Does Social Security Redistribute to Low Income Groups?

Summary of working paper 8625
Featured in print Digest

One in five individuals in the top 20 percent of the lifetime income distribution receive greater net transfers than the average for people in the bottom 20 percent of the income distribution.

Because the Social Security benefit formula replaces a greater fraction of the lifetime earnings of low earners than of high earners, the U.S. Social Security system is generally thought to be progressive. However, in Redistribution in the Current U.S. Social Security System (NBER Working Paper No. 8625), NBER Faculty Research Fellow Jeffrey Liebman shows that much of the redistribution in the U.S. Social Security system is related to factors other than lifetime income. Total income-related transfers from Social Security are relatively modest.

The pattern of redistribution that occurs through Social Security is complicated because lifetime taxes and benefits are influenced by the different mortality rates of people in different demographic groups, by differences in marital status, and by variations in the earnings levels of secondary earners in married couples, among other factors. Thus the income-related redistribution that occurs because of the progressive benefit formula is partially offset by the longer life expectancies of higher income individuals and by the larger spouse-and-survivor benefits received by the spouses of higher earners.

In his study, Liebman examines the distribution of internal rates of return, net transfers, and lifetime net tax rates from Social Security that would have accrued to people born from 1925-9 if the present Social Security rules had been in place for their entire lives. His simulation is based on matching the 1990 and 1991 Surveys of Income and Program Participation with Social Security administrative earnings and benefits records.

Liebman finds that within this group, Social Security provides net transfers equal to just 13 percent of Social Security benefits paid (when taxes and benefits are discounted at the sample's 1.29 percent rate of return ). Because much of this 13 percent redistribution is related to non-income factors, the annual income-related transfers from Social Security are only 5-9 percent of Social Security benefits paid, or $19-34 billion at 2001 aggregate benefits levels. Using a higher discount rate than 1.29 percent, Social Security appears to be more redistributive on some measures and less so on others.

Moreover, the range of transfers received at a given level of lifetime income is very wide. Liebman shows that one in five individuals in the top 20 percent of the lifetime income distribution receive greater net transfers than the average for people in the bottom 20 percent of the income distribution.

A number of proposed Social Security reforms would increase the link between workers' Social Security contributions and their retirement income by supplementing or replacing the current system with defined-contribution personal retirement accounts (PRAs). These proposals have led to concerns that some of the redistributive and poverty-reducing components of the system would be lost.

Liebman suggests that his results have two implications for such Social Security reform. First, they suggest the magnitude of redistribution that a PRA-based plan would need to achieve in order to maintain the current level of redistribution from high-earners to low-earners. A mixed plan in which PRAs are responsible for about one-third of the retirement income from Social Security would require the equivalent of $7 to $10 billion per year in transfers. However, most PRA plans would mandate that retirees convert at least part of their account balances into annuities and thus would redistribute from demographic groups with short life expectancies to groups with long life expectancies). Therefore, several billion additional dollars of transfers would be necessary to maintain the current level of redistribution.

Second, if no explicit steps are taken in a PRA-based plan to redistribute income to groups with low life expectancies, then certain demographic groups with low life expectancies could end up doing substantially worse than under the current system. For example, blacks and high school dropouts currently receive rates of return from Social Security that are roughly the same as the population average, because the progressive benefit formula offsets the impact of their relatively high mortality rates. If a PRA-based plan required individuals to convert their account balances into annuities at a single price for everyone in the population, then mortality would have the same effect on benefit payments as in the current system, but there would be no progressive benefit formula to offset it. Providing explicit redistribution as part of the individual accounts would be one way to ensure that these groups are not made worse off by Social Security reform. Valid comparisons between the current system and possible alternatives, Liebman argues, must take into account the actual pattern of redistribution under current Social Security rules. However, income redistribution is not the only potential benefit that Social Security provides for low-income families. Some of the other benefits - such as the inflation-protected annuity and the absence of market risk - may be particularly valuable to low-income families.

-- Andrew Balls