Luigi Briglia

This paper investigates the role of surprises in expenditure and income, in explaining household wealth dynamics. Using the Panel Survey of Income Dynamics (PSID), I estimate a flexible semi-structural consumption model to identify income and expenditure risk and their link to observed wealth dynamics. Deviations between observed consumption and its empirical prediction counterpart can be modeled as an autoregressive stochastic process indicating the existence of persistent shocks to household expenditure. Both income and expenditure risk affect wealth dynamics, with the latter assuming more relevance for households at the bottom of the wealth distribution. Moreover, while income risk can predict episodes of falling into non-positive net worth positions, it is expenditure risk that plays a bigger role in the persistence of these low net worth statuses. I propose a theoretical framework to model expenditure risk, exploring its properties wrt income risk, considering both shocks to marginal utility and resource-draining shocks to expenditure. Marginal utility shocks hitting only a fraction of household non-durable consumption provide the best fit explaining wealth dynamics around zero net worth, suggesting a fundamental difference in the nature of income and expenditure risk; the latter capturing a persistent change in the household intertemporal trade-off that does not lie in the same space of the distribution of available resources across time and states of the world.

We pose a new rationale for precautionary savings based on type one extreme value preference shocks. In this framework, the errors of the Euler equation that predicts consumption based on observables have the same properties as the data: the log errors are an affine-increasing function of cash in hand. This is not the case with the more standard shocks to the marginal utility of consumption. We characterize the solution to the households’ problem showing how the familiar formulas from static and discrete settings adapt to dynamic and continuous environments. We also show how the existence of these shocks provides utility-enhancing consumption opportunities that justify additional savings beyond those generated by environments with shocks to earnings, something that does not happen with standard shocks to marginal utility. The properties of the Euler equation errors allow us to estimate jointly risk aversion and the variance of the preference shocks.