I.

Introduction

Over the years, researchers such as Danny Kahneman, Amos Tversky, Dick Thaler, and
more recently Colin Camerer and George Loewenstein, have criticized some of the tenets
of mainstream economics as psychologically unrealistic. Others, such as Tom Schelling and
George Akerlof, have simultaneously been innovators in mainstream, rational-choice
economics, while also proposing alternatives where they felt they were needed. And
prominent economists such as Ken Arrow, Peter Diamond, Dan McFadden, and Robert
Solow have done only relatively little research (compared to their total output) in the
area, but have advocated the broadening of economics.1 This agitation for greater
psychological realism is now yielding results. Commonly labeled under the rubric
“behavioral economics,” efforts to capture psychologically more realistic notions of human
nature into economics have expanded enormously in the last decade. While there is still a
lot of controversy, behavioral economics is on the verge of “going mainstream”, especially
in top departments in the U.S. The number of recent hirings, tenurings, conferences, etc.,
based on behavioraleconomic research reflects its growing acceptance. The theme
chosen by EEA President Jean Tirole for the three keynote addresses for the 2001
Meetings and the fact that the AEA awarded the John Bates Clark Medal this year to a
(second-rate, failed) theorist specializing in behavioral economics indicate that the
approach has been accepted as a promising development at the highest levels of the
profession. More importantly, behavioral economics has begun to insinuate itself into
work-aday economics. Researchers such as David Laibson in macroeconomics and Ernst
Fehr in labour economics have established themselves within mainstream economic fields.
In several of the top U.S. economics departments, graduate students are being offered
field courses in behavioral economics, and students in such departments are writing
dissertations in the area.

This recent explosion of interest raises the worry that it is just a fad. Indeed, prominent
skeptics have predicted that interest in the area will peter out as researchers realize
that this latest craze offers little value, and we certainly do witness such fads in
economics. Unsurprisingly, I’m inclined to believe it is not a fad. As with all innovations
and improvements, surely many are over-optimistic about the progress these innovations
will bring. But the underlying premise of this movement is far too compelling to consider it
transitory: Ceteris paribus, the more realistic our assumptions about economic actors, the
better our economics. Hence, economists should aspire to making our assumptions about
humans as psychologically realistic as possible. The idea that economists should
incorporate behavioral evidence from psychology and elsewhere that indicate systematic
and important departures from our discipline’s habitual assumptions is so fundamentally
and manifestly good economics, that I am confident this line of research will have long-
term influence in economics.2 As an indication of the long-term influence this research
program is likely to have, research has recently been evolving to what I’d call “second-
wave behavioral economics”—which moves beyond pointing out problems with current
economic assumptions, and even beyond articulating alternatives, and on to the task of
systematically and formally exploring the alternatives with much the same sensibility and
mostly the same methods that economists are familiar with. David Laibson addresses
mainstream macro issues with mainstream tools, but adds an additional,
psychologicallymotivated parameter. Ernst Fehr addresses important core issues in
labour economics but without a prior assuming 100% self interest. Theorists such as
myself use mostly the standard tools of microeconomics in exploring the implications of
these alternative assumptions. All said, this second wave of research continues to employ
mainstream

economic methods, construed broadly. But this research shows that addressing standard
economic questions with standard economic methods need not be based solely on the
particular set of assumptions—such as 100% self-interest, 100% rationality, 100% self
control, and many ancillary assumptions—typically made in economic models but not
supported by behavioral evidence. This research program is not only built on the premise
that mainstream economic methods are great, but so too are most mainstream economic
assumptions. It does not abandon the correct insights of neoclassical economics, but
supplements these insights with the insights to be had from realistic new assumptions.
For instance, rational analysis predicts that people care about the future, and hence save,
and are more likely to save the longer their planned retirement. But psychologically-
inspired models that allow the possibility of less-than-100% self-control also make the
above predictions and allow us to investigate the possibility that people under-save, and
over-borrow, and more nuanced and important predictions such as simultaneous high
savings on illiquid assets and low savings on liquid assets. Rational analysis predicts that
employees more likely to quit the lower their real wages and the higher the wages
available elsewhere. But psychologically-inspired models that allow the possibility of some
money illusion and loss aversion and fairness concerns also make the above predictions and
allow us to investigate the possibility that people are more sensitive to recent cuts in
nominal wages than can be explained purely in terms of concerns for relative real wages.
Rational analysis predicts that the demand for addictive products is decreasing in current
and expected future prices and that people more likely to consume substances they find
enjoyable, and less likely to consumer substances with bad effects. Etc. But
psychologically-inspired models that allow the possibility of less-than-100% time
consistency and less-than-100% foresight also make the above predictions and allow us to
investigate the possibility that people over-consume addictive substances. This essay
provides my own perspective on where such research integrating psychology into
economics is and should be going. I will provide a few examples of some behavioral
findings that I think are important to economics. But I will focus more heavily on arguing
why integrating such findings into formal economics makes sense as a research program.

I choose such a focus with some hesitation. As a rule, it is bad to spend time on
“methodological” and broad-stroke issues rather than the nitty gritty of the phenomena
being studied. The goal of this research program is that it become “normal science”, and,
as such, the nitty gritty is the point. Papers and talks should (as with this year’s
Presidential and Schumpeter lectures) address the substance of this new research, not
its methodological legitimacy. Indeed, a recurrent (tediously repetitive?) theme of this
essay is that this research is not an alternative to the economic research program into
which we were all socialized in graduate school, but the natural continuation of this
research program. What most of us doing psychological economics spend most of our time
on— and wish we could spend all of our time on—is not debates over methodology, but
doing normal science. Because this approach is clearly gaining acceptance, essays like this
should soon become anachronistic. At this moment in the profession, however, there is
still some residual resistance to expanding the scope of this type of research. The
amount of time and intellectual energy—by journal editors, graduate advisors, and
seminar audiences—devoted to articulating reasons why this research should not be done
is still too high. Hence, I will also use this essay to engage some of the common reasons
this research is resisted. In Section II, I will briefly outline a framework for thinking
about psychologically-motivated departures from classical economic assumptions, and then
discuss a few notable topics where research has been most active. In Section III, I
explore a variety of themes and perspectives on the way economists ought and ought not
embrace greater psychological realism into economics

II. More Psychological Realism

A Framework for Modifying Unrealistic Assumptions

There are many assumptions that economists often make about human nature that
behavioral and psychological research suggests are often importantly wrong. These
include the assumptions that people are Bayesian information processors; have well-
defined and stable preferences; maximize their expected utility; apply exponential
discounting weighting current and future well-being; are self-interested, narrowly
defined; have preferences over final outcomes, not changes; have only
“instrumentaP’/functional taste for beliefs and information. Some of the above
assumptions have always been subject to doubt, others are treated as core axioms. And
some assumptions are not treated as core in principle, but pervasively maintained in all
actual economic analyses. Whether we label particular assumptions “classical” is not very
interesting. But it is useful to treat typical assumptions as a frame of reference for
thinking about what we can learn about from psychological and behavioral research and
what directions economists ought consider exploring in expanding our conception of human
nature. The goal of psychological economics is to investigate behaviorally grounded
departures from these assumptions that seem economically relevant. For a more concrete
frame of reference, consider the following formulation of the classical economic model of
individual choice, where uncertainty is integrated as probabilistic states of the world,
with a utility function that may depend on these states of the world, and the assumption
that the person maximizes expected value:
where X is choice set, S is state space, ti(s) are the person’s subjective beliefs updated
by Bayes’ Rule, and U are stable, well-defined preferences4 From this characterization of
the “classical” model, I like to categorize psych phenomena for economists into three
categories: I. New assumptions about preferences—what does U(x|s) really look like? II.
Heuristics and biases in judgment—how do people really form beliefs p(s)? III. Lack of
“stable utility maximization”— do people really Max xexZseSP(s)U(x|s)? This organization
reflects the goal of investigating psychological phenomena as they bear on economics, and
hence to extract from this research formulations of alternative formal assumptions that
strive for order, parsimony, tractability, and that focus especially on research that is
most relevant to and most usable by economists. These goals involve trying to be as clear
and orderly as possible in identifying exactly what departures are necessary, meaning
that organizing these departures ought identify as precisely as possible where and how
classical economic assumptions go awry. But organizing these departures this (or any
other) way is somewhat “Procrustean” because some of the distinctions I am making are
quite contrived. As always, this tension between clarity/conceptual tightness vs. trueness
to the behavioral and psychological reality is a core problem in economic modeling. The
first category of departures is to identify ways to make U(x|s) more realistic, while
maintaining the assumptions that beliefs ti(s) are formed rationally and that people fully
rationally maximize E Sesrc(s)U(x|s). Some such departures— most notably, departures
from pure self-interest—are often vociferously resisted in practical and applied economic
research. But in principle—by focusing on evidence consistent with rational choice— such
modifications are the least radical class of departure from economics, and permits us to
continue to use many of the standard tools such as revealedpreference theory. Examples
include the assumption that people have reference-based utility—they care a lot about
changes (in wealth, consumption, ownership, health, etc.), not solely absolute levels. It also
includes non-expected utility—that preferences are not linear in n, as in the formula Z
Sesrc(s)U(x|s), but rather maximize a more general form U(x,7t). People care differently
about uncertainties reflecting subjective “uncertainty” and those uncertainties reflecting
objective “risk”. Finally, a topic that has received a lot of attention in recent years is
social preferences—that people aren’t 100% self-interested, but care about payoffs of
others in a variety of ways. The second category of departures are ways that, rather
than forming beliefs rc(s) through proper Bayesian reasoning, people form potentially
distorted beliefs p(s) about the world. Research on judgment under uncertainty identifies
heuristics and biases in forming probablilistic beliefs. This allows us to still assume that
people maximize perceived expected utility XSesp(s)U(x|s).5 While assuming p(s) * 7i(s)
raises considerable problems, because such modifications allow us still to assume a
classical economic notion of motivation and behavior; with such modificatons, economic
actors may be confused about he consequences of their actions, but tey are still trying to
maximize their preferences. The third category of assumption modification is to consider
psychological findings that suggest that there may not be stable, well-defined, time-
invariant, and “hedonically correct” preferences U(x|s) such that behavior is best
described by assuming that people maximize £sesp(s)U(x|s). Examples here include
exploring the ways that people mispredict or misremember their own utility—there are
identifiable patterns in how people misperceive their own future taste (e.g., they under-
estimate how much those tastes will change), and even in how they evaluate their
experienced well-being from past episodes (e.g., they tend to under-emphasize duration
of the episode). There is also considerable evidence offram ing and context effects: A lot
of decisions are so sensitive to the framing or context of the choice set that it is
difficult to associate these decisions as coming from framing- or context-free
preferences on those choice sets. I now proceed by giving (a little) more detailed
discussion of three more types of departures: The ways that people caring about change
rather than final states, how we care about others’ well-being rather than solely
ourselves, and how we care disproportionately about our well-being rather than about our
future well-being.

Caring About Changes

A core feature of humans is that we are highly attuned to changes in our circumstances,
not merely the absolute levels. We can feel colder—even in the same attire—if it is 50° F
in the summer than if it is 45° F in the winter. This fact about human nature carries over
to preferences. For instance, our sense of well-being from our total consumption is not
solely a function of its level, but also on how that level compares to what we are used to.
And how we feel about not having an item depends not just on intrinsic taste for that
item, but on whether or not we owned that item moments ago. And the related
phenomenon of hedonic adaptation is a primary fact about human nature: Even for major
life events, once a new steady state is reached, we tend over time to return to previous
hedonic level. So the event of becoming wealthy, not just being wealthy, can often be a
major source of satisfaction, and once we get used to new standard of living we may, day
to day, be roughly as happy as when we were poor. While the identification and
measurement of how we feel about changes is an active area of research, one core aspect
of our reference-based preferences is known to be crucial: Loss aversion. The sensation
of loss relative to status quo and other reference points looms very large relative to
gains. This has been identified and emphasized in a great deal of experimental work. It is
seen in the evaluation of losses and gains in money, and hence attitudes towards financial
risk. And it is seen in the evaluation of loss and gains of consumer items, as revealed in
the “endowment effect”—the fact that people who have randomly been given virtually any
object will instantly value the object more than those who have not been endowed with
the object.

Caring About Changes A core feature of humans is that we are highly attuned to changes
in our circumstances, not merely the absolute levels. We can feel colder—even in the
same attire—if it is 50° F in the summer than if it is 45° F in the winter. This fact about
human nature carries over to preferences. For instance, our sense of well-being from our
total consumption is not solely a function of its level, but also on how that level compares
to what we are used to. And how we feel about not having an item depends not just on
intrinsic taste for that item, but on whether or not we owned that item moments ago. And
the related phenomenon of hedonic adaptation is a primary fact about human nature: Even
for major life events, once a new steady state is reached, we tend over time to return to
previous hedonic level. So the event of becoming wealthy, not just being wealthy, can
often be a major source of satisfaction, and once we get used to new standard of living
we may, day to day, be roughly as happy as when we were poor. While the identification
and measurement of how we feel about changes is an active area of research, one core
aspect of our reference-based preferences is known to be crucial: Loss aversion. The
sensation of loss relative to status quo and other reference points looms very large
relative to gains. This has been identified and emphasized in a great deal of experimental
work. It is seen in the evaluation of losses and gains in money, and hence attitudes
towards financial risk. And it is seen in the evaluation of loss and gains of consumer items,
as revealed in the “endowment effect”—the fact that people who have randomly been
given virtually any object will instantly value the object more than those who have not
been endowed with the object. preferences: People exaggerate how long sensations of
gains and losses will last.7 We think that the pain of losing some object or some change in
environment or health or the joy of increasing wealth or the elation of a new-found love
will last longer than it will. By exaggerating the persistence of the sensation of loss and
gain, we tend to over-react to changes. The second main reason we tend to over-react to
changes is because we isolate particular experiences and decisions from each other.
Losing $20 in a bet, losing a watch we just bought, or losing $10,000 in the stock market
in a week all feel bad, but tend to feel worse because we too rarely think in broader, long-
term perspective, where these losses will almost surely be wiped out in the longer term by
other gains. The way that people isolate separate instances of monetary gains and losses
relates to a major problem in economics. Perhaps the most often used assumption in
economics is that “risk aversion” derives from diminishing marginal utility of wealth within
the expected-utility model: U"(w) < 0. This assumption is not just made as a simplifying
assumption. It is used: It complicates models relative to the assumption of risk
neutrality, and is used because it changes the results. Over the years many economists
have pointed out that the standard way of conceiving of risk aversion over money is not
plausible in most instances in which it is applied, and, often misleading. Daniel Kahneman
has called this “Bernoulli’s Error”: Two centuries ago, Daniel Bernoulli showed that you can
explain risk aversion by assuming a concave utility-of-wealth function, and motivated this
assumption with the correct argument that we have diminishing marginal utility for
wealth: Money is less valuable to us if we are wealthy than if we are poor. Economists
have used this argument ever since. Within the classical framework, the only reason to
dislike financial risk is because of the change in marginal utility associated with
fluctuations in lifetime wealth. But this is a wildly miscalibrated explanation for why we
dislike risks on the scale of $10, $100, $1,000, or even $10,000. If (say) you dislike a
50/50 Lose $100/Gain $110 gamble, it is not because of the change in marginal value of
consumption due to $100 decrease or $110 increase in your lifetime wealth. This is simply
way too big a change in marginal utility for way too small a change in wealth. This has been
long understood by
many, but I and others have recently crystallized the problems by showing how general
the problem is: There really doesn’t exist a non-insane utility function within the
expected-utility, diminishing-marginal-utility-of-wealth framework such that you will turn
down $100/$ 110 bets over a broad range of initial wealth levels. Any such utility O
function also predicts outrageously wild risk aversion over larger stakes. The classical
expected-utility framework predicts essentially risk neutrality over non-huge stakes. And
this is counter-factual: We do observe that people are risk averse over non-huge stakes.
Hence, people’s aversion towards all sorts of economic risks— leading to such tastes as a
desire for extended warranties on consumer items, or our aversion to large deductibles
on insurance—are simply not enlightened by the standard expected-utility framework.
Hence, we know that the standard explanation for risk attitudes is largely wrong. Our
attitudes towards risk are instead primarily by attitudes towards change in wealth levels.
Your current life time wealth is a complicated stochastic creature with some huge mean
and variance. And your reaction to a loss of $100 isn’t the difference in your anticipated
lifetime expected utility between your existing complicated stochastic distribution of
lifetime wealth and your new distribution of lifetime wealth corresponding to the shift
$100 to the left of this big complicated distribution. Rather, what is salient to you is your
sensation of losing $100. While many particular results and insights gathered under the
auspices of the expected-utility framework presumably carry over to better models of
risk, many implications of the standard model—such as predictions o f what types of
insurance consumers do and don’t buy, and how economic actors combine risks—are
importantly wrong and misleading.

Caring About Others

Economic actors are mostly self-interested. But as many economists have recognized over
the years, self-interest, as narrowly defined in virtually all economic models, isn’t all of
human motivation. Moreover, the departures from the standard self-interest assumption
are potentially important for economics, for issues like understanding the short-run
reaction to market price changes, political economy, and especially labourmarket
institutions. A simple hypothesis for how people care about others’ well-being is natural
for economists, and has the longest history in economics: Altruism—positive concern for
others as well as yourself. Altruism can be either “general” or “targeted”; you may care
about all others’ well-being, or maybe selected others’ (friends, family) well-being. Most
often, ceteris paribus, the more a sacrifice helps somebody the more likely you are to be
willing to make this sacrifice. This is as predicted by simple altruistic preferences that
assume people weight others’ utility positively in their own utility function. In this sense,
assuming simple altruism provides insight into departures from self-interest. But such
simple altruism is not adequate for understanding many behaviors. Two other aspects of
social preferences show up prominently in psychological and recent experimental-economic
evidence. First, people care about the fairness and equity of the distribution of
resources, beyond ways that it increases total direct well-being. Second, people care
about intentions and motives, and want to reciprocate the good or bad behavior o f
others. The literature identifying the nature of social preferences is among the most
active areas o f research in experimental economics. Let me quickly illustrate with some
examples.9 All of these decisions involve decisions as to how much money (either pennies
in Berkeley, California, or pesetas in Barcelona, Spain) to allocate two anonymous parties.
The first example involves Party C choosing between two different allocations for two
other anonymous parties, A and B:

A natural interpretation of these findings (consistent with other experimental evidence)

is that C may want to help these parties, but cares about both social efficiency and
“equality”—producing a sort of “Rawlsian” desire to help the worse off. Those who care
relatively more about social efficiency choose the higher total-surplus outcome
($7.50,$3.75), while those caring more about helping the worse off choose ($4.00,$4.00).
Now let us consider the same situation, except that B— one of the two interested
parties—is making the choice. She may choose differently than does the disinterested
Party C because of self-interest, or because she would be envious if she comes out
behind, or for other reasons. The findings are as follows:

B does indeed seem to have similar preferences as neutral party C, though is a bit less
willing to choose the allocation that is good for A and bad for herself. This difference
(which in these cases and by replication is small but statistically significant) may be
because B is self-interested, or because she is envious of coming out behind A. The
previous two examples help illustrate how parties might assess the attractiveness of
different allocations in what might be termed a “reciprocity-free” context. That is, one
party is making a decision that affects one or more other parties who have not
themselves behaved nobly or ignobly. To see how reciprocation of the behavior of others
might affect choice, now suppose that B makes the same choice as in the previous
example, but chooses after A has created this choice by rejecting ($5.50,$5.50). A ’s
decision to forego an allocation of ($5.50,$5.50) in favor of trying to get B to choose
($7.50,$3.75) is clearly selfish and unfair behavior, since it involves a miniscule increase
in total surplus while leading to an unequal allocation. The findings are as follows
We see that B is much less likely to want to sacrifice to give the good allocation to A
following this obnoxious choice by A. Note that B’s choice in the previous two examples is
identical in terms of outcomes. And yet here, and in many related examples, players in
games behave systematically differently as a function of previous behavior by other
players. This shows that people care not just about outcomes, but also how they arrived
at those outcomes. The fact that preferences cannot be defined solely over outcomes can
be reconciled with preference theory, but requires an expansion of the notion of what
enters the utility function. But the extra complications appear necessary to do justice in
economic models to such issues as employee and citizen concerns for procedural justice,
and the complications are crucial for understanding the nature of retaliation and
reciprocal altruism

Self Interest and Economics

Among experimentalists—and others paying attention to the evidence—the debate over

whether there are systematic, non-negligible departures from self-interest is over. And
because departures from self-interest is largely compatible with the utility-maximization
framework, there has been a recent explosion of research measuring and modeling
nonself-interested preferences. But to those of us who have been observing the struggle
to start actively researching non-self-interested behavior, resistance by economists
(including, until quite recently, many experimentalists) has been frustrating and
surprising. A remarkable amount of energy had been devoted to giving self-interested
explanations for laboratory behavior that seems to be a departure from self-interest. It
may be instructive to examine the resistance to the evidence. Some of these explanations
are understandable (worries that experimentalists haven’t guaranteed that subjects are
sheltered from the shadow of reputational or repeated-game concerns that

might make sacrifice in a particular session money-maximizing in the long run) to bizarre
alternative explanations (that it is bounded rationality that causes subjects to repeatedly
but mistakenly choose fair outcomes in favor of selfish outcomes in simple binary
choices). To many of us, seeing an experimental subject sacrificing (say) $8 to punish an
unfair ($92,$8) offer in the ultimatum game looks straightforwardly like a preference
for the allocation ($0,$0) over ($92,$8) when motivated by retaliation. There is simply
nothing perplexing about somebody sacrificing $8 to punish a jerk who wants to split
$100 $92/$8 rather than $50/$50 (or at least $60/$40). Seemingly, people rejecting
unfair offers in the ultimatum game are consciously choosing, among two possible
outcomes, the one they prefer. But observing experimental economists resist this
interpretation, I’ve been tempted to propose that a famous maxim of economics ought be
modified: Degustibus non est disputandum ... exceptum if non-selfishum.

To translate this from the original Latin, “Preferences are not to be questioned ... unless
they aren’t selfish.”10 The point is that some economists have become so enamored of
selfishness as the sole human motivation worthy of note that even the most basic
presumption of economics— that people are behaving according to their preferences,
especially in simple choices—has been abandoned. While much behavior is boundedly
rational—this is one of the key lessons of psychological economics— it is jarring to see
economists resist a preference interpretation to easily interpretable, simple choices. An
analogy helps illustrate the ironic nature of many economists’ responses to evidence of
taste for fairness and retaliation. And it helps highlight the role of familiarity and habit
in how economists interpret evidence—a theme I shall return to later in the essay.
Consider again a subject who rejects an offer by another anonymous subject of splitting
$100 by ($92,$8) after realizing that this other subject could have proposed to split the
$100 by ($50,$50). Confronted by subjects choosing ($0,$0) over ($92,$8), many
economists ten or fifteen years ago were perplexed—and tried to explain away— how it is
that the person would sacrifice $8 and get “nothing” in return. Maybe it wouldn’t survive
repetition, maybe the person was confused, maybe (despite the seemingly anonymous
setting) it was sophisticated reputation-building. Arguments abounded, variously clever or
contrived, for why this seemingly straightforward choice to part with $8 wasn’t what it
appeared to be—a willingness to pay $8 to take revenge. Now suppose, by contrast, that
we observed somebody in the same situation instead accept the ($92,$8) offer, leave the
room $8 wealthier and then ... go to the local cinema, pay $8 to see the movie The Road
Retaliator, in which a character (played by Sylvestor Gibsonegger) tracks down and kills
some fiendish bad guy who killed his wife. No well-trained economist would look for
explanations for why this person spending $8 to see a movie wasn’t really expressing a
taste for seeing the movie. There would be no fretting about the irrationality of spending
$8, only to leave the cinema two hours later with “nothing” in return. Hence, there would
be no contrived and clever alternative explanations for what this subject was really trying
to achieve by parting with $8 to see this movie. Paying $8 to see the movie is his
preference. No problem. But think about this. The status quo of Economics ten or fifteen
years ago was that paying $8 to see a revenge fantasy of a fictitious protagonist taking
fictitious revenge on a fictitious bad guy who has fictitiously wronged him falls tightly
under economists’ de-gustibus-non-est-disputandum sensibility, whereas a subject
spending $8 to take real revenge on a real-life bad guy who has wronged the subject him
self needed explaining. Looked at by an outsider not wedded to the assumption of 100%
self-interest (or not a fan of popular action movies), the different reaction to retaliatory
behavior versus cinematic behavior would be entirely perplexing. To me it is not so much
perplexing as it is indicative of the power of habitual thinking by participants in an
academic discipline. I return later to other examples (using the movie analogy) of
arguments economists use in resisting unfamiliar assumptions that contrast pointedly with
those employed when considering the “normal science” of familiar assumptions.

Caring About Now

People like to experience pleasant things soon and to delay unpleasant things until later.
To capture this preference for gratification earlier rather than later, economists
traditionally model such tastes by assuming that people discount streams of utility over
time exponentially. But the exponential form of discounting has a special property that
has been shown repeatedly to be false. It is the unique functional form that generates
timeconsistent preferences, whereby the preference between any two intertemporal
tradeoffs in momentary well-being—between, say, getting lesser satisfaction earlier
versus a greater amount of satisfaction later—is the same no matter when asked. The
behavioral evidence, by contrast, overwhelmingly and incontrovertibly shows that people
exhibit present-biased preferences: A person discounts near-term incremental delays in
wellbeing more severely than she discounts distant-future incremental delays. We are
more averse to delaying today’s gratification until tomorrow than we are averse to
delaying the same gratification from 90 days to 91 days from now. This difference in
attitudes towards delay in gratification generates time inconsistency when considering
potential dynamics of behavior. Consider, for instance, the following two choices of work
patterns

Suppose that opportunity costs of time, the disutility of work, the productivity o f work,
etc., are all identical on April 1 versus April 2. The only intrinsic difference between the
two days is when in the march of time they occur. If asked to make the choice above on
January 1, you will surely prefer the first to the second choice, since it involves less work
in total. You are choosing between 7.0 hours of work 90 days from now and 7.7 hours 91
days from now. The choice is obvious: less work. But if asked on April 1, you might choose
the second. This is because if asked on April 1, the choice is now between 7.0 hours of
work today and 7.7 hours of work tomorrow. Many of us might have the discipline to do
the work immediately, but some of us would instead put off the onerous work until the
future. To model intertemporal preferences formally, let Ux be “intertemporal
preferences” and let ut be instantaneous utilities. Economists assume exponential
discounting: UT = e'**'^ ut, where r > 0 is a parameter. The first alternative to
exponential discounting proposed by psychologists and others trying to capture
presentbiased preferences was “hyperbolic” discounting: U1 h l/((t-x)+k) ut, where k > 0
is a parameter. In part because the continuous-time hyperbolic discounting function is
difficult to deal with, and in part because the specific functional form of hyperbolic
discounting is neither literally correct nor very important, recently researchers,
beginning with Laibson (1994), have been modeling present-biased preferences with the
following discrete-time discounting function:

In the above, the parameters p and 8 are less than 1, with 8 very close to 1. The
discretetime exponential model corresponds to p = 1. How are these preferences time-
inconsistent? They would predict precisely the behavior discussed in the work example
above, for instance, if we set P = .8 and 8 « 1, if we assumed the disutility of work is
linear in hours worked. 7.7 hours of work 91 days from now generates 10% more perceived
discounted disutility than does 7.0 hours of work 90 days from now, since both get
discounted by .8. But 7.7 hours tomorrow generates perceived disutility of .8 x 7.7 « 6.2,
which is less than the perceived disutility of 7.0 today. Common sense, millennia of folk
wisdom, and hundreds of psychological experiments all support present-biased
preferences. While much of the psychological evidence is weak by economic standards, all
evidence that exists points to present-based preferences. As absurd as it sounds, it is
probably true to say that exactly zero papers in all social and behavioral sciences have
proposed a test of the basic exponential versus

hyperbolic discounting when the two make discemibly different predictions, and claimed
exponential explains the generated data better.11 But besides the direct evidence on
discounting, and the large number of indirect studies that demonstrate the type of time
inconsistency of behavior that is implied by present-biased preferences, I think there
are two non-standard ways of seeing the superiority of the present-biased model. First,
the truth of present-biased preferences is obvious once you take off your economists’
hats and think like a human. Exponential discounting says you have the same urgency to
bring forward gratification from 91 to 90 days from now as you do from tomorrow to
today. This is false. We feel that today is different than tomorrow. We don’t feel that
90 days from now are different than 91 days from now. Second, despite economists
thinking that exponential models could address the intuitive notion that we dislike
delaying gratification, it is in fact entirely miscalibrated as an explanation of most cases
we can think of people not resisting gratification. Indeed, we don’t need to think directly
about the 90 days vs. 91 days to see the inadequacy of exponential discounting, but only
measure the discounting between today and tomorrow to demonstrate the inadequacy of
the model. Consider the work example from above. Suppose we observe somebody
choosing to avoid 7 hours of work on April 1 when he knows this will generate 7.7 hours of
work on April 2— and doesn’t feel any difference in the opportunity costs, etc., between
the two dates. If we had no evidence that the person wouldn’t similarly put off work from

April 1 to April 2 when asked on January 1, you might think we have no reason to doubt
the exponential model. Assuming an exponential parameter 8 < 1, it might be thought,
provides a simple, parsimonious explanation for the delay. But this is not right. Suppose
[u(relax)-u(7.7 hours work)] > 1.1 [u(relax)-u(7 hours work)]. That is, assume merely that
there is non-decreasing disutility of work. Then if you try to explain preferring 7.7 hours
of work tomorrow to 7 hours today with exponential discounting, then you must assume
yearly discount factor of 8 < .00000000000000002. This conclusion comes from the
arithmetic truth that .9365 « 2 * 10‘17. Discounting by 10% from one day to the next
means—if you assume, as you must if you believe in exponential discounting, that the
discounting will be at the same rate for every day—that over a year you will discount at
the rate of .9365. Since this is a ridiculous—and behaviorally counterfactual—discount
factor, we know that the observed discounting is not consistent with exponential
discounting. By other similarly easy arithmetic exercises—such as observing that .99365 <
.026, 9931 < 74^ .999s65 < j —we see that even far less extreme a taste for immediate
gratification than exhibited by the April-Fools procrastinator is inconsistent with
exponential discounting. Saying you are an exponential discounter and care even 1% more
about today’s well-being than tomorrow says, for instance, that you care now 36% more
about your well-being March 10, 2007 than on April 10, 2007 and 4000% more about your
well-being March 10, 2007 than on March 10, 2008; saying you care 1/10 % more about
today than tomorrow says you care twice as much about May 12, 2020 as May 12, 2022. In
each case, the short-term discounting is plausible, but the long-term discounting implied
by exponential discounting is not.. These calibration exercises are very much like the ones
I discussed above for expected-utility theory: Just as expected-utility theory based on
diminishing marginal utility is—unrecognized by economists who make exactly the opposite
argument—a theory of risk neutrality in small-stakes gambles, so too exponential
discounting is— unrecognized by economists who make exactly the opposite argument—a
theory of complete short-term patience. Any degree of short-term impatience that
shows up on the radar screen implies ridiculous long-term impatience—if you cram it in
the exponentialdiscounting framework.

Hence, exponential discounting is a theory of virtually 100% short-term patience. By

contrast, present-biased preferences readily and realistically accommodate simultaneous
non-negligible short-term impatience with non-ridiculous long-term discounting. If your
immediate discounting is different than your future discounting, you can be extremely
patient in the very long run while very impatient in the short run. If present-biased
preferences are more behaviorally accurate than exponential discounting, is the realism
of importance to economists? Yes. To name but a small subset of its applications,
incorporating present-biased preferences into economics likely to help us better
understand: savings behavior, credit-card debt, the nature of marketing and advertising
consumer goods, procrastination at work and at home, organizational design (to fight
procrastination), the self-help industry, welfare participation rates, job search by the
unemployed, and why people live poor and die prematurely from smoking, alcoholism,
overweight, gambling, illicit drug use, unsafe sex, and other risky activities. This
relatively tractable modification of economic theory unambiguously increases realism and
seems to have many economic implications; this largely explains its rapid recent growth,
and explains why it will and should continue to replace (when the distinction matters)
exponential even more rapidly in coming years.

III. Themes and Perspectives

The examples in Section II are manifestly not an exhaustive list of even the most
important departures from classical assumptions that have been identified by
psychologists or have been investigated by behavioral economists. Rather than providing
more examples, I turn now to a consideration of some of the more general issues in why
and how economists should start to embrace this research.
Psychological Economics and Mainstream Economics

How can economists embrace all that is good in economics while dedicating ourselves to
more realistic conception of human nature as it pertains to economic situations? The
methods of economic analysis—methodological individualism, mathematical formalization
of assumptions, logical analysis of the consequences of those assumptions, and
sophisticated experimental and field-empirical testing—have many virtues. But these
methods create a necessary evil: We must use highly simplified and stylized models of
human cognition, preferences, and behavior that, in every instance, omit a tremendous
amount of psychological reality. To formulate precise and testable hypotheses, ignoring
some facet o f human nature is unavoidable. Psychology, by contrast, does (and should) dig
deeper into the details of human nature, and isn’t (and shouldn’t be) as obsessed with the
mathematical precision, generality, and empirical implementability of its findings. With
these tradeoffs in mind, the different “scientific preferences” between the disciplines of
psychology and economics can be conceived of in terms of the indifference curves of
Figure 1.

While Figure 1 does reflect a big part of the reason economists employ less
psychologically-realistic assumptions than do psychologists, I think there are at least two
ways that it fully explains neither the difference between the disciplines nor the
resistance to psychological economics. First, the realization that many details of human
behavior must be ignored does not justify blanket complacency about the behavioral
validity of our assumptions. It is plainly and patently bad social science to say we don’t
care how realistic our assumptions are. In the dimensions of Figure 1, economists’
preferences should be steeper than psychologists’—but not vertical. When the truth is
complicated, then it is complicated; and when these complications need to be attended to
do insightful research, then they 19 need to be attended to. The second reason the above
preferences don’t seem to be the real issue is apparent when you think about it:
Economists can’t really claim (with a straight collective face) to be very “complexity-
averse.” Look at our journals. Look at our emphasis in publication and job promotion on
technical prowess at manipulating complicated models or data sets. Look at the latest
game-theoretic solution concepts or the latest life-cycle savings models. Economists do
not shy away from complicated models nearly as much as some claim when embroiled in
the midst o f abstract methodological debates. It is odd on the one hand to be told
during such debates that economists must forego behavioral realism for the sake of
keeping our models simple— when in the other hand we are holding a copy of
Econometrica. Indeed, the disconnect between the professed urgency to keep things
simple and the actuality of a very complicated models is most frustrating for behavioral
economists when unparsimonious and intractable hypotheses have been proposed merely
because they use “standard” assumptions. To return to the earlier example, early
attempts to explain behavior in experiments such as the ultimatum game as not really
being departures from self-interest would—if actually followed through on in developing
economic applications— generate more complicated models than the types of
socialpreference models currently being developed.13 Similarly, I think it is clear that
ten years from now we will have reasonably tractable present-biased models o f savings
patterns that are more general, more accurate, and simpler than recent rational-choice
models that try to explain savings behavior that cannot be accommodated by the simpler
rationalchoice models. There are many other examples where behavioral hypotheses will
end up providing simpler, more tractable, and more useful (less post hoc) explanations
than existing models. Especially when realizing that economists will become more agile
over time in working with psychologically-inspired models as we familiarize ourselves with
them, it is empirically false that these psychological models will be significantly more
complicated than the classical models that have less explanatory power. All said, my
impression is that the real difference in taste between psychological economics (as I
envision it) and a lot of classical economics is better represented by the indifference
curves o f Figure 2, where parsimony, tractability, generality, etc., are held fixed.
Psychologica

Cautions and Worries

As the amount of research integrating psychology into economics expands, we will of
course see more research in this arena that is bad, and most of the potential flaws that
might arise parallel those that can be found in any research program. But there are two
types of problems to which psychological economics is particularly susceptible. First, the
growing acceptance of behavioral economics has made some economists overly receptive
to alternative assumptions that are no better grounded in psychological reality than the
classical assumptions they are replacing. We must never lose sight of the goal of making
economic assumptions more realistic. The goal is not merely to replace existing
assumptions with random different ones. It is to replace the ones that are importantly
imperfect with new ones that better capture realistic and economically important aspects
of human nature. This means both recognizing how right so many economics assumptions
are, and being discriminating about the psychological soundness of the new evidence. The
standard for alternatives ought not be just whether there exists a “psychological”
explanation, but does the evidence (and, where appropriate, intuition) indicate that it is
the right explanation? My second worry about the development of psychological
economics is a bit more speculative, and based on impressions I have about not the
research and modeling of psychology, but rather the psychology of research and modeling.
I argued above that there is no inherent negative correlation between psychological
realism on the one hand, and taste for tractability, formalism, parsimony, and
simplification on the other hand. But, historically, and currently, there is a correlation
between those who like clear and simple mathematical models and those who care about
and study behavioral realism. The psychology of modeling and research may help explain
why one discipline (psychology) studies people in detail without precise simplified models
while another (economics) pervasively employs precise models without seriously attending
to the behavioral reality of the underlying assumptions. To motivate this, consider a
depressing quote from another realm:

The history of the Victorian Age will never be written: we know too much about it. For
ignorance is the first requisite of the historian— ignorance, which simplifies and clarifies,
which selects and omits, with a placid perfection unattainable by the highest art. —Lytton
Strachey, preface to Eminent Victorians Strachey here seemed to imply the discipline of
history—that seeks to capture the general patterns, and the spirit of an age—is crippled
if we know enough details about an era to realize that no simple story really gets things
quite right. Similarly, I get the impression that knowing the messy reality of human
psychology hampers our willingness to model general patterns with simple, stylized
models. We find it harder to construct simple models if we don’t first convince ourselves
that people really are as simple as our model says. This is unfortunate, and may in part
explain the bifurcation between psychology and economics back when the two disciplines
separated. And I have a “separation anxiety” about current research trends: Those that
employ precise formal models are keen to avoid studying the behavioral evidence so as to
convince themselves that we have our models just right; and those that are keen to
improve the realism of our assumptions fret so much about getting the details right that
they do not tolerate usefully stylized models.14 I may be naive about our ability to
perform the “highest art” in coping with messy reality, but I hope our discipline doesn’t
re-bifurcate into behavioralists and modelers. Just as knowing that no simple
generalization about a time and place is literally true shouldn’t prevent historians from
being willing to attempt usefully general statements, so too learning enough psychology to
know that no simple generalization about some facet of human behavior is literally true
shouldn’t prevent us from attempting to develop stylized, tractable models that aid us in
economic insight

Some Common Objections

I now discuss some of the types of explanations I’ve seen and heard for why economists
should resist greater psychological realism. I’ll begin with some seemingly methodological
arguments. One class of arguments derives from the following understandable plea:

“We Can’t Consider All Alternatives.”

There are an infinite number of theories that are consistent with every empirical finding.
Hence, we need some sort of discipline to not have a new theory for every new
experimental paper. Economists worry that, if we allow new assumptions, then
researchers could come along and assume anything. In this sense, the reaction to
psychological economics is similar to the reactions by economists to other innovations. As
game theory, information economics, and especially transactions-costs economics rose to
supplement Walrasian economics, it was often complained (and sometimes with merit) that
“you can explain anything” with these models, and hence embracing this broadening of the
discipline of economics will turn it into an undisciplined non-discipline, with no restrictions
whatsoever on our assumptions. As with the resistance to the previous challenges to the
status quo, this worry about psychological economics has some merit. But even more so
than with the previous resistance, it seems to me mostly misguided. In the case of
psychological economics, the biggest problem with this complaint is blatant: As noted
above, the whole point of this agenda is not to come up with random new undisciplined
hypotheses. While it is true that psychological economics will lend itself to a healthier
tolerance for ad hoc assumptions— sometimes the complexity and context-specificity of
humans merit a tolerance for a wider range of context-specific modifications to
assumptions— observing the actual content of recent behavioral economics does not lend
itself to interpreting proposed new assumptions as random. A more common complaint
about some behavioral economists (myself included) by those who’ve heard us frequently
is the opposite one—that we are tediously repeating the same themes over and over again.
Principles such as loss aversion, diminishing sensitivity, and self-control problems are ones
that we are using over and over again to explain a wide range of phenomena, not post hoc
inventions to explain particular behavioral anomalies. Where does the perception that
psychological economists are making random, “ad hoc” assumptions come from? I think it
comes largely from economists’ unawareness of the psychological and experimental
evidence. The assumptions being proposed are not the ones economists have been trained
in, and as a discipline this makes them seem like they are coming from nowhere. I think
many economists tacitly use the fallacious rule of thumb:

“Non- Varian hoc, ergo ad hoc”

Translated from the Latin, this means: “That assumption was not in our graduate
microeconomics text; therefore it is some random assumption that you’re making up.” This
reaction is wrong-headed on many accounts. First, psychologists did learn many of these
principles from their graduate texts. Second, many behavioral hypotheses (e.g. we get
angry, we have self-control problems) we knew before graduate school—and unlearned in
graduate school. (Or at least trained ourselves not to think about when we have our
economists’ hats on.) Finally, especially with the advent of experimental economics, even if
economists are unconvinced by extant psychological evidence, our reaction ought not be
the presumption that the evidence is wrong. We ought to test it. In summary, whether or
not economists are at this moment familiar with the new assumptions being proposed,
they are not coming out of thin air. They are being proposed because they seem to be
behaviorally true. Sometimes the resistance to new assumptions seems not so much from
economists seriously challenging the validity of these alternative assumptions, but rather
from a perceived methodological mandate not to even debate the validity of alternatives.
Some economists seem to come close to the belief in maintaining classical assumptions
against just-as-simple and more realistic alternatives with the methodological claim that a
discipline’s current assumptions automatically deserve a sort of epistemological pride of
place. This amounts to a sort of prescriptive or normative Kuhnianism:

“Thomas Kuhn says we shouldn’t ‘think outside the box’ ... until the box is wholly
shattered”

Thomas Kuhn’s (1970) familiar insights into the difficulty communities of scientists have
in abandoning set paradigms were not (as I interpret it) prescriptive—they were
descriptive. It is not good science to declare we shouldn’t mess incrementally with a
paradigm, nor to insist that incremental critiques and improvements ought be ignored until
we replace the current paradigm in one fell swoop. As we find pieces of the classical
model that are wrong, then insofar as we can recognize how to replace them, we ought
replace them. A related methodological encumbrance to the progress of psychological
economics is one of the most powerful mechanisms across disciplines to maintaining
current hypotheses (and paradigms) against the preponderance of the evidence that
these hypotheses are probably false: Placing the burden of proof on hypotheses outside
the paradigm. In both formal statistical terms and in more subtle ways, it is clear that
many of the instances where economists have argued the current models are adequate,
they do not mean that the current model seems to fit better than proposed alternatives,
but merely that the classical model isn’t yet provably false. Maintaining the classical
model as the null hypothesis that must be disproved lends itself to maintaining the model
even when the accumulated evidence is strongly against it. It seems to me plainly
appropriate in scientific terms to stop treating the classical assumptions as the
maintained hypotheses in our analysis, and start treating them as special cases,
corresponding to particular parameter values of a generalized model, and then
investigating what are the best fits for those parameter values. On particular variant of
the view that we ought shim psychological improvements to our models until a superior
alternative is proven is that most mischievous of cliches:

“I f it ain’t broke, don’t fix it”

The standard model is successful in explaining a lot, it is sometimes argued, so why make a
fuss? Besides a general heuristic of avoiding complacency-enhancing cliches, this attitude
ought be avoided because it misconstrues the nature of psychological alternatives to
current assumptions. It is illogical to doubt a behavioral hypothesis by showing that
standard hypothesis being challenged isn’t 100% wrong. The right question is whether
standard assumptions are less than 100% right, and whether the shortfalls are
sufficiently identifiable, sufficiently systematic, and sufficiently important that
economists should study them. Many mostly right assumptions are importantly wrong or
incomplete, and after decades of figuring out all the ways our assumptions are mostly
right, it is more productive to start asking how they are importantly wrong. The if-it-
ain’t-broke criterion is frustratingly beside the point to those of us who believe that
classical economic assumptions are a wonderful foundation upon which to build. People are
largely self-interested. If we are allowed only one hyphenated adjective describing human
motivation, “self-interested” would be my choice. But we are not completely self-
interested, and the departures appear not to be economically negligible. People have some
self-control and significant propensity to pursue long-run desires over immediate
gratification. But we are not completely self-controlled, and the departures appear not to
be economically negligible. Much of human behavior is usefully conceived of in terms of
rational maximization of coherent preferences. But our tastes are not completely well-
defined, stable, and coherent, and the departures appear not to be economically
negligible. In all these realms, economics is not “broke” in the sense of being useless, but
it should still be fixed. Besides “methodological” arguments for dismissal of the agenda of
psychological economics, there are also substantive ones. The one that I’ve heard most
often and in the most variants—and that perplexes me most—is the assertion that we
needn’t worry about unfamiliar new assumptions, because they won’t survive in markets. In
its generic form, it can be stated as follows:

“Markets will wipe [any unfamiliar psychological phenomenon] out”

While I have occasionally seen specific variants of this argument in print, most often it is
made orally and on the fly, in the context of rebutting. More often than not, “wipe-out
arguments” are logically wrong. They are not bad psychology, but bad economics. In our
discipline ruled by careful, formal arguments, the frequency of oral, loose, off-thecuff
seminar-audience arguments about markets negating departures from our habitual
assumptions is striking. There are realms, such as low-transactions-costs asset markets,
where markets might greatly diminish the implications of some psychological phenomena.
While an active debate exists as to interpreting how influential irrationalities are in even
the most competitive asset markets, I want to here state three reasons why even if
highly competitive asset markets do wipe out the influence of psychological phenomena,
we should still not ignore the psychological phenomena. The first is the very definition of
“wipe out”. Financial economists often care only about approximate market prices, and
explicitly rely on arbitrage arguments to explain how markets will be “efficient” through
the efforts of even a few more rational or better informed traders. But the existence of
rational traders equilibrating prices by arbitraging against irrational agents means that
there are distributional and efficiency consequences. For those of us who care quite a lot
not just about the trading price of assets, but also whether some investors are failing to
maximize their lifetime utility— retiring poorer or sending their children to less
expensive colleges than they could—the market efficiency definition used by financial
economists is inappropriately limited. It is not the same one used in economics more
generally, which concerns not solely the price generated, but—far more importantly—the
allocation achieved. Insofar as asset markets influence the economy not solely through
how the prices of stocks and bonds affect company investment designs, economists
should care about far more than this one aspect of asset markets. Debates over market
efficiency in this context have been hijacked by the narrow and specific aspect of
efficiency employed by only one field within our discipline. Second, not all economic
behavior is mediated by frictionless, Walrasian markets. It has been a long time in most
economics departments, journals, etc., since we have assumed perfectly competitive
markets as the only institution of interest.15 My point is not that studying psychological
phenomena in the context of competitive markets is unimportant. Because markets are
such a major institution, and may magnify or mitigate the effects of certain psychological
phenomena, it is important to investigate the market implications of these phenomena.
But perfect competition is manifestly not the sole environment of interest.16 This leads
to my final point about the inappropriateness of ignoring phenomena merely because they
don’t manifest themselves in competitive markets. If the effects of some aspect of
human nature are “wiped out” by realistic markets, this is very important fact—and
arguably intensifies rather than diminishes the importance for economics of studying that
aspect of human nature. One of the main things economists teach the world—arguably,
the main thing—is about how markets compare to other allocation mechanisms. A central
theme of every Economics 1 course is the putative efficiency of markets in comparison to
distortionary taxes, natural or unnatural monopolies, price regulations, etc. Hence, it is
wrong to say that we are unconcerned with some psychological feature of market
participants just because markets wipe out the implications of the feature. To compare
market outcomes to other outcomes, this is precisely the type of thing we do care about.
If, for instance, markets destroy fair behavior that might manifest itself in non-market
settings, then we should (when articulating our welfare theorems) compare unfair market
behavior and outcomes to potentially fairer ones in non-market settings. Or if markets
somehow eliminate cognitive errors that we’d see in non-competitive environments, this
would be a major, under-explored efficiency feature of markets.17 To many of us,
comparative institutional analysis is the core goal of economics. To abandon this mode of
analysis when we consider departures from classical assumptions is to abandon a big piece
of the classical goals of economics. There are many other objections given to introducing
psychological phenomena into economic analysis. In lieu of a more complete list of such
objections and my rebuttals, and to elucidate some of the above arguments, a shall
instead relate a none-toosubtle parable that continues from the movie example earlier in
the essay, and that continues with that theme of how differently economists react to
unfamiliar assumptions than to classical ones.

At the Movies: Economics from Another Planet

I earlier illustrated an untenable difference in economists’ reaction—and standards of

proof—in accepting the evidence for the unfamiliar assumption that people have a taste
for retaliating against unfair treatment to such paragons of acceptable preferences as
enjoying action movies. This example can be stretched further. Suppose that in another
galaxy there is a planet—Planet Nonhollywood—where the actual economy developed
exactly has it has here on Earth, and that the economics profession evolved almost
exactly the same as here. There is just one difference between the economic professions
on the two planets: On Planet Nonhollywood, economists had traditionally not studied
preferences over “entertainment” items—things that weren’t physically consumed, but
merely “psychologically consumed”. I don’t want to be

judgmental about the coherence of our colleagues across the universe; for whatever
reason, they had developed what they thought in their own minds was a sensible
distinction. Roughly corresponding to what we would label entertainment items, they had a
strong sense that things that could not be eaten, touched, pushed, etc., and especially
that were temporary, were simply not a direct part of anybody’s utility function. By
contrast to the economists, psychologists on Planet Non-Hollywood talked all the time
about such phenomena, but economists dismissed them as wooly-headed and unscientific.
Because the economy is identical to that of the Earth’s, many people (including
economists) would go see movies and entertain themselves in other ways. And a few
economists started to suggest that maybe people did intrinsically enjoy “nontangible
consumption,” noting all the money in the economy devoted to the enjoyment of others’
company, entertainment industry, the role of ambience in restaurants, the fact that
people travel to see beautiful buildings and paintings even if they couldn’t touch them. But
most economists had simply managed to ignore these phenomena, or come up with
alternative explanations using (what to them were) acceptable, “classical” assumptions.
Now imagine you travel to Planet Nonhollywood to give seminars on the movie industry,
arguing that people intrinsically value movies. The following are some of the responses you
might encounter.

“But there are alternative *standard’ explanations!”

Your evidence was very weak, and acceptable standard explanations abound for why
people would pay to see movies even if they didn’t intrinsically enjoy them. When so many
standard explanations exist, why introduce a crazy new assumption? For instance, movie-
goers could be going just for the food; and indeed, cinemas make more money from the
sale of food than from the movies themselves. The large amounts of food consumed—and
the convenient, efficient seating by which to eat it—support this explanation. Indeed, all
sorts of predictions of the standard model are borne out, making

the new-fangled wanting-to-see-the-movie interpretation redundant. For instance, the

better and cheaper the food, the more likely people are to go.18 Also, you didn’t
adequately demonstrate that people weren’t going to the movies intending to make money
rather than spending it—hoping to find money underneath the seats in front of them.
When you admitted not having the data to rebut this hypothesis, but said you found it
implausible, you are dismissed as unscientific, and treated to an anecdote of an audience
member once having found twenty Nonhollywood dollars on his seat. Or, you were told,
movie-going could be a signal of wealth by wasting money: There is lots of evidence that
people (especially males trying attract mates) like to pay for movies

. The fact that when wealth-signaling is most likely—as on courting rituals— people are
most likely to pay for the movies strongly supports this explanation. Audience members in
fact provided anecdotes of all the times they went to movies they didn’t want to see
merely to be on a date— and they paid for both tickets, hence doubly signaling their
wealth. When you asked them why their date wanted to go, you got told that the move
(and the food) were free to the date, all supporting the standard interpretation of no
intrinsic taste for movies. Other audience members argued:

“but the alleged ‘preference’ is ‘unstable’”

It was often pointed out, and backed up by research, that this alleged preference for
seeing movies is highly sensitive, and therefore not a real preference. While it is true
that some people like going to the movie, it varies a great deal. It depends on mood, time
of day, etc. Indeed, while behavioral researchers claim to have evidence of people willing
to pay for movies, a great deal of experimental evidence by economic experimentalists
show that this taste goes away under only slightly different conditions. Moreover, when
the experiment was done properly—in the way economic experimentalists understood how
to do experiments—the taste for movies nearly completely went away. Evidence from
well-run economic experiments shows that this alleged taste for movies is highly
ephemeral.

“But evidence shows people learn they don 7 like movies . . . ”

While a few psychologists have argued that they have evidence that people seem to like
movies, these experiments are run under novel conditions, and don’t allow learning.
Indeed, the standard in psychology experiments was to only ask people to see a movie
once. Hence, you were told, we do not learn whether this behavior represents a robust
preference. But experiments showed that, while a person might pay $8 to see the movie
once, maybe twice, if you keep asking him for $8 to see the movie, eventually stops
paying. Clearly he learns he doesn’t want to see the movie! Once play “converges” to
“equilibrium” behavior by subjects, we see no genuine preference for movies.19 Indeed, an
audience member pointed out that this provides further support for the money-under-
the-seats interpretation of movie going: People start going to the movies, V. then when
they don’t find money under the seats in front of them, they learn it involves losing money
rather than gaining money, so they stop going. Another audience member pointed out that
this movie-seeing couldn’t last, since it is clearly irrational:

“But this behavior is ‘non-consequentialist’, and hence irrationall”

The notion that people might care directly for the movie they see is so much
“psychobabble.” Movie-goers walk out of the theatre with no more than they walked in, $8
poorer. To pay $8 for a two-hour process that puts nothing in a person’s hand or

stomach is, you are told, irrational. Some researchers had started to toy with such
“nonconsequentialist” preferences, and conceded it was an interesting possibility, but
realized such preferences couldn’t be rational—getting nothing for your $8 can’t really be
rational. Indeed, if there were people who went around giving $8 for nothing in return,
they would quickly be driven from the market, so that there behavior would not matter:
“But those behaving like this will be driven from the market!” An audience member
assured you that somebody willing to pay $8 for a movie could be “Dutch-booked”: If
people paid $8 just to sit in front of a screen, then somebody could make money off of
them!20 When you respond that, yes, somebody could and is making money off of those
willing to pay the $8, another audience member assures you that if people were really
willing to pay $8 for nothing in return, they would in short order be bilked of all their
money by an arbitrageur. When you shyly suggest that a consumer’s willingness sometimes
to give some of his money to see a movie doesn’t mean he’ll pay infinite amounts to
anybody who offers movies, or suggest it might be costly to provide these movies, you get
scoffed at for being ad hoc, changing your story, and being very loose about what
preferences you were proposing. Somebody else also points out that, because people are
irrationally paying $8 for nothing, such movie-goers will be driven from the market by
those who are as happy not seeing the movies. These others will have all this money that
movie-goers won’t have, and hence have greater survivorship in the goods market.
Moreover, because the spending needs of non-movie-goers are lower, they will undercut
movie-goers in the labour market, accepting the same jobs at lower wages, so that would-
be movie-goers will be left unemployed and not be able to go to see movies anyhow! That
line of argument makes no sense to you, and as you are trying to articulate a rebuttal to
such a far-fetched story, you get hit with the seminar-stopping argument you always
fear: “B ut none o f the evidence is fo r ‘real stakes’” Somebody raises his hands, and
while acknowledging that maybe your experimental results are robust, you haven’t
demonstrated willingness to pay for movies for stakes that really matter. “Sure,” an
audience member starts, “they will go see the movie for eight dollars—but would they for
eight hundred dollars?” You know you are beaten. You hem and haw and explain you would
love to get more funds from the Nonhollywood Science Foundation to write bigger-stakes
experiments, and note that a colleague has run similar experiments on a poorer nearby
planet, and while the willingness to pay for movies is smaller, it is still non-negligible. You
can’t convince the audience you still care about movie-going even though it involves non-
huge stakes. You are not going to convince this audience

Back On Earth

Is this tale of arguments against the people-enjoy-movies hypothesis a fair parable of

what kinds of arguments we see here on Earth against the sorts of psychological
assumptions I have discussed? It would of course be unfair to say that all economists who
have resisted psychological assumptions have made arguments analogous to those above.
In fact, few of these arguments are written down (though some are) because they would
look transparently silly if they were made about what we (on Earth) consider standard
preferences. The intended moral of my tale is, of course, precisely that many of the
arguments used against unfamiliar assumptions are awful economics and don’t hold up to
scrutiny. They would be embarrassing to authors when made in print, and would in any
event not be accepted by editors. But these types of arguments were the types of
arguments frequently made at seminars ten years ago, and occasionally made today,
against unfamiliar but psychologically sound assumptions. In fact, many of the examples
parallel economists’ reactions to those researchers trying to argue that people have
intrinsic taste for fairness and other non-self-interested preferences. These are the
types of arguments I used to hear all the time (but far less frequently now). All manner
of self-interested explanations for rejecting unfair offers in the ultimatum game were
offered. To those of us who see nothing mysterious about rejecting such offers, the
pursuit of money-maximizing interpretations is as strange as a moneymaximizing
interpretation of movie-going. The fascination many economists have had with the
“instability” of the taste for fairness—and the predisposition to argue that the
experimental manipulation that produces the least concern for fairness yields the “true”
stable preferences—has struck some of us as wholly disproportionate in comparison to
the lack of focus on the “instability” of classical assumptions about preferences. Showing
that there exist conditions where a preference for fairness “goes away” isn’t a
demonstration that these preferences don’t exist. And the tendency of experimental
economists to assume that all temporal changes in behavior represent “learning” would be
recognized as transparently misguided if applied to standard assumptions. It is silly to
label the diminishing marginal utility of seeing the same movie repeatedly as a person
learning he doesn’t like the movie. But experimental economists studying the important
topic of learning, or wedded to a methodology of repetition in running experiments, are
ignoring the possibility that people with a taste for revenge and similar non-
selfinterested tastes might similarly exhibit satiation.22 And the consequentialist
arguments that those who reject money in the ultimatum game get “nothing” in return
similarly derives from habitual thinking what constitutes “something” and what
constitutes “nothing”. And the related arguments that paying money for nothing but the
satisfaction of revenge is a money-losing behavior and therefore unable to survive in the
market or in evolution seem highly confused. Every time a consumer buys anything he
loses money. And every time he buys something without the highest caloric and nutritional
survival payoff we have an evolutionary mystery on our hands for those that want to solve
an evolutionary mystery. For those interested in economic outcomes, it will be more
sensible and more adaptive to assume that people are willing to spend money on whatever
they are willing to spend money on. And at die lion’s share of seminars on non-self-
interested behavior five years ago the speaker would be cross-examined about the low
stakes in the experiments being discussed. It is perfectly reasonable to care a lot about
what the shape of the demand curve for revenge looks like. (And the answer as far as the
evidence so far suggests is that it is not very steep—many people are willing to pay quite
a lot for justice.) But the mere fact that the taste for revenge and fairness is finite, and
diminishes when it is more costly to purchase, makes it like every single other taste
economists study, not something

to be dismissed. And the presumption that economists don’t consider even $8

transactions to be “real stakes” is plainly false, as the movie-going example illustrates.

Final Thoughts

Above I have proposed that economists have until recently resisted new assumptions
about non-self-interested preferences based on economically flawed arguments. Similar
flawed arguments are used to resist other modifications that will improve the
psychological realism of economics. As economists are starting to realize that
metaarguments for dismissing these modifications are not helpful to economics, and to
realize that adding greater psychological realism will improve it rather than undermine
economics, such defensive arguments are decreasing. Happily, the trend is towards
integrating apparently true and apparently relevant new psychological assumptions into
economic analysis.