How to Forecast Long-Run Asset Returns

Abstract

Abstract This article shows how to use components to estimate an expected return for various asset classes, including stocks, bonds, cash, real estate, small stocks, foreign stocks, and foreign bonds. In principle, these components can be used to estimate expected returns on any asset class. Of the many kinds of asset allocation models in use, perhaps the most common is that which uses historical and current data to form long-term equilibrium capital market expectations. These expectations are combined with an analysis of the liabilities of the investor (typically a pension fund). The output consists of policy, or normal, weights for asset classes held or desired to be held by the investor. This article shows how to use historical and current data to form expectations. Asset allocation models require as inputs three kinds of data for each asset class: This list of required inputs also applies to tactical asset allocation models and other variations on the theme of setting asset class weights. In tactical models, however, the expected return is typically a short-term forecast that differs from the long-run equilibrium return. The forecasts of standard deviation and correlations may also differ from long-run equilibrium in tactical models. This article focuses on the policy or strategic, rather than the tactical, situation.