Policy
Simulation
Group

Pension Projection Models

Key Feature: Extensively Validated

The validity of output generated by the PSG models has been tested in several ways over the years. The overall conclusion from these tests is that the PSG models produce realistic samples of individual life histories that can support sophisticated pension and social security analysis.

First, demographic and economic statistics calculated from samples of individual life histories simulated by PENSIM have been compared with the same statistic tabulated from real survey data. The results of a number of these validation tests are reported in Chapter 8 of the PENSIM Overview.

Second, future-year aggregate demographic and economic statistics calculated from PENSIM life histories closely match the projected statistics in the social security Trustees Report. This means that the PSG models can be used with confidence when undertaking social security policy analysis. And the use of these same Trustees Report intermediate-cost assumptions for pension analysis provides a much-needed standard for a literature in which widely differing results across papers are often caused by the use of substantially different economic and demographic assumptions.

Third, employer-sponsored pension output of the PSG models has been compared to statistics tabulated from real survey data. The results of a number of these validation tests are reported in Chapter 10 of the PENSIM Overview.

Fourth, estimates of the solvency effects of nearly two dozen social security reform provisions produced by the PSG models operating in OLC mode have been compared with estimates produced by the SSA Office of the Chief Actuary (OCACT). These comparisons, which have been performed by SSA, show that in almost all cases, the two estimates are the same or very close. And in the few cases where there are meaningful differences, investigations raise as many, if not more, questions about the OCACT estimates than about the estimates from the PSG models.

And fifth, PSG model estimates of the effects of numerous multiple-provision social security reform proposals on individual benefits and trust-fund solvency have been compared with OCACT estimates. These comparisons, which have been performed by SSA and GAO over the years, have been favorable to the PSG models. For example, in the case of the Commission to Strengthen Social Security reform proposals, the individual benefit estimates differed by less than one percent.

Other key features.


This page was last revised on February 17, 2016.