Knowledge that Transforms

We study the effect of fair value accounting on the association between net income and cash pay following the 2005 worldwide adoption of IFRS. We find that, while IFRS's non-fair-value provisions are associated with an increase in this association, its fair value provisions are associated with a decrease in this association. Overall, we contribute to the literature on the usefulness of fair value accounting by presenting evidence that fair value accounting is associated with a decrease in this association. Under assumptions that we detail and subject to caveats that we detail, our evidence suggests that fair value accounting may reduce the usefulness of earnings in evaluating management performance.

We test the political cost hypothesis using local soldier fatalities as a source of as-if-random variation in the threat of political costs for local defense firms. Soldier fatalities vary the threat of political costs for defense firms because the U.S. tradition of shared sacrifice during war vulgarizes war profits amid dead soldiers. Local defense firms record more income-decreasing accruals, equal to 1.17 percent of total assets, in response to a one standard deviation increase in local soldier fatalities (an additional 29 soldier fatalities in the average state-year). A wide variety of robustness tests corroborate our inferences.

We study how the potential for discretionary disclosure affects the way a firm designs its reporting system. In our model, the firm's primary but nonexclusive concern is to induce beliefs that exceed a threshold. Such thresholds arise in numerous contexts, including investing decisions, liquidation/continuation choices, covenants, audits, impairments, listing requirements, index inclusion, credit ratings, analyst recommendations, and stress tests. The optimal reporting system is characterized by informative good reports when the threshold is high and, potentially, uninformative reports when the threshold is low. Under an optimal impairment-type reporting system, the likelihood of reported impairments and the information content of non-impairment reports both increase in the probability of the firm observing private information. We provide a novel motivation for the quiet period around an IPO and empirical predictions relating the probability of discretionary disclosure to the properties of financial reports. In extensions, we consider disclosure mandates, report manipulation, endogenous thresholds, and alternative payoff functions.

This paper studies strategic interactions between public and private enforcement of accounting regulation and their consequences for the deterrence of financial misreporting. We develop an economic model with a manager, a public enforcement agency, and an investor and derive equilibrium strategies for manipulative effort, routine investigative effort, and costly private litigation. Our main results are as follows. (i) Strengthening private enforcement unambiguously enhances deterrence, whereas strengthening public enforcement can exacerbate misreporting, due to a crowding out of private enforcement. We provide conditions under which (ii) the enforcer's investigation incentives first increase and then decrease in the strength of private enforcement, (iii) public and private enforcement are strategic substitutes, (iv) the number of enforcement actions is misleading about public enforcement effectiveness, and (v) strengthening private enforcement decreases litigation risk. We also discuss implications of our results for empirical research.

Effective tax rates (ETRs) are often used to compare tax avoidance across firms and time. Using firms' detailed tax footnote data, we find that the effect of valuation allowances (VA) related to prior-period losses biases GAAP ETRs. This downward bias explains almost all of the downward trend in domestic firms' ETRs over the last 20 years. We also find that VAs explain cross-sectional differences in ETRs for both domestic and multinational firms. We show this bias extends to cash ETRs and the Henry and Sansing (2018) tax avoidance measure. We develop a methodology for substantially reducing the bias in both time-series and cross-sectional analyses of cash and GAAP ETRs. Overall, our results suggest firms’ loss histories and GAAP rules influence inferences from tax avoidance proxies.

We examine the role of concurrent information in the striking increase in investor response to earnings announcements from 2001 to 2016, as measured by return variability and volume following Beaver (1968). We find management guidance, analyst forecasts, and disaggregated financial statement line items are more frequently bundled with earnings announcements, and each of these items explains part of the increase in market response. Furthermore, collectively, these concurrent information releases explain a substantial fraction of the increase in market response to earnings announcements since 2001. This is in contrast to the decline in market response to management guidance issued separately from earnings and the much smaller increase in market response to analyst forecasts issued separately from earnings over this time. The findings indicate that information arrival at earnings announcement dates has increased significantly over the past two decades, and that key components of this are increased disclosures by management of guidance and financial statement line items and forecasts by analysts.

We examine how the media influences retail trade and market returns during the “quiet period” that follows a firm's IPO. We find that more media coverage during this period is associated with more purchases by retail investors and that such purchases are attention-driven, rather than information-based. Further, these retail trades are negatively associated with stock returns at the firm's first earnings announcement post-IPO. Our results suggest that media coverage, combined with market frictions that limit price efficiency in the post-IPO period, leads to worse investing outcomes for retail investors.

In an effort to motivate firms to more rapidly detect potential misconduct, legislators, regulators, and enforcement agencies incentivize firms to have integrity or “whistleblowing” hotlines. These hotlines provide individuals an opportunity to report alleged misconduct and seek guidance about how to appropriately respond. Beyond some isolated examples, little is known about the responsiveness of hotlines to actual claims of alleged misconduct. I undertake a field study to investigate how hotlines function in practice by making four different inquiries involving alleged misconduct to nearly 250 firms. I find that one‐fifth of firms have impediments (e.g., phone line disconnected, email bounce back, direct to incorrect website) that hinder reporting and approximately 10% of firms do not respond in a timely manner. Overall, this investigation illuminates several differences between integrity hotlines “on paper” and how they actually perform in practice.

We use a machine learning technique to assess whether the thematic content of financial statement disclosures (labeled topic ) is incrementally informative in predicting intentional misreporting. Using a Bayesian topic modeling algorithm, we determine and empirically quantify the topic content of a large collection of 10‐K narratives spanning 1994 to 2012. We find that the algorithm produces a valid set of semantically meaningful topics that predict financial misreporting, based on samples of Securities and Exchange Commission (SEC) enforcement actions (Accounting and Auditing Enforcement Releases [AAERs]) and irregularities identified from financial restatements and 10‐K filing amendments. Our out‐of‐sample tests indicate that topic significantly improves the detection of financial misreporting by as much as 59% when added to models based on commonly used financial and textual style variables. Furthermore, models that incorporate topic significantly outperform traditional models when detecting serious revenue recognition and core expense errors. Taken together, our results suggest that the topics discussed in annual report filings and the attention devoted to each topic are useful signals in detecting financial misreporting.

We develop a state‐of‐the‐art fraud prediction model using a machine learning approach. We demonstrate the value of combining domain knowledge and machine learning methods in model building. We select our model input based on existing accounting theories, but we differ from prior accounting research by using raw accounting numbers rather than financial ratios. We employ one of the most powerful machine learning methods, ensemble learning, rather than the commonly used method of logistic regression. To assess the performance of fraud prediction models, we introduce a new performance evaluation metric commonly used in ranking problems that is more appropriate for the fraud prediction task. Starting with an identical set of theory‐motivated raw accounting numbers, we show that our new fraud prediction model outperforms two benchmark models by a large margin: the Dechow et al. logistic regression model based on financial ratios, and the Cecchini et al. support‐vector‐machine model with a financial kernel that maps raw accounting numbers into a broader set of ratios.