Jiri Novak*

* Uppsala University, Sweden E-mail: jiri.novak@fek.uu.se October 2007 Abstract: The CAPM beta is arguably the most common risk factor used in estimating expected stock returns. Despite of its popularity several past studies documented weak (if any) association between CAPM beta and realized stock returns, which led several researchers to proclaim beta “dead”. This paper shows that the explanatory power of CAPM beta is highly dependent on the way it is estimated. While the conventional beta proxy is indeed largely unrelated to realized stock returns (in fact the relationship is slightly negative), using forward looking beta and eliminating unrealistic assumptions about expected market returns turns it (highly) significant. In addition, this study shows that complementary empirical factors – size and ratio of book-to-market value of equity – that are sometimes presented as potential remedies to beta’s deficiencies do not seem to outperform beta. This suggests they are not good risk proxies on the Swedish stock market, which casts doubt on the universal applicability of the 3-factor model. Keywords: asset pricing, CAPM, beta, factor pricing models, 3-factor model, market efficiency, Sweden, Scandinavia JEL classification: G12, G14 Acknowledgements: I would like to thank Dalibor Petr, Tomas ... and Johan Lyhagen for their help with the empirical analysis used in the earlier drafts of this paper. I would also like to thank my supervisors Jan-Erik Gröjer and Mattias Hamberg for their comments and suggestions. Finally, I am grateful to all participants if the Accounting and Finance Workshop help at the Göteborg University for their feedback.

1

Is CAPM Beta Dead or Alive? Depends on How you Measure It

…

October 2007 Abstract: The CAPM beta is arguably the most common risk factor used in estimating expected stock returns. Despite of its popularity several past studies documented weak (if any) association between CAPM beta and realized stock returns, which led several researchers to proclaim beta “dead”. This paper shows that the explanatory power of CAPM beta is highly dependent on the way it is estimated. While the conventional beta proxy is indeed largely unrelated to realized stock returns (in fact the relationship is slightly negative), using forward looking beta and eliminating unrealistic assumptions about expected market returns turns it (highly) significant. In addition, this study shows that complementary empirical factors – size and ratio of book-to-market value of equity – that are sometimes presented as potential remedies to beta’s deficiencies do not seem to outperform beta. This suggests they are not good risk proxies on the Swedish stock market, which casts doubt on the universal applicability of the 3-factor model. Keywords: asset pricing, CAPM, beta, factor pricing models, 3-factor model, market efficiency, Sweden, Scandinavia JEL classification: G12, G14 Acknowledgements: …

2

1. Introduction

The introduction of the Capital Asset Pricing Model (CAPM) (Sharpe, 1964, Lintner, 1965, Mossin, 1966, Black, 1972) initiated a stream in empirical research aimed at verifying the significance of CAPM beta and at identifying the determinants of expected stock returns in general. CAPM implies that there is a positive linear dependence of expected stock returns and CAPM betas (that capture the sensitivity of asset return to market return) and that CAPM beta is sufficient for explaining expected stock returns. Black, et al. (1972) performed one of the first empirical studies in the area that tested whether portfolios consisting of stocks with high betas one average generate higher returns. It soon became clear that CAPM beta does not suffice to explain the cross section of expected stock returns. Basu (1977) documented the positive significance of earnings-to-price (E/P) multiples. Banz (1981) found that size measured as the market...