Earnings response coefficient: Applying individual and portfolio methods

Abstract

This paper reports new findings from applying portfolio method, which shows a much bigger earnings impact on share prices (ERC) compared to the erstwhile reports of ERC using individual events, averaged over the sample. We estimate cumulative abnormal returns, CAR, across a test window for each quarterly earnings announcement event across one accounting year. The CARs are then regressed against earnings changes of individual firms and portfolios. The findings show a significant positive CAR when earnings increases; and a negative CAR if earnings declines. The ERC is very small in the test period of 2001-14, which is consistent with published results for years before 2000. The ERC size magnifies substantially due to the grouping effect used through portfolio formation. What is significant is that the use of portfolio method, by removing the idiosyncratic errors, show a price response very close to the size of earnings. The last evidence supports strongly the value relevance accounting theory that has not seen much support from averaging the price responses of individual event responses.