Evidence on Surrogates for Earnings Expectations Within a Capital Market Context

Abstract

This study compares the abilities of statistical model forecasts with financial analyst forecasts to serve as surrogates for market expectations of quarterly and annual earnings per share. For both annual and interim earnings expectations, statistical model forecast errors were found to be associated with risk-adjusted security returns, even with financial analysts' forecast errors held constant. Similarly, financial analysts’ forecast-error associations were found to be significant after controlling for statistical model forecast errors.

Additional tests were performed on the null hypothesis that the financial analysts exploit all information used by the time-series models. The data indicate rejection of this hypothesis for both annual and interim forecasts. Finally, forecast error analysis supports previous research in finding that analysts’ forecasts are more accurate than those of statistical models. However, this superiority disappears after controlling for hypothesized timing advantages favoring the analysts.