Imagine a system where the regulation of financial products turns not on their ability to comply with a pre-set standard but is instead framed by empiricism.
The goal of regulation is to make sure that institutions comply with the law. That is a good thing. It is also
something that is unlikely to change. Rule-making complements the law. But one problem is that those methods can never keep pace with innovation. Private enterprise changes quickly. Regulation does not. The process produces an ongoing collaboration between industry and government. Industry develops internal compliance once it has clarity from regulators. Then regulators use the established guidances to interpret products and to justify their enforcement.
In most cases, the net effect means that any evolution is met by lots of skepticism. It can cost too much to change.
Alternatives based upon the review of empirical resuts did not exist a few years ago. But big data has now arrived. The time has come to look more closely at how big data could make a difference.
For a critique of backward-facing standards, look no further than to what what happened with subprime mortgage lending. Banks came up with all kinds of incredibly exotic products. Everyone knew that they were dangerous. Interest-only, pick-n-pay, stated income....you didn't have to be a genius to see that there were problems on the horizon. Some people say that regulators were too cozied up with the banks to see a problem. But there were plenty of research papers coming out of the Federal Reserve, state banking commissioners, and the FDIC arguing that trouble was on the horizon. But very little was done to stop the problem. In fact, the thing that stopped sub-prime lending (with the exception of efforts in North Carolina and in a few other states) was the
But some of the most important rules out there were unable to catch up. Many of our fair lending rules trigger enforcement when it can be found that borrowers of protected classes are being denied credit. But hardly anyone was denied credit. Many of the rules had an ability-to-repay standard, but rule-making didn't contemplate the possibility of rate resets and on one had much in the way of a bar over maximum DTIs. The Community Reinvestment Act required banks to meet the credit needs of certain borrowers in areas where they had branches, but it never imagined that credit needs would be satisfied by a) non-covered subsidiaries of covered banks b) with originated loans at high interest rates and poor underwriting criteria. Safety and soundness rules should have stopped sub-prime lending, but regulators didn't seem to be too bothered. Without that pressure, few institutions set aside enough in loan loss reserves.
Imagine if the regulators could have used more in the way of empirical evidence to stop the problem before it blew up the economy of the entire world.
Big data makes that possible in so many ways.
I've spent some time putting forward the idea that big data could go a long way toward enhancing consumer-facing disclosures for prepaid cards. Right now, there is no standard at all. But if rule-making on prepaid mimics the approach used for credit cards, then the result will be a list of fees. Once that approach is established, it will take years to change it. But if the past is any predictor, that set of fees will be out of date by the time it appears on the first j-hook. There is a new iteration in product functionality almost every month. Right now it looks like cash loading at the register is about to take off. Even the most well-funded and broadly conceived disclosure box on prepaid never contemplated listing such a fee on its box. Never mind that listing fees is itself somewhat compromised from the start, given that there is only a finite amount of space on retail packaging. Is it really possible to list 19 fees?
The alternative approach would be to monitor how much consumers are actually paying for their cards, how often they run into trouble, and how often entities fail to effectively implement their KYC or Patriot Act obligations. Couple that with cluster analysis that allows regulation to address the incredibly fragmented customer segment scenario and suddenly there is a system in place that keep up with consumer protections in real-time. Big data and cluster analysis could easily tackle many challenges. Imagine if in the future, the CFPB could use its big data to let people better understand how much they were likely to actually spend on a card. Here are some future scenarios:
- Evidence suggests that cardholders of XYZ card are frequently making the wrong choice when they select from the variety of cards offered by the company. Many consumers could be saving money if they had picked one of the alternative choices that better fit with their actual usage habits. Already, some card companies are implementing systems that help consumers to make the right choice. See the Rush Card's selection software here.
- Some studies say that fewer than half of all full-time prepaid card account holders are using free in-network ATMs. The result is that ATM costs constitute the largest cost-item in prepaid. But some cards offer both AllPoint and MoneyPass. What if the disclosures made it clear how much less cardholders at ABC card (where there are both free networks) were spending?
- NMO card has an opt-in overdraft function. It turns out that one cluster of account holders - college students - are overdrafting six to ten times per year. Absent a rule-making to require the issuer to establish a plan to intervene (as has been the case with deposit advance ANPRs), there is little that can be done to highlight the problem. Opt-in requirements are then changed to include a line that tells consumers how much people are spending on this service.
Ebay published a fairly detailed paper yesterday which outlines how evidence-based rule-making might be applied in the regulatory framework governing payments. The pdf can be downloaded through a link in this article.
Payments is an interesting place to apply this idea (and I hate to imagine what BitCoin will involve). It helps that rule-making on this topic is still fairly wide-open. I think there is a good chance that mobile payments will foreshadow the end of homegenous pricing. Retailers will be able to develop robust e-scores which describe things like lifetime customer value. In so doing, many will offer "deals" to the best spenders. The poor will pay more. Of course, there are no rules in place against offering discounts, but at some point in time these pricing tremors will become systematic inequalities.