Improving Information for Social Policy Decisions -- The Uses of Microsimulation Modeling

This volume, second in the series, provides essential background material for policy analysts, researchers, statisticians, and others interested in the application of microsimulation techniques to develop estimates of the costs and population impacts of proposed changes in government policies ranging from welfare to retirement income to health care to taxes.The material spans data inputs to models, design and computer implementation of models, validation of model outputs, and model documentation.Table of ContentsFront MatterIntroductionModel DesignModel EvaluationDynamic Simulation of Income Model 2 (DYNASIM2)Social Policy Simulation Database/Model (SPSD/M)REFERENCESDatabases and Methods of Data EnhancementINTRODUCTIONDATA NEEDS FOR MODELING INCOME SUPPORT PROGRAMSSURVEY DESIGN OF THE MARCH CPS AND THE SIPPPopulation CoverageHousehold and Individual NonresponseItem NonresponseReporting ErrorsSample SizeData DeliveryIncome Accounting PeriodIncome DetailHouseholds and Families Versus Program Filing UnitsAsset HoldingsOther Variables for Related Policy IssuesEVALUATION OF ESTIMATES OF PROGRAM PARTICIPANTS FROM CPS AND SIPPTHE DIVISION OF LABOR FOR PRODUCTION OF DATABASES FROM CPS AND SIPPTHE SUITABILITY OF CPS AND SIPP FOR INTEGRATED MODELING OF TAXES AND TRANSFERSCONCLUSIONSREFERENCESINTRODUCTIONFile TreatmentConstrained and Unconstrained Statistical MatchingChoosing the Matching VariablesThe EM-AF Statistical MatchMerge File of the Office of Tax Analysis1966 Merge File for Household Income DataStatistics Canada SCF-FEX MatchAFDC-Census MatchConditional IndependenceLimitations in ModelingReweighting of File B Data Resulting From Statistical MatchingIterative Proportional FittingMore Data CollectionMultiple Matching and File ConcatenationRough Sensitivity AnalysisCONCLUDING NOTEREFERENCESModel DesignINTRODUCTIONCONVERTING ANNUAL TO MONTHLY VALUES IN TRIM2, MATH, AND HITSMHITSMTRIM2MATHHITSMTRIM2Input DataMATHCommentEvaluation of Alternative Months ProceduresNeed for Further ValidationPARTICIPATION FUNCTIONS FOR SSI, AFDC, AND FOOD STAMPS IN TRIM2, MATH, AND HITSMTRIM2HITSMTRIM2TRIM2MATHEvaluation of Alternative Participation FunctionsAPPENDIX: PERFORMANCE OF THE AFDC CALIBRATION PROCESS IN TRIM2REFERENCESTHE MODELSSTARTING DATABASESSIMULATION OF LONGITUDINAL HISTORIESLabor Force SimulationsPRISM Pension Coverage and CharacteristicsInteractions Between Pension and Social Security Eligibility and Retirement in PRISMLabor Force SimulationsRetirement in DYNASIM2DYNASIM2 Pension Coverage and CharacteristicsCONCLUSIONSREFERENCESComputing TechnologyINTRODUCTIONLayered StructureSingle Period Versus Future ProjectionsStatic Versus Dynamic SimulationHistorical BackgroundHistoryDatabase CreationDatabase Structure and SizeSPSD/M ParametersModel ExecutionComparing Model RunsOutput FacilitiesOther Software FacilitiesAvailability, Customer Base, and Technical SupportFuture WorkHistoryDatabase CreationDatabase Structure and SizeDatabase AdjustmentOperating CharacteristicsTRIM2 ParametersModel ExecutionOutput FacilitiesOther Software FacilitiesOperating EnvironmentComparisonBackgroundInformed Industry ObserversSpecific Technical Product InformationAdvances in Computer ProcessorsAdvances in Primary MemoryAdvances in Computer Systems ArchitectureEconomic Studies of Industry PerformanceAssessment and PredictionsNature of the Demand for Microanalytic Simulation ModelsFactors Affecting the Availability of Microsimulation ModelsFunctions of a System for Microanalytic SimulationLarger Model Execution on the DesktopAdvances in Software MethodologyModel SpecificationObject-Oriented ComputingSimulation Module ConsistencyAdvantages of Desktop-Based EnvironmentsSummaryRECOMMENDATIONS AND CONCLUSIONSShort-Term IssuesPorting TRIM2 to an MS-DOS EnvironmentTranslating TRIM2 Modules to the C LanguageBenefits of Porting TRIM2Medium-Term IssuesREFERENCES AND BIBLIOGRAPHYModel Validation6 Variance Estimation of Microsimulation Models Through Sample Reuse COMPUTATIONS OF MICROSIMULATION MODELSNONPARAMETRIC VARIANCE ESTIMATIONMODELS AND VARIABILITYBOOTSTRAPPING MICROSIMULATION MODELSIncorporation of Sensitivity Analyses - Measurement of Total UncertaintyNumber of Replications NeededUse of Resampling Techniques in Constructing Confidence IntervalsREFERENCES7 Evaluations of Microsimulation Models: Literature Review HENDRICKS AND HOLDEN (1976A)HENDRICKS AND HOLDEN (1976B)GENERAL ACCOUNTING OFFICE (1977)HOLDEN (1977)HAYES (1982)JEFFERSON (1983)HAVEMAN AND LACKER (1984)ICF, INC. (1987)KORMENDI AND MEGUIRE (1988)BETSON (1988)DOYLE AND TRIPPE (1989)BEEBOUT AND HAWORTH (1989)BURTLESS (1989)DISCUSSIONREFERENCES8 A Validation Experiment with TRIM2 OVERVIEW OF THE EXPERIMENTCHOICE OF MODEL YEAR, PROGRAM YEAR, AND COMPARISON VALUESCHOICE OF COMPONENT MODULES AND ALTERNATIVES FOR STUDY IN TRIM2OUTPUTS EXAMINEDTotal Variability Due to Use of Alternative ModulesAnalysis of Variance Methods for Sensitivity Analysis and External ValidationNonparametric AnalysisAnalysis of Categorical DataMAJOR CONCLUSIONSLIMITATIONS OF THE PRESENT STUDYDESIGN OF EXPERIMENT ISSUESREFERENCESINTRODUCTIONNATURE OF POPULATION PROJECTION MODELSVALIDATING POPULATION PROJECTIONSSensitivity AnalysisExternal ValidationCONFIDENCE INTERVALSCONCLUSIONSREFERENCESModel DocumentationINTRODUCTIONPURPOSES OF DOCUMENTATION AND EVALUATION CRITERIACritiqueSuggestionsCritiqueSuggestionsCritiqueSuggestionsCOMPARISONS WITH IEEE STANDARDSREFERENCES