We examine how removing user anonymity affects social media’s ability to generate value-relevant information for the stock market. Using a difference-in-differences design that exploits the differential timing of adopting real-name verification policies by the two most popular investment-related social media websites in China, we find that content on the treated site becomes significantly more informative about future stock returns and earnings after the policy takes effect. This effect is primarily driven by continuing users who post more actively in the pre-period. Although these users post less after the policy, the informational content of their posts increases, suggesting greater prudence in expressing opinions. Our results strengthen for firms that attract regulatory scrutiny. Overall, our study suggests that real-name verification policies can discipline internet users and potentially improve the informativeness of investment-related social media, particularly in emerging markets where retail investors predominate.

Machine learning (ML) could strengthen banks’ resilience through improved credit risk screening and ultimately benefit financial stability. Yet, ML adoption in banking remains limited, with simpler linear models still predominating. We argue that the emergence of highly flexible ML models has created a new challenge for forecasting tasks: ‘model multiplicity’—where equally accurate ML models at the aggregate level produce divergent individual-level predictions (‘predictive multiplicity’) or differ in their decision surfaces (‘procedural multiplicity’). These issues raise fundamental questions: Why should an individual or firm be subject to an adverse credit risk model outcome when there is an equally accurate model that treats them more favorably? Using the world’s largest loss database of corporate defaults, we examine these two phenomena in recovery rate ( RR ) modeling and propose heterogeneous ML ensembles as a natural solution. By combining predictions and decision surfaces from multiple well-performing ML models, ensembles mitigate risks associated with predictive multiplicity by ensuring that borrowers are not subject to the fluctuations of a single model, and reduce procedural multiplicity by providing a robust measure of features that ultimately improve out-of-sample RR predictions. By addressing the ‘multiplicity of good models’ problem, our study emphasizes the importance of model stability and provides new insights for the future development of ML models.

We survey public company executives and directors to understand what audit clients want from their auditor in today’s regulated environment. Our results provide at least three important takeaways. First, executives and directors view elements of auditors’ service quality (e.g., timeliness of communications) as at least as important as auditors’ technical competence. Second, we find no evidence that stakeholders’ preferences for service quality replaces their expectation for technical competence. Third, we find that 57% of executives and 67% of directors feel they can very accurately assess audit quality after an audit’s completion, but we are unable to identify clear determinants of participants’ self-reported ability to assess quality. Our results highlight the challenges facing auditors as they attempt to please clients, protect shareholders, and comply with regulations, all while offering a product whose quality is inherently challenging for others to assess. Our findings motivate future research and inform regulatory efforts.

This article explores the relationship between a firm’s legal contracting environment and its innovation incentives. Using granular data from the pharmaceutical industry, we examine a contracting mechanism through which incumbents maintain market power: “pay-for-delay” agreements to delay the market entry of competitors. Exploiting a shock where such contracts become legally tenuous, we find that affected incumbents subsequently increase their innovation activity across a variety of project-level measures. Exploring the nature of this innovation, we also find that it is more “impactful” from a scientific and commercial standpoint. The results provide novel evidence that restricting the contracting space can boost innovation at the firm level. However, at the extensive margin we find a reduction in innovation by new entrants in response to increased competition, suggesting a nuanced effect on aggregate innovation.

We document that firms vary in their timeliness of support for the Black Lives Matter (BLM) movement following the death of George Floyd in May 2020, and that timeliness is an indicator of authenticity. We predict that firms that speak out quickly in support of BLM (via Twitter or their websites) have made more investments in diversity and inclusion, relative to firms that speak out slowly (via conference calls or annual reports) or that remain silent. Consistent with this prediction, quick -disclosing firms have greater workforce diversity, have boards with greater ethnic diversity, and are more likely to tie executive compensation to diversity and inclusivity. Furthermore, quick -disclosing firms increase their hiring of both Black American employees and Black directors relative to firms that stay silent. We also document that quick-disclosing firms are part of more supportive stakeholder networks. We develop an inclusivity index and show that firms with higher index levels are more likely to speak out on the Capitol Riots, the Asian Spa Shootings, and voting rights.

Research suggests that board members value financial reporting quality because executive dismissal often follows low reporting quality events. However, inferences about the board’s demand for reporting quality in these studies are confounded by board members’ reputation incentives because the events examined are public (e.g., restatements). We investigate boards’ demand for reporting quality by exploiting a private signal of misreporting: audit adjustments communicated to the Board by the external auditor. We first survey 29 audit committee chairs to understand whether boards use audit adjustments in their oversight of management and then conduct an empirical investigation to answer our research question. We find an increased likelihood of CFO dismissal following audit adjustments. This association is driven by adjustments that reduce income and by firms with better board oversight. These findings suggest that boards proactively use nonpublic signals of reporting quality and incorporate information from auditors into their monitoring function.

This paper examines investors’ perceptions of auditor involvement in non‐GAAP reporting as captured by non‐GAAP disclosures in 10‐K filings. We find that firm‐years with auditor involvement in non‐GAAP reporting have higher CARs, lower bid–ask spreads, lower stock volatility, and lower abnormal trading volume on 10‐K filing dates. To sharpen identification of auditor involvement, we hand‐collect non‐GAAP measures and reconciliations for S&P 500 firms and identify which exclusions reconcile directly to the audited financial statements. As the percentage of exclusions that reconcile directly to the audited financial statements increases, bid–ask spreads and stock volatility on 10‐K filing dates decrease. We find that the results are not driven by strategic reporting or managers’ responses to perceived litigation risk. This study provides new insights into non‐GAAP disclosures outside of earnings announcements, which have been largely ignored in prior literature. Collectively, our results suggest investors value auditor involvement in non‐GAAP reporting and inform policymakers and standard setters considering the usefulness of assurance over non‐GAAP measures.