The Power of Two-Pass OLSs Test of the CAPM : A Simulation Analysis of the CSE

Inconclusive evidence for the CAPM is due to one or both of two reasons. First, the theory does not exist in the real world, second, the testing methods use to test the CAPM are incapable of detecting it. This study isolates the second reason by developing an artificial market in which the CAPM is true. The parameter values of the simulated market (market excess return, etc.) are extracted from Colombo Stock Exchange (CSE) to exemplify the Sri Lankan context. Then, the two-pass OLS regression test is applied to the simulated market which holds the CAPM and ascertains whether the test is capable of detecting the CAPM. The simulation extends to study how the power changes in response to two power determinants with conflicting roles, the standard deviation of market risk premium and the number of portfolios formed in the test. The sensitivity of power curve in response to changes in the standard deviation of market risk premium is nonlinear and peaks and troughs are observable on it. Thus it is not a continuously decreasing function as indicate by previous research. Thus, the paper demonstrates importance of having considerable standard deviation of market risk premium to detect the CAPM when it exists. Therefore, the test perform very poorly at low standard deviation of market risk premium as opposed to common understanding that the standard deviation of market risk premium only affects negatively on the power of two-pass OLS regression test of the CAPM. Further, it is revealed in the simulation that the number of portfolios formed in the test has no major impact on the power, doubting usefulness of forming portfolios. Further, our results show that the chance of detecting the CAPM by the two-pass OLS regression approach (power of the test) is less than 10%