EXEC command takes no arguments when used to terminate collect mode 2 The EXIT command may be used to return to interactive mode without executing the collected lines The lines are stored and EXEC may be used on them interactively at any time While entering commands in collect mode you may find you want to fix a typo or modify something you ve entered before requesting execution For this reason there are two commands that are always executed immediately even while in collect mode these are EDIT and DELETE Of course if you really mess things up or simply change your mind you can always abandon the effort with EXIT HINT If there are PROCs or sequences of commands you find you use frequently whether in collect mode or not you may want to store them in external files to save having to type them more than once Any group of TSP commands may be read from disk with the INPUT command INPUT is functionally equivalent to collect mode the only difference being that the commands are read from a file instead of your terminal See the section on INPUT for more details on how to use this feature WARNING You will confuse TSP if you begin a control structure in collect mode
260. es name or a list of series names On the plot the series will be differentiated by colors or the style of the lines used to plot them Parameters to control the appearance and printing of the plot may be included in parentheses following the word PLOT The graph will be displayed on the screen if a DEVICE is specified a prompt is also displayed which instructs you to type P if you wish to print the graph If you type anything else the graph will not be printed this is useful if you decide you do not like its appearance after you have seen the Screen If there are observations with missing data or if the FREQ PANEL option is set there will be breaks in the plotted lines Output A high resolution plot is produced on the screen with time the observation index on the horizontal axis and the series on the vertical access If there is more than one series they are differentiated by means of colors see below for other options Options General DOS Win only only 331 Commands For convenience the DEVICE FILE AND HEIGHT options of PLOT that you set are retained in the next PLOT s or GRAPH s until they are overridden explicitly They may also be set in a LOGIN TSP file by not specifying any series to plot DASH NODASH specifies whether the lines for different series on the screen are to be distinguished by using different dash patterns of which there are seven The default is no dashes just color on the sc
261. es name gt lt obs id gt lt value gt lt series name gt lt obs id gt value 1 Usage After NORMAL list the name of the series the observation identifier of the base observation and the value to be assigned to the base observation The normalized series will replace the original series for those observations in the current sample The observation identifier must include the period if the frequency is neither NONE nor ANNUAL It is written in the form YYYY PP or YY PP where YYYY or YY is the year and PP is the period You may normalize as many series with the same statement as you wish just include three arguments series name observation identifier and value for each one Output NORMAL produces no printed output One or more series are replaced in data storage Examples This example normalizes the CPI to have the value 100 in 1975 NORMAL CPI 75 100 This is equivalent to the following statements SET BASE CPI 75 CPI 100 CPI BASE This example normalizes a set of quarterly price series so they have the value 1 in the first quarter of 1972 NORMAL P1 72 1 1 P2 72 1 1 P3 72 1 1 289 Commands NOSUPRES NOSUPRES turns off the suppression of output for the selected results from procedures NOSUPRES list of result names Usage The arguments to NOSUPRES can be any of the output names beginning with described in this help system The printing of the output associated with th
262. es the title The title is a string of up to 60 characters enclosed in quotes There can be no quotes or imbedded in the title Output NAME causes a jobname to be printed in the upper right hand corner of each page of TSP output If a title is included on the command the title is also printed at the top of every page in columns 21 through 80 If the terminal CRT option is on no paging of TSP output is done and no titles are printed unless requested by a PAGE command Examples NAME KARLMARX NAME ILLUS44 ILLUSTRATIVE MODEL FOR TSP VERSION 4 4 NAME KLEINLSQ 3SLS ESTIMATES OF KLEIN MODEL I 273 Commands NEGBIN Output Options Example References NEGBIN obtains estimates of the Negative Binomial model where the dependent variable takes on only nonnegative integer count values and its expectation is an exponential linear function of the independent variables In the Negative Binomial model the variance of the dependent variable is larger than the mean in contrast to the Poisson model where the variance equals the mean see the POISSON procedure NEGBIN MODEL 1 or 2 nonlinear options dependent variable list of independent variables Usage The basic NEGBIN statement is like the OLSQ statement first list the dependent variable and then the independent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of
263. es with the program currently it takes the generalized inverse of a symmetric matrix and stores the corresponding eigenvalues and eigenvectors Note that this feature requires that you have the source code version of TSP usually not available for PC or Mac but sometimes available on request for unix Output The USER procedure will produce whatever output you print or store in data storage Reference Cummins Clint and Hall Bronwyn H Time Series Processor Version 4 0 Programmer s Manual TSP International Stanford CA 1985 445 Commands VAR Output Options Example Reference VAR performs vector autoregressions which are a set of unrestricted reduced form linear regressions with lags of the dependent variables on the right hand side Impulse response functions dynamic simulations based on the estimated coefficients forecast error decompositions and block exogeneity tests are also performed VAR NHORIZ length of impulse response NLAGS lt number of lags in VAR SHOCK ALL or CHOL or STDDEV or UNIT or matrix name gt SILENT TERSE list of dependent variables list of exogenous variables Usage First list the dependent variables and specify the number of lags desired in the options list If there are any exogenous variables give their names after a If you want to have intercept terms in the regressions usually recommended include the special variable C or CONSTANT in the list of exo
264. ese names will be suppressed throughout the TSP program unless a NOSUPRES or REGOPT command with these codes is issued The output results are still stored in memory and may be accessed See also SUPRES and REGOPT 290 OLSQ OLSQ Output Options Examples References OLSQ is the basic regression procedure in TSP It obtains ordinary least squares estimates of the coefficients of a regression of the dependent variable on a set of independent variables Options allow you to obtain weighted least squares estimates to correct for heteroskedasticity or to obtain standard errors which are robust in the presence of heteroskedasticity of the disturbances see the GMM command for robustness to autocorrelation OLSQ HCOMEGA BLOCK or DIAGONAL HCTYPE lt robust SE type gt HI NORM or UNNORM ROBUSTSE SILENT TERSE WEIGHT lt name of weighting variable WTYPEz weight type dependent variable list of independent variables Usage In the basic OLSQ statement you list the dependent variable and then the independent variables in the equation To have an intercept term in the regression include the special variable C or CONSTANT in the list of independent variables The number of independent variables is limited by the overall limits on the number of arguments per statement and the amount of working space obviously it is also limited by the number of data observations available The observations over which the regression is co
265. evious command and re executes it Interactive REVIEW lists range of TSP command lines Interactive UPDATE replaces observations in a series nteractive 27 Command summary Obsolete Commands Old command INPROD x y z INV a ai deta MADD x y z MATRAN x xt MDIV x yz MEDIV x y z MEMULT x y z MMULT x yz MSQUARE x y MSUB x y z NOSUPRES x SUPRES x VGVMLT x yZ YFACT x y YINV a ai YQUAD 2 28 Replacement command MAT 2 MAT ai a MAT deta DET a MAT z x y MAT xt MAT z x y MAT z x y 2 2 2 MAT xx 2 REGOPT NOPRINT MAT z y MAT y CHOL x MAT ai YINV a z x y x Cross Reference Pointers Cross Reference Pointers Command synonyms For LOAD COVA CORR NOSUPRES PRINT SUPRES 2515 UNIT See READ MSD MSD REGOPT PRINT WRITE REGOPT NOPRINT INST COINT Non command entries For Functions FORMAT NONLINEAR PDL Examples See BASIC RULES READ and WRITE convergence options used in LSQ ML FIML BJEST etc Polynomial Distributed Lags used in OLSQ AR1 etc For examples of See AR1 CDF DOT ELSE EQSUB FRML GENR IF INST LIST LOGIT LSQ MAT FORM PDL RANDOM ELSE LIST SORT GOTO ANALYZ ML FORM LIST DIFFER GOTO PDL DUMMY LENGTH MMAKE ANALYZ EQSUB MMAKE ORTHON 29 Command summary OLSQ PR
266. ew matrix to have the same dimensions TRANS NOTRANS specifies whether the input variable is to be transposed to produce a matrix of the type and dimensions specified If the input variable is a matrix the output matrix will have the number of rows equal to the number of columns of input and the number of columns equal to the number of rows of input MFORM TRANS is the same as the MATRAN command TYPEZGENERAL or SYMMETRIC or TRIANG or DIAG specifies the type of the new matrix GENERAL the default may be used for any rectangular or square matrix SYMMETRIC implies that the matrix is equal to its transpose only the lower triangle will be stored internally to save space TRIANG implies that the matrix is upper triangular has zeroes below the diagonal DIAG means a matrix whose off diagonal elements are zero Only the diagonal is stored and it is expanded before use A series may be used as input to MFORM DIAG Examples Suppose that we have a nine observation series called X containing the values 10 20 30 40 50 60 70 80 90 Each of the following MFORM commands will yield a different result 255 Commands SMPL 1 9 MFORM TYPE GENERAL NROW 3 NCOL 3 X yields X 20 50 80 30 60 90 SMPL 1 6 MFORM TYPE GENERAL NROW 2 NCOL 3 X XMAT yields XMAT 20 40 60 MFORM TRANS XMATT XMAT yields XMATT MFORM TYPE GENERAL NROW 4 NCOL 3 X gives an error message the resulting matrix is too big
267. ew York 1977 pp 237 242 Ortega J M and W C Rheinboldt Iterative Solution of Nonlinear Equations in Several Variables Academic Press New York 1970 Chapter 7 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 Chapters 10 11 12 Saaty T L and J Bram Nonlinear Mathematics McGraw Hill Book Co New York 1964 Theil Henri Principles of Econmetrics John Wiley amp Sons Inc New York 1971 432 439 407 Commands SMPL Output Examples SMPL is used to define the observations of the data which will be used in the following TSP procedures The SMPL vector is a set of pairs of observation identifiers which define the range s of observations which are in the current sample SMPL lt beginning obs id gt lt ending obs id gt lt beginning obs id gt ending obs 1 or SMPL SMPL_vector_name SMPL is often used in conjunction with FREQ which sets the frequency of the data Usage The sample of observations may be specified in four ways 1 SMPL statement listing the beginning and ending pairs of observations to be used 2 ASMPL statement containing the name of a variable that contains a SMPL vector of pairs of observations ids 3 A SMPLIF statement with an expression which is true for the observations to include in the sample see the SMPLIF section 4 A SELECT statement same as SMPLIF ex
268. f instruments and the variables in the equation LIML determines the list of endogenous variables included exogenous variables and excluded exogenous variables by comparing the instrument list with the variables in the equation If there are no endogenous variables OLSQ is used and a warning is printed If the right hand side equation is exactly identified number of endogenous variables equals number of excluded exogenous variables LIML is equivalent to 2SLS so 2SLS is used and a warning is printed If the equation is under identified an error message is printed just like 2SLS Output The output of LIML begins with an equation title the name of the dependent variable and the lists of endogenous included exogenous and excluded exogenous variables The LIML eigenvalue and an F test of the overidentifying restrictions are printed If FULLER is used the FULLER constant and the computed K class value are also printed The log of the LIML eigenvalue plus a constant is equal to the log of the likelihood function This is followed by various statistics on goodness of fit the sum of squared residuals the standard error of the regression the R squared and the Durbin Watson statistic for autocorrelation of the residuals 219 Commands The estimated concentration parameter mu squared and Cragg Donald F statistic CDF are also shown and stored When there is a single right hand side endogenous variable CDF is an F statistic which
269. f squared residuals the standard error of the regression the R squared the Durbin Watson statistic for autocorrelation of the residuals and an F statistic for the hypothesis that all coefficients in the regression except the constant are zero The objective function e P Z e stored as PHI is the length of the residual vector projected onto the space of the instruments This is analogous to the sum of squared residuals in OLSQ it can be used to construct a pseudo F test of nested models Note that it is zero for exactly identified models if they have full rank A test of overidentifying restrictions FOVERID is also printed when then number of instruments is greater than the number of right hand side variables It is given by PHI S2 m k All the above statistics are based on the structural residuals that is residuals computed with the actual values of the right hand side endogenous variables in the model rather than the fitted values from a hypothetical first stage regression Following this is a table of right hand side variable names estimated coefficients standard errors and associated t statistics If the variance covariance matrix has not been suppressed see the SUPRES command it is printed after this table Finally if the RESID and PLOTS options are on a table and plot of the actual and fitted values of the dependent variable and the residuals is printed INST also stores most of these results in data storage fo
270. f the density to the distribution function is also known as the inverse Mills ratio This is used in the derivatives and with the MILLS option Options LOWERc the value below which the dependent variable is not observed The default is zero MILLS the name of a series used to store the inverse Mills ratio series evaluated at the estimated parameters The default is MILLS WEIGHT the name of a weighting series The weights are applied directly to the likelihood function and no normalization is performed UPPER the value above which the dependent variable is not observed The default is no limit Nonlinear options see NONLINEAR Examples Standard Tobit model with truncation below zero TOBIT CAR C INCOME RURAL MSTAT Truncation below two TOBIT LOWER 2 CAR C INCOME RURAL MSTAT Truncation above ten 436 TOBIT TOBIT UPPER 10 CAR C INCOME RURAL MSTAT References Amemiya Takeshi Tobit Models A Survey Journal of Econometrics 24 December 1981 pp 3 61 Greene William H On the Asymptotic Bias of the Ordinary Least Squares Estimator of the Tobit Model Econometrica 49 March 1981 pp 505 513 Maddala G S Limited dependent and Qualitative Variables Econometrics Cambridge University Press New York 1983 pp 151 155 Tobin James Estimation of Relationships for Limited Dependent Variables Econometrica 31 1958 pp 24 36 437 Commands TREND Options Examples TREND
271. fects AR1 estimator the estimated fixed effects are stored in the matrix COEFAI and in the series QAI Options 43 Commands FAIR NOFAIR specifies whether the lagged dependent and independent variables are to be added to the instrument list automatically when doing instrumental variable estimation combined with serial correlation correction FEI NOFEI specifies that an AR 1 model with panel fixed effects is to be estimated by means of maximum likelihood or GLS if OBJFN GLS is specified INST ist of instrumental variables This list should include any exogenous variables that are in the equation such as the constant or time trend as well as any other variables you wish to use as instruments After any instruments are added by the FAIR option there must be at least as many instruments as the number of estimated coefficients the number of independent variables in the equation plus one for rho OBJFN GLS is implied the actual objective function is E PZ E where the Es are rho transformed residuals See the Examples for a way to reproduce the AR1 estimates with FORM and LSQ Fair once argued that the lagged dependent and independent variables must be in the instrument list to obtain consistent estimates when doing instrumental variable estimation with a serial correlation correction TSP adds them automatically if you use the FAIR option the default if you want to specify a different list of instruments you must suppress
272. ferences SAMPSEL estimates a generalized Tobit or sample selection model where both the regression and the latent variable which predicts selection are linear regression functions of the exogenous variables Either a censored regression variables not observed for non selected observations or truncated all variables not observed model may be estimated SAMPSEL MILLS lt series gt nonlinear options lt probit dep var gt lt probit indep vars gt lt regression dep var gt lt regression indep vars gt Usage The model estimated by SAMPSEL is the Tobit type II model described by Amemiya or the censored regression model with a stochastic threshold described by Maddala see the references It can be written as X B e if y gt 0 regression equation 0 otherwise X U selection equation 1 1 i e i and u i are assumed to be joint normally distributed O po 1 In the output the standard deviation of the regression equation is denoted SIGMA and the correlation coefficient is denoted RHO The variance of the selection probit equation is normalized to one without loss of generality To use the procedure to estimate this model supply the name of a zero one variable which tells whether the observation was observed or not 1 gt 0 as the probit dependent variable the regressors X1 as the probit independent variables y 2 as the regression dependent variable and X2 as the regre
273. for the Multinomial Logit Model Journal of Econometrics 34 1987 63 82 McFadden Daniel S Quantal Choice Analysis A Survey Annals of Economic and Social Measurement 5 1976 pp 363 390 McFadden Daniel S Conditional Logit Analysis of Qualitative Choice Behavior in Zarembka P ed Frontiers in Econometrics Academic Press 1973 Nerlove Marc and S James Press Univariate and Multivariate Loglinear and Logistic Models Rand Report No R 1306 EDA NIH 1973 Train Kenneth Quantitative Choice Analysis The MIT Press Cambridge MA 1986 241 Commands LSQ Output Options Examples References LSQ is used to obtain least squares or minimum distance estimates of one or more linear or nonlinear equations These estimates may optionally be instrumental variables estimates LSQ can compute nonlinear least squares nonlinear two stage least squares or instrumental variables nonlinear multivariate regression with cross equation constraints seemingly unrelated regression and nonlinear three stage least squares The equations for any of these estimators may be linear or nonlinear in the variables and parameters and there may be arbitrary cross equation constraints LSQ can also be invoked with the SUR 3SLS THSLS and GMM commands LSQ DEBUG FEI HETERO INST lt list of instrumental variables gt ITERU COVU OWN or lt name of residual covariance matrix gt nonlinear options lt list of equation names gt
274. forms unit root and cointegration tests These may be useful for choosing between trend stationary and difference stationary specifications for variables in time series regressions See Davidson and MacKinnon 1993 for an introduction and comprehensive exposition of these concepts Most of these tests can be done with OLSQ and CDF commands on a few simple lagged and differenced variables so the main function of COINT is to summarize the key regression results concisely and to automate the selection of the optimal number of lags COINT ALL ALLORD COINT CONST DF EG FINITE JOH MAXLAG lt number of lags gt MINLAG lt number of lags PRINT RULE AIC2 SEAS SEAST SEASTSQ SILENT TERSE TREND TSQ UNIT WS lt list of variables gt lt list of special exogenous trend variables gt or UNIT ALL NOCOINT CONST DF FINITE MAXLAG lt number of lags gt MINLAGz number of lags PP PRINT RULE AIC2 SEAS SEAST SEASTSQ SILENT TERSE TREND TSQ UNIT WS lt list of variables gt lt list of special exogenous trend variables gt 1 Usage List the variables to be tested and specify the types of tests maximum number of augmenting lags and standard constant trend variables in the options list The default performs augmented Weighted Symmetric Dickey Fuller and Engle Granger tests with 0 to 10 lags If there are any special exogenous trend variables such as split sample dummies or trends give their names a
275. fter a see the explanation under General Options below The observations over which the test regressions are computed are determined by the current sample If any observations have missing values within the current sample COINT drops the missing observations and prints a warning message or an error message if a discontinuous sample would result Output The default output prints a table of results plus coefficients for each test on each variable Two summary tables are also printed with just the optimal lag lengths one for all the unit root tests and one for the Engle Granger tests if ALLORD is used 84 COINT For each variable all the specified types of unit root tests are performed A table is printed for each type of test Usually the rows of this table are the estimated root alpha test statistic P value coefficients of trend variables number of observations the log likelihood AIC and the standard error squared The columns of this table are the number of augmenting lags A summary table is also printed which includes just the test statistics and P values for the optimal lag length COINT usually stores most of these results in data storage for later use except when a 3 dimensional matrix would be required The summary tables are always stored If more than one variable is being tested TABWS TABDF and TABPP are not stored If ALLORD is used TABEG is not stored but EG EG and EGLAG will be stored In
276. g for interactive use or the Windows interface programs Givewin or Through the Looking Glass for editing batch files EDIT line numbers Usage Only one argument is allowed with this command which must be a TSP line number If this argument is omitted EDIT will prompt you for modifications to the previous line This command is generally used for correcting typos or moditying lists of series etc before requesting re execution of a procedure RETRY is identical to EDIT except that execution of the modified command is automatic upon exit Also EDIT is treated as a special case in COLLECT mode its execution is not suppressed while RETRY is not The editor will first echo the command you wish to modify then issue the prompt gt gt Responses to the prompt must consist of an editing command followed by appropriate arguments Any unique abbreviation for an editing command will suffice including a single letter Complete arguments must be entered even if you only wish to replace a single character Only one modification per edit prompt may be made Prompting will continue until an EXIT command or carriage return is given in response Arguments TSP breaks up everything you type except data into a string of arguments Each series or variable name is an argument as is each operator Statement terminators semi colons and item separators commas and blanks are not considered arguments This example has 10 arguments
277. ge for later use using names but with B T V W REI and REIT appended to distinguish between the different estimators For example COEFW is the within coefficients RESW are the within residuals SSRV is the sum of squared residuals from VARCOMP RESB is a matrix In the table below vars is equal to the number of right hand side variables plus one for the constant for the T B W and V estimators For the REI estimator vars includes the estimate of RHO the within group correlation and SIGMA the total standard error For the REIT estimator vars includes the estimate of RHO the estimate of RHO T the within time correlation and SIGMA the total standard error variable type length description LHV list 1 Name of the dependent variable SSRT I B W V REI REIT scalar 1 Sum of squared residuals SSRI BYID etc S2T B W V REI REIT scalar 1 Variance of residuals S2B BETWEEN etc ST B W V REI REIT scalar 1 Standard error of the regression YMEANT B W V REI REIT scalar 1 Mean of the dependent variable SDEVT B W V REI REIT scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations APUI scalar 1 Ahrens Pincus unbalancedness in SPUT scalar 1 Ahrens Pincus unbalancedness in t RSQT B W V scalar 1 R squared ARSQT B W V scalar 1 Adjusted R squared NCOEFT B W V REI REIT scalar 1 Number of coefficients NCIDT B W V REI REIT scalar 1 Number of identified coefficient
278. ged twice with the part of the series which is orthogonal to the first lag The options on the BUIDENT command allow you to specify the differencing you want performed on the series whether there is a seasonal component and the output you wish to see plotted or printed You can also increase and decrease the number of autocorrelations which are computed 68 BJIDENT If you want to analyze the log or other transformation of your original series as suggested by Box and Jenkins or Nelson transform the series using a GENR statement before submitting it to BJIDENT Output The output of BJIDENT begins with the plots of the raw and differenced series if the PLOT option was requested These plots are high resolution graphics plots in TSP Givewin and low resolution in other versions Next a table of the autocorrelations is printed with their standard errors and the Ljung Box portmanteau test or modified Q statistic These modified Q statistics are distributed independently as chi squared random variables with degrees of freedom equal to the number of autocorrelations The null hypothesis is that all the autocorrelations to that order are zero Following this table the partial autocorrelations are printed for each series If PLOTAC has been specified BJIDENT then plots the autocorrelation and partial autocorrelation functions of the series and its differences with standard error bands The inverse autocorrelations are printed if requested
279. generates a series with a linear growth trend The trend may be repeated under the control of several options TREND FREQ PERIOD lt value gt PSTART lt value gt START lt initial value gt STEP lt increment gt lt series gt or TREND series lt initial value lt increment gt Usage TREND followed by a series name makes a simple time trend variable and stores it under that name This variable is equal to one in the beginning observation of the current sample and increases by one in every period The starting value of the series may be changed by including the second argument and the increment may also be changed by including the third argument If you want to change the increment but not the starting value use the STEP option Since the TREND command creates a series under control of SMPL you must be careful to specify a SMPL which covers the whole period in which you are interested so you don t have missing values or strange jumps due to gaps in the sample Output TREND produces no printed output A single series is stored Options FREQ NOFREQ causes the trend to be restarted every time there is a new year This option is valid only when FREQ Q or FREQ M has been specified PERIOD z specifies the number of observations after which the trend starts repeating itself For example PERIOD 4 would have the same effect as FREQ Q PSTART specifies the starting period for the first observation in the sam
280. genous variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space obviously the number is limited by the number of data observations available The observations over which the regression is computed are determined by the current sample If any observations have missing values within the current sample VAR drops the missing observations and prints an error message Output VAR output begins with an equation title and the names of the dependent variables followed by the log likelihood value and a table of the regression coefficients Various statistics on goodness of fit are printed for each equation the sum of squared residuals the standard error of the regression the R squared and the Durbin Watson statistic for autocorrelation of the residuals biased unless NLAGS 0 because of the presence of lagged dependent variables Block exogeneity tests Granger causality tests are computed as F tests to see if lagged values of other dependent variables are significant in each equation Next are the impulse response functions see Method and variance decompositions The variance decomposition is for residual variances only it does not include sampling error in the regression coefficients VAR stores most of these results in data storage for later use Here are the results available after a VAR command variable type length description 4
281. he same command except that UNIT has a default of NOCOINT UNIT may also seem more appropriate if you are just testing one variable General Options these apply to both unit root and cointegration tests CONST NOCONST include a constant term in the tests NOCONST implies NOTREND FINITE NOFINITE computes finite sample vs asymptotic P values when possible augmented Dickey Fuller and Engle Granger tests See the discussion and references under Method in the CDF entry of this manual These are distinguished by different labels P valFin or P valAsy normally the finite sample P values will be slightly larger than the asymptotic ones MINLAG smallest number of augmenting lags default 0 This is denoted as L in the equations under Method Note that 7 is the total AR order of the process generating y So L is the number of lags in excess of the first one For the Phillips Perron test L is the number of lags in the autocorrelation robust covariance matrix MAXLAG maximum number of augmenting lags The default is min 10 2 NOB 1 3 which is 10 for 100 observations or below the factor 2 was chosen arbitrarily to ensure this If the number of observations in the current sample NOB is extremely small MAXLAG and MINLAG will be reduced automatically 90 COINT RULE AIC2 or specifies the rule used to choose an optimal lag length number of augmenting lags assuming MINLAG lt MAXLAG The default is 2 which is
282. he DEBUG statement in your TSP program directly before the command s for which you want additional output The DEBUG switch will remain on until a NODBUG statement is encountered For DEBUG output during the compile phase of the program see the ASMBUG statement Output DEBUG should be used with care since it normally produces a great deal of printed output For example every fetch and store to data storage VPUT VGET is printed which facilitates following the progress of the program but can be voluminous In the nonlinear procedures all the input data and the results of the differentiation of the equations will be printed For any estimation procedure the matrices involved in the computations will be printed at every iteration Every input command line will be printed before and after it has been interpreted for dates dots and lists Several of the non estimation procedures also produce special debug output when this switch is on Example DEBUG LSQ PRINT EQNAME NODBUG This example causes debug output to be produced during the execution of a nonlinear least squares estimation 110 DELETE DELETE DELETE removes TSP variables from the symbol table DELETE list of variables Usage Useful for long or complex TSP programs DELETE deletes variables from the symbol table Data for the variables is not actually deleted from memory until an automatic compression occurs when space is needed to store a new variab
283. he default values of the options are DYNAM METHOD NEWTON and TAGznothing The print options are all off except PRNSIM which prints only a one line summary of each iteration and a table of results Examples This example solves the Illustrative Model described in the User s Manual SIML PRNDAT TAGzS ENDOG GNP CONS I R LP CONSEQ INVEQ INTRSTEQ GNPID PRICEQ After this model has been solved the solved series are stored under the names GNPS CONSS IS etc The input data and the solved series are printed The next example shows how to set up the well known Klein Model for simulation The equations are linear so they are formed after the corresponding instrumental variables estimation LIST Z C P 1 K 1 1 TM W2 G TX 2SLS INST Z CX C P P 1 W FORM CONS 2SLS INST Z 1C P P 1 K 1 FORM INV 2SLS INST Z W1 C E E 1 FORM WAGES IDENT WAGE W W1 W2 IDENT BALANCE CX I G TX W P IDENT PPROD E P TX W1 IDENT CAPSTK K K 1 I SIML 5 ENDOG CX I W1 W E P K CONS INV WAGES WAGE BALANCE PPROD CAPSTK 406 SIML This model solves for CX consumption investment W1 wages in the private sector W total wage bill E production of the private sector P profits and K capital stock using TM time W2 government wage bill TX taxes and G government expenditures as exogenous variables References Maddala G S Econometrics McGraw Hill Book Company N
284. he file FOO TSP which would contain Re create TSP Databank FOO TLB END OUT FOO PARAM B 3 14 FREQ Q SMPL 60 1 85 4 LOAD X 1234 LOAD Y 11 22 33 44 106 DBDEL Databank DBDEL Databank Options Example DBDEL deletes one or more variables from a TSP databank DBDEL COMPRESS lt filename gt lt list of variables gt Usage Supply the filename and one or more variable names to delete This is a way of getting rid of variables which were put in the databank by mistake or which are no longer needed It can also be used to crudely rename variables in a databank if the variables are first copied to the new names and then the old names are deleted An alternative for small databanks would be to use the DBCOPY command and then use a text editor on the TSP file to delete rename or generally edit variables and DOCumentation Options COMPRESS NOCOMPRESS compresses the databank after the variables are deleted This is the same operation as the DBCOMP command the space formerly used by the variables is recovered for use by other files Example Suppose that you have a databank called FOO TLB which contains two time series X and Y and a parameter B The command DBDEL FOO X would delete the series X from the databank 107 Commands DBLIST Databank Output Options Example DBLIST shows the contents of TSP databanks with information on variable names types frequencies and sample cove
285. he horizontal axis are different PRINT NOPRINT tells whether the histogram is to be printed or just stored WIDTH the width of the bars the default is 2 printer lines Examples HIST X produces a plot with the vertical axis containing ten cells running from the minimum value of X to the maximum value of X and the horizontal axis showing the number of observations of X which take on values within each of the cells HIST MAXz100 MINZO NBINSz40 PERCENT Y1 Y2 produces two histograms each with 40 cells which have a width equal to 2 5 The percent of observations of Y1 or Y2 which fall in each cell are shown Suppose the variable REASON takes on the values 0 1 2 and 3 The command HIST DISCRETE REASON will produce a histogram with four cells containing the number of observations taking on each one of the four values of REASON 184 HIST graphics version HIST graphics version Output Options Examples HIST produces histograms bar charts or frequency distributions of series It is convenient for obtaining a rough picture of the univariate distribution of your data The graphics version options NOPRINT PREVIEW is the default for TSP Givewin HIST CDF DENSITY DISCRETE HIST MAX lt maximum X value gt lt value gt NBINS lt number of bins gt NORMAL PRINT PREVIEW STANDARD TITLE lt text string gt lt list of series gt Usage Follow HIST with the names of one
286. he use of the NOPRINT command to suppress printing of the entire data section NOPRINT FREQ SMPL 2041 LOAD YEAR YT K1 PW1W2Y 1920 39 8 2 7 4 6 47 1 180 1 12 7 28 8 2 2 43 7 1921 41 9 2 6 6 48 3 182 8 12 4 25 5 2 7 40 6 isses more data input END 287 Commands NOREPL Examples NOREPL turns off the replacement mode option REPL REPL specifies that series are to be updated rather than completely replaced when the current sample under which they are being computed does not cover the complete series NOREPL Usage The REPLace mode is the default use NOREPL if you do not want previously existing series updated when they are modified For example use the REPL mode to create a series element by element with a DO loop and SET statements Later on if you want to recreate the same series with a slightly different sample the REPL mode could cause the old observations to be mixed in with the new so you might want to use NOREPL just for safety Examples REPL SMPL 110 GENRD 0 SMPL 1120 GENRDz 1 NOREPL This creates a series named D which is zero for observations 1 through 10 and one for observations 11 through 20 288 NORMAL NORMAL Examples NORMAL normalizes a series so that a chosen observation has a predetermined value It accomplishes this by dividing all the observations of the series by the ratio of the supplied value to the chosen observation s value NORMAL lt seri
287. hieved 0 otherwise NCOEF scalar 1 Number of parameters to be estimated NCID scalar 1 Number of identified parameters RNMS list params Names of right hand side variables COEF vector params Coefficient estimates GRAD vector params Gradient of the log likelihood at convergence SES vector params Standard errors 262 ML T vector params T statistics VCOV matrix params Variance covariance of estimated params coefficients See the NONLINEAR section for the alternative names of VCOV stored when the HCOV option is used Method The method used is a standard gradient method explained in somewhat more detail in Chapter 9 of the User s Manual Briefly at each iteration a new parameter vector is computed by moving in the direction specified by the gradient of the likelihood uphill weighting this gradient by an approximation to the matrix of second derivatives at that point in order to adjust for the curvature Convergence is declared when the changes in the parameters are all small where small is defined by the TOL option The ML procedure normally uses analytic first and second derivatives for the FRML method The function can also be maximized numerically HITER F or HITER D See NONLINEAR for more information on the options HCOV N gives standard errors based on analytic second derivatives MLPROC uses HCOV U numeric second derivatives to compute the standard errors Options Standard nonlinear opt
288. hings do not improve the estimation will be terminated When you encounter a problem like this you can often get around it by estimating only a few parameters at a time to obtain better starting values Use CONST to fix the others at reasonable values For details on the estimation method see the Berndt Hall Hall and Hausman article Options COVU residual covariance matrix same as the old WNAME option below DEBUG NODEBUG specifies whether detailed computations of the model and its derivatives are to be printed out at every iteration This option produces extremely voluminous output and is not recommended for use except by systems programmers maintaining TSP FEI NOFEI specifies that models with additive individual fixed effects are to be estimated The panel structure must have been defined previously with the FREQ PANEL command The equations specified must be linear in the parameters this will be checked and variables Individual specific means will be removed from both variables and instruments 247 Commands INST ist of instrumental variables If this option is included the LSQ estimator becomes nonlinear two stage least squares or nonlinear IV if there is one equation and nonlinear three stage least squares if there is more than one equation The list of instrumental variables supplied is used for all the equations See the INST section of this manual and the references for further information on the choice
289. hts applied to each component in expressing each input series as a function of the components PRIN stores the correlation matrix of the input variables under the name CORR and stores the components themselves under the names P1 P2 P3 etc as time series If you supply a different prefix for the names from P PRIN will use that when making the names Method 341 Commands TSP standardizes the variables subtracts their means and divides by their standard deviations before computing the principal components The resulting components have the following properties 1 They have mean zero standard deviation unity and are orthogonal 2 The correlation coefficients between a principal component vector and the set of original variables are identical to that component s loading factor 3 The sum of squared loading factors equals the characteristic root In some other principal component packages the sum of squared factor loadings equals unity this is a matter of arbitrary scaling In calculating the principal components the factor loadings are divided by the characteristic root to obtain a principal component with standard deviation of unity Other programs treat the scaling differently 4 The fraction of the variance of the original variables explained by a principal component is its characteristic root divided by the number of variables The TSP commands below obtain the same results as the PRIN X Y Z command CORR
290. iance is a quadratic function of the mean E Y X alpha E Y X 2 In both cases the parameter alpha is restricted to be non negative alpha 0 corresponds to the Poisson Nonlinear options see the NONLINEAR entry Examples 276 NEGBIN Negative Binomial 2 regression of patents on lags of log R amp D science sector dummy and firm size NEGBIN PATENTS C LRND LRND 1 LRND 2 DSCI SIZE Negative Binomial 1 regression for the same model NEGBIN MODEL 1 PATENTS LRND 1 LRND 2 DSCI SIZE References Cameron A Colin and Pravid K Trivedi Regression Analysis of Count Data Cambridge University Press New York 1998 Cameron A Colin and Pravin K Trivedi Count Models for Financial Data Maddala and Rao eds Handbook of Statistics Volume 14 Statistical Methods in Finance Elsevier North Holland 1995 Hausman Jerry A Bronwyn H Hall and Zvi Griliches Econometric Models for Count Data with an Application to the Patents R amp D Relationship Econometrica 52 1984 pp 908 938 277 Commands Nonlinear Options Options References These options are common to all of TSP s nonlinear estimation and SAMPSEL SIML SOLVE ML etc See Chapter 10 of the User s Guide for further information DROPMISS EPSMIN lt value gt GRADCHEC GRADIENT method HCOVzmethod HESSCHEC HITERzmethod MAXIT lt of iterations MAXSQZ lt of squeezes NHERMITE lt value gt
291. ible for seeing that there are enough instruments for these variables after the constraints implied by PDL are imposed If the PDL variable is exogenous the most complete list would include all the lags of the variable over which the PDL is defined These variables may be highly collinear but will cause no problems due to TSP s use of the generalized inverse when computing regressions a subset of the variables which contains all the information will be used If the PDL variable is endogenous you must include enough instruments to satisfy the order condition for identification The number required can be computed as inst lags less order of polynomial less number of endpoint constraints Method Let y be the dependent variable X be the T by k matrix of independent variables and Z be the T by m matrix of instrumental variables the included and excluded exogenous variables for two stage least squares Then the formulas used to compute the coefficients their standard errors and the objective function are the following P Z Z Zzyz b X RX X Ry X Zz Z Z Z X x z z z pz y e y Xb s e eK T k V b S X P X e P e 195 Commands The structural residuals e are used to compute all the usual goodness of fit statistics Output The output of INST begins with an equation title the name of the dependent variable and the list of instruments This is followed by statistics on goodness of fit the sum o
292. ic triangular and diagonal matrices If a matrix is symmetric only the lower triangle needs to be read i e elements 1 1 2 1 2 2 3 1 8 2 3 3 and so forth The FULL option specifies whether the full matrix is being specified or only the lower triangle If the matrix is triangular you need only specify the transpose of the upper triangular portion 1 1 1 2 2 2 1 3 2 3 3 3 and so forth If the matrix is diagonal only the diagonal needs to be given 1 1 2 2 3 3 It will be filled out with zeroes when used Stata files TSP reads stata version 2 7 dta files The variables have the same names as they have in stata although string text variables are generally not supported Spreadsheet files TSP can read series and matrices directly to or from spreadsheet files The following files are supported by TSP spreadsheet version filename TSP extensions support Lotus 123 Symphony 1 2 Wks wk1 wrk Read and 363 Commands wr1 write Lotus 123 3 Lotus 123 J Japanese 1 2 Wj1 wj2 wk2 Read and wt2 write Microsoft Excel 2 xIs Read and write Microsoft Excel 3 4 xis Read Microsoft Excel xlw Read 5 7 8 97 98 2000 2002 Quattro Pro wq1 Read and write Note that TSP generally writes the oldest file formats which are always readable by more recent spreadsheet releases Spreadsheet files should be in the format of the following example SM
293. ics GRAPH command GRAPH 4 DASH DEVICE lt name of printer FlLE lt name of file gt HEIGHT lt height of letters gt HIRES LANDSCAP or PORTRAIT LINE ORIGIN PAIR PREVIEW SORT SURFACE SYMBOL TITLE text string to be used as WIDTH XMIN lt x axis minimum XMAXz x axis maximum YMINz y axis minimum YMAXz y axis maximum x axis series list of y axis series or x series 1 y series 1 x series 2 y series 2 for the PAIR option or x series y series z series for the SURFACE option Usage GRAPH has several forms depending on the options specified If GRAPH is followed by a pair of series names a scatter plot will be produced using the x axis for the first series and the y axis for the second If there are more than two series names and NOPAIR the default is specified the others will also be plotted on the y axis and their points will be connected with a line This makes it easy to plot fitted and actual values from a regression using the default options see the example below If the PAIR option is specified there must be an even number of series names and each pair will be plotted versus each other the first series on the x axis and the second on the y axis TSP Givewin automatically displays graphs in different windows by default If there are gaps in the SMPL observations with missing data or the FREQ PANEL option is set any lines being graphed w
294. ilename option FORMAT LOTUS or EXCEL is optional If the filename contains one of the extensions listed earlier WKS XLS etc TSP checks the first few bytes of the file to confirm that it is one of the spreadsheet versions listed above Conversely if FORMAT LOTUS or EXCEL is specified but the filename does not contain a extension then WKS or XLS is appended to the filename To read a matrix bypassing column names and dates use the TYPE GEN option TYPE CONSTANT is not supported for spreadsheet files series will be defined instead NCOL NROWS IFULL UNIT etc options are ignored but SETSMPL is supported If no series names are supplied on the READ command TSP looks for column names in the file and creates series with those names If you supply series names TSP attempts to match them to column names in the file If the file does not have column names you must supply a READ argument for each data column If you are unsure of the file s contents check it with your spreadsheet or read it as a matrix If you think the file has column names but you don t know what they are try supplying a dummy name which won t be matched TSP will print an error message listing the column names in the file If you are reading a matrix using TYPE GEN as mentioned above TSP will create a matrix named LOTMAT unless you supply an argument matrix name to READ If for some reason your series are in rows instead of columns y
295. iles more readable Of course TSP automatically provides the file BKUP TSP as an undocumented session photo 310 PAGE PAGE PAGE is used to force the paging of the printed output of TSP It only operates when the option HARDCOPY is in effect PAGE Usage Include a PAGE statement any place in your TSP program where you wish to force the printed output to start on a new page This might be at the beginning of a series of regressions or when doing a large simulation or just to force a new title to be printed Normally TSP pages the output as well as it can so that the major procedures start on a new page but it is not always possible to format exactly as the user would want without wasting a large amount of blank paper the PAGE statement lets you control the printing to a certain extent Output PAGE produces no printed output itself but causes paging to the next page to occur and the title line to be printed 311 Commands PANEL Output Options Examples References PANEL obtains estimates of linear regression models for panel data several observations or time periods for each individual Total between groups within groups and variance components may be obtained In addition one and two way random effects models may be estimated by maximum likelinood The data may be unbalanced different number of observations per individual PANEL can also compute means by group and perform F tests between groups PANEL ALL
296. ill contain breaks In the DOS Win or MAC version of TSP the graph will be displayed on the screen if a DEVICE is specified a prompt is also displayed which instructs you to type P if you wish to print the graph If you type anything else the graph will not be printed this is useful if you decide you do not like its appearance after you have seen the screen 177 Commands Output The graphics version of GRAPH produces a multi color scatterplot by default The first series is graphed with points and the remaining series with lines The LINE option can be used to graph all series with lines GRAPH differentiates the series using different colors Options General Givewin only DOS Winonly MAC only DASH NODASH specifies whether the lines for different series on the Screen are to be distinguished by using different dash patterns The default is no dashes just color on the screen and dashes on printed output There are 7 dash patterns LINE NOLINE specifies whether the first series should be plotted with a line rather than as a scatter of points The default is to use dots to represent the series ORIGIN NOORIGIN causes a horizontal line to be drawn starting at zero on the vertical axis PAIR NOPAIR specifies whether all the series except the first are to be plotted on the y axis the default or whether the series are to be used in pairs of x versus y Since GRAPH always draws a line for every pair of series except the first
297. imated parameters T vector params asymptotic T statistics vector p values for asymptotic T statistics based on normal distribution SSR vector eqs Sum of squared residuals for each equation stored in a vector S vector eqs Standard error of each equation stored in a vector DW vector eqs Durbin Watson statistic for each equation stored in a vector RSQ vector eqs R squared for each equation stored ina vector if eqs are normalized ARSQ vector eqs Adjusted R squared for each equation if eqs are normalized YMEAN vector eqs Vector of means of dependent variables if eqs are normalized SDEV vector eqs Vector of standard deviations of dependent variables if eqs are normalized VCOV matrix par par Estimated variance covariance of estimated parameters COVU matrix eqs eqs Residual covariance matrix the of eqs is the number of stuctural equations FIT matrix obs eqs Matrix of fitted values if equations are normalized RES matrix obs eqs Matrix of residuals Options ENDOG list of endogenous variables This defines the endogenous variables including those that do not appear on the left hand side of any behavioral equation There is no requirement that endogenous variables appear on the left side of any equation since FIML estimates are invariant to this normalization However the number of endogenous variables must be equal to the number of equations 147 Comman
298. include non alphabetic or lowercase characters it does not need to be enclosed in quotes FORMAT BINARY or DATABANK or or EXCEL or FREE or LOTUS or RB4 or RB8 or STATA or format text string specifies the format in which your data is to be read The default is free format which means the fields numbers are separated by blanks or tabs and may be of varying length Each of the format options is described in more detail below and under the FORMAT entry FORMATzZBINARY specifies that the data is in binary single precision REAL 4 format To read data in this format it must be on an external file since binary data cannot be intermixed in a TSP program input file This method of reading data is quite fast about the same as a TSP databank but not quite as easy to use This is the same as FORMAT RBA FORMAT RB8 is for double precision binary FORMAT DATABANK specifies that the data are to be read from a TSP databank FORMAT EXCEL reads Excel spreadsheet file similar 10 FORMAT LOTUS If the filename ends with XLS this is the default It handles version 5 7 95 97 98 2000 2002 Excel files Excel 97 and later files can have up to 65536 rows of data FORMAT FREE is the default If the number of values read does not match the expected number of observations times the number of variables an error message is printed and TSP tries to make its best guess as to what was meant FORMAT LOTUS reads Lotus 123 Excel or Quattro Pro wor
299. including these variables in the regression The estimates of these coefficients are not easily interpreted and so TSP unscrambles these results for each PDL variable after the regression and presents the estimates in terms of the original variable and its lags The results begin with a title Distributed Lag Interpretation for variable name followed by the estimated mean lag and its standard error computed as the average lag weighted by the lag coefficients If any of the lag coefficients are less than zero this quantity does not have very much meaning The estimated sum of the lag coefficients and its standard errors are also shown Both these standard errors are computed by taking into account the covariance of the estimates of the lag coefficients Following these summary statistics a table and a plot of the individual lag coefficients is printed This table also shows the standard errors of the estimates and plots standard error bands around the lag coefficients These coefficients are also stored in data storage with the following names variable type length description SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag LAGF vector lags Estimated lag coefficients after unscrambling PDL1 2 etc series obs Scrambled right hand side variables 325 Commands Method PDL estimates are obtained by forming linear combinations of the underlying variable and its lags with weights determined by the order of
300. ing series before the statistics are computed so that the series should be proportional to the inverses of the variances of the variables If the weight is zero for a particular observation that observation is not included in the computations nor is it counted in determining degrees of freedom The quartile estimates including the median are also weighted estimates Examples LIST VARS PAT RND ASSETS DRND DPAT MSD CORR VARS MSD CORR COVA WEIGHT POP INCOME PHONES NEWBUS References Davidson Russell and James G MacKinnon Estimation and Inference in Econometrics Oxford University Press New York NY 1993 Chapter 16 Godfrey L G Misspecification Tests in Econometrics Econometric Society Monograph Cambridge University Press Cambridge England 1988 pp 143 145 272 NAME NAME Examples NAME often the first statement in a TSP job is used to supply a job or user name to be printed at the top of each page and optionally a title for the run NAME lt jobname gt text string to be used as title Usage The only required argument on the NAME statement is the jobname which may be any descriptive name of up to 8 characters which you wish to give your job or could be your name to distinguish your jobs from others if they are being run together The job title is optional but recommended if included it will be printed at the top of each page of output until a TITLE statement is executed which replac
301. ing way if the value of the expression for an observation is greater than zero the observation is kept otherwise it is dropped The resulting SMPL vector replaces the previous one Note that SMPLIF unlike SMPL chooses only observations within the current sample you should reset the sample to cover the whole data set if you want to select a different group of observations later Successive SMPLIFs will nest within each other resulting in a non increasing set of observations Use SELECT for non nested observation selection If the expression is false for all current observations an empty sample would result NOB is stored as zero for this case but the sample is left unchanged NOB should be tested when empty samples are possible Output SMPLIF produces the same output as SMPL the resulting sample is printed if printing has not been suppressed and the variables SMPL and NOB are stored in data storage Options PRINT NOPRINT prints the full set of sample pairs resulting from SMPLIF Normally only one line is printed SILENT NOSILENT prints no output at all Examples Delete the first observation for every individual in a panel data set which has six years of data for each of 20 people 410 SMPLIF SMPL 1 120 TREND PERIOD 6 YEAR SMPLIF YEAR gt 1 This example shows how logical expressions can be used to select data for estimation SMPLIF gt 0 amp gt 0 amp DER lt ELAG 411 Commands SOLVE
302. ingly Unrelated Regressions and Tests of Aggregation Bias JASA 57 1962 pp 348 368 Zellner Arnold Estimators for Seemingly Unrelated Regression Equations Some Exact Finite Sample Results JASA 58 1963 pp 977 992 423 Commands SYMTAB Example SYMTAB prints the TSP symbol table showing the characteristics of all the variables in a TSP program It is useful primarily to programmers for debugging TSP programs Others may prefer the SHOW command which does not print out information on program variables SYMTAB Usage SYMTAB can be used anywhere in the program it will print out the names locations types lengths and file pointers for all the variables used up to that point in the program A description of the table is given in the output section below Output The symbol table printout has 6 items for each variable 1 Variable name you will see all the variables you have created as well as all the variables which contain results of procedures In addition there are a large number of variables which begin L 0001 or F 0001 and so forth These variables are the TSP program lines and the equations which are created by the GENR and SET commands 2 Location this is the address of the variable in the upper end of blank common in single precision words 3 Type this is the variable type Legal types are the following type description scalar or constant double precision time series see OPTIONS
303. ion TSP It is expected that the most frequent use of this command will be to locate files to INPUT The second form DIR filename lists all files in the current directory that fit the description filename You may find it convenient to locate related input output and databank files in this way Filename however must meet the requirements of a TSP variable name since it will be processed in the same manner This means that no filename extension is possible or needed here The third form is the most flexible and is simply the command with no arguments prompting you for the file specification DIR files computer response in response to which you may type anything you would ordinarily include with a DOS Windows directory command In this way you may specify directories other than the current lists of files other wildcard combinations or even command qualifiers such as date size etc 115 Commands DIVIND Options Examples References DIVIND computes Divisia price and quantity indices from a set of n price and quantity series A Divisia index of prices is obtained by cumulating the rate of change to the values of an index of price change observation by observation The index of price change is the weighted sum of the rates of change of the component prices The weights are the current shares of the component goods in the total current expenditure on all the goods in the index A Divisia index is the ultima
304. ion and Data Analysis Commands 20 Nonlinear Estimation and Formula Manipulation Commands 21 QDV Qualitative Dependent Variable Commands 22 Hypothesis Testing Commands 23 Forecasting and Model Simulation Commands 24 Time Series Identification and Estimation Commands 25 Control Flow Commands 26 Interactive Editing Commands and or Data Commands 27 Obsolete Commands 28 Cross Reference Pointers 29 31 ACTFIT 31 ADD interactive 33 ANALYZ 35 30 Table of Contents 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 AR1 41 ARCH 49 ASMBUG 54 BJEST 55 BJFRCST 63 BJIDENT 68 CAPITL 72 CDF 74 CLEAR interactive 81 CLOSE 82 COINT 84 COLLECT Interactive 95 COMPRESS 97 CONST 98 CONVERT 99 COPY 102 CORR COVA 103 DATE 104 DBCOMP Databank 105 DBCOPY Databank 106 DBDEL Databank 107 DBLIST Databank 108 DBPRINT Databank 109 DEBUG 110 DELETE 111 DELETE Interactive 112 DIFFER 113 DIR Interactive 115 DIVIND 116 DO 119 DOC 121 DOT 122 DROP Interactive 125 DUMMY 127 EDIT Interactive 129 ELSE 132 END 133 ENDDO 134 ENDDOT 135 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96
305. ions Note that for all the Box Jenkins procedures BJIDENT BJEST and BJFRCST TSP remembers the options from the previous Box Jenkins command except for nonlinear options so that you only need to specify the ones you want to change CONSTANT NOCONS specifies whether a constant term is to be included in the model CUMPLOT NOCUMPLO specifies whether a cumulative periodogram of the residuals is to be plotted The number of computations required for this plot goes up with the square of the number of observations so that it may be better to forego this option if the number of observations is large EXACTML NOEXACT specifies exact versus conditional maximum likelihood estimation EXACTML is recommended for models with a unit root in the MA polynomial NARz the number of autoregressive parameters in the model The default is zero the number of back forecasted residuals to be calculated The default is 100 NDIFF the degree of differencing to be applied to the series The default is zero NLAG the number of autocorrelations Q statistics to calculate The default is 20 the number of moving average parameters to be estimated The default is zero 5 the number of seasonal autoregressive parameters to be estimated The default is zero NSDIFF the degree of seasonal differencing to be applied to the series The default is zero no differencing NSMA the number of seasonal moving average paramete
306. ions see NONLINEAR section HITER B HCOV B is the default for the FRML method HITER F HCOV U GRAD C2 is the default for the PROC method Starting values are from the PARAM and SET statements Note that the CONST command allows fixing parameters during an estimation for the method Examples FRML method For example in the Probit model the likelinood is CNORM XB for Y lt 0 and 1 CNORM XB for Y gt 0 This could be written in the following way see the User s Guide for alternate coding and many more examples Y0 lt 0 GENR Y1 gt 0 FRML EQ1 LOGL LOG YO CNORM XB Y1 1 CNORM XB The XB expressions can be filled in later with the EQSUB command Note that this allows for nonlinear equations as in this example FRML NLXB XB BO B1 X1 B2 B1 X2 263 Commands EQSUB EQ1 NLXB ML EQ1 PROC method Here is a simple concentrated log likelihood function where we estimate the mean of a time trend and concentrate out the variance parameter to reduce the nonlinearity of the function make sure that residuals are stored in double precision OPTIONS DOUBLE SMPL 1 9 TREND T PARAM 2 ML NRMLC MT PROC form of the ML command PROC NRMLC E T MT residual SIG2 sigma squared variance SET 4 ATAN 1 tan pi 4 1 SET LOGL NOB 2 LOG SIG2 1 LOG 2 PI ENDPROC References Berndt E K B H Hal
307. iplication the Hadamard product is denoted by the operator In the descriptions of the matrix operators that follow we use the following symbols to denote the inputs and outputs of operations scalar or subscripted variable integer scalar any matrix if scalar treated as 1 by 1 matrix square matrix N by N symmetric matrix assumed positive semi definite diagonal matrix assumed positive semi definite upper triangular matrix assumed positive semi definite column vector N by 1 Here are the symbolic operators understood by the MAT command in addition to the ordinary operators used also in GENR Remember that the operands must be conformable for the operations that you request TSP will check the dimensions for you and refuse to perform the computation if this condition is violated m matrix product 251 Commands m m s or s m scalar multiplication matrix transpose m m m matrix transpose with implied matrix product m qm matrix inverse matrix inverse with implied matrix product Kronecker product m Hadamard product element by element When TSP processes a MAT command it recognizes several operations where great savings of computation time can be made by eliminating duplicate calculations These situations include but are not limited to the cross product operation which generates a symmetric matrix and the calculation of a quadratic form the expre
308. is y XB e Instead of y we observe a category value Y which implies that lies between known limits where the limits may include minus or plus infinity In the usual application the set of possible limits are the same for all observations but this is not necessary The underlying model is the same as that used for Ordered Probit that is e is assumed to be normally distributed but with known limits For example suppose there are 3 categories with category values 0 1 and 2 where the first and the last are open ended The model is Y 0 if MUO lt XB e lt MU1 infinity MU1 a known value Y 1 if MU1 lt XB e lt MU2 MU2 a known value Y 2 if MU2 lt XB e lt MU3 Note MU3 infinity The terms in the likelihood function for observations with each of the values 0 1 or 2 are the following log L Y 0 logl d X 8 log L Y 1 log gu Jdu log dX 4 X B dX X B log L Y 2 log 1 dX 4 X 8 where and q denote the cumulative normal and normal distribution respectively INTERVAL uses analytic first and second derivatives to obtain maximum likelihood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly TSP uses zeros for starting parameter values START can be used to provide different starting values see NONLINEAR Multicollinearity of the independent variables is handled with generalized inverses
309. is required The dimension of this Symmetric or diagonal matrix is the number of series to create If only one argument is supplied before the semicolon the series are stored in a matrix with this name The VMEAN option should be used if a non zero mean vector is desired 354 RANDOM To create a random variable from an empirical distribution function a series usually a set of residuals generated by the distribution function must be supplied This feature is useful for computing bootstrap standard errors A set of residuals generated by the model is used as input to the DRAW option and a new series with the same distribution as the old one is obtained by drawing observations from a discrete distribution with probability mass equal to one divided by the number of observations placed on each observed value of the residuals This new sample of residuals may then be used in further computations to obtain estimates of functions of these random variables To draw without replacement use the NOREPLACE option Method The method used by RANDOM is the multiplicative congruential method The new uniform generator GEN 2 has period 2 319 and is a combination of 2 multiple recursive generators with 8 seeds See L Ecuyer 1999 generator MRGS32k5a The old uniform generator GEN 1 has period 2 31 1 and multiplier 41358 this choice is described in L Ecuyer 1990 it has optimal randomness in its class Both are implemented in integer math for
310. is required The following FRML is invalid FRML EQNL LOG Y B X It should be rewritten as an implicit FRML for use in FIML or SIML FRML EQNL LOG Y A B X Here are some more examples see the DIFFER section also for examples using the normal density and cumulative normal function FRML ZERO A72 A73 FRML TRIGEQ COSX COS X FRML TRIGEQ2 COSXY X Y X X Y Y 0 5 FRML RCONSTR RHO 2 See the DOT command for an example of defining several similar FRMLs in a DOT loop 166 GENR GENR Options Examples GENR computes transformations of one or more series over the current SMPL and stores the result as a new series with the name specified A previously created TSP equation may also be GENRed and the result stored as a series TSP equations can be created with FRML EQSUB IDENT FORM or DIFFER lt new series name gt lt algebraic formula gt GENR SILENT STATIC lt new series name gt lt algebraic formula gt GENR SILENT STATIC lt equation name gt lt new series name gt Usage The first two forms of GENR are the most commonly used GENR followed by the new variable name an equal sign and a formula which should be composed according the rules for TSP equations given in the Basic Rules section of this Help System Note that the GENR keyword is not required This formula can involve any of the legal TSP functions and as many variables as desired su
311. ist of instrumental variables assumed to be orthogonal to the residuals of the supplied equations by assumption In some models these variables are referred to as the information set Don t forget to include C the constant unless your model does not require one INST 11511 list2 specifies a list of different instruments for each equation There must be as many lists as there are equations 172 GMM ITEROC NOITEROC causes iteration on the COVOC matrix Normally it is left fixed at its initial estimate If MASK and LSQSTART are used one iteration is made on the COVOC matrix This also occurs if NOLSQSTART is used and is not specified ITERU NOITERU causes iteration on the COVU matrix This is the same as the old MAXITW option LSQ KERNEL BARTLETT or PARZEN The spectral density kernel used to insure positive definiteness of the COVOC matrix when NMA gt 0 BARTLETT is discussed by Newey and West 1987 while PARZEN is discussed by Gallant 1987 Both are reviewed by Andrews 1991 LSQSTART NOLSQSTART specifies if 3SLS should be used to obtain starting values for the parameters and COVOC NOLSQSTART should be specified if you are restarting iterations with old parameter values and a COVOC matrix This will be important for testing see the discussion above MASK a matrix of zeroes and ones which specifies which instruments are to be used for which equations The matrix is of instruments by of
312. istic may be meaningless See the REGOPT command for a large variety of additional regression diagnostics A table of right hand side variable names estimated coefficients standard errors and associated t statistics follows The variance covariance and correlation matrices are printed next if they have been selected with the REGOPT command If there are lagged dependent variables on the right hand side the regular Durbin Watson statistic is biased so an alternative test for serial correlation is computed The statistic is computed by including the lagged residual with the right hand side variables in an auxiliary regression with the residual as the dependent variable and testing the lagged residual s coefficient for significance See the Durbin reference for details this method is very similar to the method used for correcting the standard errors for AR1 regression coefficients in the same lagged dependent variables case This statistic is more general than Durbin s h statistic since it applies in cases of several lagged dependent variables It is not computed if there is a WEIGHT or gaps in the SMPL and the presence of lagged dependent variables is not detected if they are computed with GENR instead of being specified with an explicit lag like OLSQ Y C X Y 1 or in a PDL If the PLOTS option is on TSP prints and plots the actual and fitted values of the dependent variable and the residuals OLSQ also stores most of these resul
313. it to appear immediately following the estimation However you must specify the coefficient vector the names of the right hand side variables and if a serial correlation correction is desired the value of RHO If STATIC or NODYNAM is specified FORCST will treat a lagged dependent variable like any other exogenous variable in computing the forecast However if you specify the default DYNAM option on the FORCST statement FORCST will feed back the fitted values into the lagged dependent variable dynamically As an initial condition the actual lagged dependent variable is used in the first period 151 Commands When FORCST follows an AR1 regression the forecast is computed including a serial correlation correction using the estimated value of rho from the regression This means that the dependent variable y must be available to the program so that y ift 1 X Y ift 1 may be computed This static forecast is that computed by FORCST when STATIC is specified To obtain a true dynamic forecast or extrapolation use the default DYNAM option In this case FORCST will look for presample data to calculate a presample residual In either case unless the FORCST statement immediately follows the AR1 estimation of interest the name of the dependent variable must appear in the DEPVARz option If there are any gaps in the SMPL vector FORCST will treat the observations between each pair of SMPL numbers separately
314. its inclusion guarantees that the standard errors are always consistent even if there are lagged dependent variables on the right hand side The fitted values FIT and residuals RES are computed as follows FIT t X b i 1 1 EXACTML FIT t X b p y X b t gt 1 RES t y FIT t AR1 also stores this regression output in data storage for later use The table below lists the results available after an AR1 command Note the number of coefficients vars always includes RHO variable type length description RNMS list vars Names of right hand side variables LHV list 1 Name of the dependent variable RHO scalar 1 Serial correlation parameter at convergence SSR scalar 1 Sum of squared residuals S scalar 1 Standard error of regression YMEAN scalar 1 Mean of the transformed dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations DW scalar 1 Durbin Watson statistic RSQ scalar 1 R squared ARSQ scalar 1 Adjusted R squared IFCONV scalar 1 1 if convergence achieved 0 otherwise LOGL scalar 1 Log of likelihood function COMFAC scalar 1 Common factor test if OBJFN GLS 42 AR1 COEF vector vars Coefficient estimates SES vector vars Standard Errors T vector vars t statistics vector vars p values for t statistics COEFAI vector vars Fixed effect estimates FEI SESAI vector itvars Standard Errors on
315. ized form that is with each endogenous variable appearing once and only once on the left hand side of an equation These equations may be specified with FRML or IDENT statements There is no difference between the two types of equations in SOLVE for either one the model solution tries to make the error as small as possible There must be as many equations as endogenous variables After the equations are specified form and order the model with the MODEL procedure This procedure takes the list of equations and endogenous variables in the model and produces a collected and ordered model which is stored under a name which you supply This is the name which should appear on the SOLVE statement SOLVE solves the model specified over the current SMPL one period at a time The starting values for the variables are chosen as follows 412 SOLVE 1 Ifthe variables already exist the actual values for the current period are used as starting values unless they are missing values 2 If the variables do not exist and it is the first period of the simulation the value zero is used as a starting value 3 If the variables do not exist and this is not the first period of the simulation the values of the last period solution are used as starting values Output If no options are specified the normal output from SOLVE begins with a title and listing of options This is followed by a table of the data series if the PRNDAT option is on
316. ksheet files most files with the extension WKx where x is any character 367 Commands FORMAT RB4 is the same as FORMAT BINARY single precision binary FORMAT RBS is used for double precision binary FORMAT STATA reads stata Version 7 or earlier files FORMAT a format string enclosed in quotes format string specifies the format with which the data are to be READ The quotes are required and should surround a Fortran FORMAT statement including the parentheses but excluding the word FORMAT If you are unfamiliar with the construction of a Fortran FORMAT statement see the FORMAT entry FULL NOFULL applies only to reading diagonal symmetric or triangular matrices It specifies whether the complete matrix is to be read or only the upper triangle in the case of triangular the lower triangle symmetric or the diagonal diagonal NCOL the number of columns in the matrix This is required for a general matrix NROW z the number of rows in the matrix This is required for a general matrix Either NROW or NCOL must be specified for symmetric triangular or diagonal matrices These options only apply to matrices PRINT NOPRINT specifies whether or not the data is to be printed as it is READed This option applies only to free format READing It may be set globally for the READ section by use of the NOPRINT statement SETSMPL NOSETSMP specifies whether the SMPL is to be determined from the number of data items read The d
317. l R E Hall and J A Hausman Estimation and Inference in Nonlinear Structural Models Annals of Economic and Social Measurement October 1974 pp 653 665 Gill Philip E Walter Murray and Margaret H Wright Practical Optimization Academic Press New York 1981 264 MMAKE MMAKE Options Examples MMAKE makes a new matrix by stacking a set of series or matrices or a new vector from a set of scalars In the series case the new matrix normally has the number of rows equal to the number of observations and the number of columns equal to the number of series In the matrix case the new matrix has the number of rows equal to the common number of rows of the matrices and the number of columns equal to the total number of columns The vector has the number of rows equal to the number of scalars MMAKE is the reverse of UNMAKE which breaks a matrix into a set of series or a vector into a set of scalars To make a matrix from another matrix or vector by changing its type dimensions or transposing it use the MFORM procedure MMAKE VERT lt matrix name gt lt list of series gt or MMAKE vector name gt list of scalars gt or MMAKE VERT lt matrix name gt lt list of matrices gt Usage MMAKE s first argument is the name to be given to the new matrix followed by a list of series or matrices which will form the columns of the new matrix The number of series is limited only by the maximum size of the argum
318. le see the COMPRESS command 111 Commands DELETE Interactive DELETE followed by numbers instead of names removes command lines DELETE lt firstline gt lt lastline gt Usage When used in the interactive version of TSP DELETE enables re execution of a range of lines with unnecessary steps eliminated The lines are permanently lost It is also useful in COLLECT mode for modifications to the range of lines just entered before their execution since DELETE is executed immediately regardless of mode If both arguments are present all lines from firstline through lastline will be deleted If the second argument is omitted only firstline will be deleted Arguments must be valid line numbers i e integers Deleted line s will be absent from the backup file DELETE will remain however to provide a more accurate record of actions taken during the session 112 DIFFER DIFFER Options Examples DIFFER differentiates a TSP equation analytically with respect to the list of arguments and stores each derivative in parsed form as another TSP equation FRML These equations may be evaluated using GENR used in estimation or printed just like any other TSP equation DIFFER DEPVARPR lt new dependent variable name gt PRINT PREFIXz new equation name equation name gt list of arguments Usage DIFFER requires the name of the equation to be differentiated followed by one or more arguments for differenti
319. less SUPRES SMPL has been specified earlier in the program The sample vector is also stored in data storage under the name SMPL The number of observations in the current sample is stored as a scalar under the name NOB This can be quite convenient if you do not know exactly how many observations a SMPLIF or SELECT command will yield for example Examples FREQ A SMPL 56 80 SMPL 21 40 46 82 FREQ Q SMPL 72 1 82 4 FREQ M SMPL 78 5 81 9 FREQ N SMPL 2 7 9 14 16 21 23 28 30 35 The last example specifies groups of six observations at a time skipping every seventh observation beginning with the first This is a common arrangement for panel data with a single lagged endogenous variable but it is easier to use FREQ PANEL which automates this sample selection Suppose we have a vector called SAMPLE loaded with 2 7 9 14 16 21 23 28 30 35 Then the last example could also be done with SMPL SAMPLE 409 Commands SMPLIF Options Examples SMPLIF is used to select a sample of observations based on the values of an expression The observations in the current sample for which the expression is true are selected SMPLIF PRINT SILENT lt logical expression gt Usage SMPLIF is simply followed by an expression This expression can be a series such as a dummy variable or it can involve several series with logical operators The expression is used to select observations from the current sample in the follow
320. ll be used in regressions with a constant term to prevent multicollinearity 127 Commands PREFIX Prefix for naming the dummy variables The default prefix is the name of the input series Examples FREQ Q SMPL 75 1 85 4 DUMMY creates Q1 Q2 Q3 Q4 quarterly dummies The series created have the following values Obs Q1 02 Q3 Q4 75 1 1 0 0 0 75 2 0 1 0 0 7 3 0 0 1 0 754 0 0 0 1 76 1 1 0 0 0 and so forth FREQ M SMPL 75 1 84 12 create M1 M11 DUMMY EXCLUDE monthly dummies with M12 excluded The next example creates a list of dummies from a variable SIZE which takes on 3 values 0 2 and 3 5 DUMMY SIZE SDLIST This is equivalent to the following statements SIZE1 SIZE 0 SIZE2 SIZE 2 SIZE3 SIZE 3 5 LIST SDLIST SIZE1 SIZE3 The next example creates a set of year dummies for panel data assuming you have a variable YEAR which takes on values from 72 to 91 DUMMY YEAR YEAR72 YEAR91 This command creates 20 dummy variables YEAR72 YEAR73 YEAR74 and so forth To create individual dummies for balanced data using TREND and INT see the example under AR1 128 EDIT Interactive EDIT Interactive Examples EDIT is a simple editor providing the capability to perform argument modifications on a TSP command during an interactive session If you are using the Windows or DOS versions of TSP you will not need this command as you will be able to use arrow key editin
321. ll set of conditional variables is used variables corresponding to all available alternatives Examples LOGITYCX LOGIT COND NCHOICE 3 Y XA XB looks for conditional variables XA1 XA2 XA3 and XB1 XB2 XB3 whereas LOGIT COND NCHOICE 3 SUFFIX CAR BUS RT Y XA XB looks for conditional variables XACAR XABUS XART and XBCAR XBBUS XBRT Note that the suffixes must be in the proper order 1 2 3 for correct interpretation of the output See the usage section for other examples of how to use this procedure References Albert A and J A Anderson On the Existence of Maximum Likelihood Estimates in Logistic Regression Models Biometrika 71 1984 Amemiya Takeshi Advanced Econometrics Harvard University Press 1985 Chapter 9 Cameron A Colin and Frank A G Windmeijer An R squared Measure of Goodness of Fit for Some Common Nonlinear Regression Models Journal of Econometrics 77 1997 pp 329 342 240 LOGIT Estrella Arturo A New Measure of Fit for Equations with Dichotomous Dependent Variables Journal of Business and Economic Statistics April 1998 pp 198 205 Hausman Jerry A and Daniel McFadden Specification Tests for the Multinomial Logit Model Econometrica 52 1984 1219 1240 Maddala G S Limited Dependent Variables and Qualitative Variables in Econometrics Cambridge University Press 1983 Chapters 2 and 3 McFadden Daniel Regression Based Specification Tests
322. llows for 16 significant digits so don t make the seed too large if you are trying to reproduce results When used with the new default uniform generator all 8 seeds are set to the SEEDIN value SEEDOUT random seed of the random generator the current random seed before any random variables are created by this commana To print the seed use OPTIONS NWIDTH 20 to provide enough digits SEEDOUT is not useful with the new uniform generator because it uses 8 seeds STDEV the standard deviation of the random variable or a series containing standard deviations This option applies only to normal gamma and negative binomial random variables The default value is one When a series is supplied each random number drawn will come from a distribution with a different standard deviation T NOT specifies that the random number generated is to follow student s t distribution with degrees of freedom given by the DF option UNIFORM NOUNIFORM specifies that the random number generated is to follow the uniform distribution between zero and one VARz the variance of the random variable or a series containing variances This option applies only to normal gamma and negative binomial random variables The default value is one When a series is supplied each random number drawn will come from a distribution with a different variance VCOV symmetric variance covariance matrix for multivariate normal random variables You can also supply the tri
323. log likelihood and a table of right hand side variable names estimated coefficients standard errors and associated t statistics NEGBIN also stores some of these results in data storage for later use The table below lists the results available after a NEGBIN command variable type LHV list RNMS list IFCONV scalar YMEAN scalar SDEV scalar NOB scalar HIST vector HISTVAL vector SSR scalar RSQ scalar LR scalar LR scalar LOGL scalar SBIC scalar NCOEF scalar NCID scalar COEF vector SES vector T vector vector GRAD vector VCOV matrix FIT series RES series length 1 vars 1 1 1 1 values values 1 1 1 r vars vars vars vars vars vars vars obs obs description Name of dependent variable List of names of right hand side variables 1 if convergence achieved 0 otherwise Mean of the dependent variable Standard deviation of the dependent variable Number of observations Frequency counts for each dependent variable value Corresponding dependent variable values Sum of squared residuals correlation type R squared Likelihood ratio test for zero slope coefficients P value for likelihood ratio test Log of likelihood function Schwarz Bayesian Information Criterion Number of independent variables vars Number of identified coefficients Coefficient estimates Standard errors T statistics p values for T statistics Gradient of
324. log likelinood at convergence Variance covariance of estimated coefficients Fitted values of dependent variable Residuals actual fitted values of 275 Commands dependent variable If the regression includes a PDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector lags Estimated lag coefficients after unscrambling Method NEGBIN uses analytic first and second derivatives to obtain maximum likelihood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly TSP uses zeros for starting parameter values except for the constant term and alpha START can be used to provide different starting values see NONLINEAR Multicollinearity of the independent variables is handled with generalized inverses as in all the estimation procedures in TSP The exponential mean function is used in the NEGBIN model That is if X are the independent variables and B are their coefficients E Y X This guarantees that predicted values of Y are never negative The ML command can also be used to estimate Negative Binomial models including panel data models with fixed and random effects See our web page for the panel examples Options MODEL type of variance function For MODEL 1 the variance is proportional to the mean V Y X E Y X 1 alpha For the default MODEL 2 the var
325. lt 17 u ay 4 51 50 U QU tPU QU e AR L Let L 7 for illustration Y ay t 51 5 7 al y a ay 5 t 1 d t 1P e a 1 a g Y A Bt Bt e All unit root tests are computed from possibly weighted OLS regressions a few lagged or differenced variables The coefficient of y t 1 is printed in the tables as alpha Accurate asymptotic P values for Dickey Fuller Phillips Perron and Engle Granger for up to 6 cointegrating variables are computed using the coefficients in the MacKinnon reference Note that these asymptotic distributions are used as approximations to the true finite sample distributions The WS test is a weighted double length regression First the variable being tested is regressed on the constant trend variables using the full current sample and the residual from this is used as the dependent variable Y in the double length regression The data setup for the first half of this regression is the same as an augmented Engle Granger test regress Y on lagged Y and lags of DY The weights are t 1 T where T is NOB in the original sample In the second half Y is regressed on Y 1 and leads of Y Y 1 using weights 1 t 1 T See Pantula et al 1994 for more details P values for the WS test are computed very roughly by interpolating between the asymptotic 5 and 10 level critical values given for the constant and no trend case in the reference
326. lues of the dependent variable Matrix of fitted values of the dependent variables Normal LSQ output begins with a listing of the equations The model is checked for linearity in the parameters which simplifies the computations A message is printed if linearity is found and LSQ does not iterate because it is unnecessary The amount of working space used by LSQ is also printed this number can be compared with the amount printed at the end of the run to see how much extra room you have if you wish to expand the model 245 Commands Next LSQ prints the values of constants and the starting conditions for the parameters and then iteration by iteration output If the print option is off this output consists of only one line showing the beginning value of the log likelinood the ending value the number of squeezes in the stepsize search ISQZ the final stepsize and a criterion which should go to zero rapidly if the iterations are well behaved This criterion is the norm of the gradient in the metric of the Hessian approximation It will be close to zero at convergence When the print option is on LSQ also prints the value of the parameters at the beginning of the iteration and their direction vector These are shown in convenient table so that you can easily spot parameters with which you are having difficulty Finally LSQ prints the results of the estimation whether or not it converged these results are printed even if the NOP
327. mat descriptor The records do not necessarily have to have identical formats although they will usually be of the same length 162 FREQ FREQ Examples FREQ sets the frequency for the series in your TSP run It may be changed during the course of a TSP run but series of different frequencies cannot be mixed in the same command see CONVERT for an exception to this rule The PANEL options are used to interpret any FREQ N series as panel time series cross section data that is to tell TSP how to identify one individual from the next These options are used by any subsequent TSP commands which support panel data such as PANEL AR1 and PRINT FREQ NONE or ANNUAL or MONTHLY or QUARTER or WEEKLY or FREQ value or FREQ PANEL ID ID series T lt value gt N lt value gt TIME series START lt date gt or A or or M or W or value Usage FREQ is very simple FREQ followed by one of the choices above Single letter abbreviations are allowed The annual monthly quarterly and weekly frequencies imply one 12 4 or 52 periods per year respectively The year is assumed to be base 1900 if it has two digits or base 0 if it has four You can reset the base using the BASEYEAR option see the OPTIONS command entry for details format of dates in TSP is always YYYY PP where YYYY is the year and PP is the period PP can be any number between 1 and the frequency The period is suppressed when the fre
328. mber of Hermite quadrature points for numeric integration used for PROBIT REI default is 20 PRINT NOPRINT produces short diagnostic output at each iteration including a table of the current parameter estimates and their change vector The value of the objective function for each squeeze on the change vector is also printed CRIT is the norm of the gradient in the metric of the Hessian which approaches zero at convergence SILENT NOSILENT suppresses all printed output STEP BARD or BARDB or CEA or CEAB or GOLDEN specifies the stepsize method for squeezing The default depends on HITER and the procedure HITER option STEP default Newton CEA BHHH CEA Gauss BARD for LSQ CEAB for FIML DFP GOLDEN SQZTOL tolerance of determining stepsize Used for STEP GOLDEN The default is 0 1 SYMMETRIC NOSYMMETRIC is an old option which has been replaced with GRADIENT method SYMMETRIC is the same as GRAD C4 NOSYM is equivalent to GRAD FORWARD TERSE NOTERSE produces brief output consisting of the objective function for the estimation procedure and a table of coefficient estimates and standard errors TOL tolerance of determining convergence of the parameters using a unit stepsize The default for most procedures is 001 for AR1 it is 000001 TOLG tolerance of determining convergence of the norm of the gradient printed as CRIT in the output The default is 001 CRIT g H 1g which is usually many orders of magnitude
329. mended Examples Assume that the following equations have been specified for the illustrative model of the U S economy FRML CONSEQ CONS A B GNP FRML INVEQ LAMBDA I 1 ALPHA GNP DELTA R FRML INTRSTEQ R D FLOG GNP LP LM FRML PRICEQ LP LP 1 PSI LP 1 LP 2 PHI LOG GNP TREND TIME P0 248 LSQ PARAM A LAMBDA ALPHA D F PSI PHI TREND CONST DELTA 15 The model as specified has four equations the parameters to be estimated are A B LAMBDA ALPHA D F PSI PHI TREND and PO There are 7 variables in the model CONS GNP I LP LM and TIME and one additional instrument G To estimate the investment equation by nonlinear least squares use the following command LSQ NOPRINT 0001 INVEQ We can obtain multivariate regression estimates of the whole model with the following command although these estimates are probably not consistent due to the simultaneity of the model there are endogenous variables on the right hand side of the equations LSQ MAXIT 50 CONSEQ INVEQ PRICEQ INTRSTEQ The example below obtains three stage least squares estimates of the model using a weighting matrix based on the starting values of the parameters which are obtained by nonlinear two stage least squares LSQ INST C LM G TIME CONSEQ LSQ INST C LM G TIME INVEQ LSQ INST C LM G TIME INTRSTEQ LSQ INST C LM G TIME PRICEQ LSQ INST C LM G TIME
330. meter restrictions and holding parameters fixed 259 Commands 3 Checking your own second derivatives when you are writing a maximization procedure for any program that uses natural language equations 4 Robust models like LOGL ABS Y XB 5 Minimization problems just negate the equation 6 General maximum likelihood problems using functions recognized by TSP including SQRT POS and the gamma factorial function which can be used for the gamma chi squared beta t and F densities Even complicated likelihood functions on large datasets may be estimable using ML Even though more computer time may be required using TSP instead of a custom program programming time is considerably reduced and the relative price of CPU time is usually small and shrinking Maximization with analytic derivatives is usually much faster than with numeric derivatives the method which used to be lowest in programming cost See Timing example below PROC method Sometimes it is extremely difficult or impossible to write down the log likelihood in a single FRML even with use of EQSUB See the list below for some examples For this form of the ML command write a PROC which evaluates the log likelihood and stores it in LOGL Give the name of this PROC as the first argument after any options of the ML command and follow it with a list of PARAMs which are to be estimated Write the PROC so that it starts by checking any constraints on the
331. ming theory suggests that the use of GOTO statements should be avoided since they tend to produce unreadable and hard to debug programs In TSP the DO loop facility and IF THEN ELSE syntax can be used to avoid the use of GOTO statements GOTO statement number or GO TO lt statement number gt Usage A GOTO statement consists of GOTO followed by a statement number which must been defined somewhere in your program The effect of the statement is to transfer control immediately to that statement Note that the GOTO statement cannot be used to transfer to the data section of your program you can use a series of LOAD statements instead Example IF A THEN GO TO 100 BzK K GOTO 200 100 B K K 10000 200 OLSQYCB This example shows the GOTO statement being used to create two branches of the program one to be executed if A is false and one to be executed if A is true The same thing can be done with an IF THEN ELSE sequence IFA THEN B z K K 10000 ELSE BzK K OLSQYCB 175 Commands GRAPH See also graphics version Output Options Example GRAPH plots one series against another using a scale determined by the range of each series Use the PLOT command to graph series against time GRAPH lt series for y axis gt lt series for x axis gt Usage Supply the series names in the order y axis series followed by the x axis series Output GRAPH produces a one page graph of the second series
332. mmand Output If the PRINT option is on the input and output values will be printed along with the degrees of freedom If a second argument is supplied it will be filled with the output values and stored see examples 4 and 6 Input and output arguments may be any numeric TSP variables Method 74 CDF BIVNORM ACM Algorithm 462 Inverse is not supplied because it is not unique unless x or y is known etc CHISQ DCDFLIB method Abramovitz Stegun formula 26 4 19 converts it to Incomplete Gamma and then use DiDinato and Morris 1986 Inverse by iteration trying values of x to yield p faster methods are also known Non integer degrees of freedom are allowed and can be used to compute the incomplete gamma function F and T DCDFLIB method Abramovitz Stegun formula 26 6 2 converts it to Incomplete Beta and then use DiDinato and Morris 1993 i e ACM Algorithm 708 Inverse by iteration Non integer degrees of freedom for F are allowed and can be used to compute the incomplete beta function NORMAL ACM Algorithm 304 with quadratic approximation for E 37 5 Inverse Applied Statistics Algorithm AS241 from StatLib DICKEYF Asymptotic values from Tables 3 and 4 in MacKinnon 1994 Finite sample critical values from Cheung and Lai 1995 augmented Dickey Fuller or MacKinnon 1991 Engle Granger To convert these to finite sample P values a logistic interpolation is used with the 05 size and either the 01 or 10
333. mputed are determined by the current sample If any observations have missing values within the current sample they are dropped from the sample and a warning message is printed for each series with missing values The number of observations remaining is printed with the regression output RES and FIT will have missing values in this case and the Durbin Watson will be adjusted for the sample gaps The list of independent variables on the OLSQ command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These PDL variables are a way to reduce the number of free coefficients when you are entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See the PDL section for a description of how to specify a PDL variable Output 291 Commands The output of OLSQ begins with an equation title and the name of the dependent variable This is followed by statistics on goodness of fit the sum of squared residuals the standard error of the regression the R squared the Durbin Watson statistic for auto correlation of the residuals a Lagrange multiplier test for heteroskedasticity the Jarque Bera test for normality and an F statistic for the hypothesis that all coefficients in the regression except the constant are zero If there is no constant in the regression and the mean was not removed from the dependent variable prior to the regression the F stat
334. n addition the LAD estimator occasionally produces a non unique estimate of the coefficient vector b TSP issues a warning message in this case When a quantile other than 0 5 is requested the formula above is modified slightly When the number of observations is greater than 100 and the model is not censored the estimated variance covariance of the estimated coefficients is computed as though the true distribution were Laplace V b a X X where depends the quantile used az when 0 5 2 1 1 a FF in general When the quantile tau is less than the median 2 and 17 and when it is greater than the median 215 Commands and o 2 7 The variance parameter lambda is estimated as though the data has the Laplace distribution f e exp e 2A This formula can also be derived as the BHHH estimate of the variance covariance matrix if the first derivative of e is defined to be unity at zero as it is everywhere else The outer product of the gradients of the likelihood function will then yield the above estimate The alternative to these Laplace standard errors is to use the NBOOT option to obtain bootstrap standard errors based on the empirical density This is the default when the model is censored or the number of observations is less than 100 In the case of censored estimation the Bilias et al 2000 resampling method is used to speed up the computations this can be
335. n equation title and the name of the dependent variable Then the starting values and diagnostic output from the iterations are printed followed by the convergence status The results printed are the mean of the dependent variable the number of lower censored uncensored and upper censored observations and a table of right hand side variable names estimated coefficients standard errors and associated t statistics The estimated standard deviation of the residual SIGMA is listed last in this table TOBIT also stores some of these results in data storage for your later use The table below lists the results available after a TOBIT command variable LHV YMEAN NOB NPOS LOGL IFCONV NCOEF NCID RNMS COEF SES T GRAD VCOV DBDX RES MILLS type list scalar scalar scalar scalar scalar scalar scalar list vector vector vector vector vector matrix matrix series series length k 1 params params params params params params params params vars 2 obs obs description Name of dependent variable Fraction of positive observations Number of observations Number of positive observations Log of likelinood function 1 if convergence achieved 0 otherwise Number of parameters params including SIGMA Number of identified coefficients list of names of independent variables Coefficient estimates Standard errors T statistics p v
336. n requires 3 variables as arguments DOS Win only A4 NOA4 specifies A4 paper size Available for DEVICE LJ3 or POSTSCRIPT only DEVICE CHAR or EPSON or LJ2 or LJ3 or LJET or LJPLUS or LJR75 or LJR100 or LJR150 or LJR300 or POSTSCRI or PS specifies the hardcopy device to be used for printer output LJ means HP LaserJet or compatible EPSON is EPSON dot matrix or compatible POSTSCRI and PS are Postcript output and CHAR is the old line printer output characters instead of graphics The LJ suffixes specify models of the printer and the LJR suffixes specify the resolution of the LaserJet printer directly rather than giving the printer type The maximum resolutions for the LJET LJPLUS and LJ2 printers are 100 150 and 300 respectively Note that default larger resolutions imply larger file sizes and printing times FILE the name of a file to which the graphics image is to be written This file can be printed later For example if you are running under DOS and your printer device is LPT1 use the command copy b file LPT1 HEIGHT letter height in inches The default is 25 Values in the range 0 1 are valid HIRES NOHIRES controls how graphs are printed in batch mode when PREVIEW is not being used Normally NOHIRES graphs are printed in character mode to the batch output file When the HIRES option is used the patched DEVICE and FILE will be used usually this will send a page to LPT1 for each graph 179 Commands
337. n the cointegrating regression SMPL 58 2 84 3 COINT JOH MAXLAG 2 SEAS NOTREND NOUNIT NOEQ Y1 Y4 92 COINT reproduces the Johansen Juselius 1990 results for Finnish data the chosen number of lags is 1 which matches the results from the paper The test statistics are smaller than those in the paper due to the finite sample correction References Bartlett M S The Statistical Significance of Canonical Correlations Biometrika January 1941 pp 29 37 Campbell John Y and Pierre Perron Pitfalls and Opportunities What Macroeconomists Should Know about Unit Roots in Olivier Jean Blanchard and Stanley Fischer eds NBER Macroeconomics Annual 1991 MIT Press Cambridge Mass 1991 Cushman David O Sang Sub Lee and Thorsteinn Thorgeirsson Maximum Likelihood Estimation of Cointegration in Exchange Rate Models for Seven Inflationary OECD Countries in Journal of International Money and Finance June 1996 Davidson Russell and James MacKinnon Estimation and Inference in Econometrics Oxford University Press New York NY 1993 Chapter 20 Dickey and W A Fuller Distribution of the Estimators for Autoregressive Time Series with a Unit Root JASA 74 1979 427 431 Gregory Allan W Testing for Cointegration in Linear Quadratic Models Journal of Business and Economic Statistics July 1994 pp 347 360 Johansen Soren and Katarina Juselius Maximum Likelihood Estimation and Inferenc
338. nal OUTPUT lt filename gt or filename string Usage You may use OUTPUT to save the results of your entire terminal session or to select portions for subsequent printing or review plots graphs regression results This command will stay in effect until you restore the output stream to the terminal with a TERMINAL command It is not possible to send results to the screen and output file simultaneously but warning and error messages will be displayed in both places as they occur OUTPUT will take only one filename as an argument and if it is not in quotes this filename must conform to restrictions placed on TSP variable names i e it must be limited to eight characters and the filename extension must be omitted If the filename is provided on the command line the extension OUT will be assumed If the filename is absent you will be prompted for it in this case you may specify a directory other than the current as well as an extension or disk unit the only limit is that the whole name must be 32 characters or less Again if the extension is omitted OUT will be assumed You may switch back and forth between your output file and TERMINAL or between any number of output files as much as you like If a file is found to exist already when you open it with the OUTPUT command subsequent output will be appended to it rather than creating a new file You may view any output you ve sent to a disk file by using the SYSTEM command
339. name ends with XLS this is the default FORMAT LABELS is for WRITE only it means that labels like those of the standard PRINT command are to be used FORMAT LOTUS reads Lotus 123 worksheet files WK1 WKS with column names at the top and optional dates in the first column This is the default if the filename includes WK FORMAT RB4 is the same as FORMAT BINARY single precision binary FORMAT RB8 is used for double precision binary FORMAT 10 FORMAT format text string specifies a format with which the data will be read The format text string in TSP is very similar to the format statement in Fortran since the Fortran format processor is used on it However you do not need to know Fortran to construct a simple format string and consequently a description of the features you will need is given here Technical note Since all data in TSP are floating point do not use integer or alphameric formats unless you re using OPTIONS CHARID Also avoid parenthetical groupings in the format unless you are sure you know what they do because the results can be unpredictable with different operating systems A format string starts and ends with a or with the parentheses immediately inside these quotes TSP checks that these parentheses exist and inserts them if they are missing Note that it is possible to use quotes within the format string just use double quotes outside the parentheses and single quotes within them
340. nction for the ARIMA model depends on the infinite past sequence of residuals If we estimate the time series model by simply setting the values of these past residuals to zero their unconditional expectation we might seriously misestimate the parameters if the initial disturbance a0 happens to be very different from zero The solution to this problem suggested by Box and Jenkins is to invert the representation of the time series process i e write the same process as if the future outcomes were determining the past Thus it describes the relationships which the time series will ex post exhibit This representation of the backward process constructs back forecasts of the disturbance series the at series and then uses these calculated residuals in the likelihood function By using a reasonable number of these backcasted residuals the problems introduced by an unusually high positive or negative value for the first disturbance in the time series can be eliminated If the process is a pure moving average process this backcast becomes zero after a fixed number of time periods consequently you can set NBACK to a fairly small number in this case When the EXACTML option is used no backcasting is done the AS 197 algorithm Melard 1984 is used 59 Commands HCOV U numeric second derivatives is the default method of computing the standard errors For iteration HITER F is the default but HITER U can be chosen as an option Opt
341. nd 0 the estimated S2 should be close to unity For m 1 if the prediction errors are collinear there may be problems estimating sigma squared the standard errors recursive residuals and log likelihood Options BPRIOR the vector of prior coefficients b 0 for measurement equation Required if XFIXED is used otherwise it will be calculated by default from a regression in the initial observations of the sample If the first m observations are not sufficient to identify the prior one observation is added and BPRIOR is estimated again BTRANS T the matrix of coefficients in the transition equation default identity matrix EMEAS NOEMEAS indicates the presence of an error term in the measurement equation NOEMEAS or NOEM is the same as VMEAS zero ETRANS NOETRANS indicates the presence of an error term in the transition equation PRINT NOPRINT prints the prior STATE RES1 RECRES SMOOTH RESD SILENT NOSILENT turns off most of the output 206 KALMAN SMOOTH NOSMOOTH computes fixed interval smoothed estimates of the state vector b t stored in SMOOTH and the direct residuals RESD VBPRIOR P 0 the variance of the prior symmetric matrix Required if BPRIOR is specified Note sigma squared is factored out of this matrix VMEAS the variance of the measurement equation symmetric matrix Default identity matrix In Harvey s notation this is SHS VTRANS the variance of the transition equ
342. ndogenous variables in the block The method uses numerical first derivatives of the objective function and a rank one updating technique to build up the second derivative matrix See the references for further information on this method Options 2 specifies a secondary convergence criterion which is applied to the sum of squared residuals for each block of equation after the variables have passed the standard TOL convergence test DEBUG NODEBUG prints the endogenous variables at each iteration It is useful for debugging recalcitrant models DYNAM NODYNAM specifies dynamic simulation Earlier solved values of lagged endogenous variables are used in place of actual values STATIC is the alternative to DYNAM DYNAM is the default unless there are no lagged endogenous variables KILL NOKILL specifies whether the simulation is to be stopped if convergence fails for any block or period If you use the DYNAM option you may wish to specify the KILL option since the simulated results will feed forward MAXPRT the number of iterations for which printout of the residuals is to be produced This has no effect unless PRNRES is on In that case the default value is 5 414 SOLVE METHOD GAUSS or JACOBI or FLPOW specifies the iteration method to be used The JACOBI method is a variation of the Gauss Seidel method in which the endogenous variables for any simultaneous block are all updated at once at the beginning of the iteration se
343. near Estimation and Data Analysis Commands 20 correlation matrix of several series covariance matrix of several series instrumental variables and two stage least squares regression computes a Kernel density estimation or regression Least Absolute Deviations estimation median regression Limited Information Maximum Likelihood Least Median Squares estimation mean standard deviation minimum maximum sum variance skewness kurtosis for a list of series Ordinary Least Squares linear regression can use weights panel data estimation total within between variance components describes specification of Polynomial Distributed Lag variables Almon lags and Shiller Lags used in OLSQ INST 1 PROBIT TOBIT Principal Components simple factor analysis Nonlinear Estimation and Formula Manipulation Commands Nonlinear Estimation and Formula Manipulation Commands CONST DIFFER EQSUB FIML ML MLPROC NONLINEAR options PARAM ie UR 1 i 42 defines scalars as fixed non estimable constants create equations with analytic derivatives of formulas FRMLs equation substitution one formula into another Full Information Maximum Likelihood estimation system of linear or nonlinear equations identities implicit equations general cross equation restrictions Multivariate Normal errors define a linear or nonlinear equation for estimation Generalized Method of Moments estimation
344. new TSP variable gt Output The new variable is stored in data storage The old variable is left unchanged If a variable with the new name already exists it is deleted Example Save the coefficients from a regression in the vector B1 Note that this can usually be done more efficiently with RENAME unless the original COEF needs to be saved for a procedure like FORCST OLSQ YC COPY COEF B1 102 CORR COVA CORR COVA See MSD CORR ALL COVA MOMENT MSD PAIRWISE PRINT WEIGHTz seriesname list of variables COVA ALL CORR MOMENT MSD PAIRWISE PRINT WEIGHTz seriesname list of variables 103 Commands DATE Examples DATE prints the current time and date This is useful when printing results from an interactive session If a variable name is supplied nothing is printed and the number of seconds since midnight are stored in the variable This is useful for timing a group of TSP statements such as a PROC DATE lt scalar variable name gt Examples DATE print the current date time to time stamp the output file print the elapsed time required by the procedure MYPROC DATE SECO MYPROC Y 2 DATE SEC1 SET NSEC SEC1 SECO WRITE FORMAT MYPROC took G12 1 seconds NSEC 104 DBCOMP Databank DBCOMP Databank DBCOMP compresses a TSP databank DBCOMP filename Usage Follow the word DBCOMP with the name of the TSP databank to be compresse
345. nning a command which may need a large part of available memory An alternative way to compress out unused variables is to use the SAVE command followed by restarting TSP and using the RESTORE command This would remove all the PROCs and leave only variables like time series scalars lists FRMLs and matrices Output The number of deleted variables if any is printed The amount of recovered working space if any is also printed Options PRINT NOPRINT specifies whether a summary message is printed describing the space freed by compression This message is always printed for automatic compression 97 Commands CONST Examples CONST defines scalar variables constants and assigns arithmetic values to them To define scalars that will be estimated in one of the nonlinear procedures such as LSQ use PARAM instead CONST varname value varname value Usage CONST may be followed by as many argument pairs as desired limited only by TSPs argument limit Each pair is the name of the scalar variable followed by the value it is to be given The variable names may be new or previously defined variables The value may be omitted in which case the variable is either given the value zero if it is new or left unchanged if it has already been defined The use of the CONST procedure is primarily to suppress the estimation of some of the parameters in a nonlinear estimation instead of using a PARAM statement to give the pa
346. ns for TSP These equations can be used later in the program for estimation simulation or they may simply be saved for later SOLVE procedures all require equations specified by FRML or FORM PARAM as input GENR and SET can also take a FRML name as an argument FRML lt equation name gt lt variable name gt lt algebraic expression gt or FRML lt equation name gt lt algebraic expression gt Usage There are two forms of the FRML statement the first has the name to be given to the equation followed by an equation in normalized form that is with the name of the dependent variable on the left hand side of the equal sign and an algebraic expression for that variable on the right hand side The expression must be composed according to the rules given in the Basic Rules section which introduces this manual These rules are the same wherever an equation is used in TSP in IF statements GENR SET FRML IDENT SMPLIF SELECT and GOTO DOTted variables can be used in FRMLs The second form of the FRML statement is implicit there is no equal sign but simply an algebraic expression This is used for fully simultaneous models where it might not even be possible to normalize the equations The ANALYZ FIML LSQ and SIML procedures can process implicit equations Equations defined by FRML are the same as those defined by IDENT except that the estimation procedures assume that a FRML has an implied additive disturbance tacked on th
347. number For datasets with multiple observations per case you can also use LOGIT NREC W Y X Z W specifies the number of records per case and you do not need to supply an ID variable with the CASE option 3 Mixed logit The general form of the command is now LOGIT COND NCHOICEzn dependent variable conditional variables multinomial variables 236 LOGIT For example LOGIT COND NCHOICE 3 Y ZA ZB XA XB XC Y takes on the values 1 2 and 3 TSP looks for the conditional variables ZA1 ZA2 ZA3 ZB1 ZB2 ZB3 corresponding to the 3 choices XA XB XC are the multinomial variables The coefficients ZA ZB XA2 XB2 XC2 XA3 XB3 XC3 would be estimated with XA1 1 XC1 normalized to zero Output The printed output of the LOGIT procedure is similar to that of the other nonlinear estimation procedures in TSP A title is followed by a table of the frequency distribution of the choices Then the starting values and iteration by iteration printout is printed the amount is controlled by the PRINT and SILENT options This is followed by a message indicating final convergence status the value of the likelihood and a table of parameter estimates and their asymptotic standard errors and t statistics The variance covariance matrix of the estimates is also printed if it has been unsuppressed The DPDX or DPDZ matrices described below are printed unless they have been SUPRESed The following are also stored in d
348. o exit completely from TSP CLEAR would be preferred if you have just been experimenting with some commands or if you have created some variables in error and you want to start with a clean slate 81 Commands CLOSE Example CLOSE closes a file which has been opened by the READ or WRITE command Subsequent READ statements will read from the beginning of the file or subsequent WRITE statements will create a new file CLOSE can also be used to control access to more than 12 files in a single run or to insure that a newly updated file will be complete in case the program or computer aborts later CLOSE UNITz unit numbers FILE filename string Usage When TSP processes a READ or WRITE statement which accesses a file the file is left open so that further READ or WRITE statements will read data from or write data to the next line after those already processed This is useful when more observations or variables will be read from or written to the file later in the program There are several cases in which it may be useful to close the file 1 During an interactive session if an error is made reading or writing data closing the file will allow correction of the error when the corrected READ or WRITE statements are executed Without the CLOSE command a new READ statement would read from the current file position usually causing an end of file error message while a new WRITE statement would append to the lines already
349. o periods usually CHOWHET Stable Parameters and similar to F k nob 2k parameters variances usually variances differ differ between two periods LRHET Homoskedasticity Two variances for split Chi squared 1 sample LMHET Homoskedasticity Heteroskedasticity Chi squared 1 related to FIT 2 WHITEHT Homoskedasticity X related Chi squared k 1 k 2 Heteroskedasticity 1 BPHET Homoskedasticity Heteroskedasticity Chi squared vars in 383 Commands related to BPLIST BPLIST 1 FST Y constant Specified regression F k nob k model JB Normal Non normal Chi squared 2 disturbances SWILK Normal Non normal Shapiro Wilk disturbances RESETx No omitted power Higher order terms in F RESETORD terms Xs needed nob k T Slope coefficient Slope coefficient not T OLS IV nob k 0 zero Normal all other procs References Bhargava A L Franzini and W Narendanathan Serial Correlation and the Fixed Effects Model Review of Economic Studies XLIX 1982 pp 533 549 Brown R L Durbin J and Evans J M Techniques for Testing the Constancy of Regression Relationships Over Time Journal of the Royal Statistical Society Series B 1975 pp 149 192 Durbin J Tests for Serial Correlation in Regression Analysis Based on the Periodogram of Least Squares Residuals Biometrika 1969 Durbin J Boundary crossing probabilities for the Brownian motion and Poisson processes and techniques for computing the power of the
350. ocedures may not have N and thus W available A label is printed below each table of standard errors and asymptotic t statistics identifying the method of calculation used More than one method may be specified for alternative VCOV matrices and standard errors In this case the first method is stored in 9VCOV and SES and the VCOV for each method is stored under the name constructed by appending the letter to OVCOV For example HCOV NB would store VCOV VCOVN and VCOVB Consult the table below to see which option is the default in a particular procedure The P and Q options are for panel data and are only available for PROBIT and PANEL via HCOMEGA at the present time HCOV P computes grouped BHHH standard errors from the gradient of the objective function using the formula 1 Vp 556 ia t instead of the usual formula 1 Vs gt 6 6 where Gj is the gradient vector for individual i and period t Unlike the usual formula this version of the estimate does not assume independence within individual across different time periods 4 HCOV Q computes the robust version of this matrix N ers N is the Newton inverse second derivative matrix See Wooldridge p 407 For linear models this matrix is exactly equivalent to that computed by PANEL OLSQ and 2SLS using the HCOMEGA BLOCK option Note that the rank of Vp is at most NI where NI is the number of individuals so that when Nl is less than
351. ocial Measurement October 1974 pp 653 665 Calzolari Giorgio and Gabriele Fiorentini Alternative Covariance Estimators of the Standard Tobit Model Paper presented at the World Congress of the Econometric Society Barcelona August 1990 Calzolari Giorgio and Lorenzo Panattoni Alternate Estimators of FIML Covariance Matrix A Monte Carlo Study Econometrica 56 1988 pp 701 714 Fletcher R Practical Methods of Optimization Volume Unconstrained Optimization John Wiley and Sons New York 1980 Gill Philip E Walter Murray and Margaret H Wright Practical Optimization Academic Press New York 1981 284 Nonlinear Options Goldfeld S M and R E Quandt Nonlinear Methods in Econometrics North Holland 1972 Greene William H On the Asymptotic Bias of the Ordinary Least Squares Estimator of the Tobit Model Econometrica 49 March 1981 pp 505 513 Quandt Richard E Computational Problems and Methods in Griliches and Intriligator eds Handbook of Econometrics Volume I North Holland Publishing Company Amsterdam 1983 White Halbert Maximum Likelihood Estimation of Misspecified Models Econometrica 50 1982 pp 1 2 White Halbert A Heteroskedasticity Consistent Covariance Matrix and a Direct Test for Heteroskedasticity Econometrica 48 1980 721 746 Wooldridge J M Econometric Analysis of Cross Section and Panel Data Cambridge MA MIT Press 2002
352. of the variables by setting its coefficient to zero and printing a warning The results for these special regressions become more accurate when OPTIONS DOUBLE is in effect because these data have an unusually high number of significant digits See the Benchmarks section of the TSP web page for more information on regression accuracy with Longley and other standard regression datasets Options HCOMEGA BLOCK or DIAGONAL specifies the form of the E uu7 matrix to use when computing ROBUST standard errors Ordinarily the default is diagonal which yields the usual robust standard errors When FREQ PANEL is in effect the default is BLOCK which allows for cross time correlation of the disturbances within individuals This feature can be used for any kind of grouped data simply by ensuring that the relevant PANEL setup has been defined HCTYPE the type of heteroskedastic consistent standard errors to compute between 0 and 3 with a default of 2 This option implies ROBUSTSE In general the robust estimate of the variance covariance matrix has the form V X xr Se ta xy The option HCTYPE specifies the formula used for HCTYPE d i description 294 OLSQ 0 1 the usual Eicker White asymptotic formula T k T use finite sample degrees of freedom 1 h i unbiased if e is truly homoskedastic 1 h i jacknife approximation squared where h i is the diagonal of the hat matrix defined below If 1
353. ogin tsp file are the following OPTIONS MEMORY approximate memory to be used by TSP Megabytes This option only works if OPTIONS is the first command in the run or the first command in the login tsp file INPUT some file that transforms the data or selects the data you are using for a number of TSP programs 14 Command summary Display Commands ASMBUG prints debug output during parsing of TSP commands DATE DEBUG DIR DOC GRAPH GRAPH HELP HIST HIST NAME PAGE PLOT PLOT PRINT SHOW SYMTAB TITLE TSTATS prints current date on screen or printout prints debug output during execution of TSP commands lists files available in current directory interactive adds descriptions to variables graph one variable against another graphics version graphs one variable against another in a scatter plot prints command syntax one way bar chart for variable graphics version one way bar chart for variable specifies name and title for TSP run starts a new page in printout plots several variables versus time graphics version plots several variables versus time prints variables lists currently defined TSP variables by category SERIES etc debug version of SHOW specifies new title for run or immediate printing prints table of coefficients and t statistics 15 Command summary Options Commands FREQ NOPLOT NOREPL OPTIONS PLOTS REPL SELECT SMPL SMPLIF 16 set the data frequency None Annual
354. oise residual series at 57 Commands If the PLOT option is on a time series plot of the residuals at is printed using high resolution graphics in the Givewin version If the CUMPLOT option is on a normalized cumulative periodogram is also plotted see the CUMPLOT option above The following variables are stored statistics based on differenced variable if differencing is specified variable LHV RNMS SSR S S2 YMEAN SDEV DW RSQ ARSQ IFCONV LOGL NCOEF NCID COEF SES T GRAD VCOV QSTAT 05 ARSTAT MAINV FIT RES ARRTRE ARRTIM 58 type list list scalar scalar scalar scalar scalar scalar scalar scalar scalar scalar scalar scalar vector vector vector vector matrix vector vector scalar scalar series series vector vector length 1 vars ol n ln 3 1 1 1 coefs coefs NLAG NLAG 1 1 obs obs NAR NAR description Name of the dependent variable Names of right hand side parameters Sum of squared residuals Standard error of the regression Standard error squared Mean of the dependent variable Standard deviation of the dependent variable Durbin Watson statistic R squared Adjusted R squared 1 if convergence achieved 0 otherwise Log of likelihood function Number of coefficients Number of identified coefficients Co
355. ollowing the THEN statement will be executed if the result of IF is true If you want more than one statement executed when the_IF clause is true enclose all the statements in a DO ENDDO group Example IF LOGL gt OLDL THEN DO SET OLDL LOGL COPY COEF SAVEB ENDDO See also the examples for the ELSE statement 429 Commands 3SLS Options Example References 3515 obtains three stage least squares estimates of a set of nonlinear equations It is a special case of LSQ with the options set for 3SLS estimation The LSQ entry has a more complete description of the command Three stage least squares estimates are consistent and asymptotically normal and under some conditions asymptotically more efficient than single equation estimates In general 3SLS is asymptotically less efficient than FIML unless the model is linear in the parameters and endogenous variables 3SLS COVU OWN or lt residual covariance matrix gt DEBUG FEI HETERO INST lt list of instrumental variables gt ITERU MAXITW 0 ROBUST nonlinear options lt list of equation names gt Usage Three stage least squares is a combination of multivariate regression SUR estimation and two stage least squares It obtains instrumental variable estimates taking into account the covariances across equation disturbances as well The objective function for three stage least squares is the sum of squared transformed fitted residuals Specific
356. ons see the MATRIX command LOG Natural logarithm EXP Exponential function ABS Absolute value LOG10 Log base 10 SQRT Square root SIN Sine argument in radians COS Cosine argument in radians TAN Tangent argument in radians ATAN Arctangent answer in radians NORM Standard normal density CNORM Standard normal cumulative distribution function CNORMI Inverse of the standard normal cumulative distribution function LNORM Log of normal density LCNORM Log of cumulative normal DLCNORM Derivative of LONORM inverse Mills ratio GAMFN Gamma function not Gamma density LGAMFN Log of Gamma function DLGAMFN Derivative of LGAMFN DIGAMMA TRIGAMMA Derivative of DIGAMMA non differentiable FACT Factorial FACT X X GAMFN X 1 LFACT Log of factorial SIGN Sign 1 for lt 0 0 for X 0 1 for gt 0 deriv 0 POS Positive POS X max 0 X Note min A B B POS B A max A B A POS B A MISS Missing 1 for X missing 0 otherwise non differentiable INT Integer truncate round towards 0 non differentiable CEIL Ceiling round away from 0 non differentiable ROUND Round to nearest integer 5 rounds to 1 non 10 differentiable Character Set for TSP Character Set for TSP Character Symbol Use Parts of names Lowercase letters are allowed letter A to on most computers they are treated like 2 uppercase letters cannot be u
357. operator NOT NE or power 11 Introduction tilde less than greater than continuation miscellaneous 12 lt gt depending on context Logical operator NOT NE Relational operator LT LE Relational operator GT GE Continuation of a line interactive Reserved for future use Missing Values in TSP Procedures Missing Values in TSP Procedures Procedures that drop Procedures that cannot execute if observations containing the sample contains missing missing values values AR1 OLSQ INST 2SLS LIML ACTFIT LAD LSQ FIML GMM ML SUR BJEST BJIDENT BUFRCST 3SLS PROBIT TOBIT SAMPSEL COINT UNIT LOGIT CONVERT FORCST GENR ARCH KALMAN MSD CORR COVA MOM DIVIND SAMA GRAPH PLOT HIST SOLVE SIML PANEL VAR PRIN 13 Introduction LOGIN TSP file login tsp is a special INPUT file it is read automatically at the start of interactive sessions and batch jobs This is useful for setting default options for a run If you use the same options repeatedly you may want to place them in a login tsp file Every time TSP starts it checks for a login tsp file and executes it first Normally TSP looks for login tsp in your working directory If it does not find one it looks in the directory in which you installed TSP for DOS and Windows in the folder in which you installed TSP for Macs and in the home directory on Unix Commands which can be usefully issued in a l
358. optimum The grid points used are 0 1 2 8 85 9 95 99 9999 1 2 8 85 9 95 99 9999 Sometimes the global optimum shows RHO 1 0000 or 1 0000 In these cases the actual estimate of RHO is slightly less than 1 in absolute value and the residual covariance matrix is nearly singular The standard error of RHO and its covariance with other parameters is set to zero in these cases Options MILLS name of variable where the inverse Mills ratio should be stored The default is MILLS Standard nonlinear options see NONLINEAR HITER N HCOVEN is the default Example SAMPSEL PRINT MAXIT 50 HCOV NBW IY C Z YCX References Amemiya Takeshi Advanced Econometrics Harvard University Press Cambridge Massachusetts 1985 Chapter 13 Bloom David E and Killingsworth Mark R Correcting for Truncation Bias caused by a Latent Truncation Variable Journal of Econometrics 1985 pp 131 135 Griliches Z B H Hall and J A Hausman Missing Data and Self selection in Large Panels Annales de I Insee Avril Sept 1978 pp 137 176 395 Commands Heckman James J Sample Selection Bias as a Specification Error Econometrica 47 1974 pp 153 162 Maddala G S Limited Dependent and Qualitative Variables Econometrics Cambridge University Press Cambridge 1983 Chapter 6 Nawata Kazumitsu Estimation of Sample Selection Bias Models by the Maximum Likelihood
359. or individuals p values for T statistics of fixed effects FEI If the regression includes a PDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling REGOPT NOPRINT LAGF will turn off the lag plot for PDL variables Options FEI NOFEI specifies that a model with individual specific effects is to be computed FREQ PANEL must be in effect FEPRINT NOFEPRINT specifies that the fixed effect estimates are to be printed as well as stored 197 Commands HCOMEGA BLOCK or DIAGONAL specifies the form of the Efuu matrix to use when computing ROBUST standard errors Ordinarily the default is diagonal which yields the usual robust standard errors When FREQ PANEL is in effect the default is BLOCK which allows for cross time correlation of the disturbances within individuals This feature can be used for any kind of grouped data simply by ensuring that the relevant PANEL setup has been defined INST list of instrumental variables or the name of a list containing instrumental variables NORM UNNORM tells whether the weights are to be normalized so that they sum to the number of observations This has no effect on the coefficient estimates and most of the statistics but it makes the magnitude of the unweighted and weighted data the same on average which may help
360. or example LMS CONS C GNP estimates the consumption function The PRINT option enables the printing of the outliers or you can define a set of outliers for printing by screening on the residuals which are stored in RES Output The usual regression output is printed and stored see OLSQ for further discussion The number of possible subsets and the best subset found are also printed The following results are stored variable type length description LHV list 1 name of dependent variable RNMS list vars list of names of independent variables YMEAN scalar 1 mean of dependent variable SDEV scalar 1 standard deviation of dependent variable SSR scalar 1 sum of squared residuals 952 scalar 1 standard error squared S scalar 1 standard error of estimate RSQ scalar 1 R squared ARSQ scalar 1 adjusted R squared 229 Commands LMHET LMHET QDW PHI STDDEV NCOEF NCID COEF SES T VCOV RES FIT Method scalar scalar scalar scalar scalar scalar scalar vector vector vector vector matrix series series 1 vars vars vars vars vars vars obs obs LM test for heteroskedasticity p value for heteroskedasticity test Durbin Watson statistic median of squares for final estimate standard deviation of residuals dropping outliers number of coefficients variables number of identified coefficients lt NCOEF vect
361. or later use The table below lists the results available after a PROBIT command variable type length description LOGL scalar Log of likelihood function IFCONV scalar Convergence status 1 success NOB scalar Number of observations NPOS scalar Number of positive observations SRSQ scalar Scaled R squared for binary probit 2l cla ll RSQ scalar Squared correlation between Y and FIT SSR scalar 1 Sum of squared residuals RNMS list params List of parameter names GRAD vector Gradient of likelihood function at maximum COEF vector params Estimated values of parameters SES vector params Standard errors of estimated parameters QT vector T statistics vector params p values for T statistics 346 PROBIT VCOV vector par par Estimated variance covariance of estimated parameters DPDX matrix vars 2 Matrix of mean probability derivatives for the two values of the dependent variable MILLS series obs Inverse Mills ratios FIT series obs Fitted probabilities NCOEFAI scalar 1 Number of fixed effects NCIDAI scalar 1 Number of identified fixed effects AI series obs estimated fixed effects stored as a series for FEI COEFAI vector individuals estimated fixed effects for FEI SESAI vector individuals standard errors for fixed effects for FEI TAI vector individuals T statistics for fixed effects for FEI 9eTAI vector individuals p values corresponding to
362. or of estimated coefficients standard errors of estimated coefficients T statistics for estimates null is zero p values corresponding to T statistics estimated variance covariance of estimated coefficients residuals dependent variable fitted values fitted values of dependent variable The LMS estimator minimizes the square or the absolute value of the median residual with respect to the coefficient vector b min median y X b t Clearly this ignores the sizes of the largest residuals in the sample i e those whose absolute values are larger than the median so it will be robust to the presence of any extreme data points outliers 230 LMS If there are K independent variables excluding the constant term LMS will consider many different subsets of K observations each An exact fit regression line is computed for each subset and residuals are computed for the remaining observations using these coefficients The residuals are essentially sorted to find the median and a slight adjustment is made to allow for the constant term If K or the number of observations is large the number of subsets could be very large and sorting time could be lengthy so random subsets will usually be considered in this case This is controlled with the ALL MOST and SUBSETS options The result is not necessarily the global Least Median of Squares optimum but a feasible close approximation to it with the same properties for outli
363. or this procedure you can execute a series of closely related regressions by entering the first estimation command followed by a series of ADD and DROP commands Since each ADD or DROP permanently alters the command each new modification must take all previous modifications into account Note that it is not possible to combine ADD and DROP into one step to perform a REPLACE function or to make compound modifications to a command In these circumstances RETRY must be used 125 Commands Output DROP echoes the modified command it will execute Any further output will be a direct result of the command that is executed by it Examples OLSQ WEIGHT POP YOUNG C RSALE URBAN CATHOLIC DROP URBAN will run two regressions the second of which is 0150 WEIGHT POP YOUNG C RSALE CATHOLIC This is also how the command will now look if you REVIEW it since it has been modified and replaced TSP s internal storage and in your backup file 126 DUMMY DUMMY Options Examples DUMMY creates a set of zero one variables which correspond to the different values taken by an input series The number of dummy variables created is equal to the number of unique values of the input series unless EXCLUDE is specified DUMMY EXCLUDE lt gt series listname or list of gt Usage DUMMY followed by the name of a series will cause the variables NAME1 NAME2 etc to be created where N
364. ormation Criterion AIC scalar 1 Akaike Information Criterion IFCONV scalar 1 1 if convergence achieved 0 otherwise NCOEF scalar 1 Number of coefficients number in probit number in regression 2 NCID scalar 1 Number of identified coefficients COEF vector itcoeffs Coefficient estimates SES vector itcoeffs Standard errors QT vector coeffs T statistics GRAD vector coeffs Gradient of log likelihood at convergence VCOV matrix coeffs Variance covariance of estimated coeffs coefficients DPDX matrix selvars 2 Mean of probability derivatives for selection equation RES series obs Residuals for the observed sample MILLS series obs Inverse Mills ratios If the regression includes a PDL variable the following will also be stored 394 SAMPSEL SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling Method The method used is maximum likelihood obtained by means of a gradient method that uses the Hessian approximation given by the HITER option Since this likelihood function is known to have multiple local optima frequently the method of Nawata 1994 1995 1996 is used to find the global optimum In Nawata s method a grid search is done on the correlation coefficient RHO to find the set of local optima Then further iterations are done to refine the optima to full precision and choose the global
365. ots 442 UPDATE ciii datas 444 USER 445 V tens intent 446 W WRITE s iic deis 449 Y 453
366. ou can read the file as a matrix transpose it and UNMAKE the matrix into series TSP checks the first row in the file for string names cells beginning with the characters or The names are truncated to 8 characters if necessary and are translated to uppercase They must be aligned above their corresponding data columns If you have dates in the first column no name is required for the date column Any names with imbedded blanks will be ignored Output READ prints the data as it is read when the free format option is used and the PRINT options is on otherwise READ produces no output and stores all the data read in data storage under the series names specified In some cases a few special variables may be stored in data storage variable type length description NOB scalar 1 Number of observations read when SETSMPL is in effect 366 READ RNMS list vars Variable names read for spreadsheets when no names are given Options BYOBS NOBYOBS specifies that the data is organized by observation the first observation for all series then the second observation for all series etc This is the default BYVAR NOBYVAR specifies that the data is organized by series all observations for the first series then all observations for the second series etc FILE filename string or filename specifies the actual name of the file where your data is stored If the filename string is 8 characters or less and does not
367. ow resolution with characters so they can be included in standard output files and implemented on all platforms The CUSUM plot stores CUSUM CSUB5 and CSLB5 The CUSUMSQ plot stores CUSUMSQ CSQMEAN CSQUB5 and CSQLB5 These variables can be stored without displaying the low resolution plot if you want to make only high resolution plots To do this use REGOPT CALC NOPRINT CUSUM CUSUMSQ or PLOTS CUSUM CUSUMSQ SUPRES CUSUM CUSUMSQ To then make high resolution plots on the 386 486 and Mac versions of TSP use PLOT ORIGIN PREVIEW 2CUSUM CSUB5 CSLB5 PLOT PREVIEW CSQMEAN CUSUMSQ CSQUB5 CSQLB5 335 Commands TSP only calculates the bounds automatically for the 5 two tailed tests Bounds for other size one tailed tests can be calculated manually with simple transformations of the bounds which are stored using the appropriate critical values from a table References Brown R L J Durbin and J M Evans Techniques for Testing the Constancy of Regression Relationships over Time Journal of the Royal Statistical Society B 1975 Durbin J Tests for Serial Correlation in Regression Analysis Based on the Periodogram of Least Squares Residuals Biometrika 1969 Edgerton David and Curt Wells On the Use of the CUSUMSQ Statistic in Medium Sized Samples Oxford Bulletin of Economics and Statistics 1994 Harvey A C The Econometric Analysis of Time Series 2nd ed Philip All
368. p 32 61 Cooper J Phillip Asymptotic Covariance Matrix of Procedures for Linear Regression in the Presence of First Order Autoregressive Disturbances Econometrica 40 1972 pp 305 310 Davidson Russell and MacKinnon James G Estimation and Inference in Econometrics Oxford University Press New York NY 1993 Chapter 10 This is the best single reference Dufour J M Gaudry M J Il and Liem C Cochrane Orcutt Procedure Numerical Examples of Multiple Admissible Minima Economics Letters 6 1980 pp 43 48 Fair Ray C The Estimation of Simultaneous Equation Models with Lagged Endogenous Variables and First Order Serially Correlated Errors Econometrica 38 1970 pp 507 516 Fair Ray C Specification Estimation and Analysis of Macroeconomic Models Harvard University Press Cambridge MA 1984 Hildreth C and Lu J Y Demand Relations with Autocorrelated Disturbances Research Bulletin 276 Michigan State University Agricultural Experiment Station 1960 Judge et al The Theory and Practice of Econometrics John Wiley amp Sons New York 1981 Chapter 5 Maddala G S Econometrics McGraw Hill Book Company New York 1977 pp 274 291 47 Commands Pindyck Robert S and Rubinfeld Daniel L Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 pp 106 120 Prais S J and Winsten C B Trend Estimators and Serial Correlation Cowles
369. pecifies a field length of 10 columns 16 to 25 and the 3 that there are 3 digits to the left of the decimal point Since the decimal point is specifically included the number is read as 98765 times 10 1 or 098765 The final format is F8 0 which specifies a field length of 8 columns 26 to 33 and no digits to the right of the implied decimal point In this case the number to be input has no decimal point and will be read as 200 Formats can be combined in many ways for example an integer number prefacing a format specification tells how many such fields should be read Here are several examples 8F10 5 specifies an 80 column record with eight numbers of ten columns each 10X 5E12 6 10X 8F3 1 specifies a 104 column record with 13 variables 5 in E format and 8 in F format Columns 1 to 10 and 71 to 80 will be skipped when reading 20F5 2 20F5 2 specifies two records per observation the means to skip to a new record each with 20 variables This last example demonstrates an important feature of formatted data loading in TSP In general one observation will be read or written with each pass through the format statement That is the format statement should allow for exactly as many numbers as there are variables to be loaded Usually one record will be read for each observation which contains all the variables for that observation but it is possible to have more than one record per observation by use of the slash for
370. pend on XB so XB is substituted in last There is no need to substitute XB into E separately A separate substitution would still operate correctly but it would result in larger and less efficient code The new FRML TOBIT1 is created and the original TOBIT is left untouched for later use On the other hand normalized input equations are recommended for parameter restrictions The same FRMLs can usually be used to both impose parameter restrictions and to evaluate the restricted parameters after estimation with ANALYZ If the input equation is normalized ANALYZ can store the restricted parameter name For example in a translog model with symmetry imposed FRML EQ1 SH1 A1 B11 LP1 B12 LP2 B13 LP3 FRML EQ2 SH2 A2 B12 LP1 B22 LP2 B23 LP3 Note The last share equation is not used in estimation due to singularity SH3 B13 LP1 B23 LP2 B33 LP3 homogeneity adding up constraints FRML R13 B13 B11 B12 FRML R23 B23 B12 B22 EQSUB EQ1 R13 EQSUB EQ2 R23 left out params from final equation FRML LO1 A3 1 A1 A2 note this also depends on the restricted B23 FRML LO2 B33 B12 B23 EQSUB LO2 R23 LSQ EQ1 2 Estimate model with restrictions in place Print and store values and standard errors for A3 and B33 ANALYZ 101 102 EQSUB also handle lagged dependent variables FRML U Y A B X G Z 2 FRML E U RHO U 1 EQSUB U E is equivalen
371. performing a second estimation without having to fully retype the command It is not however restricted to this usage and may be used in any circumstance where this type of command modification is needed The command ADD var1 var2 and the sequence RETRY gt gt INSERT 1 gt gt INSERT var2 gt gt EXIT are identical in function since both permanently modify the previous command by inserting var1 and var2 at the end of the command The command is then automatically executed in both cases The only potential difference between these approaches besides the amount of typing is in the definition of previous RETRY with no line number argument assumes you want to modify the last line typed ADD will not accept a line number argument and always modifies the last line that is not itself an ADD or DROP command ADD and DROP allow you to execute a series of closely related regressions by entering the first estimation command followed by a series of ADD and DROP commands Since each ADD or DROP permanently alters the command each new modification must take all previous modifications into account Notes It is not possible to combine ADD and DROP into one step to perform a replace function or to make compound modifications to a command In these circumstances RETRY must be used Examples 33 Commands 0150 WEIGHT POP YOUNG C RSALE URBAN CATHOLIC ADD MARRIED will run two regressions the second of which is
372. ple when PERIOD is used The default is one For example FREQ Q SMPL 70 2 79 4 TREND FREQ QT 438 TREND is equivalent to SMPL 2 40 TREND PERIOD 4 PSTART 2 QT 5 gives an initial value to the trend The default is one STEP supplies a value for the trend increment The default is one Examples FREQ SMPL 46 75 TREND TIME The above example creates a time trend variable called TIME for the illustrative model The variable is 1 in 1946 2 in 1947 3 in 1948 and so forth FREQ Q SMPL 70 2 79 4 TREND FREQ QT This example makes a series QT equal 2 3 4 1 2 3 4 1 2 3 4 etc SMPL 1 400 TREND PERIOD 5 TIME This example makes a series which is 1 2 3 4 5 starting in every fifth observation this might be useful for panel data where a trend was needed SMPL 1 400 TREND PERIOD 5 START 71 YEAR This example is the same as above except the trend is equal to 71 72 73 74 75 71 72 73 74 75 instead of 1 2 3 4 5 1 2 3 4 5 439 Commands TSTATS Output Option Example TSTATS prints a table of names coefficients standard errors t statistics and if it is not suppressed using REGOPT a variance covariance matrix TSTATS is useful when you compute your own estimate of a variance covariance matrix since it is tedious to take the square roots of the diagonal elements and generate a readable printout of the results TSTATS NAMES lt list of names
373. pp 275 302 Gallant A Ronald and Alberto Holly Statistical Inference in an Implicit Nonlinear Simultaneous Equation Model in the Context of Maximum Likelihood Estimation Econometrica 48 1980 pp 697 720 40 AR1 AR1 Output Options Examples References AR1 obtains estimates of a regression equation whose errors are serially correlated These estimates are efficient if the disturbances in the equation follow an autoregressive process of order one The estimates may be obtained using one of two different objective functions exact maximum likelihood which imposes stationarity by constraining the serial correlation coefficient to be between 1 and 1 and keeps the first observation for estimation or by Generalized Least Squares GLS which drops the first observation FAIR FEI INST list of instrumental variables METHOD CORC or HILU or ML or MLGRID OBJFN EXACTML or GLS REI RMIN lt minimum rho value RMAX maximum rho value RSTART lt siart value for rho RSTEP lt step value for rho TSCS nonlinear options dependent variable name list of independent variables To obtain estimates of a regression equation which are corrected for first order serial correlation use the AR1 command as you would an OLSQ command PDL polynomial distributed lag variables may be included in an AR1 statement See the PDL section for a further description of how to specify these variables TSP automatically d
374. ption equation FORM and the second generates an identity to define the variable on the left hand side This ensures that the identity will hold in the data which is necessary for proper estimation SMPL 1 10 A 1 1 SMPL 2 10 GENR A A 1 1 GENR STATIC B B 1 1 This creates A as a time trend just like the TREND A command B is 1 2 2 2 2 2 2 2 2 2 2 169 Commands GMM Output Options Examples References GMM does General Methods of Moments estimation on a set of orthogonality conditions which are the products of equations and instruments Initial conditions for estimation are obtained using three stage least squares The instrument list may be different for each equation see the MASK option or the INST option and the form of the covariance matrix used for weighting the estimator is under user control the HETERO option for heteroskedastic consistency and the NMA option for moving average disturbances GMM FEI HETERO ITEROC ITERU LSQSTART COVOC OWN or lt covariance matrix of orthogonality conditions gt COVU lt covariance matrix of residuals gt INST lt list of instruments gt KERNEL lt spectral density kernel type gt MASKz matrix of zeros and ones NMA number of autocorrelation terms OPTCOV nonlinear options list of equations Usage List the instruments in the INST option and list the equations after the options the products of these two are the orthogonality con
375. quency is annual The weekly frequency assumes exactly 52 weeks per year and CONVERT s to quarterly but not to monthly TSP does not have a calendar If the frequency specified is a number it represents the number of periods per year The default frequency is none This frequency is provided for convenience in dealing with non time series data the data are assumed to be numbered from observation one unless you specify the sample otherwise Output The scalar variable FREQ is stored with the value of the current frequency This variable can be used to restore a frequency and sample by a PROC For example COPY SMPL SMPSAV COPY FREQ FRQSAV 163 Commands can be used at the start of the PROC and FREQ FRQSAV SMPL SMPSAV can be used at the end Examples FREQA SMPL 1890 1920 specifies data with annual frequency running from 1890 to 1920 FREQ QUARTER SMPL 47 1 82 4 specifies data with quarterly frequency running from the first quarter of 1947 to the fourth quarter of 1982 FREQ 26 specifies data with a biweekly frequency FREQ PANEL Tz5 specifies balanced panel data with 5 observations for each individual and no particular time series frequency FREQ PANEL IDZCUSIP STAHTz 1974 A specifies possibly unbalanced panel data with an ID series CUSIP that distinguishes each individual and annual data starting in 1974 for each individual 164 FRML FRML Examples FRML defines equatio
376. r NOVID scalar RNMS list COEF vector SES vector T vector SSR vector YMEAN vector SDEV vector S vector DW vector RSQ vector ARSQ vector OC vector COVOC matrix COVU matrix length 1 1 1 1 params params params params eqs eqs eqs eqs eqs eqs eqs eqs inst eqs inst by eqs inst eqs eqs description E HH E the objective function for instrumental variable estimation test of overidentifying restrictions PHI NOB P value of the above test using degrees of freedom number of overidentifying restrictions degrees of freedom Parameter names Estimated values of parameters also stored under their names Standard Errors of estimated parameters T statistics Vector of sum of the squared residuals for each of the equations Vector of means of the dependent variable for each of the equations Vector of standard deviations of the dependent variable for each of the equations Vector of standard errors for each of the equations Vector of Durbin Watson statistics for each of the equations Vector of R squared for each of the equations Vector of adjusted R squared for each of the equations Estimated orthogonality conditions Estimated covariance of orthogonality conditions Residual covariance matrix 171 Commands oW matrix eqs eqs Inverse square root of COVU the upper triangular weighting matrix COVT matrix eqs eqs Covariance matrix of
377. r Complex eigenvalues are sorted by their norm qm Computes the matrix of eigenvectors columns If EIGVEC qm qm is not symmetric positive semi definite the imaginary parts of the eigenvectors are stored as EIGVECI v VEC m Creates a vector of all the elements of m column by column v VECH m Creates a vector of all the unique elements of m column by column qm N N elements sm tm N N 1 2 elements dm N elements dm DIAG m Creates a diagonal matrix from a matrix sm tm take the diagonal from input matrix v convert the vector to a diagonal matrix s illegal use s IDENT to create a diagonal matrix with s on the diagonal sm SYM qm Creates a symmetric matrix from a square matrix the upper triangular elements are ignored m GEN qm Creates a general matrix from a symmetric or diagonal matrix series SER v Converts a vector to a series under control of SMPL Output MATRIX produces no printed output Typically one matrix is stored in data storage Examples MAT B X X X Y produces OLS regression coefficients not a very accurate way to do this The example below computes the Eicker White estimate of the variance covariance of the estimated coefficients after a regression OLSQYCX MMAKE XMAT C X XXI XMAT XMAT MAT VCOV XMAT XXI DIAG RES 2 XMAT XXI 253 Commands MFORM Options Examples MFORM forms or reforms matrices used to make a matrix
378. r END If used in collect mode this command will return the user to the interactive level of the program WITHOUT executing the commands just collected use the EXEC command to leave collect mode with automatic execution EXIT may also be used to replace END at the very end of an INPUT file to suppress automatic execution upon completion of reading the file see INPUT 142 FETCH FETCH Example Reference FETCH reads microTSP and EViews format databank files FETCH list of series FETCH disk seriesname disk seriesname Usage MicroTSP format databank files may be useful for transferring data between microTSP EViews and TSP They are plain editable non binary files containing comments frequency starting and ending dates and data values one per line See the microTSP EViews documentation for details They are not efficient in terms of disk space usage or the time required to read or write them However they are easy to edit for manual data revision To fetch a series from a microTSP databank name series DB use the command FETCH series series DB must be in the default directory To move regular TSP databanks between machines such as VAX VMS to a personal computer use the DBCOPY command The STORE command creates the files read by FETCH Example FETCH X Y reads the series X and Y and any imbedded comments and documentation from the files X DB and Y DB Reference Hall Robert E and Lili
379. r later use The table below lists the results available after an INST command The fitted values and residuals will only be stored if the RESID option is on the default variable type length description LHV list 1 Name of the dependent variable RNMS list vars Names of right hand side variables SSR scalar 1 Sum of squared residuals S scalar 1 Standard error of the regression 952 scalar 1 Standard error squared YMEAN scalar 1 Mean of the dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations 196 INST DW scalar 1 Durbin Watson statistic RSQ scalar 1 R squared ARSQ scalar 1 Adjusted R squared FST scalar 1 pseudo F statistic for zero slopes 1 FOVERID scalar test of overidentifying restrictions when inst gt vars PHI scalar 1 The objective function e P Z e COEF vector vars Coefficient estimates SES vector vars Standard errors T vector vars T statistics VCOV matrix vars vars Variance covariance of estimated coefficients RES series obs Residuals actual fitted values of the dependent variable FIT series obs Fitted values of the dependent variable AI series estimated fixed effects stored as a obs series for FEI COEFAI vector estimated fixed effects for FEI individuals SESAI vector individuals standard errors of fixed effects for FEI TAI vector individuals T statistics for fixed effects FEI vect
380. r of observations for Dickey Fuller also see the NOB option for Dickey Fuller Non integers allowed for the chi squared DF1 the numerator degrees of freedom for the F distribution can be non integer DF2 the denominator degrees of freedom for the F distribution can be non integer EIGVAL vector of eigenvalues for the weighted chi squared distribution INVERSE NOINVERSE specifies the inverse distribution function input is significance level output is critical value Normally the input is a test statistic and the output is a significance level INVERSE is not defined for bivariate normal LOWTAIL TWOTAIL UPTAIL specifies the area of integration for the density function TWOTAIL is the default for most symmetric distributions normal and t UPTAIL is the default for chi squared and F and LOWTAIL is the default for bivariate normal and Dickey Fuller TWOTAIL is not defined for bivariate normal NLAGS the number of lagged differences in the augmented Dickey Fuller test This number is used to compute the approximate finite sample P value or critical value The default is zero assume an unaugmented test The NOB option must also be specified for the finite sample value NOB the number of observations for the augmented Dickey Fuller or Engle Granger tests This number is used to compute the approximate finite sample P value or critical value The default is zero to compute asymptotic instead of finite sample value NVA
381. r of the forecast for the N th period following the origin the N th step ahead The psi weights are the coefficients obtained by expressing the model as a pure moving average The N th row entry for Psi Phi or Theta gives the coefficient for the N th lag The next feature of the output is a table of the forecasts their upper and lower confidence bounds the actual values of the time series when available and the discrepancies between the forecasts and the realizations of the time series If PLOT is specified the values in this table are plotted in high resolution graphics in the Givewin version The following variables are stored variable type length description FIT series NHORIZ Forecast for last origin if there was more than one FITSE series NHORIZ Forecast standard errors still in logs for EXP option Method 64 BJFRCST BJFRCST forms the generalized autoregressive and moving average lag polynomials phi and theta These lag polynomials define simple difference equation for the time series The forecasts are generated using this difference equation As in BJEST backcasted residuals are calculated to obtain initial conditions for the difference equation Chapter 5 of Box and Jenkins 1976 contains a detailed explanation and analysis of this procedure Options Note that for all the Box Jenkins procedures BJIDENT BJEST BJFRCST TSP remembers the options from the previous Box Jenkins command so that
382. rage DBLIST DATE DOC SILENT list of filenames Usage Follow the word DBLIST with the filenames of the TSP databanks to be listed A list of all the variables contained in each databank is printed along with brief information about each variable in the format of the SHOW command If the databank contains wasted space which could be removed with the command the amount of wasted space is indicated The FREQ and SMPL information on series may be useful for setting up a DBPRINT command Output A list of the contents of the databanks will be printed The list of names of the databank variables is stored RNMS for further use Options DATE NODATE specifies whether the date is to be printed DOC NODOC specifies whether series documentation is to be printed SILENT NOSILENT suppresses all printed output RNMS is stored Example Suppose that you have a databank called FOO TLB which contains two time series X and Y and a parameter B The command DBLIST FOO would print the following information CONTENTS OF DATABANK FOO TLB Class Name Description SCALAR B parameter 3 14000 SERIES X 104 obs from 1960 1 1985 4 quarterly Y 104 obs from 1960 1 1985 4 quarterly 108 DBPRINT Databank DBPRINT Databank Example DBPRINT prints all the series in a TSP databank under the control of the current FREQ and SMPL DBPRINT lt filename gt Usage Follow the word DBPRINT with
383. rameter a starting value use a CONST statement to fix the parameter throughout the estimation Output CONST produces no printed output it stores the variables named in data storage Examples CONST DELTA 15 CONST 1 2 4 LIST ALIST A1 44 CONST ALIST PARAM ALPHA 1 0 BETA 5 CONST ALPHA BETA GAMMA 9 The second and third of these examples have the same effect The fourth assigns a value only to the third variable GAMMA the other two variables have the same value as they did previously but their type is changed from PARAM to CONST 98 CONVERT CONVERT Options Examples CONVERT changes series from one frequency to another The options specify the method used for conversion averaging the data using the first middle or last observation or summing the data Interpolation is optionally available for converting to higher frequencies CONVERT AVERAGE or FIRST or MID or LAST or SUM INTERPOL lt series gt SMPL lt list of series names gt or lt newseries gt lt oldseries gt Usage Use CONVERT after specifying the frequency you want to convert to with a FREQ statement The frequency you convert from will be that of the series to be converted The SMPL information is ignored so that the entire series is converted this avoids the confusion of possibly mixing frequencies in the same series The first form of the command simply converts the series and stores it back in data storag
384. rder the SORT command can be used to reorder them you could also sort the data by year and then individual if you wish to do variance components in the time dimension Usually it is best to use the FREQ PANEL command at the top of your run to specify such ID variables internal frequency and starting date etc Then these options will be used for all PANEL AR1 GENR etc commands within the run 312 PANEL The models you wish to estimate are specified in the options list The default is to estimate the total between within and variance components models For the VARCOMP random effects model there are additional options that specify how to compute the variance components Small or large sample formulas may be used or the user can supply the values directly If negative variances are computed using the small sample method the method switches over to the large sample formulas which always result in positive values PANEL also computes a Hausman test for correlated effects by comparing the WITHIN fixed effects and VARCOMP random effects estimators The REI and REIT options are used to obtain maximum likelihood estimates of the one and two way random effects models Output The output begins with a title and a summary of the panel structure number of individuals NI number of time periods T and total number of observations NOB If the data are unbalanced TMIN and TMAX will be printed For each estimator a table of
385. re 19 December 1981 pp 1483 1536 Cameron A Colin and Frank A G Windmeijer An R squared Measure of Goodness of Fit for Some Common Nonlinear Regression Models Journal of Econometrics 77 1997 pp 329 342 Estrella Arturo A New Measure of Fit for Equations with Dichotomous Dependent Variables Journal of Business and Economic Statistics April 1998 pp 198 205 Maddala G S Limited dependent and Qualitative Variables Econometrics Cambridge University Press New York 1983 pp 22 27 221 223 231 234 257 259 365 350 PROC PROC Examples PROC defines a TSP user procedure It is always the first statement in a procedure and must be matched by a corresponding ENDPROC statement PROC procedure name lt list of arguments Usage PROC gives the procedure name any legal TSP name and the list of dummy arguments for the procedure When the procedure is called by specifying its name somewhere else in the TSP run any dummy arguments which are used in the procedure are replaced by the actual arguments on the statement which invokes the procedure Always include an ENDPROC statement at the end of your procedure A user procedure can use any of the TSP variables which are at a higher level that is any variables in the procedure s which call it or in the main TSP program Variables which are created in a lower level procedure are not available after leaving that procedure unless they are
386. re three or more arguments FORM does not look for a previous estimation instead it creates a new equation In this case the variable names are the remaining arguments and the coefficient names are constructed mechanically zero starting values are used Note having exactly two arguments is invalid unless the previous estimation was a VAR with two equations because it implies a dependent variable but no right hand side 155 Commands The output equation can be used for estimation or for simulation or in any application which can use equations such as ANALYZ DIFFER EQSUB etc The PARAM option is used for estimation in this case parameter names are inserted into the FRML and values are also stored under these names as if a PARAM statement had been issued These names are normally created by appending 0 1 2 etc to a coefficient prefix such If there is a constant term C the number start with 0 otherwise they start with 1 PARAM names can also be created from the original variable names using the VARPREF option In either case the PARAM names are stored in the list RNMSF which can be used in commands like UNMAKE to fill them with new starting values For simulation use the FORM command after you estimate a linear equation constant values are inserted into the FRML FRMLs with parameter names can also be used for simulation using constant values is just slightly more compact and it also prevents the values from b
387. re to be found for five variables TIME GOVEXP and EXPORTS If 95 of the variance of the five variables can be explained by fewer than three components the program will stop there The principal components found will be stored under the names PC1 2 and PC3 for further use in the program References Harman Harry H Modern Factor Analysis University of Chicago Press First Edition 1960 Sec 9 3 or Third Edition 1976 Sec 8 3 Judge et al The Theory and Practice of Econometrics John Wiley amp Sons New York 1980 Section 12 5 Mundlak Yair On the Concept of Non Significant Functions and its Implications for Regression Analysis Journal of Econometrics 16 1981 pp 139 149 Theil Henri Principles of Econometrics John Wiley amp Sons Inc 1971 pp 46 56 343 Commands PRINT PRINT is a synonym for WRITE PRINT lt list of variables gt 344 PROBIT PROBIT Output Options Example References PROBIT obtains estimates of the linear probit model where the dependent variable takes on only two values Options allow you to obtain and save the inverse Mills ratio as a series so that the sample selection correction due to Heckman can be estimated also see the SAMPSEL command PROBIT FEI FEPRINT MILLS lt name for output inverse Mills ratio NHERMITEz number of points for hermite quadrature gt REI nonlinear options dependent variable list of independent variables Usag
388. reen and dashes on printed output MIN minimum value for the y axis This value must be less than or equal to the minimum value in the data MAX maximum value for the y axis This value must be greater than or equal to the maximum value in the data ORIGIN NOORIGIN causes a horizontal line to be drawn starting at zero on the vertical axis PREVIEW NOPREVIE specifies whether the graph is to be shown on the screen before printing or saving The default is PREVIEW for interactive use and NOPREVIE for batch use This option is not used when working inside the Givewin shell SYMBOL NOSYMBOL specifies that symbols are to be used for plotting TITLE a string which will be printed across the top of the graph DOS Win only A4 NOA4 specifies 4 paper size Available for DEVICE LJ3 or POSTSCRIPT only DEVICE CHAR or EPSON or LJ2 or LJ3 or LJET or LJPLUS or LJR75 or LJR100 or LJR150 or LJR300 or POSTSCRI or PS specifies the hardcopy device to be used for printer output LJ means HP LaserJet or compatible EPSON is EPSON dot matrix or compatible POSTSCRI and PS are Postcript output and CHAR is the old line printer output characters instead of graphics The LJ suffixes specify models of the printer and the LJR suffixes specify the resolution of the LaserJet printer directly rather than giving the printer type The default resolutions for the LJET LJPLUS and LJ2 printers are 100 150 and 300 respectively Note that larger resolutions
389. rent sample ORDPROB will print a warning message and will drop those observations ORDPROB also checks that the observations on the dependent variable are integers and are not negative The list of independent variables on the ORDPROB command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when entering a large number of lagged variables in a regression by imposing smoothness on the coefficients See the PDL section for a description of how to specify such variables Output The output of ORDPROB begins with an equation title and frequency counts for the lowest 10 values of the dependent variable Starting values and diagnostic output from the iterations will be printed Final convergence status is printed 302 ORDPROB This is followed by the number of observations mean and standard deviation of the dependent variable sum of squared residuals scaled R squared likelihood ratio test for zero slopes log likelihood and a table of right hand side variable names estimated coefficients standard errors and associated t statistics ORDPROB also stores some of these results in data storage for later use The table below lists the results available after a ORDPROB command variable type length description LHV list 1 Name of dependent variable RNMS list params List of names of right hand side
390. results globally It is an alias for REGOPT NOPRINT SUPRES lt list of output results gt REGOPT NOPRINT lt list of output results gt NOSUPRES undoes any previous SUPRES that has been issued Usage The arguments to SUPRES can be any of the output names beginning with described in this manual The printing of the output associated with these names will be suppressed throughout the TSP program unless a NOSUPRES or REGOPT command with these codes is issued The output results are still stored in memory and may be accessed Examples To suppress all regression output use the following command SUPRES REGOUT SMPL NOB COEF FST SBIC LOGL This can be done more simply by using the SILENT option on an estimation OLSQ INST etc command If you wish to see only a particular result such as the Durbin Watson follow the SUPRES command with a NOSUPRES command NOSUPRES DW To suppress the printing of the sample every time it changes use SUPRES SMPL To cancel all previous SUPRES commands use SUPRES with no arguments SUPRES 421 Commands SUR Options References SUR obtains seemingly unrelated regression estimates of a set of nonlinear equations It is a special case of LSQ with the options set for SUR estimation 150 command has a more complete description of the procedure SUR COVU OWN or covariance matrix of residuals gt MAXITW 0 HETERO NOITERU NOROBUST nonlinear options list of equa
391. riable are integers and are not negative The list of independent variables on the POISSON command may include variables with explicit lags and leads as well as PDL Polynomial Distributed Lag variables These distributed lag variables are a way to reduce the number of free coefficients when entering a large number of lagged variables in a regression by imposing smoothness on the coefficients Output The output of POISSON begins with an equation title and frequency counts for the lowest 10 values of the dependent variable Starting values and diagnostic output from the iterations will be printed Final convergence status is printed 337 Commands This is followed by the number of observations mean and standard deviation of the dependent variable sum of squared residuals correlation type R squared a test for overdispersion likelihood ratio test for zero slopes log likelinood and a table of right hand side variable names estimated coefficients standard errors and associated t statistics The default standard errors are the robust QMLE Eicker White estimates These are consistent even for a model whose variance is not equal to the mean as long as the mean is correctly specified For most economic data the overdispersion test rejects the Poisson model and you may wish to use the Negative Binomial model instead although as a member of the linear exponential class the Poisson model with Eicker White standard errors may be more robus
392. ries by the moving average method SAMA ARITH PRINT lt input series gt lt output series gt Usage SAMA is followed by the name of the series to be seasonally adjusted and then the name to be given to the new series The two series may be the same in which case the new one will just replace the old one SAMA should not be used on series which can be negative Output When the print option is off SAMA produces no printed output The seasonally adjusted series is stored along with the intermediate results under the following names variable type length description SFAC vector periods Seasonal factors which divide the old series to make new series MOVA series obs ratio of the old series to its moving average If the print option is on these quantities are also printed in table form Method Denote the series to be adjusted by X indexed by t and T observations in length The periodicity of the series the number of periods per year is p which is customarily 4 in the case of quarterly series or 12 in the case of monthly series The ratio of the series to its moving average is formed in the following way MOVA t X ty Moving Average of X where the moving average of X is defined as 4 1 4 X t 2 2 X t 1 X t X t 1 X t 2 2 pz12 1 12 X t 6 2 X t 5 X t X t 5 X t 6 2 391 Commands This equation describes a weighted moving average over p 1 observations centered
393. ring of the model the number of simultaneous and recursive blocks and whether or not the simultaneous blocks are linear in the variables for which they are to be solved This information is also stored as part of the collected model A table is printed which shows for each equation in the model the number of its block whether the block is simultaneous S or recursive R and which endogenous variables appear in that equation See the example output in the TSP User s Guide Options DONGALLO NODONGAL specifies that each simultaneous block should be ordered for a near minimal feedback set It prints an F next to the feedback variables to identify them and a summary of the blocks This ordering is sometimes useful when the Gauss Seidel method is used to SOLVE the model FILE filename writes a file containing input for the CAUSOR program CAUSOR provides detailed information on model structure such as essential feedback sets When FILE is used the equations do not have to be uniquely normalized as long as the endogenous variable list is supplied CAUSOR may be obtained from Manfred Gilli Departement d Econometrie Universite de Geneve 268 MODEL PRINT NOPRINT specifies whether the older more voluminous output format is to be used SILENT NOSILENT suppresses printing completely Examples This example shows how to set up the well known Klein Model for simulation INST CX C W P 1 INVR C P 1 K 1 E 1 TM W2 TX FO
394. rivatives internally for the nonlinear functions ANALYZ can also be used to select reorder a subset of a VCOV matrix and COEF vector for use in making a Hausman specification test ANAL YZ COEF lt input parameter vector HALTON NAMES lt list of names NDRAW lt number of draws PRINT SILENT VCOV lt matrix name gt lt list of equation names gt Usage ANALYZ is followed by a list of equation FRML names After estimation procedures with linear models OLSQ INST LIML PROBIT these equations specify functions of the estimated coefficients which are to be computed by referring to the coefficients by the names of the associated variables After estimation procedures with nonlinear models LSQ and FIML the equations specify functions of the estimated parameters ANALYZ has no provision for combining the variances from more than one estimation because it cannot obtain the associated covariance of the coefficient estimates The equations must be previously defined by FRML statements if the FRML statements have variable names on the left hand side the computed value of each function will be stored under that variable name 35 Commands If series names other than the names of right hand side variables from the previous OLSQ INST LIML or PROBIT estimation are included in the FRML s a series of values will result One application for this kind of FRML is an elasticity which depends on estimated parameters and
395. ro standard errors They match HCOVEN standard errors to 3 5 digits in the tests we ve performed and HCOV C use a discrete Hessian for iteration and computing the variance estimate respectively This is a numeric difference of analytic first derivatives This is even more accurate than HCOV U in terms of matching HCOVEN results it s usually good to 6 digits in standard errors It also requires 2 K derivative evaluations for a model with K parameters Unfortunately it is very specialized it is really only useful in a command that has analytic first derivatives but not analytic second derivatives The most important such command in TSP is FIML FIML also stores its results in data storage The estimated values of the parameters are stored under the parameter names In addition the following results are stored variable type length description RNMS list params Parameter names LOGL scalar 1 Log of likelihood function SBIC scalar 1 Schwarz Bayes information criterion with nobs NOB NEQ AIC scalar 1 Akaike information criterion NCOEF scalar 1 Number of parameters in model NCID scalar 1 Number of identified parameters in model lt NCOEF IFCONV scalar 1 Convergence status 1 success 146 FIML GRAD vector Gradient of likelihood at maximum COEF vector Estimated values of parameters also stored under their names SES vector Standard errors of est
396. ro divides When these conditions occur the appropriate limit is taken instead This may result in some slight inaccuracy in the likelihood function but it is certainly preferable to halting the estimation Observations subject to these problems can be identified by exact 0 and 1 values in FIT Before estimation LOGIT checks for univariate complete and quasi complete separation of the data and flags this condition The model is not identified in this case because one or more of the independent variables perfectly predict Y for some observations and therefore their coefficients would slowly iterate to or infinity if estimation was allowed to proceed The scaled R squared is a measure of goodness of fit relative to a model with just a constant term it replaced the Kullback Leibler R squared beginning with TSP 4 5 since it has somewhat better properties for discrete dependent variable problems See the Estrella 1998 article Options The standard options for nonlinear estimation are available see the NONLINEAR section in this manual Note that HITERZN HCOVEN are the defaults for the Hessian approximation and standard error computations In addition the following options are specifically for the LOGIT procedure CASE case series for multiple observations per case This variable holds a case identification which is equal for adjacent observations that belong to the same case Note that any such variable may be used it does not neces
397. rpreted form of the command is shown exactly as it is stored in data storage for later execution Example ASMBUG FRML EQ Y A EXP GAMMA X B LOG HSQ Z NOABUG This example will cause various intermediate results during the parsing of the equation to be printed 54 BJEST BJEST Output Options Examples References BJEST estimates the parameters of an ARIMA AutoRegressive Integrated Moving Average univariate time series model by the method of conditional or exact maximum likelihood The technical details of the method used are described in Box and Jenkins Chapter 7 TSP uses the notation of Box and Jenkins to describe the time series model See BJIDENT for identification of a time series model and BJFRCST for forecasting using the estimated model BJEST CONSTANT CUMPLOT EXACTML NARz number of AR parameters NBACKz number of back forecasted residuals NDIFF lt degree of differencing gt NLAG lt number of autocorrelations gt lt of parameters NSAR lt number of seasonal params gt NSDIFF lt degree of seasonal differencing gt NSMA lt number of seasonal MA parameters gt NSPAN lt span of seasonal gt PLOT PREVIEW ROOTS START nonlinear options lt series name gt START parameter name gt lt parameter value FIX parameter name parameter value ZERO parameter names ZFIX parameter names Usage To estimate
398. rs of parameter names and starting values The names available are AR lag or PHI lag an ordinary autoregressive parameter MA lag or THETA lag ordinary moving average parameter SAR lag or a seasonal autoregressive parameter GAMMA lag SMA lag or a seasonal moving average parameter DELTA lag 56 BJEST The lag in parentheses should be an actual number giving the position of the coefficient in the lag polynomial i e AR 1 means the coefficient which multiplies the series lagged once 0 is always unity Sometimes it is useful to fix a parameter at a certain value while others are being estimated This may be done to limit the number of parameters being estimated or to incorporate prior information about a parameter value or to isolate an estimation problem which is specific to one or a few parameters The keyword FIX followed by pairs of parameter names and values can be used to achieve this FIX and START can be used on the same BJEST command to give starting values to some parameters and hold others fixed The ZERO keyword is used to override the automatic starting values for some parameter s with zero s The ZFIX keyword fixes some parameter s to zero BJEST can also be used to check roots of polynomials for stationarity invertability without doing any estimation Just supply the coefficient values in START use NAR p and or NMA q and do not supply a dependent variable The PRINT option will print the root
399. rs to be estimated The default is zero 60 BJEST NSPAN the span number of periods of the seasonal cycle i e for quarterly data NSPAN should be 4 The default is the current frequency that is 1 for annual 4 for quarterly 12 for monthly PLOT NOPLOT specifies whether the residuals are to be plotted PREVIEW NOPREVIEW TSP Givewin only specifies that the residual plots are to be displayed in a high resolution graphics window PRINT NOPRINT specifies the level of output desired NOPRINT should be adequate for most purposes ROOTS NOROOTS specifies whether the roots of the polynomial should be printed START NOSTART specifies whether the procedure should supply its own starting values for the parameters Nonlinear options control iteration and printing They are explained in the NONLINEAR entry One special use of these options in BJEST is the combination EXACTML NOCONST MAXIT 0 MAXSQZ 123 which forces the program to simply evaluate the ARMA likelihood at the starting values of the parameters in START Examples This example estimates a simple ARMA 1 1 model with no seasonal component no plots are produced of the results BJEST NAR 1 NMA 1 NOPLOT NOCUMPLO AR9MAS This example uses the Nelson 1973 auto sales data a logarithmic transformation of the series is made before estimation GENR LOGAUTO LOG AUTOSALE BJEST NDIF 1 NSDIFF 1 NMA 2 NSMA 1 NSPAN 12 NBACK 15 LOGAUTO START THETA 1 0 12
400. rsonal computers Areas where TSP can be useful include Applied econometrics including teaching Macro economic research and forecasting Econometric analysis of cross section and panel data Sales forecasting Financial analysis Cost analysis and forecasting TSP is installed on thousands of computers worldwide Although TSP was developed by economists starting with Version 1 in 1967 and most of its uses are in economics there is nothing in its design that limits its usefulness to economic time series Any statistical or econometric application involving data sets of up to about 20 000 or even more observations is suitable for TSP For more information prices ordering and upgrades see our website http www tspintl com Examples of TSP Programs Examples of TSP Programs TSP is a powerful program because it is not limited to its preprogrammed commands To help you develop your own specialized procedures we have compiled some examples of TSP programs The example programs can be found in the Program Files TSP 5 0 examples directory the exact path may vary depending on where you installed TSP If you did a custom install of TSP you may not have installed the examples In order to get the examples you will have to reinstall TSP and do either a Typical install or a Custom with the Examples option check marked The TSP examples are further subdivided into five categories Miscellaneous ML PROC Panel Data Qualitati
401. ry values are not allowed Just recode such values to integers preserving the proper ordering 304 ORDPROB ORDPROB uses analytic first and second derivatives to obtain maximum likelihood estimates via the Newton Raphson algorithm This algorithm usually converges fairly quickly TSP uses zeros for starting parameter values except for the constant term and the MUs START can be used to provide different starting values see NONLINEAR Multicollinearity of the independent variables is handled with generalized inverses as in the other linear and nonlinear regression procedures in TSP If you wish to estimate a nonstandard ordered probit model e g adjusted for heteroskedasticity or with a nonlinear regression function use the ML command See our website for an example Before estimation ORDPROB checks for univariate complete and quasi complete separation of the data and flags this condition because the model is not identified in this case Without this check one or more RHS variables perfectly predict the dependent variable for some observations and their coefficients would slowly iterate to plus or minus infinity The Scaled R squared is a measure of goodness of fit relative to a model with just a constant term it is a nonlinear transformation of the Likelihood Ratio test for zero slopes See Estrella 1998 Although the paper is concerned with dichotomous dependent variables the scaled R squared applies to any model with a
402. s References Haerdle W Applied Nonparametric Regression Cambridge Cambridge University Press 1990 Silverman B W Density Estimation for Statistics and Data Analysis London Chapman and Hall 1986 213 Commands LAD Output Options References LAD computes least absolute deviations regression also Known as L1 regression or Laplace regression This type of regression is optimal when the disturbances have the Laplace distribution and is better than least squares L2 regression for many other leptokurtic fat tailed distributions such as Cauchy or Student s t LAD LOWER lt lower censoring limit NBOOT lt replications gt QUANTILE lt value gt RESAMPLE lt method for computing s e s gt SILENT TERSE UPPER lt upper censoring limit lt dependent variable gt lt list of independent variables gt Usage To estimate by least absolute deviations in TSP use the LAD command just like the OLSQ command For example LAD CONS C GNP estimates the consumption function using L1 median regression instead of L2 least squares regression Various options allow you to perform regression for any quantile and censored L1 regression Standard errors may be obtained using the bootstrap see the options below Output The usual regression output is printed and stored see OLSQ for a table The likelihood function LOGL and standard error estimates are computed as though the true distribution of the disturb
403. s number with non zero standard errors LMHETT W V scalar 1 LM heteroskedasticity 314 LMHETT W V DWT W V DWUT W V DWLT W V LOGLT I W REI REIT SBICT W REI REIT AICT W REI REIT HAUS HAUS HAUSDF RNMST B W V REI REIT COEFT I B W V REI REIT SEST B W V REI REIT TT B W V REI REIT COEFAI SESAI TAI TAI VCOVT B W V REI REIT REST I B W V REI REIT scalar scalar scalar scalar scalar scalar scalar scalar scalar scalar list vector vector vector vector vector vector vector series matrix series 1 vars vars vars vars individuals individuals individuals individuals vars vars obs PANEL test P value of LM heteroskedasticity test Durbin Watson autocorrelation test Upper bound on P value of DW Lower bound on P value of DW value of the log likelihood Schwarz Bayes information criterion Akaike information criterion Hausman test value Hausman test p value Hausman test degrees of freedom List of names of right hand side variables Coefficient estimates Standard errors T statistics Estimated fixed effects Standard errors on fixed effects t statistics on fixed effects p values associated with TAI Fixed effect estimates as a series Variance covariance of estimated coefficients Residuals actual fitted values of the dependent variable 315 Commands Method Th
404. s or the ARSTAT and MAINV variables can be used Output The output of BJEST begins with a printout of the starting values followed by an iteration log with one line per iteration giving the value of the objective function and the convergence criterion If the PRINT option is on the convert values of the options and the exact time series process being used are printed In the iteration log parameter values and changes are printed for each iteration when PRINT is on When convergence of the iterative process has been achieved or the maximum number of iterations reached a message to that effect is printed and the final results are displayed These include the conventional statistics on the model the standard error of at the R squared and F statistic for the hypothesis that all the parameters are zero The parameter estimates and their standard errors are shown in the usual regression output If the ROOTS option is on and the order of any polynomial is greater than one its roots and moduli are shown so that you can check that they are outside the unit circle as is required for stationarity A table of the autocorrelations and Ljung Box modified Q statistics of the residuals is printed after this If the PRINT option is on the exact model estimated is again printed in lag notation along with some summary statistics for the residuals Then comes a printout of the lagged cross correlations between the differenced series wt and the white n
405. s Diagonal of hat matrix if the HI option is on If the regression includes a PDL or SDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector lags Estimated lag coefficients after unscrambling REGOPT NOPRINT LAGF will turn off the lag plot for PDL variables Method OLSQ computes the matrix equation 293 Commands b X X X y where X is the matrix of independent variables and y is the vector of independent variables The method used to compute this regression and all the other regression type estimates in TSP is a very accurate one which involves applying an orthonormalizing transformation to the X matrix before computation of the inner products and inverse and then untransforming the result see ORTHON in this manual See OPTIONS FAST to compute faster and slightly less accurate regressions without orthonormalization OLSQ has been tested using the data of Longley on everything from an IBM 370 to a modern Pentium computer it gives accurate results to six digits when the data is single precision For the artificial problem suggested by Lauchli see the Wampler article OLSQ gives correct results for the coefficients to about five places until epsilon becomes so small that the regression is not computable Before this happens OLSQ detects the fact that it cannot compute the regression accurately and drops one
406. s for all the variables listed Only observations in the current sample with no missing values are included The variables may be weighted before the statistics are computed MSD ALL BYVAR CORR COVA MOMENT PAIRWISE PRINT SILENT TERSE WEIGHT lt series name gt lt list of series gt Usage For univariate statistics on a set of variables use the MSD command with no options The CORR MOMENT and COVA options enable you to get several forms of descriptive statistics on the variables at once saving on computation time The option ALL allows you to obtain additional statistics such as the median The TERSE option restricts the statistics computed in order save space and time Output MSD stores the statistics which are requested as well as printing them variable type length description NOBMSD vector vars Number of non missing observations for BYVAR MEAN vector vars Means STDDEV vector vars Standard Deviations MIN vectpr vars Minimums MAX vector vars Maximums SUM vector vars Sums VAR vector vars Variances SKEW vector vars Skewness KURT vector vars Excess kurtosis MEDIAN vector vars Median for ALL option Q1 vector vars 1st quartile Q3 vector vars 3rd quartile IQR vecotr vars Inter quartile range CORR matrix vars vars Correlation matrix COVA matrix vars vars Covariance matrix 270 MSD MOM matrix vars vars Moment matrix divided by number of observations N
407. s if they happen in the first function evaluation using the starting values They are not fatal during iterations the program automatically uses a smaller stepsize but they can be inefficient Often these problems can be avoided by reparametrizing the likelihood function The standard example of this is estimating SIGMA or SIGMA inverse instead of SIGMA squared If you are getting numerical errors and you can t rewrite the likelihood function try using SELECT to remove the problem observations After you get convergence use the converged values as starting values and reestimate using the full sample Choose starting values carefully See previous Use EQSUB for less work rewriting equations and more efficient code The Working space message gives an indication of the length of the derivative code If the second derivative matrix is singular you may have sign errors in the log likelihood function the inversion routine assumes the second derivative matrix is negative definite If you are using derivatives make sure the functions you are using are differentiable Logical operations are not differentiable everywhere although they are differentiable at all but a finite number of points TSP will do the best it can with them but if you end up on a kink corner it may stall The following results are stored variable type length description LOGL scalar 1 Log of likelihood function IFCONV scalar 1 1 if convergence ac
408. s to the PROC It prevents the local variables from being used instead of the PROC arguments The disadvantage is that the labels in the output will be the formal argument names rather than that actual argument names BASEYEAR value used to make dates from 2 digit numbers The default is 1900 For example by default 86 2 means 1986 2 If you set BASEYEAR 2000 86 2 would mean 2086 2 BASEYEAR can also be set to Zero so that you can use dates in the first two centuries CHARID NOCHARID treats the ID series as characters instead of numbers when printing observation labels CHARID requires use of the DOUBLE option also To use this option read in your ID series called ID using an A8 format statement CRT NOCRT sets several output format options to values suitable for viewing output on a 24 line by 80 character screen These are LIMPRN 80 LINLIM 24 and LEFTMG 0 The page headings date time and page number are also suppressed in CRT mode DATE NODATE specifies whether the date and time headings at the top of each page of TSP output are to be printed A user title and page number if present is still printed This option only applies when page headings are being printed NOCRT DEBUG NODBUG sets the debug option causing intermediate results to be printed This is described more fully in the DEBUG section and is of use primarily to TSP programmers DISPLAYz monitor type for PC 386 graphics essentially obsolete DOUBLE NODOUBL
409. sarily have to be the case identification COND NOCOND for conditional or mixed models versus pure multinomial estimation see the Usage section If CASE or NREC is used COND is assumed and need not be specified NCHOICE number of choices can be supplied when the number is equal for all observations The program then checks to make sure the data satisfy this constraint This is not used with CASE or since then it is valid to take the first choice every time 239 Commands NREC choice count series for multiple observations per case This variable specifies the number of observations in each case Usually the number is repeated in each observation but only the count in the first observation for each case is used You cannot say NREC 3 but you can say NCHOICE 3 SUFFIX a list of short names suffixes for the alternatives These names are used in 4 places in the initial table of frequencies for the dependent variable as coefficient names for multinomial variables as labels for the probability derivatives dP dZ and as the suffixes for conditional variable names when there is one observation per case Note that SUFFIX does not imply COND See the examples below The SUFFIX names need to be in the proper order relative to the values of the dependent variable In the examples below Y 1 is CAR Y 2 is BUS and Y 3 is RT If some of the alternatives are never chosen be sure to use SUFFIX or NCHOICE to ensure the fu
410. sed as part of a name in the MATRIX command digit 0 to 9 Parts of numbers or names Marks the decimal point in numbers sets off decimal point logical operators specifies string substitution in the DOT procedure Separates the words in a list and arguments in multivariate functions like CNORM2 spaces i may be used but commas are often preferred for clarity colon Part of date semicolon Marks the end of a statement dollar sign Equivalent to Semicolon is preferred quotation Marks the beginning and end of a text string mark title or filename specifies matrix inversion Marks the beginning end of a text string apostrophe B Ss specifies matrix transposition Encloses a list of options or expressions lags parentheses 000 in algebraic antics P Delimits the beginning of comments question mark Comments are terminated by the end of the input line or logical record plus sign Specifies addition minus sign Specifies subtraction a lag or a list star ii Specifies multiplication or is part of power slash Specifies division pound sign Matrix Kronecker product 9 o Matrix Hadamard product element by percent Jo element equal sign E Specifies equality or definition of data relational operator EQ NE LE GE ampersand amp Logical operator AND Logical operator OR Also used to separate vertical bar lists of variables in INST KALMAN LOGIT caret or hat SAMPSEL and VAR Logical
411. series when NCHOICE 2 or when there are multiple observations per case If NCHOICE 2 FIT will contain the probabilities for the highest choice only just like binary PROBIT Method If C t is the choice set for the tth observations and observation t chooses the ith alternative out of C t then the expression for the choice probability is exp Z b Pr i chosen from C 2 7 gt exp Z b j y The likelihood function is LogL E log 2 Zy b j j t l The coefficient vector to be estimated is b If some of the Zs do not vary across the choices these equations would apply to an expanded Z vector formed by taking the Kronecker product of the fixed Zs with an identity matrix of order of the number of choices less one for a normalization The actual implementation does not expand the Zs but treats the conditional and multinomial variables differently to conserve space 238 LOGIT Newton s method is used to maximize this likelihood function with respect to the parameter vector b The global concavity of the likelihood function makes estimates fairly straightforward to obtain with this method Zero starting values are the default unless START is supplied See the NONLINEAR section in this manual for more information about TSP s nonlinear optimization procedures in general The evaluation of the EXP functions in the likelihood function and derivatives avoids floating overflows and ze
412. ses nearly singular matrices to fail rather than letting through some exactly singular ones Examples This example uses the dates 53 BC to 86 AD OPTIONS 0 FREQA SMPL 53 86 Here are some other OPTIONS commands OPTIONS REPL PLOTS TOLz1 E 10 OPTIONS NWIDTH 10 SIGNIF 3 301 Commands ORDPROB Output Options Example References ORDPROB obtains estimates of the linear Ordered Probit model where the dependent variable takes on only nonnegative integer ordered category values The scaling of the category values does not matter although they should be positive and integer for convenience only information about their order is used in estimation ORDPROB nonlinear options lt dependent variable gt lt list of independent variables gt Usage The basic ORDPROB statement is like the OLSQ statement first list the dependent variable and then the independent variables If you wish to have an intercept term in the regression usually recommended include the special variable C or CONSTANT in your list of independent variables You may have as many independent variables as you like subject to the overall limits on the number of arguments per statement and the amount of working space as well as the number of data observations you have available The observations over which the regression is computed are determined by the current sample If any of the observations have missing values within the cur
413. sing a default 20 points when RHO is high it may be necessary to increase this using the NHERMITE option Options FEI NOFEI specifies that the fixed effects Probit model should be computed FREQ PANEL must be in effect FEPRINT NOFEPRIN specifies whether the estimated effects and their standard errors should be printed MILLS the name of a series used to store the inverse Mills ratio series evaluated at the estimated parameters The default is MILLS number of points for the Hermite quadrature in computing the integral for the random effects Probit model The default is 20 The value set is retained throughout the TSP run REI NOREI specifies that the random effects Probit model should be computed FREQ PANEL must be in effect The usual nonlinear estimation options can be used See the NONLINEAR entry Examples Standard probit model PROBIT MOVE C WAGE1 WAGE2 COST1 COST2 Heckman sample selection model see the SAMPSEL command for ML estimation of this model PROBIT MILLS RMILL WORK C OCC1 OCC2 TENURE MSTAT AGE SELECT WORK 01 50 LWAGE SCHOOL EXPER IQ UNION OCC1 OCC2 RMILL Computing fitted probabilities and inverse Mills ratios explicitly PROBIT MOVE C WAGE1 WAGE2 COST1 COST2 FORCST XB MOVEP CNORM XB MILLSR MOVE DLCNORM XB 1 MOVE DLCNORM XB 349 Commands References Amemiya Takeshi Qualitative Response Models A Survey Journal of Economic Literatu
414. sis Testing Commands ANALYZ CDF COINT REGOPT computes standard errors for functions of parameters from a previous estimation distribution functions and P values unit root and cointegration tests controls printing and storage of regression diagnostics 23 Command summary Forecasting and Model Simulation Commands ACTFIT BJFRCST FORCST FORM MODEL n IML LVE e o 24 compares actual and forecasted series forecasts Box Jenkins ARIMA models computes forecasts for estimated linear models OLSQ INST AR1 constructs an equation FRML from an estimated linear model orders large simultaneous equation systems for use by SOLVE simulation of general nonlinear systems of equations simulation of large usually sparse systems of equations Time Series Identification and Estimation Commands Time Series Identification and A Estimation Commands AR1 regression with correction for AR 1 autocorrelated error ARCH estimates GARCH M models BJEST estimates Box Jenkins ARIMA models BJFRCST forecasts Box Jenkins ARIMA models BJIDENT identifies the order of Box Jenkins ARIMA models COINT unit root and cointegration tests KALMAN Kalman filter estimation UNIT synonym for COINT VAR Vector autoregressions 25 Command summary Control Flow Commands COLLECT COMPRESS DO DOT ELSE END ENDDO ENDDOT ENDPROC EXEC EXIT GOTO IF INPUT LIST LOCAL OUTPUT PROC QUIT STOP SYSTEM TERMINA
415. smaller than 001 TOLS tolerance of determining convergence of the parameters using the squeezed step The default is 0 that is ignore the squeezed change in parameters and use the regular TOL instead 283 Commands VERBOSE NOVERBOSE produces lots of diagnostic output including the gradient Hessian and inverse Hessian at each iteration and the non inverted Hessian for each output VCOV Starting values The default values depend on the procedure For the standard TSP models which are potentially nonlinear in the parameters LSQ FIML ML the user provides them with PARAM and SET PROBIT and LOGIT use zeros TOBIT uses a regression and formulas from Greene 1981 SAMPSEL uses probit and a regression The default is overridden in the linear model procedures ARCH BJEST PROBIT TOBIT LOGIT and SAMPSEL if the user supplies a matrix named START The length of START must be equal to the number of parameters in the estimation otherwise it is ignored The easiest way to create START is with a statement like MMAKE START 12 3 4 56 33 44 55 The order of the parameters in START is obvious for most of the linear models except for the following TOBIT SIGMA comes last SAMPSEL the probit equation is first then the regression equation and then SIGMA and RHO References Berndt E K B H Hall R E Hall and J A Hausman Estimation and Inference in Nonlinear Structural Models Annals of Economic and S
416. so itis performed by default See Pantula et al 1994 or the Method section for details UNIT NOUNIT use NOUNIT to skip all unit root tests if you are only interested in cointegration tests and you are sure which individual variables have unit roots Cointegration Test Options these apply only if you have more than one variable ALLORD NOALLORD repeat the Engle Granger tests using each variable in turn on the left hand side of the cointegrating regression 89 Commands EG NOEG perform augmented Engle Granger tests Dickey Fuller test on residuals from the cointegrating regression The Engle Granger test is only valid if all the cointegrating variables are 1 1 hence the default option to perform unit root tests on the individual series to confirm this before running the Engle Granger test Note that if you accept 1 i e reject 1 0 you will also want to difference the series and repeat the unit root test to make sure you reject 2 in favor of 1 Note that you need to reduce the order of trends when testing such a differenced series for example if the original series had a constant and trend in the equation the differenced one will only have a constant JOH NOJOH perform Johansen trace cointegration tests COINT NOCOINT use NOCOINT to skip all cointegration tests if you are only interested in unit root tests You may prefer to use the UNIT or UNIT NOCOINT command for this UNIT and COINT are synonyms for t
417. ssion This occurs even when the arguments to these expressions are complicated expressions themselves Thus you should be careful to express any such complex arguments the same way whenever they appear in the matrix expression The following functions take matrices as their input and produce scalars as output They may be used anywhere in a MAT statement where scalars are allowed keeping in mind that a scalar is also a 1 by 1 matrix s DET qm determinant truncated to zero when lt 1 E 37 s log of positive determinant no truncation LOGDET qm s TR qm trace sum of diagonal elements s MIN m element with minimum value s MAX m element with maximum value s SUM m sum of elements i NROW m number of rows i NCOL m number of columns i RANK m rank number of linearly independent columns or rows The following functions are matrix to matrix that is they take a matrix perform some computation on it and produce another matrix as output They can be used anywhere in a MAT equation tm Choleski factorization matrix square root CHOL sm sm YINV sm Positive semi definite inverse via CHOL dm IDENT i Creates an identity matrix of order v Computes the vector of eigenvalues of qm If qm EIGVAL qm is not symmetric positive semi definite the imaginary parts of the eigenvalues are stored as 252 MATRIX EIGVALI Real eigenvalues are sorted in decreasing orde
418. ssion independent variables Missing values for the regression variables are allowed for those observations for which y 1 0 If y 1 is always greater than zero the truncated conditional model is estimated Bloom and Killingsworth 1985 This is flagged with the message Latent Selection Variable The identifying condition that there be variables other than the constant in the probit equation is not checked 393 Commands Output The output of SAMPSEL begins with an equation title and the name of the dependent variable Starting values and diagnostic output from the iterations will be printed Final convergence status is printed This is followed by the mean of the dependent variable number of positive observations sum of squared residuals R squared and a table of right hand side variable names estimated coefficients standard errors and associated t statistics SAMPSEL also stores some of these results in data storage for later use The table below lists the results available after a SAMPSEL command name type length description LHV list 1 Name of dependent variable RNMS list vars list of names of right hand side variables YMEAN scalar 1 Mean of the probit dependent variable NOB scalar 1 Number of observations NPOS scalar 1 Number of positive observations in probit equation SSR scalar 1 Sum of squared residuals regression equation LOGL scalar 1 Log of likelihood function SBIC scalar 1 Schwarz Bayesian Inf
419. st for ARCH 1 residuals Recursive residuals CUSUM plot CUSUMSQ plot CUSUM test statistic CUSUMSQ test statistic F test for stability of coefficients split sample Test for stability of coefficients with heteroskedasticity LR test for heteroscedasticity in split sample White het test on cross products of RHS variables Breusch Pagan het test on user supplied list of vars simple LM het test on squared fitted values F statistic for zero slope coefficients Ramsey s RESET test of order x Jarque Bera LM normality test Shapiro Wilk normality test Akaike Information Criterion Schwarz Bayesian Information Criterion 379 Commands LOGL Log of likelihood function Method Notes on specific diagnostics DW ignores sample gaps except when there is PANEL data The DWPVALUE option can be used to choose one of the 3 methods of calculating its P value EXACT computes the nonzero eigenvalues of the matrix and then uses the Farebrother Pan method to compute the P value from the DW and these eigenvalues The APPROX method is a small sample adjustment to the asymptotic distribution using a nonlinear regression fit to the 5 dL lower bound table p value DW 2 040 58325 1 04 0 54522 14 1 50451 K 1 T 20342 where phi is the cumulative normal This usually provides conservative test i e P value larger than the EXACT method like the larger number from BOUNDS The BOUN
420. st one argument all prompting and storage will be defined by the most recent SMPL and FREQ prior to the ENTER command In response to the prompt for data you may enter as many items per line as you like the prompts will adjust accordingly Prompting will cease when the current SMPL has been satisfied and the series stored If more than one series is being entered you will be prompted to enter them sequentially If a numeric error is made in data entry make a note of which observation s and finish entering the data Then the UPDATE command may be used to correct the error If a non numeric entry is encountered you will be prompted to go back and correct it Missing values may be stored by creating a SMPL with gaps prompting will adjust or using the missing value code Example 1 FREQA 2 SMPL 1946 1975 3 ENTER GNP Enter data for GNP 1946 475 7 468 3 487 7 490 7 1950 533 5 576 5 598 5 621 8 613 7 1955 654 8 668 8 680 9 679 5 720 4 736 8 755 3 1962 799 1 830 7 874 4 925 9 981 0 1007 7 1968 1051 8 1078 8 1075 3 1107 5 1171 1 1973 1233 4 1210 7 1186 4 GNP will be stored with 30 observations 137 Commands EQSUB Options Example EQSUB substitutes one or more equations into another This is useful for estimation with several parameter restrictions with long equations which have common terms or for setting up a complicated model where the exogenous variables can be changed later by just changing one of
421. st replace the listname Lists may also be subscripted or lagged PROC arguments are also treated as lists the same type of replacement occurs LIST FIRST lt number gt LAST lt number gt PREFIX names SUFFIX lt name gt or value list name list of variable names or LIST DROP lt list name gt lt list of variable names gt or LIST PRINT DELETE list of listnames Usage The LIST name can be any legal TSP variable name Follow this with any number of legal TSP variable names they may include lags or leads including the names of other lists Lists may be nested indefinitely that is the names following the LIST name may be LISTs The only limitation on length is that on the ultimate list of variables LISTs may not be defined recursively the list name cannot appear in its own list when a list is first defined An example of an illegal recursive definition is LIST LL A B LL if LL has not already been defined as a list A LIST of variables may be specified as a range e VAR1 VAR2O A list like this is interpreted as VAR1 VAR2 VAR3 VAR19 VAR2O You also use a range expression for numbers or for lags X 1 X 20 Such an implicit list can also appear directly in a command see the DOT command below You can combine this type of list with a DOT loop to operate on individual elements of the list very conveniently LIST VARLIST VAH1 VAR20 DOT 1 20 compu
422. such as 1 5 be used directly in a statement without making an intermediate listname See the LIST command for a complete description of implicit list syntax The end of a statement is marked by a semicolon or dollar sign Introduction Composing Algebraic Expressions in TSP In general TSP rules for formulas are similar to Fortran or other scientific programming languages A lag is indicated by putting an integer or a name in parentheses after a series name The integer is negative for lags and positive for leads A sign is not necessary for leads If the lag or lead is a name it must have no more than four characters A series may have a single numeric or variable subscript or lag lead A matrix may have a single or double subscript numeric or variable See the SET command for detailed rules and examples Arithmetic operators are p add multiply or raise to the power See TSP Functions for a detailed list of functions and the MATRIX command for matrix functions Relational and logical operators are the following OP Description EQ gives the value 1 when the variables on the left and on the right are equal otherwise it is zero NE gives the value 1 when the variables on the left and on the right are not equal otherwise it is zero LT gives the value 1 when the variable on the left is less than the variable on the right otherwise it is zero GT gives the value 1 when the
423. sults showing names and names eed ee eee ee le ee ie 1 00600095 MINIMUM MAXIMUM show scalar Class Name Description SCALAR NOB constant 10 00000000 FREQ constant 0 00000000 YMEAN constant 38 50000000 SDEV constant 34 17357654 SSR constant 528 00000000 S2 constant 66 00000000 85 constant 8 12403840 RSO constant 0 94976452 ARSQ constant 0 94348509 LMHET constant 0 39160497 SLMHET constant 0 53145697 constant 0 45454545 SDW constant 0 012096704 QJB constant 1 01478803 SJB constant 0 60206250 RESET2 constant 8 5070592D 37 SRESET2 constant 0 00000000 QFST constant 151 25000000 SEST constant 0 00000177754 SBIC constant 36 32452638 AIC constant 36 02194129 LOGL constant 34 02194129 NCOEF constant 2 00000000 NCID constant 2 00000000 LMAR1 constant 8 5070592D 37 SLMAR1 constant 0 00000000 LMAR2 constant 8 5070592D 37 LMAR2 constant 0 00000000 QSTAT1 constant 3 33333333 SQSTAT1 constant 0 067889155 QSTAT2 constant 3 38842975 SQSTAT2 constant 0 18374343 ARCH constant 0 25822990 SARCH constant 0 61133885 CSMAX constant 1 26364964 SCSMAX constant 0 0031685821 CSQMAX constant 0 46590909 CSOMAX constant 0 050848751 CHOW constant 53 57142857 SCHOW constant 0 00014913251 CHOWHET constant 53 57142857 SCHOWHET constant 0 00014913251 LRHET constant 26 49209701 SLRHET constant 0 00000026462 WHITEHT constant 3 38983051 SWHITEHT constant 0 18361479 BPHET constant 1 7490
424. suppress all the default regression models These individual means are stored in the NI x 1 NX matrix MEAN where the first column is the dependent variable PRINT NOPRINT prints COEFI in conjunction with BYID and prints FIXED for within REG NOREG is used with the MEAN option above To suppress some regression models but print others use the individual options NOBETW to suppress the BETWEEN output etc REI NOREI specifies that ML estimates of the one way random effects model are to be obtained START may be used to supply starting values REIT NOREIT specifies that ML estimates of the two way random effects model are to be obtained This requires the TIME option for unbalanced data in FREQ PANEL START may be used to supply starting values ROBUST NOROBUST calculates heteroskedasticity robust standard errors HCTYPE 1 see OLSQ for the WITHIN coefficients If this option is used the Hausman test comparing WITHIN and VARCOMP is not computed SILENT NOSILENT can be used to turn off all the regression output T the number of time periods for each individual for balanced data only For unbalanced data use the ID option TERSE NOTERSE can be used to turn off most of the regression output except the coefficients and standard errors 318 PANEL TIME the name of a time period series which increases in value for each individual and decreases between individuals Alternatives are the ID and T options Example T
425. t Equation 1 Method of estimation Ordinary Least Squares Dependent variable T2 Current sample 1 to 10 Number of observations Mean of dep var Std dev of dep var 38 5000 LM het test 34 1736 Durbin Watson Sum of squared residuals 528 000 Jarque Bera test Variance of residuals 66 0000 Ramsey s RESET2 Std error of regression 8 12404 F zero slopes 391605 531 454545 012 1 01479 602 850706 38 000 151 250 000 H og Y wok o om ow wed 373 Commands R squared 949765 Schwarz 36 3245 Adjusted R squared 943485 Log likelihood 34 0219 Estimated Standard Variable Coefficient Error t statistic P value 22 0000 5 54977 3 96412 1 004 iE 11 0000 894427 12 2984 000 Example 2 short label output regopt shortlab olsq t2 c t Equation 2 Method of estimation Ordinary Least Squares Dependent variable T2 Current sample 1 to 10 Number of observations 10 YMEAN 38 5000 S 8 12404 DW 454545 lt 012 SBIC 36 3245 SDEV 34 1736 RSQ 949765 JB 1 01479 602 LOGL 34 0219 SSR 528 000 ARSQ 943485 RESET2 850706E 38 000 52 66 0000 LMHET 391605 531 FST 151 250 000 Estimated Standard Variable Coefficient Error t statistic P value 22 0000 5 54977 3 96412 1 004 T 11 0000 894427 12 2984 000 Example 3 maximal output regopt pvprint stars bplistz c t Ilmlagsz2 qlags z2 noshort all options signif 8 increase width of displayed num
426. t Cauchy exponential gamma or an empirical distribution The user may specify optionally parameters of the distribution RANDOM CAUCHY DF lt scalar gt DRAW lt series gt or matrix EDF lt series gt EXPON GAMMA GEN 1 or 2 LAMBDA lt scalar gt LAPLACE MEAN lt scalar gt or series NEGBIN POISSON REPLACE SEEDIN lt scalar gt SEEDOUT lt scalar gt STDEV lt scalar gt or lt series gt T UNIFORM VAR lt scalar gt or series VCOV lt variance matrix gt VMEAN lt mean vector gt lt series gt or lt matrix gt or lt list of series gt Usage RANDOM with no options causes a normal random variable with mean zero and variance one to be generated and stored as a series under control of the current SMPL If you want a non standardized random variable include the MEAN and STDEV options Other options cause the random variable generated to follow the Poisson negative binomial uniform t Cauchy exponential gamma Laplace double exponential or general empirical distributions See the Examples Section to learn how to obtain random variables from other distributions TSP randomizes the seed to start every run based on the current time so you need to specify a fixed seed if you want to reproduce results from run to run To change the starting seed for any given run or to save it for future use use the SEEDIN and SEEDOUT options If multivariate normal random deviates are desired the VCOV option
427. t SHOW listname SHOW PROC or SHOW procname are useful if you want to call one of the procedures you have defined but do not remember the number or order of the arguments to pass it SHOW by itself lists the internal array dimensions in TSP This is helpful if you have a very large problem Output SHOW SERIES stores a list of all the series under RNMS class information provided by SHOW EQUATION name type formula or identity number of arguments and operations LIST name of members MATRIX name dimensions type 401 Commands MODEL name PROC name formal arguments SCALAR name type value SERIES name obs beg date end date frequency Options DATE NODATE prints the last date modified on a separate line if the variable has documentation created with the DOC command DOC NODOC prints any documentation on a separate line When DOC is off a portion of the documentation which fits on the end of the current line is printed Examples Assume the following TSP commands have been given SMPL 1 10 TREND T T2 T T OLSQ T2C T SHOW would then produce the following results SHOW SERIES Class Name Description SERIES RES 10 obs from 1 10 no frequency FIT 10 obs from 1 10 no frequency T2 10 obs from 1 10 no frequency T 10 obs from 1 10 no frequency SHOW MATRIX Class Name Description MATRIX VCOV 2x2 symmetric SES 2 1 general COEF 2 1 general
428. t Squares Programs for the Electronic Computer from the Point of View of the User JASA 1967 pp 818 841 Mackinnon James G and Halbert White Some _ heteroskedasticity consistent covariance matrix estimators with improved finite sample properties Journal of Econometrics 29 pp 305 325 Maddala G S Econometrics McGraw Hill Book Company New York 1977 pp 104 127 257 268 Pindyck Robert S and Daniel L Rubinfeld Econometric Models and Economic Forecasts McGraw Hill Book Company New York 1976 Chapter 2 3 4 Wampler Roy H Test Procedures and Test Problems for Least Squares Algorithms Journal of Econometrics 12 pp 3 21 297 Commands OPTIONS Options Examples OPTIONS is used to set various options for the TSP run OPTIONS APPEND ARGSUB BASEYEAR value CHARID CRT DATE DEBUG DISPLAY lt monitor type gt DOUBLE FAST HARDCOPY INDENT lt of spaces LEFTMG lt left margin LIMCOL lt column width for input LIMERR lt maximum of errors LIMNUM maximum of numerical errors LIMPRNz printer line width LIMWARNz maximum of warning messages printed LIMWMISSz maximum of missing value warning messages printed LIMWNUMCz maximum of numeric warning messages printed per command LINLIM lt lines per printer page MEMORYz size of memory for TSP NWIDTH lt of digits printed PLOTS REPL RESID SECONDS lt of seconds SIGNIF lt of significant digits printe
429. t against misspecification even when the data are overdispersed see Cameron and Trivedi for further information on this point POISSON also stores some of these results in data storage for later use The table below lists the results available after a POISSON command variable type length description LHV list 1 Name of dependent variable RNMS list vars List of names of right hand side variables IFCONV scalar 1 1 if convergence achieved 0 otherwise YMEAN scalar 1 Mean of the dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations HIST vector values Frequency counts for each dependent variable value HISTVAL vector values Corresponding dependent variable values SSR scalar 1 Sum of squared residuals RSQ scalar 1 correlation type R squared OVERDIS scalar Overdispersion test OVERDIS scalar 1 p value for overdispersion test LR scalar 1 Likelihood ratio test for zero slope coefficients LR scalar 1 P value for likelihood ratio test LOGL scalar 1 Log of likelihood function SBIC scalar 1 Schwarz Bayesian Information Criterion 338 POISSON NCOEF scalar 1 Number of independent variables vars NCID scalar 1 Number of identified coefficients COEF vector vars Coefficient estimates SES vector vars Standard errors T vector vars T statistics vector vars p values for T statistics GRAD vector vars Gradient of log likelihood at conv
430. t be used for equations FORMAT DATABANK specifies that the data are to be written to a TSP databank This option requires the FILE option also FORMAT EXCEL writes an Excel spreadsheet file similar to FORMAT LOTUS If the filename ends XLS this is the default This option requires the FILE option also FORMAT FREE specifies that the data is to be written to an external file in a format determined by TSP This option causes the data to be represented by six numbers per record with a field width of 15 characters and at least 7 significant digits The exact format is chosen by the program to represent the numbers most conveniently 450 WRITE FORMAT LABELS specifies that the data is to be formatted as for printed output with observation labels if they are series and row and column numbers and titles if they are vectors or matrices This option is the default if the output is being written to the output file or screen or if the item being written is an equation FORMAT LOTUS writes a Lotus 123 WKx worksheet file The variable names are written in the first row atop the series columns If a FREQ other than N is in effect dates are written in the leftmost column FORMAT LOTUS is the default if the filename string includes WK See the READ command FORMAT RBA is the same as FORMAT BINARY single precision binary FORMAT RBS is used for double precision binary FORMAT format text string specifies a fixed format with whi
431. t to FRML E Y A B X G Z 2 RHO Y 1 A B X 1 G Z 3 Note that when the EQSUB command is given all the variables in the equations must exist either as series PARAMs CONSTs or other FRMLs so that EQSUB will know which ones need to be lagged the series and FRMLs and which ones don t need lags the PARAMs and CONSTS Output 139 Commands Normally the output equation is stored silently replacing the input equation or creating a new equation If the PRINT option is on the output equation is printed Options LAGS NOLAGS controls substitution for the dependent variable name when it is lagged When the NOLAGS option is specified only the unlagged appearances of the dependent variable are substituted for NAME new output equation name supplies a new name for the output equation If this option is not present the main equation is overwritten by the new one PRINT NOPRINT controls whether the output equation is printed Example See above See also the TSP User s Guide for many additional examples of using EQSUB to set up log likelihood equations for estimation by ML Here is one more example illustrating the use of DOT to substitute input equations F1 to F20 into main equations E1 to E8 DOT E1 E8 EQSUB F1 F20 ENDDOT 140 EXEC Interactive EXEC Interactive EXEC forces execution or re execution of a range of lines consisting of TSP commands that have already been entered in th
432. ta matrix X as it is done in TSP s regression calculation to obtain more accurate results Even if S is not determined very accurately due to inaccuracy in forming the cross product matrix X X a regression run on the transformed Xs and then untransformed will produce extremely accurate results since the actual matrix inversion is performed on an X X matrix from which most collinearity has been removed Output ORTHON produces no output but two matrices are stored in data storage Method ORTHON forms X X from the X matrix factors it using the Choleski factorization algorithm inverts the result using the method described in MATRIX and postmultiplies X by the resulting upper triangular matrix Example The following example shows how to use ORTHON in programming ordinary least squares explicitly in TSP MMAKE X C X1 X2 ORTHON X S XTILDA MAT XTXINV z XTILDA XTILDA MAT BETA z S XTXINV XTILDA Y MAT XXINV S XTXINV S The resulting BETA and XXINV are estimates of the untransformed coefficients and the inverse of the matrix 306 OUT Databank OUT Databank Examples OUT specifies a list of external files on which all TSP variables created or modified will be stored OUT lt list of filenames gt or filename strings Usage Follow the word OUT with the names of the TSP databank s on which you wish to store your variables On most computers these are binary TLB files Up to 8 databanks may be acti
433. tain dates This will ensure the series are read with the proper frequency and starting date regardless of the current FREQ and SMPL in TSP If you have dates in other columns they will be read as numbers If you are reading a matrix the date column will be ignored i e it will not be read into the matrix Dates can be strings such as 48 1 or numbers formatted as dates Mar 48 3 31 48 etc You only have to supply enough dates so that TSP can detect the frequency 5 is enough to distinguish between quarterly and monthly TSP ignores any dates after these assuming that the data is contiguous no missing periods years or SMPL gaps in TSP terminology If you have missing periods years for all series leave the corresponding rows blank Below is a table of examples showing recommended ways of defining the starting date and frequency with a dates column string dates first character is second date resulting frequency 2 A if current FREQ is A otherwise N any A any M 49 1 Q 1949 1 A any invalid numeric dates second date resulting frequency 12 31 48 12 31 49 A any dates 365 366 days apart 12 31 48 1 31 49 M any dates 28 31 days apart 12 31 48 3 31 49 Q any dates 90 92 days apart 12 31 48 1 1 49 N any other date range 365 Commands 4 Missing values can be represented by blank cells or by formulas which evaluate to NA NA N A etc To read in spreadsheet file use the FILE f
434. tations on VAR ENDDOT operations VARLIST 225 Commands This enables you to use the variables as a group or as individual variables in GENR since listnames are not allowed in GENR SET or equation specifications A special list named ALL is always available it contains the names of all the series in the current TSP run Output LIST produces no printed output A TSP variable list is stored in data storage SHOW LIST can be used to see the currently defined lists and LIST PRINT can be used to view the contents of one or more lists Options DELETE NODELETE deletes existing lists DROP NODROP drops variables from an existing list FIRST starting integer for a sequence of names in a list The default is 1 FIRST can be larger than last to make a list in decreasing order It cannot be negative LAST ending integer for a sequence of names in a list The default is 1 and it cannot be negative PREFIXz name to be used as the prefix in constructing a list in the form namefirst namelast See the examples PREFIX can also be used to append a list of different names to a common prefix SUFFIX name or number to be added to all the variable names in the list PRINT NOPRINT prints the contents i e included names of existing lists Examples A few simple examples LIST EQ EQ1 EQ4 creates LIST EQ EQ1 EQ2 EQ3 EQ4 LIST EQS 001 004 creates LIST EQS EQ01 EQ EQ03 EQ04 LIST LAGS X
435. te extension of a chain index A Divisia index of quantity can be obtained by applying the same strategy to quantities in place of prices or alternatively by dividing total expenditure by the price index However the two quantity indices will not be exactly the same DIVIND PNORMz obs id PRINT PVAL lt value gt QNORM lt obs id QVAL lt value gt TYPE Q or P or WEIGHT CONB or ARITH or GEOM name of output price index name of output quantity index list of pairs of input price and quantity series Usage DIVIND has as its arguments the name to be given to the computed price index then the name to be given to the computed quantity index and finally the names of the series for prices and quantities of the components to be used as input to the calculations The order is price for input one quantity for input one price for input two quantity for input two and so forth No warning is given for non positive prices for a quantity index and vice versa the formulas still hold unless WEIGHT GEOM When a quantity is zero for one or more periods the series are spliced and the price is temporarily omitted from the index Output Normally DIVIND produces no printed output but stores the two computed index series in data storage If the PRINT option is on DIVIND prints a title the options the names of the input and output series and a table of the two computed series labelled by the observation name Options PN
436. tely DBUS United States DBUK United Kingdom DBSW Sweden and DBD Germany The same set of series is stored in each of the databanks using the KEEP statement 211 Commands KERNEL Options References KERNEL computes a kernel density estimation or regression Kernel estimation is semi parametric method for approximating a probability distribution KERNEL BANDWIDTH lt bandwidth gt RELBAND crelative bandwidth gt IQR variable Or KERNEL BANDWIDTHz bandwidth RELBANDz relative bandwidth gt IQR dependent variable independent variable Usage When KERNEL is used with a single argument a Gaussian kernel density of the variable is computed and stored in DENSITY You can display the result using a GRAPH command with the variable and DENSITY as arguments When KERNEL is used with two arguments a Gaussian kernel regression of the first variable on the second is computed the smoothed values of the dependent variable are stored in FIT The default bandwidth for both estimators is RELBAND 1 which uses Silverman s default bandwidth h RELBAND h0 9 NOB 2 where 0 is the standard deviation of the independent variable for the default NOIQR option and the minimum of the standard deviation and the interquartile range divided by 1 349 for IQR When the number of observations is one h 1 is used For values of RELBAND lt 1 the fit is closer to the data less smooth while values of RELBAND
437. terpreted as the ratio of the harmonic and arithmetic means of the T i over the sample of individuals Note that AP is always less than or equal to 7 and that it equals one only when T i T for all i Options ALL NOALL turns all regressions on or off equivalent to the combination of TOTAL BETWEEN WITHIN VARCOMP REI REIT BETWEEN NOBETWEEN selects the between estimator a regression on the means for each individual BYID NOBYID does a separate regression for each individual and computes F tests for equality with the TOTAL and WITHIN estimators FEPRINT NOFEPRINT specifies that the fixed effect estimates are to be printed as well as stored 317 Commands HCOMEGA BLOCK or DIAGONAL specifies the form of the Efuu matrix to use when computing ROBUST standard errors Ordinarily the default is BLOCK for PANEL which allows for cross time correlation of the disturbances within individuals This feature can be used for any kind of grouped data simply by ensuring that the relevant PANEL setup has been defined HCTYPE 0 or 1 specifies whether to apply a degrees of freedom correction to the robust s e s 0 is no and 1 is yes ID the name of a series which takes on a different value for each individual The default is ID alternatives are the T and TIME options MEAN NOMEAN causes the means for each individual to be printed in a table This can be used in conjunction with the NOREG option to print means only to
438. the filename of the TSP databank whose series are to be printed Make sure the FREQ if any and SMPL have been set The FREQ and SMPL information on series from a DBLIST command may be useful for setting FREQ and the SMPL range Any series not stored with the FREQ currently in effect will not be printed TSP variables other than series such as matrices and FRMLs will not be printed Parameters and constants can be printed with the DBLIST command while matrices and equations can be printed using IN and PRINT together The series printed are only temporarily brought into memory they are only stored when they have been accessed with an IN command This protects any currently stored series from being overwritten by a series in the databank with the same name Only one databank can be printed in a DBPRINT command Example Suppose that you have a databank called FOO TLB which contains two time series X and Y and a parameter B The commands FREQ Q SMPL 60 1 60 4 DBPRINT FOO would yield VALUES FOR ALL SERIES IN DATABANK FOO TLB X Y 1960 1 1 00000 11 0000 1960 2 2 00000 22 0000 1960 3 3 00000 33 0000 1960 4 4 00000 44 0000 109 Commands DEBUG Examples DEBUG turns the DEBUG switch on When this switch is on TSP produces a great deal more printed output than it usually does This output is normally not of interest to users but may be helpful to a TSP programmer or consultant DEBUG Usage Include t
439. the input equations EQSUB LAGS NAMEz new output equation name gt PRINT main equation name list of input equation names Usage Define the main equation and the input equation s with FRML or IDENT statements The EQSUB command substitutes each input equation into the main equation in order from left to right For each input equation EQSUB looks for its dependent variable or equation name if there is no dependent variable in the argument list of the main equation If the dependent variable is found the code from the input equation is inserted into the main equation and the old variable name is deleted For example FRML EQ1 Y A XB FRML EXB XB X1 B1 X2 B2 EQSUB EQ1 EXB is equivalent to FRML EQ1 Y A X1 B1 X2 B2 The resulting equation replaces the main equation unless the NAME option is supplied The DOT command is useful when there are several different main equations If you have many component input equations to define it may be convenient to leave out the dependent variable name so that you don t have to invent both a dependent variable and an equation name for each one Such an equation is called unnormalized in TSP For example FRML E Y1 XB FRML XB BO B1 X1 B2 X2 change XB to add delete exog errors variables FRML TOBIT LOGL YPOS LNORM E SIGMA LOG SIGMA YZERO LCNORM XB SIGMA EQSUB NAME TOBIT1 TOBIT E XB 138 EQSUB Note that both TOBIT and E de
440. the lag polynomial and the value of the constraint options These scrambled variables are used as regressors and then unscramblea after estimation to obtain the implied lag coefficients Further details are given in the TSP User s Guide Almon 1965 The method TSP uses is described Cooper 1972 It uses Lagrangian interpolation polynomials that are orthogonal in order to minimize multicollinearity problems See the Shiller 1973 reference for further details on the method of estimation for Shiller lags Examples OLSQ I C GNP 4 16 FAR specifies a distributed lag on GNP which covers 16 periods where the lag coefficients are constrained to lie on a third degree polynomial and go to zero at the 16th lag OLSQ I C GNP 4 16 FAR R 4 24 NEAR adds another distributed lag in R which covers 24 periods and is constrained to go to zero at the first lead PDL variables can also be used with INST and AR1 INST I C GNP 3 5 NONE INVR C LM LM 1 LM 2 AR1 METHOD CORC CONS C GNP 3 5 NONE The instrumental variable estimation specifies just enough instruments for the number of independent variables which will appear in the estimation the constant and 3 weighted combinations of GNP and its lags References Almon Shirley The Distributed Lag between Capital Appropriations and Expenditures Econometrica 33 January 1965 pp 178 196 Cooper J Phillip Two Approaches to Polynomial Distributed Lags Estimation
441. the left margin The default value is 132 which is correct for most high speed printers Occasionally these printers have only 120 positions and your local installation may change the default accordingly LIMWARN maximum number of warning messages to print The default value is 100000 LIMWMISS maximum number of warning messages about missing values to print The default is 10 LIMWNUMC maximum number of numeric warning messages to print in any particular command The default value is 10 This means that each command will print at most 10 numeric warning messages and then the remainder for that command will be suppressed LINLIM number of lines per printer page The default is 60 which is correct for most conventional printed output MEMORY approximate memory used by TSP in MB This option only works if OPTIONS is the first command in the run or the first command in the login tsp file The default is 4MB and the minimum is 2 1MB Calculate memory as 2MB plus 4MB per million words of working space desired MEMORY 4 should be enough for most time series datasets and small cross sections The memory actually used is printed at the end of the TSP run NWIDTH maximum number of digits to be printed for numbers in tables This is the number of columns allowed for each number and the default value is 13 300 OPTIONS PLOTS NOPLOT tells whether residual plots are to be printed following each estimation See the PLOTS command des
442. the number of coefficients the grouped panel estimates of the variance covariance matrix will be singular and therefore probably inappropriate However in most cases NI K and this problem will not arise Note also that for fixed effect estimation the gradient is always zero for the fixed effects at the optimum so the block of V corresponding to these effects is zero For this reason standard errors for the fixed effects are always computed using the Newton second derivative matrix HESSCHEC NOHESSCH compares analytic and discrete Hessian differenced analytic gradient 280 Nonlinear Options HITER B or N or G or F or D Specifies the method of Hessian second derivative matrix approximation to be used during the parameter iterations The options are the same as those described above for the estimate of the covariance matrix of the parameter estimates Table of HCOV and HITER options Used Name and description Procedures to for which it is iterate the default 2 yes BHHH Berndt Hall Hall ML Hausman Covariance of the analytic gradient yes Newton Analytic second ARCH derivatives iterations AR1 PROBIT TOBIT LOGIT SAMPSEL yes GAUSS Gauss Newton LSQ FIML Quadratic form of the iterations analytic gradient and the residual covariance matrix yes BFGS Broyden Fletcher None Goldfarb Shanno Analytic or numeric first derivatives and rank 1 update approximation of the Hessian from iterations Usuall
443. the values of the endogenous variables and the vector of changes at each iteration If PRNRES is on the residual error from each equation is printed when convergence is achieved or the maximum number of iterations is reached The Jacobian is printed for the first two periods After solution of the model over the whole sample a message is printed if the variables are being saved in data storage Following this message a table of the results of the simulation is printed labelled by the observation IDs To suppress the table use the NOPRNSIM option IFCONV is stored as a series with ones if the simulation for the observation converged and zeroes otherwise This may be useful for a convergence check since this information may otherwise be hidden in a large output file full of iteration information Method If a simultaneous model is linear it can be written as Ax b where Ais a matrix of coefficients x is the set of endogenous variables and b is a vector of numbers which may include functions of exogenous variables This model can be solved directly by inverting A and multiplying b by it Newton s method applies this idea to the iterative solution of nonlinear models in the following way At each iteration the model is linearized in its variables around the values from the previous iteration The linearized model is solved by matrix inversion The resulting set of new values is treated as a direction vector for a linear search for a be
444. this feature with a NOFAIR option Fair retracted his claim in 1984 it has since been disproved by Buse 1989 but the alternative instruments for consistency involve pseudo differencing with the estimated rho Theil s G2SLS which is tedious to perform by hand Buse also showed that the asymptotically most efficient estimator in this case S2SLS includes the lagged excluded exogenous variables as well but he cautions that in small samples this may quickly exhaust the degrees of freedom METHOD ML or MLGRID or CORC or HILU was formerly used to specify the estimation algorithm This is now specified by the OBJFN option METHOD ML or MLGRID imply while METHOD CORC or HILU imply OBJFN GLS METHOD ML formerly used the Beach and McKinnon algorithm while METHOD CORC used the Cochrane Orcutt algorithm Now iterations are done using the Newton Raphson algorithm HITER N the nonlinear options which is quadratically convergent about the same speed as Beach MacKinnon but much faster and more accurate than Cochrane Orcutt METHOD HILU refers to Hildreth Lu a simple grid search method 44 AR1 OBJFN EXACTML or GLS specifies the objective function EXACTML retains the first observation and includes the Jacobian term log 1 rho 2 which guarantees stationarity GLS drops the first observation and does not impose stationarity It is the same as nonlinear least squares on a rho differenced equation and can also
445. this period and the previous period WEIGHT COMEB specifies that the weights are the geometric averages of i the arithmetic average ii the share this period and iii the share in the previous period Either QNORM or is required if TYPE P Q and both are required if TYPE N The default values of the options are the following TYPE Q WEIGHT COMB NOPRINT QVAL 1 PVAL 1 Examples DIVIND WEIGHT ARITH TYPE P PNORM 67 PRICEIN QUANTIN PS QS PND QND PD QD 117 Commands FREQ Q DIVIND WEIGHT GEOM TYPE N PNORM 75 1 PVAL 100 75 1 QVAL 100 PRINT GI P1 01 P2 Q2 P3 Q3 P4 Q4 The first of these example computes a Divisia price index as a weighted average of changes in PS PND and PD using the shares of PS QS PND QND and PD QD in total expenditure as weights The series PRICEIN is normalized to have the value 1 0 in 1967 and QUANTIN is derived by dividing PRICEIN into total expenditure The second example computes a price index and quantity index QI independently from quarterly data Both indices are normalized to have the value 100 in the first quarter of 1975 The weights are geometric averages of the shares in the adjacent years References Jorgenson Dale W and Zvi Griliches Divisia Index Numbers and Productivity Measurement Review of Income and Wealth Vol 17 2 June 1971 pp 227 229 Diewert W Erwin Exact and superlative index numbers Journal of
446. tic 1 0000 2 0000 Lower tail area 01327 7 Dickey Fuller unit root test TREND TIME SMPL 2 50 DY Y Y 1 OLSQ DY TIME C Y 1 CDF DICKEYF T 3 The above is equivalent to the following UNIT MAXLAG 0 NOWS Y 8 Augmented Dickey Fuller unit root test with finite sample P value TREND TIME SMPL 2 50 DY Y Y 1 SMPL 5 50 OLSQ DY TIME C Y 1 DY 1 DY 3 CDF DICKEYF NOB NOB NLAGS 3 T 3 The above is equivalent to the following UNIT MAXLAG 3 NOWS FINITE Y 9 Engle Granger cointegration test TREND TIME 01 50 Y1 TIME 2 4 EGTEST 01 50 2 TIME 1 4 EGTEST 01 50 TIME 1 2 4 EGTEST 01 50 Y4 TIME Y1 Y2 EGTEST PROC EGTEST SMPL 2 50 DU RES RES 1 01 50 DU RES 1 CDF DICKEYF NVAR 4 T SMPL 1 50 78 CDF ENDPROC The above is equivalent to the following COINT NOUNIT MAXLAG 0 ALLORD Y1 Y4 10 Verify critical values for Durbin Watson statistic for regression with 10 observations and 2 RHS variables SMPL 1 10 OLSQ X1 SET 4 ATAN 1 SET PI 2 NOB TREND I EIGB 4 SIN I F 2 SELECT lt NOB NCOEF use largest eigenvalues for dL CDF WTDCHI EIG EIGB 879 dL for 5 n 10 k 1 use smallest eigenvalues for dU SELECT NCOEF lt 1 amp I lt NOB 1 MMAKE dU 1 320 1 165 1 001 dU for 5 2 5 1 n 10 k 1 CDF WTDCHI EIG EIGB PRINT dU
447. tics INTERVAL also stores some of these results in data storage for later use The table below lists the results available after an INTERVAL command variable type length description LHV list 1 Name of dependent variable RNMS list vars List of names of right hand side variables IFCONV scalar 1 1 if convergence achieved 0 otherwise scalar NOB scalar LOGL scalar AIC scalar SBIC scalar scalar Mean of the dependent variable Number of observations Log of likelihood function Akaike information criterion Schwarz Bayesian information criterion Number of independent variables vars NCID scalar 1 Number of identified coefficients COEF vector vars Coefficient estimates SES vector vars Standard errors T vector vars T statistics vector vars p values for T statistics GRAD vector vars Gradient of log likelihood at convergence VCOV matrix vars Variance covariance of estimated vars coefficients FIT series obs Fitted values of dependent variable If the regression includes a PDL or SDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag coefficients after unscrambling Method 201 Commands Like the binary and ordered Probit models the Interval model is based on an unobserved continuous dependent variable y The model
448. tima with rho gt 1 which are usually not reached during iterations when rho starts below 1 RSTART specifies a starting value of rho for the iterative methods Ordinarily zero is used for OBJFN EXACTML but faster convergence may be achieved if a value closer to the true answer is chosen RSTART can also be used to override the default grid search for OBJFN GLS but multiple local optima would not be detected TSCS NOTSCS specifies EXACTML estimation for time series cross section data when the FREQ PANEL command is in effect then TSCS is the default or when SMPL gaps have been set up to separate the cross section units see the example below OBJFN 2GLS is not implemented for panel data 45 Commands Obsolete WEIGHT is a former AR1 option which is no longer supported The ML or LSQ commands should be used instead to implement a weight Nonlinear options are described under NONLINEAR in this manual HITER N HCOVEN second derivatives the default and G first derivatives are both available MAXIT 0 can be used to avoid iterations and to perform a simple grid search without the additional accuracy of iterations Also AR1 uses a special default TOL 1E 6 000001 Examples This example estimates the consumption function for the illustrative model with a serial correlation correction first using the maximum likelihood method and then searching over rho to verify that the likelihood is unimodal in the relevant range
449. tion function LIST RES RES1 RES10 RANDOM EDF RES RES When we are done the list RES consists of ten series which have the same distribution as the original residual series RES Example with inverse distribution functions RANDOM SEED 94298 UNIFORM U EV LOG LOG U Type Extreme Value CDF INV CHISQ DF n CHIV Chi square n An example using the matrix version of DRAW to draw data from an empirical distribution function in order to investigate the potential rate of convergence of a particular estimator original data set has 100 observations compute residuals 359 Commands SMPL 1 100 OLSQYCX UNMAKE COEF AB EDF RES estimation using 1000 obs drawn from EDF SMPL 1 1000 RANDOM DRAW EDF E Y A B X E OLSQYCX estimation using 10 000 obs drawn from same EDF SMPL 1 10000 RANDOM DRAW EDF EX Y A B X E OLSQYCX Draw 5 cards from a deck of 52 without replacement SMPL 1 52 TREND OBS SUITE 1 INT OBS 1 13 TREND PER 13 NUMBER MMAKE CARDS SUITE NUMBER SMPL 15 RANDOM DRAW CARDS NOREPL SUITE NUMBER Permute a series of residuals SMPL 1 100 OLSQ YC X1 2 RANDOM DRAW RES NOREPL U Verify that the new uniform generator is properly implemented check sum of first 10 000 000 r v s for seed 12345 OPTIONS DOUBLE MEMORY 5 SMPL 1 100000 RANDOM GEN 2 SEEDIN 12345 SET TOTAL 0 DO 1 1 100 RANDOM UNIFORM
450. tion names gt Method Seemingly unrelated regression estimates are obtained by first estimating a set of nonlinear equations with cross equation constraints imposed but with a diagonal covariance matrix of the disturbances across equations These parameter estimates are used to form a consistent estimate of the covariance matrix of the disturbances which is then used as a weighting matrix when the model is reestimated to obtain new values of the parameters These estimates are consistent and asymptotically normal and under some conditions asymptotically more efficient than the single equation estimates The seemingly unrelated regression method is a special case of generalized least squares with a residual covariance matrix of a particular structure V 91 where T the number of observations and Sigma is the matrix of cross equation variances and co variances It is sometimes called Zellner s method since it was originally proposed for linear models by Arnold Zellner or the Aitken estimator of which it is a special case Options These are the same as for LSQ except that NOITERU and MAXITW 0 are the defaults in order to obtain SUR estimates References Judge et al The Theory and Practice of Econometrics John Wiley and Sons New York 1980 pp 245 250 Theil Henri Principles of Econometrics John Wiley and Sons New York 1971 pp 294 311 422 SUR Zellner Arnold An Efficient Method of Estimating Seem
451. tly higher median bias than plain LIML but it is mean unbiased and it has a smaller MSE than LIML it has finite moments Following this is a table of right hand side variable names estimated coefficients standard errors and associated t statistics If the variance covariance matrix has not been suppressed see the SUPRES command it is printed after this table Finally if the RESID and PLOTS options are on a table and plot of the actual and fitted values of the dependent variable and the residuals is printed LIML also stores most of these results in data storage for later use The table below lists the results available after a LIML command The fitted values and residuals will only be stored if the RESID option is on the default variable type length description LHV list 1 Name of the dependent variable RNMS list vars Names of right hand side variables SSR scalar 1 Sum of squared residuals S scalar 1 Standard error of the regression 952 scalar 1 Standard error squared 220 LIML FOVERID scalar LAMBDA scalar F test of overidentifying restrictions LIML eigenvalue YMEAN scalar 1 Mean of the dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations DW scalar 1 Durbin Watson statistic RSQ scalar 1 R squared ARSQ scalar 1 Adjusted R squared FST scalar 1 F statistic for zero slopes PHI scalar 1 The objective function sum e 1 1 MU2
452. ts in data storage for your later use The table below lists the results available variable type length description LHV list 1 Name of the dependent variable RNMS list vars Names of right hand side variables SSR scalar 1 Sum of squared residuals S scalar 1 Standard error of the regression 952 scalar 1 Standard error squared scalar 1 Mean of the dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations 292 OLSQ DW scalar 1 Durbin Watson statistic DH scalar 1 Durbin s h lagged dependent variable DHALT scalar 1 Durbin s h alternative lagged dependent variables RSQ scalar 1 R squared ARSQ scalar 1 Adjusted R squared FST scalar 1 F statistic for zero slopes LMHET scalar 1 LM heteroskedasticity test JB scalar 1 Jarque Bera LM test for normality of residuals RESET2 scalar 1 Ramsey s RESET test of order 2 for missing quadratic Xs LOGL scalar 1 Log of likelihood function SSRO scalar 1 SSR for Original data in weighted regression 0 scalar 1 9520 SO ARSQO all for the unweighted data COEF vector vars Coefficient estimates SES vector vars Standard errors T vector vars T statistics VCOV matrix vars vars Variance covariance of estimated coefficients RES series obs Residuals actual fitted values of the dependent variable FIT series obs Fitted values of the dependent variable HI series ob
453. tter set of values 404 SIML The criterion function for SIML is the sum of squared deviations of the equations At the solution all the deviations will be zero Away from the solution the deviations are computed by substituting the current values of the variables into the equations and evaluating them This sum of squared deviations is the objective function printed out by SIML at each iteration When the model is linear in the variables the model is simply solved by the matrix equation above and no iteration is done TSP s implementation of Newton s method uses an analytic Jacobian evaluated at the current variable values as the matrix A this is an extremely powerful method for finding the solution of a nonlinear model but it can still run into trouble primarily because of near singularity of the Jacobian If this happens a message is printed and you may wish to try the GAUSSN method which uses a generalized inverse to try to get past a locally singular point Options DEBUG NODEBUG prints the endogenous variables and direction vector at each iteration for debugging recalcitrant models DYNAM NODYNAM specifies dynamic simulation Earlier solved values of lagged endogenous variables are used in place of actual values STATIC is the alternative to DYNAM DYNAM is the default unless there are no lagged endogenous variables ENDOG a list of the endogenous variables in the model This is the list of variables for which the
454. uch components obtained may be a fixed number or it may be determined by the amount of variance in the original series explained by the principal components Principal components are a set of orthogonal vectors with the same number of observations as the original set of series which explain as much variance as possible of the original series Users of this procedure should be familiar with the method and uses of principal components which are described in many standard texts such as Harman 1976 or Theil 1971 PRIN FRACz fraction of variance NAME lt name of components NCOMz number of components PRINT list of series Usage To obtain principal components in TSP give the word PRIN followed by a list of series whose principal components you want The options determine how many principal components will be found The resulting principal components are also series and are stored in data storage under the names created from the NAME option Output If PRINT is on the output of the principal components procedure begins with a title the list of input series the number of observations and the correlation matrix of the input series This is followed by a table for the components showing the corresponding characteristic root and the fraction of the variance of the original series which was explained by all the components up to and including this one Finally a table of factor loadings is printed this table shows the weig
455. um likelihood estimation with FIML or for simulation with SIML or SOLVE IDENT equation name variable name z algebraic expressions or IDENT equation name algebraic expressions Usage There are two forms of the IDENT statement the first has the name to be given to the equation followed by an equation in normalized form that is with the name of the dependent variable on the left hand side of the equal sign and an algebraic expression for that variable on the right hand side The expression must be composed according to the rules given in the Basic Rules section in this manual These rules are the same wherever an equation is used TSP in IF statements SET FRML and IDENT The second form of the IDENT statement is implicit there is no equal sign but simply an algebraic expression This is used for fully simultaneous models where it might not be possible to normalize the equations The FIML and SIML procedures can process implicit equations although the SOLVE procedure cannot Equations defined by IDENT are the same as those defined by FRML except that the estimation procedures assume that a FRML has an implied additive disturbance tacked on the end while an IDENT does not The distinction is useful only in FIML where identities may be necessary to complete square the Jacobian An equation defined by a IDENT statement can contain numbers parameters constants and series The equation can alwa
456. upplied for S the standard error of the disturbance and if no estimate from BJEST is available BJFRCST will calculate an estimate of S using the values of the time series up to and including the ORGBEG th observation 63 Commands BJFRCST calculates forecasts for the time series for the NHORIZ periods following the current origin The origin is the final period for which conceptually the time series is observed i e the final period for which no forecast is required In practice it is often useful to generate forecasts for periods that have already been observed These forecasts provide evidence on the reliability of the estimated time series model It is also useful to generate sets of forecasts for different origins BUFRCST calculates a complete set of forecasts and associated confidence bounds for each of the origins from ORGBEG to ORGEND inclusive Output The output of BUFRCST starts with a printout of the option settings The parameter values are also printed This is followed by a printout of the expanded Phi and Theta polynomials implied by these parameters The Phi and Theta polynomials are the generalized autoregressive and moving average polynomials respectively that are obtained by taking the product of the ordinary seasonal and pure difference polynomials Next a table is printed of the forecast standard errors and the psi weights Note that the standard error in the N th row of the table is the standard erro
457. urement and transition equations respectively 204 KALMAN The y t and X t variables are the dependent and independent variables just as in ordinary least squares If you want to use more than one dependent variable list all the y variables first then a and then list the X variables for each y duplicate the X list if they are the same for every y You may want to insert zeros along with the X variables to prevent cross equation restrictions see the Examples If X t is fixed over the sample use the XFIXED option To get a time varying parameter model specify VTRANS Q the noise to signal ratio and BTRANS T if it is not the identity matrix To evaluate the likelihood function for general ARMA p q models fill the BTRANS and VTRANS matrices with the estimated coefficients for the model see Harvey p 103 for the general form Output A standard table of coefficients and standard errors is printed for the final state vector along with the log likelihood The following items are stored and may be printed variable type length description COEF vector m Final state vector SES vector m Standard errors T vector m T statistics VCOV matrix Variance covariance matrix LOGL scalar 1 Log of likelihood function SSR scalar 1 Sum of squared recursive residuals S2 scalar 1 Variance of recursive residuals RES1 matrix obs n Prediction errors one step ahead RECRES matrix obs n Recursive residuals i i d
458. ursby J A Comparison of Several Exact and Approximate Tests for Structural Shifts under Heteroskedasticity Journal of Econometrics 1992 363 386 Statlib http ib stat cmu edu apstat 385 Commands RENAME Examples RENAME changes the name of an old TSP variable series matrix constant etc RENAME lt old variable name gt lt new variable name gt Output The name of the variable is changed If a variable already exists with the new name it is deleted Examples Save the coefficients from a regression in the vector B1 Note this is more efficient than COPY if there is no reason to save the original COEF 0150 RENAME COEF 1 386 REPL REPL Examples REPL turns on the replacement mode option after it has been turned off with a NOREPL command This option specifies that series are to be updated rather than completely replaced when the current sample under which they are being computed does not cover the complete series OPTIONS REPL is the same as REPL REPL Usage While in REPL mode if data are created for a sample which overlaps with the SMPL previously used to create the same series the old values which fall within the current SMPL definition will be replaced but those outside the current SMPL will remain untouched REPL is the default mode and remains in effect until a NOREPL is executed Examples The result of the following sequence of statements REPL SMPL 1 20
459. us Equation Model Econometrica May 1977 pp 955 975 Berndt E K B H Hall R E Hall and J A Hausman Estimation and Inference in Nonlinear Structural Models Annals of Economic and Social Measurement October 1975 pp 653 665 Gallant A Ronald and Dale W Jorgenson Statistical Inference for a System of Simultaneous Non linear Implicit Equations in the Context of Instrumental Variable Estimation Journal of Econometrics 11 1979 pp 275 302 Jorgenson Dale W and Jean Jacques Laffont Efficient Estimation of Nonlinear Simultaneous Equations with Additive Disturbances Annals of Economic and Social Measurement October 1975 pp 615 640 Theil Henri Principles of Econometrics John Wiley and Sons New York 1971 pp 508 527 Zellner Arnold and Henri Theil Three Stage Least Squares Simultaneous Estimation of Simultaneous Equations Econometrica 30 1962 pp 54 78 432 TITLE TITLE Options Examples The TITLE statement is used to change the title on the top of the page of TSP printed output or to print a centered title underlined with equals signs on the current page TITLE PAGE Text string to be used as title Usage To change the TSP title follow the word TITLE with up to sixty characters of text enclosed in single quotes If OPTIONS HARDCOPY the default for non interactive jobs is in effect new page is started with the new title printed at the top of the page All furth
460. values or significance levels or critical values for several cumulative distribution functions This is useful for hypothesis testing CDF BIVNORM or CHISQ or DICKEYF or F or NORMAL or T or WTDCHI CONSTANT DF lt degrees of freedom for CHISQ or T gt DF1 lt numerator degrees of freedom for F gt DF2 lt denominator degrees of freedom for F gt EIGVAL lt vector of eigenvalues for WTDCHI LOWTAIL or UPTAIL TWOTAIL INVERSE NLAGS lt number of lags for augmented Dickey Fuller test gt NOB lt number of observations for unit root or cointegration test gt NVAR lt number of variables for cointegration test gt PRINT RHO lt correlation coefficient for BIVNORM gt TREND TSQ test statistic significance level a or lt significance level gt lt critical value gt for INVERSE or x value y value lt significance level gt for BIVNORM Usage CDF followed by the value of a scalar test statistic is the simplest form of the command In this case a two tailed probability for the normal distribution will be calculated and printed If the INVERSE option is used the first argument must be a probability level a critical value will be calculated Arguments need not be scalars they can be series or matrices Distributions other than the normal and or a choice of tail areas may be selected through the options For hypothesis testing using a wide variety of regression diagnostics see the REGOPT PVPRINT co
461. variable on the left is greater than the variable on the right otherwise it is zero LE gives the value 1 when the variable on the left is less than or equal to the variable on the right otherwise it is zero GE gives the value 1 when the variable on the left is greater than or equal to the variable on the Composing Algebraic Expressions in TSP right otherwise it is zero AND gives the value 1 when both the variable on the left and on the right are positive OR gives the value 1 when both the variable on the left and on the right are positive NOT gives the value 1 when the variable on the right is negative or zero Note the OP form of the relational and logical operators is the alternative to the symbolic notation but it cannot be used in nested DOT loops As many parentheses as necessary may be used to indicate the order of evaluation of a formula The special parentheses and are treated as In the absence of parentheses evaluation proceeds from left to right in the following order Functions Exponentiation Multiplication and division Addition subtraction and negation unary Relational operators AND amp and OR Introduction TSP Functions These functions can be used in any GENR FRML IF SET SELECT or MATRIX command Include the argument value series name or algebraic expression in the parentheses For additional matrix functi
462. variables IFCONV scalar 1 1 if convergence achieved 0 otherwise YMEAN scalar 1 Mean of the dependent variable SDEV scalar 1 Standard deviation of the dependent variable NOB scalar 1 Number of observations HIST vector values Frequency counts for each dependent variable value HISTVAL vector values Corresponding dependent variable values SSR scalar 1 Sum of squared residuals RSQ scalar 1 correlation type R squared SRSQ scalar 1 Scaled R squared LR scalar 1 Likelihood ratio test for zero slope coefficients LR scalar 1 P value for likelihood ratio test LOGL scalar 1 Log of likelihood function SBIC scalar 1 Schwarz Bayesian Information Criterion scalar 1 Number of parameters params NCID scalar 1 Number of identified coefficients COEF vector params Coefficient estimates SES vector params Standard errors T vector params T statistics vector params p values for T statistics GRAD vector params Gradient of log likelinood at convergence VCOV matrix params params Variance covariance of estimated coefficients 303 Commands FIT series obs Fitted values of dependent variable RES series obs Residuals actual fitted values of dependent variable If the regression includes a PDL or SDL variable the following will also be stored SLAG scalar 1 Sum of the lag coefficients MLAG scalar 1 Mean lag coefficient number of time periods LAGF vector flags Estimated lag
463. vations grows this is an alternative to the conventional testing procedure which is certain to reject all point null hypotheses when sample sizes become large It is based on a Bayesian flat prior and computed from the formula in the Leamer reference E 5 1 316 PANEL Where T total number of observations k number of estimated parameters in the unrestricted model and p the number of restrictions All regressions are computed with the standard orthonormalized data matrices to insure accurate coefficients and variance estimates under possible multicollinearity methods using moment matrices are less accurate The Durbin Watson test and bounds on its P values are computed following the Bhargava et al reference extended to the unbalanced data case The P values are computed using the Farebrother Imhof method since there can be multiple equal eigenvalues The REI estimates are obtained with a grid search over RHO in order to avoid the problem of multiple local optima Estimates are then refined to choose the global optimum and multiple optima are reported RHO is bounded between 1 Max T 1 and 1 where Max T is the maximum number of observations per individual See Maddala and Nerlove 1971 The REIT estimates are obtained using the method of Davis 2002 The Ahrens Pincus measure of unbalancedness in dimension is defined as follows t N a where 42 17 i l This can be in
464. vatives A message is printed if either form of linearity is found The amount of working space used by FIML is also printed this number can be compared with the amount printed at the end of the run to see how much extra room you have if you wish to expand the model Next FIML prints the starting conditions for the constants and parameters and then iteration by iteration output If the print option is off this output consists of only one line showing the beginning value of the minus log likelinood the ending value the number of squeezes in the stepsize search ISQZ the final stepsize and a criterion which should go to zero rapidly if the iterations are well behaved This criterion is the norm of the gradient in the metric of the Hessian approximation It will be close to zero at convergence When the print option is on FIML also prints the value of the parameters at the beginning of the iteration and their direction vector These are shown in convenient table so that you can easily spot parameters with which you are having difficulty Finally FIML prints the results of the estimation whether or not it converged these results are printed when the print option PRINT NOPRINT or TERSE but not when SILENT is specified The names of the equations and endogenous variables are printed the value of the log likelihood at the maximum and the corresponding estimate of the covariance of the structural disturbances Following this is
465. ve Dependent Variables and Time Series Each of these categories has their own directory in the Examples directory For a description of each examples look at the file examples txt in the Examples directory Updated examples may be found on the TSP International website at http www tspintl com examples Introduction Composing Names in TSP Every name must begin with a letter 96 or exceptions 2515 and 3SLS commands Subsequent characters name may be letters 4 or digits or may not be used in names that appear in MATRIX commands The maximum number of characters permitted in a name is 64 in versions prior to TSP 4 4 it was 8 Composing Numbers in TSP Composing Numbers in TSP Every number must begin with a or a digit No spaces may appear within a number One decimal point may appear One E or D may appear followed immediately by a one or two digit number sign This is interpreted as a power of 10 to multiply the first Example 1E2 100 With free format LOAD or READ commands a is interpreted as a missing value and a repeat count with a may be specified The largest value of a series in absolute value that may be stored in TSP is 1 E 37 unless OPTIONS DOUBLE is used Values larger than this are set to missing Scalars and matrices are always stored in double precision Examples 3 0 is treated as 000 53 100 is treated as 53 missing 100 Introduction
466. ve for output at one time After the OUT statement in your program TSP marks any of the variables you modify or create so they will be stored on the databank files at the end of the run Variables created before the OUT statement was executed and not modified later will not be stored OUT remains in effect until another OUT statement is encountered To stop writing data to any files include an OUT statement with no arguments to cancel the previous statement this will also cause the variables to be stored on the previous OUT file When time series are stored with an OUT statement the whole series is stored rather than just the observations in the current sample The frequency of the run where you use the series later should be the same as the frequency of the run when the series was stored Since all variables to be saved on databanks are actually saved only upon execution of a new OUT statement or at the end of your TSP run the variables marked by the last OUT statement will not be stored if the run later aborts for any reason Output OUT produces no printed output except a message when a new databank is created Examples OUT FOO creates EXP TLB in the C CONSUME directory on a PC OUT C CONSUME EXP 307 Commands Also see the examples under the KEEP command 308 OUTPUT Interactive OUTPUT Interactive Examples OUTPUT sends all subsequent output to a specified external file rather than to the termi
467. versions of TSP which allowed only integers If the increment is negative obviously the index will be decremented by its absolute value so the start value should be bigger than the end value The index variable is updated every time through the loop so it may be used in computations or as a subscript However the DO loop in TSP is not a very efficient procedure so that you should be wary of doing a substantial amount of variable transformation or computation with large DO loops If you want the accumulated sum of a series use a dynamic GENR for example ACSUM z X SMPL 2 N ACSUM ACSUM 1 X or MSD NOPRINT X The result is SUM 119 Commands or INPROD X C SUM Loops with IFs are best done with logical expressions on the right hand side of a GENR or with a SMPLIF Examples DOIl 2TO7BY1 SET IM1 1 1 SET X I X IM1 ENDDO DOI 22T07 SET IM1 1 1 SET X I 1 ENDDO The first two examples here have the same effect since the default value of the increment is one OLSQYC X1 X2 IF ABS DW 2 gt 5 THEN DO YC X1 X2 FORM EQ1 ENDDO ELSE FORM EQ1 This example runs OLS on an equation checks the Durbin Watson and runs AR1 on the same equation if the Durbin Watson is sufficiently different from two The DO ENDDO statements bracket the set of statements which are to be executed if the Durbin Watson test fails 120 DOC
468. verwritten by the new file If you write a matrix the file will consist only of numbers If you write series their names will be put in the first row The first column will contain dates or observation numbers and each series will be put in a column below its name Series are written under the control of the current sample If there are gaps in the sample observation numbers will be used instead of dates in the first column Dates are written as the last day of each period and formatted as Month Year Output WRITE produces printed output in the output file or screen unless the FILE option is used in which case data are written to an external data file Options FILE filename string specifies the actual name of the file where the data is to be written FORMAT BINARY or DATABANK or EXCEL or FREE or LABELS or LOTUS or RB4 or RB8 or format text string specifies the format in which the data is to be written or printed The default is LABELS unless the FILE option is also specified in which case the default is FREE Each format option is described below FORMAT BINARY specifies that the data is to be written in single precision REAL 4 format on the external file This format for data is by far the most efficient if you do not plan to move the data to another computer and should be used if possible if you have a large amount of data To read such a data file use the READ command with the FORMAT BINARY option This format canno
469. whether the theta f h t term appears in the regression This is labelled THETA in the output MEAN indicates a GARCH M ARCH M or OLS M model and the GT NAR or NMA options are required The constraints on theta can be relaxed if necessary by using the RELAX option See the HEXP option the number of terms where h t depends its past values These coefficients are labelled BETA1 BETA2 etc in the output This indicates a GARCH model NARz the number of terms where h t depends on past squared residuals These coefficients are labelled ALPHA1 ALPHA2 etc in the output This indicates a pure ARCH model if NMA 0 RELAX NORELAX relaxes the constraints on theta and phi ZERO NOZERO specifies the method of imposing constraints on the parameters If the default ZERO option is used a parameter is set equal to the bound if the trial value violates the bound Note that ZERO does not apply to parameters with strict inequality constraints such as h t With NOZERO the stepsize is squeezed when the bound is violated until the constraint is met NOZERO appears to be much slower than ZERO Nonlinear options These options control the iteration methods and printing They are explained in the NONLINEAR section of this manual Some of the common options are MAXIT MAXSQZ PRINT NOPRINT SILENT NOSILENT The default Hessian choices are HITER N and HCOV W Other choices like B F and D are also legal but Fiorentini et al 199
470. will not be saved on exit from the procedure LOCAL lt list of variables gt Usage The LOCAL statement should be placed following the PROC statement to which it applies Follow the word LOCAL with the names of variables which will be created during the execution of the PROC but which will not be needed on exit This can save storage space or allow the use of duplicate names which can be especially convenient if you wish to build a library of PROCs and don t want variable name conflicts Output LOCAL produces no output Example The procedure below computes moving averages of variable length LEN The local variables LAST which is the index of the last observation but one and LAG the loop index are not saved on return from PROC MA PROC MA X LEN XMA LOCAL LAST LAG XMA X SET LAST 1 LEN DO LAG LAST TO 1 XMA XMA X LAG ENDDO XMA XMA LEN ENDPROC MA 234 LOGIT LOGIT Output Options Examples References LOGIT is used to estimate a conditional and or multinomial logit model The explanatory variables in the model may vary across alternatives choices for each observation or they may be characteristics of the observation or both There is no limit on the number of alternatives LOGIT CASE lt series name COND NCHOICE lt number gt NREC lt series name gt SUFFIX lt list of names gt nonlinear options lt dependent variable gt lt independent variables gt or LOGIT CASE lt series name
471. wrong number of arguments or an unrecognized command option HELP or HELP COMMANDS or HELP lt command name gt or HELP FUNCTION or HELP NONLINEAR or HELP ALL or HELP GROUP or HELP lt group number gt Usage HELP by itself lists the various ways of using the HELP command which are given below HELP COMMANDS lists all the TSP commands ten per line HELP command name gives details on a particular command HELP FUNCTION lists functions and operators both general and matrix HELP NONLINEAR lists the options for nonlinear estimation procedures HELP ALL gives a one line description of each TSP command from Ato Z HELP GROUP gives the same description but sorted by functional groups HELP group number gives the same description for a single group The group numbers can be obtained using the HELP command with no arguments Examples 181 Commands HELP AR1 summarizes the arguments and options of the AR1 command HELP 1 gives a one line description of each command in the linear estimation group 182 HIST HIST See also graphics version Output Options Examples HIST produces histograms bar charts or frequency distributions of series It is convenient for obtaining a rough picture of the univariate distribution of your data HIST BOT DISCRETE MAX lt maximum for x axis gt lt for x axis gt NBINS lt number of bins gt PERCENT PRINT WIDTHz width of bin
472. xecution of the EDIT command is not suppressed during collect mode The documentation for EDIT in this manual describes the editor its commands and provides examples In Givewin Mac and DOS Windows TSP it is easier to re execute a single command using the cursor keys 389 Commands REVIEW Interactive REVIEW displays a line or range of lines entered previously in the terminal session or read from an external file REVIEW firstline lt lastline gt Usage REVIEW provides a means of going back and re examining earlier portions of your terminal session If the second argument is omitted indicating end of range a single line will be displayed If no arguments are present the entire terminal session will be listed The uses are numerous e Several commands use line numbers as arguments EXEC EDIT DELETE RETRY you will probably need this procedure from time to time to locate commands you want to use again in some way e f you get results that puzzle you you may want to review the sequence of commands that produced them e When using INPUT files and directing the output to disk you may want a reminder of what has just been executed before proceeding e When recovering a session that was terminated abnormally you will want to REVIEW the command stream before executing selected portions of it 390 SAMA SAMA Output Options Examples Reference SAMA performs seasonal adjustment of time se
473. y HITER F is superior to HITER D The HCOV F option is valid only if HITER F yes DFP Davidon Fletcher None Powell Analytic or numeric first derivatives and rank 1 update approximation of the Hessian from iterations This option is valid only if HITER D For upward compatibility it implies a 281 Commands default of GRADIENT C4 no Eicker White A ARCH combination of analytic variance second derivatives and estimate BHHH see the White Reference no Robust Robust to None heteroskedasticity This is equivalent to W and used in LSQ only no Panel grouped estimate None allows for free correlation within panel PROBIT FEI REI and PANEL only no Panel grouped estimate None except robust to PANEL with heteroskedasticity across ROBUST units allows for free option correlation within panel PROBIT FEI REI and PANEL only yes Discrete Hessian numeric None second derivatives based on analytic first derivatives yes Numeric second BJEST derivatives MLPROC variance estimates no Print all three standard None error estimates MAXIT maximum number of iterations The default is 20 MAXSQZ maximum number of squeezes in the stepsize search The default depends on STEP STEP option MAXSQZ default BARD 10 BARDB 10 CEA 10 CEAB 10 GOLDEN 20 282 Nonlinear Options Note that some routines BJEST LOGIT reserve MAXSQZ 123 for special options NHERMITE nu
474. y difference is that UPDATE is expecting to modify an existing series You will be prompted to specify the observations you wish to update your response must conform to the rules for specifying a SMPL A single number will be interpreted as a request to update a single observation Prompting and storage will be defined by the SMPL you specify and the most recent FREQ prior to the UPDATE command If the frequency is not appropriate for the series you will encounter errors In response to the prompt for data you may enter as many items per line as you like the prompts will adjust accordingly Prompting will cease when the observations you specified have been updated and the series will be re stored If more than one series is being updated you will be prompted to update them sequentially 444 USER Mainframe USER Mainframe Reference USER allows a user written TSP subroutine to be linked to the program The list of arguments you have written is passed directly to the subroutine USER USER lt list of arguments gt Usage To use the USER feature you will have to write your own subroutine in Fortran and link it to the TSP program To do this you should consult the TSP Programmer s Guide and your local TSP consultant Attempting to write a TSP subroutine requires understanding how the program works internally and is not recommended for novices or those who are unfamiliar with Fortran A default USER subroutine com
475. you only need to specify the ones you want to change CONBOUND specifies the probability for the confidence bounds about the forecasts The default is 0 95 CONSTANT NOCONST specifies whether there is a constant term in the model EXP NOEXP is used when BJEST was used to model log y and you want to forecast the original y The forecast is increased by using the forecast variance for each observation to make it unbiased and the forecast confidence interval will be asymmetric See the Nelson reference for more details NAR the number of ordinary autoregressive parameters in the model The default is zero NBACK the number of back forecasted residuals to be calculated The default is 100 NDIFF the degree of ordinary differencing required to obtain stationarity Note that the undifferenced series is forecasted The default is zero NHORIZz the number of periods ahead to forecast the series The default is 20 NLAG the number of periods prior to the forecasting origin to include in the plot of the series and the forecasts The default is 20 the number of ordinary moving average parameters in the model The default is zero NSARz the number of seasonal autoregressive parameters in the model The default is zero 65 Commands NSDIFFz the degree of seasonal differencing required to obtain stationarity Note that the undifferenced series is forecasted The default is zero 5 the number of seasonal moving
476. ys be computed at any point by use of GENR see GENR for the form of the statement When it is computed the parameters and constants are supplied with their current values before computation and the identity is computed for all the values of the series in the current sample Output IDENT produces no output A single equation is stored in data storage Example 188 IDENT Here is the identity that completes the five equation illustrative model in the User s Guide IDENT GNPID GNP CONS I G This identity states that GNP Gross National Product is always equal to the sum of CONS consumption investment and government expenditures 189 Commands IF Examples IF provides conditional execution of commands in TSP It requires a subsequent THEN statement and may also be followed by an ELSE statement if you wish to have a branch for a false result IF lt scalar expression gt Usage The result of the scalar expression following IF is interpreted as true if it is larger than zero and false otherwise If the result of the expression is a logical value itself for example TEST 2 0 it will be given the value one for true and zero for false by TSP The expression must be formulated using TSP s Basic Rules for formula construction Only if the expression gives a true result will the statement following the next THEN statement be executed If that statement is a DO statement all the statements up to the

TSP 5.0 Reference Manual - National Bureau of Economic Research

Contents

Download Pdf Manuals

Related Search

Related Contents