Analyst Forecasting Errors and Their Implications for Security Analysis

Abstract

Dreman and Berry (1995) have offered a perspective on analyst earnings forecast errors and their implications for security analysis. Among other arguments, they contend that the errors are too large to be reliably used by investors, the forecasts are less accurate than forecasts by time-series models, the errors are increasing over time, the analysts' forecasts are too optimistic, and the investment community relies too heavily on analyst forecasts. An alternative perspective on these issues is offered. The argument is that analysts' forecast errors are within 3% of an appropriate benchmark (namely, stock price), that their forecasts generally are significantly more accurate than forecasts by naive or sophisticated time-series models, that analyst forecast errors have not been increasing over time, that analysts have been too pessimistic in recent years, and that the investment community, by placing too much weight on forecasts made by time-series models, relies too little on analysts' forecasts.