The use of discretionary revenues to meet earnings and revenue targets

摘要

This dissertation addresses the use of discretionary revenues to meet earnings and revenue targets. First, I develop measures of discretionary revenues on both annual and quarterly levels. Second, I provide evidence on the reliability of various measures of discretionary accruals and discretionary revenues by assessing their ability to detect both simulated and actual manipulation. Third, I use the models to provide evidence as to whether firms use discretionary revenues or discretionary accruals to meet several annual and quarterly earnings and revenue targets.;In assessing the models, I find that measures of discretionary revenues are less biased and better specified than measures of discretionary accruals. For firms with high growth or high profits, accrual models over-reject the null hypothesis of no discretion, but the revenue models generally do not. Using a sample of firms subject to SEC enforcement actions, I find that both accrual and revenue models detect discretion by firms with revenue-related misstatements, but the accrual models fail to detect discretion by firms with expense-related misstatements. These findings provide support for studying discretion in revenues, and for using measures of discretionary revenues to do so.;In testing the hypotheses, I find the following. The evidence is consistent with firms using discretionary revenues, but not discretionary accruals, to meet annual and quarterly earnings forecasts. Thus, measures of discretion in revenues can detect earnings management (via revenues) where measures of discretionary accruals do not. At the annual level, I find evidence of discretionary accruals, but not discretionary revenues, to report increases in earnings, but the opposite is true at the quarterly level. This is consistent with quarterly revenue models being more powerful than annual revenue models. I find little evidence of discretion in accruals or revenues to report profits, which suggests a zero earnings target is less important for managers than earnings forecasts and prior period earnings. Finally. I find evidence of the use of discretionary revenues to meet annual and quarterly revenue forecasts, especially for growth firms. I also find some evidence of the use of discretionary revenues by loss firms.