Washington, DC – Behavioral Economics Symposium (9/19/19) — consumerfinance.gov

Good morning to everyone here in the room
and to those watching via livestream. My name is Andrew Duke. I serve as the Bureau’s Policy Associate Director
for External Affairs and for Consumer Education and Engagement. Welcome to the Consumer Financial Protection
Bureau’s symposium on behavioral economics, which is being held here at the Bureau’s headquarters
in Washington, D.C. This symposium is the second in a series aimed
at stimulating a robust dialogue with experts reflecting diverse viewpoints across a range
of subject matter areas. In a moment I will have the honor of introducing
the Bureau’s Director, Kathy Kraninger. Before I do so, let me tell you about what
you can expect at today’s symposium. Following the Director’s opening remarks,
the Bureau’s Section Chief for Decision-Making and Behavioral Studies in the Office of Research,
Melissa Knoll, will moderate a panel discussion entitled “Methodological Foundations of Behavioral
Economics.” The panel discussion will include brief statements
from each of our distinguished panelists, to be followed by a moderated discussion. This first panel is scheduled to end at about
10:35 a.m., after which there will be a 15-minute break. Following the break, the second panel, entitled
“Behavioral Law and Economics and Consumer Financial Protection” will begin. It will be moderated by Jason Brown, the Bureau’s
Assistant Director of the Office of Research. This panel will also include brief statements
from each panelist as well as a 55-minute moderated discussion. The symposium will conclude at about 12:30
p.m., with brief remarks from the Bureau’s Deputy Director, Brian Johnson. For those unable to attend in person or watch
via livestream today, a recording will be made available on the Bureau’s website. Finally, as a friendly reminder, the views
of our panelists today are their views. They are greatly appreciated and welcomed,
yet they do not necessarily represent the views of the Bureau. It is now my honor to introduce Director Kathy
Kraninger. Director Kraninger became the second confirmed
Director of the Consumer Financial Protection Bureau in December 2018. From her early days as a Peace Corps volunteer
to her role establishing the Department of Homeland Security to her policy work at the
Office of Management and Budget, to the CFPB, Director Kraninger has dedicated her career
to public service. It is my privilege to welcome her to the podium. Director Kraninger, the floor is yours. Thank you, Andrew. Good morning, everyone. I am excited to welcome you to today’s symposium
on behavioral economics and consumer financial services policy. I would like to take a brief moment to thank
all of our panelists for participating today. They will be further introduced at the start
of each panel, so I won’t go into great detail, except to say that they are truly experts
and they are going to bring some fantastic perspectives to this conversation. I would also like to thank our moderators,
Melissa Knoll and Jason Brown, and the members of the working group for all of their work
developing today’s program. Today is part of a broader symposia series,
as Andrew mentioned, that seeks to gather experts in a variety of different fields to
tackle legal and policy issues facing the Bureau. We have a number of topics we plan to explore. A few of the topics, like abusive acts or
practices, at the symposium we recently held, and the symposium this later this year on
Section 1071 of the Dodd-Frank Act, deal with statutory provisions enacted by Congress and
focus on the various issues the Bureau must deal with when interpreting statutes. Some of the other topics, though, touch on
the broader issue of how the Bureau approaches consumer protection policy, in general. Next year’s symposium on cost benefit analysis
falls in that category, and so does today’s, with a focus on behavioral economics. Before turning to our first panel, I want
to offer a few observations on how the Bureau crafts its consumer protection policies. Most fundamentally, the Bureau is guided by
the objectives Congress set forth in our statute, ensuring that all consumers have access to
markets for consumer financial products and services that are fair, transparent, and competitive. So in its rulemaking, the Bureau seeks, consistent
with its legal authorities, to articular clear rules of the road for regulated entities that
promote competition, increase transparency, and preserve fair markets for consumer financial
products and services. First and foremost, the Bureau promulgates
regulations to implement statutes, and we, of course, promulgate the rules Congress directs
us to issue. In other circumstances, though, the Bureau
has the discretion to create rules in a new space. If we have this sort of discretion, the Bureau
should focus on regulating to prevent the harm to consumers. The Bureau must approach this work extremely
judiciously, however, cognizant of unintended consequences. To effectively achieve intended policy outcomes,
agencies should be able to articulate good reasons for what they do, and those reasons
should rest on solid evidence. This includes whether the benefits of the
proposed action justify the cost. Indeed, to formulate good policy, a substantive
analysis and estimation of costs and benefits, both direct and indirect, must be conducted. This transparent process informs the public
of the underlying reasons for both proposed and final regulations, and it can both help
identify practices that warrant regulatory action and lessen the chances that the Bureau
prohibits or restricts business practices that, in fact, benefit consumers or competition. It cannot be stressed enough that to develop
sound policy in these cases we need a demonstration and not just an assertion of a market failure,
and we need to offer a remedy carefully tailored to address that failure. Markets are often imperfect. It is often said, wisely, that we should not
let the search for the perfect be the enemy of the good, but the Bureau should only address
market imperfections if it is clear that intervention would improve the status quo. It is also generally well understood, and
there is overwhelming evidence to demonstrate it, that markets allocate resources more efficiently
than government agencies. That is why the presumption should be in favor
of the market and the onus should be on the government to demonstrate it can improve that
status quo. For consumer protection policy, specifically,
the Bureau generally seeks to empower consumers to make the best decisions for themselves,
and in developing such policies we start from the presumption that with timely and understandable
information consumers will be empowered to make decisions in their best interests. Today’s symposium explores whether behavioral
economics can offer insights on practices in the markets we regulate. In recent years, the field of behavioral economics
has offered criticisms of standard economic models, and the field of behavioral law and
economics has attempted to apply behavioral economics to legal and policy-making issues. This symposium provides an opportunity for
academics and policymakers to discuss the use of behavioral economics and psychology
in consumer protection. It will explore the merits of, and drawbacks
to, the application of behavioral economics within the regulatory environment. I am looking forward to today’s symposium
and the opportunity to hear from experts on these issues. Broadly speaking, my hope is that this symposia
series will serve as a launching pad for further economic and empirical research in the consumer
protection space. We have a great need to understand how consumers
evaluate the information they receive about consumer financial services. We must understand what makes consumers tick
so that we can use our supervision, enforcement, regulation, and education tools to be the
best consumer protection agency we can be. Fortunately, we have great panelists here
today to share their research, experience, and thoughts, to help us understand consumers
and consumer protection policy even better. I look forward to an enlightening and spirited
exchange of view. Thank you very much. Thank you, Director. At this time I would like to invite Melissa
Knoll and the panelists to please take their seats up here on the panel. And, Melissa, the floor is yours. Great. Thank you. Good morning, everyone, and thank you for
joining us today. We are pleased to get started with our discussion
related to the methodological foundations of behavioral economics. Fortunately, as the director said, we have
four very accomplished experts on our panel to help us work through some important issues
in this area. They come to us with a variety of perspectives
and extensive experience in topics related to these issues. With us today are Dr. Michael Baye, Bert Elwert
Professor of Business at Indiana University’s Kelley School of Business; Dr. David Gal,
Professor of Marketing from the University of Illinois at Chicago; Dr. John Lynch, Senior
Associate Dean for Faculty and Research at the University of Colorado Leeds School of
Business; and Dr. Brigitte Madrian, Dean and Marriott Distinguished Professor, Brigham
Young University Marriott School of Business. As a supplement to today’s discussion, our
panelists have also submitted written statements outlining some of their views on behavioral
economics, and behavioral science, more generally, and its relevance to policy-making. I would encourage all of you to take a look
at these statements after today’s panel. They were quite interesting to read, and,
in many cases, provide an even deeper dive into the issues that we will discuss today. This symposium is intended to generate discussion
about behavioral economics, a topic potentially relevant to many aspects of the Bureau’s work. I will be posing questions to the panelists
that the Bureau developed in collaboration with them. As a friendly reminder, again, the views of
our panelists today are their views. They are greatly appreciated and welcomed. Yet the questions and the panelist statements
do not necessarily reflect the views of the Bureau. So to kick off the panel this morning I am
going to ask each of the panelists to speak for about 5 minutes to provide some opening
thoughts that will help frame this morning’s discussion, and then we will follow with a
series of questions that we hope will generate some interesting dialogue. And so we would like to kick it off with Brigitte
Madrian giving her opening statement. So I’m going to start with what I hope is
an obvious assertion, but I think it sets the stage for at least how I think of the
role of behavioral economics and public policy-making, and that is the purpose of all public policy
intervention is, in fact, to change behavior by either individuals or organizations. Public policy is designed to either encourage
more of social welfare-enhancing behaviors, like getting more vaccinations, or less of
behaviors that reduce social welfare, like pollution or financial fraud. Maybe that would be more relevant to the Consumer
Financial Protection Bureau. I’m going to make another assertion, and that
is that the purpose of models—and as a trained economist I’m thinking of economic models
but I think this is true of models in the hard sciences as well—the purpose of models
is to provide a simplified view of a problem in a way that elucidates important mechanisms
that explain the problem at hand, and models are an abstraction from reality. I’ll come back to that in a minute. As Kathy mentioned in her opening remarks,
traditional economic analysis gives us a taxonomy of market failures that motivate policy intervention. The traditional set of market failures are
externalities, public goods, market power, or information failures. An additional motivation for policy intervention
would be to achieve different distributional aims that you get from a market left to its
own devices. And from these traditional market failures
we get a set of traditional policy tools that have been used for a long time. These include attempts to change market prices
through taxes or subsidies, information provision or disclosure to combat information asymmetries,
regulation of how firms and individuals can behave in markets, or direct public provision
of services by the government. So what does behavioral economics bring to
this table? I’m going to argue that there are a few things
that behavioral economics brings to the table. First of all, it brings a recognition that
some of the abstractions in traditional economic models ignored potentially important factors
for certain economically important outcomes, in certain circumstances. To give one example, traditional economic
models do a pretty good job of predicting the types of choices that consumers make when
they are making tradeoffs over time between things in the distant future and things in
the far distant future. They do a really poor job at predicting the
choices that individuals make for outcomes that involve something in the immediate present
versus something in the future. So traditional economics, the abstraction
away from present bias—I’ll get to that in a minute—does matter so much for a certain
set of choices. It matters a lot for a whole other set of
choices. So behavioral economics forces us to recognize
that some of the abstractions that have made historically don’t hold water for some of
the outcomes we are trying to analyze. Two, behavioral economics gives us additional
motives for public policy intervention beyond the traditional taxonomy of market failures. Congdon et al. outlined three broad categories
of additional motives for policy intervention, number one, that individuals are imperfect
optimizers. They are inattentive, they have biased reasoning. Individuals do a really lousy job in dealing
with probabilistic outcomes, for example. Individuals have bounded self-control. They procrastinate. Sometimes they are subject to addiction. And third, individuals’ preferences, to the
extent we can use that term, are very context-dependent. They don’t exist in a vacuum. So preferences depend on how the options are
framed. They depend on how individuals perceive themselves
relative to others. Things like fairness matters. Social norms matter. And this poses a real problem for thinking
about revealed preference as a tool for assessing what individuals actually want and what is
in their welfare. I should note that these biases don’t necessarily
imply a market failure. In some contexts, firms may have incentives
that are such that they want to help individuals overcome some of these behavioral problems,
so the market might actually correct these issues. But in many cases firms might have an incentive
to exacerbate or even exploit some of these biases. Finally, behavioral economics gives us additional
policy tools to help combat market failures. One that has received a lot of attention in
the literature, and one on which I’ve written, is the broad category of choice architecture,
which would include things like what is the default option, what are the number of choices
individuals have to make, are they simple or are they complicated, when do individuals
make choices, how often do they make them, how are they framed, are they given help to
help make choices? Those are all elements of choice architecture. So that would be one category of additional
policy tools, but there are many others. Behavioral economics also gives us ways to
increase the effectiveness and the cost-effectiveness of traditional policy tools, and I think this
is really important. Take, for example, the use of incentives,
a very traditional public policy tool. What we have learned from behavioral economics
is that incentives can be more or less effective at changing behavior, and in some circumstances
a probabilistic incentive, so think a lottery-like incentive, can be much more motivating than
a guaranteed incentive. There is plenty of research suggesting that
immediate incentives are much more effective than delayed incentives. That is important for thinking about, for
example, how the tax code might be an effective or an ineffective tool at changing behavior. People are more responsive to incentives that
are simple to understand than incentives that are complicated. That also speaks to the effectiveness of,
for example, the tax code, at changing behavior. And individuals respond differently to intrinsic
versus extrinsic incentives. And if you can intrinsically motivate people
to do something, that is often a much more cost-effective way to change behavior than
relying on extrinsic motivation in the form of actual monetary incentives. Another example would be disclosure. Disclosure is another very traditional public
policy tool, and what we have learned from a lot of behavioral economics research is
that there are many circumstances in which disclosure is ineffective, and there are,
indeed, some circumstances in which disclosure can be counter-productive. And behavioral economics gives us a framework
for thinking about when disclosure, or how to make disclosure more effective, rather
than just an assumption that if we provide information, individuals will consume it and
use it accurately. So, for example, disclosure about financing
terms is more effective if the disclosure actually does the calculations. Individuals do a poor job of thinking in terms
of percentages. They do a much better job if you give them
actual dollars to think about. If the disclosures are translated into objectives
that matter to individuals—so calorie information is more effective if you tell people how many
hours they are going to have to run to burn the calories off than if you just tell them
what the calories are. Individuals do a better job of assessing disclosures
if they are relative, if you are comparing A relative to B, rather than just telling
individuals information about A in the abstract. If you expand important outcomes, for example,
if the cost over a 10-year period you actually multiply by 10 rather than relying on individuals
to do the math. Disclosure is more effective if you can standardize
things across multiple outcomes, and more disclosure if you can make what you’re talking
about vivid. Okay. I think I’m done. Next we have David Gal. Go ahead. Yeah. What is behavioral economics? I think that’s an important question. Well, it arose from the behavioral decision-making
tradition in psychology that sought to identify deviations from normative standards of decision-making. And these fun, predictably irrational, systematic
anomalies documented in effects such as anchoring, the endowment effect, the compromise effect,
and many other effects, have captured the imagination of business people, policymakers,
and the public. In part, the attraction was, and is, that
through subtle changes in how information is presented these effects could be used to
induce large behavioral changes. Yet, through a bit of bait and switch, some
boring, quote/unquote, old information design, and other interventions, like even sending
text reminders, have been sold under the fun and exciting label of behavioral economics. Indeed, the term behavioral economics appears
to have expanded to include any intervention that make assumptions about psychology. In other words, any intervention intended
to influence behavior, which, in my view, renders the term essentially meaningless. Now, at the same time, from a policy perspective,
it is important to note that these original fun and interesting, irrational anomalies
that gave rise to behavioral economics don’t inherently reveal any suboptimal behavior
that needs to be corrected. So take the case of the endowment effect. One example of the endowment effect, the majority
of people given a mug kept it rather than trade it for a chocolate bar, whereas the
majority of people given the chocolate bar kept it rather than trade it for the mug. This apparent preference and consistency is
commonly viewed as an irrational mistake, typically explained via loss aversion. However, the mistake is not in the behavior
but in the assumption that preferences are stable and well defined. If, instead, we assume preferences are fuzzy,
then the effect can be trivially explained without any resort to error. Many people might have had no clear preference
between the mug and the chocolate bar, and so they stick with whatever they had initially,
simply out of inertia. Indeed, I think many of the systematic preference
and consistencies documented by behavioral economists and psychologists can more parsimoniously
be explained by fuzzy preferences than by invoking errors in decision-making. Moreover, when policymakers have sought to
take advantage of these effects to influence behavior, the impacts have generally been
small and unreliable. For instance, policymakers have sought to
nudge consumers towards lower electricity consumption by informing them when they use
more electricity than their neighbors. However, when tested, the effects of these
interventions have been small to negligible. Now does this mean that the psychological
insights frequently relied upon or derived by behavioral economists and others in related
fields, like psychologists are not of value? In my view not at all. It is that many of these ideas and insights
such as the fuzziness of preferences, that people often lack insight into their preferences,
that identity drives behavior, confirmation bias, they don’t lend themselves to obvious
simple answers or even to obvious questions. In fact, the focus on these interventions
that involve slight changes to how information is presented or choice represented, I think
they are important but I think they can also distract from the big picture. For example, I agree with John, your comments,
just a little bit, you made the comment that the more important mistake people made in
the financial crisis was choosing the wrong house, not choosing the wrong mortgage. And so, accordingly, even if we frame information
in a way that would have allowed people to pick a better mortgage, that’s a value, but
it would have had comparative little impacts on their financial well-being compared to
the house they chose, choosing the wrong house, meaning choosing one that put them in a precarious
financial position. So why did people pick the wrong house? Well, we can think about a mix of favors. For example, the lay belief, reinforced by
a lot of social proof that house prices always go up, trusted institutions that promoted
mortgages and home ownership, status seeking, low financial literacy, cultural factors,
low future time orientation, and so forth. So understanding these factors and considering
policy to address them requires stepping well outside the heuristics and biases tradition
of decision-making on which behavioral economics was founded. As another example, lottery winners and well-compensated
professional athletes routinely end up in financial distress or bankrupt. What can explain this? Again, it is unlikely that heuristics or biases
associated with how information is presented has anything to do with this phenomenon or
can do much to correct it, and by way of inference, why many other people can’t or don’t save. Instead, we might consider as factors low
financial literacy, status seeking, low future time orientation, which itself could be a
function of personality culture, family environment, structure of incentives in society, and related
factors. So both these examples illustrate the point
that psychological insights have potentially important policy implications, but these insights
are not well realized through simple changes in how information or choices are presented. Instead, successful interventions are likely
to take advantage of psychological insights, but to be much more multi-faceted complex,
having ended and sustained and to require a more careful evaluation of uncertainty and
tradeoffs than many of the ones advocated until now by behavioral economists. And to take a final example, confirmation
bias, which is the tendency of people to interpret new information to be in line with their existing
strongly held beliefs, suggests no easy fixes, but it does offer potential insights for policy. For example, it suggests the need for structures
that facilitate redundancy in policy evaluation and for creating paths for contrarian ideas
to receive a hearing. And I thank the CFPB for providing this forum
for such ideas. Thank you. And next up we have John Lynch. So by way of background, I am different from
many of the people in this room. I was an undergrad econ major but my PhD is
in psychology, and I have studied consumer behavior for my 40 years as a business professor
in a marketing department. And, like David, I am in a minority, because
most of the people are experts in economics and law. I have studied consumer memory, attention,
and learning. On the topic that both Brigitte and David
talked about, I’ve studied how people construct their preferences when they do not have established
preferences, making them highly sensitive to various context and framing effects. I also do work about how people think about
the future. And finally, relevant to this session, where
I think we are going to talk about methodological issues, I do work on external validity, which
is the issue of examining how general findings are coming from a given study. And if you read the papers associated with
this panel, my fellow panelist, Michael Baye, has a great point in his that you have to
target the right pathology to have an effective policy remedy. And so I would say that economics gives concepts
and findings to help us understand what pathology is underlying a set of symptoms, but it is
not alone in allowing such insights, and other behavioral sciences do so. So I consider myself a behavioral scientist. Probably apropos of David’s things, I don’t
consider myself a behavioral economist. Like I said, a behavioral scientist. Every discipline has its strong suit. In my view, the strong suit of economics is
that it has, first of all, a powerful toolkit, a single unifying theory, and alone among
behavioral sciences, it models the interactions of buyers and sellers and how they adapt to
each other. Other behavioral sciences, like mine in psychology,
tend to ignore the strategic adaptation of sellers to buyers in the markets. However, other behavioral sciences, in some
cases, have an advantage in descriptive accuracy about buyer behavior, and help us understand
how people actually behave and why they do so. So often that is in the form of giving us
process insights about why we are observing what we do and why certain kinds of seemingly
plausible interventions intended to help consumers do not work as intended, much as what Brigitte
was describing. I would also say—and this is agreeing with
one of David’s themes—that in all cases it’s wise to do the careful work to test a
proposed policy remedy and understand all the—and I think Janis is going to talk about
this in her remarks—to do the very careful mapping to understand exactly how this is
playing out before launching, on some broad scale, a policy action that will affect particular
segments differentially. I would say just three ideas and sets of findings
from my own field that I think are highly relevant to the work of the Bureau. The first is the idea of consumers’ consideration
sets, which is probably, I think, the most important idea in marketing, understanding
consumer choice. That is the set of alternatives actively considered. And anything that the Bureau can do to promote
transparency in the financial competitive marketplaces will involve trying to enlarge
and study the nature of consumers’ consideration set. Work in marketing shows that if you’re trying
to understand what actually gets chosen, most of the explainable uncertainty has to do with
whether an alternative was considered or not, not the issue of what the relative evaluation
is to Alternative A versus another, Alternative B, that is considered. And most information remedies by policymakers
tend to target the ladder effect, which, again, work in marketing shows a second order, explaining
a much smaller percent of the variants than the issue of what alternatives are considered. Second, I would say that information remedies,
such as financial education and product-specific disclosures often have limited effects, and
the reason for those small effects are originally surprising but they can be understood in terms
of psychological principles. And then my third point is one that the previous
speakers have touched upon. This is work that I’ve done some of the early
work on. One of the most important conclusions from
the last about 35 years of research in judgment decision-making is that people often lack
stable utility functions to guide their decisions. Not always, but if I haven’t had to make a
decision yet I don’t have it in my head what my preferences are. And so in those cases, it turns out, people
are highly sensitive to various context and framing effects that affect decisions they
make. However, if people have thought about it before,
and they have inputs to their decisions that are stable, acceptable, and they perceive
to be diagnostic and relevant, then they are largely insensitive to many of these behavioral
effects that we have discussed. And so why do you get that discrepancy? In the case where I have those stable preferences,
there is going to be a close match between my expected utility and what Kahneman has
called experienced utility, that is my ex post utility. But in cases where people lack those prior
preferences, that isn’t the case, and people may systematically mis-predict both the likelihood
of events but also how they would feel about events once they happen. And so, in many cases, that is noble in advance
to the policymaker or the seller, but not to the consumer. And so the implication of that is that the
more laissez-faire approaches to financial services regulation are more appropriate in
cases where people do have these established preferences but less so when people do not. And in this regard, one of the remarks in
Thaler and Sunstein’s book is when would these nudges be desirable—decisions that are hard,
infrequent, no opportunity to learn from feedback, no established preferences, and the markets
are not going to correct our mistakes. So my thought is that that is where the role
of the more psychological research can come into play. I would just say, in closing, that in May
I attended a conference at Wharton that was honoring this distinguished cognitive psychologist
and consumer decision researcher, Wes Hutchinson, and the whole thing was about whether consumers
are boundedly rational and where, in fact, they are more or less rational. And his conclusion, after serving this wide
array of research, was the following. He said, “We are very good at selecting important
information, learning from past experience, similarity-based reasoning. We are pretty good at verbal and non-verbal
communication, simple symbolic inference, and predicting the very near future. But we truly suck at complex problem-solving
and understanding the long-term consequences of our current actions.” And so that angle of these complex problems,
where there are long-term consequences, those are the areas where I think are most in need
of consumer protection, and behavioral science is going to help us understand that psychology. Thanks so much. Thanks, John. And finally we have Mike Baye. Well, first of all, thanks to the director
and her staff for putting this together. I know how much work it is to herd a bunch
of academic cats up here to give a little talk. I promise to keep my opening remarks here
at a very high level. I think it’s more important, given this is
a behavioral conference, to frame where I’m coming from so you understand my biases so
they are on the table. I had the pleasure of serving as the Director
of the Bureau of Economics at the Federal Trade Commission in 2007-2008, a couple of
years before the Bureau was created, and in that capacity I was obviously working with
staff on antitrust matters, but also consumer protection matters, like fair lending, payday
loans, FACTA study, mortgage disclosures, and a whole host of other things. And I have to say, as an academic, I found
the consumer protection issues far more interesting than I found the antitrust work. But just to put this into context, I have
a vivid memory of going into meetings with parties on an antitrust matter. It was a hospital merger. And staff’s concern—this would have been
staff in on the antitrust side of the Bureau of Economics at the FTC—staff’s concern
about the merger was that it was going to create anti-competitive harm because it was
going to reduce choices of customers in hospital markets. That very afternoon, I went to a seminar in
the Bureau of Economics where a behavioral economist came in, and was explaining that
unless we limit the choices on Medicare, people were just going to be unable to make decisions. So I throw that out just to point out that
it’s easy for us to say “I’m for this,” “I’m for that.” But these are very complex issues and that’s
why I applaud the Bureau for tacking these issues. They are not simple answers to these questions. That said, I also want to disclose that my
academic work has done theoretical work as well as empirical work, using models of irrational
or behavioral models, if you like, models with rational players. My empirical work has featured analyses with
field data, I’ve used survey data, and I’ve used experimental data. So I’m all over the map as an individual. And what I’d like to share with you, at a
very high level, is some issues that I think are important for the Bureau to think about
as it especially goes into next year’s symposium on cost benefit analysis, because in my mind,
that’s the tough issue, and that was the toughest issue that I dealt with in trying to understand
staff’s concerns on that matter. Let me just say, at a high level—we’ll get
into this in the back-and-forth, I’m sure—my view is both models of rational behavior,
traditional economic models, if you will, and models of behavioral economics, they are
both useful in explaining behavior. So it’s not an either-or. It’s a both. In fact, I would argue, consistent with what
some other panelists have suggested, is the bright-line differences between the predictions
of behavioral models and traditional economic models aren’t as bright-line-ish as you might
think. One can explain many behavioral anomalies
with a more sophisticated model of rational behavior. And on the behavioral side there are competing
models of behavioral economics that explain this same phenomenon. Differences in behavior are important, but
the real challenge here is not to identify something that might impact consumers’ behavior
or an irrational choice that consumers might make. In my experience, the difficulty is attempting
to do that difficult cost benefit analysis, to quantify the harm to consumers, to contemplate
the potential unintended consequences of the remedy, and to ensure that when you make an
enforcement decision or you adopt a policy that you are considering all costs and benefits. So just to give you one real simple example,
one might think of a policy that encourages entities to restrict the number of options
they give people, in the medical area, in the pension area, or whatever, to get people
over that shackle of just indecision, being unable to make a decision because they are
bombarded with choices. When I think about that as an economist, I
think about, wait a minute, that might be true in some environments but not necessary
every environment. So if you think about a restaurant, for example,
does a restaurant only want to have four choices on a menu, because people are unable to make
a decisions? As an economist, I would say a rational business,
if people get bombarded with information can’t make decisions, you’re only going to have
four items on the menu. So that’s a situation where the market may
solve a problem. On the other hand, let’s imagine that people
aren’t putting enough money in their 401(k)s, or they’re not making health care choices
because they’re just bombarded with too many choices and they are unable to make decisions. As we restrict the number of choices that
consumers have, which choices do we, as a regulator, allow them to make? Who determines that those choices are? Are there unintended consequences as we limit
their number of choices? Does that give advantages to small numbers
of players? Might that impact the prices that consumers
are paying for those products? So those are the types of issues that, as
an economist, irrespective of whether I’m a behavioral economist or I’m a traditional
economist, that I’ve got to help the commission, be it the Federal Trade Commission or the
Bureau, do to rectify its different policies. So hopefully we will have a great back-and-forth
here. Thanks. Thank you so much to all the panelists. Now we will dive into a number of questions
that I think will bring up a lot more of these issues. First up, in recent years, empirical findings
from the field of behavioral economics have been used as a way to explain consumer choices
and behaviors that deviate from a standard neoclassical model of consumer decision-making. Based on your expertise and review of the
literature, what findings from behavioral economics have the strongest support and what
findings have the weakest support? Where does behavioral economics do well? Where do neoclassical models do well? What implications are there for consumer protection
when they diverge? As part of this, feel free to bring in findings
from other social sciences that you think also have relevance to our understanding of
consumers and consumer protection. Brigitte, you can kick us off here. Yeah. So I’m going to take a couple pieces of your
question in reverse order. In general, where do traditional economics
models do well, and, in general, where does a behavioral lens help, and I’m going to reiterate
something that John said near the end of his remarks. Traditional economic models do a pretty good
job of predicting behavior for decisions that consumers make frequently and where the costs
and benefits are happening at the same point in time, and where individuals can learn from
feedback and they can map their preferences into the course of action that will give them
what they want. Traditional models do much less well with
choices that involve intertemporal tradeoffs, with risk and uncertainty, and with complicated
decisions where it’s difficult to map which choice is actually going to deliver the outcome
that you actually want. And a lot of the financial decisions that
individuals make, and a lot of the concern around consumer financial protection, has
to do with these intertemporal choices—saving, insurance against long-term risks, things
like that. In terms of the evidence on behavioral economics,
I think we have volumes of evidence that choice architecture matters a lot, and the choices
that individuals make can be swayed in significant ways, not in trivial ways, not in nudge-like
ways, but in significant ways in certain circumstances through the use of choice architecture. My own research documented, now almost 20
years ago, that changing the default option in the case of retirement savings plans, can
lead to up to a 50 percentage point change in the fraction of people who are participating
in a 401(k). That is not a small effect. That is a huge effect, and it dwarfs, by several
orders of magnitude, the largest estimated effects in the literature, from traditional
approaches to facilitate savings, namely financial incentives. So the change in behavior you get from a substantial
financial incentive is a tiny fraction of what you get from changing choice architecture. We find similar effects with organ donation. The magnitude is just as large, if not larger. We find similar effects with getting consumers
to sign up for energy plans that give encouragement to use less energy during periods of peak
demand, and defaulting consumers into an energy plan like that you get 60, 70 percentage point
differences in the fraction of consumers who are on that type of a price plan versus another
plan, depending on what the default is. So choice architecture and the default is
just one example of that. It has pretty robust support that it can have
large effects in certain circumstances. I would say the area of evidence where at
least I think the generalizability and the predictability of the results is more in question
would be social norms. We know that individuals respond to information
about what other people are doing. Those effects can sometimes be big. Often times they’re small. Sometimes they’re actually perverse. You tell people what other people around them
are doing and they act in a reactionary way, and you get effects going the other direction. So I would say that’s one area where I think
the evidence is weaker, because it’s a little bit more all over the map, and we need more
research into understanding when those types of interventions work and when they don’t. Great. We can just go down the line, or— Yeah,
maybe I can say something. I agree that the framing and kind of the opt-in
and opt-out type of decisions can be very, very important. Certainly, as a behavioral matter, there is
lots of laboratory evidence that suggests that is the case. It creates a challenge, though, when you are
attempting to do some type of cost benefit analysis of the effects of a policy. For example, one of the issues that we grappled
with at the FTC was privacy policies. So it turns out if you look at the behavioral
literature on privacy, if you ask someone, “How much are you willing to pay to keep your
data private?” they say, “Not much.” If you ask that same person, “How much would
I have to pay you to give up your privacy?” they say, “A whole lot.” That’s a preference reversal. That’s a behavioral anomaly. So it happens. What do you do as a policymaker? Which of those preferences do you use to evaluate
the cost and benefits? That’s one of the challenges. At a broader level, I would say that behavioral
economics does very well in the lab. It does less well in the field. It works in some instances but not in all. I think one of the challenges, in addition
to just quantifying harm in an environment where preferences aren’t stable or they are
reversing, or whatever, is the notion that many behavioral theories aren’t falsifiable. There is a paper by Hale and some co-authors
in the AER that looks at quantal response equilibria, which is a behavioral model of
decision errors, and shows that it imposes no falsifiable restrictions on the model. That is not true of all behavioral models. I’ve got a paper with John Morgan in the RAND
Journal with the same quantal response equilibria, only with a logit specification, and it does
impose falsifiable restrictions. I’m not dissing behavioral models, generally,
but falsifiability is an important element of science. So I would agree that there is abundant evidence
that a number of traditional economic theories are rejected in the data. But let me remind you what rejection in the
data means. Let’s imagine we’ve got 100 consumers. Let’s imagine that 85 of those 100 consumers—I
won’t do percentages because people may not understand. So you have 100 people and 85 of those people
behave exactly as homo economicus says they should behave—rational. Fifteen of those consumers don’t. At the 5 percent level, at the 10 percent
significance level, guess what you’re going to do? You’re going to reject traditional economic
theory in favor of the alternative that consumers are not behaving rationally. From a policy perspective, I think that whole
framing of that information is absolutely wrong. I think the way I would frame that information
is, look, we’ve got 85 people who are behaving rationally, 15 people who are behaving irrationally,
or by some other method. We need to recognize that not all consumers
are cut out of the same cookie cutter. There is heterogeneity out there. And from a policy perspective we want to make
sure that if we’re protecting those 15 percent, we’re not harming the 85 percent in a way
that, if we thought of a representative consumer, that harm more than offsets the gains to the
other. We also want to contemplate what the costs
to the policy are. We also want to contemplate what the impact
on social welfare might be. In Section 5 of the FTC Act, when an unfairness
claim is made, the FTC is required to do that analysis. I’m happy to chime in a bit. I agree with Brigitte that certainly certain
ways in which information and choices are presented or designed can be highly influential
on consumer behavior, and particularly when consumers don’t have a clear sense of what
they want to do, then giving them a recommendation or a default can be effective. But I think a lot of these systematic anomalies
that have been identified by behavioral economists, the focus has been too much on identifying
deviations from rational models and not necessarily understanding why those deviations occur,
and oftentimes it’s the rational model that’s taken as kind of objectively correct, even
though the rational model itself invokes certain assumptions about psychology, about the nature
of preferences, the nature of goals, and so forth. And so we oftentimes identify deviation, we
tend to generalize it without really understanding why it’s happening. I gave the example earlier of the endowment
effect and the attribution of that to loss aversion, which I believe is incorrect. I think also with respect to present bias,
for example, certainly many people are present biased, but also many people are future biased. We see now, of course, around the world these
very low interest rates for the long term and a lot of that is caused by, especially
in countries like Japan, people being, maybe from the perspective of central bankers, too
future biased. So I think we have to really carefully look
at these effects. I think many of the documentations that have
shown these systematic present bias, like hyperbolic discounting, some of those effects
could be attributed to other causes, in particular also to inertia and to transaction costs. So there is some bias and many people have
present bias, but I think this idea that it is a universal bias or almost universal bias
and systematic, I think is overstated. And I think more attention needs to be paid
to why people deviate from models and try to understand their behaviors as opposed to
just focusing on, hey, there’s a deviation, let’s generalize that. Okay. So I, in my work, do not frame the work that
I do in terms of some sort of deviation from some reactional model. That’s not the frame of people in psychology. So I take what things have stronger or weaker
support to be kind of like asking a biologist what things in biology have stronger versus
weaker support. There are hundreds of things. So it sounds like we focused on a few of them. I think the evidence is actually quite strong
that people exhibit this present bias. I agree with David, however, that it has multiple
causes, and that the underlying mechanism for observing something that looks like present
bias may be different in different circumstances. I guess a couple of things that Michael said,
on the angle, if you think that unfettered choice is a good policy in the face of present
bias, just ask yourself the question, imagine that consumers, for 1 month only, could take
money out of their Social Security accounts if they so wished. How many people in the room would advocate
that? The key thing with present bias is that people
say they prefer one thing when they are far away from the decision, they do something
different when they’re in the decision, and then after the decision, and they go back
to what they thought here before. So in those circumstances where the change
in their preferences is predictable, I think that that’s a situation where it’s a relevant
situation for policymakers to intervene. I don’t know if we’re going to talk about
this methodological stuff later. We’re going to talk about it later? I’ll hang back, then, on the methodological
things. Great. Thank you. The next question relates to some of what
was already brought up during the first question, but I’ll dig a little bit deeper. Some of the deviations from the neoclassical
model have been posited to stem from consumers’ use of heuristics, or mental shortcuts or
rules of thumb. In particular, what are the implications of
consumers’ use of heuristics, and how do consumers’ use of heuristics relate to whether consumers
have well-defined preferences related to financial products and services? As part of this you can obviously get into
the question of rationality if you choose, as well. David, do you want to kick us off? I mean, I’ll reiterate some of what I said
earlier. Heuristics, people use them to simplify decisions. We could use them because we don’t have the
motivation to exert a lot of effort, or because we don’t know how to optimize, so we might
use heuristics in those cases. They simplify decisions. They’re not inherently wrong or right. We can use, for example, one of the heuristics
that’s frequently talked about in the consumer financial decision-making literature, particularly
with respect to investment decision-making, is diversification heuristic. So let’s say you give me a portfolio, or I
have options in my 401(k). Let’s say I have two potential options, stocks
and bonds. Most people kind of will just evenly allocate
their money across those options. So maybe I’ll put 50 percent in stocks, 50
percent in bonds. Now you might say, okay, well, you shouldn’t
do that. You shouldn’t just have this 1/N heuristic,
or whatever. But, on the other hand, you can’t necessarily
tell me that that’s a wrong decision, because we don’t know what the optimal allocation
between stocks and bonds is, certainly not for a particular individual. And most individuals, they don’t know what
the right allocation is for them, and there is no objectively right allocation. So it’s a simplifying rule that’s used to
make a decision, but it’s not necessarily right or wrong. There is probably a wide range of different
allocations that could be equally right, and so it is not clear that using a heuristic
makes it any more right or wrong than any other type. Of course, in some cases heuristics might
be used to an extreme, but there is no inherent way we can say this is wrong. We just must use our judgment and say, hey,
this is too extreme an allocation, based on use of a heuristic. But there is no automatic decision rule that
you could say just because someone used a heuristic that they made a mistake. I agree with what David said. People use heuristics because heuristics are
an efficient way to simplify a complex decision, and we use them because most of the time they
work well. If they didn’t work well we would probably
stop using them. The problem is that even if most of the time
they work well, the small or even medium fraction of the time when they don’t work well could
be potentially problematic. So let me give you an example of a heuristic
that works well some of the time but doesn’t work well some other times. Individuals have a really hard time, as noted
earlier, with intertemporal problems. In the saving or the borrowing domain, where
interest rates matter and compounding over time matters, individuals have a hard time
with compounding. How many of you remember how to deal with
exponents, back from high school math? So what’s the heuristic that individuals use? The heuristic is that they linearize, because
it’s a lot easier to think in linear terms than in exponential terms. Well, if you’re dealing with a short time
horizon and, say, savings, over a short time horizon—2, 3, 4 years—this linear approximation
is not going to be too far of a deviation from reality. But if you’re talking about how much will
my money grow over 30 years, linearization is going to give you a very, very poor approximation
of what would actually happen over time, and you’re going to get similar effects when it
comes to borrowing. So heuristics matter, and I think one of the
problems, if individuals can’t always apply them well, is that nefarious firms can try
and exploit the fact that individuals have a difficult time with these heuristics. So come to the diversification rule. If people are using a 1/N heuristic, allocate
my money 50/50 between the two options in the savings plan, you’re going to get dramatically
different outcomes if those two options are a single stock and a single bond, or a stock
equity fund versus a bond fund versus a fund that’s already diversified, you know, a 75/25
stock fund and a 100 percent stock fund. You’re going to get different outcomes, and
you could start having firms pay to be on the menu because they’re going to get half
of the assets under management if individuals are using a 1/N rule, or giving bribes to
the person who is setting the menu. I mean, there are ways that firms can exploit
the inappropriate application of a heuristic in a way that actually harms consumers, above
and beyond just the fact that if the heuristic is misapplied there will be some inherent
harm in the first place. To me, the single idea that I would, again,
try to put forward is this distinction between expected utility and experienced utility. If it’s the case that what people expect,
on average, doesn’t match what they experience, then I think that policymakers should be concerned
about maximizing people’s experienced utility. I’ll leave it at that. Any other comments on this question, or should
we move on? You’re looking at me, so, of course, I’m an
academic. I’ll comment on anything. No, I mean, I actually kind of agree with
everybody here. I believe you can think of heuristics as just
an efficient choice. I’m an economist. I believe in making efficient decisions in
the face of scarcity. If my time is scarce, I might use a rule of
thumb instead of going through all the calculations. You don’t have to go to some behavioral story
to say that. That said, I would argue that if people are
systematically not understanding the terms of a mortgage, or the terms of an investment
vehicle, then a disclosure that provides that in a way that helps an individual understand
what the ramifications of his or her choice are to help them through that computation,
help them economize in that computation, might be a reasonable public policy. Thank you. So let’s move on to the next question which
deals a little bit more with the methodology, as John was talking about before. Many behavioral science findings in the literature
come from surveys or laboratory or field experiments. Are there particular merits or drawback to
studying consumer behavior in controlled settings, and what factors related to generalizability
or effect size should policymakers consider when attempting to apply these findings to
broader, real-world settings? I would disagree with something that Michael
said about the idea that field studies are inherently more generalizable than lab studies. The truth is that all studies have issues
of generalizability, because human behavior is moderated so that any given intervention
has differential effects, depending on certain side conditions. I don’t believe that there is any rigorous
evidence suggesting that there is more generality if you try to replicate from field settings
than across laboratory settings. It’s just that because it’s so expensive to
do field studies they don’t get replicated, so we don’t know. I was on sabbatical at Columbia and I just
came from a conference that was all about all the brave new world of online new media
stuff in marketing, and all the papers were these very large-scale field experiments about
the effects of online advertising, and they did not replicate each other. So it is not the case that there is anything
about the field versus lab setting that distinguishes in generality. It just means that we need to look very carefully
at subgroup level, how does the same treatment affect people differentially, in different
subgroups, and we should be interested in replication and seeing, in some systematic
way, how things do or don’t generalize. And on that I would say that I would love
it if the Bureau would get in the business of meta-analysis, that is, of systematic recording
what has been found on a given topic, and not just cherry-picking studies but rather
looking at all of them and trying to line them up and try to explain why some find different
results than others. Go ahead, Mike. If I could just respond, I apologize because
I think John and I might be talking across purposes. I didn’t say I prefer field experiments to
field data. In economics, we have economic experiments,
we use surveys sometimes, we use field experiments, and we use field data. When an economist uses the term “field data,”
he or she means like real-world data, what really goes on in the real world. Steve Levitt and John List have a paper that
kind of talks about some of the problems that you might have with field experiments, or
experiments or surveys, and they include things like the subject pool—how does that match
up with the population of people in the actual environment who are making decisions?—and
the nature of the information. Typically, in at least the surveys and the
experiments that I’ve done, you have very stark, simplified environments to isolate
the impact of one thing, like a disclosure, for example. Can you see the small print or not? In reality, when someone makes a purchase
decision it might be a journey. You might start out with that one little ad
that sucks you in to have a conversation with a person trying to sell you mutual funds,
but then you might have a conversation that lasts an hour or more, where additional information
is coming, and it’s difficult to include that in a laboratory setting. And I’ve used all of them. I’m not saying I’m against field experiments. I just think that you are subject to potential
framing when you do experiments or surveys, where you’re potentially impacting the behavior
of the individuals that you’re studying. And that’s why I prefer field data, real-world
data, although to isolate, for example, the impact of competition on prices, I’ve used
experimental data where I can zero in on how many sellers there are. That’s something that’s difficult, if not
impossible to do in every market environment. Can I just chime in with one point that relates
back to the previous question. I think many of us, either as policymakers
or as researchers, use a heuristic when we are evaluating evidence—and I’m going to
put “evaluating” in air quotes—which is that you read the abstract and you look for
the top-level finding. This has this impact. And then you want to say, okay, I’m going
to go take that and use it to make policy, or generalize it, something like that, because
we don’t want to do the hard work. That’s our heuristic. We read the abstract. And I think the really important thing that
we need to recognize is that every study has strengths and weaknesses, and if you want
to know the extent to which you can generalize the results of a particular study to your
application at hand, you’ve got to read the actual study and figure out what are the potential
factors that might bias the results of the study in a different context, and that is
hard work. And we don’t like to do hard work. We like to take the shortcut. And if you read a laboratory study, yeah,
there might be situations where it generalizes pretty well to the real world, and there might
be situations where it doesn’t. But if you read a field study, that’s going
to be true as well. I would say if I were a policymaker, I would
put more weight on a finding that had replicated across a number of laboratory studies and
across a number of field studies, so it’s been shown to generalize across a number of
settings. If it generalizes across a number of different
types of populations and if it generalizes across many different types of application,
then I’d feel like that’s a pretty robust finding. But I don’t think we’re at the stage in the
literature where we have that type of robust evidence for many types of, quote/unquote,
“behaviorally informed interventions.” I think in the majority of cases we are dealing
with severe limitations, just because enough replications haven’t been done in laboratory
settings where it’s cheap, but even more importantly, in field settings where it’s very expensive. I’ll just quickly piggyback on what Brigitte
said in terms of I once heard Daniel Congdon say that he’s never cited a paper that he
hasn’t read, which I think is impressive because I’ve never read a paper I’ve cited. But I think I agree with a lot of what has
been said here, and I just would reiterate, again, the importance of understanding why
an effect is happening. Often when we run a study, the key is to understand
why it’s happening. If you understand why it’s happening, then
you can generalize. It doesn’t mean the effect will generalize. Effects don’t generalize. Psychological insights generalize. Effects don’t generalize because of the reasons
that John mentioned—the contexts are different, there’s lots of moderators that can moderate
the effect. So you have to make a judgment call, is the
context that you’re trying to apply, the effect from one context to another, is similar in
the important aspects that is going to lead to a similar effect? The effect doesn’t occur everywhere. It occurs depending on having similar context
and similar attributes relative to ones that are key to driving the effect. So you have to really understand why an effect
is happening. And the other thing I’ll just mention, is
oftentimes—I agree with John, also, that there is no difference between whether you’re
going to replicate something just because it occurred in a lab or the field. But the lab studies are often cheaper, and
we tend to calibrate our stimuli in order to get an effect. And that’s not necessarily a bad thing. It just means we’re trying to understand—it
could be used to try to understand the psychological process. But it could mean that the effect is fragile,
in the sense that it won’t translate to another context. We’re doing a really good job of linking these
questions together. The next question relates to how policymakers
use incoming information, additional studies, to update potentially our knowledge in this
area. Science progresses as researchers, like yourselves,
continue to explore important issues in your field, often finding additional evidence that
confirms, disconfirms, or adds an important nuance to existing evidence. How should policymakers weigh incoming evidence
that challenges our previously held beliefs about consumer behavior, and is there a standard
that we, as policymakers, should apply to the weight of a single study or an additional
two or three, in terms of our thinking on consumer behavior? I think that’s the big question when it comes
to making policy is how do you weigh the evidence, and I think most of us would be well advised
to exert some degree of caution on the basis of one particular study. We’ve got a lot of evidence that’s been accumulating
over the past few years on replicability in the social sciences, and a number of important
studies have failed to replicate upon further examination. So I think with one study you want to try
and really understand why do we think that a particular study might be delivering different
results from what the literature out there is telling us. That said, there are also lots of good examples
of where there was a prevailing belief and a single study challenged that belief, and
it’s actually held up pretty well over time, once the evidence came to bear. If it’s not on a really significant matter,
then you might say let’s wait and see whether further evidence support this. But in some points you have to make a decision
on the basis of limited evidence, especially if we’re dealing with significant financial
harm or we’re dealing with life or death in the medical sciences, for example. But I think this is where government agencies
can help both researchers and policymakers by trying to bake evaluation into things that
are done, so we can get more rapid accumulation of evidence that helps inform everyone else
down the road, as they are trying to make better decisions. So I’m happy that the CFPB is trying to at
least do retrospective analyses that gives us a better evidence base going forward. I want to just reiterate or expand on a point
I was making earlier. In my mind, the single most important methodologist
in psychology is a guy named Lee Cronbach. Many of the things that people do in psychology
are attributable to this guy. He was the one who introduced the idea of
how it was you could actually have scientific statements about things that were not directly
observable. But one of his most important influences is
on the study of the generality, or lack of generality, of findings. One of his basic insights is that we, when
we write these individual articles, grossly overgeneralize, and grossly overestimate the
degree to which the finding is going to be robust to different variations. So that’s why it’s really critical to have
the—and I would say not just direct replications, or attempts at direct, but rather conceptual
replication. There, the thing that I think that the Bureau
should do in relation to this is to rely heavily on this technique of meta-analysis. Meta-analysis is this technique to summarize
what has been done, not just to say, overall, did it work, but to understand where it works
better or less, that will allow these kinds of cross-benefit analyses that others have
been showing. For instance, I’ve done some work on financial
education and financial literacy that shows the relatively small effects of financial
education on behavior. However, the effects are bigger if the interventions
are close in time to some behavior you’re trying to influence. Well, that dictates like how you should try
to operate on financial education, to try to get it at a point in time close to a decision. So I think the Bureau should be very cautious,
like Brigitte says, about any given study, but should really be in the evidence and trying
to accumulate and systematically record and analyze what is the pattern of the data show. Great points. Just an observation, something that, again,
is very different on the antitrust side and the consumer protection side of the Federal
Trade Commission, in any event. On the antitrust side, investigations are
specific to the specific investigation. One doesn’t try to come up with some generalized
rule that a 4-3 merger is anticompetitive or not. One looks at specific facts. So you might have one 4-3 merger where firms
are competing in prices, selling homogeneous products, something that an economist would
say resembles homogeneous product competition, and there may not be any competitive concerns. There might be another industry where those
firms have a history of colluding in the past and/or they’re producing heterogeneous outputs
or other factors that might make the price effects of a 4-3 merger more specific. So keeping that in mind, I guess this whole
notion of generalizability, it seems to me that certainly you want to be careful not
to generalize—take some study in a lab and then based on that impose some onerous restrictions
on businesses in the real world. But if you’ve got data from the real world
that shows a particular business practice is harming consumers, and you can demonstrate
that using but-for analysis, then, by all means, protect consumers in that instance,
even if it’s not generalizable. In my view, the only place generalized rules
come into play on the antitrust side is in the so-called Hart-Scott-Rodino premerger
notification, where the agency uses basically Herfindahl indices, which are just glorified
measures of market share, squared and summed, but basically just kind of heuristics to determine
whether or not this is likely to be among— So the meta-studies could be helpful for the
Bureau in coming up with some heuristics that kind of do what they used to call triage at
the FTC, kind of separating cases that look problematic from ones that don’t. But that triage is only to help you allocate
resources to the cases where you really ought to be discovering the facts there. That’s my view. I’ll just quickly add that while I agree with
what was said, I think you also have to look, when we’re trying to evaluate the totality
of evidence for a particular phenomenon or a particular intervention, it’s important
to look at all sorts of evidence and not just controlled experiments but also secondary
data and also historical cases, I think, could be very useful. For example, if I’m thinking about a particular
intervention, let’s look at cigarette smoking. What was done in that case? So there it’s more qualitative interpretation,
but that could potentially inform the totality of evidence for certain kinds of interventions. So I think this integrative view is quite
useful. Great. Thank you. So moving on to the next question, many consumer
financial decisions involve making tradeoffs over time, as we talked about earlier. Some findings from behavioral economics suggest
that many consumers may display time-inconsistent preferences and, therefore, are susceptible
to preference reversals. Are there findings from this literature that
might help policymakers better understand consumers’ financial decisions? John, do you want to start? Mike and the mic. Is that just curious that Mike never turns
on the mic? All I want to say is, I mean, I do believe
there is laboratory evidence that consumers engage in preference reversals. One of the first papers I remember, as a graduate
student, it’s a paper by Grether and Plott, where some economists doing experimental economics
were out to prove that the psychologists were flat wrong, and the reason you get preference
reversals is that when these silly psychologists—and I don’t really mean that, but this is the
spirit of the Grether and Plott paper—these silly psychologists just ask people hypothetical
questions. And if you ask people hypothetical questions,
you’re going to get hypothetical answers. So what Grether and Plott did is they replicated
the preference reversal studies using experimental economic tools, which requires that you pay
subjects, you provide them with economic incentive to make decisions. And lo and behold, when they incentivized
subjects to respond, the preference reversals got worse instead of better. So there is laboratory evidence that preference
reversals are a real phenomenon. That’s all I’ve got to say about that. What I will say is that many phenomenon that
your knee-jerk reaction is a preference reversal can easily be explained by the difference
between an ex ante and an ex post decision. So I don’t know about you but I bought house
insurance last year. I also had automobile insurance. And I didn’t have a wreck. Ex post, that was a mistake. I made a bad decision, right? Ex ante, when I made that decision, it was
a good decision. So I would warn you, as an enforcement agency
and someone trying to protect consumers, make sure that if you have a phenomenon that you
believe is consistent with a preference reversal, that it is, in fact, that and not just an
ex post outcomes that differs from an ex ante expectation by consumers. So you’re not silly. Heck no. You’re smart. So in a case that Mike just described, the
fact that he paid for insurance and he didn’t use and now he regrets it, like that is not
a mistake. But again, I would distinguish a case like
that from when, systematically, if people bought a product, and systematically, after
the fact, they concluded they had made a mistake, and it was predictable in advance that that
was going to be the case, then I think that’s a real phenomenon. That’s where I would focus my efforts. I think we’ve talked a bit about intertemporal
decisions so far. Maybe I’ll just consolidate, I think, some
of what we know about the problems that come up in the domain of intertemporal decisions. One is that intertemporal decisions almost
always involve risk and uncertainty, that is to say we don’t know what will happen in
the future. Mike did not know that he was not going to
get in an accident and a tornado was not going to rip the roof of his house. And that uncertainty complicates decision-making. Individuals do not do a good job of making
decisions in probabilistic domains. So intertemporal decisions almost always involve
uncertainty, and the further the horizon is for the outcome of the decision, the greater
the amount of uncertainty. Individuals also don’t do good with things
that grow over time, so exponential decision-making. We are just not any good at that. That is also a key feature of many intertemporal
decisions. And then individuals make different choices
for today versus the future than they do comparing things at two different points in time in
the future. So when it comes to intertemporal—and another
factor is we don’t have a lot of opportunity to learn from feedback or from experience
when you’re dealing with something that you do very infrequently. So many people who purchase homes, that’s
not something they’re doing every six months. They’re doing it once a decade. And the financial market changes and economic
conditions change from when you bought your home 10 years ago and when you’re trying to
repeat that transaction. Something like life insurance or retirement,
you don’t get a do-over at all. You don’t learn from experience at all. So the types of products that involve long-term
intertemporal choices are the ones that I think are most problematic, from the consumer
decision-making standpoint, and, therefore, the ones where consumers are probably most
in need of a good regulator trying to facilitate better decision-making and trying to make
sure that firms in the industry are not unfairly exploiting the types of biases that might
arise from all of the things that I just mentioned. I largely agree with the other panelists here. I would add, though, again, it’s important
to understand why people are having these timing-consistent preferences because they
could be due to different reasons, and some examples that look like timing-consistent
preferences are not, or examples of present bias. So if you ask me, “Would you take $10 today
or $20 a month from now?” I’ll take $10 today. It’s not because I’m present biased. I think that’s because of the mental transaction
costs that keep in mind. I have to wait a month. I’m actually very future biased or future
oriented. I’m very good at saving, and so forth. But if you give me that particular choice,
I will take $10 today. So we have to really thinking about what illustrates
intertemporal preference and consistency and what doesn’t really try to understand why
these effects are happening. They can happen for various reasons, like
some of the other panelists said. It could be because I’m bad at forecasting
the future. It could be, in many cases, because I have
low willpower or because I have different motivations that are active at different times. But we have to understand why those are happening
in order to try to remedy them. But I generally agree that present bias is
common among many consumers, not all, and it’s probably more harmful than future biases
to the individual consumer. So it is something that I think should be
addressed, but we really have to understand why it’s happening and in which consumers
it’s happening. Great. We have about 3 minutes, so likely not enough
to get to our last question, which we touched on earlier, a little bit about mental counting
and some framing issues. I want to thank the panelists very much for
your thoughts this morning. We, I believe, are going to a coffee break
now. Thank you so much. You will see another set of panelists after
the coffee break, talking a little bit more about the policy applications to a variety
of these types of topics. So thank you so much. Thank you. So everybody is up here. Thank you for that. Jason Brown is going to take us through and
moderate our next panel. Jason, the floor is yours. Thank you, Andrew. I’m very happy to continue the conversation
that we started this morning. This session is intended to generate discussion
on behavioral law and economics, a topic potentially relevant to many aspects of the Bureau’s work. We were slated to have five accomplished experts
to offer their perspectives on the subject. Unfortunately, one of the Joshua Wrights on
this panel had to excuse himself, because of a family emergency. So, nonetheless, the remaining four, I’m sure,
will provide a wonderfully diverse set of perspectives. We have Jan Pappalardo, who is the Assistant
Director of the Division of Consumer Protection at the Federal Trade Commission’s Bureau of
Economics. We also have Josh Wright, who is the Executive
Director of ideas42; Greg Elliehausen, who is a Principal Economist at the Federal Reserve
Board of Governors; and finally, up for another round, Brigitte Madrian, Dean and Marriott
Distinguished Professor, Brigham Young University Marriott School of Business. Thank you all for joining us. Without further ado, let’s hear from our panelists. Jan, would you like to start? First of all, it’s a wonderful pleasure to
be here today with such distinguished panelists. I learned a lot this morning. I’m sure I’m going to learn a lot more as
the morning progresses. Today I want to reflect on how behavioral
economics and its role in consumer protection policy has evolved during my three decades
at the Federal Trade Commission. I came straight out of graduate school. I think John talked about his background being
a little bit unusual. I think mine may be a little bit unusual too. I went to Cornell University, got a PhD, with
my major field being consumer economics, which is a field that’s not widely studied, with
minor fields in statistics and industrial organization, which I thought was a great
background for coming to the FTC. At the outset I’d like to be clear on two
points. First of all, I am speaking for myself and
not for anybody else at the Federal Trade Commission. And secondly, I am not a behavioral economist. Now have I seen our work on mortgage disclosures
has been cited in the literature as an example of behavioral economics, but I know at least
one friend who is a bona fide behavioral economist who would disagree with that assessment. There’s a lot of ground to cover in only a
few minutes, but in the time I will try to address four big-picture questions from my
perspective, as somebody working in the trenches of consumer protection for my whole career. First, what is behavioral economics and how
does it differ from other fields of economics or marketing? Second, why is there debate over the role
of economics in consumer protection policy? Third, how has behavioral economics evolved
and contributed to consumer protection policy over the past few decades? And finally, how can behavioral economics
improve consumer protection policy in the future? If I don’t get to the end, to the punchline,
I always tell my guys in my group that I would like to see the bottom line on the first page,
so here’s the bottom line on the first page. Overall, I conclude that behavioral economics
has evolved substantially over the past few decades; debate over the role of behavioral
economics and public policy is real but narrowing, as scholars clarify and discuss the terms
of the debate; and behavioral economists develop welfare analysis to assess consumer policy
interventions. What matters for improving consumer protection
regulatory policy is not the field with which the research is identified, but the quality
of the research and the extent to which the research gets us good estimates to answer
two questions. One, what are the likely costs and benefits
of a business practice, and two, what are the costs and benefits of possible remedies? Moreover, policymakers and analysts need to
clarify the objective of a consumer policy to assess the costs and benefits of any intervention. Is the goal to improve consumer welfare or
total welfare? Is to remove deception without privileging
a particular consumer choice or change consumer behavior in a particular direction, or is
it promote efficiency or equity? The policy best suited for one policy objective
may not be best for another policy objective, and I think this is something that we have
to be really clear about. In the academic literature, there is discussion
of various goals of behavioral interventions, but there is this question of how do these
goals in the academic literature marry up with the mandates of regulatory agencies. There has been discussion earlier about what
is behavioral economics, and I think before you can discuss the role of behavioral economics
you have to have a definition. Camerer, Loewenstein, and Rabin define behavioral
economics as “a subfield of economics that seeks to increase the explanatory power of
traditional models by incorporating more realistic psychological foundations.” Now there is some debate about this definition. A critic of some behavioral economics findings
and recommendations is David Levine, and he writes, “Behavioral economics is hard to define,
because it’s a terribly trendy term, some research that antedates the invention of the
word, and has little to do with psychological theory or data, such as learning theory, is
sometimes referred to as behavioral. Sometimes it seems as if anything these days,
besides a purist irrational model, sells itself as behavioral.” One might say, informally, that behavioral
economics is concerned with systematic deviation from standard economic models. The question is, which standard economic model? There are many different models of rational
behavior. Some incorporate more constraints than others. So I was trained in a consumer economics department
where the focus was on household decision-making, and in household decision-making models, the
idea is the household seeks to maximize utility. They have some higher-level good that they
want to achieve, maybe education, happiness, health, pleasure, be it what it is. And then you have constraints that make it
very difficult to reach the things that you would like to reach. You might have time constraints—we all have
24 hours in the day. You might have wealth constraints—very few
of us can write an open check whenever we want to. We also have household production constraints. Some of us are more or less efficient with
combining scarce time and wealth to produce these higher-level goods. The bottom line of this is that constraints
matter, and what a constrained consumer would do might be different from what an unconstrained
consumer would do. So it’s very important to think about what
the counterfactual model of rational choice is, or the economic model that you’re using,
to compare to various behavioral or other models. I’ve wondered how much daylight there is when
you look at some behavioral findings, when you look at very constrained maximization
problems by consumers, because constraints matter, as I said, and time and information
processing constraints matter. So any time you make it easier for people
to comprehend something or understanding something, you’re essentially loosening a constraint. In addition to debate over the definition
of behavioral economics among economists, I’ve often wondered what’s the difference
between behavioral economics and marketing. When I came to the FTC 30 years ago, there
was already a long tradition of collaboration between economists and marketing researchers
who either worked at the Federal Trade Commission, or maybe there was collaboration with people
who were outside of the Commission, and we learned a lot from our marketing colleagues. One thing that I would say, as a take-away,
what I thought then is what I still think now, which is that economics is a powerful
tool for framing up problems, and we think about consumer choice, we think about what
things belong in a demand function, what motivates people to choose A versus B. When we estimate
demand function, economics says price of that good belongs in there, the price of other
goods belong in there, income belongs in there, information, taste and preference shifters. And what I thought then, and I still continue
to think now, is that people trained in psychology often have the comparative advantage, and
giving us insight into how to measure these taste and preference shifters and information. We also learned from our marketing research
colleagues that these terms that are now popular mattered to them decades ago—things like
framing, mattering, the whole field of advertising—but how do you frame up a product so that people
want to buy the product? And there are many, many years of research
that the marketing colleagues have done to try to understand these things better. Regardless of the definition of behavioral
economics, it’s important, it’s here to stay, and I would say that it would be useful for
colleagues in all these areas to talk more with one another, to find out what we know
from these different traditions and from the past. One might wonder why consumer protection economists
and policymakers have not universally accepted behavioral economics. I can tell you a little story of my own. When I was at the FTC, behavioral economics
terms were new, and one thing that was put forth was that mortgage disclosures were an
example of a great behavioral intervention, because, of course, a disclosure is going
to be helpful to people. And I can tell you from our own research at
the FTC that we found that mortgage disclosures that were designed to be helpful to people
could actually be confusing or downright misleading, suggesting that policymakers really need to
be careful before they use an intervention, even when it seems to be simple, like a mortgage
disclosure, because how you give the information to consumers, and whether you even have a
metric that’s useful to impart information, like the APR, which can actually be misleading,
is really important. Time is running out. Just a few final notes. I think that as we go forward in discussion
about the role of behavioral economics in public policy, it’s really important to think
about how these terms—rational consumer, behavioral consumer—are going to be used
in consumer protection policy. Because much of consumer protection policy
law is based on the concept of the reasonable consumer, and do we need to rethink that or
not? That’s a big question. Thank you, Jan. Josh? Thank you. Thanks for having me here. I’m also a little bit odd to other people
on the panel, or the previous panel, in that I’m not an economist. I am trained as an MBA. I spend lots of time talking to economists
and psychologists, and my wife would say I have a problem reading too many papers about
economics. That’s not a problem. I would agree. So I’m really glad that the Bureau is having
this conversation, and hopefully, from my perspective, this is not just one conversation
but an ongoing effort by the Bureau to reach out and have conversations with experts in
many fields. To me, ultimately what we’re talking about
today is just good public policy. I’m not going to read my remarks that I wrote
verbatim. I’ll just touch on some highlights. You know, behavioral economics, in the larger
field of behavioral science—and that’s really what we think of at ideas42. We think of our work as using behavioral science
to help solve social problems. And that’s because it draws from behavioral
economics, psychology. I mean, traditional economics is also the
study about how people are making decisions and how actors behave in the marketplace. So, to me, it’s all behavioral science. Behavioral economics, specifically, I think
now undoubtedly has a rich set of evidence that it’s useful, and that evidence comes
from both lab and field studies, as well as, to use Michael’s term, data, the use of data
and analysis of real-world data. It’s a powerful tool. It’s just one tool. It’s not a panacea. Traditional economics, and the models that
that brings to the table, is also a super powerful tool. Psychology is a powerful tool. Sociology is a powerful tool. So, to me, it’s really about not being wedded
to any one of these tools but saying what is the evidence that the tools produce, and
how rigorous is that evidence? I think, to John’s point very much, you have
to look at the preponderance of all the evidence and try and do meta-analysis. And every one of us who wrote our opening
statements cherry-picked. We all had a point we wanted to make and so
we cited three or four studies that happened to support that point. We didn’t go and look at all of the evidence
because that would be time-consuming, and it doesn’t make for a good study when we write
something. But the Bureau’s job, really, is about looking
at the totality of the evidence and doing more meta-analysis. And then I would say that all of these insights
are very context specific. The insights from psychology, from economics,
the models, as was said in the first panel, are simply just a stylized or a way to think
about the world, simplify the world, in many cases, to see if we can draw insight from
it. Choice overload was talked about in the first
panel. It’s very context specific. And Michael brought up the issue of, well,
you have pensions and you have decisions you make in a restaurant. Why isn’t there a problem with decision-making
when you go the restaurant? It’s the context. The restaurant is a forced-choice model. If you do forced-choice with pensions you
get pretty close to Brigitte’s findings around defaults. How many people have gone to a restaurant,
had a reservation, looked at the menu, and said, “There are 40 choices. Forget it. I’m leaving.” You don’t do that. You’re there. There is a forced-choice structure where you’re
going to order something. Do you exactly order the perfect meal? Is your utility maximized? Who knows. And as regulators we shouldn’t care about
that. But when it comes to pensions, if there is
no forced choice situation created then people are going to do what people often do when
they have a choice. It’s not that they’re not making a choice. They’re saying, “Eh, maybe I’ll make the choice
later.” So that’s all to say that this context in
which people are operating is very specific, and we have to be very cautious about generalizing
any of what we’ve learned to any one context. We have to look at the totality of the evidence,
think about the specific products or practices we’re thinking about, and the context in which
people are operating in that world. And in think in behavioral science, and behavioral
economics in particular, there is a lot of focus on the form that people look at, or
the disclosure that happens. That’s just part of the context. That’s part of the thing that we should think
about. But context goes all the way up to poverty. So scarcity and the idea of chronic scarcity—great
work by two of our founders, Eldar Shafir and Sendhil Mullainathan—is a context. It causes people to often be more present
biased. It causes people to tunnel on very short-term
goals. Yes, they’re also very, very good at managing
the day-to-day, most immediate needs associated with their finance. Like if you go up to a lower-income person
who has come out of a CVS and asked them how much the toothpaste costs, they know exactly. If I ask you, you have no idea. But it also means that they tunnel on solving
this very immediate problem they have, and they may not factor in all the implications
of taking on, for example, credit. And this brings me to my last point. I’m very much a fan that we should be thinking
about cost benefit and consumer welfare. In my document I say I think we should be
a little bit more expansive about that. There are lots of connections between financial
health and physical health, so we might want to look at physical health. There is a great new study around payday lending
and suicide rates. So we want to look at, I think, a broad expanse
of those things. But I also think that we don’t want to do
that in a blanket way, just roll it up for all consumers. I do think we have an obligation, as society
and as regulators, to think about who are the most vulnerable among the Americans in
our society. You know, there’s a reason there’s an Office
for Older Americans in the statute for the CFPB. We have to think about older Americans. We have to think about young Americans who
are newly entering the financial market. We have to think about Americans who might
have disabilities that would make making financial decisions challenging. And it also means we need to think about lower-income
Americans or Americans that, because of their race or culture, might encounter systematic
racism. So we have to factor those things in as we
do that analysis, and that’s going to make some people uncomfortable, because we’re going
to say, “Well, some parties’ consumer benefit might not be as great because we’re going
to do something to protect one set of consumers.” And that’s a tough decision. Now I think if we do regulation very well
and we look at the evidence, we can limit those situations. So to go back to Michael’s situation, can
we design a regulation so you prevent the harm to the 15 percent but still the 85 percent
get the benefit? So I’ll just close on that and I look forward
to the discussion. Thank you, Josh. Greg? Psychology has influenced economics beyond
what we normally think of today as behavioral economics. Around the mid 20th century, behavioral research
was developing information processing models of decision-making. The work of Herbert Simon contributed significantly
to this effect. Simon proposed that “cognitive and time limitations
prevent individuals from obtaining full information and evaluating all possible alternatives to
achieve optimal outcomes.” Instead, he argued, “Individuals simplify
using heuristics, shortcuts that make decision processes easier. Heuristics enable them to limit the decision
process, ending the process as soon as a satisfactory alternative is achieved. Decisions processes are rational if they lead
consumers to achieve desired goals.” He called this process satisfying in the concept,
bounded rationality. Bonded rationality has stimulated research
in several areas. In one area, researchers use surveys to question
consumers about the extent of their decisions processes when making economic choices. Much of the survey research involved purchases
of durables and the financing of durables by consumers. The findings indicate the consumers simplify
and take shortcuts in the decision process but tend to be purposive and deliberate when
required. Work by George Katona and his colleagues at
the Survey Research Center at the University of Michigan are major contributors in this
area. Also notable is Day and Brandt’s study of
consumer decision processes for the National Commission on Consumer Finance. And I might add that when the board of governors
had to deal with consumer protection rule-writing, we relied on surveys, often by the University
of Michigan Survey Research Center, to inform our decision-making. Another area, called the heuristics and biases
program, which we talked about quite a bit this morning, is concerned with biases in
behavior arising from the use of heuristics. Researchers relied primarily on experiments
to detect deviations from rationality, as defined by statistics or logical rules. Frequent deviations are attributed to faulty
application of heuristics to solve problems. Individuals are capable of overriding heuristic
biases, but the heuristics and biases literature argues that due to limits in cognitive ability
and motivation, consumers often fail to do so. Much of the recent work in behavioral economics
is heavily influenced by this research program. A third area is also concerned with heuristics. It views heuristics as specialized tools guiding
decisions in specific environments. This program is called the fast-and-frugal
heuristics program. Such heuristics allow fast decisions with
limited information in uncertain environments. I note that they emphasize uncertain environments
rather than risky environments, which much of the heuristics and biases literature deals
with. The program argues that in the right environment
heuristics provide for accurate decision-making. It does not involve a trade-off between accuracy
and cognitive effort. Gerd Gigerenzer at the Max Planck Institute
in Berlin is a frequent contributor to this area. We talked earlier this morning about the 1/N
portfolio allocation rule. There have been studies that suggest that
the rule performs about as well as mean variance portfolios. It is interesting to note that Harry Markowitz,
the developer of portfolio theory, admitted to using a 1/N rule for allocating his retirement
savings. As mentioned, behavioral research indicates
consumers do not always make cognitive efforts required for an extensive decision process. Individuals often take shortcuts, simplify,
and use heuristics. Cognitive effort tends to be reserved for
situations where commitments in money and duration are great. Past experience and information are insufficient
or obsolete, and outcomes of previous decisions are regarded as unsatisfactory. In situations where consumers have previous
experience and were satisfied with past decisions, they often make choices with little further
deliberation. That cognitive biases in time-inconsistent
discounting exists is well established in the behavioral literature. Some research suggests these psychological
considerations influence consumers’ credit behavior, however, the extent to which cognitive
biases and time-inconsistent behavior affect actual credit decisions is unclear, and I’ll
talk about that in a few minutes. Experimental evidence supporting cognitive
biases in time-inconsistent discounting is sensitive to format, context, and content
of the problems presented to study participants. When problems are framed differently, the
results sometimes contradict previous findings. Individuals may be predisposed to impulsive
behavior but they also have capacity to exert self-control and implement forward-looking
plans. An individual facing an impulse might yield
to the impulse if it does not perturb the plan too much. To be effective, self-control requires that
internal inhibitions become stronger as awareness of the cost of impulsive behavior increases. Empirical evidence indicates that consumers
generally use credit to finance purchases of relatively expensive durable goods, not
to smooth consumption. In doing so, behavioral concepts such as free
commitment and mental accounts may be used to manage behavior. Such concepts may not be optimal in the sense
of global utility maximization but they may be sensible when future prospects, preferences,
and resources are uncertain. Katona described the concept of rationality
emerging from behavioral studies as follows: “If careful deliberation were to find as comprising
all the features of decision-making, the conclusion would emerge that almost all people proceed
in a careless way in purchasing large household goods This conclusion is not justified. Deliberation may be strongly focused on one
aspect of the purchase to the exclusion of other assets. Therefore, it may be considered careful deliberation
if some, but by no means all of the features of problem-solving and thinking are present.” In reaching this conclusion, Katona focuses
on the process rather than the outcome of assessing rationality. For Katona, rational behavior involved choosing
appropriate means for achieving improvements in one’s well-being. Katona argued that the predominant alternative
to rational behavior is habitual behavior, not irrational behavior. Habitual behavior occurs when consumers are
satisfied with previous decisions and do not experience any stimulus for change. Consumers’ use of credit cards is frequently
seen as vulnerable to biases and impulsive behavior. However, analysis of actual credit card behavior
indicates that consumers are sensitive to price consistent with the predictions of economic
theory. When credit card companies increase interest
rates on an account, consumers reduce new charges, reduce existing balances, and shift
charges to credit card accounts that have lower interest rates. Over the course of the year, they reduce total
credit card balances from the level before the price increase. Also, based on subsequent account use, evidence
suggests consumers generally make cost-minimizing choices in trading off interest rates and
annual fees when choosing new credit cards. When they make mistakes, the mistakes are
usually relatively small. If mistakes are large, consumers generally
correct the mistakes. Although some consumers do not correct large
mistakes, persistent large mistakes are not the rule, and also, it is the case that frequently
people chose an account with an annual fee because it limited their maximum loss. So that’s not an optimal decision, limiting
maximum loss, but it seems perfectly reasonable. Limited theoretical evidence indicates that
a satisfying heuristic produces long-run optimal outcomes in some circumstances. Empirical evidence from Experimental Economics
supports this theoretical conclusion. The studies consistently indicate that individual
decisions based on limited information in experimental markets produce prices and allocations
that converge quickly to the neighborhood of optimum equilibrium values. The results occur even though participants
do not engage in extensive weighing of alternatives. The behavior of participants using various
heuristics with limited information produces efficient market outcomes. Experimental studies have also found that
market environments reduce the incidence of preference reversals for risky prospects and
losses from failure to recognize some costs and opportunity costs. Vernon Smith argued that “the findings of
these studies suggest that markets reinforce or even induce individual rationality.” Smith speculated that market prices provide
the stimuli that cause individuals to take actions to better their solutions. These actions move prices and allocations
to competitive equilibrium. Focusing on whether or not individual decisions
are optimal misses the point. To conclude, I would say that at this time
neither existing behavioral evidence nor standard economic evidence supports the general conclusion
that consumers’ financial decisions are not rational or that markets do not work reasonably
well. In examining decision processes, behavioral
research complements the findings of standard economics and provides useful guidance for
regulatory policy. Thank you, Greg. And Brigitte, we will give you a pass on making
an opening statement. It was very appreciated you signed up for
two sessions. I’d also like to echo Melissa’s statement
earlier today, that the statements that panelists submitted are really excellent and they are
very insightful, so I would encourage everyone to take a look at them. And I would also like to reiterate Melissa’s
earlier point, the disclaimer that I will be posing questions to the panelists that
the Bureau developed in collaboration with them, and as a friendly reminder, the views
of our panelists today are their views, and they are greatly appreciated and welcome. Yet the questions and the panelist statements
do not necessarily represent the views of the Bureau. So let’s get started. Generally speaking, behavioral economics focuses
primarily on exploring the existence of biases, in contrast with behavioral law and economics,
which is prescriptive in nature and aims to improve consumer outcomes and make consumers
better off when mitigating or ameliorating these biases. So the first question is, since many BLE prescriptions
doubt the usefulness of revealed preference, how can we know if consumers are better off? And I will ask Jan to start that. I think that’s a really tough question. The whole field of consumer protection economics
is really underdeveloped, and really understanding even what the right question is to ask about
whether or not consumers are injured by a particular practice is something that is not
well defined. So you can say, okay, consumers are injured. Is it by opportunity cost? Is it by expectations? How do you define when consumers are harmed
by a particular business practice? So I think that even at a theoretical level
there is a lot of work that needs to be done to try to understand the terms of the debate
better than we do now. The whole field of consumer protection economics
does not even have a handbook, like a handbook of consumer protection economics. We don’t have guidelines for how should we
even frame up the problems of how to think about consumer injury. So I would say that it’s a field that needs
a lot more work. Greg, would you like to— I think it’s very
difficult to assess situations. So, for example, suppose a consumer has a
car repair and he pays for it, and later in the month he can’t make his rent payment. Is the cause of the problem the auto repair
or is it rent? So suppose, then, he doesn’t pay off a credit
card bill or has to take out a payday loan or incur a credit card balance that he hadn’t
planned on. Is that optimism bias or is it just uncertainty? And, furthermore, in the future, some other
problem may occur. An appliance may break down and so he finds
himself having to continue to pay a balance or renew a payday loan. It’s awfully difficult to say whether someone
is better off or not. Others? This isn’t an issue unique to consumer financial
protection issues. I think we could make the same arguments in
a whole variety of domains that haven’t been a deterrent to policymaking. I think we could ask the same questions in
the health domain. If we require children to be vaccinated, are
we making them better off or worse off? And there is definitely a segment of parents
out there who think that vaccination is not a good thing for their children, and there
is a whole other segment of the population who thinks that most of the recommended childhood
vaccines are probably a good thing, and there is no shortage of medical interventions for
which we think we have pretty good evidence that there’s benefit. But there’s also no shortage of examples of
things that, 20 years ago, we thought were pretty good prescriptive advice and have been
overturned by subsequent research, so that we now think something else is the better
course of action, and 20 or 30 years from now we might discover that our current state
of knowledge is also not the best course of action. And I think this kind of relates back to some
of the discussion in the previous panel. How do you use the evidence? You know, we try to do the best that we can,
and you give more weight to evidence that’s been replicated and corroborated and holds
up across different domains, holds up over time, holds up across different populations
and different contexts. But sometimes you’re in the difficult position
of having to make policy on the basis of inadequate and incomplete information. But just because information is inadequate
and incomplete doesn’t mean that not doing something is the right course of action. Not doing something is, in fact, doing something. It’s just a different manner of doing. So there really is no neutral and no default
when it comes to policymaking. Whatever you do or don’t do will have an impact
on outcomes, and, unfortunately, policymakers are in a difficult position of having to decide
what is the outcome they are going to support, and their actions will either support one
outcome or another. I’d like to echo that. Back in the old days when we were taught public
policy there was a definition. I think it was Thomas Dye, and I’m going to
paraphrase it. It’s something like public policy is everything
that government chooses to do and what it chooses not to do. And I think that’s always important to keep
in mind when we’re weighing the options that we have for public policy. Status quo may sometimes be the best option. Maybe it’s not. I’d also like to go back to the notion of
how to figure out whether or not consumers are harmed. In the concept of deception, there is a framework
that we have in mind that at least I use, and many of my colleagues. They ask the question, number one, is there
evidence that consumers were misled? So that often requires getting copies of test
evidence to understand how consumers interpreted a marketing claim. Then the second question is, if you compare
what people did when they were misled versus when they had non-deceptive information, would
they change by buying more of the product or pay a higher price for it? So the framework of basic Economics 101 or
300, it’s very helpful in at least framing up and setting up the problem. I would make two points. One, I would return to the comment I made
earlier that I think this is very hard to do, to gauge the harm and the benefit. But even when we’re doing it with maybe suboptimal
tools we have, we have to think about the subgroups within the population in the U.S.,
and who is impacted in what way. And then the second thing I would say is,
there are opportunities for us to think about this not just in consumer harm devoid of some
other option. Greg made the point about, well, if someone
takes a payday loan to repair their car, you don’t really know if they didn’t take the
payday loan, did they lose their job because they couldn’t get to work? But you can figure out a way to facilitate
credit, so that benefit happens, and then the question is how do you have the product
structured in a way to reduce the chance of the harm on the other side, which we presume
would be them defaulting or damaging their credit score, something like that. And there are ways to think about how you
structure that product so people are more likely to be able to repay. Now some of that might have to do with the
economic benefit that the business gets or doesn’t get, but I think we can look at situations
where the business still gets a profit. The measure is like how large is that profit,
you know, this tradeoff, between the consumer benefit versus the business benefit. So you can get a product that’s better off
for the consumer, and it might not necessarily be truly better off for the business, and
that’s the rules of the road. And sometimes there are win-wins. Like you can criticize the CARD Act, but there
is a fair bit of evidence that shows that the CARD Act caused people to, in totality,
pay less for credit, and that there’s more transparency up front about what is the ultimate
rate they’re going to pay. And it doesn’t seem to have massively restrained
credit either. So there can be win-wins, and I think it’s
finding those win-wins and then doing the analysis to make sure that it’s playing out
properly. You’re going to use the evidence, make a decision,
and then the market is going to react, and there’s going to be unintended consequences. And then I think the Bureau often does this
very well. They have to go back and see, years later,
how has the regulation played out, and then adjust from it. I think also you have to remember that consumers
are heterogeneous, and in one situation where the consumer really does need a short-term
loan, a payday loan is cheaper than an installment loan. In other situations where the consumer needs
more money or needs to finance it over a longer period of time, then an installment loan makes
more sense than a payday loan. Simply saying that payday loans are bad because
the APR is higher really doesn’t address the issue. Great. Thanks. Okay, so the next question is, many BLE interventions
are consistent with the nudge approach, which attempts to shift consumers towards a particular
welfare enhancing outcome. Nudges are often implemented through changes
to the decision environment through choice architecture, which Brigitte talked a little
bit about earlier. Are there any findings from the choice architecture
literature that might help policymakers better understand consumers’ financial decisions
in the context in which many of these decisions are made? Jan, we’ll start with you. We know from the literature that things that
change people’s perceptions or change costs matter. So if you decrease the cost of decision-making
you’re going to probably get a different decision. We talked about the defaults before. We know that defaults matter. Changing out of default takes times. It takes effort. You have to think about it. So I think that there is a whole range of
policy tools that one can use, and the question is trying to marry the right tool to the right
objective. And I think the real question is, what is
the objective? I think much of the discussion thus far has
been about changing behavior, and then the question is how do you know which direction
you want to change behavior in? I don’t think that’s always the objective
of a regulatory agency. You know, it might be to improve the information
environment to make sure that it’s non-deceptive. Where I have worked, at the Federal Trade
Commission, when we are looking at deception issues, the notion that you want to change
behavior in a particular way is really quite different. Now if you want to change behavior, ban a
product. There’s this long list of information remedies. There is an information remedies task force
convened at the Federal Trade Commission. They issued a report I think back in 1979,
and the idea was to rank different remedies that you could use for different problems
in the marketplace dealing with market failure problems, with a focus on the market failure
of having asymmetric information available to consumers. And the rank ordering would go from status
quo to provide metrics to make labeling comparisons easier for firms to convey to consumers, have
like a standard metric. You could have mandatory labels. Later on you might want to add to that, you
can educate consumers, the notion that educating consumers requires more of a value judgment
than trying to get people just the information. And they can go to banning claims about particular
product characteristics or banning product characteristics themselves. So if the real objective is to change behavior,
you can change behavior by banning the product. But if the idea is you want to preserve choice,
then you might want to do something like provide people more information or more education. You have to be really clear on what the goal
is and what the tradeoffs are. Can I jump in with a couple of thoughts? First of all, I’m going to start off with
something that might be a little bit controversial, or might get me in hot water down the road. I have a tremendous amount of respect for
Dick Thaler and Cass Sunstein, who have both been colleagues of mine over the years, for
many years, and they are both, I would say, good professional friends. But I find the term “nudge” to be largely
unhelpful when it comes to thinking about public policy decisions. And it goes back to the very first thing I
said in my opening statement before the first panel, which is that the goal of all public
policy is to change behavior. So it’s not that we have public policy tools
that are not changing behavior and then we have nudges, which are designed to change
behavior. All public policy, whether it’s behaviorally
informed or not, has the goal of trying to change behavior. And I think what we have to do is what Jan
was just saying, which is you have to take a stand on what outcome you want. And a lot of times I think what it boils down
to is how much heterogeneity do you think there is in terms of what the best outcome
is for the population. Jan mentioned if you don’t want something
to happen you can ban it, or if you want it to happen you can mandate it. When are those appropriate public policy tools? Well, a mandate is probably appropriate if
it’s bad outcome for everyone. Did I say mandate? If it’s a good outcome for everyone then you
want to mandate it. I think I got things backwards, right? Yeah, I think you did. If it’s a good outcome for everyone, then
you mandate it. You just say this is a good outcome for everyone,
and if everyone knows it’s a good outcome no one is going to think the mandate is controversial. If it’s a bad outcome for everyone, then you
prohibit it. And then there are these other things where
you’ve got a range in the middle, where, yeah, the payday loan might be good for some people
but it’s bad for others. And then you’re in the difficult position
of having to figure out how do you design policy so that you make the outcome an option
for the people for whom it’s appropriate and you discourage its use for the people for
whom it’s not a particularly good outcome. And there are a range of tools that might
help you in doing a better job of achieving that type of heterogeneous outcomes, and some
of those tools might be what we would call behaviorally informed, and some of those tools
might not. So price tools tend to be a pretty blunt instrument,
and we know that disclosure works better for some individuals than others. In general, I would say many disclosures don’t
work well for anyone but the people for whom a traditional disclosure works well it would
be people who are highly motivated to read a large amount of small print, and that’s
probably a narrow subset of the population. So then the question is, how do you design
a set of tools that get you to the outcome that you want? When I talk about savings, which is where
my expertise is, what types of tools would you use to generate policy outcomes, I tell
my students to think about it as a continuum. On the one hand you have, we mandate that
everyone has to save. You could think of the Social Security system
as that type of a mandate. On the other end, we prohibit. We don’t really do that in the savings domain
but we could if we thought it was bad. And then there are different tools that are
going to get you along this continuum. Automatic enrollment gets you pretty close
to the outcome that you get if you were to mandate that everyone has to do it. Providing a financial incentive gets you pretty
close to the outcome where you just build it and they will come, and then there are
other tools that kind of get you in the middle. Active choosing gets you not quite to an automatic
enrollment but it gets you pretty close. Simplifying the process by reducing options,
creating simple forms, gets you kind of somewhere in the middle. So I think you have to think about what the
outcome you think is right, what’s the outcome you want—that may or may not be the outcome
that you think is right; I’m not saying that should be the case but sometimes people make
compromises—and then what’s the best tool to get you there? And any tool that does that is a behaviorally
informed tool, whether it’s what would be categorized as a nudge or whether it’s a traditional
tool like regulation or fines and subsidies or something like that. Can I just comment? But I think it is also the case you may not
always want to change behavior, right? So it depends on whether or not you think
that you know better than the consumer about what they ought to do, and that might be more
obvious in some cases than in others. Because going back to public policy sometimes,
the right decision is not intervening. Right, but I think you have to recognize that
if you’re deciding that the right outcome is the status quo, there is some set of policies
that somebody else put in place that got you to the status quo. Let me just give a concrete and specific example. Some people would say, “Well, if I’m changing
the default option, that’s using a nudge. That’s a nudge to change behavior.” And I would push back, and I would say, “Well,
somebody already made the choice for what the existing default is. So the current policy that you have is a nudge
towards a particular outcome, and the new policy that you have is a nudge towards a
different outcome.” So it’s not the case that the status quo is
devoid of any nudging to get us where we are. The status quo is very much the result of
a system in which policy decisions were made that moved outcomes to a direction which is
the status quo. And then we can quibble about do we think
that’s the right outcome or not, but it’s not that you didn’t use behaviorally informed
policy to get there. You might not have realized you were using
it, but the policies that got you there were impacting an outcome that we can then have
a discussion about. I anticipated the next question and I want
to bring in Josh and Greg, if they’d like to chime in. Are nudges, as a matter of policy, an appropriate
tool for the Bureau to use? What should policymakers consider before implementing
a policy decision with a nudge component? And do nudges require determination of the
value of a particular ends-oriented goal and a comparison of what would happen in the alternative? I think, to just build on what Brigitte says,
it depends for two reasons. One, to Brigitte’s point about what’s your
objective of the policy, but it also depends on what the context of the product and market
that you’re talking about. I also don’t love the term “nudge” because
I think it boxes behavioral science and behavioral economics into some small set of things that
are about notices and forms and framing. So, to me, it depends, and then, yes, as much
as possible you should be looking at the totality of the evidence. Here I think there is some judgment involved. Sometimes you’ll say, oh, we think the evidence
is pretty strong that would point us in this direction. Other times you might say, well, the evidence
isn’t there, and the CFPB has a research department, and part of their job is to develop questions
that need to be answered, and the best possible answer to those questions, within the regulation
or the nudges, can be more informed by that. Yeah, that depends, and to me it’s about a
process you pursue, and it’s a process that’s imperfect in terms of information, and there
will be judgment involved. All right. Thank you. We’ve had a lot of references to how BLE has
been used in policymaking, and I want to kind of do a sort of deep drill down into some
concrete examples. If we take an example or two of one of these
BLE interventions and start from like what was the evidence that was used to justify
it? How was it implemented and what did we learn
from that intervention? Josh, I will go back to you and see if you
can give us some examples. Sure. I mean, Brigitte’s work in retirement savings
I think is probably the most well-known example. The research started 20 years ago. You know, people talk about randomized controlled
tests all the time. That original research is in a randomized
controlled test, right? It’s a before and after, and it’s turned on
and off in a number of different companies. But the effect size is so large that it’s
pretty hard to argue against the fact. And then we had to figure out, okay, how do
we actually get that implemented, and the Pension Protection Act was—I can’t remember
if it was the first iteration or being redone. And then they thought about, well, how do
we get companies to do this? They didn’t require them to do it, right? They created a set of incentives, again, some
behavioral informed policy about safe harbors and what you got running a pension plan, if
you implemented things like defaults and even then percentages of automatic contributions. So I think that’s one example. I think another example is in the financial
aid space. So there is a fair bit of research by academics
that the difficulty and the hassle and the problem of filing out the financial aid form
cause people who, in any kind of traditional economic analysis, would say there’s tens
of thousands of dollars on the table in terms of aid, grants, and loans, as well as a much
higher income if you complete college. People aren’t filling out this form that could
take 4 hours. They actually prepopulated the form, and then
not only did people fill out the form at a much higher rate, they actually matriculated
to college at a much higher rate. I think it’s 40 percentage points and 20 percentage
points. So that’s a second example. I think a third example, I would argue even
you could take the CARD Act, to me, as an example of behavioral informed policy. There was concern—and I think here there’s
a situation where the evidence was not as rigorous—there was a concern that some of
these what we call back-end fees, like the idea that you could get a credit card, and
then on that balance they could change the rate on you later on, not on new balances
but on the existing balance. It suggests everything in behavioral science
to all the things that Brigitte said before about complex decisions, uncertainty, things
that are in the future. We’re going to have trouble with that decision. So what did the CARD Act do? It disallowed that one particular practice,
not all back-end fees, and then you see the result, and it’s a positive impact. So I think, in each of these cases there is
some set of research that’s happened already, and some of it can be very concrete. They’ve tested the exact thing, like in the
retirement case or the FAFSA case. In other cases you’re trying to make a leap,
and you should be careful when you make those leaps. People, I think, have been reasonable to argue
that maybe there should have been more research done on the CARD Act before they made that
leap, to see if you could test that out. It’s a little bit harder to test exactly. So those are just some examples. I have other ones but I don’t want to take
up too much time. I could talk all day about the examples. Can I add a little bit more color to the savings
research that Josh described, just because I’ve been in the center of it, and I think
it’s a really interesting story. When I wrote the first paper on the impact
of automatic enrollment in 401(k) savings plans, which was about 20 years ago, the initial
reaction, at the University of Chicago, which interestingly is both the bastion of the traditional
economic models but also Dick Thaler is there, it’s also kind of the ground zero for behavioral
economics. It was an interesting place to be doing that
research. I had a number of colleagues who were extremely
skeptical that this finding would hold up. They were convinced there was some other explanation
that I wasn’t looking at that was generating this dramatic result. And the policymaking in response to that research
started out with a set of rulemakings in the U.S., in the Treasury Department. So it was kind of undercover, stealth regulation. And as that was being done, more evidence
started accumulating that, you know what, these findings are actually pretty robust,
and they were replicating at other companies. And that then led to the Pension Protection
Act of 2006, which, as Josh described, encouraged employers to use automatic enrollment but
didn’t require it. And the UK, a few years later, went one step
further, and they decided, as part of a national pension reform, that they were going to require
employers to use automatic enrollment, and that has been rolled out between 2012 and
2017, and they’ve seen nationally something like a 30 percentage point increase in the
fraction of people who are saving for retirement. And then in the U.S., this is now playing
out at the state level, with a number of states now adopting UK-style legislation, requiring
employers to use automatic enrollment. An interesting difference between the UK and
the state initiatives in the U.S., at least, is that in the UK they started out with a
very low default contribution rate of 1 percent, because they were afraid that consumers would
opt out, and then they are increasing that contribution rate over time, and that’s going
to give us a piece of case-based evidence in the UK. The three states in the U.S. that currently
have something similar operating, which would be California, Illinois, and Oregon—and
these are all pretty recent, meaning we barely have 2 years of data, in some cases, and others
much less—have opted to go with a much higher default contribution rate of 5 percent. So they’re getting more money into the plans
quickly, but they are having a higher opt-out rate than in the UK. So it’s an interesting space because we’ve
seen different approaches to using the evidence and policymaking, in terms of the political
jurisdiction, in terms of making it mandatory versus not, and in terms of the risks that
policymakers are willing to take in implementing it. I think another interesting piece of using
behavioral research to think about policymaking in the savings domain has been on the asset
allocation side. So another piece of the Pension Protection
Act was legislation that required regulators to come up with a set of rules around what
would be a good default investment option. And that was in response to a number of papers
in Behavioral Finance suggesting that individuals really, the vast majority, don’t do a very
good job in thinking about investment allocation. They don’t even understand the options, let
alone how those options would map into the outcomes that they want. And so the regulations were put in place to
set up a default option, or encourage employers to have a default option that was well diversified. And what we’ve seen in the U.S. since that
time is that almost all plans have adopted a default that’s consistent with these qualified
default investment alternative rules. And now the vast majority of money is flowing
into these default funds, that is to say participants, even when they have a choice, stick with the
default. So one response by firms has been we adopt
a default consistent with the regulations. Another response by firms, not necessarily
to the QDIA rules but just to behavioral research in general, has been a dramatic shrinkage
in the number of investment options on the fund menus in the U.S. So when I started as a professor at Harvard
in 2006, amongst the investment alternatives in the plans that Harvard sponsored there
were about 350 different options to choose from. And a few years ago Harvard dramatically pared
back the number of options, and that’s been happening all over the country, on the basis
of behavioral research. If you look, in contrast, to what’s been going
on in Sweden, Sweden, as part of their Social Security system, has a private account component,
so 2 percent of everyone’s pay goes into this private account. It’s a mandate; you can’t get out of it. And the Swedish government has taken the view
that as long as you satisfy some fairly minimal regulatory constraints you can be an option
on the fund menu. And they now have more than 1,000 options
on the menu. When they first started they had about 400
or 500, and initially about 80 percent of participants were at the default option and
about 20 percent were making their own choice. And as the number of options has proliferated,
they are now down to about 1 percent of people making a choice out of these 1,000 different
investment options. And facing this difficult conversation around,
well, does it make sense to have this whole regulatory infrastructure around making sure
that the investment alternatives are good, when hardly anyone is actually choosing any
of them—does this even make any sense? I’ve kind of tipped my hand on how I think
about it. Yeah, we can argue it’s nice to have consumer
choice, but if 99 percent of people are choosing the same thing, you might think differently
about the value of choice than if people are actually exercising those options. I actually think there are a lot of fascinating
things that we learn, both about the nature of human decision-making and about the process
by which research gets turned into public policy, and about the cycle between research
and public policy, and the research gets used to make policy, the policy changes, then inform
a next generation of research, and then the next generation of research feeds into the
next generation of public policy. I would add, also, that the Bureau can take
a role in saying like these are the questions that would be helpful to have answered. They don’t even, I think, need to put money
on the table to do the research, but there’s a set of researchers who are particularly
interested in doing things that matter in the real world, and so I think being more
explicit about the questions that you’re struggling with, as a Bureau, and getting researchers
to work on those could be a powerful interplay. Okay. Well, speaking of the Bureau, disclosures
are an important tool for us. However, in many cases, mandated disclosures
are often, by necessity, a one-size-fits-all approach to information provision. And as has been discussed several times this
morning, consumers may want and will make use of different pieces of information when
trying to make the decision they think is right for their personal situation. Consumers are heterogeneous. So, Jan, what are some examples of disclosures,
and what has the experience with them been? I have personal experience with two sets of
disclosures. One is mortgage disclosures and one is energy
labels. So, mortgage disclosures. As a staff economist at the FTC, I was noticing,
as well as my colleague, that consumers could get every document required under the law
and still not understand the terms of their loans. So in the back of my head I had this question,
when I looked at the forms myself, saying, gee, are these forms really helpful to people? Do people really understand the terms in these
forms? So we did a research project, a randomized
controlled study. Actually, we did two studies at the FTC. The first study was in response to proposed
changes by HUD, and they wanted to change their disclosures to how people focus on yield
spread premiums. This is a mortgage broken compensation disclosure. And as staff economist I was assigned to look
at this regulation and say, well, what do you think about that and what are the costs
and benefits? And I looked at the disclosure and I said,
“Can you tell me again what the form looks like if I get the mortgage from a banker versus
I get the mortgage from a mortgage broker?” And the forms would have been different. So we posed the question, in a comment to
HUD, that we were not sure that these forms were going to be helpful to people, because
it might bias their understanding of which loan costs more or less, and it might lead
them to worse loan choices. So we went back to our management and said,
gee, you know, we’d like to test this. And borrowing the techniques that we learned
from our marketing research colleagues—you know, randomized controlled testing of information—we
were incredibly surprised to find how much the proposed disclosures misled people about
the relative costs of loans from one source versus another. So that was one study. Then we had these back-of-our-mind questions
about these mandated disclosures that were in the marketplace, the good faith estimate
and the HUD one, and so we asked our management if we could do research on that, and we got
some funding to do, once again, some research. This is a two-part research study. The first part was doing qualitative research,
and we found consumers who had recently bought houses and had mortgages. We identified, for our contractor, people
who had loans that were in the sub-prime category and in the prime category, and the contractor
was able to identify consumers to participate in the qualitative research. And they were asked to bring their disclosure
documents with them to the research meeting. And then they were asked a series of questions
about their experiences in procuring their homes, in procuring their mortgages. What was really interesting was that people
tended to think that they understood the terms of their mortgages. They felt generally pretty confident, until
we started asking—well, it wasn’t us. It was actually our contractors who asked
questions. And the more people were asked questions,
the more uncomfortable they became about whether or not they really knew what the deal was. And even in this small qualitative research
we could see that people were actually misled by required terms on the documents. Is a discount fee a discount or a fee? So we did randomized controlled testing and
we did a study where we looked at what we thought were good examples of mandated disclosures
in the marketplace. And we compared those to prototype disclosures
that were developed by two economists, myself and my colleague, Jim Lacko, asking the question,
what would you want in the form if you were telling your best friend or your mother how
to shop for a mortgage? And I’m sure the forms could have been better
if we had professional help in developing them, but nevertheless we tested our form,
the prototype, against what was in the marketplace, and what did we find? Our form did much, much better, and it was
because there was no testing that went into the development of these forms that the Federal
Government had mandated for many, many years. So it gets to this point that disclosures
themselves, although you think that they could be benign, they’re not necessarily benign. And it’s really important to do consumer research
to make sure the consumers at least comprehend the disclosures correctly. Now comprehension might change behavior, it
might not change behavior, but the question, I think, for policymakers is actually an ethical
one. If you’re going to put out a disclosure, do
you have a responsibility to make sure that you’re not going to be misleading people? So I put that out there as a question. Now we also did research on appliance and
energy labels. That was sort of a fun project. Congress had suggested that the FTC move from
a continuous label, where kilowatt hours was featured, on those yellow labels—you know,
those yellow and black labels that you see on the refrigerator? And the question was whether or not we should
move from a continuous variable, like kilowatt hours, to a categorical label, something like
a star label. And an energy group had tested various categorical
labels, so we decided to use their version as the one label to test for categorial. We tested that against kilowatt hours being
featured. And then we said, while we’re at it, consumers
understand dollars, so maybe we should test dollars as a featured metric as well. Long story short, controlled quantitative
testing, people did pretty well rank ordering which appliance cost more or less, no matter
which form you used, but here’s the thing. People were confused and potentially misled
by stars, because when people saw a star it seemed that they thought that it pertained
more to a statement about the energy characteristic of the product and could say something about
how reliable the product was. So I think in all these efforts to try to
simplify things for people, it’s really important to think, you know, are we sure that we’re
bringing them closer to the truth when we’re trying to do our interventions. One of the things that caused a great deal
of difficulty and debate when the board was doing rulemaking in truth in lending was dealing
with disclosures that required some assumption about future events. For example, the mortgage disclosure, you
basically assumed that the person stays in the house for 30 years, doesn’t refinance,
and that’s not the case for, by far, most mortgages. So the board sought a way to inform consumers
about tradeoffs between fees and interest rates, and they worked with a market research
firm, they conducted focus groups, and tried various disclosures involving APRs, to get
people to understand what’s involved. And most people did miserably. One consumer, who answered virtually every
other question about the mortgage disclosure, understood everything, but when he got to
this question he also didn’t have a clue. So we have to think about how we can make
a disclosure that is meaningful to a consumer. Now we were predisposed towards APR. APR is an internal rate of return. If you read every finance book it will tell
you that you need to use an internal rate of return or net present value. There is also a way to evaluate things called
a payback period. Every finance book tells you you shouldn’t
use that, and we didn’t consider it. But that’s a simple heuristic. People can understand how long it takes to
pay back with your savings and your monthly payments, an initial fee. Did we do it? No. Another example is we require a number of
months to repay credit cards. That assumes that there is no increase in
the balances and that people make a minimum payment each period. The board conducted surveys of consumers,
asking them about their payment behavior on credit cards. Virtually nobody did that. We also obtained data from credit card companies. We looked at the actual transactions. Nobody did it. So the question is, is that useful, and what’s
more concerning is whether that might mislead people, because that’s not what they do. So I think we need to understand consumers’
decision processes, their understanding of information available in the market, just
what Jan said, and do considerable testing, and we have to be willing to consider rules
that make sense to consumers, are not likely to mislead them, and enable them to, in many
cases, substitute what they expect their circumstances will be and calculate values that are more
meaningful to them. Josh, I was going to ask you, are there new
approaches to disclosure or information provision that can allow for the tailoring of disclosures
to consumer heterogeneity? I had two comments on this. One is first around this idea of sophisticants
versus non-sophisticants. David Laibson and Antoinette Schoar—and
I cite one of Antoinette’s papers in my comments—have started to look at this issue related to disclosure
and credit cards and this idea of price shrouding. So there is some sense that—and I think
this is an area that needs more research before anyone would make a regulation around it—there
is heterogeneity but are there big segments that you could actually think about acting
the same, related to disclosure? And then, again, how do you care about those
segments reacting? There is some research to show that unsophisticants,
which is often measured by education level, have derived much less benefit from disclosure. The second one is this idea of smart disclosure,
which is not something I came up with but is out there in the policy world, that I still
think needs more development. But it’s this idea that humans make a bunch
of decisions, they buy products, they interact with their credit card, their cell phone,
basing anything that they’re buying. And there’s data now, in most cases, around
that, that consumers should be allowed to have access to that data in machine-readable
format, and be able to give decision-making algorithms and companies access to that data
and then the companies can help them actually make the decision that’s best for them, based
on their historic performance. It’s not perfect. I think there are lots of questions about
how do you regulate the people who get access to that data and the privacy issues that Mike
was talking about before, but it seems like an area that is rich for opportunity and the
technology will allow us to derive some benefit from over time. But we should proceed cautiously. Can I add something? I think individualized disclosures and decision
recommender tools, like John has been working on, are great. I think this also raises the question of what
segment of the marketplace or of the whole universe should maybe focus on what tools. Behavioral economics tells us a lot. There are some things that I might want to
use as an individual consumer, like I might want to sign up for a service that reminds
me to do A, B, or C. You know, you might want a service that reminds you that you should
take your medication. You might want to have recommender tools that
you sign up for by yourself, which is a very different question from saying at what point
does the government provide those tools. But I think the notion of individualized tools
is really very exciting, and it’s very important, and I think it has a great future. We’re almost out of time and I wanted to offer
a final chance for some last bits of wisdom. Last bits of wisdom. My last bit of wisdom, I’ll reiterate what
I said before, every policy tool is about changing behavior. There is nothing unique about behaviorally
informed tools. They’re just another way to change behavior. And why would you not want to understand all
of the tools in the toolkit and use the ones that are going to best deliver the outcome
at the lowest cost that you’re trying to achieve? We’ll leave it at that. Thank you very much. Thanks to the panelists for the excellent
discussion. Well, thank you, Jason, and thanks to our
panelists. Again, great discussion, and it’s been a great
day, so thank you. At this time it is my honor to introduce the
Bureau’s Deputy Director, Brian Johnson. Deputy Director Johnson first joined the Bureau
in December of 2017 as senior advisor to the acting Director, and served as acting Deputy
Director from July 2018 until May 2019, when he was announced as the Bureau’s Deputy Director. Mr. Johnson joined the Bureau from the House
Financial Services Committee where he spent over 5 years serving in various capacities,
including senior counsel. Mr. Johnson received his BA in economics as
well as his JD from the University of Virginia. It is my privilege to welcome him to the podium. Deputy Director Johnson, you have the floor. Thank you, Andrew. This has been a fantastic morning, and now
afternoon, and I recognize I’m coming in at the end of this, so I don’t intend to be here
long. Thank you all for your careful thoughts, for
your insightful papers. Brigitte, thank you especially for pulling
double duty. We had a last-minute withdrawal, and based
on your expertise we are very happy to have you twice. We also thought we were going to have double
Josh Wrights. He also had a family emergency and pulled
out, so we didn’t get to see one Josh debate the other. Maybe in the future we will have that. Thank you all for coming here in person. Thank you to everybody who tuned in. I would be remiss if I didn’t thank all of
our staff who did an excellent job pulling this together. These things don’t come together on their
own, and for those of you who have done faculty panels like this I know you appreciate that
as well. So thank you for all the effort that our team
put in. Our symposium today is part of a broader effort
on the part of the Bureau to engage experts in various fields on legal and policy issues. An important motivation for this symposia
series is to give these legal and policy issues a fresh look, and that’s precisely what we’re
going to do. The Bureau has relied on behavioral law and
economics, in the past and part of the fresh look I just mentioned includes examining how
behavioral law and economics should play a role in Bureau policymaking going forward. As a matter of public policy, the director’s
remarks this morning, including her observations about the value of focusing on market failure,
captured the essence of what good government looks like. When articulating and implementing public
policy there is general agreement that the primary motivation for regulation is to address
market failure, and it can’t be stressed enough that demonstrating a market failure requires
more than just general notions of incomplete or asymmetric information. Justifying intervention in such circumstances
should require clear showing that the agency can improve on the status quo and utilize
symmetrical treatment of both market and government action, that is, a recognition that markets
have costs that may impede desirable outcomes but government action faces similar, if not
greater, impediments. In prior remarks, I have elaborated on the
role I believe the Bureau should play when crafting consumer protection policy. There, I emphasized that rather than attempting
to replace market outcomes, the Bureau should focus on regulatory and policy initiatives
that reinforce market institutions. I also emphasized my view that the Bureau’s
guiding principles should be a presumption in favor of consumer choice. During the more than 200 years of American
history, ordinary consumers making choices in the marketplace has resulted in them realizing
direct and extraordinary benefits that have dramatically improved their lives. Market-reinforcing activity includes promoting
competition and prosecuting unlawful acts or practices that impede or undermine consumer
choice. Now some may think that agencies like the
Bureau should engage in more prescriptive market-replacing regulation. In academic settings, some have argued that
financial goods and services are like toasters, requiring agencies like the Bureau to impose
heavy-ended, prophylactic measures to try and protect consumers. I believe this view is mistaken. Today’s symposium, along with the written
statements, is helping me evaluate how behavioral law and economics fits into the framework
of articulated. To add to today’s discussion, I want to offer
a few observations that I hope can move the debate forward and prompt future research. First, regulatory proposals should avoid presenting
the choice as between a purportedly ideal government solution and an imperfect market
solution, what economist Harold Demsetz referred to as the nirvana fallacy. Indeed, humans, by our nature, are imperfect
so it should come as no shock to anyone that markets are imperfect. We should also not forget that governments
are made up of humans, and—this might be shocking—we are all equally imperfect, and
perhaps have a few of our biases. So in an imperfect world, we need to critically
examine both market and governmental solutions to identify which is best for consumers. Second, though the debate we have had today
often relates back to the concept of rationality, no serious thinker on this issue would suggest
that humans act with perfect information or that transaction costs in the market are zero,
nor do we expect humans to always choose the efficient solution. The relevant policy issues relate to how consumers
adapt to mistakes, whether they can learn from their errors within market settings,
and if intervention would create a net beneficial outcome. Third, when contemplating a particular intervention,
the market failure and the potential problem that is being analyzed should be clearly defined
and demarcated. Doing so will enable the agency to tailor
a solution to the actual problem. For example, one focused on what consumers
actually harmed without imposing an unnecessarily broad remedy that could undermine the effective
functioning of the overall market and potentially undermine consumer welfare. Finally, any analysis done regarding the behavioral
law and economics must include a discussion about the nature of the interventions being
considered. Much of this was discussed today. Behavioral economics can potentially inform
market-reinforcing regulation, but we should tread very carefully where proposed policies
would restrict or change the nature of a private relationship or contract. When government follows this course, it does
not politely request the firms change their behavior or seek to convince firms of the
propriety of the intervention. It is a legal command, backed by force or
threat of force. While there are many examples in the literature
that relate to private or non-government interventions in the marketplace, those examples importantly
do not involve commands from an agency. On this score, 200 years before much of the
modern debate on behavioral law and economics, Adam Smith—and this wouldn’t be my remarks
if I didn’t quote Adam Smith in some way—had many insights into essentially many of the
issues that we are discussing today. In his Theory of Moral Sentiments, he used
social psychology to better understand the nature of human behavior. Smith observed that we desire mutual sympathy,
and because of this desire, Smith suggested that order can emerge through decentralized
processes. Smith was also careful to note the important
difference between what we call today private and government paternalism. Because of local knowledge and feedback that’s
generated through private or voluntary iterative interactions, Smith saw generous benefits
to private tinkering, what we might today call framing or choice architecture. On the other hand, he was less enthusiastic
about centralized control. His thoughts here were best highlighted by
what he called the “man of system.” Smith wrote that “the man of system seems
to imagine that he can arrange the different members of a great society with as much ease
as the hand arranges different pieces upon a chess-board. He has not considered that the pieces have
no other principle of motion besides that which the hand impresses upon them.” He further observed that “when the individual’s
principle of motion diverges from the central authority’s preference, such instances have
negative consequences on the individual, and society, more broadly.” Though Smith incorporated many of today’s
modern behavioral methods into his analysis, he also noted that “the benefits of such methods
were highly dependent on whether they were implemented by a centralized authority or
through decentralized processes.” His insights, in my view, highlight the importance
of Bureau action that is market-reinforcing and promoting of competition and consumer
choice. The issues highlighted by today’s symposium
and all of the excellent debate and back-and-forth and insightful questions by our moderators,
have far-reaching implications on the nature of good government and human agency. Today’s symposium has helped move forward
that conversation. For that I want to thank, again, the panelists
for all of your contributions today. Thank you for coming. And thank you all for attending.

Leave a Reply

Your email address will not be published. Required fields are marked *