McVay (2006) presents evidence that managers inflate core earnings by shifting operating expenses to special items. In this paper, we improve her model to estimate core earnings by controlling for a firm's fundamental operating performance effectively. McVay (2006) suggests that poorly performing firms, which are more likely to report income-decreasing special items, may elect to temporarily reduce discretionary spending leading to a positive association between income-decreasing special items and unexpected core earnings. This positive association may then be incorrectly inferred as evidence of classification shifting. Hence, we control for those managerial actions which can impact the level of discretionary expenses and inventory, in order to estimate unexpected core earnings. Using the modified model, which exhibits better explanatory and forecasting power, we continue to find evidence of classification shifting for our full sample (1990-2010). Focusing on financial distress, we also show that McVay's (2006) model seems to overstate magnitude of shifting due to insufficient control for performance, and likelihood of presence of poorly performing firms in the sample. Further, her model doesn't capture classification shifting using shiftable income-decreasing special items, while the model proposed is able to do so. When we use special item subtypes reported by Compustat from 2001 onwards, we find that classification shifting continues to exist in the recent period (2001-2010) and in fact, its magnitude has increased. Overall, the proposed model improves identification and understanding of classification shifting.
