EVIDENCE ON THE "GROWTH-OPTIMUM" MODEL.

If investors wish to maximize the probability of achieving a given level of wealth within a fixed time, they should choose the "growth-optimum" portfolio; that is, the portfolio with highest expected fate of increase in value. This paper has examined the implications for observed common stock returns of all investors selecting such a portfolio. Given some widely-used (and useful) aggregation assumptions, the growth-optimum model implies that the expected return ratio E[(1 + R[sub j])/(1 + R[sub m])] is equal for all securities. This implied equality of expected return ratios was utilized in analysis of variance tests with New York and American Exchange listed stocks from 1962-1969 in order to ascertain the basic validity of the growth-optimum model. The model was well-supported by the data. The growth-optimum model was compared algebraically to Sharpe-Lintner theory, which is probably the most widely-used portfolio result in empirical work. A close correspondence was demonstrated between their qualitative implications. For example, both models imply that an asset's expected return will equal the risk-free interest rate if the covariance between the asset's return and the average return on all assets, Cov(R[sub j], R[sub m]), is zero. For most cases, the growth-optimum model also shares the Sharpe-Lintner implication that an asset's expected return will exceed the risk-free rate if and only if Cov(R[sub j], R[sub m]) > 0. There are, however, some cases of highly-skewed probability distributions where this implication does not follow for the growth-optimum model. A close empirical correspondence between the two models was demonstrated for common stock returns. The procedure (1) estimated returns and risk premia implied by the two models from time series; (2) calculated cross-sectional relations between estimated returns and risks; and (3) compared the cross-sectional relations to the theoretical predictions of the two models. They could not be distinguished ... [ABSTRACT FROM AUTHOR]