And the Financial Institution Most Likely to Be Late Responding to a Consumer Complaint Is . . .

One day last spring, I saw in a Google alert that the Consumer Financial Protection Bureau (CFPB) had announced that it was for the first time making public a consumer complaint database.  At the time, I was teaching a course in Empirical Law & Economics at Yale and decided to call an audible.  I came into class that day and projected the raw data (which you can see for yourself by clicking here) and asked the class how we might make use of the information.

With incredible dispatch, Jeff Lingwall and Sonia Steinway merged the complaint data with other datasets and together we started to put together an initial draft analyzing the complaint information.  When the CFPB made the database public, they actively encouraged “the public, including consumers, analysts, developers, data scientists, civic hackers, and companies that serve consumers, to analyze, augment, and build on the public database to develop ways for consumers to access the complaint data or mash it up with other public data sets.”  This paper is our attempt to respond to the Bureau’s call to action.

One advantage of having outsiders analyze the data is that it might be easier to name names.  The Bureau’s decision to release the names of the companies that consumers complained about was initially controversial: in 2012, lobbying groups like the Consumer Bankers Association launched significant efforts to disallow the Consumer Complaint Database from including firm names implicated in consumer complaints.  It’s easy to see why: the complaints paint a clear picture of which companies are responsive to consumers, including how timely they are in responding and whether their responses are adequate in resolving consumers’ issues.  A contestable marketplace of analysts might be able to come to some consensus about how well or badly particular creditors rank.

We ranked financial companies on several metrics: timely response to complaints (within the CFPB-mandated number of days), and percentage of responses disputed by consumers.  For timely and disputed responses, we constructed lists of the companies that had proportions that were statistically significantly above the average.  Here are the five worst-performers (of companies with at least 50 complaints):

5 Worst Performers in Percent Untimely and Percent Disputed 

Note: lists include companies with more than 50 complaints and companies that have a rate that is statistically significantly above the average.  Derived from tables 3 and 4.

We also “mashed up” the complaints with other datasets based on the ZIP code from which the complaint was lodged. We compared the outcomes in the table above to demographic factors like race, income, senior population density, military employment, and college enrollment.  But there are important limits to what can be inferred.  We found, for example, that more responses are disputed in ZIP codes with a higher proportion of senior residents.  The result could be because the complaints of seniors are being ignored or it could be because seniors are more likely to complain in general.  Even with these caveats, we hope the results might help guide follow-up research and potentially even Bureau investigations.