Looming Risk For Mortgage Credit And Mbs Investors From “lender Choice”

Since 1996 when Freddie Mac introduced the industry to the first automated underwriting system (AUS) for GSE-eligible mortgages, a borrower’s credit score has served as one of the most important predictors of mortgage delinquency.  For many years thereafter, FICO score was the sole provider of scores to both GSEs’ AUS scorecards.  The arrival of VantageScore as a competitor to FICO has turned up the heat on the use of credit scores in mortgage underwriting with FHFA’s decision to allow lenders to choose between FICO and VantageScore in delivering loans to the GSEs.  This “lender choice” policy poses risk to both credit and MBS investors.

Lenders clearly have an economic incentive to deliver loans that have the best chance of being purchased by the GSEs.  Loans that pose higher risk may be subject to additional fees (LLPAs) or may not be an acceptable risk to the GSE.  Ever since the rollout of GSE automated underwriting, lenders have always been on the hunt for ways to find weaknesses in these scoring systems.  

In economics this practice is known as adverse selection, where one side of a transaction has more information than the other and sells a product with a higher credit risk unbeknownst to the buyer at the time.  In the specific case of lender choice, while the technical definition of an information advantage by lenders might not exist over time given that the GSEs expect such a strategy from lenders regarding their AUS, giving a lender the option to select which credit score to deliver invites a form of adverse selection that poses greater credit risk to the GSEs.  Don’t take my word for it, other analysis, including a recent one by Milliman finds evidence of this strategy.

In a new study of the effects on credit risk to the GSEs from lender choice, I examined the potential for adverse selection to take place by conducting a statistical analysis of GSE mortgages consistent with how a sophisticated and analytically oriented lender would approach the problem.  In this analysis, I developed two statistical scorecards similar to those used by the GSEs to estimate the likelihood of mortgage delinquency (defined as a loan ever going 90 days past due or worse in its life (D90+)).  One model used FICO as the proxy for creditworthiness, and the other used VantageScore along with a set of other borrower, loan and property characteristics.  That hypothetical lender would then choose to send the credit score that generated the lowest expected default rate between the two scorecards for a borrower.  In some cases that might be VantageScore, in others it could be FICO.

There is a fair amount of uncertainty over the degree of adverse selection that would be observed and so I looked at a range of possible scenarios from complete (100%) adoption of adverse selection (the extreme scenario) to variations in between (25% – 75%) and compared actual D90+ rates from those adverse selection scenarios to a baseline random credit score selection scenario.  

What I found is that over a range of credit score categories shown in Figure 1, any adverse selection scenario winds up with higher D90+ rates than the random score selection baseline and that difference widens as credit scores decline. On a weighted average basis of loans across all credit score categories from the loan sample, the 100% adverse selection scenario would raise D90+ rates by .44%.  Relative to actual D90+ rates for the sample period, that is an increase of about 18%.  

Figure 1: Actual D90+ Rates Across Lender Choice Scenarios by Credit Score Category

The D90+ estimates from this analysis underestimate the effects of potential adverse selection for several reasons including the fact that it does not account for another policy that eliminates the minimum credit score requirement for loans submitted through the GSEs’ AUS process which together with lender choice could further amplify credit risk.

To be sure, the GSEs will implement controls to monitor any potential for adverse selection over time, however, a fair amount of uncertainty exists for credit investors such as private mortgage insurers and credit risk transfer security (CRT) investors over the impact lender choice will have on credit risk and their investments.  Likewise, MBS investors must worry about the impact greater involuntary prepayments (defaults) under lender choice would have on mortgage security prices.  In the end, while the intent of lender choice to bring competition into the market on credit scores seems well-placed, it has the potential to backfire and actually pose higher costs on borrowers over time.  The FHFA would be well-served to suspend lender choice until a comprehensive assessment of its impacts on the mortgage market and borrowers is conducted.

Clifford Rossi is the Principal at Chesapeake Risk Advisors, LLC.
This column does not necessarily reflect the opinion of HousingWire’s editorial department and its owners. To contact the editor responsible for this piece: zeb@hwmedia.com.