In the Dodd-Frank Act, Congress included an exemption from the five percent risk-retention requirements applied to residential mortgages. Unfortunately, in their proposal to define what qualifies as a "qualified residential mortgage" under that exemption, regulators included only a very narrow test of borrowers' credit risk.
In fact, regulators' proposed test is based on derogatory credit history and other factors that are retrospective and may not accurately predict whether borrowers will default in the future.
This would have the unintended consequence of discouraging loans to deserving borrowers while potentially including riskier borrowers in the qualified mortgage exemption.
Ultimately the approach does not meet the Congressional intent of creating an exception for a "prime mortgage."
Regulators and financial institutions have traditionally relied on credit scores to assess a borrower's future risk. Credit scores do not tell the whole underwriting story, but they are designed specifically to predict credit risk and their predictive power has been validated time and time again.
Yet unlike past federal guidelines, the QRM exemption does not include credit scores. Instead, it requires financial institutions to perform an inflexible manual review of derogatory factors in the borrower's credit report.
This is like asking scientists to abandon their computers and use slide rules. There's a reason the mortgage lending industry has wholeheartedly adopted credit scores over the past 20 years: They work better than any other means of assessing risk based on credit data, which means lenders can make fairer and risk-appropriate decisions.
To understand the potential impact of the proposed QRM rule on consumer borrowing, FICO analyzed a large national sample of mortgage loans and found that under the proposed rule:
- Many high-risk borrowers would be inadvertently included in QRM loan portfolios. The minimum FICO Score of consumers who would meet the proposed QRM delinquency standards is 472, which is very high risk. Recently more than 25 percent of borrowers at that score level have been defaulting on their mortgage within two years. The median FICO Score of U.S. consumers is about 711 on the model's 300-850 range.
- Many low-risk borrowers would be inadvertently denied access to QRM loans. The maximum FICO Score of consumers who would fail to meet the proposed QRM delinquency standards is 845, which is very low risk.
- The lack of consistent treatment under the QRM standard should alarm lenders and borrowers alike. In our analysis we found sharp discrepancies in the proposed QRM rules' treatment of consumers over a broad range of FICO Scores that would impact more than 25 percent of the US population. For instance, at any given score level within that range some consumers would qualify and others would not qualify under the proposed QRM credit history rules. As a result, consumers in this score range would not only be confused but in thousands of cases could be put at a serious and unfair disadvantage.
The delinquency-based standards proposed by regulators would create other unintended consequences. The standards would be difficult for lenders to implement, and would force them to create new systems with associated costs, delays and errors.
In fact, some of the data relied upon in the proposed standard, such as the timing of short sales and repossessions, is not included on credit reports and lenders would have no way to comply.
The proposed requirements also allow for vital information used in the credit review to be pulled as much as 90 days prior to the closing of a loan. Today the industry's best practice is to confirm that the consumer profile has not changed within days of closing and funding the mortgage. By instead following regulators' proposed test, lenders could make important loan decisions based on stale information. Applicants' credit profiles could either improve or decline during this time period, leading to unequal treatment of consumers. FICO research has shown that scores for more than 25 percent of the U.S. population move up or down by more than 20 points during a quarter.
The solution to this problem is straightforward: Regulators should set a QRM standard that requires the use of credit scores to ensure compliance. The QRM population could be set at a targeted maximum default rate, or a specified percentage of the lowest-risk borrowers. This can be implemented by regulators in a vendor-neutral way that allows credit scoring models to be calibrated to either standard.
Andrew Jennings is the chief research officer at FICO.
