"The life-cycle accounts proposal for Social Security reform has been justified by its proponents using a number of different arguments, but these arguments generally involve the assumption of a high likelihood of good returns on the accounts. A simulation is undertaken to estimate the probability distribution of returns in the accounts based on long-term historical experience. U.S. stock market, bond market and money market data 1871-2004 are used for the analysis. Assuming that future returns behave like historical data, it is found that a baseline personal account portfolio after offset will be negative 32% of the time on the retirement date. The median internal rate of return in this case is 3.4 percent, just above the amount necessary for holders of the accounts to break even. However, the U.S. stock market has been unusually successful historically by world standards. It would be better if we adjust the historical data to reduce the assumed average stock market return for the simulation. When this is done so that the return matches the median stock market return of 15 countries 1900-2000 as reported by Dimson et al. [2002], the baseline personal account is found to be negative 71% of the time on the date of retirement and the median internal rate of return is 2.6 percent"--National Bureau of Economic Research web site.