Abstract
Executive Summary. A bootstrap analysis is used to estimate confidence intervals for the optimum level of real estate investment in mixed-asset portfolios over a five-year holding period. The data used in the study extends from 1973 to 1994. Lengthening the holding period to five years provides much narrower confidence intervals in comparison to those generated in earlier studies that examined shorter holding periods of one year or less. The optimum amount of real estate in mixed-asset portfolios was remarkably stable at all levels of investor risk preference along the efficient frontier when investors held their portfolios for five years. Furthermore, the estimates of optimum real estate composition produced by the bootstrap are highly consistent with the behavior of pension fund managers. This suggests that the markets are efficient and fund managers are exactly where they should be in terms of real estate investment.