Reforming Social Security with Personal Accounts

"Retirees can be insured against all downside investment risk at a very small cost to taxpayers through a government guarantee of the benchmark benefit."

If it pays benefits at promised levels as baby boomers retire, the Social Security trust fund will be exhausted by 2034. The payroll tax will have to rise to 17.3 percent by 2050 as the population ages, and to 18.7 percent by 2070. However, the payroll tax could stay at 12.4 percent if the government deposits 2.3 percent of taxable earnings in Personal Retirement Accounts (PRAs), according to research by Martin Feldstein, Elena Ranguelova, and Andrew Samwick. Investing in the stock market offers lower cost provision of pensions because the return from investing in the market is greater than the return inherent in the pay-as-you-go system of Social Security ( equal to the rate of growth in the economy). The government deposits to the PRAs can be financed by a portion of the projected budget surpluses.

In The Transition to Investment-Based Social Security when Portfolio Returns and Capital Profitability are Uncertain (NBER Working Paper No. 7016), Feldstein and his co-authors also show that a shift to a fully investment-based system would be possible with contributions to PRAs that are less than one third of the current payroll tax rate. However, a mixed system that maintains Social Security while introducing a funded element is probably a more realistic proposition.

Under the mixed system, people would continue to pay the 12.4 percent payroll tax. On top of this, the 2.3 percent of earnings (up to the payroll tax threshold) that is deposited into PRAs by the government would be invested in a 60:40 stock-bond portfolio. The calculations in the paper assume that PRAs earn a 5.5 percent real rate of return (this is the average real postwar portfolio rate of return for such a portfolio, after a 0.4 percent allowance is made for administration costs). The researchers use the same forecasts for average wage growth and changing demographics that the Social Security Administration uses, and take into account the rise in the retirement age to 67 under current law.

With these assumptions, retirees would receive a pension equal to the benefits promised, under current law, by Social Security, plus an extra annuity based on 25 percent of the value of their PRA. The remaining 75 percent of the PRA annuity would be paid to the Social Security Trust Fund. This would cover the costs of supporting an aging population, and mean that the payroll tax would not have to rise. In the early years of this transition, the combined pension would be only slightly greater than the promised Social Security benchmark benefit. However, by 2050, the combined average pension would be 11 percent greater than the benchmark and by 2070 it would be 15 percent greater.

Feldstein and his co-authors carry out 10,000 simulations of the investment risk, based on the postwar experience associated with a 60:40 equity-debt portfolio. The younger the worker at the start of the transition, the greater the risk. But even people aged 21 in 2000 -- who make PRA contributions throughout their working lives -- face only a 10 percent chance of a combined benefit of less than 79 percent of the benchmark benefit when they are 71. They also have a 10 percent chance that the combined pension will be nearly three times the benchmark.

The researchers show that retirees can be insured against all downside investment risk at a very small cost to taxpayers through a government guarantee of the benchmark benefit. This means that the only risk that retirees would face is the upside risk of higher than expected benefits. This is an inter-generational guarantee; the working population bears the risk. But by 2070 -- when the risk is greatest -- the median value of the transfer is equal to just 0.07 percent of earnings that year. There is only a 1 in 10 chance that the transfer will exceed 5.32 percent of earnings.

However, the additional corporate tax revenues generated through the PRA saving mean that in 2070 there is only a 1 in 10 chance that any net transfer will be required at all, and only a 1 in 50 chance that the net transfer will exceed 3.7 percent of covered earnings. Even then, the total burden is less than the rise in the payroll tax to 18.7 per cent if Social Security is left as it is.

In this paper, the researchers assume that if an individual dies before retirement age, the funds in his PRA are divided among the remaining employee accounts (survivor benefits are included in the benefits calculations). However, public discussion and the experience of private pension plans suggest that any Social Security reform that involves investment accounts will also involve leaving a "pre-retirement bequest" if an individual dies before reaching the normal retirement age, and may even involve bequests if the individual dies after retirement. In The Economics of Bequests in Pensions and Social Security (NBER Working Paper No. 7065), Feldstein and Ranguelova investigate the implications of incorporating bequests into a system of PRAs.

They find that pre-retirement bequests would be possible without any reduction of the potential retirement annuity if the PRA savings are increased by one sixth -- from 2.3 percent of payroll to 2.7 percent of payroll in the mixed system described above. They also investigate the possibility of "post-retirement bequests" when individuals die after retirement. A "ten year certain" annuity which provides a bequest whenever a retiree dies before the age of 77 would only require raising the PRA savings from 2.7 percent of payroll to 2.9 percent.

-- Andrew Balls

The Digest is not copyrighted and may be reproduced freely with appropriate attribution of source.
NBER Videos

National Bureau of Economic Research, 1050 Massachusetts Ave., Cambridge, MA 02138; 617-868-3900; email:

Contact Us
Undefined variable: video_footer