Book Description
This dissertation consists of three chapters, two of which address consumer choices and tastes and one that addresses risk aversion in lending. The chapters are organized as follows: Chapter 1, "Tastes in the United States: Convergence or Divergence?": This study investigates how tastes for consumption goods across the United States have changed in the 2000s. Using Nielsen Homescan barcode-scanner data, I find evidence of an overall divergence in food tastes from 2007 to 2016 between households of different regional markets and different income, education, and race groups. There is evidence of convergence between rural households, while urban/rural groups and some race groups do not show an overall change in taste differences across groups. Across regional markets, political distance explains taste differences better than geographical distance.Chapter 2, "Human Capital and Quality Choice": This paper argues that human capital increases a consumer's ability to evaluate and appreciate the quality of information-intensive goods. We show theoretically that consumers with more human capital have steeper Engel curves for higher-quality varieties. We test this prediction using Nielsen Homescan data and focus on organic milk consumption in the United States. Using education as a proxy for human capital, we show that households with college-educated heads have steeper Engel curves for organic milk than those without college-educated heads.Chapter 3, "Risk Aversion in Lending Following the Financial Crisis": Following the 2007-2008 financial crisis, bank lending dropped dramatically, and the decline persisted well into the recovery period. This paper attempts to explain the drop in lending using a consumer default model with risk aversion in loan pricing. The version of the model with risk aversion is compared to one that uses the standard assumption of risk-neutral financial intermediaries to see if a model with risk aversion more accurately predicts lending in the post-crisis period. The risk-averse pricing model consistently generates more accurate interest rates than the risk-neutral pricing model, while the risk-neutral pricing model more accurately predicts the debt-output ratio.
