Analysts’ Earnings Forecast Errors and Cost Behavior

The objective of this study is to examine whether analysts fully incorporate a nonproportionate algorithm of cost-change relative to revenue-change in forecasting earnings. We conjecture that the errors in analysts’ earnings forecasts made in the first month after the fiscal year-end are largely due to the errors in estimating expenses as a result of analysts’ use of a proportionate cost-change model. Using 19,666 consensus analysts’ earnings forecasts for fiscal years 1983-2001, we find that, ceteris paribus, the earnings forecast errors are lower in the years when revenue decreases than when revenue increases. Our results of estimating a rational expectations model using a nonlinear generalized least squares method are consistent with our hypothesis that these differential earnings forecast errors between revenue decreasing and increasing years can be attributed to analysts’ systematic under (over) estimation of expenses in revenue decreasing (increasing) years.