Time-Varying Expected Returns and Investor

Abstract

What is the optimal rebalancing policy for a portfolio’s equity and bond holdings? The classical answer, building on the seminal contributions by Mossin (1968), Merton (1969, 1971), Samuelson (1969) and others, is that investors should hold a constant proportion of bonds and equities in their portfolios. During the 1980s and 1990s, empirical and theoretical research began to challenge the fundamental premises for this result. It has, for example, been documented that risk premia and expected returns vary over time. An implication of this is that the Mossin-Merton-Samuelson result is incompatible with market clearing. The risks associated with a portfolio with a constant proportion of bonds and equities might be very different from what was previously understood. A framework to analyse the expected returns and risks associated with a dynamic rebalancing regime must be based on models that account for investors’ risk preferences and time-varying risk premia. Analytical frameworks based on recursive risk preferences, long-run risk and time-varying volatility give a rationale for a dynamic rebalancing regime where different investors put different weights on different risk factors, such as short-run risk, long-run risk and volatility. Investor “preference heterogeneity” should not be interpreted literally as heterogeneity in “preferences”, but rather as reduced-form representations of other forms of heterogeneity associated with, for example, market frictions or market incompleteness.