Taking center stage in the debate over reform in the mortgage industry is the definition of a qualified residential mortgage. A QRM is broadly defined as a loan that carries very low risk of default, which the Dodd-Frank Act permits lenders to fully securitize without retaining any portion of the loan on their balance sheets. Because a QRM carries less risk, it will likely be less expensive for borrowers.  

Until regulators complete their rulemaking, the definition of what is and is not a QRM is far from clear. The stakes are high.    

Moody’s Analytics chief economist Mark Zandi recently framed the implications appropriately when he wrote: “Too narrow a definition … could significantly raise the cost of mortgage credit and reduce its availability for a large number of potential borrowers. Too wide a QRM definition could blunt the risk retention rule’s ability to raise market confidence in securitization.”

I agree with the authors of the proposed rule that credit scores should not be used as part of the QRM definition. Some may suggest that it would be politically palatable and perhaps even sensible to use a minimum credit score as part of the standard QRM definition. As someone who is immersed in the credit scoring world on a daily basis, I can assure you that the unintended consequences would be severe.

Indeed, a minimum credit score could create precisely the type of extreme environments Zandi presages. If a minimum credit score is part of the QRM standard it will create more uncertainty about risk, instead of establishing a floor for it. That is because the risk levels behind credit score values are not static: the default levels associated with credit scores shift over time. For example, in the case of VantageScore, between June 2003 and June 2005, a consumer with a score from 691-710 had a default rate of 5.99%. More recent default rates tell a much different story. The default rates for that same credit score range increased to 10% in the period between June 2008 and June 2010. This principle holds true with other credit score models as well.

This dramatic difference in default rates demonstrates that a minimum credit score would not cap or regulate risk, or even measure it consistently. Instead it would ensure that the intended risk floor would actually fluctuate. On the one hand, the score cutoff chosen by regulators could expose the system to too much risk when the risk levels inevitably rise during periods of stress. In the opposite scenario, as default rates dip below the intended cut-off target during economic improvement, qualified borrowers who would otherwise be good credit risks would be frozen out of QRM eligibility.

Lenders account for risk shifts by revalidating their models.  If regulators were to attempt similar revalidations it would get messy. Consider that there are dozens of credit score models with varying ranges, which themselves may be changed at the creators’ discretion. Literally there would be dozens of score cutoffs that would each need continual revalidation.  Regulators would continuously have to revisit the QRM issue through a complex, time-consuming and uncertain rule-making process. It would cause confusion for lenders and borrowers.

While some market participants may favor usage of a minimum credit score as part of the definition of QRM, the interagency task force rightly disagrees, finding that “in order to ensure that creditors continue to choose among different credit score providers, the Agencies would have to determine a cutoff score under multiple scoring models and periodically revise the regulation in response to new scoring models.”

Fortunately, there’s a better option: use a measurement known as “propensity for default.” This is the default rate aligned with a particular level of risk, or credit score. Score developers provide performance charts to lenders so they understand the relationship between the three-digit score from that provider and their risk. Lenders use the propensity for default measurement every day. Unlike with credit scores, a propensity for default rate goal is not subject to drift. Credit score models are revalidated according to propensity to default trends and consumer behaviors, not the other way around.  If propensity for default is incorporated in the QRM standard, lenders would simply have to calibrate their own models and their preferred credit score model accordingly.  

There are many ways rule makers may frame the QRM definition. Adding credit scores to the mix clearly is ill-advised. If it is necessary to implement a risk threshold, propensity for default is the best path forward.

 

Barrett Burns is the president and CEO of VantageScore Solutions LLC, a credit score provider.

 

By Barrett Burns
Taking center stage in the debate over reform in the mortgage industry is the definition of a qualified residential mortgage. A QRM is broadly defined as a loan that carries very low risk of default, which the Dodd-Frank Act permits lenders to fully securitize without retaining any portion of the loan on their balance sheets under. Because a QRM carries less risk, it will likely be less expensive for borrowers.  
Until regulators complete their rulemaking, the definition of what is and is not a QRM is far from clear. The stakes are high.    
Moody’s Analytics chief economist Mark Zandi recently framed the implications appropriately when he wrote: “Too narrow a definition … could significantly raise the cost of mortgage credit and reduce its availability for a large number of potential borrowers. Too wide a QRM definition could blunt the risk retention rule’s ability to raise market confidence in securitization.”
I agree with the authors of the proposed rule that credit scores should not be used as part of the QRM definition. Some may suggest that it would be politically palatable and perhaps even sensible to use a minimum credit score as part of the standard QRM definition. As someone who is immersed in the credit scoring world on a daily basis, I can assure you that the unintended consequences would be severe.
Indeed, a minimum credit score could create precisely the type of extreme environments Zandi presages. If a minimum credit score is part of the QRM standard it will create more uncertainty about risk, instead of establishing a floor for it. That is because the risk levels behind credit score values are not static: the default levels associated with credit scores shift over time. For example, in the case of VantageScore, between June 2003 and June 2005, a consumer with a score from 691-710 had a default rate of 5.99%. More recent default rates tell a much different story. The default rates for that same credit score range increased to 10% in the period between June 2008 and June 2010. This principle holds true with other credit score models as well.
This dramatic difference in default rates demonstrates that a minimum credit score would not cap or regulate risk, or even measure it consistently. Instead it would ensure that the intended risk floor would actually fluctuate. On the one hand, the score cutoff chosen by regulators could expose the system to too much risk when the risk levels inevitably rise during periods of stress. In the opposite scenario, as default rates dip below the intended cut-off target during economic improvement, qualified borrowers who would otherwise be good credit risks would be frozen out of QRM eligibility.
Lenders account for risk shifts by revalidating their models.  If regulators were to attempt similar revalidations it would get messy. Consider that there are dozens of credit score models with varying ranges, which themselves may be changed at the creators’ discretion. Literally there would be dozens of score cutoffs that would each need continual revalidation.  Regulators would continuously have to revisit the QRM issue through a complex, time-consuming and uncertain rule-making process. It would cause confusion for lenders and borrowers.
While some market participants may favor usage of a minimum credit score as part of the definition of QRM, the interagency task force rightly disagrees, finding that “in order to ensure that creditors continue to choose among different credit score providers, the Agencies would have to determine a cutoff score under multiple scoring models and periodically revise the regulation in response to new scoring models.”
Fortunately, there’s a better option: use a measurement known as “propensity for default.” This is the default rate aligned with a particular level of risk, or credit score. Score developers provide performance charts to lenders so they understand the relationship between the three-digit score from that provider and their risk. Lenders use the propensity for default measurement every day. Unlike with credit scores, a propensity for default rate goal is not subject to drift. Credit score models are revalidated according to propensity to default trends and consumer behaviors, not the other way around.  If propensity for default is incorporated in the QRM standard, lenders would simply have to calibrate their own models and their preferred credit score model accordingly.  

There are many ways rule makers may frame the QRM definition. Adding credit scores to the mix clearly is ill-advised. If it is necessary to implement a risk threshold, propensity for default is the best path forward.