ABSTRACT
This chapter discusses the existing literature by employing robust regression techniques in composite model building. Executive compensation should be associated with earnings growth because of the linkages among earnings, stock prices, and stock option exercise. Security analysts’ forecasts may be improved when combined with time series forecasts for a diversified sample of 261 firms with a 1980–1982 postsample estimation period. Better forecasting models may identify potentially high-growth securities overlooked by analysts, or may produce more accurate forecasts of growth rates than those produced by analysts. The use of ordinary least squares in estimating composite earnings models as developed by Granger and Ramanathan reduces the average estimated mean square regression error relative to the analyst forecasts for the 261 firms for the 1980–1982 postsample period. The information coefficient measures the efficiency of identifying the relationship between actual and forecasted rankings among the sources of information.
