Essays in Household Finance and Consumer Protection

Abstract

This dissertation studies consumer decision-making and how household outcomes are shaped by technological change and regulatory context.

Chapter 1 (with John Beshears and Kartik Vira) approaches the study of household finance by analyzing household consumption decisions as they manage the day-to-day ebbs and flows of their liquid assets and debts. Using high-frequency account-level data from a retail bank in Australia, we show that consumption declines in the week after households make their monthly mortgage payment. Focusing on households that receive their income on a fortnightly basis, consumption drops more sharply over the course of a pay cycle when the cycle contains a mortgage payment than when the cycle does not contain a mortgage payment. This behavior is inconsistent with a neoclassical model augmented by incorporating liquidity constraints and is most consistent with models of present bias or mental accounting.

Chapter 2 (with Jens Frankenreiter and Dan Svirsky) focuses on consumer data usage by online service providers. We argue that privacy policies play an important role not only in describing current data practices but also in allocating among websites and consumers the power to decide whether a website can use consumer data in novel ways. By adapting standard models of incomplete contracts to privacy policies we explain the role of policies in allocating residual data rights and the potential role of regulation in limiting the allocation of rights to firms, similar to the approach of the E.U.’s General Data Protection Regulation (GDPR). We then use the model to explain how U.S. firms reacted to the GDPR, showing that U.S. websites with E.U. exposure are more likely to change their U.S. privacy policies to have more permissive modification rules akin to allocating residual data rights to firms.

Chapter 3 (with Jann Spiess) discusses tensions between old law and new technology in the context of consumer credit markets. The ability to distinguish between people in setting the price of credit is often constrained by antidiscrimination laws. These laws were developed when human discretion played a central role in credit allocation and may be ineffective as pricing increasingly relies on intelligent algorithms that extract information from big data. Using a simulation exercise based on real-world mortgage data, we highlight the shortcomings of legal approaches that focus on algorithmic inputs and pricing rule interpretation. Instead, we argue that testing pricing outcomes is a promising path forward for evaluating discrimination claims in the algorithmic setting.