Auto Title Lending: Exploding Toasters

The CFPB has a new report out on auto title lending, and the findings are jaw-dropping. If ever there was a consumer financial product that looks like an exploding toaster, it is an auto title loan.  Default rates on auto title loans are one in three, with one in five resulting in a repossession. Is there any consumer product that is tolerated when one out of three products blows up? Even one in five? 

There's a lot of good data in the report (which assiduously avoids any interpretation, but just presents the facts), but beyond the default rates, here's what really jumps out at me: over 80% of the loans roll over and around half result in sequences of 10 or more loans.  That means that rather than viewing auto title loans as short term products with an extension option, they are really used more like longer-term products with a prepayment option. But more importantly, it tells us something about how to interpret default rates.  

Defenders of the auto title lending industry, such as Todd Zywicki, have argued (and also here) that the default rates are low. As I have previously pointed out, Zywicki's calculations are based on treating each loan in a multi-rollover sequence as a separate loan, thereby massively inflating the denominator in the default rate, and others have unfortunately relied on Zywicki's calculations. The CFPB study is the first study I've seen that calculates the real default rate and ignores the artificiality of treating rollovers as separate loans. 

Based on these default rates, let me suggest a new metric for thinking about abusive or unfair or unconscionable lending. The primary consumer protection concern in lending regulation is that consumers will end up with unmanageable debt burdens that will in turn result in consumption deprivations for their households and possibly a drain on public resources.  Historically society has tried to use the cost of the loan as the measure of whether a debt burden is unmanageable--that's usury laws. But how about looking not at the costs, but at the results, or at the anticipated results:  if a loan product's default rate is over a certain threshold, isn't that prima facie evidence that there is something wrong with the product?  Now sometimes a default rate ends up higher than anticipated--lenders miscalculate.  But what if the product is designed to have a default rate of 30%?  That's like selling tasty food that one knows has a 30% chance of being contaminated with a harmful microbe. We would never allow that in the public health context, so why is it ok in the financial context? 

Now come on Adam, you say, no lender would ever design a product to have a default rate of 30% because the losses will eat away the lender's profits.  Nuh-uh.  The traditional amortized loan model aligns the interests of the lender and borrower--if the borrower doesn't repay, the lender will lose principal and interest. But that's not how non-amortizing sweatbox-type products like payday and auto title loans work. (Ronald Mann via Jay Westbrook coined the term sweatbox in the context of a political economy explanation of credit card lenders support for BAPCPA, but I think the concept is even more applicable to payday and auto title loans.)

Sweatbox products are non-amortizing, and are designed to be rolled over.   Sweatbox lenders anticipate that borrowers will in fact rollover the products multiple times by lending an amount and on a payment schedule that the borrower cannot easily repay the loan from his or her regular income.  The key to understanding the sweatbox model is that even if the loan ultimately defaults, the lender will have recovered all its costs and made a profit provided that there are enough rollovers.  In such a situation, a lender may be willing to tolerate--heck, might embrace--higher ultimate default rates in order to grow the borrower base that will be profitable in the sweatbox.  Sweatbox products like payday and auto title loans pervert the traditional alignment of borrower and lender interests. 

Let's say you agree with me that actual or anticipated default rates might be an appropriate metric for evaluating abusiveness or unfairness or unconscionability. (Or maybe even deceptive lending, as I think it's material to know that a lender anticipates that there's a ⅓ chance you won't be able to repay your loan.) Aren't we back to the good old usury law problem of figuring out what is an appropriate cutoff?  Yes and no.  I'm not sure exactly where the line should be drawn (I'm all ears for suggestions), and one might think of any line drawing as a starting point for a safe harbor or rebuttable presumption, but there are some situations where the default rates are pretty clearly over any line that might be drawn. 

So what we have with title lending are real exploding toasters. At least they're not exploding Ford Pintos. Or Chevy Volts. 

Comments

[Robert White]

Wasn’t the government’s motive in the Chrysler and GM bailout similar to consumer protection concerns with auto loans; ie, “The primary consumer protection concern in lending regulation is that consumers will end up with unmanageable debt burdens that will in turn result in consumption deprivations for their households and possibly a drain on public resources.”

When automakers go bankrupt jobs are lost which “in turn result[s] in consumption deprivations for [] households and possibly a drain on public resources”.

The conclusion is that bad loans are tolerated as long as they insure demand for purchased products and create jobs.

So how do you manage an economy?

With debt, bankruptcy, and laissez-faire economics?

Not a criticism, just a question.

[Robert White]

I should have said "When automakers go out of business jobs are lost ..."

[Ken]

Pretty minor corrections, I believe, based on how I read the study.

First, the study is on single-payment title loans, so I think that it may overgeneralize to just say title loans.

People re-borrow or extend non-single payment title loans, so I do not think the discussion of reborrowing/rollover sufficiently implies that it's talking about single payment title loans only.

Second, and even more minor, the default rates/repossession rates are not 1 in 3 and 1 in 5.

Those are the default rates/repossession rates on sequential single payment loans. Now, 80% to 90% of the single payment loans end up being sequential loans, so 1 in 3 and 1 in 5 are pretty close. But I think it's worth mentioning because of the first correction.

I imagine -- and this is literally me just guessing -- that rollovers on non-single payment loans are lower than the single payments ones. So even if the 1 and 3/1 in 5 figures could also be applied to sequential non-single payment loans, the default figure may still be overstated because of the reduced rate of re-borrowing.

[Adam Levitin]

Ken is right regarding the limited scope of the report. It's only single-payment title loans. Seven states (according to Pew) have only single-payment products, while five have only multiple-payment products. The other thirteen states that allow title lending have both single and multiple payment products. I don't know how the breakdown of borrowing goes overall by number of loans (installment loans tend to be larger loans, I believe). The difference might explain some of the difference in the statistics in the CFPB report and in those in Jim Hawkins' Credit on Wheels (2012).

Ken is wrong, however, regarding the default rates. The report provides the statistics in terms of "loan sequences," which a casual browser might reasonably take to mean only sequences of multiple loans, but that's not how the report is using the term. On p.8 it defines a sequence to include single, non-rolled over loans: "A loan sequence consists of an initial loan and any subsequent loans that are taken within a specified period of time after the prior loan." Accordingly, on p. 4: "Only about one-in-eight loan sequences consist of a single loan that is repaid without reborrowing."

[Adam Levitin]

Regarding Robert White: there's lot of bad behavior tolerated in the economy. GM and Chrysler should never have been allowed to perilously underfund their pension and healthcare obligations. Once that was the situation, however, the government did the best it could to salvage the companies and was relatively successful (although people seldom count all of the jobs shed at dealerships as part of the casualty county).

[Ken]

I agree. I erred. I simply put too much stock into the word "and" in the sentence quoted, which basically appears on page 3 as well. I interpreted it to require at least one subsequent loan in that 14/30/60 day period rather than reading it as including those if such subsequent loans were made.

That said, yes, the "key findings" on page 4 and the summary of possible outcomes on page 11 establish that a single loan - repaid or defaulted -- constitutes a loan sequence.

[Knute Rife]

So Zywicki's been caught out again. Quelle surprise. Wish I could make a lucrative living making the arguments he makes, except that in actual court there are these things called Rule 11 sanctions.

[Robert White]

Here is what I don't understand.
A Bankruptcy Court can sanction, hold you in contempt, and incarcerate you but can't decide the debtor's breach of contract action against a 3rd party.
If the BK Court doesn't have authority to decide common law claims against 3rd parties they how is incarcerating a 3rd party constitutional?

[Knute Rife]

Robert White:

The common law claims are beyond the Constitutional subject matter jurisdiction of the bankruptcy courts, and the parties can not confer subject matter jurisdiction by acquiescence. The parties CAN, however, confer Personal jurisdiction by acquiescence, and sanctions are imposed on parties who have misbehaved after submitting themselves personally to the jurisdiction of the court.

[Robert White]

Submitting oneself to the jurisdiction of the BK Court might explain Article III authority to adjudicate the Debtor’s issues but not creditors and third parties - at least that is what the majority said in Stern v Marshall.
For a creditor who complains about sale procedure and estate counsels’ fees “He ha[s] nowhere else to go if he wished to recover from [the] estate. “, id .
I sense an axiom in play - "If the Court has authority to rule it has authority to punish".
Can there be one without the other?
And of course there should be an exception to the rule for sanctions and contempt leading to incarceration.

[not_a_lawyer]

Why do lawyers and their professors insist on equating borrower default with lender abuse--as if the lender's make money when borrowers fail to repay. Even secured lenders, like auto title lenders, prefer repayment to default. The fact they are willing to lend despite a 20 percent chance of loss is to their credit (unintended pun...)

[Adam Levitin]

not a lawyer misses the point, which is about how sweatbox lending changes things. Traditionally lenders prefer repayment to default. But in a sweatbox model, a lender will trade higher default rates for the larger borrower base because of the lower losses given default.

[not_a_lawyer]

Lenders tradeoff higher default rates for higher volume in traditional, economic models of credit market and in reality-it's called moving out the risk-return frontier. There is nothing remotely novel about that tradeoff-except perhaps to lawyers and their professors. What's novel is the tendency of legal profession and CFPB to equate borrower default with lender abuse. You praised them for avoiding interpreting their 20 percent default finding as problematic but they didn't have to-you are doing it for them.

[Adam Levitin]

Sorry not_a_lawyer. The sweatbox model isn't simply moving out the risk-return frontier. It's not the same as trading off higher default rates for higher returns. In the traditional model, consumers sort into two groups: defaulters and performers. The higher rates paid by the performers offset the losses from the defaulters...up to a point. In the sweatbox model, it doesn't matter if _everyone_ were to default because the defaults occur after the fees and interest collected exceed the principal loaned. In a well-executed sweatbox, there are never any real losses given default, even if default rates are 100%. (Yes, one can play with the amortization to argue that there are losses, but if $100 is loaned out and the borrower repays $300 before defaulting, it's hard to call that a loss in any meaningful sense.) That is a (relatively) novel business model, although I sometimes cite to Faulkner for an example of Old Man Snopes' lending operation.

In any event, to make a sweatbox loan work, the borrower does not need to be able to pay off the loan, but only to pay enough in fees and interest to offset the principal and costs before defaulting. That's what's novel.

fwiw, I didn't praise the CFPB for not interpreting the 20% default finding as problematic. I merely pointed out that the CFPB's report doesn't offer any opinions or interpretations. While I happen to think that it is smart for the CFPB to avoid doing so at this juncture, I didn't praise or criticize this in the original post.

[not_a_lawyer]

No need to apologize.

"in a well-executed sweatbox there are never any real losses given default" First, you make it sound like the "sweat box" is a real thing, with some well-executed and some not. In fact, it's a law review article in a not particularly distinguished Review that does not have a single equation. I am not aware of any theoretical counterpart of a "sweatbox" in economic or financial models of credit markets. Second, if default is costless to lenders, why aren't the borrowers being charged the risk-free rate (there are plenty of auto-title lenders competing for borrowers)? The fact that borrowers have to pay a risk premium suggests default is not, in fact, costless. For example, what if default occurs before the lender covers her cost of capital, time, and the risk of law suits if too many borrowers default and the lawyers come-a-knocking with a class action lawsuit? Then it's the lender in the sweatbox.

[Adam Levitin]

1. Ronald Mann explained the sweatbox idea in the context of credit cards and the political economy of BAPCPA in an excellent symposium issue of the Illinois Law Review. The idea is pretty well accepted within legal scholarship, even if it hasn't been picked up on by other disciplines. My point in this blog post was extending that concept to short-term small-dollar loans, as I have in certain expert reports.

2. Why on earth does it matter where the idea was published or whether it has equations? Neither of those factors has any relation to whether the idea is correct. Law review placement is a poor proxy for quality, and equations are simply one of many ways of conveying information; they don't mean that an idea is right.

3. Economic and financial models do not account for sweatboxes because very few economists and finance scholars study short-term small-dollar lending and most of those who do simply aren't thinking about this issue. (Let's put aside the problem that economists are often not good at reading the legal literature.) I hate to burst your methodological bubble, which clearly favors number crunching, but legal scholarship is probably the strongest field of consumer finance studies because legal scholarship is attuned to the institutional and situational framework of consumer finance. That's something equations don't really catch.

4. I gave an extreme example of a perfectly executed sweatbox in my previous post--the sort of idealized model that one might find in, hmmm, an economics article with equations. In real life, no one is going to have a perfectly executed sweatbox--some borrowers default before their payments exceed principal and costs. Thus default isn't costless to lenders, but it is less costly in a sweatbox model.

5. Lenders can screw up the sweatbox if they lend to borrowers who never repay anything (or who don't repay enough to cover costs). That doesn't disprove anything.

6. When thinking about costs, be aware that there are almost no class action lawsuits against sweatbox lenders because virtually all of them have binding mandatory arbitration provisions in their contracts. The only exception I'm aware of is the litigation against CashCall in the Northern District of California, and even CashCall eventually adopted an arbitration provision. (My expert report in that case is public.) There is a separate risk of litigation from the CFPB or state attorneys general.