The following very useful explanation appearedon Governance Metrics International’s web site. GMI is the leading independent provider of global corporate governance and ESG ratings and research. Corporate stakeholders – including leading investors, insurers, auditors, regulators and others – use GovernanceMetrics services to identify and monitor risks related to non-financial measures covering key environmental, social, governance and accounting risk factors. (Memo by James James Kaplan, Co-Founder and Chief Executive.)
At first glance, GMI Ratings’ Accounting & Governance Risk (AGR) rating may appear to be a “black box,” a purely computerdriven algorithm divorced from analysts’ experience. In fact, however, the AGR is built upon a handful of very human intuitions. Software and statistics just help to amplify those insights, providing clear, comparable, predictive ratings on tens of thousands of companies.
As I like to sum it up: the AGR is common sense, quantified.
Programming the Lie Detector
The core insight that led me to found our predecessor firm, Audit Integrity, is that even though accounting deals with numbers, it’s really a kind of language. And like any language, it can be used to lie.
To create our AGR algorithms, we put together all the financial statements that had been found by regulators to be fraudulent. Then we used statistical analysis to identify characteristics that these statements had in common, but that distinguished them from the broader group of financial statements issued by all companies in our universe. When we see new financial statements come out that have a lot of the traits commonly associated with fraud, we think the odds are higher that they could be fraudulent too.
We never claim to know that for sure—and we don’t conduct in-depth analysis of each financial statement. It’s a quantitative model, remember. But its power comes from recognizing patterns, based on the experience of the past 13 years, and raising red flags when there are signs that someone, right now, could be cooking the books. In a host of cases, we’ve turned out to be right: companies that score low on the AGR have been more likely to face regulatory action, and also to show stock price underperformance.
Revenue Recognition: A Key Indicator
Publicly-traded companies are always hoping to make themselves look good to investors and the public, and one of the simplest ways to make yourself seem richer than you are is through improper revenue recognition.
In principle, companies should recognize revenues only after they have been earned (because the product or service in question has been delivered) and if there is reasonable assurance that the customer will pay. But in practice, some companies book sales before the deal is clinched, or when there’s still work left to be done. Companies that do this may ind up appearing much stronger than their peers on accounting ratios that involve revenues (for example, Accounts eceivable/Sales, Operating Revenue/Operating Expense, or Unearned evenue/Revenue). It turns out that in many past ases of proven fraud, these ratios were extreme. So the AGR model flags companies that are outliers in these respects.