Related Blogs


« FICO Award Winners Demonstrate How Analytics Fuels Success | Main | The "Computer Says No" Myth »

No Shortcuts to Objectivity

My colleague Frederic Huynh recently posted that mortgage short sales continue to represent high credit risk in our FICO® Score algorithm. Some readers have responded that our predictive science may be a tad too rigid, which naturally begs the question: Are they right?

They point out that starting in November, Fannie Mae and Freddie Mac reportedly will begin approving short sales for underwater borrowers who are current on their mortgage payments, provided the borrowers face an imminent “hardship.” Since these loans never have gone delinquent, do the borrowers deserve a break in the way FICO® Scores assess their short sales?

Score developers have trouble answering judgmental questions like that without any data to prove or disprove the point. That’s actually a good thing: data is crucial for objectivity. Because scores are objective, data-driven measures of risk, they help lenders make fairer and more consistent decisions. This objectivity also benefits consumers by democratizing access to credit for people from every walk of life.

In some ways the objectivity is hard earned. FICO® Scores are designed to rank-order people based on their default risk over a 24-month period. This means that our scientists need a minimum of 24 months of data about consumer credit behavior following a significant change or innovation in the credit industry to thoroughly evaluate a potential change to the model. Only by analyzing that data can we determine if a change should be made to our scoring model.

So it will be two years before we can determine if this new breed of short sales poses a different risk from past short sales. Is that delay fair to the people undergoing short sales today? Another judgmental question!

The delay is clearly integral to the objectivity that benefits both consumers and lenders. I believe the situation is reasonable based on the fact that through good economic times and bad, short sales have consistently correlated with high credit risk. That’s not surprising since a short sale is a type of default.

When we looked at data from October 2009 to October 2011, we found that one out of every two borrowers who went through a short sale also defaulted on another account within two years. That is exceptionally high risk. Such a strong risk predictor needs to be given full weight in the scoring model until we see compelling evidence that short sales—or at least short sales of loans without prior delinquencies—represent a different level of risk.

Bottom line, not enough performance data has been reported to credit bureaus to allow us to measure the risk of borrowers who go through short sales without prior delinquencies. Until sufficient data is available for us to analyze, it's not prudent to change scoring models.

First time on the Banking Analytics Blog?
Subscribe to the Banking Analytics Blog Feed or check out some other recent posts:

Comments

Post a comment

Comments are moderated, and will not appear on this weblog until the author has approved them.

If you have a TypeKey or TypePad account, please Sign In.