Behavioral Law and Economics and Bank Overdraft Protection

In the past few months, at least two articles have come out that apply behavioral law and economics to the analysis of bank overdraft protection. One by Bubb and Pildes (forthcoming in the Harvard Law Review) and the other by Lauren Willis in The University of Chicago Law Review. Both articles make the same claim–that there are situations in which consumer behavioral biases are so pronounced and banks are so “nefarious” (a term actually used by Bubb and Pildes at one point) that it is no longer sufficient to simply have mere “nudges” or “sticky defaults,” instead it is necessary to have more aggressive interventions.

Both use the same example to purportedly demonstrate their point–federal regulation of bank overdraft protection. Here’s the story: a few years ago the Federal Reserve changed the rules governing how consumers are enrolled in bank overdraft protection programs for overdrafts of ATMs and point-of-sale debit card transactions. Previously, it was the practice of many banks to automatically enroll consumers in overdraft protection when they opened their account unless consumers opted-out. Regulation E changed it to an opt-in system, so consumers have to sign up for it. The Fed’s expressed rationale was that they were concerned about the propensity for some small percentage of consumers to use overdraft protection “excessively” (i.e., more than 10 times annually) and that moving to opt-in would protect these heavy overdrafters from themselves.

So what happened? First, while estimates vary, the percentage of people enrolled in bank overdraft programs fell (some banks simply eliminated the program because of regulatory risk). Second–and more important–the tendency of consumers to opt-in was precisely opposite of that predicted by the Fed–the tendency of consumers to opt-in was positively and linearly related to the number of overdrafts that they make. For example, the CFPB reports in its recent white paper on overdraft protection that while 15.2% of all consumers opted-in to overdraft protection after the Reg E amendments, that number ranged from 11.3% for accounts with no overdrafts or bounced checks to 44.7% of customers with 10 or more overdrafts. Other studies and surveys provide different point estimates but all find the same linear trend–the propensity to opt-in is positively correlated with the number of overdrafts. As I concluded in my article on “The Economics and Regulation of Bank Overdraft Protection” published early last year, “This real-world experience rebuts one of the proffered rationales offered by Federal Reserve—but one for which it offers no evidence or even serious theoretical support—that opt-in would paternalistically protect frequent users of overdraft protection from overusing the product. According to Federal Reserve, requiring opt-in would make it more difficult for these consumers to access overdraft protection, which ‘could therefore best prevent these consumers from entering into a harmful cycle of repeated overdrafts.’ But experience shows that heavier users of overdraft protection are those who are most likely to opt in to overdraft protection.”

Now consider Bubb and Pildes’s analysis of this experience (TZ: heavily edited to just provide the essence of the argument :

The case of automated overdraft services provides an instructive example of the failures of both the sticky default and total cost disclosure approaches of BLE. Historically, banks extended loans to cover overdrafts on an ad hoc basis and only for check transactions. ATM withdrawals and debit card transactions that would take the consumer’s balance below zero were simply declined. But in recent years banks have instituted automated overdraft protection services that both automatically enroll depositors in the service and automatically cover overdrafts caused not only by checks but also by debit card and ATM transactions. Banks typically charge a fixed fee per transaction—$27 was the median as of 2007—which typically amounts to an over 1,000 percent APR on the loan. Consumers can avoid these fees by linking a savings account or credit card to their deposit account to handle overdrafts—a service provided by most banks at much lower cost (typically $5 per transaction, sometimes free) than their automated overdraft service. But many consumers don’t take action to sign up for these alternative lower cost overdraft services. As a result, automated overdraft services have been major revenue generators for banks, amounting to some $2 billion in 2006 alone.

Banks’ automated overdraft services for debit card and ATM transactions are used to exploit consumer mistakes and arguably provide little social value…. Analysis of data on those who trigger overdraft fees reveals “nearly all could have avoided fees by using a much cheaper source of liquidity (usually a credit card with available credit).” Not surprisingly, many banks structured these programs in misleading ways. Banks sometimes operated “nonpromoted” services, meaning they automatically enroll customers in the service without informing them of the program and that they can opt out. More deviously, half of large banks processed all of a consumer’s overdrafts in a batch once each day and, rather than processing the transactions in the order they occurred, processed them in order of largest to smallest transaction. Because banks charge the fixed fee on each transaction after the one that puts the consumer below zero, this opportunistic reordering of transactions results in even more fees….

Notably, in adopting the rule the Federal Reserve gave no consideration to a mandatory product regulation alternative, such as requiring overdraft protection for ATM and debit card transactions to be provided using linked account products or capping overdraft fees. Professor Willis provides a damning account of the failure of this regulatory approach.220 Facing the new sticky default rule, banks used a variety of aggressive marketing tactics to induce many consumers to opt out of the default and into the bank’s automatic overdraft program. First, for new customers, many banks effectively undid the new legal default (nonenrollment) by forcing consumers to make an affirmative choice whether to enroll in the overdraft service or not. For existing customers, particularly those with a history of frequent overdrafts, banks aggressively marketed the benefits of opting out and enrolling (through phone calls and in-person in bank branches). And they were persistent. As Willis puts it, “Consumers realized that there is an immediate intangible benefit to opting out — the marketing stops. The calls and emails will cease, the tellers will stop asking, and those who bank on-line will be able to navigate directly to their personal account without clicking through a pop-up screen asking whether they would like to opt out first.”

By the end of 2010, 45% of those who overdraft more than 10 times during the first half of 2010 had, in one survey, opted out of the new nonenrollment default, compared to only a 15% opt-out rate for all customers.

There is only one problem with this analysis–it is all made  up.

Consider the core assertion of Bubb and Pildes’s argument: “Consumers can avoid these fees by linking a savings account or credit card to their deposit account to handle overdrafts—a service provided by most banks at much lower cost (typically $5 per transaction, sometimes free) than their automated overdraft service. But many consumers don’t take action to sign up for these alternative lower cost overdraft services.”

So what’s the real reason why heavy overdrafters don’t sign up for linked credit card, line of credit, or savings accounts? Because they can’t. The reason why people use overdraft protection is because they don’t have credit cards and cannot qualify for a line of credit. As I stated in my article:

A survey conducted by the Raddon Financial Group of customers of a large regional bank asked customers who used overdraft services where they would turn for emergency funds if they no longer had access to overdraft protection. 53% of “elevated users” of overdraft protection reported that if overdraft protection was not available they would “not be able to get money,” as opposed to only 16% of non-users. While 26% of non-users of overdraft protection said that they would “use a credit card” if overdraft protection were unavailable, only 10% of elevated users said they would use a credit card. This presumably reflects their lack of access to credit cards or that use of a credit card would cause them to exceed their credit lines  leading to penalties. Similarly, while only 6% of non-users said that they would seek a payday loan if overdraft protection was unavailable, 24% of elevated users reported that would be their option (the second-highest response after “[n]ot able to get money” for elevated users). Moreover, while 56% of non-users said in such situations they would simply transfer the needed money from another account, presumably a savings account, only 13% of elevated users said that they would do so, presumably reflecting the simple truth that they have no other accounts available.

Regular users of overdraft protection have low credit quality and limited credit alternatives. According to the Raddon survey, for example, only 7% of elevated users of overdraft protection describe their personal assessment of their credit rating as “excellent,” while 70% describe their credit rating as “fair” (38%) or “poor” (32%). By contrast, 74% of non-users of overdraft protection describe their credit rating as “excellent” or “good,” and only 9% consider their credit rating to be “poor.” Thus, reducing access to overdraft protection would simply exacerbate the plight of those who rely upon it because of the lack of better alternatives.

Another survey, conducted by Baselice & Associates, Inc., of one bank’s customers found similar results. According to that study, 54% of those who self-identified as having “poor credit” thought that overdraft protection was “extremely important,” compared to only 18% of those who said that they had “excellent credit.” When asked how upset they would be if overdraft protection was eliminated, 62% of those with poor credit said they’d be “extremely upset” compared to only 20% of those with excellent credit. In addition, while 41% of lower-income customers reported that they’d be “extremely upset,” 29% of customers with annual incomes over $60,000 also said that they would be extremely upset if overdraft protection were eliminated.

So according to this research, about 10% of elevated overdraft users say that they could use a credit card if they didn’t have overdraft protection and only 13% said that they even had a savings account that they could link (those who use alternative credit products typically have little savings).

As I also noted in my overdraft protection article, while a bank of line of credit also is a theoretical possibility as an alternative to overdraft protection heavy users of overdraft protection typically cannot qualify for a bank line of credit, which typically requires a much higher credit score than those who use overdraft protection regularly. Moreover, a bank line of credit typically requires being approved for minimum line of credit of $2,500, far above the $300-$500 for which overdraft protection lines of credit.

To be sure, this data is limited too–there is surprisingly little research out there on overdraft protection. But it is consistent with what we know about consumer usage of alternative credit products generally. Moreover, I am not aware of any contrary evidence (the one study that Bubb and Pildes cite is not actually apposite to the specific question that is relevant here, which is why do heavy users of overdraft use overdraft as opposed to less-expensive alternatives). So while it is possible that future research might find that heavy users of overdraft protection actually have access to credit cards, bank credit lines, and linked savings accounts (heck, I was sort of wondering why if they were going to just magically give credit cards to overdraft users, why didn’t they just go all the way and magically give them houses and home equity loans too, which would be even better), one cannot simply assume that is the case in the real world. Nor do I think that is particularly likely given what we know about how consumers actually use overdraft protection and other consumer credit products. But if your entire model depends on that fact being true, and your policy recommendation to restrict access to this product is premised on that basis, I would think you’d want to be certain that the foundation is correct.

What this also demonstrates is that for most frequent overdraft users, the most realistic real-world alternatives aren’t credit cards or lines of credit–it is payday lending. Or going without–and while we don’t have good numbers on what this means for overdraft protection, for payday lending one study found that 81% of payday lending customers said that they would go without necessities such as food and clothing if they couldn’t access payday lending. Considering the similarities of payday lending and bank overdraft users, this is probably representative of overdrafters as well.

So the reality is that those who use overdraft protection do so not because they are browbeat or hounded into it but because it is typically their best available option among the limited options that they confront. These are people with poor credit, limited access to credit cards, limited savings, and for which their actual realistic option is payday lending. Moreover, as I discuss, the evidence indicate that when it comes to the choice between payday lending and overdraft protection, these consumers actually choose rationally between those two products.  The tendency of heavy overdrafters to use overdraft protection and to opt-in is thus perfectly rational. Or to put it another way, it is not a matter of making default rules more “sticky” in order to bring about the Fed’s desired result–it is simply that the Fed’s rule is simply incorrect for heavy overdrafters.

Or, what I actually suspect might be the case (because the consumer credit economists at the Fed typically get this), the Fed (unlike our group of BLE authors) actually realizes that they don’t want to make the rule too sticky precisely because they want to make it relatively easy for heavy overdrafters to switch the default rule so that they can opt-in at relatively low cost. Thus, the whole point of the Fed’s rule is to make the default rule opt-in, but at the same time to also specifically make it relatively unsticky precisely so that consumers can readjust it. To put it another way, I suspect that the Fed has more confidence in the intelligence and local knowledge of those who use overdraft protection than do the law professors. The lack of stickiness is a fundamental part of the switch in the default rule. So the law professors simply cannot assume that in adopting the opt-in rule that the Fed actually meant to also make it sticky. In this sense, the opt-in and unsticky nature of the default rule is by design–with respect to heavy overdrafters it might actually be an efficient minoritarian default rule (although perhaps majoritarian for overdraft consumers as a whole, so the nature of the default rule could potentially make sense, although I think we need more data before we supplant the market rule by a regulatory-imposed new rule, but it is possible).

In this sense, the rule for heavy drafters may be similar to something like the national do-not-call registry, which for transaction cost reasons suggests that might be an efficient minoritarian default rule. The DNC registry illustrates the point that you cannot necessarily understand the default rule without also understanding how “sticky” it is. The FTC obviously knew that most people would want to opt-out, but instead of making it opt-out and making it difficult for people to opt-out (thereby solidifying a regime where most people would get bothered by telemarketing calls), it instead made it opt-out but easy to switch the rule. Similarly, the Fed could have been wanting to do that for overdraft protection–make it opt-in for ATM and POS, which could be the majoritarian rule for most consumers, but also making it somewhat easy heavy users of overdraft protection (who are likely hurt the most by losing access to overdraft protection) to opt-in. But that, of course, is based on the recognition that the Fed recognizes the value of overdraft protection for those who use it (I know, that’s a pretty elementary economics principle).

Fortunately the Fed didn’t make the rule too sticky–as the BLE theorists would have it–or those who desire overdraft protection the most would have difficulty getting access to it. I can see why the payday lenders would support that policy but not the rest of us.

The are other problems with the BLE analysis of overdraft protection. The “nefarious” practice of banks ordering payments from large to small, rather than vice-versa is, of course, more complicated than Bubb and Pildes imply (and, of course, they provide no analysis of why it is “nefarious,” it is just a conclusion). It may be that the FDIC’s limits on this are reasonable, but the issue is more complicated than just simply asserting that there is no justification for the practice (Michael Flores and I discuss this practice in our new comment on the CFPB’s overdraft protection white paper). Larger payments tend to be more important payments (such as rent, mortgage, and utilities) and the limited survey evidence we have indicates that a majority of consumers would rather have their most important (and largest) payments cleared first. I’m certainly not prepared to call that a definitive conclusion based on the limited information that exists–but one certainly cannot simply assert that the only reason for ordering checks from largest (most important) to smallest (least important) is only to “nefariously” jack up overdraft fees.

A final point pertains to the lack of institutional context for these BLE analyses. I am aware of only one study that tries to estimate the overall welfare effects of consumer access to overdraft protection, and that study estimates a consumer welfare surplus of about $50 per person ($2 billion overall in the economy) through avoided bounced-check fees and reduced precautionary balances. More generally, there is an intrinsic link between overdraft protection and access to free checking. I discuss it in my paper, but one analysis notes that in the year following the amendments to Reg E, the percentage of bank accounts eligible for free checking declined from 76% to 65%, meaning that new bank account fees (such as monthly fees and higher minimum balances) were being assessed to offset the reduction in consumer use of overdraft protection. Now there is an interesting economic discussion to be had as to whether consumers on net are better off with more access to free checking (combined with overdraft fees) or higher monthly bank account fees and related questions about consumer cross-subsidies. According to the data that was provided to me in writing my article, at one about 2/3 of low-balance free-checking bank customers never incur an overdraft fee and thus have entirely free checking. Imposing monthly maintenance fees on those customers (say $60-$180 a year) plus fees for other services would likely push many of them out of the banking system entirely. Once you combine the higher checking fees for these consumers, the costs to society of increasing the unbanked population (and their reliance on high-cost products such as check cashers, pawn shops, and auto title lenders), plus the harm to heavy overdrafters forced to turn to payday lending and other expensive alternatives, it is at least an open question as to whether consumers would be better off.

But a serious welfare analysis of reducing overdraft access has to take all of these factors into account. Heavy overdrafters use the product because they have limited options available to them. It is possible that taking away a preferred option (as demonstrated by their choices) from consumers with already limited options can theoretically make them better off. But it sure isn’t obvious.

A final word that I just can’t resist observing–what is up with the Harvard Law Review and the Chicago Law Review? I think that the fact that leading law reviews are rife with this type of BLE articles–hand-waving papers that treat alternative hypotheses as straw-men and ignore contrary evidence–is why law reviews are so lacking in credibility with the rest of the academy. Yet we seem to be trapped in a suboptimal cartel-type market where top law reviews use proxies for quality measures that they can’t observe directly. In my view, it really is a problem. At GMU we compensate by largely treating the “placement” of a law review article as essentially having no signaling value for quality, especially for any articles outside of Constitutional law, which strikes me as probably the optimal policy given the ill-suited nature of law reviews for processing these sorts of articles.