When valuing a firm, investors must assess not only its expected future cash flows but also the systematic risk inherent in these cash flows. In this paper, we model the process by which investors may learn about firms' betas from earnings and how this learning process affects the relationship between earnings, announcement returns, and expected future returns. The model's main predictions are: (i) earnings response coefficients vary with macroeconomic conditions and are lower in upswings than downturns; (ii) earnings positively and negatively predict future returns in economic upswings and downturns, respectively, leading to return autocorrelation; and (iii) real earnings management rises in economic downturns and contributes to systematic risk in the economy. These predictions are directly attributable to investors' uncertainty regarding firms' exposures to systematic risk.

Extant literature documents that aggregate accruals positively predict future market returns and attributes this relation to either changes in discount rates or systematic earnings management. We offer an alternative explanation: aggregate merger and acquisition (M&A) activity drives this relation. M&A activity affects the magnitude of accruals, which in turn drives the market return predictability of aggregate accruals. We find that the ability of both aggregate accruals and discretionary aggregate accruals (a measure of systematic earnings management) to predict market returns disappears after controlling for aggregate M&A activity. Furthermore, aggregate M&A activity predicts future market returns, consistent with a price response to improvements in macroeconomic outcomes due to aggregate M&A activity.

Regulators have expressed concerns that an emphasis on non-audit services (NAS) could distract from the audit function, even for clients with minimal NAS purchases. Motivated by this concern, we examine whether a greater emphasis on providing NAS to audit clients generally (i.e., not to a specific client) can distract from the audit function, thus reducing audit quality. We find evidence of an NAS distraction effect, where a greater emphasis on NAS at the audit office-level results in more client financial statement restatements, even after controlling for client-specific NAS. Further, the association exists among clients that purchase minimal NAS, suggesting that this association relates to distraction effects in addition to independence issues examined in prior research. This study should be of interest to audit firms, audit committees, and regulators because it provides new evidence regarding issues related to a business model that includes both audit and non-audit services.