Making Firms Liable for Consumers’ Mistaken Beliefs: Applications to the U.S. Mortgage and Credit Card Markets

• Published date: 01 December 2018
• Date: 01 December 2018
• Author: Alexei Alexandrov
• DOI: http://doi.org/10.1111/jels.12203

Journal of Empirical Legal Studies

Volume 15, Issue 4, 800–841, December 2018

Making Firms Liable for Consumers’

Mistaken Beliefs: Applications to the

U.S. Mortgage and Credit Card Markets

Alexei Alexandrov *

In theory, an introduction of a liability on firms, related to the difference between con-

sumers’ beliefs and the effective terms of purchase/contract, can improve both social

welfare and consumer surplus, depending on the relative magnitudes of: (1) decrease

in the gap between the beliefs and the effective terms of the contract due to the intro-

duction of the liability, (2) output decrease or price increase, and (3) efficiency of

administering the liability (and the amount transferred). I do not find statistically sig-

nificant evidence of (2) in two examples of instituting a similar liability: when several

large U.S. credit card issuers dropped mandatory arbitration clauses (that effectively

precluded class action lawsuits) and when U.S. residential mortgage creditors became

liable for failing to consider a borrower’s future ability to repay the mortgage, suggest-

ing that these events improved consumer surplus and might have improved social

welfare.

I. INTRODUCTION

In many markets consumers hold mistaken beliefs. Although in some markets consumers

are correct in their beliefs on average, this is not always the case. Sometimes, mechanisms

such as reputation, competition, or mandated disclosure, all discussed later, might

*1221 First Ave, Apt. 503, Seattle WA 98101; email: alexei01@gmail.com. The views expressed are those of the

author and do not necessarily represent those of his current or past employers. The author is grateful to Ron Bor-

zekowski, Kelvin Chen, and Will Wade-Gery for conversations about these topics. Portions of the article would

either not have been possible or would have taken years longer if not for the work done by Brian Bucks, Albert

Chang, Arland Kane, and Stefano Sciolli on projects using the same data. The author also thanks the participants

of CFPB’s Office of Research Brown Bag, of the International Industrial Organization Conference (especially dis-

cussant Wilko Bolt), of the Conference on Empirical Legal Studies (especially discussant Florencia Marotta -Wurg-

ler), and Oren Bar-Gill, Chris Drahozal, Andrew Rhodes, and Jean Sternlight for their detailed comments on an

earlier draft. He is also grateful to the editor and to an anonymous referee. Any remaining errors in the draft are

solely the author’s.

800

nonetheless result in a well-functioning market. However, in many cases, consumers’ mis-

taken beliefs lead to a market failure.

1

In many such markets firms can either alter the effective terms of contract to

better correspond to consumer beliefs or alter consumer beliefs to better corre-

spondtotheeffectivetermsofthecontract.First, using a theoretical model to place

the following empirical tests in the right context, I show that such an intervention,

under certain conditions, increases both social welfare and consumer surplus in

such markets, even in the ones where other mechanisms might not have succeeded.

Empirically, I do not find statistically significant price increase (in the U.S. credit

card market) or statistically significant quantity decrease (in the U.S. mortgage mar-

ket) due to such interventions (with all standard caveats). To test for the magnitude

or presence of positive effects, one would need data on deterrence that are unavail-

able, at least for these particular interventions.

In the theoretical model, introducing this liability results in (1) firms shrinking

the gap between the effective terms of the contract and consumers’ beliefs by shifting

at least one of these two variables.

2

However, this (2) increases firms’ cost and there-

fore lowers quantity produced, to the extent of the marginal cost increase adjusted by

the cost pass-through rate. Also, there are (3) the liability transfer itself and the

administrative costs of effecting this liability transfer, for example, court costs and law-

yer fees.

3

The overall impact on both social welfare and consumer surplus depends on

the interplay between these three effects.

Throughout the article I focus on two real-world examples of this type of a legal

liability: one connected to the U.S. credit card issuers’ exposure to class action law-

suits that started in late 2009 due to the issuers agreeing to drop their mandatory arbi-

tration clauses, and the other connected to the ability to repay (ATR) regulation in

1

DellaVigna and Malmendier (2004), Heidhues and Ko

¨szegi (2010), and Grubb (2015) all show that in particular

settings with particular behavioral biases, perfect competition does not maximize social welfare, and thus an inter-

vention might increase social welfare. Alexandrov (2017) shows that this is the case for virtually any behavioral bias,

and I adopt the model in that paper to allow for legal action. The model can also be viewed as addressing the lack

of salience of form terms (and other attributes). Policy interventions, like the ones that I analyze, arguably move

us toward the “reasonable expectations” doctrine discussed by Korobkin (2003). Such interventions are another

way to influence sellers’ ex-ante incentives, instead of (or, possibly, together with) Korobkin’s (2003) suggestion

that the courts should consider leaving sellers with the worst ex-post default options to ensure that sellers have the

correct ex-ante incentives. Note that, under some assumptions, the market can ameliorate or fix behavioral biases

completely; see, for example, Schwartz (2008). See also Schwartz (2015) for a discussion on regulation when con-

sumers might have behavioral biases.

2

This is, effectively, deterrence; see, for example, Becker (1968) and Kaplow and Shavell (2001).

3

As discussed further, effect (3) has different implications for consumer surplus and for social welfare. From the

point of view of social welfare, effect (3) in and of itself is weakly negative: this is a transfer, and at best it is per-

fectly efficient, but sometimes there will be administrative costs. From the point of view of consumer surplus, effect

(3) in and of itself is weakly positive: even if there are significant administrative costs, consumers still receive at

least a little of that monetary transfer.

Making Firms Liable for Consumers’ Mistaken Beliefs 801

the U.S. mortgage market that became effective in early 2014.

4

The available data only

allow me to attempt to estimate causally the magnitude of effect (2), increased prices

or decreased quality in these two markets.

5

The estimates are not statistically signifi-

cant: I cannot rule out that the introduction of such liability had no adverse impact

on prices or quantity. However, in the credit card application, the magnitude of the

standard errors is, arguably, economically significant, and thus I cannot rule out an

economically significant response.

For each of the two empirical analyses, I use administrative account/mortgage-level

datasets, accounting for close to 90 percent of each of the markets. For identification, I

use a difference-in-differences setup in the credit card market (four of the issuers have

dropped mandatory arbitration clauses, while others have not) and regression discontinu-

ity in the mortgage market (lenders making fewer than 500 loans were effectively

exempted from the regulation).

In the credit card application, finding not statistically significant effects, but not

being able to rule out economically significant responses, is disappointing from the statis-

tical power perspective. However, it serves as a warning to future researchers regarding

the difficulty of isolating effects (or noneffects) of contractual changes on prices, even

with excellent data. In particular, close to 50 percent of the U.S. credit card market

(by volume) dropped arbitration clauses, and I used a detailed account-level monthly

administrative dataset covering over 90 percent of U.S. accounts, and this was still not

enough to categorically rule out economically significant effects (or to rule out no

effects). One setting change that could help is having many more treated firms than four

(albeit very large) firms that I analyze.

Much of the behavioral literature’s suggestions for addressing the market failure

imposed by behavioral biases is ways of making consumers behave as if they are closer to ratio-

nal: nudges, choice architecture, or taxes on products like cigarettes, fatty foods, and sugary

beverages.

6

Another way of fixing the internality imposed by consumers’ behavioral biases is

making firms internalize consumers’ internality.

7

The liability that I analyze is an example of

alleviating the market failure by making the firms accountable for consumers’ biases. The idea

is connected to some of the more recent literature on disclosures. There is evidence that

4

See the following sections and New York Times (2015) for the first article in a three-piece long-form series on

mandatory arbitration clauses. Also see, for example, New York Times (2016) on the U.S. Consumer Financial Pro-

tection Bureau’s notice of proposed rulemaking that would not allowing mandatory arbitration clauses to be used

to block class action lawsuits in the consumer finance space in the United States. The ability to repay rule was the

key part of a package of rules that the U.S. Congress instructed the Consumer Financial Protection Bureau to pass

as a part of the Dodd-Frank Act, with the intention of preventing future mortgage crises; see New York Times

(2013) and Slate (2013) for examples of popular press coverage when the rule was finalized. After the rule was

finalized, Congress exercised its power under the Congressional Review Act and repealed the rule.

5

The causal claim is subject to interpreting the Ross case as exogenous variation, discussed in further detail below.

6

See, for example, Jolls et al. (1998), Camerer et al. (2003), Sunstein and Thaler (2003), Jolls and Sunstein

(2006), Johnson et al. (2012), Allcott et al. (2014), and Madrian (2014).

7

See, for example, Chetty (2015), discussing literature on externalities.

802 Alexandrov