Empirical testing of the CAPM on the JS']JSE

Fama's (1970) efficient market hypothesis (EMH) and the capital asset pricing model (CAPM), jointly ascribed to Markowitz (1952), Treynor (1961), Sharpe (1964), Lintner (1965) and Mossin (1966), remain the foundation of most finance and investment courses. This is surprising, given the sustained criticism of the model and its assumptions, and is a reflection of the elegance and parsimony of the theory over the empirical evidence. On the Johannesburg Stock Exchange (JSE), several authors have examined and noted significant inadequacies relating to the single factor CAPM, particularly with regard to the dual nature (resources versus industrial shares) which characterise this bourse. Van Rensburg and Slaney (1997) advocate the use of a two factor arbitrage pricing theory (APT) model, but show that (at least for industrial shares), additional parameters are required (Van Rensburg, 2001). We revisit this ground using an improved methodology and data set over the period 31 December 1986 to 31 December 2011. We find that portfolios constructed on the basis of ranked beta exhibit a monotonic, inverse relationship to what the CAPM prescribes for most of the time-series. The use of the single beta CAPM is therefore inappropriate.