Bias in quarterly estimates of annual effective tax rates and earnings management

We investigate whether quarterly annual effective tax rate (ETR) estimates are systematically biased in comparison to year-end actual ETRs. We find that estimated annual ETRs in the first, second, and third quarters are systematically higher than yearend ETRs. We then investigate whether firms' overstatement of quarterly ETRs creates slack that is used for earnings management. We find that quarterly ETR increases are more likely to be reversed in subsequent quarters when firms would have missed their analysts' earnings forecast absent the reversal. Finally, we show that in the years subsequent to the passage of the Sarbanes-Oxley Act (SOX), changes in the ETR continue to be associated with earnings management. These results, documenting patterns of annual ETR estimates and revisions, contribute to research about how earnings management is accomplished.