In an analysis of the stock market from 1802 to 1997, John Bogle³ found that rates of return for one-year periods averaged 7% above inflation and annual rates of return for twenty five-year and fifty-year periods averaged 6.7% above inflation. He went on to calculate a standard statistical measure of variability, the standard deviation, as a measure of risk. The standard deviation for one-year periods was 18.1%, for twenty-five year periods was 2%, and for fifty-year periods, 1%. The common statistical interpretation of the standard deviation is that most (about two thirds) of the observations would fall between the average plus or minus 1 standard deviation and practically all observations would fall between the average plus or minus 2 standard deviations. Using this interpretation, an investor who invested in the entire market for many one-year periods could expect to get a return between -11.1% and +25.1% (the mean plus or minus 1 standard deviation) in most years and a return between -36.2% and +42.2% in practically all years. But an investor who invested for periods of fifty years could expect returns of between 5.7% and 7.7% above inflation in most fifty-year periods and between 4.7% and 8.7% practically all periods. At an inflation rate of 3.2%, one would expect that a fifty year investor would almost never experience a return below 7.9% after inflation.

These results are significant for the Social Security system for two reasons. They show that, historically, fairly high long-term returns can be obtained without trying to pick stocks or groups of stocks. Acceptable returns can be expected if funds are invested in the market as a whole. And since most of the investments made by workers in the stock fund would be made over about forty years, workers can take advantage of these long-term returns without any significant risk. Such investments may not realize the high returns possible in a few well-selected stocks, but neither would they suffer the high losses of a bad selection of individual stocks. This is nothing more than the age-old advice to diversify taken to its extreme. During long periods of investments, market returns may vary widely from year-to-year, but ups should balance the downs, and one could reasonably expect to have long-range returns in the range of 8.5% to 13.7%.

This does not mean that there can't be forty-year periods in which the return is lower, even though such periods should be very infrequent. Workers whose investment years happen to coincide with one of those periods could have their retirement accumulations put in jeopardy. Using the Trust Fund as an insurance reserve, the Social Security system should guarantee that, at the time of retirement, accumulations in the stock fund would be no less that which could have been attained with a 5% annual return. This would assure all workers that when they retire, the investments in their stock and bond fund accounts will have earned at least 5% during their work years. Since long-run returns that low would be expected so infrequently, the cost to the Social Security system for assuming this risk should be minimal, and workers would enjoy an investment period which is practically risk free.

However, these happy circumstances do not hold for the retirement years. While for the twenty or so years of retirement the retiree may expect to realize a return of about 10%, it is likely that it could be lower than 7.5% (1 1/2 standard deviations) for some twenty-year periods. (Using standard statistical analysis, this would occur less than 5% of the time). But, more importantly, since retirement income from the stock fund would be paid in the form of a variable annuity, yearly variations in returns from the stock fund would affect yearly retirement pay, and these variations could be quite large in some years. Retirement income would be expected to rise and fall over the retirement years and would probably average out over the years to around 10%. These variations in income from the stock fund would be partially balanced by a steadily increasing income from the bond fund, and in those cases where monthly income fell below the income insured level, income insurance would protect retirees from having very low retirement incomes. While variations in retirement income are not a desirable feature, they represent the cost of the opportunity to get higher total returns over the long run. Still, some retirees may find this risk unacceptable and should be offered the option of converting some or all of the accumulated value of their stock fund accounts into bond fund annuities paying a guaranteed 5%. Retirees can then select the amount of risk they are willing to accept during their retirement years.

In summary, the long run-risk associated with market returns is very low if history is any indication of future market performance. During the last century, through the major wars, the robber barons, the Teapot Dome scandals, the 1929 stock market crash, and the Great Depression, long-range (forty to fifty years) annual returns for the entire stock market have averaged about 7% above inflation with little variation above or below the 7%. Ibbotson Associates found that the lowest return for all the thirty year periods between 1926 and 1999 was 8.47% after inflation. Since investments made by workers covered by Social Security would be made over comparable periods, one could expect workers to enjoy similar appreciation of their accounts in the Social Security stock fund. What little risk of long-term low-market performance exists would be eliminated by a guarantee of a minimum 5% return (or the bond fund rate of return) on stock fund investments.

There are, however, significant short-run risks of low yearly returns during the retirement years. To minimize these short-run risks, the proposed system would balance the stock fund annuities with guaranteed annuities from the government bond fund and would provide income insurance which guarantees a specified minimum retirement income. In addition, retirees would have the option of converting some, or all, of their stock fund accumulations to bond fund annuities which involve no risk at all.

Links: Home     Appendix A     Appendix B     Appendix C     Appendix D     Appendix E     Appendix F    

Send comments to David C. Dellinger.