This paper documents a differential role of R&D versus selling, general, and administrative expenses (SG&A) for real earnings management. The distinction of these two components is important because prior studies mostly examine their combined use, but firms could manipulate them differently given the differing valuation implications. Reduced SG&A is viewed positively by investors as evidence of cost reduction, while reduced R&D is viewed negatively by investors as such expenditures are critical signals of expected growth. I examine their use in the context of seasoned equity offerings (SEOs) as well as firms receiving accounting and auditing enforcement releases (AAERs). Although both groups face strong incentives to manage earnings upward by reducing expenses, I predict and find that firms will reduce SG&A but increase R&D. During the manipulation period, SEO and AAER firms exhibit lower discretionary SG&A and higher discretionary R&D, relative to control firms, and investors positively value low discretionary SG&A and high discretionary R&D. Overall, this study confirms the importance of distinguishing between R&D and SG&A in real earnings management contexts and suggests a complementary (substitutive) relation between cutting SG&A (R&D) and accruals management.

This study investigates the mispricing of market‐wide investor sentiment by exploring the relation between sentiment and investor expectations of future earnings. Prior research argues that sentiment‐driven mispricing should be most pronounced for hard‐to‐value firms, such as those reporting losses (Baker and Wurgler 2006). Using investor expectations of future earnings, we provide empirical results consistent with this behavioral finance theory. We predict and find that investors perceive losses to be more (less) persistent during periods of low (high) sentiment; that (in contrast) investors perceive profit persistence to be lower (higher) during periods of low (high) sentiment; and that the effects appear stronger for loss firms relative to profit firms. We also document predictable cross‐sectional variation within losses (with the mispricing mitigated for losses associated with activities expected to generate future benefits), R&D, growth, large negative special items, and severe financial distress. Overall, our results document a new and important channel—investor expectations of future earnings—to explain sentiment‐driven mispricing.