Essays in Industry Cost of Equity and Return Dynamics

Abstract

Industry-level equity research is a relatively small but rapidly growing area of interest to finance academics. Costs of equity estimation as well as investigations into various anomalies such as the momentum, contrarian, value and size effects have mostly focused on individual stock returns rather than industry returns. Using a sample of U.S. industry returns, this thesis investigates two broad areas: industry cost of equity estimation and aspects of industry return predictability. This thesis presents three empirical chapters that cover industry cost of equity estimation, an examination of the long-term return reversal of industry returns, and finally research into the existence of value effects at the industry level. Previous research on industries shows that the Fama-French three-factor model and the Cahart four-factor model do not produce significantly better cost of equity (CE) estimates than those produced by the Capital Asset Pricing Model (CAPM). The first empirical study investigates the bias of the standard CAPM approach for each industry separately, and examines the effectiveness of a number of alternative beta estimators in producing CE estimates. The alternative beta estimators are a Blume-adjusted beta and a range of constant betas. CE estimates are judged on the basis of how well they predict the industry’s expected return (as proxied by average annual return over the next five or eight years). Performance evaluations are based on mean absolute forecast error (MAE).