When investment takes place over many periods, the portfolio selection criteria that are valid in the earlier periods narrow to a very small set of short-run objectives for a wide variety of long-run goals. These goals are sufficiently comprehensive to span a continuum of risk attitudes all the way from risk neutrality to infinite risk aversion.

The authors apply criteria drawn from multiperiod investment theory to the construction and rebalancing of portfolios composed of U.S. stocks, corporate bonds, government bonds, a risk-free asset and borrowing. The results indicate that active rebalancing among the major asset classes can substantially improve investment performance. The most conservative portfolio strategy, for example, achieved a geometric average return of 4.45 per cent over the 1934–83 period; a portfolio invested entirely in the risk-free asset over this same period would have returned only 3.51 per cent and would actually have experienced more variability.

Use of leverage did not consistently hinder or help portfolio performance. The best return achieved without leverage was 11.41 per cent. This beat out the riskiest strategy, which relied heavily on borrowing and attained a compound return of 9.89 per cent. But the “growth-optimal” strategy with leverage achieved the highest geometric average—14.99 per cent.