Title Lending’s Big Question: Dude, Where's My Car?

In a new paper on title lending with Katie Fritzdixon and Jim Hawkins, we report data from a survey of over 400 title lending customers across three states. To introduce this work, we wanted to start off by talking about the important issues that title lending raises. The biggest question, by far, is how many title borrowers end up losing their car?

Academics, consumer advocates, and policymakers have focused on title borrowers' risk of losing car and hence their way to work, sometimes exclusively, as a reason for restricting title lending. To figure out how many borrowers are losing their only way to get to work, we must unpack how many borrowers are losing their cars and how many of those have no other way to get to work. The best answer to the first question is gleaned from reports from state regulators. Several states require title lenders to provide information about their operations. The repossession rates of a few states’ lenders are summarized in the table below:

State	Year	# of New Title Loan Agreements	# of Repossessions	Repossession Rate
TN	2008	161,417	14,832	9.18%
	2007	139,319	18,199	13.06%
	11/2005–6/2006	92,489	10,933	11.82%
	2004	250,593	17,313	6.91%
OR	2009	17,820	2	0.01%
	2008	10,136	1	0.00%
	2007	8,568	32	0.37%
	2006	15,726	125	0.80%
	2005	17,801	114	0.64%
ID	2010	34,247	2382	6.96%
IL	10/2009–6/2011	155,094	7,334	4.73%
MT	2009	12,727	599	4.71%
VA	10/2010–12/2010	24,975	194	0.78%
TX	1st 3 Quarters 2012	19,144	259,691	7.37%

While backing out precise repossession rates from these figures is difficult, it appears that an educated estimate of--at most--around 10% of new title loans result in the car being repossessed. Thus, at least 90% of borrowers don’t lose their car. 

For data on the second question—whether the vehicle borrowers used to secure the loan was their only means of getting to work—we turn to our survey. We ask customers the following question:

If you lost the vehicle that you used to get your loan, how would you get to work?
Do not work outside the home	12.16%
Use public transportation	14.14%
Another vehicle I own	40.20%
Walk to work	8.93%
Rides from friends/family	17.37%
Buy a new car	8.68%
No other way	14.64%
Number of Observations	403

So, based on our survey, around 15% of people using title loans are using their only means of getting to work as collateral for the loan. Assuming that our subjects are representative of title loan customers generally and assuming that people losing their cars have the same likelihood of being in that 15% as all customers, it appears that only around 1.5% of borrowers using title loans will end up losing their only way to get to work. These assumptions may be heroic, but the data we present here is literally the only data generated by academics available to answer this question--a question clearly at the heart of the debate over title lending. Based on the best estimates available, it appears that the frequency with which title borrowers end up stranded at home and unable to get to work is quite minimal. 

We don’t think the risk of losing one’s only way to get to work is the real problem with title loans, and we will turn in subsequent posts to the other risks title borrowers face.

Car graphic from Shutterstock.

Comments

[Alan White]

There are some assumptions underlying the proposed interpretation of data here. If 10% of loans result in a repo, it does not follow that 90% of borrowers don't lose their car, if borrowers have multiple loans.

As for the alternative transport to work issue, getting a ride with friends and buying a new car may be ambiguous responses, especially allowing for the optimism bias. Those categories may represent borrowers with at least some difficulties posed by losing their car. In other words, one could say 66% of borrowers report not working or having reliable alternative transportation.

Finally I would want to inquire a bit about the Oregon and Virginia data to see if some legal environment or reporting issue skews them from the other states.

[Jim Hawkins]

Good points Alan.

On the issue of losing one's car, the borrowers would have to have multiple loans from multiple lenders for the 90% number not to be true. All of the figures are new loans, not rollovers of existing loans, and almost no lenders allow borrowers to get title loans if there is an existing lien on the vehicle.

But, it is possible people changed lenders and then had their car repossessed. I don't think the risk of this is very high though because, in some of the states at least, the average loan duration was very long, making it impossible for borrowers to have loans from more than one lender. (Virginia's average was 305 days; Illinois' average was 300 days.)

On the other hand, the numbers might overstate how many people that lost their vehicles because in some states (like Texas) the figure is how many cars were repossessed and not how many cars were sold. It is possible that people redeemed their vehicles after they were repossessed.

[Paige Marta Skiba]

Thanks Alan. A goal of ours is to import some empirical analysis into the debate so we appreciate comments on what data would help.

[Adam]

Jim, Paige,

Interesting paper, but it leaves me more perplexed than anything, as the answers from your sample (volunteers at 9 outlets of one lender, correct) just don't make a lot of sense to me. I'm willing to go with the data, but I find myself wondering about what is really being reported and the reliability/representativeness of the data. Two issues in particular jump out at me.

First, are you sure that "new loans" does not include rollovers? From a legal perspective, a rollover is a new loan, just as a refinancing of a mortgage is an "origination" even if it is from the same lender. Unless the states are also tracking rollovers as a separate category, I'd be skeptical about the data.

Based on a perusal of Schedule D to TitleMax's bankruptcy disclosure statement (ECF 390), I would think that repo figures are somewhat higher. The disclosure statement doesn't have repo figures, per se, but it does have loss allowances and a valuation of repossession property. Loss allowances run at approximately 12-14% of title pawn receivables (TitleMax goes up to 80% LTV, and doesn't require collision insurance), while repossessed property is on a monthly basis about 1% of receivables. That suggests to me that we are looking at repo rates of more like 12% annually. Perhaps others will have a different take on the financial statement, which is available here:

http://documents.epiq11.com/docket/docketlist.aspx?pk=f24417f7-5c0f-4968-8443-c8031f5caa47

On the other hand, if the repo rate data is right, then the question is why are auto title loans so expensive? I get that small dollar lending has fixed costs that have to be amortized over the life of the loan, but with loan default rates, I would think title lenders would have to be raking it in (TitleMax thinks it hit a gold mine--they just had a refi hickup in 2009), and that competition would be driving down rates. Something here doesn't seem right.

I would also take issue with your interpretation that only 15% or so of borrowers are using their only means of getting to work as collateral. i think this sort of survey data has to be taken with a grain of salt, even if it is representative. First, the idea that 40% of auto title borrowers have a spare car just sitting around seems hard to believe. If consumers have a spare monetizable asset like that sitting around, why are they taking out a title loan? Second, even if that number held true for a broadly representative sample of title borrowers, what are we to make of the 8.68% who would "buy a new car"? Do title borrowers really have the funds sitting around to buy a new car when they can't pay off a title loan? Third, it is not clear that the 40% or so who would take public transit, walk, or get a ride with family/friends, really have sustainable transit methods or at least ones that don't have major impacts on their lives. And fourth, isn't it possible, indeed likely, that the respondents are overoptimistic about their ability to get to work without their car? I would think that a population that would gamble with their cars would, by its very nature, tend to downplay the costs of losing their cars.

I don't mean to be too harsh here--the paper is really thought provoking and underscores how limited our knowledge is in this area.

[Jim Hawkins]

Thanks for the questions Adam.

I am confident that the reports separate new loans v. rollovers at the same lender. I emailed the regulators at states where the report was unclear to verify, and in other states, the reports do, as you say, report rollovers as a separate category. For instance, check out Texas' here: http://www.occc.state.tx.us/pages/industry/CAB/12142012%20CAB%20Third%20Quarter%20Report.pdf

In terms of profitability, I don't think low repo rates means low default rates, which I take your comment to assume (sorry if I have misread it). It could be that lenders do not repo because the collateral will not pay enough to cover the costs of repossessing it. If the car is worth $500, the lender would barely get the repo costs back from a sale. Also, lenders say that people default when the car loses its value completely (because of engine trouble or an accident, for instance).

I think your concerns (and Alan's) about people's perceptions about the consequences of default make sense. I am curious how we could design a question or do research to get a better answer to the question, but I am sure there is a better way.

As you say, maybe people are wrong about what would happen to them if they lost their car. But, I think this information is more reliable than claims by academics who are removed from these borrowers' lives that borrowers are frequently losing their only way to work.

I guess I also see their responses are more plausible than you do. To me, it seems rational to take out a title loan even if you have another car sitting around because the extra car might have some function that is worth more than the cost of the loan or the hassle in selling the car might not be worth it. For whatever it is worth, in an earlier attempt to ask people this question, I asked a small number of borrowers how many cars they had in their household, and 80% (n=28) had more than one car. Obviously, this is too small to mean much, but until someone produces data that title lending customers have no other way to get to work, I think we should be cautious about regulating the transaction on that basis.

[Stephen Brown]

why do comments disappear?

[Stephen Brown]

I think that your 15% * 10% = 1.5% is relevant only if the probability distributions are independent. I would be surprised if these two distributions were not positively correlated. For example, I expect that those who are wealthy enough to have a spare vehicle are less likely to default on the title loan than those who do not have a spare vehicle.

[Bob Lawless]

To Stephen Brown: Our blog hosting service, Typepad, has been updating and improving its fight against comment spam. The amount of comment spam is a problem, and their efforts are welcome. According to Typepad, the performance of its new spam filters will improve over time and that should reduce the incidence of false positives.

[Benjamin Martineau]

I am a title lender here in Idaho - one location. From my perspective the statistics here are quite reasonable. My repo rate is just above 4% and of those I repo, usually it is a second car - not the family's only vehicle. In fact 2 of the 3 most recent repos I've done have been classic cars that were sitting in a garage not being driven at all.

I doubt it is policy at larger title lenders, but I know myself and another local lender here in Idaho consider the impact a repo will have before deciding to get the car. If it is someone's only transportation to work, I am much more hesitant to repossess.

[Paige Marta Skiba]

Thanks for your comments Mr. Martineau. We are hoping the paper I mentioned is the first in a larger research program on title lending. So we need to hear about how lenders operate. Thanks!

[Charlie Tanner-Mot]

I would like to know what the loans that were taken out were initially used for - even if they were separated into categories of loan-spend type, I think it would be interesting to see if there is a correlational cause based on this