Most households have a considerable proportion of their wealth in nonfinancial assets. The value of human capital, for example, has been estimated to be at least four times as large as the value of all other assets combined. Around two-thirds of US households have been homeowners since 1965, and residential real estate can be a significant household investment.

Economic theory suggests that investors should consider the risk of these assets when making investment decisions: Investors with riskier nonfinancial assets should construct less risky portfolios, and vice versa.

While there is much literature on what investors ought to do, the authors set out to investigate the extent to which participants in defined-contribution (DC) plans actually consider nonfinancial asset risks in their investment portfolio decisions.

The authors test two hypotheses:

1) Investors with riskier human capital will have lower equity exposure than those with less risky human capital.

2) Investors living in states with more volatile residential real estate prices will have lower equity exposure than those in states with less volatile home prices.

They obtain 401(k) allocations from a US recordkeeper as of three annual end dates for each participant: 31 December 2016, 2017, and 2018. They focus on proactive investors (i.e., those who build portfolios themselves rather than defaulting into an investment). In total, 36,755 of these investors are included in the study. The risk level of all the investments held by the study participants is identified for each year, using data on equity allocation from Morningstar.

The authors use panel regression to investigate whether equity allocations (a proxy for investment risk) differ by industry and US state of residence.

They then examine the risk of nonfinancial assets and run an additional set of regressions to investigate whether the participants in their sample are investing rationally:

• Human capital risk in 19 industries is measured with data from the US Bureau of Economic Analysis from 1998 to 2017 for each industry’s total compensation (equivalent to total industry wages multiplied by total employment).

• Residential real estate risk is measured with monthly return data by state on the Zillow Home Value Index for All Homes Time Series from 1996 to 2017.

The authors determine that equity risk levels vary by industry of employment and US state of residence.

They also find that as the risk from nonfinancial assets increases, investors have less equity exposure, consistent with economic theory. Their results suggest that human capital risk has a stronger influence on equity allocation than location risk—but the coefficients for both are statistically significant.

However, they note that, according to their results, the human capital adjustments would lead to a range of around eight percentage points (pps) in equity differences in portfolios, and the location adjustments, to a range of around two pps. This is significantly lower than optimal differences noted in past research, which raises questions about the economic significance of the findings.

The authors’ findings suggest that proactive investors do generally consider the riskiness of their human capital and real estate when constructing portfolios. However, many may not be giving nonfinancial assets enough weight.

Investment professionals who run DC plans could take steps to improve participant outcomes by explicitly considering nonfinancial assets—potentially defaulting participants into more appropriate target-date funds, depending on their home equity and expected future earnings streams. When allocating assets within their DC plans, investors should not allow their portfolios to become “sticky,” but should adjust them according to changing nonfinancial circumstances.