Essays in corporate finance
This dissertation consists of three essays in empirical corporate finance. The first two essays focus on the impact of securitization on the corporate loan market. The third essay studies the role of equity blockholders in corporate governance.
The first essay, written jointly with Efraim Benmelech, examines the credit rating process for collateralized loan obligations. CLOs were one of the largest segments of the structured finance market from 2003-2007, fueling a boom in syndicated loans and leveraged buyouts. Investors in CLOs rely heavily on credit ratings provided by rating agencies, yet little is known about their rating practices. This paper attempts to fill that gap. Using novel hand-collected data on 3,912 tranches of collateralized loan obligations we document the rating practices of CLOs and analyze their structures.
The second essay, written jointly with Efraim Benmelech and Victoria Ivashina, examines agency problems in securitization. We investigate whether securitization was associated with risky lending in the corporate loan market by examining the performance of individual loans held by CLOs. Our results indicate that adverse selection problems in corporate loan securitizations may be less widespread than commonly believed. Controlling for firm characteristics, we find that securitized loans perform no worse, and under some criteria better, than unsecuritized loans of comparable quality. Furthermore, we find that banks that participate on both sides of the market, arranging loans and underwriting CLOs, may use private information gained in the lending process to direct loans with more stable credit quality towards their own CLOs.
The third essay, written jointly with Rüdiger Fahlenbrach, Paul Gompers, and Andrew Metrick, contributes to the empirical corporate governance literature. Large blocks of stock play an important role in many finance studies, yet there is no standardized data set that tracks them. Further, the best available data source has many mistakes and biases. We document these mistakes, show how to fix them, and demonstrate the impact in a regression framework. For researchers who need to work outside of this sample, we test the efficacy of alternative fixes and fmd that truncating or winsorizing can reduce about half of the bias in our representative application.