Investment horizon and beta coefficients

Three important issues are examined concerning the pricing of securities and the investment horizon, namely, the length of the investment horizon, the sensitivity of the beta coefficient to the assumed horizon, and the correlation between the beta and the horizon estimates. The analysis is based on a constant elasticity of substitution (CES)-type functional form of the capital asset pricing model (CAPM) and employs monthly returns for 75 stocks and 15 portfolios. Shown empirically, by the Bayesian approach incorporating with a diffuse prior, the estimated horizon is finite but often short, the beta coefficient is generally quite responsive to the assumed horizon, and the beta and the horizon estimates are correlated, but the directions of the sensitivity and the correlation vary for different assets. The empirical evidence has implications for the choice between the CES-type functional CAPM with a finite horizon and the Cobb-Douglas (CD)-type model with an instantaneous horizon. Other related findings include the following: The Bayesian estimates of a portfolio's beta and horizon reflect the averages of its component beta and horizon estimates, the Bayesian and the ordinary-least squares (OLS) estimates of the horizon parameter differ considerably, and there is no particular relationship between the investment horizon and trading volume.