Is Earnings Fraud Associated with High Deferred Tax and/or Book Minus Tax Levels?

Abstract

The objective of this paper is to provide preliminary evidence whether SFAS No. 109 tax data might be useful in distinguishing between firms that do versus do not engage in earnings overstatement fraud (hereafter fraud). We examine the associations of various versions of deferred tax expense (DTE) variables and book income minus taxable income (BMT) variables with fraud, in the year of fraud onset and the year prior to fraud. The analysis is based upon a sample of 65 firms with positive pretax income, sanctioned by the Securities and Exchange Commission (SEC), in Accounting and Auditing Enforcement Releases (AAERs). A set of control firms are matched by asset size, two-digit SIC code, year, and nature of income (positive versus negative pretax income). We also perform analyses using a larger, nonmatched control sample. Our results indicate that, for firms with positive pretax income, DTE-based variables have strong incremental associations with fraud occurrence, beyond discretionary accruals and selected other explanatory variables, in the year of fraud onset. DTE-based variables have modest incremental power to explain future (next-year) fraud occurrence (but only when using matched samples). BMT-based variables generally lack explanatory power. In summary, this study provides new information about managers' tax reporting behavior in the presence of fraud, and suggests that DTE-based variables are likely to be useful in detecting fraud.