Pension Projection Models
The PSG models provide a broad range of federal income and payroll tax analysis capabilities that are logically consistent because all aggregate and distributional estimates are built up from simulated samples of cohort individuals.
The validity of the tax module has been tested in two ways. First, the simulated tax liability for each tax unit in a randomly generated sample of 2,200,000 (spread evenly across the years from 2004 to 2014) was found to be the same as generated by the Internet TAXSIM 9.3 model maintained by the National Bureau of Economic Research. And second, when embedded in GEMINI the tax module produces an aggregate estimate of the revenue generated by the federal income taxation of social security Disability Insurance benefits in 2004 that is very close to the actual 2004 amount. Details are in Chapter 7 of the PENSIM Overview.
When switching from aggregate to distributional analysis, the simulation sample size can be increased and the focus narrowed to a few birth cohorts or to cross-sectional samples of everyone alive in a specified calendar year. Most other models use a fixed sample size that may be adequate for aggregate analysis but can be too small for statistically reliable distributional analysis of a single birth cohort or a small population subgroup.
The number of federal income and payroll tax reforms that can be specified using existing policy parameters is large. In the future it is likely that there will be reform proposals that contain provisions not already included in the tax module, in which case the PSG models (like any other model) would need to be enhanced to add the policy parameters required to characterize the reform.
Other key features.