How Slippery Language Clouds the Debate Over Auto Lending Risk
The chief executive officer of Exeter Finance responds to critics who argue that it's become too easy for Americans with low credit scores to finance a car purchase.
Government investigations of auto lending are beginning to pile up, and they are weighing on investors' perceptions of the fast-growing industry.
Earlier this year economists at Equifax published a commentary that boldly asserted "The Subprime Auto Bubble is Fiction, Not Fact."
To support that claim, Equifax compared the growth in auto loans to those it called "nonprime" borrowers with those to consumers who had "prime" and "super-prime" credit scores. The firm found that loan originations to the less creditworthy group grew by 2.4% from 2013 to 2014, while loans to the more creditworthy borrowers swelled by 5.1%.
The implication was clear: why all the hand-wringing about subprime auto lending when loans to prime borrowers have actually been growing at a faster rate?
But a close reading of the report showed that Equifax was counting consumers with credit scores as low as 620 as prime borrowers — quite a low threshold, and lower than the one that Equifax had used previously.
At the bottom end of Equifax's so-called prime segment, more than one out of every three borrowers is expected to become severely delinquent, according to a document published by the company.
The low cutoff skewed the numbers in a way that made the growth in prime auto lending look larger, and the growth in subprime lending appear smaller, than they otherwise would have.
When asked about the report's methodology, Dennis Carlson, one of the Equifax economists who authored it, said in an email: "We chose 620 as our cutoff based on conversation with industry experts inside and outside of the company."
"Had we chosen a different cutoff, we believe the trends would remain materially the same," he added.
The Equifax report is an example of how, in consumer lending, the terminology used to describe the creditworthiness of borrowers often sows more confusion than understanding.
At a time when regulators, prosecutors and the media have stepped up their scrutiny of subprime auto lending, some in the industry have changed their definitions of key terms in ways that downplay the risks involved. Others have left the buzzwords undefined.
The parallels with subprime mortgage lending are hard to miss. During the mortgage boom a decade ago, government-backed mortgage giants Fannie Mae and Freddie Mac used language in ways that minimized their perceived exposure to high-risk loans.
"This is what we saw in the mortgage-backed markets," said Mark Williams, a lecturer at Boston University's business school and a former federal bank examiner. "So it's kind of repeating the lessons still not learned from the mortgage crisis."
"Subprime," "Otherwise Subprime" or "Subprime-Like"
Between 2005 and 2007, Fannie defined a subprime loan as one that was originated by a company, or part of a company, that specialized in subprime loans, the Financial Crisis Inquiry Commission later found.
Using that definition, Fannie stated that subprime loans accounted for less than 1% of its business volume during those three years. But at the same time, 5% of Fannie's conventional, single-family loans went to borrowers with FICO scores of less than 620, the commission reported.
A similar issue involving the use of terminology arose in a recently settled government lawsuit against three former Freddie Mac executives.
The Securities and Exchange Commission alleged that the officials, including former Chief Executive Officer Richard Syron, drastically understated Freddie's exposure to subprime mortgages between March 2007 and August 2008.
One senior executive told investors that Freddie had "basically no subprime exposure," even as the company had more than $141 billion in exposure to loans that were described internally as "subprime," "otherwise subprime" or "subprime-like," the complaint stated.
But the SEC's case turned out to be a dud, dragging on for four years before finally being resolved on terms that barely amounted to a slap on the wrist. The three former executives agreed to temporary employment restrictions and to fork over a total of $310,000, payments that will be covered by insurance.
A settlement document filed in U.S. District Court last month explained why the SEC's case fizzled.
"During the relevant period," the document stated, "there was no one universally accepted definition of subprime that was used by market participants."
The SEC's problem was this: it is hard to prove that someone made a misleading statement about subprime mortgages when no one can even agree about what a subprime mortgage is.
"Nonprime," "Near-prime" and "Aspiring Prime"
In auto lending, the terminology used to describe various market segments is perhaps even more slippery than in the mortgage realm.
In the much larger mortgage market, the Home Mortgage Disclosure Act provides some agreed-upon terminology, as do the conforming loan standards established by Fannie and Freddie. But there are no parallel standards in auto lending, said Chris Kukla, senior vice president at the Center for Responsible Lending.
"I think unless and until some kind of data reporting scheme is put in place, you're going to see these very different definitions," he said.
In auto lending, industry participants have not even found common ground on what the various risk categories should be.
The word "subprime" has fallen out of favor in auto lending in the wake of the subprime mortgage fiasco. "The concept's been tarnished by what happened in the financial crisis," said Chris D'Onofrio, a senior vice president at the ratings firm DBRS.
Sometimes, the euphemistic word "nonprime" is used as the word "subprime" used to be — describing borrowers who do not qualify for prime credit offers. "Subprime" is frequently the descriptor for a smaller subcategory of loans.
In a September 2014 report, Standard & Poor's segmented the auto lending market into "subprime," which referred to FICO scores below 620; "nonprime," which ranged from 620 to 659; and "aspiring prime," which went from 660 to 719; as well as "prime" and "super-prime."
Other market participants use different categories, and establish different cutoffs to distinguish between them. "It's not a hard and fast, solid line between the different categories," D'Onofrio said.
The lack of standard terminology can make it hard for investors to assess the public statements of executives at U.S. auto lenders.
For example, Ally Financial CEO Jeffrey Brown said this during the firm's most recent earnings call: "While we're not doing any of the deep subprime, we have widened the buy box a little bit."
But when asked to clarify how Ally defines "deep subprime," a company spokeswoman declined to say. Instead, she stated in an email: "We view nonprime as below 620 FICO."
Further confusion can arise when the definitions of key terms are changed.
Experian publishes a widely cited quarterly report on the U.S. auto finance market. As recently as the second quarter of last year, Experian defined "nonprime" loans as those to borrowers with scores of 620 to 679, "subprime" loans ranged from 550 to 619, and "deep subprime" loans were those that went to borrowers with credit scores below 550.
But in the third quarter of 2014, Experian lowered its ranges, even though it was still using the same scoring model. Now "nonprime" borrowers ranged from 601 to 660, "subprime" encompassed borrowers with scores from 501 to 600, and the "deep subprime" category included borrowers who scored from 300 to 500.
One effect of the revisions was to mask the prevalence of new loans to less creditworthy borrowers.
In the second quarter of 2014, 36% of new auto loans went to borrowers with credit scores below prime, according to Experian. The following quarter, once the new definitions were in place, that figure had fallen 28%.
Melinda Zabritski, senior director of product and marketing in Experian's auto finance group, said that the definitions were revised in order to conform to the use of auto-lending terminology elsewhere in Experian.
"There was really no market reason why we made the changes that we did. It was more internal business reasons," she said.
What's in a Name?
None of this proves that there is in fact a bubble in subprime auto lending. The industry's supporters make some strong points to support the contention that there's no cause for alarm.
They note that auto loan originations to consumers at the bottom end of the credit spectrum were significantly higher a decade ago than they are today. They also point out that the sector weathered the crisis surprisingly well, as Americans who needed cars to get to their jobs prioritized paying their auto loans ahead their mortgages.
On the other hand, the industry's critics point to the growing trend of 72-month and even 84-month loan terms as evidence that lots of consumers are driving home in vehicles they cannot really afford. Longer loan terms result in smaller monthly payments, but they also mean sharper falls in the value of the cars used to secure the loans.
The key question, in terms of assessing credit risk, is whether lenders are pricing subprime loans appropriately. And auto loans to consumers with damaged credit are clearly expensive.
A recent WalletHub report compared what borrowers with high and low credit scores would be charged for a five-year, fixed-rate new car loan of $20,000. Borrowers with credit scores of 720 or higher would pay $1,557 in interest over the life of the loan, while those with scores from 620 to 659 would pay $6,918, the report found.
The labels that get affixed to the various risk bands — "nonprime," "subprime," etc. — may not matter much inside the companies that specialize in auto lending. After all, lenders look far beyond credit scores, assessing factors such as the ratio between the loan and the value of the car, as well as the ratio between the monthly payment obligation and the borrower's income, in an effort to gauge risk.
"As you get into the folks that have had credit problems of various types, I think the use of just a FICO score as a predictor of behavior is a little harder," said Christopher Gillock, a managing director at Colonnade Advisors.
Likewise, investors in bonds linked to auto loans typically have access to a wide range of information that is relevant to an assessment of credit quality, including average interest rates, average loan terms and geographic mix.
But for regulators and researchers seeking to assess the big picture in U.S. auto lending, the industry's confusing and often contradictory use of terminology can be an impediment to understanding. That is troublesome, given that industry participants have revised key definitions in ways that make the risks appear smaller.
"If definitions were changing to be more conservative, then that would be very positive," Boston University's Williams said. "But what you're finding is the opposite."