![](/sites/default/files/styles/post_image/public/2024-08/21%20Aug%202024_1.jpg?itok=Ov6mPUK_)
Do credit rating actions predict accounting fraud? The answer is yes, according to Allen H. Huang and Shiheng Wang of HKUST and their colleague. Their study also suggests that evidence of fraudulent accounting is strongest when it comes from issuer-paid credit rating agencies (CRAs); their investor-paid rivals are less able to detect fraud signals.
Accounting fraud has been the target of numerous attempts at prediction. However, credit ratings are rarely included in such predictive models. This is surprising, considering that “CRAs give lower ratings to firms with more aggressive financial reporting, worse accruals quality, and less timely earnings,” suggesting that rating changes may correlate with undetected fraud.
Thus, the researchers examined “the ability of rating actions to depict accounting fraud before it is publicly revealed.” They investigated whether the power of the F-score model—the most widely used fraud-prediction model—was improved by including the rating actions of two representative CRAs, Standard and Poor’s (S&P) and Egan-Jones Rating Company (EJR).
When the downgrading of firms’ ratings by these CRAs was added to the F-score, the model better predicted when those firms were involved in “settled securities class action lawsuits alleging accounting misstatements” or were the subject of Accounting and Auditing Enforce Releases by the SEC.
CRAs are not all equal. S&P is issuer-paid, while EJR is investor-paid. Although both CRAs’ ratings separately improved the model, the effect of EJR’s ratings was subsumed by that of S&P. Moreover, S&P was found to act sooner during fraud periods. “This observation is consistent with EJR’s reliance on public information,” say the researchers.
Overall, this research shows that issuer-paid CRAs should not be overlooked when constructing fraud-prediction models. “Although CRAs may not explicitly acknowledge fraud detection as part of their mission,” the researchers conclude, “their rating actions can contribute to fraud detection in a meaningful and significant manner.”