Rounding Up the Net Income: Earnings Cosmetics or Material Earnings Management?

Abstract

To the extent that cognitive thresholds are relevant in the investors’ decision process, managers have an incentive to report earnings that exceed the given critical value. If the net income is below this threshold, managers might simply use accounting discretion to round up the earnings number. As a consequence, the second digits of the net income in a sample of firms rounding up do not follow Benford’s Law. However, the magnitude of rounding up (i.e., the difference between the (unobservable) pre-managed and the reported earnings) is important in assessing whether rounding up reduces the decision usefulness of financial statements. Rounding up reduces the financial reporting quality only if the earnings manipulation is material but not in cases of non-material “earnings cosmetics”. To differentiate between these cases, we investigate whether deviations from the Benford distribution are more likely in subsamples of firms with lower earnings quality. We find deviations of earnings numbers from the Benford distribution for firms that maximize their earnings via discretionary accruals, for firms with a non-Big 4 audit firm, and for firms with an auditor who is not an industry specialist. We do not find divergences for the subsamples of firms with higher financial reporting quality as approximated by these metrics. In contrast, earnings smoothing and overstepping investors’ cognitive thresholds seem to be opposing goals because only firms with low degrees of earnings smoothing deviate from the Benford distribution. Taken together, we find evidence for the materiality of the manipulations used to round up the net income.