Fraud or Not, Measure the Risk!

It’s a large, ugly fact in sub-prime auto finance: many dealers advertise used cars for one price but then sell those cars to sub-prime buyers for more than that advertised price. For example, a dealership officially advertises a used car on their webpage for $10,000 and then sells that car to a sub-prime consumer for $12,000. This practice is prevalent, observable in 25 percent of independent dealer used car sales and 15 percent of franchise used car sales. It’s called “dealer price inflation.”

We need to pay attention to this practice for at least three reasons. First, it happens a lot, upwards of 15 percent of sub-prime financing. So it impacts a significant portion of the retail installment contracts we finance. Second, it negatively impacts loan performance. These loans perform worse: they default 39 percent more frequently, and most of the additional default risk is not captured in financing terms or deal structuring. These loans are simply more expensive to our operations.

The third reason we must pay attention is this practice might actually be considered a form of consumer fraud. There must be a rule about it, but no one has really complained, yet it feels wrong. It feels like the practice could be construed as price gouging of unwitting consumers. But that’s not for me to call. If it is determined this sales practice is fraud, it will significantly shake up our industry, including how we do business and even how much business we do. For now, it’s a huge looming unknown.

Measuring the risk 
At minimum we need to understand the impact this phenomenon has on loan performance. To do that let’s compare the advertised prices for used cars on dealer websites to the sale price shown in loan contracts. Then we can season the loans and measure how dealer price inflation impacts loan performance. The results are as expected, but there are also a few surprise kickers. Here is a summary of the findings:

Expected losses (E[L]) increase a whopping 39 percent when dealers inflated the sale price by more than $1,500. The increase makes logical sense. Consumers who overpay this much are some combination of uninformed and impetuous. Not to mention these consumers and even the lenders end up being far deeper in LTV than they ought to be.
The next two results were more fascinating. There was a 72 percent increase in E[L] if we couldn’t find an advertisement for the car. As if to beg the question, “where did this car come from?” or “of all the cars advertised by the dealer, how did they end up with this one?” Finally, E[L] doubled from baseline when the car is advertised online but the dealer did not show the price on their website. This implies that if a dealer doesn’t make price transparency obvious, it will show up dramatically in loan performance.

Analysis and takeaways
Without regard to the fraud issue, what does this tell us about lending to the consumers who took these loans? We shared these results with Joel Kennedy, the COO of Tru Decision, who offered a valuable insight. “Today we can expect a consumer to shop online ahead of time and go into a dealership with a specific car and price in mind. The fact that consumers are buying these cars that were not advertised, or advertised without a price, suggests the consumers either didn’t do their homework or were flipped. The increase in defaults is a reflection of that.”

Now with regard to the fraud issue, it’s prudent to monitor dealer pricing and selling behavior in case the issue becomes regulated. We can see in the data that many dealers do not practice price inflation on sub-prime consumers regularly. Usually a dealer first tries it out a little before making it a habit. Therefore, if you observe the issue, I advise you to have a conversation with your dealer and they will likely stop. If they don’t, then you may want to step away from them. You will find some dealers employ price inflation on all their sales. These dealers may be lost causes.

Try it yourself
You can run your own quick experiment to capture any price inflation in your loan apps. Here is how. Choose a few dealers that have good submission volume. Visit their websites and record the list of cars they are advertising and the price shown for each of those cars, along with the date you saw this. Record this in a document or spreadsheet. As loan requests arrive over the next month, you can compare the sale price you see on the app to the advertised price you captured in the spreadsheet. The point of this experiment is to show you it happens often. However, to scale this effort and leverage it in an operation you need to automate this kind of process.

Whether or not you consider price inflation to be fraud, be mindful that it happens on a significant portion of our sub-prime originations and understand the negative impact to loan performance. Upon discovering an instance of price inflation in a loan application, the response should be at least one of two things: adjust the loan pricing to account for the increase in default risk, and to communicate to the dealer that selling over advertised price is not acceptable thereby encouraging them to eliminate the behavior.