Professional analysts' estimates of earnings per share (EPS) provide a rare source of forward-looking information regarding the financial performance of publicly traded firms. Although numerous studies in the economics, finance, and accounting literatures have examined the properties of these forecasts and provided general insight into their performance, no known research explicitly examines the performance of analysts' EPS estimates for publicly traded food companies. This issue is particularly relevant given the influence that publicly traded agribusiness companies maintain in the agro-food supply chain (Vickner, 2002). Focusing on quarterly consensus estimates of EPS for 11 of the largest publicly traded food companies based on capitalization, the authors examine the point accuracy of these estimates through the introduction of the mean absolute scaled error measure, their performance over time, as well as their optimal forecast properties of bias, efficiency, and forecast encompassing. Results suggest that professional analysts, on average, produce EPS estimates that are more accurate than time series alternatives, yet the differences are often not statistically significant. For many of the firms examined, analysts' EPS estimates are found to be biased, inefficient, and do not encompass information in simple time series alternatives. For many firms in the sample, forecast accuracy has decreased over time. However, it is difficult to determine if this decline in forecast accuracy is due to turnover of analysts in the wake of increased financial market regulation (e.g., Sarbanes-Oxley), decline in forecasting skill, or structural changes in the food industry, which make it more difficult to forecast earnings over time. [EconLit citations: Q140; G170; M490]. © 2010 Wiley Periodicals, Inc.
Read full abstract