Insights

How many times have we heard that the Consumer Financial Protection Bureau is a "data-driven" agency? The short answer is plenty. It appears to me that the Bureau is well organized to be a truly (and objectively) "data-driven" agency. The question is, "Is it?"

Interestingly, it's not as easy as you might think to find a definition for "data-driven." My very exhaustive (two-minute) Google search for a definition not specific to some particular discipline came up empty, except for one from that bastion of all things true and accurate, Wikipedia (which I think pretty much nailed the way most folks - including me - would define the term):

"Data driven means that progress in an activity is compelled by data, rather than by intuition or personal experience. It is often labeled as the business jargon for what scientists call evidence based decision making."

Sounds pretty good, right? Yes and no. In fact, it's pretty squishy. There are too many judgment calls to be made.

Justice Potter Stewart famously wrote "I know it when I see it" when trying to define pornography in 1964. Bill Clinton famously parsed himself into oblivion in 1998 when he said in grand jury testimony, "It depends on what the meaning of the word 'is' is," in response to a question about the veracity of a statement his attorney had made about Clinton's relationship with Monica Lewinsky.

I think the "data-driven" definition is less Stewart and more Clinton. Justice Stewart defaulted to the most subjective standard possible. Clinton, like it or not, had a valid point. Does "is" mean something happening right now, or does it mean (as the prosecutor seemed to imply) something that happened at any time - past, present, or future? He was asserting that the definition of "is" is purely objective and purely present tense.

I expect that most reasonable people would define "data-driven" by some objective standard. So, if the definition says "compelled" by data, what does that mean? Dictionary.com (another great research tool) defines "compelled" to mean, among other things, "to have a powerful and irresistible effect, influence, etc." Ergo, "data-driven" must/might/could mean that the data used to reach a conclusion are so powerful and irresistible that one outcome or rationale outweighs all others.

How one interprets data is wholly reliant on how well the interpreter has eliminated inappropriate biases in the analysis. This is, in itself, a subjective analysis - which is why every disparate impact case is a battle of the statistical experts, each advocating for his/her interpretation of the data and why, in my view, the cases themselves are built on flawed premises. It is far too easy to mistake a "correlation" for a "cause," and any decent statistician can manipulate the data to support either outcome. So, why does it go on?

Industry has objected to the Bureau's interpretation of fair lending data from day one, arguing that it doesn't use appropriate controls in its proxy methodology, i.e., how it figures out who is likely to be African American, Asian-Pacific Islander, Hispanic, etc. To provide industry with the means to test its fair lending data using the Bureau's methodology, the Bureau published a white paper describing its methodology last fall. Shortly after, Charles River Associates published a study that, using my shorthand, identified the flaws in that methodology and how the lack of appropriate controls seriously overestimates minorities.

When questioned about the CRA study's conclusions by a member of the House Financial Services Committee, Director Cordray replied that the Bureau did not agree with the study and that it is not obligated to respond. He also said that the Bureau is still looking at the conclusions, which begs the question, "How can the Bureau disagree with conclusions it is still analyzing?" Ah, politics.

Discrimination is a serious issue, as the Bureau and industry well know. However, when a government agency is basing serious and expensive claims on disputed methodology, and has in its possession a statistically sound and scientifically derived rebuttal of that methodology, I think it behooves the Bureau, a "data-driven" and transparent agency, to take it seriously and respond to it - whether it's obligated to or not, and before it next imposes massive restitution and penalties for questionable harm to hard-to-identify victims.

I've always said that I think the concept of the CFPB is a good idea. However, my concern is that the Bureau somehow seems to think that it would diminish itself if it were to admit to taking a wrong turn or change tack when the data demand. To me, that is simple intellectual honesty and institutional awareness that enhances its mission, particularly at this time when it is under so much political attack. With so many unscrupulous actors doing real and identifiable harm to consumers, focusing its energy on all that low-hanging fruit might do a lot for the Bureau's political image.

Michael A. Benoit is a partner in the Washington, D.C., office of Hudson Cook, LLP. Michael can be reached at 202-327-9705 or by email at mbenoit@hudco.com.