Read Full Paper

Accounting fraud imposes significant costs on firms and their stakeholders in the forms of higher costs of capital, inefficient resource allocation, regulatory sanctions, and investment losses. Not surprisingly, stakeholders, regulators, and researchers are interested in fraud detection. Credit rating agencies are one of the most important information intermediaries and have unique access to material non-public information from management, investment community would expect them to play a gatekeeper role against accounting fraud.

However, it is not clear whether credit ratings can be used to predict fraud. On the one hand, the conflicts of interest faced by issuer-paid credit rating agencies, such as S&P and Moody’s, weaken these rating agencies’ incentive to incorporate fraud related information into ratings in a timely fashion. Indeed, both investors and regulators have criticized issuer-paid credit rating agencies for failing to effectively assess financial reporting quality, especially in the wake of high-profile fraud cases in the early 2000s. The 2008 financial crisis sparked further debate on whether issuer-paid credit rating agencies have the expertise to analyze complex transactions and products.

On the other hand, the primary mission of credit rating agencies is to predict default risk. Deteriorating performance and financial distress are found to be one of the most important triggers of accounting fraud. In addition, accounting fraud exacerbates financial distress by imposing various direct and indirect costs, such as litigation expenses, lost sales, and higher costs of capital. Therefore, by diligently examining default risk related information, credit rating agencies have opportunities to collect information that systematically correlates with accounting fraud. Furthermore, issuer-paid credit rating agencies enjoy privileged access to management according to non-disclosure agreements. For example, rated firms provide financial forecasts of future cash flows to rating analysts and grant them on-site visits to observe firms’ operations and have face-to-face meetings with top executives.

Using 177 accounting frauds in the U.S. from 1999 to 2018, we find that rating downgrades and issuances of negative watch by S&P improve fraud prediction models’ performance after considering accounting factors and information from other market participants. Moreover, S&P’s rating actions are useful for fraud prediction as early as four quarters before fraud revelation. We further find that S&P’s rating actions are more useful for fraud prediction after the 2008–2009 financial crisis.

Overall, our findings support the view that rating actions of issuer-paid rating agencies are useful for predicting fraud among the firms they cover. We show that although credit rating agencies may not explicitly acknowledge fraud detection as part of their mission, their rating actions nonetheless can contribute to fraud detection in a meaningful and significant manner.