Reducing the Risk of Investment-Based Social Security

Collars could be used to completely eliminate the risk of future annuities falling below the Social Security benchmark level. Or, they could allow individuals to decide the level of risk they are prepared to accept in return for the possibility of higher retirement annuities.

Private financial markets could provide a minimum guarantee for Personal Retirement Account (PRA) annuities as part of an investment-based reform of Social Security. Under such a system, the "collar" would allow an individual to enjoy the higher returns offered by investing in stocks and bonds -- rather than simply in paying into the pay-as-you-go system -- while managing the market risk.

In Accumulated Pension Collars: A Market Approach to Reducing the Risk of Investment-Based Social Security Reform (NBER Working Paper No. 7861) NBER President Martin Feldstein and Elena Ranguelova show how individuals could, in effect, purchase a "put option" guaranteeing a minimum level of retirement annuity from their PRA. They would pay for it by selling a "call option" on returns above a specified level. The put and call options, which would be priced by the financial intermediary using standard techniques, define the collar and limit the extent of uncertainty of the PRA returns.

Feldstein and Ranguelova examine a number of ways in which collars could guarantee that the combination of traditional Social Security benefits and the PRA annuity could match the "benchmark" case -- the level of future benefits projected in current Social Security law. Their analysis is based on simulations using 50 years of historical data on the level and variability of market returns.

Collars could be used to completely eliminate the risk of future annuities falling below the Social Security benchmark level. Or, they could allow individuals to decide the level of risk they are prepared to accept in return for the possibility of higher retirement annuities. The greater the level of protection that the individual chooses, the more of the potential excess returns he or she would agree to give up.

The results of the Feldstein/Ranguelova analysis show that the combination of the current payroll tax and an additional PRA contribution rate equal to 2.5 percent of covered earnings would guarantee the benchmark Social Security benefit. Raising the PRA contribution to 3 percent means that future benefits would fall in the collar range of 100-145 percent of the benchmark. Reducing the guaranteed level of benefits to 90 percent of the benchmark with a 2.5 percent PRA contribution rate means that future benefits would fall in the collar range of 90-150 percent of the benchmark.

In general, a higher PRA savings rate reduces the risk that the PRA benefits will fall below the benchmark level, and so substantially increases the maximum amount of the annuity that the individual can keep. It also increases the average level of PRA benefits for any given options strike price. Reducing the level of guaranteed annuity slightly -- a lower put option strike price -- increases the maximum amount of the annuity that the individual can keep substantially.

Another possibility would be to combine a minimum guarantee and share the upside gain above some threshold level. For example, the individual could buy a put option that guarantees that benchmark benefit and in exchange sell a put option that gives the annuity holder 50 percent of the gains above the threshold level. With a PRA contribution rate of 3 percent, this would allow the individual to keep all of the annuity up to 119 percent of the benchmark and half of all returns above 119 percent.

Introducing PRAs, alongside the existing system, means that the U.S. pension system could be made solvent without the hefty increases in payroll taxes that will be required under the current rules, owing to the higher returns offered. The use of collars could meet one of the main criticisms of an investment-based system, the exposure of individuals to excessive market risk. The analysis extends several earlier studies by Feldstein and other researchers, based on historic data and the demographic and economic forecasts of the Social Security actuaries.

The results are based on a standard portfolio made up of 60 percent stocks and 40 percent corporate bonds. In the 50-year period from 1945-95, such a portfolio delivered an average real return of 5.9 percent and a standard deviation of 12.5 percent. The researchers subtract 0.4 percent for administrative costs and use the remaining 5.5 percent real return to simulate PRAs and retirement annuities for 10,000 individuals. The shift to PRAs would be phased-in over a number of years, with those retiring in earlier years more dependent on Social Security benefits and those in later years more dependent on PRA returns. The researchers focus on the long-run position when the PRA accounts for one-third of the total benefits.

This market-based approach to insuring the level of retirement income could be completely voluntary, with individuals deciding what level of guarantee they want, depending on their risk preferences. Alternatively, the government could mandate a minimum level of protection, for example, a minimum guarantee of 90 percent of the benchmark.

-- Andrew Balls