Essays on heterogeneity over the business cycle

Abstract

The three chapters of this PhD thesis look at how heterogeneity and business cycles interact. The first chapter features heterogeneity in the form of multidimentional tasks in occupations, and the composition of job-to-job movers over the business cycle. The second and third chapters focus on heterogeneity in consumption and saving decisions over the business cycle. Chapter 1 presents a jointly co-authored paper in which we match UK LFS employment transition data to US O*NET data on multidimensional tasks. We present two measures to capture task and skill differences between occupations. We document a set of stylised facts relating to the task and skill content of job transitions over the business cycle in the UK. During recessions, the overall number of transitions decreases and the task content of transitions becomes more similar both in terms of tasks and overall skill requirements, relative to non-recessionary periods. However, we find that the magnitude of all the estimated relationships is very small, and partially offset by selection effects in the types of people who make job-to-job moves during recessions. We do find that those who upskill tend to capture greater wage increases than those who down-skill or whose skills are unchanged. However, we find no cyclical relationship in the wage changes of those who up-skill, down-skill or with no skill change. Chapter 2 describes the key features of a Bewley-type heterogeneous agent incomplete market models with aggregate and idiosyncratic uncertainty, the Krusell and Smith (1999) model. It reiterates the common result that, in its benchmark form, the model does a poor job of fitting the empirical wealth distribution. It shows that this result is robust to large changes in the key parameters. I show that a commonly used addition to improve this fit, dispersion in discount factors, implies a contradiction when the model is calibrated to US PSID data before and during the Great Recession. In particular, it implies that the preferences of agents shifted substantially, resulting in a shift of individual policy functions for consumption. However, internal cyclical dynamics of the model imply only a movement in mass along the policy function in recessions. I also show that fitting the empirical fraction of individuals with zero or negative wealth implies that the borrowing constraint should have loosened in the Great Recession, contrary to empirical evidence that the availability of credit fell. Chapter 3 takes the contradictions of chapter 2, and asks whether the increase in the aggregate marginal propensity to consume could be explained by individual policy functions for consumption shifting over the Great Recession. The mechanism I examine is whether an increase in the variance of income shocks could have caused a shift in the consumption function. This mechanism is also known to shift the consumption function in the model of chapter 2. Using PSID data on consumption and income, I apply the two-step method of Blundell et al. (2008) to first extract the transitory components of consumption and income for individuals in a pre-recession and recession sample. I then use an instrumental variables regression to estimate the marginal propensity to consume out of transitory income at the income quintiles of the distribution. I find that the aggregate marginal propensity to consume increased over the recession, and that the marginal propensity to consume varies across the distribution. However, I do not find evidence that marginal propensities to consume shifted across the income distribution, consistent with consumption functions not shifting. This suggests that increased variance in transitory income is unlikely to explain the contradictory findings of chapter 2.