Earnings Forecasting Models: Adding a Theoretical Foundation for the Selection of Explanatory Variables

Abstract

The purpose of this study is to determine if an earnings forecasting model based on factors hypothesised to result in differential profits across firms (industries) reduces model error relative to the model developed by Ou (1990). Initial research attempting to forecast earnings found that the random walk model, where current year's earnings are the prediction for next year, provides the best forecast of annual earnings (Ball and Watts 1972; Foster 1973; Beaver, Kettler, and Scholes 1970; Albrecht, Lookabill, and McKeown 1977; Brealey 1969). Ou (1990) developed an earnings forecasting model using financial statement information beyond prior years' earnings as the explanatory variables that outperformed the random walk model in predicting annual earnings.

Citation

Sneed, J.E. (1996), "Earnings Forecasting Models: Adding a Theoretical Foundation for the Selection of Explanatory Variables", Management Research News, Vol. 19 No. 11, pp. 42-57. https://doi.org/10.1108/eb028504