Job Market Paper
Assessing Bank Deposit Market Power Given Limited Consumer Consideration LinkAccurate assessments of bank deposit market power are essential for antitrust and monetary policy. Regulators and researchers currently assess market power under the assumption that all consumers consider every bank operating in a given geographic region. However, in practice, consumers only consider a small fraction of the available banks. I propose a new model of bank deposit competition that specifically accounts for this limited consideration. I estimate the model for the twenty largest US Metropolitan Statistical Areas over 2004 to 2018 using rich data on bank deposit interest rates, bank advertising, and the distance between consumers and bank branch locations. I find that accounting for consumers' limited consideration of banks shows that current assessments underestimate the market power of online direct banks, poorly capture how market power contributes to heterogeneous bank pass through of federal funds rate changes, and support an inefficient proposal for strengthening bank deposit competition. Whereas current assessments recommend further bank deregulation, I demonstrate that a better plan for strengthening competition is to facilitate the launch of an online direct bank by a major technology company.
Heterogeneous Effects of Bank and Mortgage Company Advertising: New Evidence from Television Commercial Content LinkI apply machine learning algorithms to extract rich text and image features from over 20,000 bank and mortgage company television commercials that aired between 2004 and 2015. I document how the commercials highlight significantly different banking, mortgage, and home equity products using diverse actors and widely varying backdrops. I then use discontinuities in advertising exposure created by television market borders to explore the heterogeneous effect of this commercial content on consumers' financial decisions. As a first pass, I examine the extent to which actors' racial backgrounds impact consumers' mortgage refinancing decisions. I find that a 1% increase in mortgage commercials featuring white actors causes a 0.16% increase in the fraction of white homeowners who apply for refinancing. In contrast, black homeowners are not more likely to apply for refinancing when more commercials feature black actors and are 0.07% less likely to apply when there is a 1% increase in commercials with only white actors.
Finding Exogenous Variation in Data with George Gui and Ali Hortacsu LinkWe reconsider the classic problem of recovering exogenous variation from an endogenous regressor. Two-stage least squares recovers exogenous variation through presuming the existence of an instrumental variable and then uses this variation in a second stage to consistently estimate a parameter of interest. We propose a new method for finding exogenous variation in certain settings where the regressor is a mixture random variable. Here one of the components of the mixture can provide sufficient exogenous variation to identify the parameter of interest. We demonstrate our method through finding pricing experiments concealed in grocery store scanner data and display experiments hidden in hotel search data.
How Can Research Registries Be Improved? An Examination of the AEA RCT Registry with Jonathan Libgober and John List LinkSeveral scientific disciplines have launched research registries for randomized controlled trials (RCTs) in order to address academic publication bias. However, current registry design is not informed by a model of a researcher's incentives to register. As such, common registry features, including allowing researchers to register an RCT even after its completion, may be inefficient. In this paper, we provide the first assessment of the extent to which the AEA RCT Registry addresses publication bias along with a rich model of a researcher's registration decision. While the AEA RCT Registry is making some progress against publication bias, we find that over 90% of new economics RCTs do not register, only 50% of the RCTs that register do so before the intervention begins, and the majority of these preregistrations are not detailed enough to tie the researchers to a specific experimental design. Given these results, we then use the model to explore possible improvements to the AEA RCT Registry. In particular, the model demonstrates that removing the option to register completed RCTs can improve the efficacy of the registry by increasing the fraction of RCTs that register.
Securities Crowdfunding: More than Family, Friends, and Fools? LinkThe US securities crowdfunding market opened on May 16th, 2016. I provide the first description of the companies and investors participating in this market using data on the full universe of Regulation Crowdfunding issues. I then demonstrate that investment in an issue is closely tied to economic fundamentals after the first week. This result is consistent with unsophisticated investors arriving in the first week and investing indiscriminately and relatively more sophisticated investors arriving thereafter. Based on this evidence, I conclude that the market provides a promising new way for high quality early stage companies to seek financing.