Home

GTAP in GAMS with GUI for class room use, PDF

1. Scenario 4 And now you can see the counterfactual and the simulation side by side in the tables Im Terini E AG vewe z Ie USER SCEMARTOS TFP 2 MT 1318 E 1315 Du 1 225 ER 1 220 5551E 1 ER rx T 000 x E i 9484 7 58 T 0 002 WT 16167 688 6620 293 TUNE 5354 00 iaaa 54n7 23 5504 54 amis 2360651 7 Rekord 1590607 155060 Copy to Chipboard bad Export Data Pivot Cuvlermuze Table bald You might now want to add relative differences click with the mouse in the table or use the Customize button in order to customize your view see screenshot below Choose Values and percentage difference as the Comparison output select Scenarios as the Data dimensions used for comparisons and make sure that the NOSHOCK scenario is used as the comparison element Tama di ano Fracton dgis and decmal separator 2 9 oe Separator between merged data dm ndions m Cohmnvedh V Aer eth na dH pan z ILL Hee empryrows 7 Hide empty cumma Fracton dgis and decmal separator 2 FEES Cut off mit te determine eenty celis an Separator between merged data dimensons ow E z ues defaut ot for tables C Show Fertogram nee casuficamon colons for tables Column width 77 5 Raw width 12 5 Shoe oniy selected tems Lor beats aniy M Hie mpy reves Hide empty cooler Cui off mit ba determine emply celis BS V Use defuit phrasing for tables M chow histogram 7 Ute
2. Select Load convert output and GAMS file into viewer from the file menu Rum Lead only GOX files Into viewer Load GAMS files and H existing GOX files Into viewer Use ref file to load GAMS file and if exiting GDM filed inte viewer lead Genrer Gulp into viewer Load conser couper and GAMS lile inda viewerr TTT TRI Derry arentiATE mock Use small forz for nan selected vor tens ceci s Peck Sie dialog to iak GOX dimemeoom to sets List of tables loaded from GDX fels TABRITZAGTAPEINGAMS RESULTS ARUNIHPTX GDX xy AAS TA it AI eel os lool ool ee eS GIAP project GARS file COM TSUDRITTGIAPIUIGAMS GAMS POSTMOC purpose Top lev file of GTAPinGAMS project py TORITTQTAPSINGAMSAGAMS IPOrs TMOC Bauthor adate Paince or p t poin EE o_o n t ikad pe ture Li mn reser years 1 7 Orgnai SAM Accounts 8 Set of sectors m GT A t Ty EM MT Heim om ml Olm m mimm l Oloo olov wine 6 4 9 4 4 4 4 4 M44 4 qs EB so onmulti uu uc the foll The viewer allows you to see e Linearized views on the equations which you can select via the selection button above The information in green behind each equation provides separated by commata 1 The absolute change in the variable 2 The absolute change multiplied with the Jacobian i e the approximate change in the equation due to the simulated change 64 3
3. nm Money metric decomposition by factor 11 T Regions c d Wold vJ N Y 10x10 Tariffs Y Yv Unskilled labor 0 57 Skilled labor 1 66 Capital 6 82 Land 7 91 Natural resources 2 58 Markets Demand by product Shows a table with demand by institutions products and totals shown in the row The items box allows for looking at quantities prices tax rates value and tax income the origins by domestic yy lt J imported and tota an aggregation over all products 48 ie The first box allows selecting the regions Note total is GGIG GUI gam View Handling Windows Demand b Y v Total International transport demand Total 137954 328 29255 543 165232 531 19 773 Crops 2325 690 679 039 3 520 1135 439 12 553 495 141 2654 458 14 135 Te and Fishery 1328 746 369 028 2 529 809 677 42 006 105 506 1574 340 18 483 o and Capleraiion 4687 890 22 361 0 001 2336 248 0 013 2329 266 5644 262 20 401 Other Minerals and Mining 1323 381 36 883 0 638 839 756 21 985 424 119 1587 975 19 994965 E 6239 726 3103 143 4 847 1874 002 8 685 1249 049 7399 113 18 581 Textiles amd Agpurel 3567 711 1105 375 4 233 1124 771 13 450 1319 881 4343 022 21 731 mee 17505 932 2301 049 59 600 9099 064 263 720 5782 497 21084 553 20 442 Heavy Manufacturing 30399 992 2108 420 49 331 12138 285 3935 351 12168 604 35848 355 17 922 Utilities 2871 889 773 766 3 358 1930 692 14 312 149 761 3467 724 20 747 Services 6
4. edur ap censes Paky Socks Test Pods user ecenaroe selglcbal SCENDES eame ecen bese scenarios noshock gas lambdaf fx r f m User type runner The window shows now the file as stored in the disk and included by the standard GTAP in GAMS model driver when you use it to run a scenario 29 Note The scenario editor can also be used to edit parameters settings for the model Several files which show the parameters used in different modules are provided Running the scenario General settings Model structure Output options Agorithm GAMS General settings Input file gdx 10x10 X Scenario description Post fix for result files Scenario file Dynamics Comparative static v Regional coverage Global model v In order to run the scenario click on Simulation select first the directory with the user scenarios as shown below and then pick the second box with scenario you just generated C GIG GUT gama Fie Unser CLE Sei Hen yeorhatepas m amine Gener selh os Modd structure and series GAME General settings tasks Defne icem ri Lumine ier EBD a eg yum Dai fie debo and press the Start GAMS button That should run your counterfactual against the reference and generate results in a GDX file The tab General settings gives you the following additional input possibilities e Entering a scenario description Normally the output files gener
5. 5 1 2 0 fom are ane oo Leeeebock ard iet Adate bru acd rrano Pspca ed Pood iem w dm won eh s E an ET Ir era Pe 8 a W oma T m E oom 23a Eos E nes Lae oa I Bor z ye 4 re T p a Le za pa E F c is xi moe e ade XR map p l Ee Beer er 7H are a non eM me cox Loa ve Ea nm omes Bm Loo Doom am nu Ea La r Ee A mS 2E Ld Tha DaF d piigi ee mal De maha Fi Hirra aral Pah mhen En rm maim Promi bon am an La adm LE O E Um 4 cm 3 3 a E 3 n z eum noo M __ i M a 4 gaa er TA mm isa non I me cow Loa Ea n m mm Bm eo no a nau m Lon d AUS mo ira LE Tila Char aru Cir Liwawiruh aul Has Ce bi Hirra are Bah aston En rm cam amd Fossil B EX LOER Aak Zwf Tcr hui nre Sere ot eho came L md nuu eras io OT eemreraet dened n BEI bes streaker oe boeme chaud B8 miS nie ScpaRIOS OT reece ened E igg oiei Sree dae epee deme 0 wes eee SINSRREJOR CRT serene oai dere The visualization of histograms and scatter plots can be improved if the graphics settings are changed such that 1 zeros are treated as missing values and 2 zero is not automatically included in the axis range Options for bar charts and histograms Maximal number of plots Maximal number of bar blocks Maximal number of bars per blocks o Foreground transparency in 10 5 7 3D effect E Stacked 7 Plot vertical E Cylinder onl
6. 5 79 19605 71 320 25 12931 68 110 98 889 01 8106 09 600 19 2134 70 920 29 15 38 1517 95 14 97 47 99 172 65 403 27 928 05 141 86 71 39 604 79 143 71 6742 14 6581 61 160 53 29 24 Total output 1 00 1 00 1 00 1 00 1 00 1 00 Total intermediate demand 0 45 0 40 0 52 0 34 0 42 0 82 Total intermediate input taxes 0 01 0 00 0 00 0 01 0 00 0 02 Total factor demand Total factor taxes 0 04 0 00 0 03 0 06 0 02 AGR 0 02 0 13 0 04 0 00 0 01 0 00 MFR 0 16 0 13 0 30 0 09 0 04 0 22 SER 0 24 0 12 0 16 0 24 0 14 0 60 ENR 0 03 0 02 0 03 0 02 0 22 0 00 LND 0 00 0 15 LAB 0 13 0 24 0 11 0 17 0 05 SKL 0 08 0 01 0 05 0 12 0 03 Capital 0 18 0 14 0 13 0 24 0 23 RES 0 00 0 02 0 00 0 05 There are further tables available to bi lateral trade which can also be visualized as flow maps as shown in the screen shot below and more might be added in the future But the examples above might be sufficient to judge if these views in addition of using a GDX viewer to look out variables are a useful addition to result analysis a ae Items Origins Meta b emw Markets gt Sectors Trade b Exports trade matrix No table Exports by product and destination Imports trade matrix Imports by product and origin Sensitivity analysis GEMPACK supports systematic sensitivity analysis Channing 1996 The standard GTAP in GAMS model also offers similar functionality drawing on experiences with other G
7. Gtap8InGams gams util hsManual2004 pdf application octet stream D Gtap8InGams gams til TaskSync bat D Gtap8InGams gams utilXitle def gms D Gtap8InGams gams util TaskSync Files bat D Gtap8InGams gams util deanMemDef gms D Gtap8InGams gams util deanMem gms D Gtap8InGams gams util Whs exe application octet stream D Gtap8InGams gams util title qms D Gtap8InGams gams util sleep exe application octet stream D Gtap8InGams gams wtil LHSManualSandia pdf application octet stream D Gtap8InGams expRefDir D Gtap8InGams data D Gtap8InGams data GTAP8set gms D Gtap8InGams data 10x 10_example gqdx application octet stream D Gtap8InGams data 10x10_example zip application octet stream D Gtap8InGams data tested D Gtap8InGams data tested 32X34 aqq D Gtap8InGams data tested 57X36 agqq D Gtap8InGams data tested 57X45 aqq D Gtap8InGams data tested 57X82 aqq D Gtap8InGams data tested 57X56 aqq D Gtap8InGams data tested 10X34 aqq D Gtap8InGams data tested 57X68 aqq D Gtap8InGams htmiDoc Starting externals And should end with a Completed At revision message Added D Gtap8InGams gui jars commons beanutils 1 7 0 jar application octet stream Added D Gtap8InGams gui jars jsch agentproxy sshagent 0 0 7 jar application octet stream Added D Gtap8InGams qui jars gmszlib 164 dll application octet stream Completed At revision 268 I Added 223 After you press OK you shoul
8. Te demand Cascade Tie horizontal rim Tide vertical T Jed sanas 487 56 And open the setting dialogue from the menu bar Fide Utiities GUI Settings Help GTAP in CAMS Vli Edit setti Dats bese con Lessee cominus fnar ini File Calibration Ganre current settings lo ini file 9 Simulation Remove task wnecllr settings GTAP in GAMS V6 Remove view specific settings 1 17 First you should enter your name Le Option gt mtd Option User Srita GTAP in GAMS Vie System Settings GAMS and R Other aptons User Type aware Next click on the tab GAMS and R and either type in text field next to Path to GAMS EXE which GAMS EXE you would like to use Alternatively you can use the button to the right of the field to navigate to the directory where GAMS EXE is found via a file selection dialogue Please do not only enter a directory but the full file name as shown below and choose GAMS EXE not the user interface of GAMS GAMSIDE EXE Option User Settings GTAP in GAMS V6 System Settings GAMS and R Other options lOp m Sme d qams24 4 qams exe Path to GAMS exe GAMS scratch Directory rbridge r302 bin rscript exe Path to R exe Path to Troll exe GAMS Options Number of processors used in GAMS Processor speed relative 100 2 4 GH Intel core 2 If you are using regularly a text editor you can register it under Other options You might also wa
9. The contribution of the variable to the change in the LHS in percent using information in the absolute changes in the LHS and the variable and the related Jacobian entries The variable which the code analyzer has detected as the LHS is indicated with a in the example below that is xd For the LHS the relative change is report In example below the change is almost entirely coming from the change in nd 100 the contribution of the other changes is very small xdeq SouthAsia MeatLstk c ProcFood a shock Agents demand for domestic goods T 0 00902671001153436 0 9999999486802 0 9085354810037336 1 0000000513198 ps SouthAsia MeatLstk c sho 1 0 005 7 2E 5 1 0 2 13050842285156 0 00464540767085857 1 00000000249911 pmt SouthAsia MeatLstk c shock 0 0033 4 7E 5 0 8 2 130 50842285156 0 469371530886303 ps SouthAsia MeatLstk c shock 0 005 7 2E 5 1 0 3 130508422 5156 nd SouthaAsia ProcFood a shock 0 67 0 006 100 0 t0 183822553739227 xd SouthAsia MeatLstk c ProcFood a shock 0 034 0 0062 0 6 E 0 e AGDX viewer which allows merging of symbols from one or several files for combined analysis e A quick view on any symbol click on any symbol in the GAMS code shown in blue and it will be loaded in the Symbol from GDX view You can directly compare your counterfactual against the base or any scenarios against each other which you loaded as GDX file i Gami js P
10. 9534 91 54631 16 12416 49 23091 78 591 84 0 4196 0 3096 0 0296 0 0196 0 69 2 03 4 30 ao 2058 32 800 55 1 87 504 73 0 57 413 50 0 43 0 29 12 82 0 58 0 73 0 78 2 91 1 99 0 57 0 20 6 36 Light Manufacturing 17319 36 3950 65 18 43 4854 85 985 41 5116 43 1 02 1 51 1 91 0 74 0 50 1 23 Pone aE 32869 18 2062 65 53 34 15055 72 2527 85 11201 62 0 68 0 66 0 27 0 24 0 77 2 35 e aaa Cau aioe 11765 99 509 02 9 54 3485 60 7415 18 292 50 0 44 0 28 0 57 0 59 0 64 5 2696 Transport and Communication 22034 38 9171 68 67 26 10126 40 593 31 1725 47 591 84 0 2096 0 2196 0 8896 0 0996 0 63 1 05 4 30 A ES 46061 50 14658 88 9384 46 18118 90 893 11 2799 73 0 0196 0 0196 0 0196 0 0196 1 15 0 40 Total 137582 97 31212 10 9534 91 54631 16 12416 49 23091 78 591 84 0 41 0 30 0 02 0 01 0 69 2 03 4 30 PENIS 2058 32 800 55 1 87 504 73 0 57 413 50 0 4396 0 2996 12 8296 0 5896 0 73 0 78 EE 21 36 6 28 0 19 11 79 1 55 1 22 0 03 1 34 1 76 1 52 Wheat 99 08 12 88 26 67 29 76 0 12 0 28 0 05 0 24 0 45 0 76 22 49 1 59 0 75 0 70 0 29 1 14 2 43 fil caadc 133 49 8 28 58 67 33 27 Model structure parameterization and factor mobility You might also select from a so far limited set of different model setups General settings Model structure Closures Output options Agorithm GAMS Model structure Vv CTAP AE Residual region NAmerica GTAP_E Simple_Energy_Nests Labor_nest CO2 E
11. South Asia North A emer d States of America Mexico Rest of North America Latin America Argentina Bolivia Brazil Chile Colombia Ecuador Paraguay Peru Uruguay Venezuela Rest of South America Costa Rica Guatemala Honduras Nicaragua Panama El Salvador Rest of Central America Caribbean European Union 25 Austria Belgium Cyprus Czech Republic Denmark Estonia Finland France Greece Hungary Ireland Italy Latvia Lithuania Luxembourg Malta Netherlands Poland Plu Slovakia Slovenia Spain Sweden United Kingdom Middle East and North Africa Rest of Western Asia Egypt Morocco Tunisia Rest of North Africa Sub Saharan Africa Benin Burkina Faso Cameroon Cote d Ivoire Ghana Guinea Nigeria Senegal Togo Rest of Western Africa Central Africa South Central Africa Ethiopia Kenya Madagascar Malawi Mauritius Mozambique Rwanda Tanzania Uganda Zambia Zimbabwe Rest of Eastern Africa Botswana Namibia South Africa Rest of South African Customs Rest of World Oman Israel Switzerland Norway Rest of EFTA Albania Bulgaria Belarus Croatia Romania Russian Federation Ukraine Rest of Eastern Europe Rest of Europe Kazakhstan Kyrgyztan Rest of Former Soviet Union Armenia Azerbaijan Georgia Bahrain Iran Islamic Republic of Kuwait Qatar Saudi Arabia Turkey United Arab Emirates Rest of the World 43 e 7 0 Y Paddy rice Wheat Cereal grain
12. and regional savings The alternative closures allow to fix private consumption spending and to let the saving rate adjust Fyrhannoe rate e The final closure relates to the exchange rates In the standard GTAP model there are no exchange rates and a world factor price index is used as the anchor in the model there is no money illusion such that only relative price matter That is basically a set up with a global monetary union The alternative closure fixes the regional factor price indices and introduces flexible exchange rates for each region as e g found in the GLOBE model The reader should note that the two closures do not give identical results These closures are only available in the global model Exchange rate Monetary Union Flexible exchange rates e For the single region layout two closures for the current account balances are offered either fixed foreign saving with a flexible exchange rate or the reverse combination Current account balance Fixed foreign savings Government Tax Fixed exchange rate Algorithm General settings Model structure Output options Agorithm GAMS Agorithm 3 Model type and solve o w O Presoves m Solution algorithm CNS v Option file Conopt 1 Substitute out prices v Presolve steps with fixed import prices b Scale transactions Solve mode Fixed foreign savings v Algorithm for NLP CONOPT v Algorithm for CNS CONOPT v Use CNS first v
13. defined scenarios are stored Results of model runs are found under results run The directory expRefDir and HtmlDoc are used to generate a HTML based documentation of the GAMS project 65 Mame svn e data doc e expRefDir gams y gui ey htmlDoc a results GUI overview The GUI is programmed in Java and based on GGIG Gams Graphical Interface Generator which is also used for other projects e g CAPRI www capri model org While binaries are distributed for free the Java code itself is currently not open source It uses third party libraries provided under licenses which allow for further distribution and often as well code changes The main jar to start is jiars ggig jar In order to modify the GUI no Java programming is necessary as the Java code reads a XML file guilGtaplnGams xml which defines tasks work steps and the individual GUI controls Equally a XML file guilGtaplInGamsTables xml defines the different views The link to GDX files is based on the Java API distributed by GAMS com as part of any GAMS installation and uses dynamic link libraries As both Java and GAMS are available also for non Windows operation systems it would be possible to also port the standard GTAP model in GAMS to other software platforms Indeed a CAPRI user had a few years back generated a native MAC version The interaction between the GUI and the GAMS processes is based on includes files which are generated anew for
14. each run and capture the state of the user operated control 66 Graphic Overview on interaction between user GUI and GAMS Based onUser input from GUI GTAPAgg exe Zip file GTAPiInGAMS xml from GTAPAgg Control and task in data folder definition in gui folder loadGTAPAgg gms Build_inc gms GDX file with SAM parameters MEELERC CED Additional modules com_inc gms GGIG Java binaries i jars folde GDX file with results in gui jars folder in results run folder GTAPinGAMSTables xml wiew definitions in gui folder TETTE tuli Hid Hlini HH Processes In order to work with the model GTAPAGG which is clearly not part of the installation must first be used to produce a zip file which comprises the necessary GTAP data parameters etc at the user chosen aggregation with regard to sectors regions and factors The GTAPAGG output should be stored in the data directory The zip file comprises data in proprietary GEMPACK formats and cannot be used directly with GAMS code 67 Next the LoadGTapAgg gms programs reads that output from GTAPAGG and converts it into GAMS symbols which are stored in a GDX container It uses the ConvertGtapAgg jar to read the content of the agg file with the aggregation information used by GTAPAgg to generate definitions with long texts which are temporarily outputted into agg gdx On demand the filter gms program processes the inpu
15. example where normally information on a sector or institution would be stored fourth dimension the activityDim the label tots is used It is chosen in the filter below based on a regex expression lt activitysSel gt tots REGEX lt activitysel gt table lt theme gt Model oueruieuc theme lt name gt Model properties lt name gt amp regionSel Wor REGEX lt regionsel gt amp regionText hideiz regionText amp productsel x REBRES amp productsel lt productText gt hide lt productText gt lt activitySel gt tots REGEX lt activitysel gt lt activityText gt hide lt activityText gt lt dimS el gt Meta dimSSel gt lt dimSText gt hide lt dimSText gt lt hideEmptyRows gt YES lt hideEmptyRouws gt lt dim Text gt hide lt dim Text gt lt defuiew gt Table lt defuiew gt lt defpivot gt GP es lt defpivot gt lt itemDim gt dim lt itemDim gt lt item gt lt itemName gt Dummy lt itemName gt lt key gt dom lt e key gt lt item gt lt table gt Another example provides the table for intermediate demand Here all products and sectors are selected instead of a regex expression the pre defined lists found in generated xml are used 71 lt table gt lt name gt Intermediate demand lt name gt lt Cheme gt Markets lt theme gt lt regionsel gt all lt regionsel gt lt product el gt products lt product el gt lt productText gt Inputs lt productText gt lt a
16. factor price index which is fixed during a global model solve to reflect that there is no money solution i e the behavioral equations are of degree zero in prices However during single model solves fob prices are fixed such that it is impossible to fix at the same time the factor price index In order to still drive the single models during iterations towards a global factor price index of unity the fob prices are divided before each solve by the currently calculated global factor price index as are the saving price indx and the factor prices of all regions which enter fixed the single model solves Finally we reflect the global bank mechanism by calculating iterations the average expected return at global level based on weighting with regional net investments A heuristic estimate a change in foreign savings which should drive the country s expected rate towards the global average There are a number of options available for the pre solves 1 The number of pre solves For larger shocks and a higher number of regions ten iterations are recommended as the pre solves are relatively cheap 2 Use of grid solves That solves the single country models in parallel with the exemption of the last pre solve but exploits less information as results from other countries can only be exploited once all single countries are solved The grid sole is the recommended option for modern multi core machines CNS models based with CONOPT have proven to bene
17. generic approach is implemented for factor supply nested CET functions can be used to supply primary factors to the production sectors The top level nest is labeled with xft The following example used in the GTAP AGR implementation shows that approach x build nesting structure for factor supply to agr non agr X fHest agr VES fHest a f agr agr fm WES FHest n F xFt agr fFm VES onegafHest r agr fm omegaf r fm fNest nonfigr YES fHest a fF nonfigr nonfigr fm YES fHest n F xft nonfigr fm YES onegafHest r nonfgr f m omegafi r fm Similar to the CES production nests the post model processing reports the structure parameterization and simulation results in the exploitation tools Finally these flexible nestings are also applicable for final demand government investments and households as shown here again with the nesting used by GTAP E dNest energy dNest i fd f energy sele c fdn YES dNest_n_fd ener OX non electric fdn YES dNest_n_fd top energy fdn YES sigmaFDNest r energy fdn 1 00 dNest non electric YES dNest 1 _fd non electric Coa c fdn YES dNest n fd non electric non coal fdn YES sigmaFDNest r non electric fdn 0 50 dNest non coal YES dNest_i_fd non coal gas Cc fdn YES dNest_i_fd non coal RE a fdn YES dNest i fTd non coal
18. not present and these equations empty That flexible and generic nesting approach is based on sets and cross sets in GAMS which define the lists of factors intermediates and sub nests comprised in a CES composite nest along with the top nest it belongs too These nesting definitions enter the above mentioned equations in the core model and matching code dealing with parameter calibration Hence the user does not need to introduce additional equations in the code to use the feature it is sufficient to provide the structure of the nesting used via set definitions and the related substitution elasticities The code also tests for potential errors such as duplicate assignments or sub nests not linked into another nests The following examples should be sufficient to show the application of that feature and demonstrate its flexibility 1 An example of a sub nest under the top VA nest which aggregates the two labor categories found in the GTAP 8 data base into an aggregate Add technology nest to model give it a name Pd The mother of the nest is the top VA nest tNest Labor em o m tNest n a VA Labor a tNest f a Labor skLab a tNest f a Labor unskLab a sigmaNest r Labor a Q XX Mmmm unu uu Linkthe factors into the bundle Define the substitution elasticity Note The code will automatically remove the factors linked into nests from the top VA nest 2 The second example shows how to introduce a CES composi
19. the GEMPACK version operates The following short section discusses the two other options Simple deletion With simple deletion transactions loaded from the HAR file are removed from the data base if they in absolute terms are below the threshold entered on the interface under Absolute tolerance With a value of 1 E 10 that deletion step is skipped Afterwards the transactions are formatted into a 20 SAM structure The resulting SAM is then cleansed with the chosen absolution tolerance No attempt is made with simple deletion to maintain the resulting SAM balanced That option is mostly maintained as a fall back in case the more refined rebalancing step normally recommended and discussed next should not work Rebalancing The Rebalancing option uses more advances tactics to select transactions to delete and perhaps more importantly rebalances the resulting SAMs As with simple deletion first transactions loaded from the HAR file which in absolute term are below the chosen absolute threshold are removed from the data base With a value of 1 E 10 that preliminary deletion step is skipped It is generally not recommended to use absolute deletion thresholds above 1 E 6 in combination with rebalancing as the subsequent relative thresholds will anyhow apply more refined rules Please note that rebalancing was only tested with CONOPT and that alternative NLP solver such as PATHNLP might not work satisfactory The consecutively app
20. the number of draws and the LHS uses stratified random sampling for space filling design The LHS is chosen as it can be easily applied also in case of non symmetric truncated distributions In the default case the LHS simply returns more or less uniform distributed draws across the factor range i e between the relative lower and upper limit chosen by the user with an average presenting the mean between the lower and upper limit If the user has chosen non symmetric bounds around zero the simulations would hence lead to a systematic deviation from the mean parameter used without sensitivity analysis The entropy estimator can therefore be used to determine probability weights for the draws which ensure that the weighted average over the draws for each relative factor change is close to unity That approach is hence quite similar to Gaussian Quadrature which also applies weights to the observations The difference here is that a uniform distribution over the chosen range is the preferred distribution while the Gaussian Qudratures used in GEMPACK tries to mimic normal or triangular distributions The user has the choice between three difference distributions all truncated at the chosen lower and upper limit 1 Uniform 2 Truncated Normal distribution 3 Triangular distribution For the truncated Normal the standard deviation must be set for the triangular distribution the peak The combination of user set truncation points and distributi
21. 07 591 89 752 71 119 47 Te lS 1410 77 674 03 665 66 71 07 Meat Products 0 05 0 08 0 04 0 12 1411 47 674 56 665 91 70 99 Mining and a 4261 06 51 20 2871 75 1338 11 j 0 09 0 14 0 10 0 07 4264 97 51 27 2874 67 1339 03 pem 4244 36 2477 29 1248 83 518 24 Food 0 02 0 03 0 02 0 01 4245 42 2478 05 1249 07 518 30 ned q L 1744 34 683 58 642 55 418 20 Clothing 0 08 0 04 0 15 0 03 1745 66 683 84 643 49 418 33 Light 12733 19 2332 11 29 84 7643 46 2727 78 Manufacturing 0 01 0 01 0 04 0 01 0 01 n 12734 61 2332 41 29 86 7644 27 2728 07 Heavy a 25936 77 2292 69 17 20 17176 41 6450 47 Manufacturing 0 03 0 02 0 04 0 02 0 03 25943 26 2293 26 17 21 17180 67 6452 12 PSS qe 10878 77 834 30 11 21 10030 96 2 31 Construction 0 03 0 04 0 12 0 02 0 38 10881 54 834 60 11 22 10033 40 2 32 Transport and meand 18524 80 7878 17 100 56 9187 13 838 46 520 48 Communicatio 0 02 0 02 0 04 0 02 0 01 0 04 n 18528 80 7879 68 100 60 9189 25 838 58 520 67 um 37266 07 12777 97 9350 83 14117 64 1019 62 0 01 0 01 0 02 0 01 0 02 37270 21 12779 25 9352 39 14118 79 1019 78 qe aes 12834 53 12834 53 the 0 02 0 02 t 12837 40 12837 40 Note In the example above relative differences to the case with standard parameters mean where added If the draws are selected an additional dimension is introduced in most tables gm RE
22. 0742 51 15718 52 102812 74 Processed Food 9131 17 65037 52 25261 48 12142 52 80825 78 22016 18 235287 66 15308 34 18758 60 61590 79 Textiles and Clothing 7145 06 79334 26 2253 65 1114933 125823 12 1988 73 21864617 1386 54 17829 08 73258 90 Light Manufacturing 50377 86 226457 47 79897 42 39464 47 645333 81 97419 51 115136538 38315 93 59522 32 333534 00 Heavy Manufacturing 96718 02 1261921 50 387924 38 154015 48 1176961 00 240079 50 237266200 101033 31 119092 05 698601 50 Utiities and Construction 1444 59 19565 64 6879 54 1591 53 9805 61 4257 45 TILSI 3427 62 6088 09 36541 52 Transport and Communication 19523 31 17382156 5078 73 18943 08 137347 89 34221 13 46040150 11411 07 108b AH 100875 77 Other Services 22045 32 165100 83 64577 36 40070 06 252669 31 46522 71 692118 50 21065 73 33369 14 150210 90 Unskilled labor Skilled labor 000000000000 iad 5 MT I ER ar Ne zd msc nem 5 np VA Labor CAP ENE energy ree non coal Paddy rice 1 00 Wheat ae 0 50 ols and fats 1 00 Similar views are available for the nesting used for factor supply and final demand Regions Rens Final demand Years 4 b fdceania Hess Private consumptien shock top energy non electric non coal F Paddy ree on CET Factor supply nests n vent zi Regins Items ee ee ee ets 4 b muss Vegetables fruit nuts on F S seeds O O o o ammm zz life Sugar cane supar best on
23. 3 32 Bio Total utility 1 00 Index Utility of private household 1 00 Index Utility of government 1 00 Index Utility of saving 1 00 Index GDP 49780 15 Bio Consumer price 0 99 Index Government price 1 00 Index Factor income 39510 97 Bio Current account balance 0 00 Bio Tax income 9547 97 Bio Population size 6620 29 Mil heads The three tables Welfare Money metric decomposition Welfare Money metric decomposition by product and Welfare Money metric decomposition by factor support an welfare analysis By products reports welfare gains or losses related to output price changes by factor linked to changes in factor prices or quantities Regions Yi 10x10 Tariffs Tote Money metric decomposition by product 1 Y ai Total 33 32 4 gt Regions Sectors and Bio i wold v Money metr Output price index 155 10 55 10x10 Tariffs Bio Y T ice index id bx Bios Grains and Crops 0 47 I 7 66 Livestock and Meat Products 2 07 Government price index d Bio Mining and Extraction 20 13 Savings price index 35 65 Processed Food Bio Textiles and Clothing 11 16 Income total 121 63 Light Manufacturing 19 70 Bio Heavy Manufacturing 26 95 Factor income 2 58 Utilities and Construction 8 42 Bio Transport and Communication 5 85 Indirect tax income 147 84 other Services 7 92 Bio
24. 75 6623 22 42 5 demand pases 1528 65 178 228 70 1298 10 0 06 1529 02 17 demand e TEE 9910 88 161 00 370644 5571 60 471 84 10417 64 181 1 demand 1990 64 1990 64 1990 64 _ demand po 888 88 31 49 626 56 219 18 11 64 1350 66 35 9 Markets Final demand nests shows the demand aggregates quantities prices and values according to the active demand nesting B Private Governme Investme Em consumpti nt nt top 576 57 184 14 271 59 energy 18 43 0 01 11 03 omer 13 66 0 00 0 00 ed 13 65 0 00 0 00 Demand per product per capita in is an example for a tabled definition which uses the in built expression evaluator it divides total demand by the population size D ma E p od a per capita fo A E v BASE SCENARIOS NOSHOCK y Total Government demand Intermediate demand Investment demand Y Total 19660 35 4602 31 1424 48 9700 63 1938 36 1924 02 Grains and Crops 217 49 89 83 111 79 15 87 Livestock and Meat 212 19 101 69 100 34 10 15 Products Mining and Extraction 638 36 7 88 444 66 185 82 SRS aa 635 65 372 83 191 11 71 71 Textiles and Clothing 244 44 103 20 89 15 52 09 Light Manufacturing 1901 86 352 23 4 66 1157 64 387 32 Heavy Manufacturing 3883 43 345 09 2 69 2614 66 920 99 Utilities and 1635 37 125 83 1 69 1507 48 0 37 Construction Transport and 2765 58 1183 57 15 02 1371 13 125 32 Communication Other Services 5587 62 1920 16 1400 42 2112 67 154 38 That must be the 193
25. 7703 383 20710 266 9446 101 24099 453 8215 841 5231 719 541 999 81628 727 20 568 A predefined view Demand by product bar charts for scenario comparison as seen below shows how the combination of a certain pivot view type while using the comparisons can be used to produce bar charts showing relative differences It might be useful to the filter out the comparison scenario from the view such that the empty upper graph is not shown Demand by oroduct bar harts for scenario com iarison 01 m Lo uw Items Origins Percentage diff to UU yi u vers ED NBS asc y v NHOSHOCK a Government demand Internal abe diem Sectors aid nedtuti ong BE ro B sss rer xm NER F os E ra E SR Beer x DER F cons Note you use the second filter bottom next to view type to view type double click on it to open a dialogue which lets you remove the benchmark scenario 49 m te 1 d Selection dialog for Table columns E Enter exarch pattern in feld and use buttons or use moans ba define selecbons Chem selecton add patiern to labels Add partem balabala That will produce a nicer graph Markets Demand by institution same content as before but different pivot could also be quickly produced by using the pivot possibility which can be applied to any view Total 21263 64 242 58 5951 03 12460 10 619 30 1990 64 21931 60 261 5 nouschold 6925 93 48 30 1389 33 5352 55 135
26. 8 36 1938 36 investment good That view can alternatively be visualized as a map in Demand per product per capita in map Scenariy BASE ZCEMARICHS _MOSHOCE E ra 2 SO lt a B Peas ed x Fees oe mulus Maps can be made from any table by putting the regions in the rows and choosing Map under View type Markets Intermediate demand break of the intermediate demand by sector Note the total in the columns is an aggregation over all sectors 51 ENR 471 84 129 12 2976 38 450176 312 88 1990 64 10417 64 18 24 125 81 15 86 0 03 1 05 181 18 49 22 1818 59 1029 74 12 45 796 45 3989 68 58 84 92214 3300 12 97 36 1193 13 573979 2 82 109 84 156 04 203 14 0 00 506 98 Sectors Sector overview reports for each sector total output output taxes total intermediate and 42 10 32 24 2 46 674 81 381 57 694 84 2414 43 1830 26 2779 58 Orii 35 02 16 09 130 72 18 17 19 82 report primary factor use price etc by sector 3416 47 total factor demand as well the individual intermediate and factor demands Total output Output taxes Total intermediate demand Total intermediate input taxes Total factor demand Total factor taxes AGR MFR JEIEIEIE Capital Sectors Sector cost shares a selection of the value item of the view above normalized by total output 52 65315 98 1050 87 35040 87 491 28 2813 62 1329 38 14741 90 16812 19 2157 39
27. AMS based models In order to keep the technical solution relatively easy the following approach is chosen 53 1 The GAMS program which is used to run a single shock is also used for single sensitivity experiments in order to use as much as possible existing code and to avoid duplicating code fragments 2 A second GAMS program in combination with utility from SANDIA labs defines the experiments and executes in parallel GAMS children which solve the same scenario using each its own parameter settings 3 That second GAMS mother program collects the results from the children and combines them into one GDX file Currently the sensitivity analysis allows changing substitution and transformation elasticities by defining relative ranges around the given parameters plus generates on demand additional draws for two shocks which needs to be translated by changes in the GAMS code into something useful These relative ranges need not be symmetric The draws can be defined generally in two modes 1 The same relative change is applied simultaneously for all sector factors shocks To give an example if the drawn relative change for the factor transformation is 0 5 all transformation elasticities are reduced by 50 2 Individual draws for each sector and factor are generated Whereas Arndt 1996 uses Gaussian Quadratures the approach in here is based on a combination of Latin Hypercube Sampling LHS and an entropy estimator The user chooses
28. DEL MRTMCEP GMS html 15 10 2013 11 27 E E e dipn E GAMS MODEL MTRMCP IMC aMS tmd 6 100 000 11 22 l Mis i GAMS_MODEL_REPORT GMS hemi 25 102013 11 22 B on E T z E GAMS MODEL SCEN USER SCENARIOS 251020121 7 a oy cc E GroupSelecthimnl manum T Parameters diwri 7 B Models html nomena ga T dxmd 7 7 EF Paracseterihted nami gp Files p d i i r E pel rmi sijnmsng W C Equations p de rr E piel 25 10 2013 11 22 C3 Variables p doti r ngi html 251020131122 L Smulaton Beye amp rts html 251020031122 M HI Medals O E reat chtmi 83152013112 EOJ Functions ft f r Specific factor transformati n shire 2302002 11 22 S10 SourceFiles EEG a r META html PAO 5 1147 im Trea vids O You can e g open all Variables used in the task simulation clicking with the mouse opens a symbol page GTAP in Top Declarations Definieons Assignments References Elements G AMS V 6 Variable ddfm technical Name ddim documentation E Automatically generated from Number of dimensions 3 TbritAGTAPInGamsumodehsSimm Usedinproject s Simulation at 25 10 2013 11 35 18 See eee Tog Declaration Detinibons Assignments References Elements File Projects GAMS MODEL MRIMCP GMS Simulation Edit ddfm is assigned in Tog D claratong Dedeeon Asabgnmants Eue enincss Elements open all ckxea all File Projects 3 vue GAMS MODEL MRIMCP GMS Simulation Parameter
29. ET nios de gt T RES Rel SUAE riim n aati dias Bw frala iaai ona NEUE Ha dns pr liim Baer T 4 L rodi Len Frsain icm PITE Fidri tai Piia W VETE te Fiia mar O TL TD iei TIT i we ee aha Fm TT ee VT E p aim miii ii II rcm b p ni P LT er T d MNT El p i TT ET imma jahr Fi demam FR o Td Jd prr Tl rt a reum ER Snes Leta bes mv s mz S un osa poem iium IET a mir iat Ra LI vix nem s E umma eum diet aa Tm imas MENT E menm msnm anna m M peta maam aii min BETLRN O e ean mT pa as PTT sida DIT i is pua meld araar mTT CEEP T pe Ano iiri T iia ici in r r pu LS miras rey amai twazan m nee Hed dk ra Dias LIE k faama aii n LIE qr dd VEL tel ak ici i That allows a user to analyze each draw if wished Alternatively the statistics built in the interface can be used In order to do this open with a right click in the table the pop up menu and select statistics 57 Select the statistics you want to be shown such as mean and stdDev and select Only show outliers to only show outliers and statistics View Handling Windows 276 37 2201 73 183 95 Nan 2 07 Nan 1011 97 9587 35 1547 67 706 44 9601 15 1457 58 830 71 8670 79 1540 18 600 08 8442 18 1356 41 21 77 351 98 104 45 2 74 44 34 13 16 159 50 1268 54 111 94 106 74 1139 17 100 42 52 11 2326 15 851 49 0 04 Nan Nan Further work is necessary to add views which aut
30. GTAPinGAMS with a GUI Wolfgang Britz Institute for Food Resource Economics October 2015 matches rev 313 of https svn1 agp uni bonn de svn Gtap8lnGams Inhalt Sic a Oh PN ORT 3 Why an implementation In GAIVISI cessi eerdesextus aisi rta Satur eias ea odd Uc e okP Mta oU a adt Pags x Eod a a ta dr CEEUEE 3 Scalability Modularity and ExteNsiONns ccccccsseccccessccccessececeesececceeceecesececeesecesseeceeseneecessusecetseneees 3 HEIDE HES IAE NERRTRNRRRORRT 5 EE EEEE EEE NE E PEE E ENES 8 WORRIE WEE ETE E E E E een eee 10 Getting the model code and setting up the GUI sssssssssessssresssrrresrrrsesrrressrerssreressreresrrressreresereeeseeee 10 SV CSCO UE asserere bEsR qune UnbE INT eR NU RRN EET HN URS INIT IER E T HIN DENIM NRI URMTE 10 SRA NU Gies 13 ab 26 T TT m 15 Making the interface working with GAMS cccccsseccccssececeeseceeeesecceceeecceesesececsunecesseaeceeeeusecetsuneees 17 Datapase Beljerdtiolisu sie vete ic bendi vet qenid isset mI Mmi vem Im id 20 Sanje sge lt 11 1 ERES 20 FEIN NS RENTE 21 Special treatment for specific regions and sectors ssesseesesreeessreesrrresrreresrrressreresrereserereserressne 22 Model size and solution behavior eeeeeeeiissseseseeeeeeeeeeee enne nnne nennen nennen nnn nnne nnns 23 Inspecting the resulting data base cecccccssseccccesecc
31. GUI offers two possibilities to ease working with the code 1 Generation of a set of static HTML pages 2 An equation parameter and variable viewer linked to a specific model instance which also allows looking at the GAMS code HTML documentation A HTML documentation of the code can be generated via Utilities Build HTML documentation L GTAP in GAMS VB Ll Hie Utilities GUI Settings Heip aT GOK Viewer ral seeing GAMI CSW te GDX MP in CAMS VE E Start Command prompt Start equation and variable viewer I cT Hald HIMI documentation Deline cena x Samalation The documentation will include any changes you introduced to the project s GAMS code First choose the directory where the files should be stored best use the HtmlDoc directory as shown below beware the utility will generate many files do not use a directory where other files are already stored The input files are found in the folder where GAMS was run in the case of standard GTAP model in GAMS typically in the model directory 61 ZEE ecto F Query ifi status You should be able to find a file named simulation ref Make sure that Query SVN Status is switched off and next press the Generate HTML documentation button Eg debe agpi puberes E Ei garmi Vasgppublseft L x pab Vagal NE ED run Vaggheuvec Ri Egi appt Magpipubisot S E Ls aa on P CAMS decent alion genera
32. Global scaling factor 1 E 3 l Use grid solve for presteps v Maximal seconds for one model solve 1 000 Threshold for transactions 1 E 10 Supress output from presolve v Number of repeated solves 2 0 Zero iterations will only test check case The model can be solved either as a constrained system of equations formally solved with a dummy objective to yield a NLP which might help CONOPT or as a MCP The latter option is interesting if tax rates are to be endogenized e g under emissions ceilings trade or production quotas features not comprised in the standard model Please note that the default maximal time for model solution of 600 seconds might be too low for very large models The settings shown above are recommended defaults Generally solving the model as a constrained system of equations CNS especially in connection with pres solves see below has proven to be generally faster for larger models 150 000 transactions in SAM The CNS solution will automatically check if the model is square MCP gives additionally flexibility such as introducing Tariff Rates or production quota Equally the MCP version will automatically check for a unpaired equations and variables The standard option file is 1 which uses lower number of minor iteration shown to speed up solution with larger models The option file 2 with relaxed convergence tolerance is automatically chosen if the model cannot be solved in a first atte
33. P applications is a project specific pre model aggregation Given the total size of the data base there are already about 1 Million non zero trade flows reported pre model aggregation cannot be avoided it yields smaller models which are faster to solve easier to debug and typically also numerically more stable with less small values being present However aggregation comes clearly at a price not only might detail of interest for the application be lost but also peaks of 26 policy instruments etc wiped out That can have important effects Ko and Britz 2013 show in example applications that one might increase simulated welfare from FTAs by simply increasing sectoral and regional detail GTAPinGAMS tries to soften the decisions on what regional and sectoral aggregation level to use based on three features Firstly filtering and rebalancing of a data base can help to solve models with rather high number of sectors and regions but clearly the filtering process will also introduce aggregation bias Secondly the equations of the model had been carefully arranged and scales and bounds for variables introduced to stabilize the solution behavior of the solvers and to speed up solution time And thirdly post model processing allows for a second aggregation step for reporting purposes To do so the user has to provide a second aggregation definition file agg can be produced by GTAPagg See also Britz and VDM 2015 for discussion of solution t
34. S statements 72 lt item lt key gt pc lt key gt lt itemName gt Consumer price lt itemName gt lt item gt lt item lt key gt esubd lt key gt lt itemName gt Subs domestic lt itemName gt lt fitem lt item lt key gt pic lt key gt lt itemName gt Price imported lt itemName gt lt item gt lt item gt lt key gt pics lt key gt lt itemName gt Import share lt itemName gt lt item p results r P i if tot p results r P int i tot p results r P i cf tot cf l i r p results r P i py tot p results r P out i dom p results r P i pc tot p results r P i hou tot References Britz W 2010a GAMS Graphical Interface Generator User Guide Institute for Food and Resource Economics University Bonn In http www ilr uni bonn de agpo staff britz GGIG user Guide pdf Britz W 2010b GAMS Graphical Interface Generator Programming guide Institute for Food and Resource Economics University Bonn In http www ilr uni bonn de agpo staff britz GGIG programming guide pdf Britz W 2014 A New Graphical User Interface Generator for Economic Models and its Comparison to Existing Approaches German Journal for Agricultural Eonomics 63 4 forthcoming Britz W Dom nguez P and Gopalakrishnan B N 2015 Analyzing results from agricultural large scale Economic Simulation Model Recent Progress and the Way Ahead German Journal fo
35. UI available for the GEMPACK version see also Britz 2014 and Britz et al 2015 or e g the GAMSIDE often used with economic models realized in GAMS As a consequence some functionalities found in runGTAP are not comprised the GUI linked to the GAMS version of the GTAP model and vice versa The development of the GUI for the GAMS version and GAMS code for post model processing feeding into the exploitation part of the GUI went partly in parallel with the development of the model codes to ensure seamless inter operability However it is important to note that the GAMS code can be used completely independent from the GUI The GUI provides two major functionalities firstly it allows running the model while also liberating a user to carry out changes that are otherwise necessary in the GAMS code such as to select a specific data base or shock file or to switch modules on and off Instead controls on the GUI allow steering the model run thereby avoiding code edits Secondly the GUI offers tools to view and analyze model results Specifically a post model reporting part maps back the variables into a structure similar to a SAM and additionally stores quantities prices values tax rates and tax incomes That structure can then be viewed in a flexible exploitation tool The post model code also provides some eventually useful aggregations e g to total intermediate input total factor input total output total demand and world totals and provides a w
36. We also allow for a version where exchange rates are flexible Use of the options extensions and modules discussed above does not require additional coding efforts they can be directly switched activated from the interface or by using non default parameter files The set up of the code should render it relatively straightforward to implement in a modular fashion additional module Table 1 Modules and extensions Data filter Tested not used with GEMPACK Optionally removes small transactions based model from SAMS trade matrices while maintaining closely important totals Thought to support model applications with highly dis aggregated data bases GTAP Standard Tested for exact replication With extensions from ENVISAGE such as non diagonal make matrix CET on export side Completely flexible Tested based on set definitions Should allow to quickly generate variants nesting of of the standard GTAP model currently production available which differ in nesting of factors functions intermediate Completely flexible Tested based on set definitions Should allow to quickly generate variants nesting for factor of the standard GTAP model currently supply available which use nested CET structure to describe factor supply Completely flexible Tested based on set definitions Should allow to quickly generate variants The code makes a differentiation between extensions which change the nesting of the production function or introduc
37. Y Peatbaedfers OOOO O on Paddy rice vit Crops nex in Wheat 2 M Catthe sheep qoats horses i v Cereal grains nec T Animal products nec us Mese abes fruit nuts on Raw mik ii W seeda 8 ae Wool silk vorm cocoons 1 Sugar cane ugar beet m Forestry xp Cropsnec P Coal on Cathie sheep goats horses v ol s on Animal products nex a Gas 0 0 08008 m Rarw milk m Minerals nec Wool sik worm cocoons n Meat cattle sheep goals horse Forestry Meat products nec Fishing ae Meqetabhe oi and fats Coal E be Dairy products ms on e Prcc soon rico on was on Saar on Miersnec 0000 xe Food products nec e Meat cattbessheep gnats horse us Beverages and tobacco products on Meat products nec n Textiles a Meqetahde oils and fate yr Miearimy apparel a Model Overview Flow schemer uses the in built mapping utility in a somewhat unusual way to draw a flow chart with all transactions in the economy some aggregated The lowest rectangle shows the total production value the related cost are split into intermediate imported and domestic demand on the LHS rectangles above and total primary factor costs on the RHS rectangle above Total primary factor income remuneration for each factor is split up into tax arrows on right and household income arrows to the left The rectangle Government shows the government account incoming arrows show tax income outgoing ones gover
38. and the cutoff used in that preliminary iteration can help to find a good compromise between model size and a too aggressive filtering where also more important transaction are wiped Out 2 Symbol free GON Arley et prag hu ad rt bal ih ard ierit mri nne Tow TL T0172 08 TIER BATI ara ato INEN aT EO TET HT EXT MTTER HIN MIROR 42e TEAT fm Xr 4 eh ere Tii Euri T 3145408 T5208 E ETE ATO b 1584 50 J4T3 B0 Jk DU T2430 aire PHL 38 T De AAT UE Yer 41 11 amd Haram HAE ANI ISIERAMR Zi Pere Turre V pat oo ar a name aia pn ie MISSE p MIT d TI2BOP Terum gg Tar Sd P 18070 00 18070 00 19070 08 16887 08 15245 08 TWO 1134208 TITZLO0 TENA POO Mii M15 E1500 BOG BHAD BHAD ADTI STAT vitm IES eo JAHH HM mo BARS OD 5124 00 AIH JEA AJAFTA IFL Ao JRER D Miro Tah DG MAT OD Lr ot B2 sn TS ML ML 3550401 142208 ID meras 34 8 ES rene Pee Baw ET2E ETE EH EHDI EHSA Mann os BAB vpm 1085 00 1080 00 Loe 122 4 fiis LT E ndo GIE 0146 bi T PT ioe ur ETIAM HTS ATA npm T7085 Dd Tasa Dd Tus pd Li rs ramos nma En E5720 Err Bap P 5D DO 55500 pi SSH IMP AHF 08 vigm 1088 00 1082 00 1082 00 MAL 260 08 r2 TA 20 amp 00 150 19140 13500 Ho TEW 476 00 176 00 176 00 AID 132 dgm TUB D SEL TOEN OF 719 04 LL Lm 1h 141 00 17234800 1711 00 TE TALIXI Tio TTLDO 1110F moy mne miL curfimitol im ios ak ga aar GAE mar gm LET on os 025 ET urba Tas nux Dx nos om om aon inm aw 00 bid Dx nox nux Pre or post model aggregation The standard case in GTA
39. ated are named of the scenario file used That name can be overwritten with content of that text field e Post fix for result files The post fix is appended to the names of the output files It can e g used to differentiate versions using the same shock e g results generated based on different closures or by using different modules e Choice between the Comparative static or Recursive dynamic version of the model Dynamics Comparative static Recursive dynamic Regional co Note that a simple example scenario which shows how to introduce shocks dynamically can be found under scen base_cenarios regDyn gms If the Recursive dynamic option is switched on a drop down box is added which allows to define the number of years Dynamics Recursive dynamic Y Humber of years 3 e Choice between the Global Model or Single Region version of the model 30 Single region Regional coverage If the single region mode is switched on a drop down box appears where the region can be chosen Regional coverage Single region v Region in model Oceania y Compile GAMS StartGAMS Hide Unihide controls Exploit resu In single region mode fob price of other regions are fixed and imports to other regions react to changes in the region s cif prices according to the share equations of the lower Armington nest at fixed total import quantities and aggregate import prices Output optio
40. bit version of Java should be used You can test that again in a DOS prompt by typing java version The response should look similar as shown below note the string 64 Bit Unfortunately many 64 bit Windows machines use a 32 bit browser and if you install Java from such as browser the 32 bit version is downloaded The link JRE points both to a 32 and 64 bit version make sure you select the 64 bit version on a computer with a 64 bit OS Installing a 64 bit version on a machine where the 32 bit version is already installed does no harm and makes the 64 bit version the default java version 1 8 68 25 JavatCIh SE Runtime Environment amp build 1 8 6_25 hi8 gt Java HotS5pot amp TM 64 Bit Server UM Chuild 25 25 hH2 mixed mode 14 e lf that still does not help open a DOS prompt navigate to the GUI folder and type GtaplnGams bat possible errors will be shown in the window In case you cannot find help in your team with regard to these error messages contact the author General information on how to work with the GUI can be found in the GGIG user guide www ilr uni bonn de agpo staff britz GGIG user Guide pdf That pfd document can also found in the folder GUMjars in your installation Please note that the hints given below in the document on using the interface are only thought as a first introduction refer to the GGIG user guide for any further details or use the inbuilt help system by pressing F1 We would finall
41. ble in the sense that it can work with differently sized but identically structured data bases The GAMS code comprises a pre solve algorithm which should speed up solution of larger models with more than 300 000 non empty transactions in the global SAM The code was intensively tested for numerical stability and acceptable solution time when applied to more dis aggregated versions of the GTAP data base As mentioned above the GTAP in GAMS code includes and supports extensions beyond the GTAP standard model Using these extensions is partly driven by the underlying SAM structure such as allowing for a multi product make matrix to support non diagonal relations between production activities and commodities consumed Other extensions are not driven by the structure of the data base but switched one depending on the parameterization of the system such as allowing for non infinite CET between domestic sales and exports and between export flows Similarly non diagonal make matrices can be applied in conjunction with a CET nest to steer output composition of multi output sectors and a CES nest to determine consumption shares in case several sectors produce the same commodity Many of these extensions are inherited from ENVISAGE however not all ENVISAGE features are found in the code The code also support already a range of different closures e g for foreign savings private and government consumption Furthermore the code structure has a modular des
42. ccesecccsusececeuecceeeesccecsuseceseunecessensceeseueeesseneses 25 Prespr post model aggregation saisons le teras mci d bu enaccesaainbese bU cM E dS e ples E Li Dr LE LADUEE 26 CENO dU a a o tomm 27 Defining th scenario esscase cedente ero Ete thao docta En re sos n eR obe S ERsa ovo dra Uk rep deso ere sa E bcsi s UM ec tup la tesi U Rd sinis 27 RUNNEL ESC SVAN 10 usce ietoidbdtenid EG victui mas uteri uoPPEd UN abis iu M SEEN E E 30 UT OD UIS ET 31 Model structure parameterization and factor mobility eeeccccsssccceeseceeeeseceeeeeeceeseeeeeeeeneess 32 Modules and parameters sccccsesicecsancnadssateeveceannccdessaususuasnwesianaiesstsndewsnanaaudaueieesedabatuseeioareenteenecnenoss 32 KUMOA OS acca sc cc m RN 33 Ger e 33 AISOPIEBI o deiEia vit Et ES EM UDR HEP TL EI IMMUNE E MM UM 35 epu m H 37 IISDECUING ENE HOSUICS so todo acea a e pte b t eet angen A sa au tata s dates obo etie hin 37 Notes for GEMPACK users or users familiar with other CGE modlels eere 39 Relative difference for all variables as in runGTAP lessen nnne nennen nnns 40 Troubleshooting WITH the VIE WF ur lteiect ne hu eben cua Ho ete eva ect vute Reb Coa eU turus Cauda atu Het baie 42 Overview on existing views for exploitation ccccccessccccesececceecceeeenececseueceeeeeeceesene
43. ceteueecesteneeeetan 43 Sensit IVEY cel S Ses tas ees catt ees cca EAE AT euet Ce D EN CLOMID REL POS ARUM esac 53 Controls Tor SENSITIVITY experiments o acetate de e e od abeo m e ue ici Peer b Cb unda t N Feb teuntens 55 miaideojo CSUN ALON NE RR TR PTT PCT 55 Drawing systematic experiments for other elements of the simulation eee 55 Parallel solves and analyzing the results nennen 56 Getting an overview on the COG Cs ioci vio t RUE TOR Sora best tado bois ie ta SETep E Hes ios EDU RL leo Reve des Use iE 61 GAT IVIL COG UIC EYE AION areca ess Recent atest Sha id utut E date ipta Dima ios odia he Cc ISn cO a Piden 61 Equation and variable viewer similar AnalyZ GE cccccccccssssecccceeeseeccceeeeeecesseeeeeeeseesueneeeeeeeas 63 TVECHAIC AI OCUINNGNUALION t TE EE 65 Overview on GIFCCLORY Structure socie c ibo Rte i DD pac ER eS Vota vnu ub Nase Ec Nbui eM DU adi UE vedt 65 C18 KeT aUuri een 66 PROCESS CS asians nota rM a IM NM M Cd I a E dE 67 POSt IMOG El PLOCES SINS e T T TEE 68 Exploitation and fHexibledggregatiOFia serius or oorr E UY A aoc viue De ev M Ra P xu FUE UE 70 References senno 73 Background Why an implementation in GAMS GTAP is certainly the far most widespread international data base for CGE modeling The data base was so far only distributed with a default CGE model template in GEMPACK called the standard GTAP model as well as with many useful GEMPACK based ut
44. ctivitysel gt sectors lt activitysel gt lt activityText gt sectors lt activityText gt lt dimS el gt all lt dimS el gt lt dim Text gt Origin lt dim Text gt lt dim Sel gt origins lt dim sSel gt lt defuiew gt Table lt defuiew gt lt defpivot gt OPsA lt defpivot gt lt itemDim gt dim lt itemDim gt amp item lt itenmName gt Total lt itemName gt lt key gt tot lt key gt amp item lt item gt lt itemName gt I mported lt itemName gt lt key gt I mp lt key gt lt item gt lt item gt lt itemName gt Domestic lt itemName gt lt key gt Dome fkey gt lt item gt lt table gt Details on how views can be defined can be found in GGIG programming guide Britz 2010b As part of the exploitation tool the composition of the producer price per unit costs and the consumer prices is visualized as shown below y HY medi That type of schemer is constructed based on registering hyperlinks to graphical symbols ona powerpoint slide and storing the presentation as html A java programs reads the coordinates of the mouse regions which linked to the hyperlinks and stores them technically as if they would be coordinates for a geographic map The hyperlinks registered in powerpoint must hence match the keys in a data dimension put in the row dimension of a tabular view the following screenshots show a part of the view definition underlying the scheme visualize above and related GAM
45. d find the new folder on your computer r GtapSInGams 02 10 2015 11 34 Dateiardner The green mark indicates that there are no local modifications to files found in the folder i e that the files on your disk are identical to those just downloaded from the repository If you introduce changes in these files a different symbol will be shown For further information on how to benefit from TortoiseSVN for your daily work with GTAP8InGAMS for other projects please refer to the web We would like to mention here only that you have always two versions of each file on your computer 1 the so called local working copy these are the files in your normal directory with which you can work as usual and 2 a second version in a hidden data base which reflects the latest download from the repository That allows the system to e g find out which files changed and to highlight changes line by line It is important to note that you will never lose edits on your local computing when you perform updates i e download newer versions from the central repository 12 Starting the GUI The GUI requires a Java 8 run time environment JRE feeware the JRE should be found on the path Please make sure that you install the 64 bit version of the JRE on 64 bit operating systems After downloading the GAMS sources and other files via SVN or having unzipped the downloaded folder navigate to the GUI folder j GAMSDOC 6 jars I gtaplogo gif Bx britzjpg gta
46. dassi amp cabon colons for tables ask etre cors geine maures mere setts aad meuran 38 After closing the dialogue you find now relative difference in small numbers beyond the simulated results in each cell of the table View Hinding Wandews i Inceme 0 K e b R egons Total or household groups Viral Total f nosno Index Savings prie Tt 5 501 index a Factor income Be eal OF Curent account balance 0 000 ris e Tax income 16033 4530 16167 088 Iris i arl t5 0 at bb 20 293 Populakkn sine AMi heads 5 ooo Note as expected total utility increases but prices drop There are many other functionalities found in the exploitation tools the GGIG user manual discusses these options http www ilr uni bonn de agpo staff britz GGIG user Guide pdf If you want to leave the viewer again use Exit from View Handling menu gr GTAP in GAMS V8 G New Data View New Data View using settings af last one Close all Data View Windows Exit Cascade Tile horizontal Tile vertical Notes for GEMPACK users or users familiar with other CGE models e The view Model overview SAM shows a SAM generated from the model results including row and column sums and potential imbalances if the option Store SAM in GDX is switched on e Many tables comprise additional information on aggregates on top of the other rows Only in some case
47. del set up Foreign savings Global equal returns ta capita Global equal returns to capital Fixed global allocation of investments Governm Fixed foreign savings 33 e The standard closure for the government is to calculate tax income at given tax rates and to let the top level utility function allocate a certain share of regional income to government consumption The reader should note that there is no immediate relation between changes in tax income and government consumption Changes in government consumption relative to private consumption and regional saving depend on the one hand on changes of the related price indices and on the other on the elasticity of the private consumption share to changes in overall utility As such there is no closed government account one could argue that implicitly government saving adjust The alternative closures endogenize tax shifters either on all or selected products in consumption or for all or selected factors while keeping the real tax income for government consumption fixed using the government price index to define real tax income These two closures come closer to a representation of separate government account Government Consumption taxes Factor taxes Final consump For the closures modifying taxes can be further detailed by applying the shifters e The standard closure for private consumptions and saving is to a have flexible budget shares for private and government consumption
48. e Define scena Remove task specific settings Remove view specific settings Simulation i 3 Ifthat does also not help try to close and re open the application 4 If the viewer still shows curious things it is most probably some programming error contact the author 42 Overview on existing views for exploitation Model overview Model properties reports some basic attributes of the model instance as seen below such as the modules switched one closures used or the size of the model s M adel properth s 0 eb EAB able Y 10x10 TARIITS Y e eiie M Foreign savings vupialasturnTaotnv CO2 Emissions 5 on ol sectors 10 00 of factors 500 eof regions 10 00 m ol SAM entries HEU Model type 5 cs Make matri diagonal E Global scaler 1000 00 aol equations 17380 sul variables 12380 00 sIterations 15 00 s seconds solution time 0 53 Time total 343 There are also two tables which report the aggregation of the data base under Model overview regional aggregation and Model overview sectoral aggregation Australia New Zealand Australia New Zealand Rest of Oceania East Asia China Hong Kong Japan Korea Mongolia Taiwan Rest of East Asia Southeast Asia Cambodia Indonesia Lao People s Democratic Republ Malaysia Philippines Singapore Thailand Viet Nam Rest of Southeast Asia South Asia Bangladesh India Nepal Pakistan Sri Lanka Rest of
49. e GTAP E model the composite of skilled and unskilled labor is already shown above top level nest aggregate of capital and energy 3 tNest CAP EHE VES tHest n a UR CAP EHE a YES tHest f a CAP EHE capital a YES tNest_ _h_a CAP ENE energy a VES sigmaNest r CAP ENE a 25 tHest energy VES tHest i a energy ele c a VES tMest n a energy non electric a YES signaHest r energy a 1 88 tHest non electric VES tHest i a non electric coa c a VES tHest n a non electric non coal a VES signaHest r non electric a p 58 tHest non coal VES tNest i a non coal gas c a YES tHest i a non coal oil c a YES tHest i a non coal p c c a YES signaMest r non coal a 1 88 The nesting used in a specific model run along with the substitution elasticities are reported in a table in the exploitation tools Equally the code reports the quantity and price aggregators for nests and the resulting values and aggregates them up over individual sectors and over regions It is important to note that the post model reporting redefines the top level VA and ND nests such that they match the usual definition i e an aggregation over primary factors and intermediates respectively Information about these nests in the definition used in the model can be retrieved with the separate tables showing all technology nests A similar
50. e chosen parameters to all sectors and factors Otherwise each sector factor will receive its own random shock in each experiment The panel on the right hand side allows choosing the type of distribution for each parameter and the related settings The Peak defines the mode for the triangular case Entropy estimator In the case of non symmetric truncation points and use of triangular or normal distributions as in the last example above the mean of the draws will deviate from unity On demand an entropy estimator will try to find probability weights for the draws which recover a mean of unity If no solution is found the estimator tries bounds around each mean of 1 2 596 and 5 If that fails the program will abort with an error Without the entropy estimator each draw receives an equal weight and the mean relative parameter draws might deviate from unity Drawing systematic experiments for other elements of the simulation Senshi analysis Draws E Mumie of exzxerimerits 10d Fs Bah Subtituson domestejmpnrtso ursfnrm fuga bruton between importers Truncated normal TAN vubrbubon between fackora Irancabesdi norm Oriy generale draws ol Gemegal Trantiomaton between latiora Truancabedi rie mad Sect Irunzated meer a Brod Truncated normal Minoru parallel CAMS proceres Set iere af shock Set length af shock 25 Lise uniform shocks monies sectors actors JJ Use entropy e
51. e nested CET structures for factor supply These extensions only require set definitions and substitution respectively transformation elasticities but no changes in the equation structure of the model or is post processing In opposite to that the implementation of a what is called a module consists typically of two files one file which declares the additional symbols parameters variables equations found in model gms and a file which calibrates the related parameters for the benchmark found in cal gms Potentially there are also additional statements in the post model reporting part 4 of sub nests under not available for the trade of the standard GTAP model currently final demand margins available which use CES subnests under the top level final demand equation ina ERN replications land cover differentiated by land cover GTAP AGR Operational not tested for exact Applicable also to regional dis replications aggregation different from original GEMPACK implementation uses the flexible nesting approach GTAP E Operational not tested for exact Based on flexible nesting approach replications Aggregate Operational not part of standard Domestic and import shares for Armington GTAP model or variants thereof intermediate demand and related tax aggregator for available from GTAP rates are not longer sector specific intermediate removes a large share of equations demand Post model Could be done in GEMPACK Generates SAM lik
52. e structure calculates reporting world totals regional and sectoral totals based on additional GTAP agg file welfare decomposition etc feeds into GUI exploitation tools Single region mode Operational not available in Fixes import prices and let export demand standard GTAP model react to lower Armington nest at export destinations Recursive dynamic Operational not tested for mode compliance with GDyn Flexible nesting While most CGE models apply the CES functional form to depict the production function quite some differences exist how input composites are defined Usually each nest is represented in the model s programming code by its own quantity and price aggregator equation Adding or changing nests thus requires coding efforts new variables equations and parameters need to be defined and properly assigned Typically also the equations relating to the top level value added aggregator and or the intermediate demand nest need to be adjusted if the nesting is changed The GAMS code underlying the GTAP in GAMS project applies a different strategy Here nested CES structures in the production function are represented by a generic approach where a small number of equations and matching variables handle basically all possible nesting structures The equations for the top level VA and ND nests are already set up to host sub nests along with equations describing sub nests In the standard GTAP implementation such sub nests are
53. eerie eiit cet Fh aum mam nae ea en HOANG see emma Pic steh E ac mnc g m i A 4 z qoem FE Me Lan a am m bot i e P it J K 4 d F ars ped L ar cum 43100 E T Az T xam ELETE x2 70n ETO SS Gan GREI ROO honor Erare Core ris Vnabisc orere sei one a ue If the view type from existing tables is changed to Scatter plots the result might look as below the example shows different types of prices in each plot and different sectors in the rows columns 59 MAP or er RABE Sas F euim ao Lo aca aom LI Ais AT 2 xz e uns 3 3 am S ose Z sse 2 ase mo ntm ean 5 ESI em as ans z ana Aii de v uw d zw E a ar ux a i narpa vae z oe rea wd ES ase s eT 1 2 Bux La ame It i 5 mcm ERE n a is ES ES ATE EE manm 2 ABS 5r oS Total Cea oa Cree ee Hars neo 2 la ey LU Ferd Loa E Lae a 120 Loe E mo am i u c Li E LE i asa am as F m n n mI n w em eM A non coms nox r 3 c0 r ERR d r ze pe ce Sm a am ncm aA L2 Loa zm n m Ll unm Loo noce Sam non mx Lon ree Bee eA d Lj Ed Tirta are areal gii Lhvmilinh aul Hamat ea Ei Eheu are Fats ao tions Co mm iameld Pac i n Am Lea 2 m z j ro i 3 L i 52 nn 3 mod 1 EE i os 1 ez ie nc Ld n AMA cd ap AS w BES elm d 5 e S ATs in E a n 2 PEL aE
54. elfare decomposition The user can also provide based on an aggregation definition file generated with GTAPAgg a second more aggregated sectoral and or regional aggregation which is used for post model aggregation of these results The exploitation tools available with the GUI provide interesting features they allows to view simultaneously results from different shocks without running additional programs support maps and a wider range of graphics as well as basic statistics and thus might hence also be of interest to seasoned modelers Furthermore the GUI comprises some additional utilities a tool similar to EXAMINER in GTAP which shows the linearized version of the model s equations and allows inspecting parameters and variables found in these equations Similar to AnalyzeGE that tool also allows decomposing changes of the LHS of each equation to changes of the variables on the RHS Another utility allows documenting the code in HTML pages Finally instead to using the GUI in interactive model a batch mode allows to perform model runs The batch reports these run in a HTML page and allows parallel execution of different shocks or of the same shock in different model variants The paper provides a hand on guide on how to use the model in conjunction with the GUI The latter is based on a GUI generator for GAMS projects http www ilr uni bonn de agpo staff britz ggig e htm Britz 2010b which is also used for some larger projects s
55. etre Li ymbil Iram GE zi T rita T AP ram eode pr rco gens theta f ai l l 7 Dim ricshock ada GET gx m 4 Vern type i Profit function Lor Liras ees E E Tube 4 T i N TT EE AAN OAE f fa E b SAI RSA RSS LUR RUNI ER v TW z z n 7 t t rT Fick H 11 F pp parameter threat f 4 F Factor share of value added mp We am eR nt T thetad i j r Demestic share of intermediate input e 2 em ax arid m thetat i 1 r Tapert share nf intermediate input en anm ern soma ps am E x 9 28 0 55 0 35 LA e 7B theca ARF rj Value added share of sectoral output Mo 08 0 29 0 25 DFF DH amp T Ar J u a O41 es tH 2 13 nlins f TFf Ee 1 j iic ma a fe oe pan 0 85 UP LE LE qa thetaE 3 zr Emm EE wrEr EE 4 2 eee ED E 3 701 gt vin Tibi ED unit via iszcin j LI etad i j rh5 vdfmi r l rztEdO i Ent r fa T vdfmii3 z 1 rtfdO i j PF tatai Ej 1 rtEdO mii rj i rj Technical documentation Overview on directory structure The screenshot below shows the directory structure of a standard GTAP model in GAMS installation The information is only provided for those who have an interest into technical detail The GUI with the Java libraries is found in GUI The results directory comprises the output from model runs The data directory comprises also the GAMS readable versions of available data bases All GAMS files can be found under gams in the subdirectory scen scenario templates and user
56. fdn YES sigmaFDNest r non coal fdn ml 00 GUI Working with a larger existing model such as GTAP can provoke serious entry barriers for newcomers partly due to the need to learn specific model mnemonics underlying the code That knowledge is necessary to analyze results if only model output directly produced by the modeling language is available e g listings or proprietary binary result files Equally the sheer amount of symbols found in larger models parameters variables equations labels for regions sectors factors etc and the dimensionality of results can overwhelm beginners That might shy them away from a fascinating research field even if excellent model documentation is available such as in the case of GTAP Memorizing model mnemonics and partly structure is rather independent from the modeling language used beginners will always have to digest a larger amount of information to make the first steps with a larger economic simulation model It might therefore be useful to reduce entry barriers such that students might become enthusiastic about the model s capabilities and then take the more tedious steps to familiarize themselves with the model s code Therefore the standard GTAP model in GAMS comes with a Graphical User Interface GUI which complements the GAMS code The reader should however note that the GUI 8 linked to the GTAP in GAMS version follows a somewhat different approach compared to runGTAP the G
57. fit much more from the pre solves than the MCP version Test have shown that the combination of presolves and CONOPTA allow to solve 57x82 models without any filtering in about 15 minutes even under larger shocks 36 GAMS General settings Model structure Output options Agorithm GAMS GTAP in GAMS V8 GAMS Options for listing Options for solve outputs Print gams code to listing offListing v Solprint Off v Symbollist offSymList Limrow 07 Limcol or Symbol list with cross references offSymxRef v Reslim 600 As all symbols variables equations are also stored in a GDX container with GUI output switched on one might normally switch solprint to Silent to keep the listing at minimal size Inspecting the results Before we turn to the exploitations tool of the GUI the reader should be reminded that GTAPinGAMS allows for a wider range of possibilities to look at results 1 As with any GAMS model solprint on will produce a listing of the model solution showing variables and equations see the GAMS documentation for details That option can be switched on from the interface as discussed above 2 If additionally LIMROW and LIMCOL are non zero the individual equations will be shown with the Jacobian entries The numbers will limit the output of individual instances shown and can be set by the GUI For more information on how to interpret that output please look at the GAMS manual 3 If the output option GDX is swi
58. g can influence substantially results obtained Unfortunately the same holds for changing the regional and sectoral resolution of the model We can therefore generally only recommend running highly dis aggregated models with limited filtering For further detail on solution times and aggregation bias due to sectoral and regional aggregation or the use of filtering refer to Britz and VDM 2015 ref who provide a larger sensitivity analysis with the standard GTAP model in GAMS The long solution times occurred under the direct tax cuts As the standard GTAP model assumes that FDI is taxed at the same rate as domestic one the direct tax cut impacts fully the expected returns to capital by foreign investors which drive the distribution of foreign savings under the global bank mechanism A 5096 reduction as simulated in the test shock can provoke very large differences in expected returns depending on the regional resolution of the data base Thus a new equilibrium can require massive changes in investments to decrease or increase the regional capital stock in order to drive marginal returns to capital up and down towards the global average For some regions that scenario might not feasible without fixing some variables at their bounds Solving the model as a MCP should yield automatically that solution However solving as a CNS Will yield infeasibilities in that case Therefore an algorithm is comprised in the GAMS code which tries to find based on equa
59. here aim at providing a good starting point for a solve of the full global model 35 Therefore their structure needs to reflect the bi lateral trade relations and other linkages between regions in the GTAP model The layout of the single country models therefore differs somewhat from the usual single country CE structure and solely uses equations and variables already comprised in the model at unchanged parameterization In the standard GTAP model there is no CET on the supply side We therefore simply assume that a change in demand for bilateral exports of a single country has a negligible impact on the supply price of each importer Accordingly we drive the single country models with fixed fob prices on the import side On the export side however we exploit the Armington structure we feed changes in the country s supply prices in the Armington share equations of the importers at fixed total import and prices The reader should note that solving the model also with fixed prices for exports would not help much in providing a good starting point for the model Without a CET the export price is by definition equal to the supply price in the domestic market If one would hence fix the export prices the domestic prices in the current solve would also be fixed There would hence be no updated price information be passed along from solving one single country model to the next Further links in the full models between regions are based on the global
60. i le s TRT PLUG T APETRAL HS LERS TSUN SF TARDE Se s 111027 Biter excort taxes Haina export taxes Y fa as ir he TiFmctur income net of deprecation MRC bade ac fainnr cf prodoctn xUiretal factcr uui IE ENEMIES om dal I julia FILE gest ae In order to show percentage differences against the benchmark open the option dialogue by pressing A Select e g Values and percentage differences under comparison output chose the data dimension which end with check and the item check E Only monospaced 44 plain sal Fraction digits and decimal separator 2 Separator between merged data dimensions Fill up merged Dims Column width 224 2 Row width 81 5 E Hide empty rows T Hide empty columns Cut off limit to determine empty cells o 7 Use default pivoting for tables 7 Show histogram T Use classification colors for tables Show only selected tems v Longtexsony v Comparison output Values and percentage difference Comparison threshold to hide values o Dim1 base check Dim2 Levl Marginal Element used for comparisons Dim1 base check check The window with the result will now also comprises the information on the relative change as shown below 41 Symbol from GDX ola factY Dim2 Levl Marginal Percentage diff to 2 me i iE View t
61. ign which facilitates introduction of new model features to the GAMS code So far implementations of GTAP AGR GTAP E GTAP AEZ and CO2 emissions accounting and taxation are available as additional modules as well as the production nesting of the GTAP E model It should however be noted that these extensions are partly not set up as a prefect replica of the GEMPACK implementation and not tested for replication of results obtained with the corresponding GEMPACK code An important feature of the model s structure is a set of equations which allow the introduction of CES nests comprised of factors intermediates demand and sub nests such that even complex nestings in the production function can be introduced via cross set definitions without additional programming work in the model s code A similar generic implementation is provided for factor supply based on nested CET functions and for CES subnests under final demand These features are discussed below in more detail The basic application mode of the model according to the GTAP standard model is that of a comparative static global model of a barter economy i e with fixed exchange rates Alternatively the model can also be run in recursive dynamic fashion again drawing on ENVISAGE A formal comparison to the features of dynamic GTAP model Gdyn is pending Equally a single regional model can be directly derived from the equation system and the single region to run selected form the GUI
62. ilities GEMPACK as a special package for CGEs has certainly distinct advantages compared to less specialized languages Some users might however prefer GAMS when working with GTAP Indeed outside the CGE community GAMS seems much more common than GEMPACK as an Algebraic Modelling Language e g in Agricultural Economics Britz and Kallrath 2012 In 2015 the Center for Global Trade Analysis GTAP decided to release in parallel to the standard GTAP model in GEMPACK a version coded in GAMS van der Mensbrugghe 2015 Intensive testing shows that both versions produce the same results CGE models realized in GAMS sharing basic elements of the standard GTAP model while adding extensions in several directions are widely available e g GTAPInGAMS by Tom Rutherford GLOBE by Scott McDonald and Karen Thierfelder or ENVISAGE by Dominique van der Mensbrugghe However these models do not provide an exact replication of the standard GTAP Model The newly released replica of the standard GTAP model in GAMS is not only interesting for research but also provides together with the GTAP data base an excellent starting point either for teaching or own projects of students especially in the context of course programs already using GAMS It offers additional some extensions compared to the standard GTAP model as discussed next inspired by features of existing CGEs Scalability Modularity and Extensions As the standard GTAP model the GAMS version is highly scala
63. imes and aggregation bias Scenario runs Defining the scenario The interface helps with the definition of scenarios it has a set of example files with shocks which can be combined and edited directly in the GUI Alternatively you can use any text editor and define a shock in GAMS and store the file somewhere under gams scen In order to use the in built scenario editor click on Define scenario and the interface should look like below Defne bass scenario fie ILR institute Ta Foo ani Arete Economes mime Gtapinsamsini Usermame undefned Usertype runner Currently there are two one base scenarios 27 e No shock that is simply an empty file e RecDyn atest implementation for a recursive dynamic baseline e In order to add the content of pre defined shock files to your base scenarios click on a tree element e g Endowment and productivity shocks You should be able to see a list of files as below Scenario categories fH base scenarios Endowment and productivity shocks L 4 Factor specific TP L 4 amp Factor supply changes Policy Shocks Test Shocks user scenarios Double click for example on Factor specific TP and the content of that file is loaded in the editor on the right as seen below ontext Eridowrser ard productety via LI Foty supply cenges t Polcy Shocks Test Socke GAMS file purpose lambdaf fx r f a t 1 1 lambdaf li
64. ixed lists of regional code and matching coordinate set which link each regional id to a list of polygons The flexible regional aggregation in GTAP required a more flexible approach The XML definition of a view can register for a region a list of components under the tag lt disagg gt 70 lt region gt lt 1itemName gt Australia New Zealand lt itemName gt lt key gt Oceania lt key gt amp sel all amp sel disagg fRUS F JI PVF KIR HCL HZL HNP PHG SLB TON VUT WSH lt disagg gt lt region gt The individual code listed under lt disagg gt indicate the regional Ids used in the coordinate set In the case of GTAP these are codes for individual countries The actual mapping between the aggregates used in the current model instance and the GTAP regions is read from an agg file the mapping between the GTAP regions code in the data base and the individual country ids is defined in model map_regions gms e g remap BRH ASE YES rrmapc HHR RSE VES That file should be currently set up to work with GTAP version 8 Other versions require an update of the mappings as the list of GTAP regions might differ A first view reports the meta information on the model It is here used as a first example for the XML based definition of views Basically each view can define filters in the different dimension The filters end with sel and start with the logical name of a dimension To give an
65. ket capitalhHarket tresHarketz 1ndMarket p results Unr Heta res Market tots dom p results Wor Meta Ind Market tots Set a aar eet a n Ww Ww m I dom Sendif That approach differs considerably from the way runGTAP allows to exploit results For a formal discussion on these differences see Britz 2014 and Britz et al 2014 The equation and variable viewer discussed above allows views on the variables equations and related parameters more similar to the runGTAP exploitation tools Exploitation and flexible aggregation The exploitation is based on views in the multi dimensional cube defined in p results Several such cubes representing different counter factual runs plus typically a benchmark can be loaded simultaneously in the viewer The views are defined in GtapInGamsTables xml As the set of sectors regions and factors can differ from run to run report gms generates a XML file generated xml which that information e g for the products loop i put lt product gt put amp itemHame i te i itemHame put amp Key 1 t1 lt key gt put amp sel all praducts sectorU ieuc sel put i product That file is included into the view definitions at run time In order to allow the viewer to work with maps a co ordinate set of individual countries is stored in GUI world zip The standard case of GGIG applications are f
66. lgorithm with five pre solve steps was applied The aim of these tests was not only to ensure a stable numerical implementation of the model but also to gain experiences with solution behavior on larger dis aggregations The times reported in the table below are for a full model run without post model processing switch on but including the time needed to store all symbols in a GDX container i e they cover loading the data base model calibration a benchmark solve and solving the shock The test show the expected more than linear increases in solution time if model size and complexity expands Solution times with a fast multi core laptop will be about double the times reported below The aggregation definition files agg used are found in the repository in the data directory The test shocks can be found in the scen directory The distribution which includes the GUI also comprises the described above test suite for the model realized as an input file for the batch utility gui testbatch txt It is recommended to run the test suite after changes to the model code 23 Many of the test solves will actually run even somewhat faster without the pre solves switched on However if the solver fails to solve the shock without any pre solves quickly it can often spent several minutes until all infeasibilities are removed As the potential further gains in solution speed by switching off the pre solves are limited we opted to show in here res
67. lian List af available EXP amd REF fibra The program will work for a while and should end with HTML documentation is ready as shown below p NTML documenation is ready persing T bratz GTAPinGams model mrtmcp gms done parsing T brice GIaPinGams model report gms done parsing Ti bratz GirAPinGams model scen user scenarios noshock gm Generzace MIML page for Gams symbols Generave HTML page for Gams source files Generate MIML pages for project Simulation HTML peges are generated Afterwards you find a in the chosen output directory a list of file index html is the starting point 62 amp TAbe GTAPIn Game Htmibectindex herd D G X B TwiGTAPInGamHmiL X Variables Used in project Simulation technical documentation open al close all Automatically generated from Dy ar amp GTAPIGAm b Her Doc olr Consumption ERES TAbrit amp GTAPInGamsimodeliSim 55 bus ei i iur Mame Jinderungsdatum at 25 10 2013 11 35 18 can 7 EV header dri T amm3 11 37 EYIT E indies himi 253012213 11 27 p etr Bl dien 15 10 2013 11 22 Beo dtres cit 25102012 11 22 i amp Equations Parmi 25 10 2013 11 3 ddfim 7 Files html 25102013 11 22 ddgmi 7 7 B Functions Mmi 2510 202 11 22 d pm i GANS DATA ASATXS GDX html 25 10 2015 11 22 d mi FJ GAMES MODEL ASATXS GET him 55 102013 11 22 dim E GAMS MODEL BTAPSRATA O45 html 25103013 11 29 dign E GAMS MO
68. lied filtering and rebalancing approach is an extension of the method and code developed by Tom Rutherford for GTAPinGAMS see e g http www mpsge org gtap6 It deletes component of the SAM depending on their shares on specific totals according to the Relative tolerance entered on the interface e Domestic and imported intermediate demand of a commodity are dropped relative to its total output e Private government investment domestic respectively import demand of a commodity are dropped relative to total private government investment domestic respectively import demand e Trade flows of a product are dropped if both shares on total exports of that product and its exporter and on imports of that product and its importer are below the relative threshold e Production is dropped if net production of a commodity i e after intermediate use of that commodity in its own production is deducted is below the relative threshold with regard to total net production The filtering process imposes restrictions which should maintain the regional SAMs balanced Additional constraints ensure that production activities require added value and intermediate inputs if not already otherwise found in the data base As filtering systematically removes elements from the SAM and the trade matrices the process implies without further corrections shrinking the economies During rebalancing the algorithm can therefore add penalties for deviations fro
69. lts r U 1 use oris The report gms programs also stores meta information on the run on the cube such as the number of sectors regions and factors model size and solution status and about how factor mobility is modeled 69 Heta information on model structure and solue p results Unr Heta H of sectors tots dom cardi p results Unr Heta H of factors tots doam cardiff p results Uor Heta i4 of regions tots dom rar r acronym HCP CHS iftheni modeltype MCP p_results Hor Heta Hodel type tots dom MCP Selse p results Unr Heta Hodel tyupe tots dom CNS Sendif Siftheni NOT modeltye NONE p results Wor Meta of equations tots dom p results Uor Heta H of uariables tots dom mnodelz numuar 5 p results WUor Heta SunInfes tots dom nodelZ sumInfes model sumInfes ne inf nodelt numequ p results Uor Heta HunInfes tots dom mnodel numInfes model numinfes ne inf p results Uor Heta H Iterations tots dom nadelz iterusd p results Uor nodel resusd Heta 4 seconds solution time t ts dom acronym sluggish mobile fixed priced p results Wor Heta Lab Harket tots dom p results Unr Heta skl Market tots dom i z labmarket 9 p results Uor Heta capital Harket tots dom sklhar
70. m the following aggregate transactions Penalty for deviation from By adding these penalties terms the non deleted entries and thus most important transactions tend to be scaled upwards It is generally recommended to use these penalties terms The code will also scale all non deleted trade flows to approximately maintain the total volume of international trade 21 The thresholds are stepwise enforced starting with 10 of the desired final ones Once the final thresholds are active filtering is still applied several times until no small values are found any longer The code should ensure that the resulting transactions are still fully consistent with each other i e both the resulting trade matrices and the SAMs are balanced The changes imposed by filtering and subsequent balancing are stored on the itrlog symbol in the GDX container with the final results Inspecting how the stepwise enforcement of the thresholds impacts on the number of non zero items can inform on an appropriate level for tolerances to be used Thanks to balancing also rather dis aggregated versions of the model with large number of sectors and regions can be used The biggest impact of the filtering is typically on transactions related to bi lateral trade flows Here often 50 or more of the flows account for only 1 of the total value of these transactions Thus tiny changes in the relative tolerance can have a considerable impact on the number of dele
71. missions Aggregate intermediate demand Numeraire products Numeraire regions Non default parameters Modules and parameters Currently different modules and extensions as shown above can be added to the standard GTAP model Please note that these modules are not a full replica of the GEMPACK implementations The simple energy nests allow for substitution of primary and secondary intermediate demand for energy inputs The labor nests which is also part of GTAP E depicts substitution between skilled and unskilled labor The Aggregate intermediate demand extensions reduced the number of Armington agents to four final government and investment demand and aggregate intermediate demand The latter replaces the sector specific nests in the GTAP standard model 32 Non Default parameters can be used with the model If non default parameters is not switched on the parameters of the standard GTAP model as provided by GTAPAgg are used If the option is switched on a file from disk can be chosen aMeters Farameter extended Parameter file Pa ameters iParameter extended Parameters Parameter extended Parameters parameter_gtap_aez Parameters parameter_gtap_agr Parametersiparameter Gtap standard The possibility to use non default parameters is important as certain model extensions are driven by parameter settings To give an example in the standard GTAP model the transformation elasticity be
72. mpt Option file 3 relaxes the convergence tolerance further and is not intended for production runs but for debugging purpose Both with MCP and CNS the above mentioned combinations of solver settings have proven to work best The user can also chose which solver to use for NLP and CNS Tests with a beta version of CONOPT4 which parallelizes e g matrix factorization has shown considerable speedups for very large models The reader should however note the so far CONOPTA is not officially released and that up to that point only users highly familiar with the code of the model should use that beta version which shows great potential If the model is solved as a NLP it will on demand first solved as a CNS CONOPT uses a somewhat simpler and faster algorithm with CNS which might however sometimes fail It is generally recommended to try Use CNS first Presolves For large models pre solves can be used during which single region CGEs for each region in the model are solved independently That process is repeated several times to inform the individual country CGE about changes in others regions The single regions are only introduced as an intermediate steps towards the solution of the full model They thus differ from regular single country CGEs which are usually solved either at fixed international prices or by rendering import and export prices depending on import and export quantities not considering different trading partners The pre solves in
73. n Q Prepare data gt Scenario 1 10X10 Simulation Scenario 2 tasks Import from GTAPAGG Scenario 3 Scenario 4 Scenario 5 Scenario 6 Scenario 7 Scenario 8 Scenario 9 Scenario 10 Scenario 11 Scenario 12 Scenario 13 Scenario 14 Scenario 15 GAMS Graphical Interface Generator Wolfgang Britz l E R 2013 University Bonn Institute for Food and Resource Economics Ini file GtapInGams ini User name undefined User type runner Once you press Load content of files into GDX viewer you can inspect the individual symbols used in the model 25 Ls GOXViewer the Views lel Selection et L l Optiom si Set AP table definition Ihe Use rami mn egoaten output vanadie selec3on Eauaton selecson Ux tabis defaut from Use smal font for non setecned war terms Sort code Int Show dialing to ink GOX dimensions to sets A maa xdmo 1 uii y mboss m GDI List of tables loaded roe GOX file s TABRIT Z G TAPRINGAPSUDATATTOEEO AG DX p 2 ve T Sep dco te Wi GrainsCrops c MeatLatk Extracton c Procfood Textivespc LightMnfcc teavyMetc c ui Conse Py ee me Y a pet Ji w 417245 7491 29 122141 ITAN 4102 32 1979941 15924 22 1146671 pet L 10 Dese 12202847 124495 21 1 096 3r6274 19 1385 56 159011 58 za 1767490 T m 49648 26480 tata oa C564 06 ttost 82108 777 a2 1542 08 i i f 12289544 12 24942 107581 0 4327542 20143 53 amna MANO I
74. n E B DET mur 4 ua nam i ae i 1 i oy sam a oc 1 m ot E and 7 ae E aes E E nml 7 a pts a E nsa om n Ju T em 4 4 amp on T cs E os T Eon fs E su E asd uw l p idi na aF LE e Ti o c Aka ar GrH ina i W ea d em 5s ini E d Ari om umm aed Encpup comes Pood j Last 4 i icm T t T a 3 Poi d H Bee lA W orem Sa yf E k u 1 Fi a amm 2 lt i a oem E s EE UJ i H oe a L SO e met a coon a nc m di f P Fo LE ewm ds Cirad Bem d qa m ual bind x5 da w sn ag tS meom i pecs 4 a Lec ar nus noma Lon ie GoT D95 uas LEO coma D Tia Chau ara Cri pue De om miam P j am Vus mue d A bii pud P ane 7 at gone pr pt 1 1 Lir P E p x 9 91 ae 3 F E ma d a Ez F Nt d E Ea i si ouace m all vdd m3 P y eF cman F uci s i ass d 4 dri T 2 nsa nsa 120 n GUT D ab Lm 4 ow Em sa Tahal eee Cri pee aud aa De aci uie scERMAADS OFT Tod Of a_i ieeRaaADOS et etd demand E ES uiek SOBRE ET Goes demne w RES LOTA Sobre GT incanrachaos denad i mrs uane OAA Gu ere demen 0 duh AA CirxEsehkRod SFT epot dered EPE LEPR CORRE PT jera ge barraa Note If the entropy estimator is used the mean calculated in the GAMS code uses the probabilities as weights but they are not reflected in the GUI when the in built statistics are used Getting an overview on the code The
75. ng file GAMS RC O Open mictude file Scroll Lock That pop up menu offers you the options to open in your favorite editor the GAMS file the listing produced by GAMS and the include file generated by the interface Notes e The use of the GTAP AGR or GTAP E module requires a data base with a full dis aggregation of primary agriculture respectively detail for energy sector switching on the module using a data base with insufficient sectoral detail will lead to compilation errors e You cannot use the global bank mechanism Global equal returns to capital if capital is declared immobile 19 e A license either for PATH or CONOPT is required to solve the model Using the pre solve algorithm requires CONOPT as does using the filtering approach when generating a data base Data base generation The GTAPAgg utility which you should have received together with your GTAP data base allows you to build a GTAP data base with a sector and regional aggregation chosen by you see https www gtap agecon purdue edu products packages asp a training video on you tube https www youtube com watch v QDBROKqNuzE amp feature youtu be or the first pages economia unipv it pa agine_personali msassi QPA materialQPA Introduction 20to 2 OGTAP pdf That data base comprises different data sets stored in HAR header array files a proprietary data format for GEMPACK Mark Horridge has developed a program called HAR2GDX which converts a HAR file in
76. ng the Exploit results button The layout will change as seen below Please click now once in the Dataset selection otherwise you will get a warning message later Figure Scenario selection for exploitation 16 Select the scenario 10x_10 example_test and press show results you should see a table similar as the one shown below Once you have produced your own results you can find the data bases you used and the scenarios generated in the drop down boxes and you might combine different scenarios for comparison The ISi view button in the upper left corner lets you select different views on your results Lu Model overview gt Model properties Markets Sector aggregation Sectors Regional aggregation Trade Income lt No table Income per capita faeces j z i Income per capita map regions elt Tax income Tax income composition relative 100 variables Schemer How to produce your own results is discussed in the following Information on the different views is given in a section below Making the interface working with GAMS In order to use the GUI to generate data sets or for simulation runs you have first to register a GAMS executable In order to do so first leave the result viewer as shown below Mes Data View New Date Verma ating teflon of lait cme ion Clese all Data View Windows bty
77. nment spending for domestic imports and saving the latter is residually calculated The rectangle Household consumption shows final consumption by the household the arrow to its left shows spending for imports net of taxes The rectangle on the top is final demand it is sourced by export and international transport demand left household saving and government demand and domestic intermediate demand for that sector The schemer view as with any view can also show differences between scenarios and provides an aggregation to total world The schemer can also be used to visualize individual sectors and commodities or an aggregation to total output demand 45 st Inter 46 1 tes Imported Similarly Model Overview Price schemer visualizes the composition of per unit costs and prices where the boxes represent price the arrows show costs shares respectively tax base multipliers and the triangles substitution elasticities Combined as shown below with a percentage difference to the baseline it allows a rather straightforward interpretation of major price changes in the model The results are available per region and sector commodity as well as aggregated to total output respectively private consumption and world level The table Welfare Income provides an overview in key results such as GDP and price indices 47 4 O 3 Regions Total or houset AG World Total Y 10x10 Tariffs Y Money metric 3
78. ns General settings Model structure Output optons Agerithm GAMS GTAP in GAMS VE Output options Output CSV CONVERT Outnut for flew schemer 7 Output fer prier hemer Store AM nCOX Post model regional ancl orclor aggregation Aggregation file Wn E Mes 10x30 X Also make sure that output for the GUI is selected Output CSV e CSV will generate a SAM and some other core results and store it in CSV format under results run e GUI will generate a GDX container with a parameter for use with the exploitation tools see below e GDX will store all other GAMS symbols parameters variables equations sets etc in a GDX container including output for the GUI if selected e CONVERT will generate output for the Equation and Variable viewer which is similar to AnalyseGE see section below If you are interested in a graphical depiction of flows and price changes see chapter Overview on existing views for exploitation below you can deselect these options to save some memory and speed up post model processing The same holds for the post model aggregation The screenshot below shows what post model aggregation contributes in most views results are not only available in the regional and sector dis aggregation of the underlying data base but additionally also aggregated results as shown below 31 Regions ep Items Word Quantity ef tot Origins Y v Total 137582 97 31212 10
79. nt to check the box Clean window with GAMS output with each new GAMS compile start Path to view definition file tables xml Sort code lists in predefined tables Path to Editor E Sort code lists if showing all elements Language to load from tables xml Pagikan V Use task specific settings in interface T Debug XML table definition output to OS prompt In order to check if it worked press the Compile GAMS button You should see now the compile output from GAMS in the window below the button 18 GTAP in GAMS V6 mo File Utilities GUI Settings Help GTAP in CAMS MEE General settings Model Shocks GAMs Data base compilation GTAP in GAMS V6 General settings Simulation GTAP in GAMS V6 tasks Simulation Input file with GTAP data gdx 10x10 Exploit results GAMS run Simulation mrtmcp gms 2792 18 Mb mrtmcp gms 2792 17 Mb Executing CONVERT elapsed 0 00 03 963 Convert 2 0 24 3 3 r48116 Released Sep 19 2014 WEI x86 64bit MS Windows Using Option File Reading parameter s from U GTAPInGams model convert opt gt gt gams convert gtap 6cns gms gt gt dict convert gtap6cns txt Finished reading from U GTAPInGams model convert opt Writing Dict convert gtap6cns txt Writing Gams convert gtap6cns gms Restarting execution mrtmcp gms 2792 8 Mb Reading solution for model gtap6Cns GAMS Graphical Use
80. nt threshold only to the intersection of the inputted regions and sectors otherwise all regions and sectors inputted will be receive different thresholds Take an example where you enter for regions xoc and for sectors pd If Reduced thresholds only in combination is NOT switched on all transactions of the region xoc and all transactions for the sector pd will be treated differently If the Reduced thresholds only in combination is active only the transaction relating both to pd and the region xoc are exemptions However filtering for the remaining sectors regions has still an impact on these exemptions For example if production of a sector in a region is dropped the related export flows need to be dropped as well affecting potentially transactions in regions and sectors where tighter thresholds are used Tests have however indicated that very few transactions are lost in regions sector where stringent thresholds are applied as long as the overall filtering thresholds are not too aggressive 22 Model size and solution behavior The model template including the GTAP AEZ and GTAP AGR modules was intensively tested under a suite of test shocks applied simultaneously to all regions and sectors 50 reductions in tax rates direct consumption factor tariffs export subsidies 20 endowment changes and 10 tfp shocks for all primary factors The model was tested in three configurations 1 the GTAP standard model with fixed all
81. ntation As seen from above the interface allows defining two shocks with related set lengths The simplest layout is shown above one shock with a set length of unity The automatically generated include files will each carry lines as follows parameter shackl x set s1 51 x51 shock1 s1 1 58398955806 The shock1 parameter can be used to e g change the population size in all regions by the same factor If regional specific shocks are needed the length of the set must be equal or higher than the 55 number of regions Assume e g a dis aggregation with 5 regions and the set length of shock1 accordingly increased to 5 The include file will comprise lines as follows parameter shockl x set s1 51x55 shack1 51 H 68917 65639 shack1 572 0 7597 044235 Shockt s3 0 6128377 057 Shocki s4 1 2482232299 Shocki s55 1 3406 050177 With a cross set between r and s1 using the pos operator from GAMS the numerical values on shock1 can be mapped e g on the population size set r_to 51 r 51 r to si r st s r paos eg s1 po5 yes pop r popi r sumir tno 51 r 51 shack1 51 The two shocks in combination with flexible set lengths should hence allow drawing from experiments for different types of shocks quite easily without the need to change the GAMS code for the sensitivity analysis Note that these shocks can be combined with changes to the parameters Parallel solves and analyzing the
82. ocation of foreign savings and 2 alternatively the global bank mechanism as a default in the standard GTAP model and 3 the global bank mechanism in combination with GTAP AGR and GTAP AEZ The tests used a full sector dis aggregation 57 sectors and varyingly sized regional aggregations 10 24 36 45 56 68 Additionally a 10x10 model was solved for which the GTAP AGR and GTAP AEZ modules cannot be used due to missing sectoral detail The parameterization was kept at defaults Most of the tested models are large with regard to the number of sectors and countries compared to applications of GTAP reported in publications All tests were also run solving the model as MCP in PATH and as a CNS in CONOPT Table 2 Data bases used in the standard tests Model size Filtering thresholds Resulting non zeros in global Model size relative tolerance SAM including trade flows GTAP Standard maximal minimum of number of variables transactions substituted out 10x10 74 700 8 000 57x10 35 000 92 000 Note The number of transactions accounted for during filtering is somewhat higher that the resulting non zero SAM entries GAMS version 25 5 4 was used on a computer server in combination with a beta release of CONOPTA which executes certain part of the NLP algorithm in parallel We would like to acknowledge the continued support of Arne Drud especially to let use CONOPTA for the tests Results with CONOPT3 are somewhat slower The pre solve a
83. omatically generated e g histograms Currently only one view with the core income results is defined Parameter Income Income histogram by draws Income stats Income per capita Income per capita compare households Total draws001 Income per capita map Tax income Tax income composition relative 100 draws003 n 58 er e EAB 2 a n r eli eR prep Cu 3 WHE OD RET Hae ERUNT LL gS rri BHABARLIERTUETE moa rm LEE T4 Rc pa qeu a anm Wu cec as mom orn z 17e zoo a mmg T la Fa ore B ses oues scenakien aer Note If several scenarios are analyzed the distributions for several scenarios will be shown together Alternatively the Scatter plot type graphics can be used to plot additionally the correlation between the variables depicted a eat heagan ing drai FI ice Lb dep Men zu A KG item Lee i We A Eh e a esie eu F T c LE am ame Az amm 1 aba i ats Am ai a Lo i LER c AZ axi J ataa 1mm ban CERTI Az 1m az qo i ax gm d Ax 15m 4d Pas Bee Fiai Fee Bao Hoa Hi a am gie ae mo a 960 LR Inablsc nreo masas Tora rears Fmt moore i en E EE d 3e anes xn i f 34m es i 3 mam Dea ar rrari indes alan z masg 32486 a i 32 mm ds LL LETT a SUI dual Pais eeraa munia Tain FLEGT Lie 120 aum 4 ros nan S308 RSLS s 3 ori nou 1 kaif eee oar a
84. ons allows for a rather flexible design of the experiments The program can also easily be extended to also apply sensitivity analysis to shocks i http www sandia gov 54 The general set up of the sensitivity analysis is defined on the left most group of controls the number of experiments parameter draws and related model solves the maximum number of parallel solves Controls for sensitivity experiments Sensivity analysis oms oo CONNNNNNEENEN T Number of expenments w SymnaM Substitution domestic fmpc ts Uniform 1 ome Substtuton between importers Truncated normal up stdDev Peak Maurum parallel GAMS processes m Somay SudsSuton between factors Truncated normal e _ Only generate dans J Omega Transformation between factors Truexated normal 0 50 1 50 1 00 1 00 Pir A Dhodi Trurcated normal v 0 30 150 140 109 F1 enam a nor G shock bhod truncated a 5 0 50 1 50 1 00 1 00 Set length of shock2 1 00 1 00 1 00 1 00 0 50 1 50 1 00 1 00 Use uniform shocks across sectors factors 7 0 51 1 50 1 00 1 00 Lice entropy etima JJ The left hand side determines the number of draws and how many GAMS processes are used in parallel to simulate the different experiments The user might also chose Only generate draws to prevent the actual model runs in order to check the layout of the scenarios Using uniform shocks across sector factors will apply the same relative change for th
85. pare deta Scenario d 57x38 TARIFFS Senulaben Sainani 2 amp Ss Scenario 3 Scenario 4 Gs Scenarin 5 Scenario amp GAMA Grophraj Liner Entazfarg Conaratnr i IL R Wolgieg tre Scenario 7 Lims Ran Datnset S x Scenario amp insida for Food and Tissue Show resulta Lond content ol fles into GDDX vievenr anaren RSS ZZ Fiuacurca Economics Shew meta Show results Load content of files into GOX viewer Return GTAP E GAMS VO Inifis GlapimGamnsni Usermame undefied Unertype acministrator Simulation Simulation That will open a new window as shown below ET Use pienis n eamb A dte Vic abla nilarin Use traal Font Sar nan dactied vis taret altal alean Let el tabla leddied fram GON Blaa TABRITZ Ye TAFREMGAMSIRISULTGRUMISTICIG TARI TEGEN uxerinput The window List of table loaded from GDX file s shows the status of all GAMS symbols variable equations sets parameters after the solution of the shock Double click on any symbol you are interested in e g the factor income facty as shown in the screen shot below That will open a window where the results are shown for the base case as initialized from the data base the benchmark test 2 check and the shock 40 ai Options Une indents in equation output Use areal fort for non iaer var lerem use table defisitinna trom mall Shan dialog to lnk GOH dimensions to sets List af taibles lade from G
86. plogo jpg amp gtaplogo ico amp 57 16 bt 57 24 bt 57 36 bt 57 45 bt svn 57_56 b e batchOutput e 57 68 bt oe data 2 batchTest txt doc I GTAPInGams bat ey expRefDir ej GTAPinGams xml gams GTAPInGamsTables xml 9 gui A aez zip e htmlDoc A flowSchemer zip 6 results ga priceSchemer zip g world zip and double click in the GUI folder on the file GTAPinGAMS bat it should open the interface You might want to put a shortcut to that batch file on your desktop to easily start GTAPinGAMS You first see a message which tells you that a GAMS executable is not yet registered ES You can first ignore that warning for now After pressing OK you should see a program opened as below If that does not work open a command prompt start from there the GTAPinGAMS bat batch file and analyze the error Most probably JAVA is not found In that case either put it on the path or change the batch file such that it calls Java from the directory where it is installed 13 n lt File Utilities GUI Settings Help GTAP in GAMS V8 worksteps General settings Exceptions GAMS Prepare data 5 UTR GTAP in GAMS V8 General settings Simulation Input data umm GTAP in GAMS V8 tasks Import from GTAPAGG Input file from GTAPAGG zip 10x10 example w Postfix for data set name Load land use data E 2o Filter method Rebalancing Absolute tolerance 1 E 10 5 Rela
87. r Agricultural Economics 107 119 Britz W Kallrath J 2012 Economic Simulation Models in Agricultural Economics The Current and Possible Future Role of Algebraic Modeling Languages in Kallrath J eds Algebraic Modelling Systems Modeling and Soving Real World Optimization Problems 199 212 Springer Heidelberg Germany Dom nguez l P Britz W amp Gopalakrishnan B N Post model Analysis in large scale models the examples of Aglink Cosimo CAPRI and GTAP Paper presented at 2012 GTAP conference Rutherford T and Harbor A 2005 GTAP6inGAMS The Dataset and Static Model Prepared for the Workshop Applied General Equilibrium Modeling for Trade Policy Analysis in Russia and the CIS The World Bank Resident Mission Moscow December 1 9 2005 http www mpsge org gtap6 gtap6gams pdf Arndt Channing An introduction to systematic sensitivity analysis via Gaussian quadrature GTAP Technical Papers 1996 2 https www gtap agecon purdue edu resources download 1209 pdf Van der Mensb Ko J H and Britz W 2013 Does Regional and Sectoral Aggregation Matter Sensitivity Analysis in the Context of an EU South Korea FTA https www gtap agecon purdue edu resources download 6313 pdf 73
88. r Interface Generator Putfile xml U GTAPInGams gqui generated xml l L R Wolfgang Britz Status Normal completion TEM es Job mrtmcp gms Stop 12 01 14 15 56 29 elapsed 0 00 05 414 E Institute for Food and GAMS RC 0 d Resource Economics m GTAP in GAMS V6 Ini file GtapInGams ini User name undefined User type administrator Generate include file If you have registered an editor you can also check its proper functioning a right mouse click in the window with the output should open up a pop up menu Comple GANS SurtGAMS StepGAMS Explot results Di game24 i gams exe mrimcp gma workdir T britz GraAPinGams model curDir TiMbrita AGTAPInGams mod GAMS 24 1 3 Copyright C 1987 2015 GAMS Development All rights reserved Licensee Institute for Food and Resource Economics 5 2130213 0232AN WIN University of Bonn ne4443 License for teaching and research at degree granting institutions Sterting Compilation mrrmep gms 14 2 Mb mtIECp inc gms i01 3 Mb mrtmcp gms 21 3 Mb grap dacta gna 23 3 Mb SDXin T NVbzitzNMCTAPInGams dataNasa7x5 gdx gtapecata gms i77 J Mb T BICECD gm3 779 3 Mb neShock gesil7 3 Mb mrtmcp gme Ho1 3 Mb feport gms 247 3 Mb mrtmep ogms 802 3 Mb RefFile T MbritzMACTAPIGSDams modelMSimulation ref Status Normal completion Open gam file JOD mrtmcCp gms Stop 10 25 13 09 39 01 elapsed 0 00 00 093 Open gama kati
89. r f a t Resource Economics ini tke GlapisGamsum User name undefined Usertype runner You can now modify the content with the editor by typing directly in the right hand window Once you shave edited the code additional buttons will appear The list with the scenario groups and shock files will now indicate that you have made changes to the shock file 28 Scenario categories i base scenarios Endowment and productivity shocks Factor specific TP user modified L Factor supply changes Policy Shocks Test Shocks User scenarios In order to save your edits press Save changes If you do not press Save changes your edits won t make it in final shock file generated You can now double click on further files to add them to your shock file The files you have currently combined are shown in blue you can also deselect some of these files again Once you have chosen the files you want to combine and potentially edited and saved your changes you should now enter name for the shock file to create e g TFP20 and a description of the scenario as shown below Scenario description Enter scenario name TFP 20 Hicks neutral TP of 20 across regions and sectors Enter scenario description Afterwards press Store scenario Enter scenerig name TFP 20 el TF of 208 across regions and sectors base scenaros Scere categories Endowment and prodxtety mocks LI
90. results The actual solves for each individual experiments start from a savepoint generated after solving the benchmark such that execution time is only spend on model generation solving and post model reporting The solve times reported above stem from a by now medium fast four core desktop and underline that from a computational view point sensitivity analysis is no longer a concern The interface will allow showing either all draws and or the weighted average of the draws meand Result exploitation Scenario 1 RES BASE SCENARIOS FUTURE Scenario 2 v Scenario 3 v Scenario 4 v Scenario 5 v Scenario 6 Scenario 7 v Draw selection m Al draws Scenario 8 Scenario 9 v Scenario 10 v Scenario 11 Scenario 12 X Scenario 13 v Scenario 14 v Scenario 15 v If only the mean over the draws is selected the result tables look almost identical to a single simulation run The tables will report two outcomes 1 the mean of the draws labeled meand and the 2 result at the original parameter mean what one would simulate without any sensitivity analysis 56 Percentage diff to Draws MA mean y RES_USER_SCENARIOS_GFT 5 y Total 131296 89 30592 51 9509 66 64336 11 12834 53 13503 61 520 48 0 02 0 02 0 02 0 03 0 02 0 03 0 04 131327 41 30598 80 9511 28 64352 24 12837 40 13507 00 520 67 eo 1462 23 591 16 751 72 119 35 Crops 0 13 0 12 0 13 0 10 1464
91. s 8 C Sets ddim is referenced in HL Files it C Equations Tog Declarations Da nitona Assignments References Elements EHEJ Variables L Simulation EHEJ Models File Projects SHO Functions GAMSIMODELIMRTMCP GMS Simulation E G SaureeFiles EHD Treaviews GAMS MODEL IREPORT GMS Simulation L madulas Toe Declarations Deteions kssianmaents References Elaments Equation and variable viewer similar AnalyzeGE Whereas the HTML pages document the project the equation and variable viewer helps you to analyze a specific model instance The necessary input files are automatically generated with each run in comparative static mode if CONVERT output is chosen 63 1 GTAP in GAMS Fle GUI Settings Help ST GOX Viewer Wet CSV to GDX MPa Start Command prompt Start equation and variable viewer GT Build HIMI documentation Define scenano Ib britz gtap8Ingams GUI results run 10x10 Cons Convert gams output expRefDir convert_gtaps gms Dictionary file expRefDir convert gtap8 txt ef 7 i GDX fle from CONVERT before model solve expRefDir convert_gtap8_before gdx Setfile ox fies od Stile J GDX fie from CONVERT after model solve expRefDir convert gtapS after gax Set file ef Set fite EaR gans con gas MM en setfite Reffle expRefDir Simulation ret i Set file note one might want to change the first GDX files to the last experiment you have run
92. s is what is called total a simple sum of the rows below Thus the information should generally not be confused with the sum information shown in viewhar e There is only in few selected cases a 1 1 relation between a variable in the model and numbers shown in table typically the results shown in the different views stem from some post model processing e g multiplying a quantity index with a price e Instead of using the button Show results which presents the different views discussed below the Load content of files in GDX viewer allows to inspect all symbols used in the code including all variables and the generated SAM see next section The GDX files are stored under results run and can be naturally inspected with another GDX viewer such as the one comprised in the GAMSIDE e fthe user want to inspect variables similar to AnalyzeGE including decomposing equations she can use the Equation and variable viewer see below 39 Relative difference for all variables as in runGTAP GEMPACK users typically compare simulated against benchmark levels for model variables That is also possible with the interface In order to do so select the file with the result and press the Load content of files into GDX viewer button GTAP in GANG VE CCIE LG a File Lilies lzi GUT Settings E Help a GTAP in GAMS VO worksteps Resull expicetalion Pre
93. s mec Vegetables fruit nuts Oil seeds Sugar cane sugar beet Plant bazed fibers Crops nec Processed rice Cattle sheep goats horses Animal products nec Raw milk Weeol silkworm cocoons Heat cattle sheep gears horse Heat preducts nec Minerals nec Forestry Fishing Cual Oil Gas Processed Food Vegelable oils and lats Dairy products Sugar Food products mec Beverages and Lobacer products Textiles and Clothing Textiles Wearing apparel Utilities and Constructinn Flolhiricily Gas maniacs divbribulion Waber Conslbructiun j Transport ane Cosrarsuriicaliam Trade Transport neg Sea Lramsport Air transport Communication ther Services Insurance Financial services nec Business services nec Recreation and other services PubAdmin Defence IMealth Lducat Dwellings Model Overview SAM reports the results in a SAM including row and columns sum and potential imbalances Due to rounding errors and feasibility tolerances of the solvers small imbalances in relative terms are possible and not of concern 7270731 Middle East and North Africa 0597 50 Sub Saharan Africa mer Rest of World TH GramsamdtCrops 1628 41 45166 80 14725 96 10667 06 3454 85 13415 00 06181 22 16060 21 8774 68 31107 81 Livestock and Meat Products 1357 37 27000 54 35067 70 Z10 45 17718 91 ATH GI bc ta b 7433 80 709 07 10343 05 Mining and Extraction 9455 21 416027 66 60202 07 90717 98 297363 31 35377 76 42404 20
94. shock Based on the flexible aggregation used in GTAP the sets used for regions sectors and factors are run specific and depend on the data set loaded The information about how the output is logically structured is also inputted in the GGIG definition file gtaplnGams xml 68 task lt name gt Simulation lt name gt lt gamsFile gt mrtmcp lt gamsFile gt amp incFile modelXmtrmcp inc lt incFile gt amp curDir madel amp curDir lt regionDim gt 6 regionDim gt lt dimSDime1 Items lt dimSDim gt lt productDim gt 2 lt productDim gt CactivityDime3 Sectors and institutions lt activityDim gt amp dim Dim riginsc dim Dim amp scenDim 5 Scenariaosz scenDim amp resdir runi resdir amp gdxSymbol p results lt gdxsymbol gt Filemask gdx 4 filemask lt task gt The report gms program also performs aggregations e g to total use m aggregate over origins domestic and imported to totals set agg PQu Q U G aggregate to total use set useCols set i hou gau intIrs set s inu p results r agg i use oris sum useCols p_results r agg i useCols oris Average prices and tax rates are calculated afterwards p results r P i use oris p_results r Q i use oris p results r U i use oris p results r Q i use oris p results r T i use nris p results r U i use nris p results r G5 i use a ris p resu
95. stributed for free However the underlying Java code is currently not open source Working with the GUI Getting the model code and setting up the GUI The code and GUI are distributed in three ways First a GAMS code only package can be downloaded as a zip archive from the GTAP web site The GAMS code is identical to the one used in the GUI version but the java binaries and other files necessary for the GUI are not included That GAMS only package might be preferred by users wanting to use tools such as the GAMSIDE develop their own GUI or may want to link the GTAP code into other GAMS projects Second a zipped archive Gtap8lnGams zip at ftp agp uni bonn de userid and pwd is gtapingams comprises both the GAMS code and the necessary files for the GUI SVN checkout The third option is to benefit from SVN a software version system on which the model codes are hosted on the repository https svn1 agp uni bonn de svn gtap8ingams The option is especially recommended for users who both want to use the GUI and foresee also own changes to the code In the SVN repository the development of the standard GTAP in GAMS continues We expect that over the next few months additional modules will be integrated into the code and clearly bugs corrected and improvements implemented once the code is more widespread used Using SVN allows updating more easily to bug fixed or new extensions Therefore as a first step we recommend installing a SVN client s
96. tched on all GAMS symbols sets parameters variables will be outputted to a GDX container and can be analyzed in a GDX viewer either the one delivered with the GAMS IDE or the GDX viewer built into the GUI The model s results can be mapped back into a SAM and the SAM will be stored in the GDX 5 Theinformation from 1 4 can also be inspected with the Equation and Variable viewer discussed in another section which also provides a decomposition of each equation similar to analzeGE 6 Finally the exploitation tools can be used which also cover the SAM as discussed in the following Which option works best depends on the use case and user preferences Model listings with limcol and limrow switched on are extremely useful during model development and debugging whereas the exploitation tools are probably the best approach to systematically analyze results from a shock Individual equation decomposition can complement both approaches We can only recommend trying out all approaches at least once to find out which one fits best one s preferences under specific use scenarios Inspecting results with GUI is nothing new You have done that already if you followed the short introduction so press on the Exploit results button and next on Show results again Use the NoShock and your own scenario 37 Result exploitation Scenario 1 NOSHOCK oy Scenario 2 USER_SCENARIOS_TFP_20 Scenario 3
97. te of intermediates linked into the top ND nest as e g applied in GTAP AGR to allow for high substitution of feed intermediates in animal production d Add technology nest to model give it a name tNest primEne YES Define a helper set for the intermediates Set primEne i not used elsewhere coa c oil c gas c J The mother of the nest is the top ND bundle litas primene prinene1y tNest_n_a ND primEne a sum r primEnel t io r primEnel a t YES tNest i a primEne primEne a sum r primEnel t io r primEnel a t YES sigmaNest r primEne a sum primEnel t io r primEnel a t 0 5 Linkthe intermediatesinto the bundle Define the substitution elasticity Note The code will automatically remove intermediates linked into nests from the top ND bundle 3 Thethird example shows how to combine intermediates and factors into a CES composite which in the example is a sub nest of the top VA nest ao Add technology nest to model give ita name sew d be Define a helper set for the intermediates Set scndEne i not used elsewhere p_c c Ade z The mother of the nest is the top VA bundle E tNest_ naC va energy a YES tNest_ f _a energy 3 capital YES tNest i a energy scndEne s YES sigmaNest r energy Vass Link the factors into the bundle Link the intermediatesinto the bundle Define the substitution elasticity 4 The last example reproduces the nesting of th
98. ted transaction and one might need to experiment with settings in the range around 1 E 1 to 1 E 4 to find a compromise between sparsity and the implied changes on structure of the economy For very large data sets e g a 1 1 version filtering thresholds above 1 might be needed to yield reasonable model sizes The user can additionally define a minimum number of transactions to be kept which reduces the need to experiment with different thresholds as the filtering process will stop once less than the desired number of transactions is reached Tests with the model have shown that models with more than 400 000 transactions cannot always be solved especially if the global bank mechanism active A close look at the filtering statistics is recommended to avoid sharp impacts on the structure of the economy Special treatment for specific regions and sectors Exceptions Regions with reduced thresholds Sector with reduced thresholds Reduced thresholds only in combination Reduced relative tolerance 1 00 Reduced relative tolerance E When building a data base for a project it might be desirable to apply less aggressive filtering thresholds for specific regions and or sectors in the focus of the application The algorithm therefore allows defining lists of regions sectors with accompanying specific thresholds The codes for regions sectors needs to be inputted in the two text fields Reduced thresholds only in combination will apply the differe
99. tions becoming infeasibility which variables to fixed which however requires to solve the model potentially several times That algorithm can also fix similar problems which might not be related to the global bank mechanism 24 The tests have shown that in combination with the pre solves solving the model as a CNS with CONOPT is usually faster if the model is larger compared of using PATH and solving as a MCP Very small models are typically faster solved with PATH Having both solvers available for tests might hence pay off if solving large models is part of a project And clearly certain policy instruments such as production quotas are best captured by a KTT condition embedded in a MCP model Checks for medium sized models have shown no differences in the results produced by CONOPT or PATH That seems to indicate that once a model is declared feasible the results can be used Care should be clearly given when using PATH as a MCP solver after structural changes to model code MICP solvers might e g accept a fixed variable paired to an equation which results in a non square system whereas solution as a CNS will throw an error in that case Inspecting the resulting data base You can check if loading and filtering worked by pressing on Exploit results selected as shown as output file generated a GDX comprising different parameter and set definitions 7 GGIG GUI game So x Fie Utilities GUI Settings Help worksteps Result exploitatio
100. tive tolerance 1 00 Relative tolerance E 25 Minimum number of transactions 50 000 Number of stepwise increases of thresholds 105 Penalty for deviation from SAMS Hide Unihide controls Exploit results GAMS output LE 7 Were EIE E innt til GAMS Graphical User Interface Generator Wolfgang Britz ILR University Bonn Institute for Food and Resource Economics GTAP in GAMS v8 Ini file GtapinGams ini User name undefined User type administrator Figure Interface at first start The left hand side allows selecting work steps and the tasks comprised in them the right hand side carries controls to steer these tasks The bar in the middle comprises buttons to start the GAMS code of a task and exploit the resulting output Typical reasons why the GUI does not open and possible remedies are described briefly below e Java is not installed or not registered on the PATH In order to test that open a DOS prompt and type Java If you receive a message that the command is not found re install Java and make sure that Java also updates the PATH during the installation e Awrong Java Runtime Environment is installed Please first check if Java version 8 is installed and not some older version that check is not only useful for the GUI of GTAP8inGAMS but also generally for security reasons if one uses Java plugins in web pages On 64 bit Windows operation systems a 64
101. to the proprietary data format GDX used by GAMS HAR2GDX is part of a GAMS installation and used by us to load output from GTAPAgg If you generated a data base with GTAPAgg intended for use with GTAPinGAMS please copy the zip file generated by GTAPAgg to the data directory or store the file directly there from GTAPAgg In order to make the data available to the model choose the workstep Prepare data and the task Import from GTAPAGG The file you copied in the data directory should be available in the dropbox under Input file from GTAPAGG zip and selected by you If you want to use the GTAP AEZ extensions copy the 2007Luv81 tzip in the data directory and check Load land use data Please note that once the land use data for the new data base version 9 is released the filename is subject to change General settings GAMS General settings Input file from GTAPAGG zip 57X34 Load land use data Filter method Rebalancing Relative tolerance 0 10 Relative tolerance E Absolute tolerance LE 10 Penalty for deviation from pavate consumption Govemimentconsampuon Investments intermediate G OTTSul TID HOTT heciol income pis Bop GDP During data base loading you can use a facility to additionally filter out small values three options are available Filter method Rebalancing Simple deletion Mone Choosing the option None will simply load the data base as it is i e without filtering which is the way
102. tted data to remove small values The result from these step are stored in the data directory The GDX files generated from loadGTAPAgg gms provide the input to actual model runs with com gms The GAMS code initializes all variables to benchmark values and starts the solver That model start should basically need no iterations and lead to zero infeasibilities it should prove that the model s parameters are correctly set up to replicate the benchmark Next the user chosen shock file is read and changes to parameters introduced Afterwards the model is solved for the shock Post model processing While the processes described above are more or less identical to the original code of Dominique and Van der Mensbrugghe and Tom Rutherford the post model processing was added to allow using the exploitation part of GGIG The file report gms stores the results back into a SAM like structure e g consumer prices p results r P i hou dom p dc l i r The p results parameter is stored in a gdx file and later read by the interface for exploitation It is generally organized as follows e First dimension regions including a world aggregated stored under wor e Second dimension tables for values V quantities Q prices P tax rate T and tax income G e Third dimension commodities factors e Fourth dimension institutions sectors e Fifth dimension origins destinations e Sixth dimension base check or
103. tween domestic output and exports is infinite such that producer prices between the two destinations are equal and physical balancing is used The GAMS version allows alternatively using a CET structure where prices differs and consequently a non linear quantity aggregator is used That feature can be switched on by providing a parameter file in GAMS format where a CET elasticity different from infinity is set Numeraires e The residual region closes global savings see equations walraseq savfeq and yieq in model gms e The numeraire regions and products define a price indices For the model only the choice of the numeraire regions matter which enter the global factor price index where the numeraire products and regions together define a producer price index currently not used in the equation setup Closures General settings Model structure Closures Outputoptions Agorithm GAMS Closures Foreign savings Global equal returns to capital v E EP Factors with factors tax udates Final consumption Spending Exchange rate Monetary Union v As shown above the code supports alternative closures different from the ones used in the standard GTAP model e For foreign savings besides the default global bank mechanism which leads equal expected returns to capital across all regions the model can in run with a with allocation of investments and with fixed foreign savings These closures are only available in the global mo
104. uch as CAPRI http i doku php id start but also a range of other modeling system such as the Policy Evaluation Model PEM by the OECD a dynamic single farm model an Agent Based Model of structural change in agriculture or a recursive dynamic Hydro economic river basin model For those interested in a discussion about interfaces used with economic simulation models see for example https www gtap agecon purdue edu resources download 5864 pdf P rez Dom nguez et al 2012 or Britz et al 2015 The GUI generator is flexible enough to also host extensions as shown by the implementations of GTAP AEZ GTAP AGR and GTAP E A first version of the GUI was developed for a class at Bonn University Institute for Food and Resource Economics in the winter term 2013 2014 building on Tom Rutherford s GTAPinGAMS version The students had already some knowledge of GAMS and CGEs and wanted to use a global 9 CGE in a project A year later again a class of students wanted to use a global CGE which led to further improvements such as flexible aggregation of regions in the mapping viewer or the introduction of sensitivity analysis In 2014 when the GTAP Center decided to release a full fledged version of the standard GTAP model in GAMS it was decided to adjust the interface to operate with that version The further work on GTAP led to improvements to the GUI generator also to the benefit of other modeling systems applying the GUI The GUI itself is di
105. uch as TortoiseSVN http tortoisesvn net index de html freeware on your machine Afterwards make a checkout of the repository https svn1 agp uni bonn de svn Gtap8lInGams i e download the code from the server to your local machine using the user id gtapingams and the pass word gtapingams Using SVN will allow you to keep your local copy automatically updated to code improvements to check which changes you introduced in the code and to switch easily back to original versions of all or specific files if deemed useful if necessary on a line by line basis The basic steps for using SVN to install GTAPinGAMS based on TortoiseSVN are detailed below First navigate in the windows explorer to the disk directory where you want to install GTAP and open the context menu with a right mouse click Los Volume Ck gH Wa ag kp F runl ECSER Feigeben f r di bp z set agp pul 41 SVN Checkout Ina a pre Doug pul aa Terbanes Vi Ha Lay werk eapi fot Moercn an umm o uro m mom iicasaahacriallaz Chose SVN Checkout and copy the URL https svn1 agp uni bonn de svn Gtap8InGams in the first field as shown below and press OK 10 E Omit externals Revisi 9 HEAD revision Revisi lA Certificate validation failed Error validating server certificate for https svnl agp uni bonn de 443 Unknown certificate issuer Fingerprint 15 6F 60 C1 36 11 0A 14 23 B8 02 A3 2D 7 C 4E 5C 18 C3 A8 90 Disting
106. uished name ILR University of Bonn DE Do you want to proceed gt Accept the certificate permanently You won t get asked about this certificate again gt Accept the certificate The certificate is accepted only this one time chose Accept the certificate permanently In the next dialogue enter the username and password check Save authentication in the checkbox in the lower part of the dialog and press OK E https svn1 agp uni bonn de 4437 VisualSVN Server Requests a username and a password gtapingams The SVN client will next download the newest release of all files to your computer which might take some time 11 D Gtap8InGams gams build filter gms D Gtap8InGams gams Ibuild yemTinyValues gms D Gtap8InGams gams build joad_gtapAgg gms D Gtap8InGams gams conopt op2 D Gtap8InGams gams conopt op3 D Gtap8InGams gams COM_ GMS D Gtap8InGams gams conopt opt D Gtap8InGams gams postModel D Gtap8InGams gams postModel gis_set qms D Gtap8InGams gams postModel output_sets gms D GtapSInGams gams postModel calcExp gms D Gtap8InGams gams postModel OutputSam gms D Gtap8InGams gams postModel mapToResults gms D Gtap8InGams gams postModel map_regions gms D Gtap8InGams gams postModel jniAggregation gms D Gtap8InGams gams postModel outputToGui gms D Gtap8InGams gams sensAnalysis D Gtap8InGams gams sensAnalysishs_to_Gdx gms D Gtap8InGams gams util D
107. ults obtained with the default settings For small models and special applications such as sensitivity analysis it might however pay off to check if the overhead of using the pre solves can be avoided Table 2 Solution time with different data bases and model configurations CNS Data GTAP standard GTAP Standard GTAP Standard GTAP AGR GTAP AEZ base Fixed allocation Global bank mechanism Global bank mechanism of global savings 10x10 not possible 57x10 35 45 sec 1 min 30 sec 3 min 1 min 30 sec 11 min Even if additional tests showed that rather large models such as a 57x82 variant can be solved even without any filtering at least on some larger shocks in around ten minutes such tests failed with a model using the full resolution of the data base 57x140 but also with certain larger shocks on the 57x82 variant in any reasonable time The tests were only possible with the beta version of CONOPTA as CONOPTS will exceeds its internal memory limit of 8GB on very large models around 600 000 equations which implies a SAM with about the same number of entries That implies that users might run into very long solution times under a combination of larger shocks and very detailed data bases or might even not be able to load the model into the solver Analysis of changes in simulated welfare changes under a multi lateral trade liberalization scenario which dismantled 5096 of all import tariffs and export subsidies indicate that filterin
108. us jT 1 amet 142429 08 so00t be AGm243 150656 1 anaou arse 223047 41 wn se 1 47004 42 6674120 7007 00 146219 97 41 70 90820 20 32203 66 3587 55 n set il ssn 1386 0 38103 01 sssrin 170401 36 570002 81 37563538 227600 04 fd pet y 4f el caman 2090100 1981 17 4208 68 285874 XT 1287132 2179529 1668 24 x iru 103633 07 3T0 7 08 1157345 69387 02 IT 2793348 23904 37 10381 31 gt ii m 9463 66 125470 64 2954 26 150526 34 23275 16 tarsat 134733 28 149084 04 m B 4637 1a0 pa acte J 1292 Domestik Gar chutes ai moreris pre mao pwameter i 1273 reert purchases at agents mee xmmQ par aeter 3 1272 Door t purchases at market price xmargi pw setir amp amp AYTT foss for made m acroe all tranaport nodes rh por orsetrr D 100 c Dt exper parameter Double clicking on any on the symbols in the table on the RHS will open a new window as shown above The effect of using filtering and the rebalancing can be checked with the symbol itrlog which shows totals aggregated over commodities or sectors for each rebalancing step per country and trace at global level the number of nonzero at global level and related changes The example below shows the output from filtering a 32x34 case Filtering removes about 7696 of the original almost 135 000 non zero items see row Total The line curRelTol shows the applied threshold in the iteration which is increased stepwise to the desired maximum of 0 2596 Inspecting the relation between the dropped items
109. y for 3D non stacked E Draw outline 7 Draw shadow 7 Filled bars E Show mean median qi q3 Options for spider charts 5 Maximal number of axis Maximal number of series un Foreground transparency in T Es To Is T JU To 8 4 Filled shapes V Show last column Options for all charts Font size relative to tables in E Use shades of blue 4 Treat zeros as missing values F Include zero in value axis range V Show axis titles x _seteder _ The resulting scatter plots and histograms look as follows 60 4 gt Raper quen eoe En m ET nm E vem e t Eat al rid ren Total 5 ME ET Saati pore T Liz LA im im et iaa 4 ib f T4 Dre 3 Las Lea 4 ios ion z m nwm z nm eoe Uomo 4 m gm LE el s er TM NA naa 4 noc 558 Lir ne A i os Bs le T F LE dea 156 wed ien D SHEETS e 3 a E E t PoS ies 3 nx EN E eit T s et ai Ciel Liiki i a Pii E ad esie Braal use Lm 1 28 T a d mma nam nme E ET E B at 4 C o ae C ade a JF n7 i L gor n p p e E D os 4 E F ume cem nm ns m l Uum com nem nm x 3 4 nno eu num npa 120 171 AA UJ Dus cus G s Toc FT kOJ m aa a Total Limb ssl aul Haai Drala Bi Vliscu aul Bar ate Era mand Peral ati d at Lt T d f
110. y like to note that we did our best to test the combination of the GTAPinGAMS code and the GUI but that improvements are certainly possible First steps The installation comprises a set of results for testing the interface without the need of actually running GAMS code You can now already look at these results by selecting on the left hand side the work step Simulation GTAP in GAMS V8 GGIG GUI Agams File Utilities GUI Settings Help GTAP in GAMS V8 worksteps Scenario description Prepare data Simulation GTAP in GAMS V8 tasks Define scenario Simulation Scenario elements gams scen base_scenarios noShock gms Define basis scenario file gams scen base_scenarios yecDyn gms ILR Institute for Food and Resource Economics GTAP in GAMS V8 Ini file GtaplInGams ini User name undefined User type administrator You will see the interface for the Scenario editor which is discussed later We skip that functionality here and move directly to the task Simulation 15 Uutp optic GTAP in GAMS V8 General settings Sensitivity analysis Dam Regonal coverage Copie ues suroes sion cans GAMS Graphical RU r Aes Institute for Food and Resource Economics GTAP in cams V8_ Inifile GtapinGams ini_ Username undefined Usertype administrator Simulation Simulation and next pressi
111. ype E 4 Dim base check B je NL Table Levl ba check rs Fus I 7 base check shock F Y TE 10 49 10 49 10 46 a 0 0096 0 28 isr 0 12 0 12 0 12 0 0096 0 58 m 0 62 0 62 0 62 0 00 40 75 0 14 0 14 0 11 0 0096 1 0456 0 02 0 02 0 02 0 0096 4 1995 3 07 3 07 347 0 0096 3 186 ipn mus ae ald 0 0096 1 29 kor 0 71 0 71 0 73 0 0096 2 B206 idn 0 38 0 38 0 38 0 0096 0 10 mys 0 16 0 16 0 16 0 0096 0 8596 phl 0 12 0 12 0 12 0 00596 0 206 Trouble shooting with the viewer If you are not sure what controls are for try pressing the F1 button when the mouse is hovering over the control In most cases the PDF user guide will open at a page offering information Especially when working in the beginning with the viewer one often ends up with a table showing no longer any data an awkward selection an unsuitable pivot etc The following strategies can be used to overcome such problems EXE yy lt Handling New Data View to open a fresh one use the close button to remove the view and use View 2 Leave the viewer View Handling Exit and click on Remove view specific settings under Settings in the menu bar That will set the viewer back to the original defaults GTAP in GAMS V8 GG File Utilities GUI Help GTAP in GAMS Edit settings ti nun Load settings from ini file I GTAP in GAMS save current settings to ini fil

GTAP in GAMS with GUI for class room use, PDF

Contents

Download Pdf Manuals

Related Search

Related Contents