Essays on Financial Markets

Thumbnail Image


Publication or External Link





This dissertation contains three essays on financial markets concerning the relationship between short interest and returns in over-the-counter (OTC) equities, the effect of obtaining a rating on municipal bond offering yields, and the use of alternative trading systems (ATS) in the corporate bond market.Chapter 1 studies short positions among over-the-counter domestic common stocks. Short selling plays an important role in maintaining price efficiency, but short-selling in over-the-counter equities is often perceived as extremely rare. Short selling constraints are indeed high in this market, with the median fee to borrow a security being 2% for stocks with no short interest and 10% for stocks with short interest exceeding 1% of shares outstanding. Despite these constraints, 27% of domestic OTC equities have outstanding short interest positions on a given reporting date. Consistent with theories of short selling constraints such as Miller (1977), these high short selling constraints imply a substantial negative relationship between short interest and future returns. A portfolio of securities with short interest exceeding 1% of shares underperforms a portfolio of securities with no short interest by 31% annually and panel regressions show the relationship is robust to accounting for other security characteristics. The negative relationship between short interest and future returns suggests short sellers are trading in the direction of correcting mispricing in the OTC market, but the large magnitude and long time horizon over which short positions outperform suggests that there are large potential price efficiency gains from reducing constraints on short selling. Chapter 2, joint work with Haluk Unal, studies how whether a municipal bond is rated affects its offering yield. Approximately 34% of local municipal bond issues were issued without ratings during 1998 to 2017. We study the circumstances that affect the decision to obtain a rating and whether unrated bonds, controlling for observable risk factors, are more expensive to issue than rated bonds. Results show that issuers are less likely to obtain ratings for smaller issues, negotiated offerings, and bonds with high proxies for risk such as coming from areas with low personal income. We estimate the effect of forgoing a rating on offering yields using a doubly-robust Inverse Probability Weighted Regression Adjustment that controls for confounding that arises from risk and other characteristics affecting both the choice to obtain a rating and the yield. We separately analyze revenue bonds, general obligation bonds, bank qualified, and non-bank qualified bonds and find ratings decrease offering yields by 47, 49, 60, and 42 basis points respectively. The higher offering yields cost municipalities $22.5B in higher interest expense during our sample period. We find the choice of issuers to forgo ratings despite the substantial potential savings appears to be influenced by the underwriters they work with. Underwriters may face a conflict of interest where not obtaining a rating lowers the price investors are willing to pay from the bond, but also lowers the price the underwriter must pay the issuer and thus increases the underwriter’s profit. Chapter 3 is joint work with Matthew Kozora, Bruce Mizrach, Or Shachar, and Jonathan Sokobin. This chapter studies the circumstances when corporate bonds trade via an electronic ATS rather than in the traditional phone market and the relationship between venue choice and transaction costs. Trades on ATS platforms are smaller and more likely to involve investment-grade bonds, suggesting market participants trade via ATS when concerns about information leakage and adverse selection are lower. Trades on ATS platforms are more probable for older, less actively traded bonds from smaller issues, indicating participants are more likely to trade via ATS when search costs are high. Moreover, dealer participation on ATS platforms is associated with lower customer transaction costs of between 24 and 32 basis points.