Dynamic Portfolio Decisions with Time-Varying
Asset Returns and Labor Income

Abstract

This article establishes a basic framework on a portfolio model for a financial fund that manages savings for retirement income and for which the inflows to the fund are asset returns as well as labor income. It suggests a practical way to rebalance portfolios as new information on returns and labor income become available. The rebalancing of the portfolio is guided by estimated low-frequency components of asset returns and labor income.

For the actual saving and asset allocation decisions, the author uses a new numerical procedure as guidance. First, low-frequency components for asset returns and labor income are estimated through spectral analysis. Second, after filtering out the low-frequency components of actual U.S. time-series data on asset returns and labor income data, the procedure is used to solve for dynamic saving and asset allocation decisions. Third, along the line of Campbell and Viceira [CV, 2002], he explores the effects of varying risk aversion and varying time horizon, across investors.

Whereas CV, in their solution technique, use a constant consumption–wealth ratio and a constant equity premium, the author allows both to be time varying. The optimal saving decisions, asset allocation, welfare of investors, as well as fate of the wealth of the investors are explored for the actual time-series data. The dynamic portfolio decision model allows for online decisions, as the low-frequency data on asset returns and labor income are available in real time. The method is set up in a way so that it also helps to make retirement fund decisions for various types of investors.