By Marc D. Joffe, Startup Boot Camp Participant
Last week’s DOJ lawsuit against S&P reminds us that five years after the financial crisis, the problem of conflicted rating agencies remains unresolved. Despite the passage of Dodd Frank, the rating business continues to be dominated by three firms who receive most of their revenue from the very bond issuers they rate.
While this problem was created in New York, it can be solved here in the Bay Area. Technology firms like EBay and Yelp have become adept at aggregating and filtering user supplied opinions – many of which are burdened by obvious conflicts. Even more relevant, Bay Area technologists have the ability to develop models which could supplant the “green eye shade” analysis conducted by rating agencies.
Software models that gauge the risk of all kinds of debt are available. While everyone knows about consumer credit scoring models (like FICO), the industry has also developed models to assess all three categories of bonded debt: corporate, structured finance and government. The most successful computerized model for non-consumer credit was developed in San Francisco by KMV Corporation in the 1990s. In fact, KMV’s model was so successful Moody’s bought the company – beating out a rival offer from S&P.
Credit practitioners usually claim that quantitative, model-based analysis is insufficient. They insist that there is no substitute for an experience analyst looking the borrower in the eye. Unfortunately, this approach gives rise to all sorts of bias, especially when scaled up to rating agency proportions. Different analysts interpret the guidelines differently. Also, it is hard for human beings to make fine distinctions between small probability differences. For example, AAA might mean a one in a thousand risk of default, while AA+ might translate into a risk of five in one thousand. As behavioral economists have shown, people just can’t estimate probabilities that precisely.
Finally, and most importantly, it is difficult for mere mortals to ignore non-credit related factors, like worries about the loss of market share. As evidence in the DOJ complaint suggests, analysts at S&P were concerned about losing rating mandates to Moody’s and Fitch. It is hard to imagine how they could compartmentalize this concern when assigning ratings.
Of course, computer models can also be biased by their designers. That is where another Bay Area concept – open source - can come to the rescue. Most existing credit models are proprietary. By making these models open source and by populating them with open data sets, users could review the assumptions and determine whether they are biased. Further, opening up credit models and modeling data facilitates the development of a worldwide community that can critique and refine them. This is an ideal way to restore the credibility traditional ratings now lack.
Regulators can accelerate the innovation process by authorizing one or more certification boards to review and approve open source credit models. This way, serious models can be separated from less worthy efforts. Just like the Worldwide Web Consortium sets standards for HTML, a Worldwide Credit Consortium can set standards for open source credit models. Certified, open source credit models could then be used instead of traditional ratings in regulations. Replacing ratings in government regulations is an as-yet unfulfilled mandate of Dodd Frank.
With millions of bonds outstanding and more being issued each day, investors need informed, objective opinions about their risk. The current rating agency model is broken, and the DOJ suit threatens to sweep away one of the key players without offering a replacement. This is an industry crying out for disruption – a task well suited to the capitalists, foundations and engineers in our neighborhood.
For readers interested in learning more about these ideas, we have started an Open Source Finance Meetup group in San Francisco. See http://www.meetup.com/Open-
Marc D. Joffe is a former Senior Director at Moody’s and founder of Public Sector Credit Solutions.
Alumni Spotlight: Mona Das is Running for Congress Meet Mona Das. Mona’s parents immigrated to America from India when she...
In 2012 I started working as a strategic consultant for a climate-change education project called The Green Ninja Project. Funded...