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.
DATABASES FOR MICROSIMULATION: A COMPARISON OF THE MARCH CPS AND SIPP 41 exact-match file of the March 1978 CPS and social security earnings histories, is the primary database for ICF's PRISM model. Because of heightened concerns about confidentiality, no such exact-match files have been constructed subsequently that are available for public use. Hence, modelers of retirement income programs and other long-range policy issues that require longitudinal histories now have to use more and more years of simulated historical data before they can even begin their projections into the future. One of the design goals for the SIPP was to develop a program for augmenting the survey data on a regular basis with various kinds of administrative records. Extra effort is made for matching purposes to obtain social security numbers from SIPP respondents, which are subsequently validated by the Social Security Administration (SSA).15 These efforts are quite successfulâSIPP has validated social security numbers for about 95 percent of respondents. (In contrast, no special effort is made in the CPS to press respondents for social security numbers, and there is currently no regular program of having them validated.) The Census Bureau has prepared an exact match of the 1984 SIPP panel with social security data at the request of SSA analysts. However, only SSA analysts may use the file (they must be sworn in as special census employees), and they are only allowed access for a 2-year period. The Census Bureau has also matched SIPP data to IRS records in order to develop ratio estimates of mean and median personal and household income with reduced variance. However, the goal of developing public-use versions of the SIPP data augmented with administrative records remains out of reach at this time. PROBLEMS OF VARIABLE MISSPECIFICATION This section discusses differences between the variables in the CPS and SIPP and those variables needed for simulation of income support programs for the following topics: income accounting period, income detail, household composition reference period, and households and families versus program filing units. The comparison is entirely in favor of the SIPP. Income Accounting Period The March CPS collects income and work experience data on an annual basis, pertaining to the previous calendar year. However, income support programs such as AFDC operate on a monthly accounting basis. The available evidence (from sources such as the income maintenance experiments and the ISDP and SIPP) is that a number of households experience sufficient intrayear income 15 Another goal was to supplement the SIPP area-frame sample with samples for special populations, such as income support program recipients, drawn from administrative records. Such sample additions were part of the ISDP program, although the data were never fully analyzed. However, this aspect of the SIPP program has never gotten off the ground.