Social Security fails---studies prove no progressive redistribution.

In this paper we seek to measure the extent to which the current social security system supports the social safety net, that is, the extent to which it redistributes resources to the poor. To do so, we build a model that incorporates all the information needed to categorize individuals by lifetime resources and to calculate their taxes paid and benefits received from the system. We have several findings.

First, we find that using comprehensive concepts of income implies that the Social Security system exhibits less overall redistribution than when it is evaluated using more narrow definitions of income. For example, when evaluated using potential labor earnings at the household level (rather than actual individual earnings), the Social Security retirement program exhibits virtually no overall impact on inequality (as measured by the change in the Gini coefficient). Second, we look at each quintile of the lifetime income distribution. We find that the lack of impact on overall inequality is largely driven by the lack of impact across the middle and upper part of the income distribution, whereas most of those in the bottom income quintile may, in fact, still get net benefits from the program. Third, even when redistribution does occur, we find that it is not efficiently targeted, with many high income households receiving net transfers, while many low income households pay net taxes. Finally, we show that the impact of Social Security on inequality differs across cohorts, but the direction and extent of those changes depends on the income concept employed.

This research suggests several areas for future work. First, this analysis explicitly ignores behavioral responses to the Social Security system, including changes in labor supply and/or savings behavior that might influence how we think about the system's impact on inequality. Second, this analysis is purely financial, and it thus misses important insurance aspects of the Social Security program, including its provision of earnings insurance, longevity insurance, and disability insurance. Analysis of the insurance value of Social Security would require a sophisticated dynamic programming model that embeds this analysis in a utility framework. Finally, given the poor long-term fiscal outlook of the Social Security program, this framework could be used to explore the distributional implications of alternative reform options.