eThesis - School of Business electronic theses

» hse_ethesis_14076.pdf   size:2 MB (1336903)

BACKGROUND AND OBJECTIVES:

Capital asset pricing model (CAPM) implies positive relation between the beta of a stock and its future returns. However, empirical studies do not support this proposition because low-beta stocks do better and high-beta stocks worse than expected. Moreover, Frazzini and Pedersen (2014) show that a betting against beta (BAB) factor buying low-beta assets while shorting high-beta assets provides ro-bustly positive returns in the U.S. stock market from 1926 to 2012 as well as in 20 international stock markets, Treasuries, corporate credits, and futures. There is an ongoing effort to uncover the beta anomaly. In this quest, the behavioral finance approach seems most promising. While market frictions such as benchmarking and leverage constraints probably are important, there is also something about high-beta stocks that makes them attractive for individual investors.

My paper takes a conservative approach in studying does the betting against beta exists in the stock market by combining empirical elements that have produced modest results in the previous studies. If the betting against beta is found, data mining etc. seem unlikely explanations. Secondly, I investigate whether high-beta stocks have certain characteristics that make them the most affected by investor sentiment, which measures the collective mood and trust of individual investors. Most importantly, I study does investor sentiment explain the betting against beta phenomenon in the stock market.

DATA AND METHODOLOGY:

I use stock return data from CRSP between 1984 and 2011 to sort stocks into quintile portfolios based on beta. To find out whether the betting against beta exists in the stock market I study the portfolio returns on both absolute and risk-adjusted basis. Secondly, I use an index of investor sentiment (Baker and Wurgler, 2007) to answer the question does investor sentiment explain the betting against beta. First, I divide the sample period into high and low sen-timent, and secondly, utilize the index as an explanatory variable in multifactor regressions.

MAIN FINDINGS:

I find that the betting against beta exist in the stock market. Empirical security market line is too flat and Sharpe ratios as well as CAPM alphas decline monotonically in beta. The results are generally robust to traditional multifactor models. Furthermore, investor sentiment influences the performance of the betting against beta. Following high sentiment the returns are monotonically declining in the beta quintiles, and following low sentiment the pattern completely reverses. My results also reveal a "flight-to-quality" phenomenon within stocks. Lastly, the results are robust to controls for market, popular empirical style factors, and alternative measures of sentiment. Additional robustness checks show that investor sentiment influences the betting against beta also contemporaneously and that the results are not specific to the time period because investor sentiment influences also the BAB-factor returns during a longer time period from 1965 to 2011.