GMI has enhanced their Accounting & Governance Risk (AGR®) ratings product. As I have indicated elsewhere, products like this one and GMI’s Forensic Alpha Model (FAM™)do a great deal in advancing corporate governance. See my September 26, 2013 post and FAMulous Basis for Achieving Alpha.
AGR 3.0 improves upon the methodology of AGR 2.5 in several ways. AGR® is designed to identify the various fingerprints of fraud and then to rank companies based on these fingerprints. AGR 2.5 did a remarkable job at separating the worst companies from the best with respect to forensic accounting and governance where the worst rated companies were 5x more likely to be caught by the SEC for accounting fraud. In the updated model (AGR® 3.0) the likelihood of the worst rated companies committing fraud increases to more than 13x.
The three most impactful improvements within AGR 3.0 include:
- The newest cases of SEC Enforcement Actions of Accounting Fraud have been incorporated into the model. As fraud changes, GMI stays a step ahead of the new fraudulent accounting games being played.
- A more sophisticated statistical approach has been implemented both on a univariate and multi-variate level. The results are more consistent across time periods and samples.
- An improved statistical methodology that for the first time takes into account the interaction between Governance and Accounting factors. In earlier versions of AGR each metric was treated independently when measuring its strength in predicting fraud. The measure of association between each metric and fraud was changed from odds ratios to Somers’ D. Although both measures are nonparametric methods used in rank statistics, confidence intervals for Somers’ D are better.
In my opinion, unmodified indexing is no longer appropriate. Removing the riskiest stocks from portfolios should be a standard part of fulfilling fiduciary duty. GMI advances the efficacy of corporate governance with this product. A technical document for AGR 3.0 is available here.