Essays on Extreme Markets

Abstract

Various phenomena of interest are often most clearly manifested in the extremes. I consider a particular type of extreme market a market ravaged by re sales. In chapter 1 I present a theoretical model of re sales that incorporates interaction between forced asset sales and declining price spirals in a heterogeneous multi-asset, multi-investor setting. I show that a unique equilibrium obtains in such a market and provide an analytical approximation of the resulting equilibrium re-sale prices. Fire sales can drive cross-asset contagion through leverage and overlapping assets holdings. In addition, the leverage decisions of individual investors can impose externalities on other investors by exposing them to higher re-sale risk. In chapter 2 I give empirical content to the re-sale model by using market-estimated parameters to calculate model re-sale prices for US stocks from 1982 to 2010. Model re-sale prices predict the cross-section of stock returns in distressed markets (when the S&P 500 index declines by more than 10% over a quarter), thus lending support to the theoretical model. In chapter 3 I turn to identifying extreme stocks. Using a simple unit root speci cation I identify stocks that deviate from a pure random walk in log prices, measured by autocorrelation. Autocorrelation predicts US stock returns and this predictability is robust to a wide range of time-series risk factors and stock characteristics. Stocks with autocorrelation substantially below unity generate unusually persistent excess returns: a zero cost hedge portfolio based on such stocks generate statistically signi cant positive excess returns in every month post-formation up to horizons of 25 years. Abnormal returns around earning announcements indicate that this persistence is unlikely to be the result of biased investor expectations of future earnings. This suggests the possibility that downward deviations from the random walk norm might be priced.