Social Security and Private Saving: Reply

Abstract

I am embarrassed by the programming error that Dean Leimer and Selig Lesnoy uncovered but grateful to them for the care with which they repeated my original study (Feldstein 1974). They set an admirable example of the tradition of replication on which all scientific work ultimately rests. As economic research increasingly involves large and complex computer programs to analyze macroeconomic data sets or to simulate models that cannot be solved analytically, replication studies like that of Leimer and Lesnoy ("Social Security and Private Saving: New Time-Series Evidence," in this issue) should become increasingly important. In their comment, Leimer and Lesnoy also suggest a variety of modifications to the social security wealth variable to incorporate alternative employee perceptions and actuarial assumptions. As I explain below, I have strong doubts about the appropriateness of these modifications. I therefore do not agree with their conclusion that "the time-series evidence does not support the hypothesis that the introduction of social security has substantially reduced personal saving in the United States." More specifically, as I show in this reply, correcting the computerprogramming error and reestimating the original equation for the original sample period produces results that are not substantially different from the results that were originally published in 1974. The point estimate of the coefficient of social security wealth implies that the level of social security wealth in the final year of the sample (1971) depressed personal saving by an amount equal to 44 percent of the actual personal saving in that year. Although the estimated coefficient of the social security wealth variable is smaller than in the original version and its standard error larger, the conventional statistical test