Aggregate Earnings: Time-series Properties and Forecasting Models.pdf
Aggregate Earnings: Time-series Properties and Forecasting Models false By:Koren M. Jo Published on 2015 by In this study I examine the time series properties of quarterly aggregate earnings and aggregate analyst earnings forecasts. Using quarterly data from January 1988 to December 2012, I show that both quarterly aggregate earnings and analyst earnings forecasts follow a random walk process in contrast with the corresponding firm level's actual and forecasted earnings, which both follow a seasonal random walk process. The fact that aggregate earnings follows a random walk process suggests that researchers using aggregate earnings need to account for the serial correlation in aggregate earnings in their models. Moreover, the level of aggregate earnings and aggregate analyst forecasts are co-integrated even though aggregate analyst forecasts are biased. I further show that prior forecast errors are correlated with growth in aggregate GAAP earnings but not aggregate Street earnings...