In March we learned that 15 mortgage lenders face $23.4 million in potential fines for improperly underwriting FHA-insured loans. In its report on the matter, the inspector general rebuked the Department of Housing and Urban Development for having its eye off the ball, highlighting HUD's failure "to verify independently that loans met FHA requirements and were eligible for insurance."
This event supports my conviction that executing loan-level reviews periodically throughout the mortgage lending life cycle and increasing the rate of quality-control sampling of individual loans are industry best practices that must become an industry standard practice.
Further, given the losses to the Federal Housing Administration insurance fund now buckling under the weight of terminal mortgage loans, every FHA file should be reviewed prior to insuring. Period.
Further still, questions must be asked regarding the origins of these noncompliant loan files. Were these fundamentally flawed loans submitted by sanctioned GNMA direct lenders or by regular lenders, who are required to submit case binders prior to being insured?
It would be telling to know how many of the 15 mortgage lenders facing fines were Government National Mortgage Association direct lenders.
The IG's report cites HUD's failure to detect loans destined for default based both on noncompliant production and early performance indicators that eventually caused an $11 million hit for the FHA insurance fund.
It is disheartening to read that " … lenders failed to comply with even the most basic of underwriting requirements. Half the 284 loans reviewed should have been disqualified because the lenders did not properly calculate or verify the borrowers' income or employment, did not properly document the source of borrowers' funds to close the loan or did not properly assess the borrowers' financial obligations."
Here's another rub: What about those loans that were not reviewed — those that weren't part of the IG's audit sample? If egregious failures were noted on half of 284 loans reviewed, just imagine if all had been reviewed.
Again, from the IG's report: "HUD's current method of selecting FHA loans for review [is] based on a statistically valid sample of loans with various risk factors." In a case of darkly understated irony, the IG report observes, "This method does not ensure that all claims are reviewed and may not target sufficient loans with claims paid to reasonably protect the fund."
That's exactly what the current level of quality-control sampling fails to do.
Lenders and investors that remain comfortable with 10% quality-control sampling are risk masochists, amnesiacs, or both.
Servicers, too, realize their need to improve intelligence on the loans in their keeping prior to accepting emerging risks and liabilities.
It is common knowledge that the servicing side of the mortgage lending life cycle is fully embroiled and has been unceremoniously marched to the woodshed for behavior that looks on the surface either like incompetence or negligence.
The truth is probably that they were flying with their eyes wide shut. It happens all the time in our industry.
Regardless, the proposed requirements taking shape for servicers are going to carry costs and liability for the way that non-performing loans are mitigated or resolved. This is a case in point for actually knowing what you are buying/servicing.
Look at it this way: If someone (stranger, friend, brother) paid you with a strap of cash for a car — say $5,000 in $100 bills — would you count the cash or just assume it was correct?
Mortgages are built into serious financial instruments upon which households are created, businesses are built and retirement funds are invested. Not only have there been tectonic shifts in secondary marketing, but there's also a new awareness of consumers' rights to protection from exploitation. All promise a fingerprint dusting on loans that are found, post-mortem, to have been defective from origination. We need to count the cash, make sure the denominations are correct and look for counterfeits.
Based on the now public IG audit of HUD, it appears that quality control is slipping through the cracks with regularity. Based on what my company is seeing, the current 10% industry standard quality-control sample fails to deliver confidence in mortgage lending standards. Would 25%, 50%, 75% deliver better results? Reduce losses? Could increasing the percentage of loan-level review across the mortgage life cycle entice private investors?
Is it feasible for lenders/investors/servicers to perform a higher level of quality-control sampling? Is it feasible to not?
This is a discussion worth having not only as an industry, but also with regulators and policymakers. A full 100% loan-level review would be a game changer for our industry, as it would ensure the appropriate level of risk is addressed in pricing. It would also provide a clearer picture of where the problems originate.
