Household Earnings Risk and its Impact on Consumption and Portfolio Decisions
Household labor earnings are unequal and risky. Their persistence and the distribution of the shocks they face depend on age, birth cohort, and the position of a household in the earnings distribution. Understanding these features is key to explain household consumption decisions, the allocation of household savings between different asset classes, and the role of different insurance mechanisms in smoothing out those risks. Using US survey data and Dutch administrative data, I document that household earnings are less persistent for the young and for the income-poor and that shocks to earnings are infrequent and negatively skewed. Although the tax and transfer system insures part of these risks, particularly in the Netherlands, they are also present in disposable income. I then turn to evaluating the implications of these rich features by comparing, in the context of a standard life-cycle model, a flexible earnings process that incorporates them against a canonical process with constant persistence and normal shocks. I find that considering richer earnings dynamics helps us to better understand the evolution of cross-sectional consumption inequality over the life-cycle and the pass-through of persistent earnings shocks to consumption.
Many of these features of earnings have changed over time. Using US data, I document that earnings are more unequal and riskier for younger generations. I argue that lower initial and lifetime earnings for the income-poor and larger earnings variability across the board are key to explain the reduction in homeownership rates between the generations born in the 1940s and 1980s. I show the relevance of this mechanism in a flexible life-cycle model with risk-free assets, stocks, houses, and mortgages, and correlated idiosyncratic and aggregate risks. Changes in financial constraints also matter: looser mortgage requirements helped the young buy houses, and lower participation costs rationalize higher stock market participation.