Measuring Long-Run Consumption Risk

A large body of recent research, beginning with Bansal-Yaron (2004) has hypothesized long-run risk in consumption, in conjunction with Epstein-Zin preferences, as an explanation for the joint behavior of asset prices and economic aggregates. Finding direct evidence of this risk has proved to be challenge. This paper argues that there is strong evidence for productivity growth as a source of long-run risk in consumption, provided the nonstationarity in hours of work is appropriately controlled for. The paper provides a model that can explain the notable increases in both consumption expenditure as a share of GDP and in per capita work hours, and at the same time provide a rationale for using consumption adjusted for work hours in the asset pricing kernel. The idea is that the increase in work hours represents a substitution of market labor for non-market (or "household") labor, and is naturally coupled with a switch from non-market (i.e. homeproduced) consumption to market consumption. The model implies that market consumption relative to market hours of work ("adjusted consumption") is a better approximation to true consumption than market consumption itself. Empirically, adjusted consumption exhibits the same trend behavior as productivity growth, modeled (as in Kahn and Rich, 2007) as a regime-switching process in the mean growth rate. The regime-switching process has very similar properties to the long-run risk component in Bansal-Yaron.