Earnings Risk and the U.S. Housing Market

Abstract

The goal of this dissertation is to develop, refine, and employ empirical measures of earnings risk—especially permanent risk—and determine their effect on behavior, with applications specific to the recent mortgage bubble. Chapter 1 aims to identify covariates of risks associated with permanent and temporary earnings shocks. The distinction is significant both because permanent shocks are more consequential and because measurement error would contaminate only measures of temporary risks. Generalizing methods used in previous work, we allow risk to vary both across individuals and over individuals’ careers, so they could be used to study behavioral responses to risk even with individual fixed effects. We find the majority of overall earnings risk is temporary, yet permanent risks vary notably across race and education. Chapter 2 uses the measures developed in chapter 1 to analyze the information structure in the mortgage market during the first decade of the 2000s. A feature of last decade's mortgage crisis was that the credit risk of borrowers was not accurately priced into their mortgages (e.g., the increased risk of delinquency was not associated with a higher interest rate). I confirm this belief by using a measure of labor earnings risk, a major source of risk at the household level. I find that borrowers who are granted a mortgage, labor earnings risk measured at origination predicts a greater probability of delinquency, yet its relationship to loan terms is weak. Further results also lend evidence to theories involving information problems: increases in permanent labor earnings risk measured at origination (i) increase borrowers' perceived risk of falling behind on payments and (ii) influences lenders' restructuring decisions. This disconnect is not due to supply side explanations I can indirectly test such as market distortions induced by lending policies. Since I cannot provide clear evidence of lenders' ability to appreciate this risk at the point origination, room remains to test the importance of the incentive for lenders to resell mortgages in the secondary market. Chapter 3 investigates the divergence of the gap in homeownership rates that occurred over the entire course of the recent housing-related business cycle. After controlling for a rich set of factors determining the willingness and ability to own a home, I find that most of the change in the white/non-white homeownership gap, particularly during the housing bust, can be explained by observable factors. The main contributors are marriage and fertility, especially during the boom; geography, income instability, and education also contribute during the bust.