Abstract
This thesis consists of three standalone studies in the fields of asset pricing and market microstructure. The first study investigates mispricing in the cross-section of stocks. It shows that mispricing can be explained in a rational equilibrium in which investors allocate investigative resources to stocks to maximize their expected profits from arbitrage. Stocks with smaller dollar profit potential are allocated less attention, and their percentage mispricing is higher on average. For such stocks, information discovery by investors is slow, and the mispricing is corrected mostly through mandatory disclosures by firms. These mechanisms are confirmed empirically with measures of institutional attention and trading discreteness. The attention allocation channel explains the cross-sectional pattern of mispricing better than any classic arbitrage frictions. In the second study, I find that short interest aggregated across stocks is negatively related to future market returns internationally. This result indicates that the strong relation between short interest and market returns recently documented for the United States holds robustly out of sample. I examine a number of potential explanations for this pattern. The time-varying aggregate risk premium cannot fully explain it. Somewhat surprisingly, I find that market-wide short sale constraints, such as regulatory restrictions, do not contribute to this pattern, while stock-level constraints do. This finding is consistent with short sellers’ practice of abstaining from informed trading in the market portfolio. As a result, overpricing that persists in hard-to-short stocks propagates to the market level. Consistent with this notion, I find that the predictive power of aggregate short interest predominantly comes from stocks with high short-selling risk. In the third study, I find that increased speed competition from high-frequency market makers (HFT MMs) discourages the participation of slower liquidity-providing MMs. The presence of HFT MMs reduces slower MMs’ profits without impacting the profitability of executed trades. These results indicate that HFT MMs do not expose slower MMs to increased adverse selection.