This study examines the resolution of ethical dilemmas in financial reporting by human participants and large language models. Participants act in the role of a CFO deciding whether to discontinue a prior policy with biased reporting; however, the bias is known and corrected by investors whereas a change may temporarily mislead investors. We find that models are less amenable to competing ethical considerations than humans, and exhibit greater preference for truthful reporting. Moreover, they respond with greater consistency to institutional ethical guidance, while humans become more indecisive under pressure from management. The models exhibit more internal coherence between their moral judgment and their policy prescriptions and are judged more persuasive by humans. Finally, humans follow model advice when accompanied by an explanation, but they seem to discount (and sometimes react against) advice offered without it. Our findings offer evidence on the misalignment between artificial intelligence and humans in tackling subjective reporting dilemmas while guiding the incorporation of such tools into corporate governance.

Organizations rely on languages that imperfectly represent the underlying state. We study the interaction between two types of costs: a lying cost, increasing in the distance from the message prescribed by the language, and a deception cost, increasing in the distortion induced in beliefs. In equilibrium, a message must be either untruthful or deceptive. Higher lying costs reduce bias and informativeness, while higher deception costs have the opposite effect. Surprisingly, full revelation is sustained as lying costs become small but only with unbounded equilibrium bias and communication costs. We characterize the most efficient combination of lying and deception costs among policies that sustain almost fully revealing communication. These policies feature high deception costs with some tolerance for lying. Extending the model to complementarities, we show that costs targeted at messages that are both untruthful and deceptive can sustain full revelation without communication costs. The analysis offers new insights into how legal and regulatory regimes approach literal truthfulness versus misleading statements in settings such as financial reporting, litigation, and professional ethics.

We study voluntary disclosure when investors observe firm reports through noisy information intermediaries such as auditors, analysts, rating agencies, or data providers. Any processing noise overturns the standard prediction of a unique partial-disclosure equilibrium. With low disclosure costs, the model unravels to full disclosure despite positive costs. With higher costs, the game admits two threshold equilibria featuring different disclosure probabilities. We characterize how the cost threshold for unraveling and the equilibrium set respond to changes in noise and fundamental uncertainty. In settings with high disclosure, both uncertainty and processing noise reduce disclosure, while higher certification costs can counterintuitively increase it. Endogenizing disclosure costs as optimal fees shows how profit-maximizing intermediaries select among equilibria, potentially generating a high-fee, high-disclosure regime. Extensions with bounded support, uncertain information endowment, endogenous noise, and competing information sources apply the insights to general information environments. The results caution against interpreting greater frictions as necessarily reducing disclosure.

ABSTRACT Using misstatement data, we find that the distribution of detected fraud features a heavy tail. We propose a theoretical mechanism that explains such a relatively high frequency of extreme frauds. In our dynamic model, a manager manipulates earnings for personal gain. A monitor of uncertain quality can detect fraud and punish the manager. As the monitor fails to detect fraud, the manager's posterior belief about the monitor's effectiveness decreases. Over time, the manager's learning leads to a slippery slope, in which the size of frauds grows steeply, and to a power law for detected fraud. Empirical analyses corroborate the slippery slope and the learning channel. As a policy implication, we establish that a higher detection intensity can increase fraud by enabling the manager to identify an ineffective monitor more quickly. Further, nondetection of frauds below a materiality threshold, paired with a sufficiently steep punishment scheme, can prevent large frauds.

Machine learning can improve empirical proxies of conservatism by detecting patterns beyond linear regression techniques assumed in prior literature. Using a neural network approach, we show that measures based on machine-learning exhibit (a) better fit adjusted for degrees of freedom, (b) fewer economically anomalous observations, (c) less unexplained year-over-year instability, (d) a secular decline in conservatism, and (e) more robust associations with periods post restatements. In simulations, proxies based on machine learning methods are the most robust to specification error and reduce the incidence of false negatives. Our approach shows that, separate from their usefulness in predictive analytics, methods from machine learning can be used to capture more informative variation than existing measures.

We estimate an infinite-horizon dynamic oligopoly model of audit firm tenure and misstatements and evaluate a policy counterfactual involving mandatory audit firm rotation. Longer tenure lowers the cost of producing audits, increasing audit quality and reducing audit fees. Thus clients are less likely to misstate and more likely to keep the incumbent audit firm as tenure increases. By reducing the value of retaining audit firms, mandatory rotation leads to large increases in auditor switches, even before the term limit, implying increases in the switching costs borne by clients. Misstatement rates increase because audit firms endogenously lower audit quality and newly hired audit firms have lower quality. Overall, the model suggests caution when evaluating the costs and benefits of government oversight over the audit profession. This paper was accepted by Suraj Srinivasan, accounting. Funding: The authors thank the financial support from the Olin Busiiness School and the Wharton School of the University of Pennsylvania. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.4944 .

We present a model in which investment capacity is reallocated in response to aggregate shocks and examine the resulting general equilibrium effects. The theory predicts a positive association between aggregate liquidity shocks, cost of capital, and conservative accounting. When capital becomes scarce, the accounting system is designed to preserve collateral, which depletes the supply of traded capital and leads to a higher cost of capital. The economy may accelerate small shocks with large (discontinuous) readjustments in financial reporting policies, cost of capital, and investment activity. We show that accounting policies set by firms to increase their market value may imply multiple equilibria, with self-fulfilling inefficient equilibria exhibiting excessive collateral requirements and reduced aggregate investment.

Verrecchia (1983, 1990) introduced the proprietary cost hypothesis in which exogenous disclosure costs are a reduced-form interpretation of lost competitive advantage in product markets. We develop a micro-foundation for this disclosure cost in a Cournot game and explicitly derive the cost as a function of market structure. When the market is sufficiently competitive, this model has a reduced-form representation similar to a standard voluntary disclosure game with a partial disclosure equilibrium. Proprietary costs are increasing in the number of competitors, the degree of product substitution, overall uncertainty, and production costs. The analysis also offers new empirical predictions on the interaction between disclosure choice, managerial horizon, and entry.

Although researchers often view earnings management as being widespread, measuring the cost and level of earnings management is a nontrivial task. We derive a measure of earnings management cost and the associated equilibrium level of earnings management from the cross-sectional properties of earnings and prices. This approach enables us to separate economic shocks from reporting discretion by modeling the economic tradeoff faced by management. The tradeoff can be easily estimated from a closed-form likelihood function. Consistent with prior studies, the measure suggests more earnings management during seasoned equity offerings, for smaller and growing firms, as well as in industries with more irregularities. This paper was accepted by Suraj Srinivasan, accounting.

How precise should accounting measurements be, if management has discretion to strategically withhold? We examine this question by nesting an optimal persuasion mechanism, which controls what measurements are conducted, within a voluntary disclosure framework a la Dye (85) and Jung and Kwon (1988). In our setting, information has real effects because the firm uses it to make a continuous operating decision, increasing in the market's belief. Absent frictions other than uncertainty about information endowment, we show that imprecision can reduce strategic withholding but always decreases firm value. We then examine plausible environments under which, by contrast, there is an optimal level of imprecision featuring coarseness at the marginal discloser. We offer additional implications in the contexts of enforcement against strategic withholding and financing with collateralized assets.

We model an information mosaic in which multiple signals-one gathered by an informed trader and the other publicly disclosed by the manager of the firm-are combined to estimate firm value. Under testable conditions, voluntary disclosures lead to higher ex ante information asymmetry and expected profits for the informed trader by allowing him to refine his trading strategy and complete his information mosaic. The informed trader's ability to combine information and enhance his advantage is more prevalent when there is more uncertainty about whether the news is favorable or unfavorable, the manager is more likely to be informed, and the manager's information is precise (i.e., disclosure quality is high).

Machine learning offers empirical methods to sift through accounting datasets with a large number of variables and limited a priori knowledge about functional forms. In this study, we show that these methods help detect and interpret patterns present in ongoing accounting misstatements. We use a wide set of variables from accounting, capital markets, governance, and auditing datasets to detect material misstatements. A primary insight of our analysis is that accounting variables, while they do not detect misstatements well on their own, become important with suitable interactions with audit and market variables. We also analyze differences between misstatements and irregularities, compare algorithms, examine one-year- and two-year-ahead predictions and interpret groups at greater risk of misstatements.

This study recovers a simple firm-level measure of disclosure costs implied by the voluntary disclosure theory of Verrecchia ( Journal of Accounting and Economics 12 (4), 365–380, 1990 ). The measure does not require knowledge by the researcher of the distribution of private information and can be implemented with three simple observable inputs: the minimum, average, and frequency of disclosure. We document a positive association of disclosure costs with proxies for existing and potential competition, information asymmetry, and insider trading. Higher values of disclosure costs are associated with lower contemporaneous and future disclosures as well as lower propensity to disclose in holdout samples. Overall, we provide future researchers with an easy-to-implement procedure to structurally estimate unobserved firm-level disclosure costs.

We estimate an infinite-horizon dynamic oligopoly model of audit firm tenure and misstatements and evaluate a policy counterfactual involving mandatory audit firm rotation. Longer tenure lowers the cost of producing audits, increasing audit quality and reducing audit fees. Thus clients are less likely to misstate and more likely to keep the incumbent audit firm as tenure increases. By reducing the value of retaining audit firms, mandatory rotation leads to large increases in auditor switches, even before the term limit, implying increases in the switching costs borne by clients. Misstatement rates increase because audit firms endogenously lower audit quality and newly hired audit firms have lower quality. Overall the model suggests caution when evaluating the costs and benefits of government oversight over the audit profession.

We derive a measure of earnings management cost and the associated equilibrium level of earnings management from the cross-sectional properties of earnings and prices. This approach enables us to separate economic shocks from reporting discretion by modeling the economic trade-off faced by management. The trade-off can be easily estimated from a closed-form likelihood function. Overall, the estimates suggest that earnings management is modest, consistent with the conjecture in Ball (2013) that earnings management is not as rampant as what prior research would suggest. Consistent with prior studies, the measure suggests more earnings management during seasoned equity offerings, for smaller and growing firms, as well as in industries with more irregularities.

Asset pricing theory postulates that a risk factor correlates with individuals' marginal utility of consumption. Hence, under plausible preferences, individuals should become more risk tolerant given favorable factor returns. We show that this wealth effect predicts a positive association between performance pay in observed contracts and factor returns, and test this prediction empirically with commonly-used asset pricing factors, such as the Fama French and momentum factors. Over the period 1992-2014, our empirical results support the hypothesized relationship for the market, book-to-market and momentum factors. These relationships appear to be driven by the delta of options, and for the market factor, stock grants. We also find that factors constructed from bond prices are positively associated to incentives, incrementally to the Fama French factors, but obtain mixed evidence for higher-order market factors, liquidity factors or factors constructed from national income accounts, including pricing kernels.

We extend the general equilibrium economy of Holmstrom and Tirole (1997) to optimal reporting of productive assets and examine when the accounting process can contribute to financial acceleration. Given a small change in aggregate capital stock, the economy may respond with large readjustments in accounting policies, prices and investment activity. A neutral accounting system, defined as a policy that does not distort decision-making, is optimal when capital is abundant but, after a contraction in aggregate capital, the accounting system becomes initially liberal and then conservative. Surprisingly, accounting policies maximizing firm value, i.e., the net cash flows to shareholders, may lead to self-fulfilling equilibria with inefficient forced liquidations. The theory offers a stylized paradigm to evaluate accounting policies in the aggregate.

This study recovers a simple firm-level measure of disclosure costs implied by the voluntary disclosure theory of Verrecchia (1990). The measure does not require a-priori knowledge by the researcher of the distribution of private information and can be implemented with three simple observable inputs: the minimum, average and frequency of disclosure. We document a positive association of disclosure costs with proxies for existing and potential competition, information asymmetry and insider trading. Higher values of disclosure costs are associated with lower contemporaneous and future disclosures, as well as lower propensity to disclose in holdout samples. Overall, we provide future researchers with an easy-to-implement procedure to structurally estimate unobserved firm-level disclosure costs.

This article offers a survey of theoretical research on disclosure and the cost of capital. We summarize the current state of the literature and discuss the channels through which information affects the cost of capital. After giving an overview of asset pricing theory, we examine the rationale for an accounting risk factor or an ex-ante effect of information on the cost of capital. Then, we discuss the role of voluntary disclosure, heterogenous beliefs, investor base, liquidity shocks, earnings management and agency problems as determinants of the cost of capital. Linkages between productive decisions and the cost of capital, and their implication for investor welfare, are also examined.