The non-relevance of the elusive holy grail of asset pricing tests: The "true" market portfolio does not alter CAPM validity conclusions

Abstract

Empirical evaluations of CAPM usually attach a caveat that rejection is conditional on the choice of market proxy. We explore the criticality of the proxy choice disclaimer. Using different proxies and comprehensive simulations of the unobserved "true" market in Fama-MacBeth tests of CAPM, we find that the significance (t-statistics) corresponding to betas remain consistently unaltered, even if the proxy is a small fraction of or has a low correlation with the true market. The constancy of t-statistics persists in a simulated true-CAPM world as well: if CAPM is indeed valid, the choice of proxy is unlikely to reject it erroneously. Identity of the elusive true market portfolio and the choice of representative proxy cannot overturn conclusions on validity of CAPM based on Fama-MacBeth tests. Roll's Critique, incontrovertible in theory, may be quite forgiving in practice - CAPM cannot be resurrected by a "closer" approximation of the elusive true market portfolio when it has commonly been rejected. © 2009.