Options for Price Indexing Social Security

10/20/2009
Featured in print Bulletin on Aging & Health

With the U.S. Social Security system facing a long-run deficit, policy makers and researchers have suggested a wide variety of policy changes to improve the system's finances. One of the most frequently-mentioned proposals is to switch from "wage indexing" of Social Security benefits to "price indexing."

This reform would affect how Social Security benefits evolve over time. Under the current wage indexed benefit formula, initial benefits to successive cohorts of retirees grow along with wages, so that replacement rates (the fraction of lifetime earnings replaced by Social Security) remain roughly constant. Under price indexing, initial benefits would instead grow with prices so that real benefits, rather than replacement rates, would be held constant. Because wages generally grow more quickly than prices, a shift to price indexing would result in a slower rate of benefit growth.

In Alternative Methods of Price Indexing Social Security: Implications for Benefits and System Financing (NBER Working Paper 11406), Andrew Biggs, Jeffrey Brown, and Glenn Springstead examine the effect of price indexing on replacement rates, progressivity, and fiscal sustainability. In their analysis, the authors distinguish between four alternative approaches to price indexing benefits.

To understand the alternatives, a brief review of how Social Security benefits are calculated is useful. In the first step, the worker's annual earnings in all past years are multiplied by a wage index to bring them up to current dollars; the best thirty-five years of real earnings are then used to calculate the worker's average lifetime earnings, which is known as the Average Indexed Monthly Earnings (AIME). Second, a non-linear, progressive formula is applied to the AIME to get the Primary Insurance Amount (PIA). In 2005, the PIA is equal to 90% of the first $627 of AIME, plus 32% of the next $3,152, plus 15% of the remaining AIME. Finally, the monthly benefit amount is equal to the PIA for those workers who retire at the Full Retirement Age, and is lower (higher) for those who retire before (after) it.

While price indexing has been a common feature of many reform proposals, exactly what constitutes price indexing has been the subject of some disagreement. The authors examine four possible price indexing schemes. Under the first option, AIME indexing, the index used to bring past earnings forward to current dollars would be a price index rather than a wage index. Under the second option, bend point indexing, the dollar amounts in the PIA formula would be adjusted annually by a price index rather than a wage index. The third option combines AIME and bend point indexing. Under the fourth option, PIA factor indexing, the PIA formula's percentage factors (the 90%, 32% and 15% figures) would be reduced gradually by being multiplied each year by the ratio of price growth to wage growth.

The authors first examine the effect of the various reforms on replacement rates. Compared to the current system, which provides a 45% replacement rate for the average worker, all of the reform proposals have lower benefits. AIME indexing would lower the replacement rate to 40% by 2050, but the replacement rate would not fall further after this. By contrast, the other three reforms would lower the replacement rate continuously over time, to 32% in 2075 for bend point indexing, 29% for AIME and bend point indexing, and 21% for PIA factor indexing. However, these rates are not dramatically different from the 30% replacement rate the authors calculate is actually payable with no changes to current Social Security law.

The authors also discuss how these reforms would affect the progressivity of Social Security. AIME indexing is expected to result in larger declines for low-income workers than for high-income workers for two reasons. First, using a price index to convert past earnings to today's dollars reduces the value of earnings early in one's career and these earnings are more important for low-income workers, who have flatter age-earnings profiles. Second, a given reduction in AIME leads to a larger decrease in benefits for low-income workers, whose average lifetime earnings are replaced at a rate of 90%, than for workers with a 32% or 15% replacement rate.

The authors' second option, indexing the bend points, leads to an uneven distribution of benefit cuts in the short run, with the lowest-income workers spared any cuts. In the very long run, the 90% and 32% phases of the PIA formula would essentially disappear, so that all workers would receive a benefit equal to 15% of lifetime earnings. Thus this proposal would eventually eliminate the progressivity in the current benefit formula. The fourth option, indexing the PIA factors, would reduce the PIA by the same percentage for all workers and thus would preserve the system's current degree of progressivity.

Finally, the authors examine the effect of the various proposals on Social Security's long-term solvency. The 1999 Social Security Trustee's report predicted that over the next 75 years, the Social Security system would run a deficit equal to 2.07 percent of taxable payroll. The AIME indexing method would reduce this projected deficit by one-third, while the bend point indexing method would reduce the deficit by two-thirds. Only the PIA factor indexing method would fully eliminate the deficit, with projected savings of 2.36 percent of taxable payroll. As the authors note, this change would leave the system with annual surpluses at the end of 75 years, so taxes could be reduced or benefits could be increased somewhat.

The authors conclude that PIA indexing would reduce benefits by approximately the same rate for all wage earners and would restore long-term solvency to the Social Security system. But this method would also greatly reduce Social Security replacement rates and would potentially increase the sensitivity of system finances to unexpected earnings changes. They suggest that any reform proposal should be examined not only for its effect on benefit levels, progressivity, and system solvency, but also for its degree of political risk and likely effects on savings and labor supply.