THE ORIGINAL Sharpe-Lintner capital asset pricing model advanced to explain the variations in risk differentials on different risky assets has now been widely questioned on the basis of the empirical evidence, and a large number of modified theories have been proposed to explain the discrepancies between theory and observation. The evidence points to a reasonably linear relationship on the average between return and non-diversifiable risk of outstanding common stock, or at least those listed on the New York and American Stock Exchanges. However, this same return-risk linear relationship does not seem to imply a riskless market rate of return consistent with any reasonable measure of the actual risk-free rates of return.' Moreover, while over the long-run the observed linear relationship between return and risk on individual stocks yields the expected positive sign of the risk coefficient more often than not, the shorter-term relationship has been erratic and has not been explained satisfactorily by the observed difference between the market rate of return of stocks as a whole and the risk-free rate. As a result of these findings, questions have been raised about the nature of the relationship between expected and actual rates of return, i.e., about the return generating model, as well as about the theory relating expected return to risk. A number of theoretical and empirical attempts have been made either to explain the apparent deficiencies in the original model on measurement and other statistical grounds or to modify that model to bring theory in closer conformance with reality. In our view, none of these attempts has been successful in bridging the gap between theory and measurement.2 However, four recent studies bearing on the plausability of the original or modified capital asset pricing models and on the relevance of past tests of these models merit brief mention prior to the introduction of the new tests presented in this paper. The first of these studies, based on an analysis of the stock portfolios as well as the major classes of assets and liabilities held by different individuals,3 found that a surprisingly large proportion of portfolios and assets were highly undiversified. It 1. Friend, I. and Blume, M., "Measurement of Portfolio Performance Under Uncertainty, American Economic Review, September, 1970; Black, F., Jensen, M., and Scholes, M., "The Capital Asset Pricing Model: Some Empirical Tests," in Jensen, M. (ed.), Studies in the Theory of Capital Markets, Praeger,