- Citation/Abstract
- Dissertation or Thesis
Abstract (summary)
The relationship between risk and return is a central theme of finance. The classic theoretical model which addresses this relationship is the Capital Asset Pricing Model, “CAPM”. The CAPM of Sharpe (1964), Lintner (1965), and Black (1972) posits that a security’s return is directly and positively related to its exposure to systematic risk as measured by beta.
Tests of the CAPM theorized relationship between risk and return predominantly support a direct, linear relationship. A minority of the research on the CAPM purports to show that the relationship between a security’s beta and its return is either too flat, or even inverse (Blitz, Falkenstein, & Vliet, 2014). This inverse relationship is known as the “low-beta anomaly”. Such tests of the CAPM largely focus on a single beta estimation interval; five years of monthly returns.
The purpose of this study was to assess the nature of the relationship between equity beta, and post-estimation return. Specifically, this study sought to address the validity and persistence of the low-beta anomaly across multiple beta estimation intervals.
Within the twenty year sample period from January of 1994 to December of 2013 this research covered ten different beta estimation intervals in order to determine whether a statistically significant and theoretically consistent relationship existed between equity beta and post-estimation realized return. This was done utilizing a double-pass, or “Fama-MacBeth” regression (Fama & MacBeth, 1973), whereby historical beta is first estimated by regressing a security’s return against the return of the market. The second pass of the regression, which provides the empirical test, is then conducted on the historical beta against post-estimation return. Further, this research appropriately accounted for the conditional nature of the beta-return relationship.
This research provided two basic conclusions: First, the low-beta anomaly is not robust across multiple beta estimation intervals. Second, with any test of the relationship between beta and return the choice of beta estimation interval matters. Different estimation intervals sometimes provide contradictory empirical results for the same period.