Aggregate Earnings, Forecasts and Revisions. Evaluation of the Information in, and Characteristics of, Aggregated Analysts' Forecasts

Abstract

I investigate the information in aggregated US equity analysts' earnings forecasts. Despite a voluminous body of research evaluating the information in, and characteristics of, equity analysts' forecasts, relatively little is known regarding aggregated forecasts. However, Kothari, Lewellen and Warner (2006) demonstrate how estimated relationships between, for example, earnings and returns may differ markedly at the aggregate level compared with the individual stock level. I generate time series of aggregated forecast earnings, aggregated forecast revisions and aggregated realized earnings for the period extending from the first quarter of 1979 through to the last quarter of 2009. These variables are employed in three examinations of aggregated earnings expectations. Firstly, prior research indicates significant information in analysts' forecasts for future realized earnings, and strong positive correlation between realized earnings and indicators of macroeconomic activity. I therefore hypothesize significant information in aggregated analysts' forecasts for future realized economic activity. Secondly, I investigate the informational efficiency of analysts' forecasts with respect to realized macroeconomic variables, and implications of earnings revision predictability for return predictability. Thirdly, I employ aggregated earnings revisions as proxies for market earnings surprise in tests of cash flow and discount rate effects in market returns. I find evidence of statistically significant information for future US industrial production growth in aggregated analysts' forecasts, the magnitude of which is a partial function of earnings smoothing by management, firm size and earnings cyclicality. I also find evidence of systematic underreaction by analysts to realized macroeconomic factors, resulting in revision predictability which in turn is able to explain significant systematic variation in future industry returns. In addition, my results suggest that the negative relationship between aggregated earnings surprise and contemporaneous returns identified by Kothari et al. (2006) is at least partially a product of the period they evaluate. In robustness tests employing both aggregated realized earnings and aggregated forecast revisions, I find evidence of positive (albeit insignificant) relationships between these proxies for earnings surprise and contemporaneous market returns. My results do not support the notion of a discount rate effect dominating a cash flow effect at the aggregate level.