User Manual
Contents
1. options default squaring options menu item to be either more stable or more responsive In order to set N to the default value the user presses the Use Run Tests to Determine N button shown toward the top of the screen on the left side Simple Average The simple average of development factors for each period for the most recent N values the highlighted values in the age to age triangle Weighted Average The weighted average of development factors for each period for the most recent N values the highlighted values in the age to age triangle Regression Factor The projected future value of the age to age factor based on linear regression This regression method employs a stabilizing technique intended to avoid distortions caused by random fluctuations of data More information about this technique can be found at www cgconsult com docs FactorMethods pdf A key parameter in this method is Return to Table of Contents 15 the significance level which can be adjusted in the tools options default squaring options menu item Prior Selection If a previous analysis file was imported this row will be populated with the development factors from that previous analysis If a previous analysis file was not retrieved this row will be blank Selected Age to Age Factor In this row the user will select the incremental development factors The user can type values into the cells or double click the row title2. The next step is the opportunity to import premium or exposure information If you choose to import premium or exposure you will select the file containing the premium or exposure and if necessary assign the Time Period Identifier and Premium or Exposure Identifier If mapping is required you will be given a mapping summary and the opportunity to save the layout for future premium importation If you choose not to import premium or exposure a value of 1 will be used in each period for calculations This will be followed by the option to import a previous analysis file This is the analysis file corresponding to a previous valuation date Importing this information has a number of benefits e Allow you to start your analysis with previously selected factors loss ratios and selected ultimate losses e Allow you to compare actual loss development against previously projected loss development as well as to compare changes in selected ultimate losses If you chose to import a previous analysis file you will be able to browse for the file The user will then be asked to provide a title for your analysis After entering the title the setup for the analysis is complete Analysis Title Title for Analysis Cancel A user may edit the title at any time by simply double clicking the title shown on the screen 1 2 Opening an Analysis File When you start a new analysis by opening an analysis file you will simply open a previou
3. 066 007 9 287 547 9 569 609 1998 1 301 045 3 105 752 5200794 7 114 889 8399399 9 439 846 9 998 307 10 371 062 10 708 011 1999 1 460 089 32425648 5 561 016 7 219154 8 483 535 9 111 338 9 813 296 10 228 039 2000 1 615 641 3 562 864 5 565 190 7 232129 8 297 800 8 874 527 9474992 EEE 2001 1 756 010 3 708 660 5 786 013 7 344 399 8 461 272 9 511 124 2002 1 894 116 3 524 704 5 625 241 7 503 179 9 060 008 2003 1 705 261 3 564 037 5 664 483 7 508 130 2004 1 850 733 3 454 746 5 387 894 2005 1 966 886 3 603 477 2006 1 916 518 Age to Age Triangle 7to8 8to9 b 1997 4 044 1 024 1998 1 037 1 032 1999 1 042 2000 2001 2002 2003 2004 2005 Esetar Palaulatiane Tin Pind Aiad Atn amp Fink Rin Fina fin 9 atai te VII Status The factor calculations are shown below the triangle of age to age factors The rows highlighted in yellow contain the values that may be edited A description of each row is given below Return to Table of Contents 14 Cognalysis Reserving System Loss Development Factor Method Paid Loss File Edit View Graphs Tools Help www cgconsult com Aree Wee Industry H1 Current View res Incurred Incurred Methods moe Determine N B F Age to Age Triangle 8to3 A 1997 1 024 1998 1 032 1999 2000 2001 2002 2003 2004 2005 Factor Calculations 1to2 2to3 3to4 4to5 5to6 Eto 7 to8 Bto9g Ito 10 10to Ult gt Periods N Simple Aver
4. 463 7 121 583 2005 24 475 213 3 603 477 6 303 684 12 519 116 8 915 639 8 346 955 8 606 431 8 904 303 9 212 484 9 498 980 10 012 996 2006 27 496 114 1 916 518 4 048 494 13 323 594 11 407 076 10 292 590 10 790 758 11 372 557 11 985 725 12 565 841 13 630 599 Total 170 182 612 76 967 802 92 700 930 121 868 097 44 900 295 42 934 846 43 842 592 44 873 640 45 928 935 46 899 984 48 619 274 Status id start E Cognaiysis User Man i Cognalysis Reservi C e ts9eN Radio buttons are available above the table on the left to choose a ratio view or to annualize the values if a monthly or quarterly analysis was performed Return to Table of Contents 21 5 3 Projected Losses The Cognalysis Reserving System provides projections of both paid and incurred losses These projections e Tie to the selected accident year ultimate losses both for paid and incurred e Are based on weighted projections of intermediate period loss from the four basic methods Paid LDF Paid BF Incurred LDF Incurred BF e Rely on past variability by development period to reconcile between method projections and selected values Users can review the projected paid and incurred losses in the Projections screen by selecting the Projections button in the upper right area of the screen Cognalysis Reserving System Loss Projections File Edit View Graphs Tools Help www cgconsult com ld Industry H1 Current Yiew p Residual Incurred Incurred
5. 515 2 003 564 1 871 261 2015 683 354 3 874 509 2 429 175 1 985 660 2016 446157 4 048 494 2659861 2113132 1 420 259 v Status FASET g TY Cognalysis User Man Cognalysis Reservi gt 7 4 j 2 1 50PM On the left side of the screen the projected calendar year payments and incurred loss changes are shown for the number of periods in the future as was provided for history These values are the sum of the each new diagonal added to complete the squared triangle To the right of the calendar period projections the squared triangle is provided The projected payments and incurred losses are shown by incurred period and development period shown with aqua shading Return to Table of Contents 22 5 4 Residual Triangle It can be useful when reviewing reserve analyses to consider the residual errors observed from applying current parameters to past development Patterns observed in these residual triangles may suggest changes to selected parameters are needed The residual triangles can be viewed by selecting the Residual Triangle button in the upper right area of the screen The residuals are provided for each method and older ages can be viewed by using the scroll bar at the bottom of each method Cognalysis Reserving System Residual Error Triangles Actual Modeled File Edit view Graphs Tools Help www cgconsult com AZIEN Industry H1 Current View Paid LDF Paid B F oime Incurred Incurred Methods Pro
6. Add Report Name Add Curent Time Add Analysis Name Add Page Number Users can preview the reports by selecting Print Preview from the File menu On the Print Preview screen the user can adjust the view by selecting from the screen view buttons or page box at the top of the screen After previewing the reports the user can print the reports by selecting the printer button or they can close out of print preview by selecting the Close button 42 OO BB cse 4 17 2009 Industry H1 Current View Paid Loss Triangle Agel Age2 Aged Age5 Age Age Age Age Agel 1 094 575 2 536 829 4 415 384 5 996 458 7 341 758 2 250 911 8 687 864 9 066 007 9 287 547 9 569 600 1 301 045 3 105 752 5 200 794 7 114 889 8 399 399 9 439 846 9 998 307 10 371 062 10 708 011 1 460 089 3 242 648 5 561 016 7 219 154 8483535 9 111 338 9 813 298 10 228 099 1 615 641 3 562 864 5 565 190 7 232 129 8 297 800 8 874 527 9 474 992 1 756 010 3 708 660 5 786 013 7 344 399 8461 272 9 511 124 1 894 116 3 524 704 5 625 241 7 503 179 9 060 008 1 705 261 3 564 037 5 664 483 7 508 190 1 850 733 3 454 746 5 387 894 1 966 886 3 603 477 1 916 518 Cognalysis Reserving System Return to Table of Contents 31 5 9 Notes Users can add notes for documentation purposes on each section of the analysis except graphs If the user selects Notes from the View menu at the top of the screen a box will appear at the bottom of the screen for the user to type comments into this area
7. Incurred LOE Region Finally you will be prompted to map incurred losses or build them using case reserves paid Return to Table of Contents Input Field Mapping Incurred Loss or Case Reserves Select the Field to Map to Data Type Cumulative Incurred Loss Cumulative Incurred Loss Accident Period Age Incremental Incurred Loss Paid Incurred Case Reserve Balance This data type ts for total incurred losses from the beginning of the accident period to the valuation date or age Your options here include mapping a field to cumulative incurred losses incremental incurred losses or case reserve balances After mapping each of these elements you will be given a summary as shown below with the option to go back and remap as needed ce Mapping Summary Me A selected Mapping Summary Data Element Field Selected Incured Accident Period Accident Period Age Age Cumulative Paid Loss Paid Cumulative Incurred Loss Incurred lt Back Finish Lancel Upon completion of the mapping the option to save the layout is provided When a mapping is saved future data imports will automatically recognize if a data source file is of the same format and make the appropriate mappings Since data filtering and length of accident period could vary from analysis to analysis from within the same data these questions are asked regardless Return to Table of Contents 8 Layout Save
8. Methods Ch LDF B F Summary m Incremental Paid Loss in Period A Period 1 Period 2 Period 3 Period 4 Period 5 Period 6 Period 7 Period 8 Period 9 Period 10 1 094575 1 442 254 1 876 555 1 583074 1 345 300 909 153 436953 378 143 221 540 282062 1 301 045 1 804 707 2 095 042 1 914 095 1 284 510 1 040 447 558 461 372755 336 949 327508 1 460 089 1 782559 2318 368 1 858 138 1 264381 627 808 701 958 414 743 1 615 641 1 947 223 2 002326 1 666 939 1 065 671 576 727 600 465 1 756 010 1 952 650 2 077 353 1 558 386 1 116 873 1 049 852 1 994 116 1 630 588 2100537 1 877 938 1 556 829 1 705 261 1 858 776 2 100 446 1 843 647 1 950 733 1 604 013 1 933 148 1 966 886 1 636 591 230067 1 916 518 Incremental Incurred Loss in Period CalYyr Incurred Period 1 Period 2 Period 3 Period 4 Period 5 Period 6 Period 7 Period 8 Period 9 Period 10 b 2007 8 744 652 2810811 2094776 1 837 182 1121791 919213 445 131 176680 239170 189218 206 660 2008 6 291 882 3331 607 2504015 1 942171 1 421 088 751 913 486 222 462833 264887 211 341 2009 4 280 248 3131 986 2562860 2 063623 1 306512 694 307 442633 547 633 293 855 2010 2 918 205 3412438 2581511 1 619 651 1 397 435 896 753 125 770 325 447 2011 2087 945 3 669 308 2 766 810 1 895 125 1 309 580 697 379 635 073 2012 1 624 971 3 945 367 2 495 772 2 172 109 1 261 373 820148 2013 1 205 833 4106 387 2810 301 1 686 758 1 295 799 l 2014 883920 4 087
9. can also be retrieved by selecting CTRL N The notes section is shown on the portion of the screen view circled on the following page Cognalysis Reserving System Summary of Ultimates Under Different Methods File Edit View Graphs Tools Help www cgconsult com i 3 led Industry H1 Current View Residual Paid LDF Paid B F Triangles Tr r Summary of Ultimates Table Earned Premium Current Paid Current Incurred Paid LDF Incurred LDF Paid B F Incurred B F Wi 5 Li Previous Selected Loss Ratio b 1997 11 505 094 9 569609 10 040 632 9 885 406 10 392 054 9 924 181 10 417 365 10 392 054 10 119 157 90 3 1998 12384141 10 708 011 11 376 077 12 024 513 11972596 12 024 513 11 722 973 97 1 1999 11 356 614 10 228 039 11 043 409 11 893 751 11 698 141 11 893 751 11 566 520 104 7 2000 11 409 056 9 474992 10 359 005 11 446 701 11183823 11 446 701 11 217 518 100 3 2001 10 867 684 9 511124 10 973 275 12 575 373 11 961 495 12575373 12 012 991 115 7 2002 16 314 360 9 060 008 10 694 769 12 790 944 12810537 12 790 944 12 491 396 78 4 J 2003 20 293 316 7 508 130 9 899 245 12878918 12743826 13194737 12878918 12 698 686 63 5 2004 24 079 420 5 387 894 7 962 340 12 023 133 12 023 133 11 828 875 49 9 2005 24 475 213 3 603 477 6303684 12 219 391 12 519 116 12519116 12 518 539 51 2 2006 27 496 114 1 916 518 4 048494 12 959 495 13 323 594 13 323 594 48 5 Total 170 182612
10. 0 2001 ol o ol 2002 ol 2003 2004 2005 2006 2007 2008 EAk VEIR EAE o 0 od o o 0 Oo 0 0 0 0 0 0 Bd 0DdD Go o 0 0o O0o O 0B od o d o o 0 0 Oo d do o o aod Lo Cancel The user will then be prompted to enter incurred values and exposure or premium values Ed Data Edit E Data Edit Exposure Premium Exposure gt 7999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Oo 0 0 0 0 0 O 0 E DK Cancel The user may then enter a title for the analysis and optionally retrieve a prior analysis Return to Table of Contents 11 2 0 Editing Data Users have the ability to edit loss incurred or paid triangle or premium exposure values The user can choose the element to be edited from the Edit Data menu option at the top of the screen or by double clicking any number of the paid triangle on the Paid LDF screen any number of the incurred triangle on the Incurred LDF screen or any earned premium number on the Paid B F or Incurred B F screen Cognalysis Reserving System Summary of Ultimates Under Different Methods Edit View Graphs Tools Help 3 33 Copy Ctrl C Copy Table Ctrl T AL Paste Ctrl mann cgconsult com Residual Tr ES oe Summary of Ultimates Table A Paid Triangle Incurred Triangle Exposure Optimal Earned Premium Current Paid Current Incurred Paid LDF Incurred LDF Pa
11. 1999 426 988 621 599 238 016 203 900 92 258 213215 82 766 1999 718546 606615 319804 76 462 101 45 252543 72 537 2000 670 926 404 784 329 866 67 207 91 809 140 304 2000 845 053 247 849 468 430 315 039 195 475 47 614 2001 808 805 645 601 360072 222 028 450 695 l 2001 1 210 558 685686 476 985 176 034 347 166 I 2002 85 594 48 861 78985 213 455 2002 159 470 326 430 11 451 37 487 2003 238 356 524 535 353 348 2003 42939 567 302 230 685 i 2004 802 966 1 079 669 2004 1 271 237 715 832 2005 898 18 2005 747 708 2006 2006 l p ste vi Status t Start TY Cognalysis User Man Cognalysis Reservi Search Desktop 2 1 50 PM Return to Table of Contents 23 5 5 Changes in Ultimate Loss Estimates and Actual vs Projected The change in estimated ultimate and the actual paid and incurred loss compared to expected can be viewed by selecting the Changes button in the upper right area of the screen If a previous analysis is retrieved at the start of the analysis for all but the most recent year change in selected ultimate is shown by period Also based on the projected payments and incurred losses from the previous analysis the variance of actual versus projected is provided If a previous analysis is not retrieved the prior estimated ultimate and projected losses will be zero Cognalysis Reserving System Actual vs Projected Incremental Loss C
12. 74 2 290 256 Incurred LDF 0 703 1 000 0 520 0 708 1 547 824 J Paid BF 0 358 0 520 1 000 0 918 5 604 829 v Status The four common methods discussed above the Paid Loss Development Incurred Loss Development Paid Bornhuetter Ferguson and Incurred Bornhuetter Ferguson methods are used to determine the optimal weighted estimated ultimate Given the standard deviation of the reserve using each method and the correlations matrix for these methods both shown at the bottom of the screen weights can be found that minimize the variance of the combined estimate The result is the first row of the Weights table These weights have the potential Return to Table of Contents 20 to be negative for individual methods While negative factors are an interesting result that bears some consideration the Cognalysis Reserving System does not allow them for the minimum variance weighting it ultimately uses The remaining 14 rows show the systematic determination of the minimum variance non negative weighting Rows 2 5 eliminate one of the methods each and optimizes Rows 6 11 eliminate two of the methods each and optimizes Rows 12 15 eliminate three of the methods each and optimizes The weighting that is used is the one that minimizes the variance among those with only non negative values For more information about this technique please see www cgconsult com docs MinVarianceReserves pdf 5 2 Summary of Results Users can review the selected ul
13. 76 967 802 92700930 5 eee 71 6 lt This is the notes section Status Return to Table of Contents 32 6 0 Batch Processing In the Cognalysis Reserving System routine triangle processing and report production for a large number of analyses may be done automatically in batch mode This can provide a useful first draft analysis for large amounts of data providing you information about where there may be issues for more focused analysis In order to build a batch process the user selects the Batch option from the Tools menu at the top of the screen Cognalysis Reserving System Paid LDF Paid BF oxi Resuts Sanne rs Status K ee S m T Cognalysis User Man Cognalysis Reserv 7 amp m Return to Table of Contents 33 After selecting Batch the user will be taken to a screen where an existing batch process can be opened or anew batch process can be created by selecting one of these options from the File menu at the top of the Batch screen Once a batch process is created or edited the user can save this process by selecting Save from the File menu to be retrieved for future batch imports EEk Loss File Path Loss Table Name Loss Filter Period Premium File Path Premium Table Premium Filter Analysis Title Prior File Path Save File Path Print Reports 3 Cancel T Start Ly Cognalysist Reservi ES Batch T Cognalysis User Man In building the ba
14. Cognalysis Reserving System User Manual Table of Contents SO ARN an EE SN EN EN 3 AES REE EEE ene saeeauereastnesenect 3 TINN Nr 9 VTS 10 EE EEE EN 12 3 0 Reviewing and Adjusting Methods ccssssescesssesssecesesssnecescesssnsnecesceeeaeeceseseeneseecessenaeesescseneeseecenessaeeeeass 14 EE ERR EEE EE 14 5 a rii GE EEE re ee ee ere 16 oe EE EEE EE 16 TNT 18 4 0 Selecting Ultimates NG 19 5 0 Reviewing EE ELG ee EE 20 ES Ge EEE EE NE 20 ENN NN 21 SFT NN 22 JR Te EE EE 23 5 5 Changes in Ultimates and Actual vs Projected ssrrrrnanvvonnnvnrnanraverennverennrensnnrensnanenenennnenenssensnsne 24 SNE 24 SJT EEE AAEE E E E E N E E E A EEE E AE AE E ET 25 NNN EMNE a Oe JE EEE EE gt 7 0 User FRE see DO ee VEGG ee EN EE OT TN 39 Return to Table of Contents 2 1 0 Starting an Analysis Starting or Opening an analysis can be done in one of three ways 1 Creating a new analysis file by pulling data from a file 2 Opening an existing analysis file 3 Creating triangles by entering data manually or by cutting amp pasting The user begins the analysis by selecting Open from the File menu at the top of the screen or by choosing the folder icon on the toolbar below the menus at the top of the screen Upon selecting open the user can choose from these three options Open Options Select Opening Option Open Data Files Start an Analysis O Open an Analysis File Return t
15. System Summary of Ultimates Under Different Methods File Edit View Graphs Tools Help www cgconsult com 13 fed Sample Analysis Residual Paid LDF Paid B F Resuts Triangles all i Summary of Ultimates Table Optimal Earned Premium Current Paid Current Incurred Paid LDF Incurred LDF Paid B F Incurred B F Weighting Previous Selected Loss Ratio gt 20081 158 492 9123 20 111 a 2008 2 155 398 4 606 15 636 2008 3 166 566 3 303 11 355 2008 4 159 326 1 762 6 747 Total 639 782 18 794 53 849 28 7 30 7 29 5 31 3 30 0 49 510 198 863 192 043 Summary of Standard Deviations Table Paid LDF Incurred LDF Paid B F Incurred B F gt 2008 1 2 254 255 1 873 607 682 2008 2 2 380 268 1 800 896 727 2008 3 2 501 306 2 209 487 1 678 2008 4 2577 331 2 745 279 2 296 Perfectly Independent 4 862 583 4 378 586 3 013 Perfectly Correlated 9 712 1 160 8 629 269 5 383 Partially Correlated 4 862 875 6 360 828 4 329 Status sa start E Coonatysis User Man 14 Cognalysis Reservi E For each period the current data is shown Earned Premium Paid and Incurred The indicated ultimate from each of the four methods as well as a blended estimate intended to minimize variance optimal weighting is also shown If a previous analysis file was retrieved the previous ultimates will be shown otherwise the previous column will contain zeroes The user can type in a selected ultimate value double click on a sing
16. age 1 994 1 625 1 322 1 181 1 103 1 064 1 041 1 028 1 030 1 033 Weighted Average 1 985 1 618 1 320 1 180 1 103 1 064 1 041 1 029 1 030 1 033 Regression Factor 1 957 1 593 1 319 1 180 1 103 1 064 1 041 1 029 1 030 1 033 Prior Selection 2 027 1 616 1 321 1 176 1 098 1 063 1 040 1 024 1 035 0 000 Selected Age to Age Factor Age to Ultimate 6 762 3 391 2 087 1 579 1 337 1 212 1 139 1 094 1 064 1 033 Age to Ultimate St Dev 0 687 0 217 0 089 0 045 0 026 0 010 0 005 0 005 0 004 0 004 Estimated Ultimates 12 959 495 12 219 391 11 244535 11 855 337 12113231 11 527 482 10 792 016 11 189 475 11 393 324 9 885 406 Ultimate St Dev 1 316 298 781 991 477 334 335 171 238 348 92 434 43 377 55 622 41 237 35 614 E Status sg Start T Cognalysis User Man Cognalysis Reservi J 2 2 1 47 PM Periods N This value can be typed in by the user or it can be set to a default which employs run tests to determine the length of history to use for factor averages Arun test is performed using all of the available data points where a run is defined as a series of factors either above or below the median value If an insufficient number of runs are observed given total number of data points and the significance level of the test then the earliest factor is dropped This procedure is repeated until the number of runs is sufficient given the remaining number of data points and the significance level You can modify the significance level in the tools
17. alue that combines methods to minimize variance will be selected Development Factors Users may choose from the following four options for default development factors e Simple Average Factors The starting selected development factors default to simple averages of historical factors Return to Table of Contents 37 e Weighted Average Factors The starting selected development factors defaults to loss weighted averages of historical factors e Regression Factors The starting selected factors are based on a regression analysis of historical development with stability adjustments see Statistical Thresholds section below e Previously Selected Factors The starting selected factors will default to the development factors from the previous analysis that has been imported if available Seed Loss Ratios Users may choose from the following two options for default development factors e Dynamic Determination of Loss Ratios The seed loss ratios are calculated using a moving average of the paid and incurred link ratio methods A radius number of periods before and after the period is derived with the goal of determining seed loss ratios that minimize the variance of the reserve estimate e Previously Selected Loss Ratios The seed loss ratios will default to the seed loss ratios from the previous analysis that has been imported if available Statistical Thresholds Some additional parameters guide behavior of particul
18. ar algorithms within the Cognalysis Reserving System e Critical Value for Correlation T test The measured correlations between accident periods are subject to a test for significance of the relationship against a null hypothesis of zero correlation If the test finds the measured relationship insignificant zero correlation is used The default critical value is 1 meaning t test of values greater than 1 will be seen as significant You can lower this number to be more sensitive to measured correlation between accident periods or raise it to be more stringent with regard to reacting to these measured correlations e Run Test for Significance The default number of periods used in the determination of weighted and simple average development factors is based on run tests to determine whether there are changes in development patterns over time The sensitivity of these run tests is governed by this parameter which is the significance level of the test The default is 5 which means that 50 of the time the test will reject the null hypothesis no change over time in development factors even when it is true You can adjust this parameter lower to be less reactive to changes in development patterns resulting in longer periods used for the factor averages or you can raise the parameter to be more reactive to changes in development patterns resulting in shorter periods used for the factor averages e Age to Age Factor Slope Significance The reg
19. curred link ratio estimated ultimate losses for the entire period specified by the radius The default value for the radius is determined as that radius which minimizes the variance of the total reserve estimate The outcome for this process will change each time new loss development factors are selected The selected seed loss ratios will only update if you press the Select Seed Loss Ratios Dynamically button Estimated Ultimates This row provides the estimated ultimate given the seed loss ratio The estimated ultimate is calculated for each age as 1 Percent of Ultimate Premium Seed Loss Ratio Cumulative Loss St Dev Factor This is the factor that when applied to the seed ultimate premium seed loss ratio gives the estimated standard deviation for the ultimate loss Ultimate St Dev This row provides the estimated standard deviation of the ultimate loss 3 4 Incurred B F Users can review and adjust the Incurred Bornhuetter Ferguson method by selecting the Incurred B F button in the upper right part of the screen The layout and functionality of this section is exactly like that of the Paid B F section Please refer to the Paid B F section above for further explanations Return to Table of Contents 18 4 0 Selecting Ultimates Users can review indications and select ultimates in the Methods Summary section shown below by selecting the Methods Summary button in the upper right part of the screen Cognalysis Reserving
20. dent Date Paid Incurred LOB Region This data type is for pre aggregated data Acceptable formats are vy The Accident Period can be determined in one of two ways It can be pre identified on the source data or it can be based on an accident date If the period is on the source data itself it must be in acceptable format YYYY YYYYQ or YYYYMM If accident date is to be used to build the accident periods you will be asked to identify the length of the periods annual quarterly or monthly You will next be asked to map Development Period Return to Table of Contents 6 Input Field Mapping Development Period Data Type 2 Age Valuation Date This data type 15 the age of the incurred loss in periode comesponding to the size of the accident Incured period Le 1 2 3 etc Select the Field to Map to Age Accident Period Age Paid Incurred LOB Region Similar to accident period development period can either exist already on the data source or it can be determined using valuation date Next you will be prompted to map paid losses These can be incremental amounts or cumulative amounts Input Field Mapping Paid Loss Data Type Cumulative Paid Loss Incremental Paid Loss This data type ts for total paid losses from the beginning of the accident period to the valuation date or age Select the Field to Map to Cumulative Paid Loss Accident Period Fn
21. e screen When building or modifying a batch process the user can use the Edit options from the menu at the top of the screen to target certain rows or cells These Edit options are also available by right clicking on a row in the Batch Analysis list Rows gt Copy Paste Batch Analysi Loss File Path Loss Table Name Loss Filter Period Premium File Path Premium Table Premium Filter Analysis Title Prior File Path Save File Path Print Reports gt e a Start 45 Cognalysis Reservi ES Batch Tt Cognalysis User Man Return to Table of Contents 35 7 0 User Preferences At any time users can define default squaring options or the default file directories Users can make changes to these preferences by selecting one of these options from the Options menu under the Tools drop down menu at the top of the screen Copnalysis Reserving System Default Directory Paid LDF Paid BF ge Status Search Desktop Return to Table of Contents 36 7 1 Default Squaring Options The Cognalysis Reserving System offers a variety of default options when you begin to square triangles By selecting Default Squaring options from the Tools Options menu choices for default calculations are available to the user Each of the default options is described below Default Calculations Initial Selections Statistical Thresholds Ultimates No Automatic Selection Minimum Variance Weighted A
22. ed Loss Ratios Rescaling would adjust the seed loss ratio to be consistent with the same level of losses before the change in premium or exposure Return to Table of Contents 13 3 0 Reviewing and Adjusting Methods Cognalysis contains four standard techniques commonly used for analyzing ultimates Those methods are the Paid Loss Development Method the Incurred Loss Development Method the Paid Bornhuetter Ferguson Method and the Incurred Bornhuetter Ferguson Method Each of the methods as presented in the model is described below 3 1 Paid LDF Users can review and adjust the Paid Loss Development Factor method by selecting the Paid LDF button in the upper right part of the screen The paid loss triangle will be shown as entered or imported Double clicking on a cell in the triangle will allow the user to edit the values in the triangle Below the triangle of loss values the triangle of age to age development factors is shown The highlighted values in the age to age factor triangle are the factors used in the weighted and simple averages Cognalysis Reserving System Loss Development Factor Method Paid Loss File Edit View Graphs Tools Help www cgconsult com 3 fed Industry H1 Current View co KE neces Incurred Incurred Methods SE Determine N B F change Paid Loss Triangle Agel Age2 Age3 Age4 Age5 Ageb Age Ages Aged Age10 n 1997 1 094 575 2 536 829 4 413 384 5 996 458 7 341 758 8 250 911 8 687 864 9
23. er Fields Peiod WE 8 0St t OSOSCS S CCCS Age ae Region Faid Incurred Once the fields for filtering have been selected and the user has selected the OK button the user will be given the opportunity to select the values within the selected fields to include in the analysis In the example shown on the next page Auto has been selected for inclusion from the list of unique values in the data file in the LOB field Subsequently the West Region is shown as being selected for inclusion from the list of unique values in the data file in the Region field Return to Table of Contents 4 ES Selected Values for LOB Available Values Data Selection Values GL Auto Wi Prop GE Selected Values for Region Available Values Data Selection Values West Return to Table of Contents If you choose not to filter data or after you filter data you will be taken through a series of screens to assign the fields of data to various data elements this may occur automatically if the field structure and names are identical to a previous import In the screen shown below we are given the list of fields in the data file that may be chosen to map to Accident Period Input Field Mapping Incurred Accident Period Select the Field to Map to Data Type Incurred Accident Period 8 Incured Accident Period lAccident Period Age Incured Acci
24. ewed based on the selected ultimate and projected losses The historical portion of each line is black while the projected portion is red The user can select a specific period to be highlighted bold from the drop down menu on the right The up and down arrow keys on the keyboard will also scroll the user through the different accident periods Incurred Development Like the paid development graph the incurred development graph shows the reported as percent of ultimate at each age for the periods reviewed based on the selected ultimate and projected losses The historical portion of each line is black while the projected portion is red The user can select a specific period to be highlighted bold from the drop down menu on the right Projection and Revision This graph shown on the next page provides a stacked bar chart that splits the ultimate loss for each period into each of its components paid loss case reserve and IBNR with the total bar reflecting the ultimate If a previous analysis has been retrieved the prior selected ultimate is also shown as well as projected values for paid case and IBNR reserves to the current point in time If the user has performed a quarterly or monthly analysis double clicking on a bar will show the quarterly or monthly view for that period Once in this view the user can return to an annual view by clicking the Back button in the lower right area of the screen Return to Table of Conten
25. for one of the averages shown above to select that set of averages The user can also double click on any factor within the age to age factor triangle to select that factor for the corresponding period Age to Ultimate This row shows the accumulated selected development factors Age to Ultimate St Dev This is the estimated factor to provide the standard deviation of ultimate losses when multiplied by the current loss amount Estimated Ultimates The estimated ultimate given the current valuation and selected development factors are presented in this row Ultimate St Dev This row provides the estimated standard deviation of the ultimate loss 3 2 Incurred LDF Users can review and adjust the Incurred Loss Development Factor method by selecting the Incurred LDF button in the upper right part of the screen The layout and functionality of this section is exactly like that of the Paid LDF section Please refer to the Paid LDF section above for further explanations 3 3 Paid B F Users can review and adjust the Paid Bornhuetter Ferguson method by selecting the Paid B F button in the upper right part of the screen As inthe Paid LDF section the Paid Loss Triangle is shown at the top of this section Below the triangle the B F calculations are listed Each row in this calculation area is described below Return to Table of Contents 16 Cognalysis Reserving System Bornhuetter Ferguson Method Paid Loss File Edit View Graphs Tools Hel
26. hange to Selected Ultimates Tools Help File Edit view i 3 bd Graphs mann cgconsult com Industry H1 Current View j Residual Incurred Incurred Methods Projecti LDF B F Summary rojections Prior Current Estimated Estimated Change to Estimated Incurred Variance Actual Projected ai Incurred Projected Actual Paid Paid Variance Incurred 1997 1998 1999 2000 2001 2002 2003 2004 2005 TOTAL Ultimate 11 722 973 11 566 520 11 217 518 12 012 991 12 491 396 12 698 686 11 828 875 12 518 539 106 176 655 Ultimate 10 392 054 12 024 513 11 893 751 11 446 701 12 575 373 12 790 944 12 878 918 12 023 133 12 519 116 108 544 503 Ultimate 272 897 301 540 327 231 229 183 562 382 299 546 194258 577 2 367 848 180 232 831 610 255 218 522 486 701 382 1 319 420 1 600 064 2 114 772 2 265 407 1 839 021 11 449 380 282 062 336 949 414 743 600 465 1 049 852 1 556 829 1 843 647 1 933 148 1 636 591 9 654 286 549 548 81 731 107 743 100 317 269 568 43 235 271 125 332 259 202 430 1 795 094 285 185 223 295 268 739 391 309 403 190 878 841 1 436 775 1 924 781 2 572 674 8 384 789 206 660 211 341 293 855 325 447 635 073 820 148 1 295 799 1 871 261 2 429 175 8 088 759 78 525 11 354 25 116 65 862 231 883 58 693 140 976 53520 143 499 296 030 Status ig Start g Rm 2 1 50PM w Cog
27. hoosing to view this graph for all periods or for an individual period by choosing from the drop down menu on the right or by scrolling through by using the up and down arrow keys Copnalysis Reserying System Ultimate Probability Density Function File Edit View Graphs Tools Help www cgconsult com AeA NE C Industry H1 Current Yiew Paid LDF Paid BF Optimize Results ae Incurred Incurred Methods Proiecti LDF B F Summary east Probability Density Function for Ultimates in Total g Print Graph amm Range of Predicted Outcomes 95th tile 75th tile Mean 25th tile Sth tile I I R 10 000 000 30 000 000 50 000 000 70 000 000 90 000 000 110 000 000 130 000 000 Ultimate Loss Status Return to Table of Contents 28 Paid LDF Perspective This graph shows the observed paid development factors for each period for a given age to age relative to averages prior and selected values The user can select the desired age to view from the drop down menu to the right Graphing can become useless when outlier values change the scale The user may also exclude or include such outliers by checking the box to the right The shaded gray region that highlights the factors that were discarded in order to calculate the simple and weighted averages see section 4 1 The line representing this cutoff may be moved to the left or right by clicking and dragging The
28. id B F Incurred B F Weighting Previous Selected Loss Ratio b 1997 11 506 094 9 569 609 10 040 632 9 885 406 10 392 054 9 924 181 10 417 365 10 392 054 10 119 157 90 3 1998 12 384 141 10 708 011 11 376 077 0 202451 MN 11 972596 12 024 513 11 722 973 97 1 1999 11 356 614 10 228 039 11 043 409 11 893 751 11 698 141 11 893 751 11 566 520 104 7 2000 11 409 056 9 474 992 10 359 005 11 446 701 11 183 823 11 446 701 11 217 518 100 3 2001 10 857 684 9 511124 10 973 275 12 575 373 11 961 495 12575373 12 012 991 115 7 2002 16 314 950 9 060 008 10 694 769 12 790 944 12610537 12 790 944 12491 3 78 4 2003 20 293 316 7 508 130 3899 2451 ee 12678918 12 743 826 13194737 12878918 12 698 686 63 5 2004 24 079 420 5 387 894 7 962 340 MG 12 023 133 12 023 133 11 828 875 49 9 2005 24 475 213 3 603 477 6 303 684 12 219 391 12 519 116 12519116 12518539 51 2 2006 27 496 114 1 916 518 4048494 12959495 13 323 594 13 323 594 48 5 Total 170 182 612 76 967 802 92 700 930 MGM 121 868 097 123 538 835 126 554 869 121 868 097 71 6 Summary of Standard Deviations T able Paid LDF Incurred LDF Paid B F Incurred B F b 1997 35 614 27 983 39 246 56 048 1998 41 237 32 371 48 325 57 781 1999 55 622 32 240 77 011 176 469 2000 43 377 34 343 120 711 91 297 2001 92 434 135 064 168 908 161 961 2002 238 348 254 917 454 883 506 092 mann mar 4 94 4904 ea 4 nn mm sm ram Status After selecting the data element to edit the user w
29. ile also presenting the values at various percentiles The user can choose to view this graph from a reserve or an ultimate perspective by selecting one of those two options from the drop down menu to the right of the graph If the user has performed a quarterly or monthly analysis double clicking on a box will show the quarterly or monthly view for that period Once in this view the user can return to an annual view by clicking on the Back button in the lower right area of the screen Cognalysis Reserving System Ultimates Box amp Whisker File Edit View Graphs Tools Help www caconsult com a A Industry H1 Current View Paid LDF Paid BF Optimize Results S re Incurred Incurred Methods Proiecti LDF B F Summary reo Box and Whisker Ultimates il Print Graph Perspective Ultimates i Ultimate Loss 95th tile 75th tile Mean 25th tile Sth tile EE 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Incurred Period Status Start Tj Cognalysis User Mar Cognalysis Reservi Return to Table of Contents 27 Range and Methods This graph shows a probability density function for the ultimates This probability density function is shown in relation to premium current paid and current incurred values Box and whisker graphs for each of the four methods are also added to give the user perspective relative to the method indications The user has the option of c
30. iles Changes to the system files directory are not recommended E Cognalysis Reserving System E EX File Edit View Graphs Tools Help www cgconsult com System Analysis Paid LDF Paid B F oxic Bess Sanne Search Desktop re gt lt i amp MEET a eee eee 39
31. ill be given a data edit screen Users can manually type in new values or paste values from clipboard data copied from other sources After all values are edited the user selects the OK button in the lower right corner to save changes and return to Cognalysis The Data Edit screen is shown below Return to Table of Contents 12 Data Edit Data Edit Cumulative Paid Loss gt 4997 1998 1999 2000 2001 2002 2003 2004 2005 2006 11 892 202 14 523 254 15 114 627 16 825 284 19 545 533 17 730 257 13 042 697 20 131 964 21 753138 20 346 006 Agel Age 15 172 357 18 634 741 19 253 876 22 973 633 26 366 132 23 290 491 24 530 848 26 390 377 29 443 606 ged ts Aged 15 936 483 16 204 095 19 457 421 19893623 20 087 639 20 516 644 24 021 935 24 455 194 27 540 141 28 087 625 24 239 259 24 792 438 25 544 817 26 020 530 27 415 363 Age5 16 421 435 20 123 469 20 723 327 24 685 311 28 357 433 25 002 346 28 489 943 8 Ages a Age 16 526 157 16 237 245 20 246 762 20 305 218 20 816 622 20 863 035 24 768 341 24 842 768 Agel Aged AgelD 46272511 16 291 290 16 305 163 20 332618 20 352 138 20 895 313 OK Cancel In the case where premium or exposure is being edited you will be prompted with a question regarding rescaling of loss ratios Cognalysis Reserving System Rescale Selected Se
32. jecti Ch LDF BF Summary rojections anges Paid LDF Residuals Incurred LDF Residuals A Agel Age2 Age3 Age4 Age5 Ageb Age Agel Age2 Age3 Age4 Age5 Ageb Age 354246 291 037 161 964 259941 152952 91 105 21 9411 32 308 248 073 291 922 36 205 227 132 58645 164 789 5 375 511468 153947 239439 3 285 175 309 45 689 37176 46 559 315 149 103950 168 963 57 589 48 387 76 663 18 509 331231 291 713 132509 42 285 246 001 118932 12 398 505145 269 747 57 398 103411 13224 170 162 14367 341 276 224 464 125 052 243 344 277 946 32 495 339539 268 443 171 645 103 782 310173 45 795 207 176 240 560 304710 212 463 178 341 366 075 132 252 31 750 151 013 180 192 252163 102 403 66 610 198 754 96 334 143 150 125 360 48 819 163 747 127 077 19 683 112 405 491 999 89 356 235 616 226 068 681 933 47 429 318 494 116 377 Paid Bornhuetter Ferguson Residuals Incurred Bornhuetter Ferguson Residuals Agel Age2 Age3 Aged Age5 Ageb Age Age8 Agel Age2 Age3 Age4 Age5 Ageb Age Ages gt 1997 196 560 162233 123 303 71 058 55 965 150 307 20750 66550 1997 121 298 96 484 63 808 176 928 35 157 177 892 26 759 85 1938 180 843 62458 212910 14145 189855 27012 24925 49734 1998 294684 196 800 239 09 11 787 77 496 109340 233 2354
33. le cell for a given period to use that value as the ultimate for that period or double click on a column to use that method as the ultimate for all periods The implied loss ratio is supplied in the last column If Premium was not entered into the model a value of one is used for loss ratio calculations Individual cells for various indications that highlighted in Deep Sky Blue are more than 5 below the selected ultimate for that period while cells highlighted in Tomato Red are more than 5 higher than the selected ultimate for that period Below the Summary of Ultimates Table the standard deviations of reserves by period are shown for each of the four methods Standard deviations for all years for each method are also provided assuming three different scenarios of correlations between years perfectly independent perfectly correlated and partially correlated Return to Table of Contents 19 5 0 Reviewing Diagnostics and Reports Cognalysis offers a variety of diagnostics and reports to help the user refine the analysis being performed The diagnostics include graphs and summaries of the optimization routine results projections residuals changes and reserve percentiles A description of each of the available options is described below 5 1 Optimization Metrics The Cognalysis Reserving System provides valuable insight into the different levels of variability of reserve methods arising from using each of the four basic methods of loss de
34. lines representing the simple and moving average factors will adjust accordingly including on the LDF screen but the selected factor and associated line will remain unchanged The selected factor line may also be moved up or down to a new selected value NOTE This will change the selected factor on the LDF pages as well as indications of ultimate loss projections etc i Copnalysis Reserving System Paid LDF Perspective File Edit View Graphs Tools Help www cgconsult com Hen NE C eee nnn Industry H1 Current View A Fy Residual y Incurred Incurred Methods Projecti Ch 44 LDF B F Summa rojections anges Paid Loss Development Factor Perspective Print Graph Age 1 to Age 2 Development Factor o Exclude Outliers from Graph 24 Discarded Points Selected Factor 23 gt OE Prior Selection Weighted Average Simple Average 22 ha Regression Factor a Observation 21 1997 1998 1999 2000 2001 2002 2003 2004 2005 Period Status start Ti Cognalysis User Mari Cognalysis Reservi Incurred LDF Perspective Similar to Paid LDF Perspective but for Incurred factors Return to Table of Contents 29 5 8 Reports Reports can be printed for each of the main screens and graphs discussed above To select reports for printing choose Print Setup from the File menu at the top of the screen The box shown below will appear for the user to check the boxes of the reports t
35. mium The earned premium or exposure that the user imported into the system will be shown in this row If the user imported or entered a zero value for premium or exposure this row will show zeroes but will use a value of 0 001 for calculations The user may edit premium or exposure by double clicking any premium cell on this page Percent of Ultimate This row shows estimated percent of ultimate based on the selected age to ultimate factor in the Paid LDF section of the analysis The Age 1 percent of ultimate will be equal to 1 Age 1 Age to Ultimate value Prior Seed Loss Ratio If a prior analysis file was retrieved this row will show the seed loss ratio used in the prior analysis If a prior analysis file was not retrieved this row will be blank Selected Seed Loss Ratio The user has the option of typing or copying in seed loss ratios or using the Select Seed Loss Ratios Dynamically option If the user chooses the Select Seed Loss Ratios Dynamically button located in the upper left quadrant of the screen the user will be prompted to enter a radius as shown in the print screen on the next page Return to Table of Contents 17 Radius Enter radius This radius is defined as the number of periods before and after the evaluated period to be used in estimation of the seed loss ratio for that period The Cognalysis Reserving System will use the minimum variance weighting of paid link ratio and in
36. nalysis User Man Cognalysis Reservi 5 6 Reserve Percentiles As can be seen on all screen views in the user manual the reserve percentiles for all periods combined is provided in the upper right corner of the analysis at all times This summary contains the mean selected reserve the optimal indicated reserve and the reserve at 6 different percentiles This gives the user a status of the analysis that can be viewed at all times regardless of which specific screen the user may be viewing 24 Return to Table of Contents 5 7 Graphs Eight main categories of graphs are available for the user to view and print The graphs can be accessed at any time by selecting the desired graph option from the Graph menu at the top of the screen A description of each graph type is provided below i Copnalysis Reserving System Paid Development SAA File Edit View Graphs Tools Help vwww cgconsult com BAA Indu stry H1 Incurred Development Projection and Revision Seed Loss Ratios Ultimate Box Whisker Incurred Incurred Methods Project i LDF B F Summary kr dr E Paid Development Range and Methods Paid LDF Perspective Incurred LDF Perspective Percent of Ultimate 100 History Projected Status Start T Cognalysis User Mari Paid Development This graph shown above shows the paid as a percent of ultimate at each age for the periods revi
37. o an Analysis gt Create Triangles Enter Data Manually 1 1 Opening a Data File The Cognalysis Reserving System allows for a wide variety of layouts for input data When you start a new analysis by opening a data file you will be able to browse for the file containing loss information Currently supported file types are Comma Separated Variable csv and Microsoft Access Database mdb If the input file is an Access Database file you will also be prompted to select the table containing the loss data After selecting the appropriate file you will be given the option to filter the data Return to Table of Contents Filters Would you like bo filter data This allows you to import a subset of records from your data source In this way the data source can be built in such a way that it includes data for multiple analyses If you choose to filter data select the fields you are interested in filtering by highlighting a field and clicking the right single arrow Selecting the right double arrow will select all fields To remove a single field highlight the field on the right and hit the left single arrow the left double arrow will move all the fields back to the Available Fields box In the example shown below the Line of Business LOB and Region fields have been selected for filtering and have been moved from the Available Fields box to the Data Filter Fields box FE Fields for Data Filter Meld Available Fields Data Filt
38. o be printed The user may also choose to print any notes added to the analysis by checking the appropriate box Print Selections Reports Results Summary Comparison to Prior Analysis Reserve Range Exhibit Indicated and Selected Estimates of Ultimate Loss Development of Chain Ladder Indication Paid Development of Chain Ladder Indication Incurred Development of Bornhuetter Ferguson Indication Paid Development of Bornhuetter Fergueon Indication Incurred Paid Loss Triangle Incurred Loss Trangle Graph Paid Development Graph Incurred Development Graph Projection and Revision Graph Seed Loss Ratios Graph Ultimate Box whisker Graph Range and Methods Graph Paid LOF Perspective Graph Incured LOF Perspective Options Include Notes Header Footer The user may also add a Header and or Footer to the printed reports After selecting the Header Footer button from the Print Setup box the user may type text in any of the header or footer boxes Alternatively the user can click in any header or footer box and then select the item they wish to have appear on the report The options for the header and footer are shown below and include Date Time Analysis Name Report Name and Page Number Return to Table of Contents 30 Modify Header or Footer Header i Date Report Footer Center amp Time amp Page Analysis Add Current Date
39. p www cgconsult com ae MEN Industry H1 Current View EE je Incurred Incurred Methods eee Paid Loss Triangle Agel Age2 Age3 Age4 Age5 Ageb Age Ages Aged Age10 gt 1997 1 094 575 2 536 829 4413384 5 996 458 7 341 758 8 250 911 8 687 864 9 066 007 9 287 547 9 569 609 1998 1 301 045 3 105 752 5200 794 7 114 889 8 399 399 9 439 846 9 998 307 10 371 062 10 708 011 1999 1 460 089 3 242 648 5 561 016 7 219 154 8483535 9 111 338 9 813 296 10 228 039 2000 1 615 641 3 562 864 5 565 190 7 232129 8 297 800 8 874 527 9 474 992 2001 1 756 010 3 708 660 5 786 013 7 344 399 8 461 272 9511 124 2002 1 894 116 3 524 704 5 625 241 7 503 179 9 060 008 2003 1 705 261 3 564 037 5 664 483 7 508 130 2004 1 850 733 3 454 746 5 387 894 2005 1 966 886 3 603 477 2006 1 516 518 B F Calculations 2006 2005 2004 2003 2002 2001 2000 1999 1998 gt Eamed Premium 27 496 114 24 475 213 24 079 420 20 293 316 16 314 960 10 867 684 11 409 056 11 356 614 12 384 141 Percent of Ultimate 14 8 29 5 47 9 63 3 74 8 82 5 87 8 91 4 94 0 Prior Seed Loss Ratio 0 0 67 7 71 0 73 2 74 4 74 4 74 4 79 2 86 9 Selected Seed Loss Ratio Estimated Ultimates 16 347 359 14 801 988 13 918 751 12 743 826 12 003 749 10 872 845 10 534 232 11 021 094 11 370 810 St Dev Factor 0 166 0 137 0 090 0 084 0 039 0 022 0 014 0 008 0 004 Ultimate St Dev 2 816 870 2 182 883 1 481 538 1 196 066 454 883 168 908 120 711 77 011 48 325 Status Earned Pre
40. ression that can be used to estimate projected loss development factors is governed by this parameter A significance test is performed on the slope rate of change in factors over time determined by the regression When the slope is found to be statistically insignificant the result will be a weighted average of factors When the slope is found to be significant the result will be a factor that reflects the trend The greater the slope is relative to its critical value the more slope will be reflected in the calculated factor You Return to Table of Contents 38 can lower the parameter shown in this box to reduce the sensitivity to the observed trend 0 gt long term weighted averages are used You can raise the parameter to increase the sensitivity to the observed trend 1 gt the factor completely reflects the observed trend 7 2 Default Directories The user can select an existing folder or make a new folder to act as the default directories A default directory can be chosen for data files analysis files and system files Data files are files containing data that will be imported into a new analysis Analysis files contain a saved analysis after information has been imported into Cognalysis parameters and estimates selected etc You can change these default directories to better streamline your data input and output processes System files are files that are used by the Cognalysis system such as working files and user preference f
41. sly saved analysis file by selecting the file from the usual file menu box The analysis file is the file used by Cognalysis that contains imported data paid loss incurred loss and or premium or exposure and any selections made loss development factors seed loss ratios ultimate values etc Return to Table of Contents 9 1 3 Create Triangles If you choose to create triangles by manually entering data you will start by defining the characteristics of the triangle This includes defining the starting incurred or accident period the number of incurred or accident periods and the number of development periods This allows the user the flexibility to create triangles in a multitude of period and development increments Cognalysis Reserving System Starting Incured ccident Pernod acceptable formats Tyv Pry Trikk Cognalysis Reserving System Number of Incured ccidernt Penods Cognalysis Reserving System Number of Development Periods Ti l SESS After defining the triangle parameters the user will then be provided a grid for entering cumulative paid loss values The user can type values or paste the values in from clipboard data copied from other sources After entering all the paid values select OK in the lower right corner of the screen Return to Table of Contents 10 Data Edit BAZ Data Edit Cumulative Paid Loss Agel Age2 Age3 Aged Ageh AgeB Age AgeB Aged AgelD gt 1999 dd i EE H mm GE 2000 0 0 0
42. tch process the user will enter information in each of the following fields e Loss File Path the path of where the loss data file is located e Loss Table Name the name of the MS Access table that contains the loss information e Loss Filter full string text that filters the field values for example LOB Auto and Region West e Period the length of the period monthly quarterly or annual e Premium File Path the path of where the premium data file is located e Premium Table the name of the table that contains the premium information e Premium Filter full string text that filters the field values for example LOB Auto and Region West e Analysis Title the title to be given to this new analysis e Prior File Path if any the location of the previous analysis file e Save File Path the path of where this new analysis file should be save e Print Reports a list of reports that should be printed upon creating this analysis file Items in each field except Period and Analysis Title can be browsed for by right clicking in each cell This will bring up a menu to aid in your selection Return to Table of Contents 34 Individual rows can be replicated by right clicking on the row header of the row to be replicated In this way a large number of similar entries can be made quickly After the batch process has been built the user can run the process by selecting the Run option in the lower right corner of th
43. timate selected reserve and reserve range percentiles 10 25 50 75 90 and 99 in the Results section shown below by selecting the Results button in the upper right part of the screen Cognalysis Reserving System Results and Ranges File Edit View Graphs Tools Help wwww cgconsult com Gel Industry H1 Current View Residual Incurred Incurred Methods Projecti Ch LDF B F Summary rojections anges Ratios Annualize Reserve Range Percentiles Famed Curent Paid kp sere a 10th 25th 50th 75th 90th 93th gt 1997 11 506 094 9 569 609 10 040 632 10 392 054 822 445 786 907 903 326 921 969 841 046 958 594 989 652 1998 12 384 141 10 708 011 11 376 077 12 024 513 1 316 502 1 275 283 1 294 460 1 316 104 1 338 110 1 358 232 1 393 569 1999 11 356 614 10 228 039 11 043 409 11 893 751 1 685 712 1 624 601 1 643 802 1 665 400 1 687 282 1 707 224 1 742 092 2000 11 409 056 9 474 992 10 359 005 11 446 701 1 971 709 1 927 893 1 948 386 1 971 410 1 994 706 2 015 909 2 052 924 2001 10 867 684 9 511 124 10 973 275 12 575 373 3 064 249 2 893 220 2 971 647 3 061 277 3 153 610 3 239 096 3 391 660 2002 16 314 960 9 060 008 10 694 769 12 790 944 3 730 936 3 410 527 3 554 900 3 722 258 3 997 604 4 062 482 4 362 696 2003 20 293 316 7 508 130 9 899 245 12 878 918 5 370 788 5 215 827 5 288 025 5 369 411 5 452 050 5 527 518 5 659 837 2004 24 079 420 5 387 894 7 962 340 12 023 133 6 635 239 6 377 007 6 496 614 6 632 132 6 770 476 6 897
44. ts 25 Cognalysis Reserving System Projection and Revision Fie Edit View Graphs Tools Help www cgconsult com 23 led Industry H1 Current View Paid LDF Paid B F orice Incurred Incurred Methods Projecti LDF B F Summary ae Residual Triangles Projection and Revision Loss 14 000 000 C vite 12 000 000 eer 10 000 000 D Case EE gt 8 000 000 6 000 000 4 000 000 2 000 000 Proj Cur Proj Cur Proj Cur Proj Cur Proj Curr Proj Cur Proj Cur Proj Cur Proj Cur Proj Cur 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Incurred Period id Start w Cognalysis User Man Cognalysis Reservi Search Desktop j 2 1 52PM Status Seed Loss Ratios This graph shown below shows a line with the seed loss ratios as well as the loss ratio implied by each of the four indications for each accident period Cognalysis Reserving System Seed Loss Ratio Perspective MEINE File Edit View Graphs Tools Help www cgconsult com TSA MEN Industry H1 Current View Paid LDF PaidBF ovine Fei Incurred Incurred Methods az Seed Loss Ratio Perspective Loss Premmum 100 50 25 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Incurred Period Status start w Cognalysis User Mar Cognalysis Reservi arch Deskte 2 ie 4 1 53 PM Return to Table of Contents 26 Ultimate Box Whisker This graph shows the selected values for each period wh
45. velopment It also develops a blended estimate which reflects the potential for diversification across these methods to reduce the overall uncertainty Users can review the details of this analysis as shown below by selecting the Optimize button in the upper right part of the screen Cognalysis Reserving System Optimization of Method Weights File Edit View Graphs Tools Help www cgconsult com i 23 led Industry H1 Current View Residual Incurred Incurred Methods Projecti Ch LDF B F Summary rojections anges Methods Optimization a Weights Applied to Each Method Paid LDF Incurred LDF Paid BF Incurred BF St Deviation Estimated Reserve pi 0 3333 1 5841 0 3635 0 6143 1 112 587 44 857 770 Four Methods Combined 2 0 0367 1 1223 0 0855 0 1 487 082 45 015 098 3 0 1329 1 331 0 0 1981 1 330 997 44 860 734 eE Ea 4 0 9543 0 0 0657 0 1114 2 276 140 38 929 889 5 0 1 2146 0 2659 0 4806 1 214 766 43 079 895 6 0 0361 1 0361 0 0 1 546 707 45 141 746 f 0 9761 0 0 0239 0 2 286 837 38 411 674 Two Methods Combined 8 0 9555 0 0 0 0445 2 280 860 38 718 085 9 0 1 0855 0 0855 0 1 488 140 44 757 447 10 0 1 1876 0 0 1876 1 347 834 44 021 057 11 0 0 0 1409 1 1409 4 992 326 50 012 026 12 1 0 0 0 2 290 256 38 211 890 14 0 0 1 0 5 604 829 46 571 033 15 0 0 0 1 5 002 229 49 587 067 Correlation Matrix and St Dey of Reserves Paid LDF Incurred LDF Paid BF IncuredBF St Dev of Reserve gt Paid LDF 1 000 0 703 0 358 0 3
46. verage MWA Run Test Significance 2 O Previously Selected Ultimates with No Automatic for Most Recent CO Previously Selected Ultimates with MAWA for Most Recent Age toAge Factor Slope Significance 0 5 Critical Value for Correlation t T est 1 Development Factors 8 Simple Average Factors Weighted Average Factors Regression Factors Previously Selected Factors Seed Loss Ratios Dynamic Determination of Loss Ratios Previously Selected Loss Ratios Restore Default Ultimates Users may choose the default ultimate from the following four options e No Automatic Selection The ultimate for each period will be zero until the user selects an ultimate value e Minimum Variance Weighted Average The ultimate for each period will be the optimized value that combines methods to produce the minimized variance e Previously Selected Ultimates with No Automatic for the Most Recent The ultimate for each period will be set equal to the ultimate for the same period from the previous analysis file that has been imported For the most recent and therefore new period the ultimate will be zero until an ultimate is selected e Previously Selected Ultimates with MVWA for Most Recent The ultimate for each period will be set equal to the ultimate for the same period from the previous analysis file that has been imported For the most recent and therefore new period optimized v
Download Pdf Manuals
Related Search