Who should pick credit-risk-scoring models?

Published by the R Street Institute.

American residential mortgage finance is the second-largest credit market in the world, behind only U.S. Treasury debt. Its immense size, at $10.4 trillion in outstanding loans, perhaps alone justifies the endless debates about how to reform it and improve its risk structures. In addition, the subject of mortgages is always political, being central to homeownership and to large housing-related industries and constituencies.

One relevant issue in the inside baseball of mortgage finance is the question of credit-scoring models. Credit scores and the models that calculate them are deeply imbedded in the credit decisions about mortgages. The purpose of the scores is to contribute to the credit consideration statistical estimates of the probability of default. Since credit underwriting is all about predicting and controlling the frequency and patterns of defaults, and therefore the credit losses experienced by the risk-taking lenders, this is a key idea.

Because Fannie Mae and Freddie Mac dominate the great middle of the mortgage market, the “conforming loan” sector, how they use credit scores has become a subject of congressional action, regulatory rulemaking and ongoing discussion. I just got a question on this issue when testifying to the Senate Banking Committee this week. In particular, the issue concerns Fannie and Freddie’s use of the FICO score, the incumbent mortgage-finance leader, or the challenger VantageScore. But it applies in principle to any other credit score that may be developed.

How to decide among competing scores is a highly technical matter. The more pertinent question from the political and economic perspective is who should decide. The answer to who is clear and definitive: whoever is taking the credit risk should decide how to use credit scores and which to use. That is the party that will bear the losses resulting from credit decisions, the one with skin in the game. Those without credit skin in the game should not get to choose which credit scores to use.

In short, the right principle is: the risk-taker decides. If Fannie and Freddie are taking the credit risk, they should decide which credit scores to use and how to use them, based on the most analytical, most objective criteria they can discover. And the same for any other risk-taker.

Should there be competition among credit scores?  Of course, competition is essential in this market, as anywhere. The right locus for competition is for the producers of credit scores to make the best cases they can to the risk-takers about the performance of their scores. May the most predictive, highest-performing score win.