Proposed Social Security Reforms Have Little Effect on Income Distribution

Summary of working paper 6989
Featured in print Digest

Income-differentiated mortality reduces the progressivity in the Social Security system by more than half.

Higher earners, on average, live longer than poor people. Thus, Social Security in its present form is not as progressive as is commonly thought: the benefits formula is progressive, but because higher earners live longer, they collect more benefits. Relatedly, the reforms to shore up Social Security's finances by cutting benefits and raising taxes - considered by the 1994-6 Advisory Council on Social Security - are not as "regressive" as they may at first appear. Indeed, according to an NBER Working Paper by Julia Coronado, Don Fullerton, and Thomas Glass, eliminating the provision for dropping certain low earning years from the Social Security benefit calculation, increasing the retirement age, decreasing benefits directly, and raising the payroll tax actually have very little effect on income redistribution within the U.S. pension system.

In Distributional Impacts of Proposed Changes to the Social Security System (NBER Working Paper No. 6989), the authors first identify high and low income groups in a sample of 1082 household heads and 696 secondary earners. Then, using data from 1968-89, they estimate lifetime incomes for these individuals, group them into income quintiles, and simulate their earnings histories.

The researchers then calculate that under the current system, those in the bottom quintile pay lifetime taxes of $64,700 and can expect to receive lifetime benefits of $125,700. Those in the top quintile pay taxes of $141,400 and receive benefits of $187,000. Using a benchmark discount rate of 2 percent, the authors find that at age 22, those in the bottom 20 percent face a net loss of $1,300, equal to 0.17 percent of the discounted lifetime endowment. The top quintile faces a present value loss of $30,100, which represents 1.33 percent of the present value of lifetime income (a difference between the two groups of 1.16 percent).

However, low income individuals have higher than average mortality rates, and so collect fewer years of benefits. For example, the poorest 20 percent of non-white females in the sample have a mortality rate which is 186 percent of the average for that group. Those in the top 20 percent of the income scale have a mortality rate that is equal to 44 percent of the average for that group.

Taking account of income-differentiated mortality, the authors show that the tax rate is 0.6 percent for the lowest quintile and 1.01 percent for those in the top quintile. This is a difference of only 0.41 percent (compared to the 1.16 percent estimated by ignoring mortality effects). The researchers conclude that income-differentiated mortality reduces the progressivity in the Social Security system by more than half. If the discount rate is increased to 4 percent, so that regressive payroll taxes are more important relative to progressive benefits, then the lifetime tax rate inherent in the Social Security system is regressive and close to 3 percent for all income groups.

The authors then assess four proposed Social Security reforms. Eliminating the drop-year provision reverses a measure designed to increase benefits for individuals with high lifetime variability of income. The researchers show that this measure reduces benefits in all income groups. Without income-differentiated mortality, the decline in net benefits is fairly flat across all groups, so this is a regressive measure. Introducing income-differentiated mortality means the impact is slightly less regressive.

Increasing the retirement age (already enacted legislation will increase the retirement age from 65 to 67 by 2020) is also regressive because individuals continue to pay the regressive tax for longer and receive the progressive benefit for a shorter period. Again, income-related mortality reduces the regressive nature of the reform because the present value loss in net benefits is greater for high income groups.

Decreasing the overall benefit level is regressive, because a cut will represent a bigger share of lifetime income for lower income groups. Again, introducing income-differentiated mortality makes it less regressive, because richer people live for longer.

However, in the case of increasing the payroll tax - a regressive measure - mortality assumptions matter less here, because the taxes are paid earlier in life when mortality rates are lower for all groups.

-- Andrew Balls