Shorting costs and the cross-section of stock returns

Abstract

This paper contains three essays on mispricing allowed by high shorting costs. If the trading of noise traders is highly correlated and there is arbitrage risk, the price of an asset can deviate from its fundamental value. In particular, due to the short-selling impediments, stock prices tend to be overpriced. The first essay hypothesizes that the idiosyncratic volatility puzzle is due to this overpricing caused by noise traders that can be proxied by retail traders, and provide evidence supporting it. The empirical findings are as follows: First, the idiosyncratic volatility puzzle is more prominent among stocks with high retail trading proportion (RTP). Second, negative relation between idiosyncratic volatility (IVOL) and stock returns is stronger following high retail sentiment. Lastly, return reversals for stocks with high RTP and high IVOL are observed. The second essay hypothesizes that overpricing shows up more often than underpricing when shortselling is costly relative to buying so that there is arbitrage asymmetry, and document the followings. First, putoptioned stocks, which are supposed to be less costly to short, have less anomaly profits than non-put-optioned stocks. Second, in high-sentiment periods, short-legs of anomaly portfolios are more profitable with put-optioned stocks than non-put-optioned stocks, while, in low-sentiment periods, short-legs are not profitable with both subsamples of stocks. Third, returns on short-legs of anomaly portfolios are negatively related to lagged sentiment only when the degree of market-wide arbitrage asymmetry is high, where arbitrage asymmetry is measured by the difference of the market impact costs between the up market and the down market or by the market-wide average change in breadth. Finally, little evidence exists that anomalies associated with capital investments are caused by the presence of short-sale constraints. The third essay documents that the predictability of option-implied skewness (OIS) wo...