Earnings transfer function modeling efficiency

In this study the efficiency of the transfer function relationship between earnings per share and a market index is examined. Previous studies found that the earnings forecasting improvement of transfer function models may be artifacts of the modeling process. The purpose of this study is to estimate annual earnings per share univariate and transfer function models for 35 chemical, food, and utility firms using the 1950 to 1975 period for estimating the models and reserving the 1976 to 1982 period to serve as a postsample period to test the relative transfer function model forecasting improvement. The transfer function models produce lower postsample mean square forecasting errors for 60 percent of the sample firms but the application of the Ashley et al. test for bivariate modeling improvement yields the results that the lower forecasting errors are not statistically different from the univariate model errors.