Not all risks that households face can be insured against or avoided by taking some actions. Some risks have simply to be borne. These risks are called ‘background risks’ precisely because they are part of the environment where decisions are taken. For many households, the most prominent background risk they face is labour income risk. This cannot be insured against and cannot be easily avoided.
How do households respond to background risks?
This question has a long history in macroeconomics and finance. A large literature has studied how the presence of uninsurable labour income risk affects the patterns of individual and aggregate savings, consumption and portfolio allocation over the life cycle, as well as the behaviour of asset prices. In theory, under plausible circumstances, consumers who face uninsurable labour income risk respond by accumulating precautionary savings, raising labour supply, or more generally changing the pattern of human capital accumulation (e.g. Levhari and Weiss 1974). Furthermore, people tend to reduce exposure to risks that they can avoid (Kimball 1993). The idea is that risk-averse individuals may control overall risk exposure by reducing their exposure to risks that are potentially avoidable. In particular, they should change the asset allocation of their financial portfolio by lowering the share invested in risky assets, thus tempering their overall risk exposure. Motivated by these theoretical predictions and the undisputable importance for many households of labour income, one strand of research has incorporated background risk into theoretical models of (consumption and) portfolio allocation over the life cycle and explored its ability to help reproduce patterns observed in the data. Another strand has tried to assess the empirical relevance of uninsurable income risk in explaining the heterogeneity of portfolio choice observed in the data. A fair characterisation of both strands of literature is that the effect of background labour income risk on portfolio allocation, though consistent with the predictions of the theory, seems to matter little in practice. As a consequence, the background risk channel seems to have lost appeal as a quantitatively important determinant of household portfolio choices or as a candidate explanation for asset pricing puzzles (such as the equity premium puzzle).
Addressing three main challenges in measuring background risk
In a new paper (Fagereng et al. 2016), we revisit the importance of background risk for portfolio allocations. We argue that the empirical literature faces three important challenges when trying to measure the effect of background risk.
- First, exogenous variation in background risk is hard to come by.
A popular solution is to measure background risk by the volatility of labour incomes. But, as recent research argues, a good chunk of such volatility reflects choice and not risk. For example, fluctuations in earnings may reflect deliberate variation in workers’ hours or effort; this is not risk, though an external observer can mistakenly interpret it as such.
- Second, unobserved preference heterogeneity may understate the effect of risk on portfolio choice.
For example, unobserved risk aversion determines both the income risk people face (through occupational choice) as well as the composition of one’s asset portfolio. Without variation over time in portfolio choices, it is hard to disentangle the effect of variation in risk from the effect of preference heterogeneity determining both risk and portfolio composition.
- Finally, most of the empirical literature is based on survey data on assets.
These are notoriously subject to measurement error and rarely sample the upper tail of the distribution (which is key, given the enormous skewness in the distribution of wealth). Moreover, both survey and administrative data show that several investors choose to stay out of the stock market. All these factors – some technical, others more substantive – tend to produce a downward-biased estimate of the effect of labour income risk, leading us to dismiss its importance perhaps prematurely.
To tackle these three empirical challenges, we use a rich administrative dataset with employer-employee data that we merge with data on household wealth holdings for the whole population of Norway. First, we obtain exogenous variation in the labour income risk that people face by computing the component of wage variation that is due to shocks to the firms that employs the worker. While variation in wages may partly reflect choice, the firm-related component does not and hence qualifies as true background risk. Second, we use long panel data on firms and their workers to deal with unobserved heterogeneity, thus circumventing the second challenge mentioned above. These data allow us to compute financial portfolio shares in risky assets at the household level and we thus study whether and how much workers adjust their financial investments to background risk.
We document three important findings.
- First, if we use the same methodology that has been used in the literature to estimate the effect of labour income risk, we reproduce the small marginal effect of background labour income risk on the portfolio allocation to risky assets that characterises the empirical literature. However, if we use firm-related variation in wages to remove the downward bias induced by the fact that not all variation in labour income is risk, we find that the effect is about 20 times larger.
This suggests a very large downward bias in prevailing estimates of the effect of background risk and, in principle, a potentially more important role for human capital risk in explaining portfolio decisions and assets pricing.
- Second, we find that the marginal effect of background risk varies considerably across individuals depending on their level of wealth.
The portfolio response of individuals at the bottom of the wealth distribution – those with little buffers to face labour income uncertainty – is twice as large as that of the workers with median wealth; the effect gets smaller as wealth increases and drops to zero at the top of the wealth distribution. Labour income risk is essentially irrelevant for those with large amounts of assets despite the fact that their compensation is more sensitive (as we document) to firm shocks.
- Third, using the estimated parameters, we provide some bounds on the economic effect of labour income risk.
The economic effect depends both on the marginal effect of background risk and on its size. We identify the latter as the sum of variation in wages due to exposure to the firm risk and variation that is not the reflection of workers’ choices. Overall, we find that the economic effect of background risk is small: individuals with the average amount of background risk have a share of risky assets in portfolio that is a quarter of a percentage point smaller than those with no background risk whatsoever. These numbers suggest that, when quantifying the effect of background risk on portfolio choice, our conclusions are not different from what found in the existing literature, despite the larger sensitivity to risk that we estimate. The key to understanding this apparently puzzling result is that the effect of risk on portfolio choice depends on two things: the response of portfolio choice to a change in the risk and the size of the risk itself.
Our estimates suggest that the true response is larger and the true amount of background risk people face smaller than typically found. In the existing literature the opposite is true: estimated risk is overstated and (because of this) the sensitivity is downward biased, thus reaching the right conclusion but for the wrong reasons. In turn, we show that wage risk is contained because firms provide workers with substantial insurance. If firms were to share shocks equally with their workers, the latter would reduce the demand for risky financial assets substantially, particularly for low-wealth workers. In sum, the economic importance of human capital risk crucially hinges on the insurance role of the firm and the amount of assets available to the individual to buffer labour income shocks.
Fagereng, A, L Guiso, and L Pistaferri (2016), “Back to Background Risk?”, working paper.
Kimball, M S (1993), "Standard Risk Aversion", Econometrica 61, 589-611.
Levhari, D and Y Weiss (1974), "The Effect of Risk on the Investment in Human Capital", American Economic Review, 64(6), 950-963.