This study intends to identify the better model in explaining variations of average stock returns of listed companies in the Colombo Stock Exchange (CSE) when time series and cross sectional regressions are employed. The sample consists of all stocks listed in the main board of the CSE except Bank, Finance and Insurance Sector during the period from 1997 to 2014. The methodology used to form factor mimicking portfolios to estimate risk factors and portfolio returns is similar to the methodology of Fama and French 1993 and 2012 and to test the performance of asset pricing models Fama and MacBeth (1973) two step procedure is employed. The Gibbons, Ross, and Shanken (GRS) (1989) F-test reveals that the Capital Asset Pricing Model (CAPM) is a poor model whereas the Fama and French (1993) Three Factor Model (FF3FM) and Carhart (1997) Four Factor Model (C4FM) are better models in explaining the cross sectional variations of stock returns of the listed companies in the CSE when time series regressions are employed. Fama-Macbeth t-test reveals that the C4FM is the only valid model in the size-BM sorted portfolios. The C4FM is found to be a superior model and performs better than FF3FM, Reward Beta Model (RBM) and CAPM and also the explanatory power of the FF3FM is comparatively better than both CAPM and RBM in explaining the cross section of stock returns of listed companies in the CSE.
Keywords: CAPM, FF 3-Factor Model, C4-Factor Model, Time Series Regression, Cross Sectional Regression, Reward Beta Model