More on the subprime auto lending ‘bubble’

By Todd Zywicki on September 5, 2014

This was published in the Washington Post on September 4, 2014

A couple of weeks ago I discussed the Shoot first, aim second editorial in the New York Times calling for federal regulators to act swiftly and aggressively with respect to what it perceived as a growing threat caused by subprime auto loans. There were only a few problems with that editorial:

  1. It provided no evidence that there was a subprime auto bubble.
  2. The only data the Times did present (in an accompanying news story) showed that the volume of subprime auto loans was growing but still well below its peak prior to the financial crisis.
  3. There was no clear substantial consumer harm.
  4. Delinquencies and defaults on auto loans are near or at their all-time lows
  5. There is no plausible threat of systemic risk to banks even if there was a bubble.

Still, of course, swift and aggressive action must be taken!

While the call for swift and aggressive action is silly at this point, there is reason to keep an eye on this market as it ramps back up. There’s still no evidence that anyone other than investors would be hurt if there really is a bubble and there is clearly no systemic risk.

Since that time, more data has come out on the question, and so far it shows little evidence of a subprime auto bubble.

The first was an analysis from four economists at the New York Federal Reserve that came out right around the time of my post on this (although I wasn’t aware of it at the time). Their analysis was based on the bank’s Quarterly Report on Household Debt and Credit. They note that the fuss about auto loans has been because there has been a substantial growth in the volume of auto loans since the trough of the recession in 2009. Moreover, the dollar value of originations for those with credit scores below 660 has risen more rapidly than for other credit segments. But they provide three important caveats. First, this growth comes relative to an exceedingly low baseline ” subprime auto lending collapsed after the financial crisis. Second, the growth has been concentrated in auto finance companies, not banks, which obviously reduces any potential issues related to systemic risk. Third, they note that the growth in the subprime sector has been with respect to the dollar value of the loans being originated, not the number of loans. As the analysis puts it:

The left panel shows that auto finance companies’ lending to subprime borrowers was hit very hard by the crisis: between mid-2006 and the end of 2009, originations to the below-620 group fell by more than two-thirds. Conversely, in the recovery, subprime lending by auto finance companies has shown considerable strength: since the trough, auto finance company lending to each of the three lowest credit score groups has more than doubled. In contrast, the right panel shows that banks’ lending to subprime borrowers has historically been lower than that to prime borrowers, and despite the increase in recent subprime originations, the share of subprime borrowers remains small.

More recently, last week Equifax released an analysis of the subprime auto market by Amy Crews-Cutts and Dennis Carlson. They too throw cold water on the concerns about a subprime bubble. As they put it at the outset of the report:

Surprisingly little data has been shared in the press. Many of the arguments have been rhetorical, based on the following premise: Subprime lending caused the financial crisis, ergo subprime lending is dangerous. However, that generalization does not always account for actual subprime loan data. This Economic Commentary will present an up-to-date analysis showing the current state of subprime automotive lending.

And they acknowledge that there might be some superficial similarities between the pre-crash housing market and the auto market today (some similarities in factors that indicate some effervescence such as quickly rising sales of new cars and an even faster rate of increase in auto lending). Yet, there are much more important differences: most notably, that while subprime auto lending has been growing in dollar value, it remains below its peak volumes in the past and, unlike the housing market, that peak era was not one in a frenzied state:

But looking closely at those similar factors, we see that the auto market of today is just now recovering to pre-recession levels and, unlike the mortgage market immediately prior to the recession, the auto market was not in the same frenzied state at that time. For example, new car and light truck sales reached their peak level in the first quarter of 2000 and averaged 16.96 million units for the six-year period ending December 2005. Even with these high sales volumes of new cars, the average age of a car on the road rose by almost a full year, from 8.9 years old in 2000 to 9.8 years in 2005. Today the average age of a registered car stands at 11.4 years and new cars and light truck sales have totaled 15.98 million units over the twelve months ending  July 2014. Clearly there is not an asset bubble in the auto market today. (emphasis added)

In Figure 1 we turn to the Equifax data on auto lending. On a year-to-date basis, the volume of loans originated has been increasing on a linear path since 2009, much like auto sales, but the share of these loans that is going to borrowers with subprime credit scores, measured here as having an Equifax Risk Score below 640, has been stable for the past three years and remains well below the high point in 2007.

Moreover, consistent with the New York Fed’s analysis, they observe that while the number of loans has not increased very much, average loan size has. But that too does not suggest a bubble is forming:

The average loan balance at origination for a subprime credit score borrower has been rising, but this is not due to relaxed lending standards. The average loan size for recently originated loans by the borrower’s Equifax Risk Score is shown in Figure 3. As a borrower’s Equifax Risk Score increases (e.g. becomes less risky), so too does the average loan amount. This chart is representative of originations over the past several years as the loan amount“credit score relationship has been quite stable. Rather than giving larger loans to borrowers across subprime credit-score bands, lenders have been keeping loan sizes the same conditional on credit score but originating more loans to borrowers with less-risky scores as we showed in Figure 2. Thus average loan sizes have been increasing because relatively fewer high-risk borrowers with relatively smaller loan amounts are getting loans today.

One additional virtue of the report is that they recognize that consumers and the economy will be hurt if regulators blunder around taking premature swift and aggressive action to address a potentially imaginary bubble:

Given all of this information, we believe that while the subprime lending segment needs to be monitored carefully, the evidence does not support that there is a bubble forming in the auto lending space. It is also beneficial to consider than an unmet need is being satisfied. Assuming originations and loan performance in the space remain as they are today, this may benefit the overall economy, as well as the individual participants, in the long run.

As the saying goes ” read the whole thing.

Perhaps the Times has some secret stash of data or knowledge unavailable to the New York Fed or Equifax that provides actual evidence in support of the theory that there is a subprime bubble that raises substantial risk to the economy. At this point though, one suspects that is not likely the case.