The precis from Robert Shiller (2005), "The Life-Cycle Personal Accounts Proposal for Social Security: An Evaluation":
The paper uses historical returns from 1871-2004 to assess the President’s personal accounts proposal. It does 91 different simulations for a worker born in 1990 assuming that he or she experiences the actual returns from 1871-1914, 1872-1915, 1873-1916, all the way through 1961-2004. This sample has an average real stock market return of 6.8% annually, slightly above the 6.5% annual return assumed by the Social Security actuaries.
These historical returns are not, however, a good guide to future returns. The United States economy and stock market performed extremely well over the last century. Many factors suggest this lucky experience is not likely to be repeated: most analysts project slower GDP growth in the next century, the risk premium required for investing in equities may have diminished, and the P-E ratio is very high by historical standards.
The Wall Street Journal recently surveyed 10 leading financial economists, the median projection for the stock market real rate of return in this survey was 4.6% above inflation. This is slightly lower than the median real return of 4.8% in a 15 countries from 1900-2000 surveyed by Dimson et. al.
As a result, the paper also use “adjusted” stock market returns designed to match the median stock return in 15 countries from 1900-2000, this is slightly above the return in the Wall Street Journal survey and is a more accurate projection of future returns.
Life-cycle Portfolio: The paper analyzes a range of potential portfolios. The featured portfolio is a “lifecycle portfolio” designed to capture the President’s proposal. According to the President’s plan, workers would be defaulted in a specific mixture of stocks and bonds. At age 47, workers would automatically be shifted into the “lifecycle portfolio” unless they signed a form to opt out. The President has not specified the portfolio allocation of this account, this paper assumes a benchmark portfolio is invested 85% in equities through age 29 and then phase-down to 15% equity investment by age 60.
Using historical returns, the life-cycle portfolio loses money 32% of the time (i.e., 32% of the time the internal rate of return is less than the 3% real return required to break even in the proposal). The median rate of return is 3.4% annually.
Using more realistic adjusted returns, the life-cycle portfolio loses money 71% of the time and has a median rate of return of 2.6%.
Discussion: These rates of return are considerably below the 4.6% that the Social Security actuaries have assumed for. In addition there is considerably more risk than one would generally associate with previous discussions of “lifecycle portfolios.” The most important reason this happens is that the life-cycle portfolio is invested in higher-yielding assets in early years and lower-yielding assets in later years. Because contributions are made annually, the returns in later years matter much more (i.e., the return in the first year only affects the first contribution but the return in the last year affects all 44 years of contributions). This effect is heightened because the typical worker reaches peak earnings in his or her fifties.
The optimal portfolio for a worker choosing the personal account as a replacement for much of the guaranteed Social Security benefit is considerably different from the optimal portfolio for a worker investing a 401(k) in addition to Social Security. If you have a Social Security benefit that is not subject to market risk, then you can invest your additional savings in a higher return/risk portfolio. But in the President’s proposal, the investments are replacing a large fraction of the existing Social Security benefit. Thus you would not want to invest them in as risky a portfolio.
A worker that has the correct balanced portfolio of stocks and bonds should not even participate in the accounts. Conditional on participating, he or she should invest entirely in bonds in order to avoid changing their current portfolio. Other psychologically constrained workers might benefit from shifting their portfolios more into equities. Social Security design has to take seriously psychological barriers to enlightened saving and investing; workers not subject to these barriers are very different from workers who already do things right. Overall, any proposals to encourage savings and investment should be designed with a variety of different types of workers clearly in mind.