Duck, Duck, Duck . . . Bankrupt!

Recent CreditSlips guestblogger Adam Levitin send me a short piece from the American Banker, "Data Tool from Visa, Experian" about a new risk model called "BankruptcyPredict." The technology is proprietary, of course, but one innovation seems to be the use of data from "all forms of payments cards" processed by Visa, not just credit cards. This means that your pin-based debit transactions processed through the VISA network get incorporated into the model, as well as all the usual stuff in an Experian credit report. This potentially gives the model a great deal of information about your bank account and non-credit spending, not just loans. The model purports to be able to identify consumers who are very likely to go bankrupt up to two years in advance.

My first thought on this: WOW--two years in advance! Although more work is needed on this point, the data that I've seen suggest that most people who file bankruptcy report struggling seriously with their debts for about a year or a bit more before they go broke. This model would outperform consumers themselves, it seems. I imagine a room full of consumers in financial trouble (maybe seeking credit counseling at a face-to-face location). From the consumers' perspectives, they are desperately hopeful to stay out of bankruptcy, but they know it is likely to happen to some people who get in deep financial trouble. They hope that they are the "ducks" in the room but wait out the next months in tense silence to see whether they tip over the edge into bankruptcy--who will be the "goose." Visa/Experian apparently are able to sort out the "ducks," who will continue to pay interest and fees and struggle along, from the "geese," who will seek legal relief through bankruptcy. I also wonder whether consumers would want this information? If the Federal Reserve or the National Consumer Law Center or some enterprising academic built a similar model and made it freely available, would consumers use it? Is there a societally beneficial use to a BankruptcyPredict model? Should financial educators pay for the service so they can counsel clients more realistically on the likelihood of that a client's financial problems will deepen and lead to bankruptcy?

Comments

[Jason Kilborn]

This would be fantastically useful to a new movement in Europe that I hope will catch on in the U.S.: diverting "hopeless" cases from counseling directly into the bankruptcy system. If this model could accurately predict which consumers had no chance of working out something reasonable with their creditors and following through on that workout, it certainly would take a load off of the overburdened counseling industry. After Sweden scrapped the idea of mandatory pre-bankruptcy counseling entirely in January, a new bill in Germany seems poised to scale back on forcing every debtor into pre-bankruptcy counseling, too. This model would vastly advance these movements. Of course, I suspect it isn't really much of a surprise at the counseling stage that many, many people simply can't be helped (either because they can't really pay, or because creditors have unrealistic expectations/demands). Sweden seems to have it right. If the counseling stage is too late, I strongly suspect that any earlier than that is really too early (as you seem to suggest). Strikes me as a case of correlation not equalling causation.