Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
FUTURE COMPUTING ENVIRONMENTS FOR MICROSIMULATION MODELING 156 ⢠Control parameters specify the input and output files to be used and control the optional model software facilities. ⢠Database adjustment parameters control the inflation of the database money items to a target year of analysis. ⢠Tax/transfer parameters control the specific functioning of the tax and transfer programs as represented by the SPSD/M operating characteristics. These parameters include a set that can be created by the commodity tax input/output model (COMTAX) (see Other Software Facilities). There are over 400 parameters to SPSD/M, and the 1984 version supplies values for 10 different scenarios: 1984 actual programs, 1985 actual programs, 1986 actual programs, 1987 actual and estimated, 1988 prereform, 1988 pre-reform deflated to 1984 dollars, 1988 reform, 1988 reform deflated to 1984 dollars, 1989 reform and estimated, and 1989 reform and estimated deflated to 1984 dollars. Each parameter type is stored in a different file type, and each scenario is supplied by providing a set of control, database adjustment, and control parameter files. The SPSD/M Reference Manual (Statistics Canada, 1989d) has a Parameter Guide that lists all the values of the parameters for each supplied scenario and gives the source of each data value. Users can interactively modify parameters during a model run. The resultant parameters used during a run are stored in an output set of parameter files to permit the run to be reexecuted easily. Coordination of different parameter files is a user function, and no central repository is provided by SPSD/M for standard or historical parameters. Model Execution The novice user can become acquainted with how to use the SPSM by referring to either the SPSD/M Introductory Manual (Statistics Canada, 1989b) or, for more detailed information, the SPSD/M User's Manual (Statistics Canada, 1989c). When SPSM is started, the user enters into an interactive dialogue (in English or French) to specify the current scenario's inputs and outputs. The user must specify the SPSD database to be used as well as the control, adjustment, and tax/transfer parameters. A text editor can be used to prepare the control, adjustment, and tax/ transfer parameter files before beginning a model run, or the user can specify a file of default parameters and then override the values of one or more parameters interactively. It is also possible to run SPSD/M in batch so that one or more runs can be executed unattended. The user also specifies the name of the current run so that SPSD/M can name all the output files for the current scenario (see Output Facilities below). For example, each model run can produce a new set of parameter files that represent the values of the parameters used in the run. These parameter files
FUTURE COMPUTING ENVIRONMENTS FOR MICROSIMULATION MODELING 157 can be used to reexecute the same model run or as the basis of a variant model run. Once the user has started the model and provided a complete set of parameters, the simulator phase begins. The simulator phase operates in the following sequential steps: ⢠Database adjustment This step inflates every income and tax-related dollar variable to the target year of analysis. ⢠Calculate base and variant model results SPSD/M permits a user to run one or two scenarios in the same run. The first or variant scenario is defined by one set of tax/transfer parameters. This scenario can then be optionally compared to a base scenario that is defined by a second separate set of tax/transfer parameters or by the contents of a result file produced during a previous execution of SPSD/M. ⢠Screening cases The next simulator step permits a user to restrict the calculation of results to a certain subset of households or to a certain type of individual or family on the database. This can be useful for efficiency and analytical purposes. The screening step can be based on modeled as well as original database variables; therefore, it follows the calculation of the variant and base results.13 ⢠Calculate user variables This step permits a user to create new user variables for reporting purposes by manipulating the database and modeled variables. ⢠Output phase The output phase is the last step of the simulator and is designed to produce a series of outputs for different purposes. SPSD/M can produce any of 10 default tables, or the user can specify custom tables using the built-in X-Tab tabulation facility. The output phase also permits the user to output files containing subsets of SPSD cases in several different formats for subsequent data analysis using SPSM or a statistical package such as SAS or a spreadsheet program such as Lotus 1â2â3. It is assumed that the user's simulation requirements can be met by simply changing one or more of the already existing SPSD/M parameters. This is sometimes referred to as the black box approach to modeling, since the user does not alter the internal workings of the underlying system. If the user's scenario is more complex, a glass box approach must be used. This requires that a C language programmer make changes to the actual SPSD/M source code. The software development environment supplied with SPSD/M has been designed to facilitate alterations to the existing operating characteristics. For example, if a user needs to redefine the formula for a calculated variable such as total income, the source code module containing 13 If the screening software can determine that a household can be rejected based only on original database variables, the tax/transfer system will not be executed.