21 Pages Posted: 28 Oct 2008
While there is growing evidence that stock prices do not follow pure random walks, the degree of existence of temporary components in stock prices is not well known. Modelling stock prices as the sum of a random walk and a general stationary (predictable) component, we propose an estimable lower bound on the proportion of total stock return variance caused by the predictable component. Contrary to the absolute value of the first-order autocorrelation coefficient estimates of Fama and French (1988), this lower bound reasonably estimates the true variance roportion in finite samples also when the temporary component does not follow a first-order autoregressive process. The estimated mean values of the lower bound reach a maximum of 10% for the equal-weighted market portfolio of NYSE stocks over the post-war period 1947-1986, while the maximum is 25% for the pre-war period 1926-46. The value-weighted market portfolio exhibits generally smaller variance proportion estimates. We also reexamine the pure random walk hypothesis using our univariate variance proportion statistic.
Keywords: Random walk, predictable components of stock prices, mean reversion, lower bound, ARIMA process, variance proportion estimate
JEL Classification: C1, E44, E47, C14
Suggested Citation: Suggested Citation
Eckbo, B. Espen and Liu, Jian, Temporary Components of Stock Prices: New Univariate Results. Journal of Financial and Quantitative Analysis, Vol. 28, 161-176, 1993. Available at SSRN: https://ssrn.com/abstract=1288734 or http://dx.doi.org/10.2139/ssrn.1288734