In a "Public Interest Comment" dated Sept. 10, 2014, researchers affiliated with the Mercatus Center, a non-profit think tank associated with George Mason University, present an interesting response to the proposed publication of narrative content in a Consumer Complaint Database (CCD) hosted by the Consumer Financial Protection Bureau (CFPB). They're against it.

A little background: The CFPB is chartered with a mission to protect consumers by carrying out federal financial law. Consumers may submit complaints about financial institutions to the CFPB, which classifies each complaint, forwards them to the institution and then tracks its resolution. Early reports suggest that banks are responding a little faster and managing complaints a little better these days. It's hard to tell from the data.

The Comment offered by the Mercatus Center concerns a proposal to publish redacted narrative content from complaints submitted to the database maintained by the CFPB: in short, to allow the public to review what their fellow citizens took the time to write in their complaints about a financial product or service.

In its current form, citizens may only review high level statistical information about the content in the database. Spreadsheets present the number of complaints submitted to each company, the service in question, and the response each complaint received (Closed with Explanation, Closed with Monetary Relief, Closed without Relief, etc.). Significantly, the response and the status of a complaint are defined by the bank, not by the customer who lodges it.

When a citizen files a complaint, they may include thousands of words of narrative content that tells their side of the story. The banks have the same opportunity to offer detailed responses, but do not, as a rule, offer more than a rote response. In my experience, it appeared the legal teams from the institution provided editorial direction, if not the entire content of the response. The institution was only interested in defending their actions, and did not respond to the complaint in any substantive form or manner.

The Comment submitted by the Mercatus Center opens with a fair description of the proposal, which is a matter of public record, and then proceeds to suggest "that there is not a market failure that would justify the public database." Given the events of 2008, this is a breathtaking statement. It demands a look behind the curtain, just to see who is funding this think tank. In this case, the wizards bear the name Koch, along with many of their right wing, deregulation happy consorts who want us all to live "free, prosperous, and peaceful lives." Apparently, the Koch brothers managed to weather the financial collapse without too much trouble; some of their beneficiaries, however, may have forgotten that the rest of us were not so lucky.

The Comment then proceeds to argue that a "private complaint database provider" can more effectively satisfy a public demand for sharing bad experiences. They note that web sites such as Yelp and Amazon offer both positive and negative comments, and hence offer a more balanced view of products and services. The Comment also notes that Yelp employs software to analyze natural language constructs to determine the relevance and reliability of the submitted content. Now, while a machine may be able to count the number of postings an individual has created, and identify a match for any of George Carlin's seven dirty words, rest assured it cannot yet make the kind of editorial decisions the authors of the Comment are looking for. People have to do that, and the audience must exercise its own judgment when reading positive or negative content others have posted.

Another issue with the argument that a "private complaint database provider" can satisfy a public demand for sharing bad experiences is that no such providers exist. Amazon is a retailer, and only hosts reviews to help visitors to their site make a purchase decision. Yelp, the other example cited, does not make any money from providing a database of complaints: they support themselves by selling advertising space to local businesses which may or may not be reviewed on their site.

Reviewing content posted on the Yelp site, one may become suspicious of positive comments about a local business posted by individuals who live in other cities and have posted hundreds of reviews. It is hard to believe anyone would make that kind of literary effort for free. Currently, the CFPB database houses more than 90,000 complaints; none of the banks listed on the Yelp site have more than a dozen reviews (good and bad). This statistical variance suggests that the "private site argument" offered by the Mercartus Center is an association fallacy: while both sites may contain complaints, they are clearly not the same thing.

As for the claim that adding narrative content to the CFPB database will increase its cost, recent financial settlements the Federal government has entered into with some of the major financial institutions should offer ample funding for such an effort. As none of the money collected in fines for committing acts of "control fraud" will go directly to those most injured by those acts, the least we could do is take a small percentage of those multi-billion dollar fines and use it to fund the CFPB effort. There is a certain measure of justice, and an incipient moral lesson, to be found in using those fines to "permit the victim to be reasonably heard."

The Comment goes on to assert that "the database presents a skewed picture for consumers. Consumers see all complaints, regardless of their veracity, merit, or legal significance." To be accurate, the consumer does not currently see any of the complaint content, only some highly abstracted, statistical data about the complaints.

The Comment then notes that consumers are more likely to read the narrative content, if published, than the current data and this, in turn, would make it more likely that the consumer would "look at and rely on the database in making decisions" and so "rely on incomplete and potentially inaccurate information -- exactly the type of practice the Bureau seeks to stop financial firms from engaging in." This awkward non sequitur first dodges the fact that consumers are currently operating with incomplete and misleading information (i.e., banks do not, and never have offered details about customer issues), and then concludes that if financial firms are expected to stop operating with incomplete and misleading information, so should consumers. The authors do not, however, offer anything that could serve an alternative to the narrative content.

After the opening salvo, the presumptions come fast and thick in the Comment. To support an appeal to consequences, the researchers state that "devoid of context, every complaint will carry equal weight, regardless of the facts and circumstances and the complainant's motivation" and "because of their location in a government database, the complaints will carry an air of official gravity." The narrative content is, first of all, required to create any meaningful context for the complaint: i.e., the current statistical data is stripped of the context that the narrative content may or may not restore. Also, given the nature of prose, it is difficult to imagine an audience giving equal weight to such a variety of complaints: those that make a more believable case will have more influence than those that are poorly written, or do not contain details we recognize as necessary elements of an authentic charge.

The Comment also states that "consumers may find it difficult to distinguish valid complaints from expressions of temporary frustration, unwarranted grievances, anger over being denied credit, or intentional efforts to cause reputational damage to a financial services provider." The authors use this unflattering picture of the audience, i.e., the American public, to distract from the core issue: that the majority of complaints may be valid, and so offer a sound and valuable basis for seeking out a more trustworthy service provider.

To support their contention that the costs of posting the narrative content will outweigh the benefits, the Comment makes a number of assumptions: 1) that there is no way for a consumer to determine if a complaint involves a violation of federal law, 2) that it will not contain metrics a consumer may use to support inferences about the rate of incidence, and 3) consumers will rely on a database that may contain factually incorrect information as a "relevant input in their financial decision-making."

First, when a financial institution is convicted of violating a federal law, it makes headlines. While a consumer may not make the effort to review all of the details of a complaint, they will recognize if a violation described in a complaint has made the news and merits consideration.

Then, given the scale and import of key financial transactions in a consumer's life, the rate of incidence for a complaint is less important than the fact that a financial institution will allow it to happen in the first place. In some regard, it is similar to malpractice in the medical field: we don't stop asking for major medical procedures because some doctors fail in their mission, but we do hold them accountable when it does happen.

The notion that consumers will "rely" on the database is, at best, specious. While a consumer may review a handful of complaints before making a decision regarding their financial destiny, they are not likely to rely on the complaint database as the sole source of information used to drive a decision.

The authors list these "straw man" costs, but do not offer any description of benefit in their Comment; it is a disingenuous strategy that leads to an unstated preposition that consumers are best kept in the dark about what their peers have experienced.

Moving to the key point, the Comment describes how the expanded database will harm financial institutions. This argument begins with an assertion that financial institutions will be "unable to effectively counteract or provide context for complaints" and could, therefore, "suffer severe reputational harm."

The weight of 90,000 or more complaints could, indeed, have a measurable impact on the brand value of the affected financial institutions. Obviously, they could respond to the complaints in a manner that addresses the substance of the issue, and document the effort. We might refer to this as "doing their job" in colloquial terms. Of course, they also have multi-million dollar PR budgets to work with, and can get their side of the story out pretty easily.

The CFPB posits that the target of any complaint will have the opportunity to post a response and tell their side of the story. In the Comment, the authors point to related obstacles the banks may face: short timeframes for posting of complaints (i.e., the bank gets fifteen days to respond before a complaint is published); companies are not allowed to publish information that could identify the customer, and the effort to respond would impose substantial costs on the banks (i.e., they would have to hire people to respond to complaints in a public forum).

First, in the CFPB database, banks may set the status of a complaint to "In Progress" while they are crafting a response: anyone capable of navigating the database to access narrative content should understand that if a bank is in the process of drafting a response, they should not draw conclusions about that complaint (except, perhaps, that the bank is not capable of responding to a complaint in less than fifteen days).

The authors of the Comment assert that "complaints about financial products and services often turn on very specific details of a person's interactions with a financial institution." The authors cannot, however, offer an example. It is not that hard to mask an identity, and the alternative is to leave it to the aggrieved consumer to publish the detailed content themselves, an option the authors fail to address altogether. Given the profits reported by the top 25 financial institutions over the last few years, claims of a burden due to expense are hard to take seriously.

Another point raised by the authors of the Comment is that a complaint about a small financial institution "could have a substantially harmful effect on the safety and soundness of a financial institution." According to an analysis of complaints in the database performed by the U.S. PIRG in 2013, twenty five U.S. banks account for more than 90% of all complaints submitted to the CFPB. The vast majority concern Wells Fargo, Bank of America and JPMorgan Chase (2,000 or more for each), and seven more banks with 500 or more complaints account for the bulk of the remainder. Under the Dodd-Frank Act, the CFPB supervises depository institutions and credit unions with total assets of more than $10 billion, and their affiliates. Perhaps the authors of the Comment should clarify what they mean by "small": while complaints may be entered for any institution, they are not managed (and definitely should not be published) unless they concern an institution under the supervision of the CFPB.

The remainder of the Comment supports a "Nirvana fallacy" in which the proposed solution to the obfuscation of statistical data, i.e., the publication of narrative complaint content submitted to the CFPB, is rejected because it cannot be rendered in a perfect form:

  • Potential costs, including those associated with verification, removing content that may identify a complainant, responding to institutions that may contest the validity of a complaint, and expenses related to litigation risk, are cited as factors to consider. Again, given the scope and nature of the fines levied against major financial institutions in the last few years, cost should not be a factor.
  • The authors then argue that the publication of unverified complaints may undermine market fairness, impede market transparency, and interfere with the "the competitive forces that should shape markets." A far more compelling argument may be made that withholding the content of the complaints will provide known offenders with cover, cloud issues of accountability and allow the dishonest to keep market share they do not deserve. Publishing the narrative content offers the general public a means to expose the effects of control fraud on its victims and increase the pressure on prosecutors to convict those who engage in criminal activity.
  • The authors then attack the jurisdiction of the CFPB, asserting the agency does not have the legal provenance required to create or host the database. The argument insists that Congress must explicitly identify a database as a deliverable, instead of using ambiguous language that suggests an agency "make public - in aggregated or other appropriate formats - information it collects in monitoring the markets." Lawmakers and CFPB staff could reasonably expect the creation of a database as a means of organizing its findings, just at the statistics are not calculated on a slide rule or presented in hard copy only.
  • The Comment also makes the tenuous case that any narrative content posted in the database would gain legitimacy by virtue of the fact that the CFPB is hosting it and may be confused by a disclaimer that is in a small font and is difficult to find. This appeal to intellectual poverty presents the foundational premise that someone who is willing to access the database, and review content, will not have the intellectual faculty required to recognize that an outrageous claim may be false and will view the U.S. Government as a paternal force that will protect them from inaccurate information.
  • Another section asserts that "the expansion of the complaint database to include consumer narratives does not achieve the goals of open government." The authors use equivocation to argue that since the proposed expansion would not result in something that exactly matches what other agencies have created, and because it cannot achieve absolute accuracy, it cannot be consistent with the Open Government directives.
  • The final section describes how other databases cited by the CFPB in its proposal are not comparable to the Consumer Complaint Database. The database hosted by the Federal Trade Commission (FTC) presents complaints about FOIA requests, and the database hosted by the Consumer Product Safety Commission (CPSC) presents complaints about product safety. The authors overlook the fact that the FTC does publish complaints on a public web site (more may be accessed via a FOIA request). As for the CPSC database, they cite an exceptional case wherein a judge enjoined the agency from publishing a report that contained inaccurate information. This example offers a nice example of confirmation bias, in that the thousands of reports that did not contain inaccurate information are not mentioned. The authors also gloss over the fact that narrative content is found in both databases, and is currently released to the public.

In their conclusion, the authors of the Comment state that "although the information in the expanded database may be timely, it will not be uniformly correct, complete, representative, or understandable." Following this line of reasoning, no complaint lodged against a corporate entity in a free market should ever be published. By placing a burden of absolute proof on the consumer, honest communication from the citizenry may never be presented by the U.S. government. This veiled notion of placing a gag order on the CFPB is a novel solution to a thorny problem for the sponsors of the Mercatus Center, but is probably not legal. Not yet, at any rate.

The conclusion also reveals a bias, and an unfounded revision of history when it states that "the Bureau initiated its database without due consideration of the problem the Bureau was trying to solve or the costs and benefits of the database. Rather than expanding the database's potential to cause unintended harm, the Bureau should return to the drawing board." Given the number of logical fallacies presented in its Comment, perhaps the research team at the Mercatus Center is the party that should return to the drawing board: our ideal of democracy is not served very well by shotgun arguments that distort the truth to serve the cause of unbridled avarice.

Beyond the particulars of the argument presented in the Comment, we find the mechanics of a form of purchased influence in our legislative process. The Koch brothers may use their millions to fund the production of such content, which is clearly designed to prevent meaningful action to address wrongs, expose malfeasance, and restore order to an industry that clearly cannot manage itself without bankrupting the world economy.

As citizens, we must make the effort to review the output of organizations such as the Mercatus Center, and show how they abuse language and present faulty logic in order to advance an agenda that serves the few at the expense of the many. There may be a good argument for not publishing the narrative content submitted to the CFPB, but it is not found in their argument.

A closing note: H.L. Mencken remarked that 'injustice is relatively easy to bear; what stings is justice." The response to the CFPB proposal mounted by the Financial Services Roundtable, to cite another example, belies the truth of Mencken's insight - the pain is just a little too evident. Americans are not really lost in the supermarket of financial services; we know who has kept all the marbles.

The industry should probably accept the fact that they are much better off allowing the CFPB to manage the kind of information found in complaints than they would be allowing the free market to really take them down.


Steven Peterson
Sept. 15, 2014