Social Security Benefits, and the Impact of Pension Reform

tive workers. This article Summary involves introducing a fundamentally new describes two new methods for kind of pension formula. estimating the career profile In order to assess the effect of Social This article presents two broad of earnings for representative Security reform on current and future approaches to creating representative workers. It then compares the results of those new methods workers, it is essential to accurately earnings profiles for policy evaluation. characterize the initial situations of First, we use standard econometric with earnings profiles as„ sumed in traditional distribu„ representative workers affected by methods to predict future earnings for a tional analysis of Social reform. For the purpose of analyzing representative sample of workers drawn Security and shows the impli„ typical reforms, the most important from the Survey of Income and Program cations of the new results for characteristic of a worker is the level Participation (SIPP). Our statistical evaluating Social Security and pattern of his or her preretirement estimates are based on a simple repre„ reform. earnings. Under the current system, sentation of typical career earnings paths *Barry P. Bosworth and Gary pensions are determined largely by the and a fixed-effect statistical specificaBurtless are with the Brookings level of the workers’ earnings averaged tion. Because our estimation file conInstitution. Eugene Steuerle is over their work life. However, several tains information on each worker’s with the Urban Institute. reform proposals would create individual annual earnings from 1951 through 1996 retirement accounts for which the as reported in the Social Security pension would depend on the investment Administration’s earnings files, we have accumulation within the account. Thus, a record (though an incomplete one) of the pension would also depend on the the actual earnings that will be used to timing of the contributions into the determine future benefit payments. Our account and hence on the exact shape of estimates of the earnings function permit the worker’s lifetime earnings profile. us to make highly differentiated predicMost analysis of the distributional impact tions of future earnings for each member of reform has focused, however, on of our sample. By combining the histori„ calculating benefit changes among a cal information on individual earnings handful of hypothetical workers whose with our prediction of future earnings up relative earnings are constant over their through the normal retirement age, our work life. The earnings levels are not first approach produces tens of thounecessarily chosen to represent the sands of predicted career earnings paths situations of workers who have typical or