LIMITING CONSUMER CHOICE, EXPANDING COSTLY LITIGATION:
AN ANALYSIS OF THE CFPB ARBITRATION RULE

U.S. DEPARTMENT OF THE TREASURY
OCTOBER 23, 2017

 EXECUTIVE SUMMARY
Nearly a century ago, Congress made private agreements to resolve disputes through
arbitration “valid, irrevocable, and enforceable” under the Federal Arbitration Act. This
longstanding federal policy in favor of private dispute resolution serves the twin purposes of
economic efficiency and freedom of contract. In the Dodd-Frank Act, Congress authorized the
Consumer Financial Protection Bureau to limit or ban the use of arbitration agreements in
consumer financial contracts only if the Bureau concludes that its restrictions are “in the public
interest and for the protection of consumers.” Against this background, in July 2017, the Bureau
issued its final rule (the “Rule”) prohibiting consumers and providers of financial products and
services from agreeing to resolve future disputes through arbitration rather than class-action
litigation.
The Rule follows the Bureau’s study of arbitration, summarized in a 2015 report to
Congress. The Arbitration Study attempted an empirical analysis of both the arbitral awards and
class action settlements that consumers obtained for a variety of claims. But the data the Bureau
considered were limited in ways that raise serious questions about its conclusions and undermine
the foundation of the Rule itself. More fundamentally, the Bureau failed to meaningfully evaluate
whether prohibiting mandatory arbitration clauses in consumer financial contracts would serve
either consumer protection or the public interest—its two statutory mandates. Neither the Study
nor the Rule makes that requisite showing. Instead, on closer inspection, the Study and the Rule
demonstrate that:
•

The Rule will impose extraordinary costs—based on the Bureau’s own incomplete estimates.
The Bureau projects that the Rule will generate more than 3,000 additional class action lawsuits
over the next five years. Meanwhile, affected businesses will spend more than $500 million
in additional legal defense fees, $330 million in payments to plaintiffs’ lawyers, and $1.7
billion in additional settlements. Remarkably, the Bureau’s estimates do not account for
expected increases in state court litigation. Affected businesses are unlikely to simply absorb
these new financial burdens. The Office of the Comptroller of the Currency recently reported
that the Bureau’s own data show that the Rule’s costs will very likely be passed through to
consumers in the form of higher borrowing costs for credit card users, among other burdens.

•

The vast majority of consumer class actions deliver zero relief to the putative members of
the class. According to the Bureau’s own data, only 13% of consumer class action lawsuits
filed result in class-wide recovery—meaning that in 87% of cases, either no plaintiffs or only
named plaintiffs receive relief of any kind. The Bureau projects that, out of the 3,000
additional class actions the Rule will generate, four in five cases will yield no recovery for the
putative class of consumers.

•

In the fraction of class actions that generate class-wide relief, few affected consumers
demonstrate interest in recovery. On average, only 4% of plaintiffs entitled to claim class
settlement funds actually do so. This suggests that consumers value class action litigation far
less than the Bureau believes they should. This is not surprising given that plaintiffs who do
claim funds from class action settlements receive, on average, $32.35 per person.

1

 •

The Rule will effect a large wealth transfer to plaintiffs’ attorneys. On average, plaintiffside attorneys’ fees account for approximately 31% of the payments that plaintiffs receive from
class action settlements—and in many types of cases, much more. In an average case,
plaintiffs’ attorneys collect more than $1 million; actual plaintiffs receive $32 each. The
Bureau’s data indicate that the Rule will transfer an additional $330 million over five years
from affected businesses to the plaintiffs’ bar.

•

The Bureau failed reasonably to consider whether improved disclosures regarding
arbitration would serve consumer interests better than its regulatory ban. The Bureau’s own
data show that the financial marketplace offers choices to consumers regarding arbitration; the
vast majority of contracts in the major market segments do not contain mandatory arbitration
clauses. If the Bureau is concerned that consumers are unaware of arbitration clauses, more
prominent disclosure of such clauses would be a lower cost, choice-preserving means to
advance consumer protection.

•

The Bureau did not adequately assess the share of class actions that are without merit.
Courts and commentators have long recognized that defendants settle even meritless lawsuits.
As Justice Ruth Bader Ginsburg has explained, the class mechanism “places pressure on the
defendant to settle even unmeritorious claims.” 1 The Bureau overlooked the force of this
argument and failed to assess the costs of meritless litigation that the Rule will generate.

•

The Bureau offered no foundation for its assumption that the Rule will improve compliance
with federal consumer financial laws. The Bureau “assumes that the current level of
compliance in consumer finance markets is generally sub-optimal” 2 and insists that the Rule
will protect consumers by remedying that assumed compliance gap. But after years of study,
the Bureau has identified no evidence indicating that firms that do not use arbitration clauses
treat their customers better or have higher levels of compliance with the law. As a result, the
Bureau cannot credibly claim that the Rule would yield more efficient levels of compliance.

In view of these defects, it is clear that the Rule does not satisfy the statutory prerequisites
for banning the use of arbitration agreements under the Dodd-Frank Act. The Bureau has not made
a reasoned showing that increased consumer class action litigation will result in a net benefit to
consumers or to the public as a whole. Based on the Bureau’s own data, it is far more likely that
the Rule will generate massive economic costs—borne by businesses and consumers alike—that
dwarf the speculative benefits of the Bureau’s theorized increase in compliance.
BACKGROUND
The Bureau issued the Rule pursuant to Section 1028 of the Dodd-Frank Wall Street
Reform and Consumer Protection Act. 3 The statute required the Bureau to study the use of
1

Shady Grove Orthopedic Assocs., P.A. v. Allstate Ins. Co., 559 U.S. 393, 445 n.3 (2010) (Ginsburg, J., dissenting).
Final Rule (“Rule”), 12 C.F.R. part 1040, 82 Fed. Reg. 33210, 33392 (July 19, 2017).
3
See Dodd-Frank Wall Street Reform and Consumer Protection Act, Public L. No. 111-203, 124 Stat. 1376 (2010),
12 U.S.C. § 5518; Rule, 82 Fed. Reg. at 33210.
2

2

 arbitration agreements in contracts for consumer financial products and services and to report its
findings to Congress. The statute also authorized the Bureau to implement consumer financial
contract regulations, consistent with its report to Congress, that advance both the public interest
and the cause of consumer protection. 4
The Bureau completed its Arbitration Study—the basis for the Rule—in 2015. 5 Among
several other inquiries in the Study, the Bureau examined three principal sets of cases. First, it
assessed the types of claims consumers assert—and their typical outcomes—by surveying
individual and class claims filed in federal and certain state courts between 2010 and 2012 across
six financial product markets (credit cards; checking accounts/debit cards; small-dollar loans;
prepaid cards; private student loans; and for class actions only, auto loans). 6 Second, the Bureau
examined the terms of consumer financial class action settlements by reviewing 419 class action
settlements subject to final approval in federal court between 2008 and 2012. 7 Along with these
two sets of cases, the Bureau analyzed arbitral awards in American Arbitration Association matters
concluded between 2010 and 2012. 8
The Bureau based the Rule on two of the Study’s findings: first, that arbitration agreements
prevent consumers from resolving their claims on a class-wide basis instead of individually; and
second, that consumers rarely seek to resolve claims on an individual basis. 9 Based on these
findings, the Bureau decided to prohibit mandatory arbitration clauses in contracts for consumer
financial products and services entirely.
ANALYSIS
I.

The Rule Fails To Account For Major Costs And Inefficiencies Of Class Action
Litigation.

The Bureau’s analysis in the Arbitration Study and the Rule fails to account for significant
costs imposed by class action litigation.
First, the Bureau’s attempt to assess costs of additional litigation the Rule will produce is
incomplete. The Rule attempts to forecast costs resulting from additional federal class action
settlements, payouts, and associated legal fees that it will generate. 10 The analysis in the Rule’s
Table 1 extrapolates a five-year prediction based on past evidence of costs the Arbitration Study
observed. But many of the key figures in the Table are not explained anywhere in the Rule or the

4

12 U.S.C. § 5518 (authorizing the Bureau to regulate arbitration in a manner that the Bureau determines to be “in
the public interest and for the protection of consumers.”).
5
Consumer Financial Protection Bureau, Arbitration Study: Report to Congress, pursuant to Dodd-Frank Wall Street
Reform and Consumer Protection Act § 1028(a) (March 2015) (“Arbitration Study”), available at
http://files.consumerfinance.gov/f/201503_cfpb_arbitration-study-report-to-congress-2015.pdf.
6
Arbitration Study, Section 6 at 4-11.
7
Id., Section 8 at 2-5.
8
Id., Section 4 at 58-60.
9
Rule, 82 Fed. Reg. 33210 (“This final rule is based on the Bureau’s findings—which are consistent with the Study—
that pre-dispute arbitration agreements are being widely used to prevent consumers from seeking relief from legal
violations on a class basis, and that consumers rarely file individual lawsuits or arbitration cases to obtain such relief.”).
10
Rule, 82 Fed. Reg. 33405 Table 1.

3

 Study. And the narrative offers no explanation of many estimates at the subsector level that add
up to the bottom-line cost projections.
Despite these limitations, it is clear that the Bureau’s cost estimate entirely omits at least
two major categories of cases and therefore underestimates the Rule’s cost. The Bureau estimated
costs only for additional federal cases—entirely neglecting the costs imposed by additional state
court litigation. The Bureau does not dispute that the Rule will generate additional state litigation,
but it excuses this omission based on data limitations. 11 More puzzling still, the Bureau
inexplicably estimates that prospective attorneys’ fees will make up a considerably smaller share
of total payouts to consumers (approximately 19%) than the Bureau’s data show (31%). 12
Even given these substantial omissions, the Bureau predicts a massive increase in class
action litigation as a result of the Rule. 13 Over the next five years, the Bureau predicts an
incremental increase of more than 3,000 federal cases asserting claims on a class-wide basis. The
Bureau projects that only one in five of these cases will produce any relief at all on a class-wide
basis. At the same time, the Rule will transfer $330 million in payments to plaintiffs’ lawyers and
impose on providers nearly $575 million in additional defense costs and payments to individual
named plaintiffs—costs separate from any relief to a class of consumers.
Second, the Rule does not adequately address the proportion of class action plaintiffs’
nominal recoveries ultimately paid to class attorneys. Courts and commentators have long noted
the “acute conflict of interest between class counsel, whose pecuniary interest is in their fees, and
class members, whose pecuniary interest is in the award to the class.” 14 Some commentators have
argued that this conflict of interest leads plaintiffs’ lawyers to “agree to settlements that better
serve their own interests—and the defendants who should be their adversaries—than those of class
members.” 15
The Bureau’s own data illustrate the point. The Arbitration Study found that, in 251
settlements studied, the average payment to class members was $32.35. 16 To assess plaintiff-side
attorneys’ fees, the Bureau studied 419 cases in which attorneys earned a cumulative $424 million,
more than $1 million per case on average. 17 The Bureau found that attorneys’ fees amounted to
21% of gross cash relief and 16% of gross cash and in-kind relief available to plaintiffs. 18
Importantly, the Bureau itself acknowledges that the relief available to plaintiffs by way
of class action settlement is not the same as what the plaintiffs actually recover. The Bureau’s
data on settlement agreements show that in approximately 60% of class action settlements,
payment to class members is not automatic. 19 These “claims-made” settlement agreements state
11

Compare id. at 33401 with Arbitration Study, Section 8 at 36.
Rule, 82 Fed. Reg. 33405 Table 1; cf. Arbitration Study, Section 8 at 36.
13
Rule, 82 Fed. Reg. 33405 Table 1.
14
Pearson v. NBTY, Inc., 772 F. 3d 778, 787 (7th Cir. 2014).
15
Deborah R. Hensler, et al., Class Action Dilemmas: Pursuing Public Goals for Private Gain (2000) at 79, available
at www.rand.org/pubs/monograph_reports/MR969.html.
16
Rule, 82 Fed. Reg. 33234 n. 335; Arbitration Study, Section 8 at 27-28.
17
Arbitration Study, Section 8 at 33.
18
Id.
19
Id. at 28 n. 46.
12

4

 the nominal amount a defendant could be required to pay, but each plaintiff is required to make a
settlement claim to recover anything. According to the Bureau, the weighted average claims rate—
the average percentage of class members who actually make a claim to recover from a claimsmade settlement—is only 4%. 20 In a typical case, then, only a small percentage of plaintiffs
recover anything.
Even when the Study focuses on payments actually made to plaintiffs, it overstates the
value of these settlements to plaintiffs, relative to the value of the settlements to their attorneys.
The Study concludes that attorneys’ fees comprise 24% of cash payments in the settlements the
Bureau studied. 21 The Bureau reaches this figure by considering attorneys’ fees as a percentage
of payments to the class and attorneys’ fees—that is, total payout by defendants. In other words,
the Bureau measures the value of a settlement to the class not by actual payment to the class, but
by transfers from defendants, including to plaintiffs’ lawyers. In doing so, it does not assess the
efficiency of class action litigation as a mechanism for obtaining relief for consumers. Attorneys’
fees as a percentage of payments that class members actually receive are higher—approximately
31% on average based on the Bureau’s data. And as scholars have pointed out, that average
number skyrockets in different individual types of cases; it is not uncommon for fee awards to
exceed the aggregate amount recovered by plaintiffs. 22 In other words, class actions achieve
consumer relief less efficiently than the Bureau seems to claim.
The Bureau believes that a “focus on the ratio of attorney’s fees to consumer payments is
misplaced.” 23 As Director Cordray explained in response to an inquiry on this topic from Senator
Flake, “the Bureau did not use this data to determine whether a certain percentage of a settlement
would be a reasonable amount to award plaintiffs’ attorneys.” 24 It is true that financial transfers
from defendants—whether to plaintiffs or to their lawyers—do not register as pure costs of the
Rule as a matter of economic theory. But the percentage of defendants’ costs attributable to
plaintiffs’ lawyers—and the percentage of settlement value nominally available to plaintiffs but
actually paid to their lawyers—do speak to the efficiency of the class mechanism as a tool for
achieving value for plaintiffs. That is an analysis the Bureau should have undertaken—particularly
in comparison to other regulatory options.
Third, the Bureau did not adequately consider the costs imposed by class action lawsuits
that lack merit. A long line of case law and legal literature acknowledges that defendants
frequently settle class action suits for reasons unrelated to the merits of the class claims. 25 It is
20

Id. at 30.
Id. at 36.
22
See, e.g., Jason S. Johnston, High Cost, Little Compensation, No Harm to Deter: New Evidence on Class Actions
Under Federal Consumer Protection Statutes (University of Virginia School of Law, Law and Economics Research
Paper Series 2016-12) at 2, available at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2777618.
23
Rule, 82 Fed. Reg. 33402.
24
Jan. 30, 2017 Letter from Dir. Cordray to Sen. Jeff Flake, available at https://www.consumerfinancemonitor.com/
wp-content/uploads/sites/14/2017/02/2017-01-30-RC-to-Flake_arbitration.pdf.
25
See Randy J. Kozel and David Rosenberg, Solving the Nuisance-Value Settlement Problem: Mandatory Summary
Judgment, 90 Va. L. Rev. 1849, 1851 (2004) (“[H]owever legally untenable the claim or defense may be, a nuisancevalue strategy can be profitable when it costs less to initiate the claim or defense than it does to seek dismissal.”); see
also Barry F. McNeil and Beth L. Fanscal, Mass Torts and Class Actions: Presented to the 1996 Judicial Conference
of the Fifth Circuit, 167 F.R.D. 483, 490 (1996) (“For defendants, the risk of participating in a single trial, and facing
a once and for all verdict is ordinarily intolerable. . . . This is so even if an adverse judgment is seen as improbable.”);
21

5

 often economically rational for defendants to settle even unmeritorious claims for “nuisance value”
(which can be quite substantial) simply to avoid the time and significant costs associated with
defending class action litigation, such as fees paid to defense attorneys. Moreover, even a small
risk of an enormous class action judgment on meritless claims can prompt defendants to enter into,
as Judge Henry Friendly put it, “blackmail settlements.” 26
The Bureau received multiple comments expressing concern that the Rule failed to
consider the costs imposed by meritless class action litigation—including costs incurred by
defendants’ decisions to settle claims that lack merit—and the potential increase in these costs
following publication of the Rule. The Bureau’s own data bolster these concerns, as the Bureau
indicates that only 13% of class actions result in any form of class-wide recovery. 27 But rather
than engage in serious analysis of the meritless litigation that the Rule will spawn, the Bureau
effectively dismissed these comments.
While it acknowledged that providers may feel “some pressure to settle contested matters
of all kinds to avoid defense costs or the risk of a judgment,” 28 the Bureau insisted that “a
defendant’s assessment of the merits of the plaintiff’s claim—specifically, the plaintiff’s
likelihood of succeeding at trial—is a key factor influencing a defendant’s decision to settle.” 29
But of course a “key factor” may not be the decisive factor in many cases, and scholars of civil
litigation recognize that settling with “the proponent of the meritless claim … rather than incurring
the greater expense of litigating to have it dismissed may well be the [defendant’s] rational (and
expected) course of action.” 30 In such cases, where the cost of defense exceeds the amount of the
settlement, the merits of the plaintiff’s claim are not likely a factor at all. That is all the more true
in class action litigation. As Justice Ginsburg has explained, class certification “places pressure
on the defendant to settle even unmeritorious claims.” 31 In cases seeking statutory damages, the
“pressure to settle may be heightened because a class action poses the risk of massive liability
unmoored to actual injury.” 32
The Bureau also noted that the law allows courts to limit frivolous litigation by dismissing
complaints that fail to state a claim, granting summary judgment in cases presenting no material
factual disputes, and sanctioning attorneys who file claims with no evidentiary basis. 33 But this,
too, is an incomplete and unpersuasive response. Dispositive motions do not reliably measure the
likelihood of a class’s success at trial because they test only legal sufficiency—and in a manner
most protective of the non-moving party. Federal courts evaluate a motion to dismiss on the
In re Rhone-Poulenc Rorer, Inc., 51 F.3d 1293, 1299 (7th Cir. 1995) (Posner, J.) (noting that class certification creates
immense pressure on defendants to settle by forcing them “to stake their companies on the outcome of a single jury
trial, or be forced by fear of the risk of bankruptcy to settle even if they have no legal liability”).
26
Henry J. Friendly, Federal Jurisdiction: A General View (1973) at 120 (describing settlements induced by the
unlikely prospect of an enormous class action judgment as “blackmail settlements.”).
27
Arbitration Study, Section 6 at 7, 36. Notwithstanding its report in the Study, the Rule suggests that this number
may be a few points higher than the data in the Study suggest because of the limitations on its review period. See
Rule, 82 Fed. Reg. 33403.
28
Id. at 33290.
29
Id. at 33272.
30
Kozel, supra n. 25 at 1851.
31
Shady Grove Orthopedics Ass’n, 559 U.S. at 445 n.3 (Ginsburg, J., dissenting); see supra n. 25.
32
Shady Grove Orthopedics Ass’n, 559 U.S. at 445 n.3 (Ginsburg, J., dissenting).
33
Id. at 33272-73.

6

 assumption that the plaintiffs’ allegations are true and consider a defendant’s summary judgment
motion in the light most favorable to the plaintiff. 34 And there are significant defense costs
associated with obtaining dismissal or summary judgment, which often comes only after costly
class discovery is complete.
But even accepting the Bureau’s reliance on motions practice to weed out meritless claims,
the Bureau’s analysis is flawed. The Study and the Rule appear to understate the share of class
actions dismissed by courts—that is, the cases most likely to be meritless. The Bureau examined
a total of 562 cases—470 filed in federal court and 92 filed in state court. 35 Because they were
filed in state court, nearly a fifth of the cases in the Bureau’s sample were not subject to the
heightened federal pleading standards that subject complaints to closer judicial scrutiny and may
result in more frequent filings of early dispositive motions in federal court. What is more,
approximately two-thirds of the state court cases the Bureau examined were filed in Cook County,
Illinois and Los Angeles, California—widely regarded as exceptionally plaintiff-friendly venues
that attract a large number of class action filings. 36 In other words, the filing and resolution of
dispositive motions in these venues—and defendants’ decisions to settle in these venues—may
overstate the extent to which class actions survive motions to dismiss nationwide. The Bureau did
not acknowledge, much less correct for, these facets of its sample.
The Rule’s net social benefit cannot be adequately measured without accounting for the
share of additional meritless class actions it produces—something the Bureau failed to estimate.
Payouts for meritless class actions are significant because they represent transfers that would
reduce fairness—by imposing costs on firms unrelated to violations affecting consumers—rather
than enhance it. And even meritless actions dismissed without payouts by defendants impose
litigation costs unconnected to any consumer or social benefit. Evidence suggesting that some
share of consumer class actions lack merit raises the possibility that even providers in full
compliance with federal consumer financial laws may be sued and may agree to settlements in
order to avoid defense costs. In this environment, it is unclear how effectively the Rule would
increase provider incentives to invest in more compliance—because even full compliance does not
guarantee immunity from class actions.
A simple approach to analyzing the potential effects of the Rule and the rate of meritless
class action suits is to apply alternative hypothetical scenarios to the Bureau’s projected increase
in class actions. The Study’s data show that in 10% of the cases examined across all regulated
industries, at least one defendant was dismissed by dispositive motion. 37 Table 1 below shows
scenarios of alternative shares of meritless class action claims, ranging from zero to 50% (along
the header row) and costs to providers (not social costs). All figures in the table assume no passthrough of costs. These shares apply to the Bureau’s projected annual number of additional class
34

See, e.g., Ashcroft v. Iqbal, 556 U.S. 662, 679 (2009); Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 255 (1986).
Arbitration Study, Section 6 at 17.
36
See, e.g., “Supreme Court ruling could make it harder to file class-action lawsuits against companies,” Chicago
Tribune, June 22, 2017 (describing Cook County as a “litigation tourism” destination), available at
http://www.chicagotribune.com/business/ct-supreme-court-ruling-mass-actions-illinois-0625-biz-20170622story.html; Daimler AG v. Bauman, 134 S. Ct. 746, 761 (2014) (describing California’s willingness to establish
jurisdiction over non-Californian defendants as “exorbitant.”).
37
Arbitration Study, Section 6 at 7. While this ten percent share appears modest, the reported average may mask an
elevated share of dismissal in one or more markets.
35

7

 actions and settlements attributable to the Rule, shown in panel A. 38 If, consistent with the
Bureau’s data, 10% of the additional class actions due to the Rule are meritless, the Rule would
generate 60 meritless actions and ten meritless class settlements per year. The figures rise
proportionally with the percent meritless class action scenarios, peaking at 50% (column to the far
right), which is associated with 302 additional meritless class actions and 52 meritless settlements.
Panel B applies the scenarios to provider costs and the value of projected payouts to
consumers. The projected costs to providers include attorneys’ fees for both consumers and
defendants as well as payouts to consumers. Note that costs to providers are distinct from overall
social costs in a cost-benefit analysis; the former includes payouts to consumers but the latter
excludes them since they are economic transfers from a societal perspective. The additional costs
and legal fees paid by providers remain constant across all scenarios because they mirror the fixed
amounts projected by the Bureau. The fourth row in Panel B shows potential outcomes for
additional fairness-enhancing payouts to consumers due to the Rule. As the share of meritless
claims rises, the amount of fairness-enhancing payouts shrinks, as shown in the columns to the
right.
Panel C shows that the net benefits to society of the rule may increase or decrease
depending on the share of additional meritless class actions it produces. Note that benefits in this
Panel exclude payouts to consumers, which are transfer payments—not benefits to society. 39 The
Bureau does not estimate social benefits in the Rule, so we consider hypothetical benefits
valuations ranging from $50 to $500 million annually—in other words, scenarios where the Rule
achieved reduction in consumer harm worth these amounts. Even positing the Bureau’s
assumption that 10% of additional class action lawsuits will be meritless, the Rule would have to
reduce harm to consumers by $500 million per year to demonstrate any net benefit to society. The
Rule does not come close to making that showing.

38

See Rule, 82 Fed. Reg. 33405 Table 1.
See Office of Management and Budget Circular A-4, Regulatory Analysis (Sept. 17, 2003) at Sec. E. (“Benefit and
cost estimates should reflect real resource use. Transfer payments are monetary payments from one group to another
that do not affect total resources available to society.”).
39

8

 Table 1:
Hypothetical Scenarios of Meritless Class Actions and Costs to Financial Product Providers
(Annual figures based on CFPB projected filings in federal court)
% meritless class actions

0%

10%

30%

50%

Panel A: Number of cases
Additional class actions filed in federal court
604
Additional meritless class actions
0
Additional federal class settlements
103
Additional meritless class settlements
0

604
60
103
10

604
181
103
31

604
302
103
52

$447
$34
$105
$308

$447
$103
$105
$239

$447
$171
$105
$171

Panel B: Value in millions of dollars
Additional costs to providers
$447
Additional meritless payouts to consumers
$0
Additional legal fees paid by providers
$105
Additional fairness-enhancing payouts to consumers
$342

Panel C: Benefits to society (B) minus costs to providers (benefits exclude transfers to consumers as payouts)
If B=$500 million
$53
$3
$-97
$-197
If B=$100 million
$-347
$-357
$-377
$-397
If B=$75 million
$-372
$-380
$-395
$-410
If B=$50 million
$-397
$-402
$-412
$-422

The simple conclusion to draw from this analysis is that it would require an extraordinarily
high estimation of the Rule’s social benefits (apart from payments from providers to consumers,
which are transfers for purposes of economic theory) to demonstrate that the Rule has any positive
expected social value.
II.

The Rule Does Not Account For Important Benefits Of Arbitration.

Congress nearly a century ago codified a policy favoring private dispute resolution in
federal law when it passed the Federal Arbitration Act. The Supreme Court has made clear that
Congress passed the Act with the benefits of arbitration to consumers and the public interest well
in mind. 40 As the Court has repeatedly explained, the Act’s primary purpose is to “ensure that
private agreements to arbitrate are enforced according to their terms.” 41 Indeed, the Court has said
40

See, e.g., Kindred Nursing Centers Ltd. v. Clark, 137 S.Ct. 1421, 1425 (2017) (rejecting state rule requiring express
delegation of right to enter arbitration agreement in power of attorney because it “single[d] out arbitration agreements
for disfavored treatment” among contracts); AT&T Mobility LLC v. Concepcion, 563 U.S. 333, 345-46 (2011)
(rejecting state law making class-arbitration waivers unenforceable) (“[O]ur cases place it beyond dispute that the
[Act] was designed to promote arbitration. They have repeatedly described the Act as ‘embodying a national policy
favoring arbitration,’ and ‘a liberal federal policy favoring arbitration agreements, notwithstanding any state
substantive or procedural policies to the contrary.’”).
41
Stolt-Nielsen S.A. v. Animal Feeds Int’l Corp., 559 U.S. 662, 682 (2010) (internal quotation marks and citation
omitted).

9

 pointedly that “Congress, when enacting this law, had the needs of consumers, as well as others,
in mind” and that arbitration is “helpful to individuals, say, complaining about a product, who need
a less expensive alternative to litigation.” 42 In a stark departure from these longstanding policies,
the Rule would prohibit mandatory consumer arbitration agreements entirely in the covered market
segments. In doing so, the Bureau overlooks benefits of arbitration that the Act was designed to
guarantee to consumers and providers alike.
First, the Bureau examined no data on arbitration settlements—only arbitration awards.
The Arbitration Study thus compares arbitral awards to class action settlements and does not
account for relief consumers obtain by settling an arbitration matter. The difference is important.
The Bureau examined American Arbitration Association (AAA) case files for consumer
arbitrations filed from 2010 to 2012. According to the Bureau, only 341 (32.2%) of the 1,060
cases it examined resulted in an arbitrator’s resolution of claims on the merits. 43 By contrast, 608
(57.4%) of the 1,060 case files indicated that the parties settled or that the arbitration otherwise
ended “in a manner consistent with settlement.” 44 Thus by its own terms, the Arbitration Study
cannot account for the result in more than half of the arbitrations it examined—where a consumer
and a company struck a settlement that may have compensated the consumer in some way.
What is more, the Bureau did not adequately account for pre-arbitration benefits to
consumers in the form of market-based provider responses to consumer complaints. Scholars and
commenters have suggested that aggrieved consumers would be more likely to change providers
than to call a lawyer over a small-dollar dispute—and that financial institutions respond in kind by
waiving various fees upon complaint instead of losing business or defending in arbitration. 45 The
Bureau responds that informal dispute resolution of this kind gives providers the incentive to
correct issues only for consumers who complain directly, and then to prioritize the most valuable
complainers. 46 But once again, the Bureau has not shown that the existing level of compliance
with consumer financial laws is sub-optimal—much less that banning arbitration entirely would
obtain optimal compliance more efficiently than market-based provider responses.
Second, the Bureau may underestimate consumer benefits from the systemic availability
of fast, efficient dispute resolution. The Bureau’s evidence that the costs of the Rule would not be
passed through to consumers is limited. The Bureau analyzes pass-through using an econometric
estimate of the price effects of a temporary injunction that disallows pre-dispute arbitration clauses
for four large banks as part of the 2009 settlement agreement of an antitrust class action, Ross v
Bank of America. 47 It tests whether the banks raised the prices of their newly originated credit
card accounts after the injunction went into effect, relative to peer institutions, and finds no
statistically significant effect. The result is presented as evidence of little or no pass-through of
costs to consumers, although the discussion provides caveats signaling their estimate does not
prove there is no pass-through.
42

Allied-Bruce Terminix Cos. v. Dobson, 513 U.S. 265, 280 (1995).
Rule, 82 Fed. Reg. 33227.
44
Id.
45
See, e.g., Jason S. Johnston & Todd Zywicki, The Consumer Financial Protection Bureau’s Arbitration Study: A
Summary and Critique (George Mason University Law & Economics Research Paper Series 15-25) at 8, available at
https://www.law.gmu.edu/assets/files/publications/working_papers/LS1507.pdf.
46
Rule, 82 Fed. Reg. 33394.
47
See Arbitration Study, Section 10.
43

10

 Neither the Rule nor the Study provides detailed results of the econometric analysis. This
omission is unusual and it detracts from the analysis by precluding the ability of readers to assess
the details and reliability of the econometric estimates.
The econometric discussion that is present suggests the main result (i.e., no change in costs
of credit cards) might not be robust to alternative analyses. First, the Bureau’s analysis itself notes
that the standard errors are quite large, which means the econometric analysis had little power to
detect what could be an economically large effect on consumers. In other words, there could be
pass-through, and it may be quite large, but the methodology does not have enough precision for
the Bureau to say definitively one way or another with the data available. In addition, the Bureau
does not provide enough information for readers to judge whether the reported effects are large or
small in magnitude. This is especially problematic given that the Bureau used a constructed cost
metric (“total cost of credit”) instead of one that is easily interpreted or familiar (e.g., interest rate).
Second, the analysis purports to identify a causal effect due to the methodology used, yet it does
not make the usual effort to show that their data meet the underlying assumptions necessary for a
difference-in-difference methodology to show a causal effect. A certain degree of similarity
between treatment and control groups is a prerequisite for using the difference-in-difference
methodology, and the analysis fails to argue that these groups are sufficiently alike to be
comparable. Finally, this analysis is focused strictly on credit card pass-through, and it is not
obvious that the pass-through (or lack thereof) in one form of credit applies to the others.
The Office of the Comptroller of the Currency (“OCC”) conducted an independent review
of the Bureau’s data and econometric analysis and identified many of these very concerns. OCC
concluded that there is an 88% chance that the total cost of credit will increase as a result of the
Rule and a 56% chance that it will increase by three percentage points or more. This independent
review underscores Treasury’s concern: that the Bureau’s analysis does not convincingly rule out
higher credit costs for consumers.
III.

The Bureau Did Not Attempt Meaningful Cost-Benefit Analysis.

Given the Rule’s failure to consider the substantial costs of class action litigation and
substantial benefits of arbitration, it is no surprise that the Rule falls short of both the statutory
command to weigh regulatory costs and benefits and established best practices for administrative
rulemaking.
A. The Bureau Should Have Weighed The Rule’s Costs And Benefits.
Congress authorized the Bureau to regulate arbitration agreements in a manner that it
determines to be both “in the public interest and for the protection of consumers.” 48 Congress also
required the Bureau to consider “the potential benefits and costs to consumers and covered
persons,” 49 the statutory term for providers of consumer financial products and services, when it
issues rules under the Dodd-Frank Act. 50 The Bureau’s statutory cost-benefit analysis must
48

12 U.S.C. § 5518(b).
12 U.S.C. § 5512(b)(2)(A)(i).
50
12 U.S.C. § 5481(14) defines “federal consumer financial law” to include “any rule or order prescribed by the
Bureau under this title”—i.e., any rule the Bureau promulgates under Dodd-Frank.
49

11

 consider “the potential reduction of access by consumers to consumer financial products or
services” resulting from arbitration regulation and the impact of the Rule on financial product
providers. 51
Longstanding Office of Management and Budget guidance further informs the cost-benefit
analysis that agencies undertake when promulgating regulations. Sound agency rulemakings
should include at least “three basic elements: (1) a statement of the need for the proposed action,
(2) an examination of alternative approaches, and (3) an evaluation of the benefits and costs—
quantitative and qualitative—of the proposed action and the main alternatives[.]” 52
The agency shows that its regulation is necessary by explaining what market failure or
other significant public need the regulation addresses—what problem it is trying to solve. The
agency then evaluates potential solutions to its problem against a baseline—by considering “what
the world will be like if the proposed rule is not adopted.” 53 Agencies are directed to examine a
long list of regulatory alternatives, including different enforcement methods, different degrees of
stringency, and informational measures that could reduce the need for regulation. For its own rule
and for these alternatives, an agency must identify the benefits and costs of regulation, in monetary
terms where possible. The agency compares its options by measuring “incremental benefits and
costs of successively more stringent regulatory alternatives” to “identify the alternative that
maximizes net benefits.” 54
While independent agencies like the Bureau are currently not covered by certain Executive
Orders and guidance concerning regulatory cost-benefit analysis, they are nevertheless encouraged
to “give consideration to all of [those] provisions, consistent with their legal authority.” 55
Accordingly, independent agencies like the Federal Reserve Board have stated that they “conduct[]
[their] rulemaking activities in a manner that is generally consistent with the philosophy and
principles outlined in the Executive Orders.” 56 As for the Bureau, it purports to “carefully assess
the benefits and costs of the regulations [it is] considering for consumers and financial
institutions.” 57
B. The Bureau’s Methods Of Cost-Benefit Analysis Are Inadequate.
First, the Bureau makes no effort to show that additional regulation of arbitration clauses
is necessary to achieve an efficient level of provider compliance with federal consumer financial
laws. An agency typically makes this showing by defining a baseline—“what the world will be
like if the proposed rule is not adopted.” 58 In this case, the Bureau would have offered some

51

12 U.S.C. § 5512(b)(2)(A)(i)-(ii).
Office of Management and Budget Circular A-4, Regulatory Analysis, at Sec. A.
53
Id.
54
Id. at Sec. D.
55
Office of Management and Budget Guidance M-11-10, Improving Regulation and Regulatory Review (Feb. 2, 2011)
at 6.
56
“Response to a Congressional Request Regarding Economic Analysis Associated with Specified Rulemakings,”
Board of Governors of the Federal Reserve System, Office of Inspector General (June 2011) at 9, available at
https://oig.federalreserve.gov/reports/Congressional_Response_economic_analysis_2011web.pdf.
57
Consumer Financial Protection Bureau, “Rulemaking,” available at https://www.consumerfinance.gov/policycompliance/rulemaking/.
58
Office of Management and Budget Circular A-4, Regulatory Analysis, at Sec. A.
52

12

 assessment of the present state of the problem it sought to fix—the existing level of compliance
with consumer financial protection laws. 59
Instead, the Bureau offers its own ipse dixit: “[t]he Bureau believes, based on its experience
and expertise in overseeing consumer finance markets, that in general the current incentives to
comply [with federal consumer financial laws] are weaker than the economically efficient
levels.” 60 The Bureau also acknowledges that “data and methodologies available to the Bureau do
not allow for an economic analysis of the optimal level of compliance on a law-by-law or marketby-market basis.” 61 Therefore, the Bureau “assumes that the current level of compliance in
consumer finance markets is generally sub-optimal.” 62
While achieving an efficient level of regulation is the stated goal, the Bureau offers no
analytical support for its claim to move the covered financial industries closer to regulatory
efficiency. The Rule lacks any quantitative assessment of the current level of provider compliance
with the federal consumer financial laws; the efficient level of provider compliance; the expected
level of compliance under the Rule; or alternative regulatory mechanisms to achieve that level of
compliance. Developing estimates of the current, optimal, and Rule-induced levels of compliance
is a difficult task. But the Bureau’s lack of substantiation seriously undermines its case that the
Rule would induce a more efficient level of compliance—the lynchpin of its conclusion that the
Rule advances the public interest as required by the statute.
In addition to assuming that current compliance levels are sub-optimal, the Bureau provides
no coherent basis for concluding that the Rule would improve the assumed compliance gap. After
years of study, the Bureau has identified no data indicating that firms that do not use arbitration
clauses treat their customers better or have higher levels of compliance with the law. In other
words, the Bureau has not shown that the Rule will solve the assumed problem it set out to fix. As
a result, the Bureau cannot credibly claim that its Rule would produce net benefits through
increased compliance.
Second, to the extent a reader can discern a baseline for the Rule, the Bureau chose the
wrong one. The Bureau examined class action settlements entered between 2008 and 2012 and
arbitral awards issued between 2010 and 2012. It also researched possible relationships between
public enforcement actions and private class actions, again from 2008 to 2012. 63 None of these
sets of data consider the effect of the Bureau’s own public enforcement actions—which did not
begin until 2012—on the class action landscape.
In other words, the Bureau assessed the need for ongoing future regulation by examining
the state of the world before the Bureau began its enforcement function. The Bureau acknowledges
59

The Bureau is positioned well to know about the current level of compliance with consumer financial laws. For
example, through its consumer complaints database, the Bureau each week sends “thousands of consumers’
complaints about financial products and services to companies for response.”
See https://www.consumerfinance.gov/data-research/consumer-complaints/.
60
Rule, 82 Fed. Reg. 33260. The Bureau offered the same response to industry commenters who noted the lack of
empirical evidence available to suggest that the level of compliance with consumer financial laws is sub-optimal. Id.
at 33392.
61
Id.
62
Id. (emphasis added).
63
Arbitration Study, Section 9 at 9.

13

 that its analysis does not consider the effect of its enforcement actions and asserts that “it has
provided significant relief to consumers since 2012.” 64 In connection with the Bureau’s spring
2016 semi-annual report to Congress, Director Cordray testified that—in only the previous six
months—the Bureau’s enforcement division had conducted “supervisory actions [that] resulted in
financial institutions providing more than $95 million in redress to over 177,000 consumers.” 65
The Bureau excuses this omission on the theory that, if anything, the chosen baseline has
caused the Rule to overstate costs to providers. As the Rule explains: “To the extent that the
existence and work of the Bureau, including its supervisory activity and enforcement actions,
increased compliance since 2010 in the markets the final rule will affect, the estimates of costs to
providers and the benefits to consumers going forward will be overestimates.” 66 But this
explanation sheds no light on the underlying need for regulation or any change in the efficient
level of regulation occasioned by the Bureau’s ongoing enforcement actions. More fundamentally,
the Bureau misapprehends how efficiency works: If the Bureau’s ongoing regulatory and
enforcement activities have already approached or exceeded the efficient level of compliance that
the Rule purports to achieve, then the Rule’s net compliance benefits vanish and additional
compliance costs are misspent. The same is true if the Rule is not tailored to the particular
compliance gap that exists—an issue that the Bureau failed to address. But rather than undertake
a serious assessment of existing compliance costs and benefits, the Bureau assumes away that
hurdle and ignores the impact of its own self-described “reorder[ing]” of “the consumer financial
marketplace.” 67 This major analytical defect undermines the economic foundation of the Rule.
Third, the Bureau’s rejection of less costly alternatives to the Rule—including welldesigned disclosure rules—falls short of traditional standards for administrative rulemakings.
The Rule is premised in part on the Study’s conclusion that “consumers generally lack[]
awareness regarding the effects of arbitration agreements.” 68 According to the Study,
approximately 57% of consumers whose contracts contained mandatory arbitration clauses
nevertheless still thought they could participate in class action lawsuits. 69 But rather than explore
options to better inform consumers, the Bureau adopted an outright ban on mandatory arbitration
clauses.
Regulators should trust informed consumers to make choices, and the Bureau’s data
demonstrate that the financial marketplace offers consumers broad choices to avoid agreements
containing arbitration clauses. For example, according to the Study, 84.2% of contracts for credit
cards and 92.3% of contracts for checking accounts do not contain mandatory arbitration clauses. 70
64

Rule, 82 Fed. Reg. 33279.
See Written Testimony of CFPB Director Richard Cordray Before the House Committee on Financial Services
(Mar. 16, 2016), available at https://www.consumerfinance.gov/about-us/newsroom/written-testimony-of-cfpbdirector-richard-cordray-before-the-house-committee-on-financial-services-20160316/.
66
Rule, 82 Fed. Reg. 33400.
67
Richard Cordray, Prepared Remarks at the Cincinnati AFL-CIO Labor Day Picnic, Cincinnati, Ohio (Sept. 4, 2017),
available at https://www.consumerfinance.gov/about-us/newsroom/prepared-remarks-richard-cordray-cincinnati-aflcio-labor-day-picnic/.
68
Rule, 82 Fed. Reg. 33224.
69
Arbitration Study, Section 3 at 3.
70
Id., Section 2 at 8. According to the Bureau, these numbers represent approximately half the market share in both
of these market segments.
65

14

 The Bureau’s own data also demonstrate that consumers have market alternatives to litigation or
arbitration for at least some important matters involving their contracts for consumer financial
products.
If the Bureau believes that consumers undervalue class action litigation, it could have
pursued measures to inform consumers about arbitration or even required firms to provide more
prominent disclosures concerning arbitration clauses. 71 Commenters on the Rule suggested that
providing consumers with more information about arbitration might incline some either to make
more individual arbitration claims or to choose providers whose contracts do not contain
mandatory arbitration clauses. Others might conclude that they do not value access to class action
litigation. The Bureau dismissed these comments and any nexus between consumer information
and deterring violations of federal consumer financial laws. In the Rule, it responded only that it
“is not persuaded that the presence of education or promotional materials would, for dispute
resolution, materially alter the dynamics that result in so few individual arbitrations.” 72
Alternatively, the Bureau could have chosen to regulate first those markets where
arbitration clauses predominate to assess the Rule’s effectiveness before imposing it on additional
market sectors. The Rule considers other regulatory alternatives through qualitative discussions
but provides no quantitative analysis of any of the alternatives because the Bureau believes none
of them would address the core problem it perceives—insufficient incentives for providers to
comply with consumer protection laws. Essentially, this makes the Rule an “all-or-nothing”
proposition. The Rule provides no clear basis for policy makers to consider the relative costs and
benefits of alternative stringency levels or approaches to regulation. Therefore, the Rule offers no
quantitative guidance on the appropriateness of the current or proposed future level of provider
compliance from a perspective of broad economic efficiency. This failure to consider regulatory
alternatives—or incremental options—falls plainly short of well-established principles of effective
regulation. 73
Even if the Bureau had a sound basis for concluding that its preferred ban is the only viable
option to address the stated compliance problem, the Bureau still should have considered the
alternative of taking an incremental approach toward implementation. That is, the Rule could be
applied initially to a single financial product market such as small-dollar loans on a trial basis, and
then be evaluated after a period to assess its effects, including both intended and unintended
consequences. This approach could generate a rich set of data that would enable the Bureau to
refine or reconsider potential future regulation of the other markets.
71

See Executive Order 13563, Improving Regulation and Regulatory Review (Jan. 18, 2011) at Sec. 4 (“Where
relevant, feasible, and consistent with regulatory objectives, and to the extent permitted by law, each agency shall
identify and consider regulatory approaches that reduce burdens and maintain flexibility and freedom of choice for
the public. These approaches include . . . disclosure requirements . . . .”).
72
Rule, 82 Fed. Reg. 33393 (“The Bureau is in general concerned about consumer awareness of contract terms and
the ability of consumers to make informed choices about consumer financial products and services. However, the
Bureau does not at this time have a basis to believe that any such lack of transparency leads to harm for consumers.”).
73
See, e.g., Executive Order 13563, Improving Regulation and Regulatory Review (Jan. 18, 2011) at Sec. 4; Executive
Order 12866, Regulatory Planning and Review (Sept. 30, 1993) at Sec. 1(b)(3) (“Each agency shall identify and assess
available alternatives to direct regulation, including providing economic incentives to encourage the desired behavior,
such … providing information upon which choices can be made by the public.”).

15

 Finally, the Bureau’s efforts to assess the costs and benefits of the Rule fall well short of
agency efforts undertaken for similar rules. The Bureau reports that it “does not have data to
quantify the level of investment in compliance across the 50,000 firms affected by this rule” and
that it would not be feasible to do so. 74
But developing credible estimated costs could be feasible in this case. Given the large
volumes of information already collected and analyzed in the Study—not to mention the large
consumer complaint database it maintains—the Bureau may already possess much of the data
needed to generate aggregate costs and benefit estimates. Deriving the marginal costs of
incremental adjustments to the proposed rule could build upon the aggregate estimates.
As an example, in the 2016 beneficial ownership disclosure Final Rule, the Treasury
Department’s Financial Crimes Enforcement Network (FinCEN) generated detailed
implementation and broader cost estimates for financial institutions’ compliance. That rule
requires covered institutions to collect and record several pieces of basic personal information for
new openings of business bank accounts in order to reduce the likelihood and aid enforcement of
illicit financial activities by undisclosed owners and controllers of such accounts. The cost
estimates drew upon a series of intensive information collection efforts from financial institutions
and information technology providers, including surveys, conference calls, interviews, and
comments on prior notices of the proposed rule. The development of the cost estimate required
assumptions based on limited information, and the result was an imperfect but credible range of
cost estimates that enabled FinCEN to conduct a break-even analysis that showed the rule would
be highly likely to yield benefits that exceed the expected costs. 75
The breadth of institutions covered by the Arbitration Rule is greater than that covered by
the beneficial ownership disclosure rule, which may complicate the cost estimation problem.
Nonetheless, the cost problem could be addressed incrementally, constructing a bottom-up
estimate that sums various types of costs at the firm-level and aggregates to the sector-level,
omitting exempted firms. This approach could be applied one financial product market at a time.
Focusing on the costs for the large credit card and checking account markets in particular may
provide a rough basis for extrapolating the costs to smaller affected markets.
Once a credible estimate of the costs is developed, then a benefits estimate could be
generated. One approach to securing indicative prevalence data may be through comparing
historic consumer complaint (or other consumer action) rates among providers with and without
pre-dispute arbitration clauses in their contracts with consumers. Such an analysis would have to
control for a range of provider and market characteristics that may independently influence
consumer dispute rates, such as provider size and competitiveness of the local credit market. If
credible, even rough incremental benefits estimates would provide a more solid foundation for
determining whether the Rule is indeed economically efficiency-enhancing or not. Absent full
benefits estimates, the Rule should at least provide a break-even analysis that estimates the level

74

Rule, 82 Fed. Reg. 33392.
Financial Crimes Enforcement Network, “Customer Due Diligence Requirements for Financial Institutions,” 81
Fed. Reg. 29397 (May 11, 2016).
75

16

 of benefits that would justify the Rule’s estimated costs—and assesses the likelihood of achieving
those benefits.
CONCLUSION
The Bureau’s Rule would upend a century of federal policy favoring freedom of contract
to provide for low-cost dispute resolution. An agency implementing such a drastic shift in policy
should typically subject its rulemaking to the rigors of cost-benefit analysis and require
incremental efficiency justification for more stringent regulations. The Bureau’s analysis fell short
of these standards for agency rulemaking, as well as its own statutory command to determine that
the Rule serves the public and consumer interests. And the Rule fails to account for significant
costs of class action litigation and benefits of arbitration in a meaningful way. At bottom, the
Bureau’s Arbitration Study and Rule do not show that the Bureau’s prohibition on arbitration will
efficiently improve compliance with the federal consumer financial laws or serve public and
consumer interests as the Dodd-Frank Act commands.

17