Social Security Does Not Redistribute Income

Summary of working paper 7520
Featured in print Digest

Social Security does not redistribute from people who are rich over their lifetime to those who are poor. In fact, it may even be slightly regressive.

Many people think that Social Security is a progressive program which redistributes income from the rich to the poor. But according to new research by Julia Lynn Coronado, Don Fullerton, and Thomas Glass, Social Security does not redistribute from people who are rich over their lifetime to those who are poor. In fact, it may even be slightly regressive.

In The Progressivity of Social Security (NBER Working Paper No. 7520), the authors concentrate on redistribution within, rather than across, generations. The Social Security benefit formula explicitly transfers money from people who earned more during their working years to those who earned less. This is why it is commonly seen as a progressive program. And, most economic studies have confirmed that Social Security is redistributive.

But the common perception and these previous studies fail to include a range of relevant individual characteristics that determine whether people really are lifetime rich or lifetime poor, the authors argue. When people are properly classified, no redistribution is found.

This research is based on estimated lifetime wage profiles and actual earnings data for a sample of 1800 individuals. The researchers use mortality probabilities to calculate expected payroll taxes and retirement benefits. They test redistribution in terms of net tax rates (the present value of expected tax minus expected benefit) and also measure levels of income inequality in the population before and after Social Security taxes/benefits are included. The result is a measure of how progressive the system really is.

As a first cut, the researchers classify individuals by income in a single year, to represent all U.S. citizens, working and retired, who take part in the system. The result is that Social Security is shown to be highly progressive: it has a measure of "effective progression" of 1.27. (A value of 1 implies that Social Security makes no difference - there is no redistribution.)

The second step is to shift the analysis to lifetime income, to reflect the fact that Social Security is like a retirement saving program, to which people make contributions during their working years and receive benefits in retirement. Taking into account the taxes paid and benefits received over a lifetime, the researchers find that Social Security is far less progressive: the measure of effective progression falls to 1.05.

The third step is to take account of the cap that Social Security puts on earnings subject to the payroll tax. If one person earns the taxable maximum during the working life, and another earns twice the maximum, both pay the same amount of tax and receive the same benefits. But the tax takes up a higher proportion of the lifetime income of the first person. Removing this distortion, the researchers show that the program is slightly less progressive: the measure of effective progression is just under 1.05.

The fourth step is to take account of the fact that people may choose to work part time, or to take career breaks, in order to do nonmarket work at home or to enjoy leisure time. Unlike periods of involuntary unemployment, the researchers say, periods of time that people choose to spend outside the labor market should not be counted as reducing lifetime income for the purposes of evaluating redistribution. Accordingly, they shift the analysis from actual lifetime income to potential lifetime income (the discounted value of a person's wage rate multiplied by 4,000 hours for each year). This measure captures the value of leisure and home production. On this basis, Social Security is hardly progressive at all, with an effective progression measure of 1.01.

The fifth step reflects the fact that husbands and wives pool their resources. For the purposes of assessing the progressivity of Social Security, a low-wage spouse of a high-earner should not be treated as poor, the researchers say. Therefore, they add the potential income of husbands and wives together, and divide by two. Because a low-earner with a high-earning spouse is no longer counted as lifetime poor, Social Security is shown to be less progressive: the measure of effective progression falls to slightly above 1.

The sixth step is to incorporate different life expectancies into the model. Mortality rates differ with potential income: rich people tend to live longer than poor people, and so collect more years of benefits. Taking this into account in the calculation, the researchers show that Social Security is no longer progressive: the effective progression measure is barely above 1.

Most previous research - and the first six steps of this paper - used a discount rate of 2 percent. Coronado, Fullerton, and Glass change the discount rate to reflect the return on an asset with comparable risk (the risk that the government will default on its obligations). Their seventh step increases the discount rate from 2 to 4 percent, in line with the return on indexed Treasury bonds. Raising the discount rate puts more weight on the earlier, regressive payroll tax and less on the later, progressive benefit formula. With a 4 per cent discount rate, the authors find that Social Security is actually slightly regressive, with an effective progression measure of 0.998.

-- Andrew Balls