Entries in credit agencies (1)

Saturday
Jun202009

Freerisk proposes a crowdsourced solution to risk modeling

The failure of the credit agencies to properly evaluate the risk of securitized assets and the role that this poor performance played in the current financial crisis has been discussed by the Wall Street Journal to Alan Greenspan. But what is the alternative paradigm? Freerisk, a new project by Jesper Andersen and Toby Segeren, suggests than open data and transparent, crowdsourced calculations can provide more accurate risk modeling of the assets that underpin our financial framework.

The common critique is that the current risk rating marketplace is an oligopoly; ratings can only come from three sources: Moody's, Fitch, or S&P. Sellers of securities must shop between these established players with the aim of receiving the best rating possible. Herein lies a fundamental structural problem in the market: The credit rating agencies are incentivized to provide higher ratings than are deserved.

And since there is no real competition, the credit agencies' performances are never seriously evaluated. In the structured finance business, assets that individually were junk could together be classified as extremely safe. Through late 2008, AIG was holding a AAA rating from Moody's. This rating was seriously flawed and provided an extreme disservice to investors seeking to understand the risk of AIG, but even this high-profile case doesn't stop sellers of securities from seeking Moody's stamp of approval.

But what if the marketplace could be re-structured so that incentives were more aligned to truth than to business? Rather than concentrating all ratings power into the hands of a small pool of analysts from Nationally Recognized Statistical Rating Organizations (NRSROs), could this risk evaluation be put out to a broader market?

Freerisk is using open data principles to make information on companies easier to understand and interpret, thus freeing risk modeling to a broader community. SEC records are exposed in RDF via a public API, enabling anyone to make his own risk calculator. The site also plans to use a prediction market-like leaderboard system to help rise top predictors to view. Then all participants can see who is doing the best job of measuring risk and learning more about the marketplace as they refine their systems. Freerisk hopes to eventually compete with the official rating agencies.

One of the best elements of this system is that the risk calculations are fully transparent, unlike the NRSROs who have no requirement to disclose their methods. How many people actually have the expertise to make these types of calculations, however, remains an open question. But at least the site can start building a public record of success. If the Freerisk community can consistently better evaluate risk than the official rating agencies, then than would give more ammunition to real credit agency reform. And it would also provide another example of how a market setting with transparent processes can enable a community to solve problems together.

Thanks to @alexismichelle for the tip!