Academic journal article Financial Services Review

Optimal Asset Allocation in the Presence of Nonfinancial Assets

Academic journal article Financial Services Review

Optimal Asset Allocation in the Presence of Nonfinancial Assets

Article excerpt

Abstract

In this paper, a comprehensive mean-variance model, which includes all major nonfinancial assets (housing, human capital, and private business) besides financial ones is calibrated with empirical data to generate optimal asset allocations between stocks, bonds, and cash. The model is used to investigate the effect that each of the nonfinancial assets has on the optimal mix of financial assets. Among others, it generates the optimal portfolio allocation for investors working in different industries and living in different cities. The model is also able to rationalize a popular investment advice recommending decreasing share of stocks in financial portfolios with increasing age.

© 2008 Academy of Financial Services. All rights reserved.

JEL classification: G11

Keywords: Asset allocation; Mean-variance model; Nonfinancial assets

1. Introduction

The importance of nonfinancial assets for optimal portfolio allocation has been known in the financial economics literature at least since the works of Mayers (1972) and Fama and Schwert (1977), who among the first investigated the effect of human capital on portfolio choice. Even though nonfinancial assets are rarely traded (houses or private business) or not traded at all (human capital), they are included into portfolio choice because their returns are correlated with returns on financial assets and because they are by far the largest assets in households' possession. As reported in Table 1 (see section 3.1.), based on the 2004 Survey of Consumer Finances, three major nonfinancial assets-human capital, housing, and private business-constituted almost 82% of the total assets for the average household in 2003.1

It was not until the last decade, however, that the asset allocation models incorporating nonfinancial assets, such as human capital, private business, and housing, became really popular. Human capital (in the form of labor income) is taken into account in the models of Campbell and Viceira (2002), Heaton and Lucas (200Ob), Gomes and Michaelides (2004). Human capital combined with life insurance enters in the asset allocation models of Huang, Milevsky and Wang (2005) and Chen, Ibbotson, Milevsky and Zhu (2006). Cocco (2005) and Waggle and Johnson (2003) include house value in their asset allocation models. Yao and Zhang (2005) incorporate both labor income and housing. The value of proprietors' business enters in the models of Heaton and Lucas (200Ob) and Willen (2003). Faig and Shum (2002) model portfolio selection for investors with "personal illiquid projects," which can include housing or private businesses.

These models, however, usually operate with a limited set of assets, such as one risky asset (stock), one riskless asset (cash or riskless bond), and one nonfinancial asset. This is understandable since those models are usually dynamic. While being more realistic than the static models of portfolio choice, they would have been extremely difficult or even impossible to solve if they included more asset classes.

In this paper, I employ the mean-variance asset allocation model, augmenting the familiar financial assets-stocks, bonds, and cash-with all major nonfinancial assets, such as housing, human capital, and private business. The model generates the optimal weights of financial assets, while the weights of the nonfinancial assets are taken as given, based on household empirical data from the Survey of Consumer Finances. The model can be readily re-estimated for investors with different weights. Asset returns, the means, standard deviations, and correlation matrix, which enter the model as parameters, are taken from the actual historical data and can vary according to each investor's circumstances. In fact, I do this in sections 4.3 and 4.4. Historical asset returns come from the following sources: (1) Ibbotson Associates-historic returns on financial assets; (2) National Income and Product Accounts (NIPA)-aggregate per-capita wages, salaries, and proprietor income needed to calculate the returns on human capital and private business; (3) the Office of Federal Housing Enterprise Oversight (OFHEO)-quarterly U. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.