CAPM of the Russian Market: Results of the Missed Cross-Sectional Regression

In this paper a systematic testing of the standard single-factor CAPM on the Russian stock market is conducted using a cross-sectional approach. The existing evidence in favor of multifactor CAPM for Russian stocks, made without proper verification of the significance of the influence of individual factors, could have ignored the risk of overfitting. To eliminate thisgap, we formed a sample of 92 companies that were ever included in the MOEX index from November 2002 to January 2022, which allowed, on the one hand, to avoid survivorship bias, and on the other, to focus only on companies of interest to institutional investors. Next, we applied three forms of cross-sectional models: the Black, Jensen, and Scholes (BJS) model and the Fama and MacBeth model without and with the use of a five-year rolling window (FMB and FMB+RW respectively) for estimating beta coefficients. The BJS and FMBmodels revealed two anomalies incompatible with the Sharpe-Lintner CAPM: a significant positive excess return of the zero-beta portfolio and a negative impact of systematic risk on excess return. The FMB+RW model with a priori five-year betas produced a result consistent with the CAPM, confirming the positive effect of beta on returns; however, its empirical explanatory power remained very low. Further analysis showed that the identified anomalies are concentrated among companies with fewer than 10 years of observations and among small capitalization companies. Adding the market capitalization factor to the cross-sectional regression does not noticeably change the obtained results, from which it follows that it most likely cannot explain the found anomalies.