Alpha, Beta and Beyond（289-310）.pdfVIP

10 * Alpha, Beta and Beyond Key words: beta estimation, OLS, robust estimate, Bayesian method, Kalman filter, shrinkage, backtesting Although academics and practitioners continue to debate its relevance,1 the beta, which is a measure of stock sensitivity to market movements, has become the best known and most widely employed measure for market risk. Similarly, the use of the beta to estimate expected returns, finds in the Capital Asset Pricing Model (CAPM) the theoretical foundation for justifying the current practice in the investment indus- try (for a presentation of the CAPM we refer to Barucci (2003), Cochrane (2001) and Sharpe, Alexander and Bailey (1999)). Indeed, as early as in 1982, Gitman and Mer- curio (1982) were already reporting that slightly more than 50 percent of managers were familiar with the beta and were using it in their financial activities. Sixteen years later, Bruner et al. (1998) confirmed that betas had, by that time, become the dominant technique to estimate the firm cost of equity. Similar evidence was then provided by Block (1999) during the following year. According to his survey, at the turn of the century, more than 30 percent of respondents considered the beta an im- portant tool in the valuation process and were using it in their business. Besides providing evidence on the affirmation of the beta, these surveys show that managers, traders and analysts in the investment management industry generally pur- chase beta estimates from commercial providers. And since multifactor return mod- els have become increasingly popular among practitioners, they have extended this habit to the purchase of other return factor estimates. In this manner, commercially prepared estimations have acquired unparalleled importance in the financial industry. A review of a variety of commercial providers (such as Bloomberg and Reuters) reveals that, in estimating betas, ordinary lea