Labor income accounts for about two thirds of national income in the U.S. and, since the seminal work of Mayers (1973), it has been assumed to play an important role in theoretical asset pricing. In studies such as Bodie, Merton, and Samuelson (1992), Danthine and Donaldson (2002), Qin (2002), Santos and Veronesi (2006) and Parlour and Walden (2010), risky labor income or more generally, human capital risk affects investors’ portfolio decisions, which in turn has general equilibrium asset pricing implications. Broadly, the theory suggests that the behavior of capital markets can only be understood together with labor markets. More specifically, the theory suggests that an important function of capital markets is to allow investors to hedge their labor income risk.
Are investors’ portfolio decisions affected by their labor income risk? Studies that use aggregate labor income find mixed evidence. Fama and Schwert (1977) find that adding a labor factor does not improve the performance of the unconditional CAPM. By contrast, Jagannathan and Wang (1996) find that an aggregate labor factor significantly improves the performance of a conditional CAPM in explaining the cross section of expected returns (see also Palacios-Huerta, 2003). Given the highly aggregate data, noisy measurements, and incomplete real-world markets, it seems unlikely that an approach based on aggregate data can lead to a conclusive answer.
In this paper, we approach the question by using data at the individual household level. Specifically, we study panel data on employment and portfolio holdings of a large subset of the Swedish population between 1999 and 2003, and examine whether there is a relation between workers’ wage structure (measured by wage level and volatility) and their portfolio holdings of risky assets.
We find that shocks to workers’ wage volatility affect their portfolio holdings of risky assets. This is consistent with the idea that human capital risk affects portfolio decisions. For example, households adjust their portfolios in response to job changes. This hedging effect, which is highly statistically significant, is especially strong for job changes that lead to large changes in wage volatility: a household that experiences an increase in wage volatility by 20% decreases its portfolio share of risky assets by 20%. This means that a household going from the industry with the least variable wage in the sample (recycling metal waste) to the industry with the most variable wage (fund management) ceteris paribus decreases its share of risky assets by up to 35%, or 15,575 USD.
Download
Hedging Labor Income Risk
