JOS Referee Report
Review of ‘The Empirical Evidence Against Utility Maximization’ (JOES-12-01-
001)
Summary of the Paper
This paper summarizes many strands of empirical evidence against the textbook
economic model of decision making of a fully rational, self interested exponentially
discounted expected utility maximizer. The authors focus on 7 topics (as I interpret
them): Social preferences, cognitive ability, the construction of value, savings behavior,
alternative models of consumption, the Poliani critique and the relationship between
income and happiness.
Overview
This paper offers an interesting guide to many of the empirical findings that have driven
the rise of behavioral economics over the last 30 years. As a behavioral economist, I
think that many of the issues that are raised in the paper are extremely interesting and
worthwhile. However, I think that there are several issues that limit the paper’s
effectiveness in summarizing the weakness of the extbook model, which I list below
The Title: Utility maximization has a technical definition, which is that
people maximize (or at least act as if they maximize) some stable utility
function. In the first instance, we don’t care what it is that determines utility,
just that there is some coherence to people’s choices. Since Samuelson, we
have known how to test whether people look like utility maximizers – we see
whether they violate WARP (or GARP, or SARP). There is indeed a literature
on whether or not people act like utility maximizers. See for example
Echenique et al [2011] and Choi et al [2007]. But this paper doesn’t look into
this problem, and as such I’d suggest a different title
The Tone: Even as someone who has a lot of sympathy with many of the
arguments in the piece, I found the tone somewhat jarring. For example, in the
conclusion “Today’s economists do not bother to check whether their theories
are actually aligned with real world behavior”. This will come as a bit of a
surprise to the thousands of applied economists who thought they did exactly
that. Even if the point is more that we don’t check whether the underlying
assumptions on which we build our theories on is correct, this will come as a
surprise to the thousands of behavioral economists who work on many of the
issues described in this paper.
This is one example of what I believe to be a theme running through the paper
which is “economists are blinkered idiots who are willfully ignoring the way
in which people behave when they build their models, and therefore say all
sorts of stupid things” This is a view that may (or may not) have been valid 30
years ago, but I don’t think it is today. (Almost) every major economics
department has someone who works on behavioral issues, and almost all
important economic issues have been considered in the light of the
phenomena that the authors discuss (social preferences, temptation and self
control, etc). A look at the February edition of the AER shows an article on
the origins of co-operation, one on the effect of confusion on asset prices, one
of the effect of conflict on social preferences and one on the behavioral
foundations of mircrocredit. It is therefore in my opinion unfair and unhelpful
to label all economists as ignorant (or dismissive) of behavioral issues.
Of course, this is not to say that all economists spend all their time using or
designing behavioral models. Some of them think that the more traditional
models of behavior capture what is important for the question they are
studying. Whether they are correct in this belief or not is an empirical
question, and an extremely important one, but not one this paper does a good
job of dealing with. In fact, I would argue that it presents an extremely one
sided view of the data: while it is true that the factors that the author presents
in this paper are important for behavior, it is also true that the forces
considered by traditional economics are also extremely important. In fact, if
you open any major economics journal at random, you will find articles that
describe how a model that boils economic behavior down to an (obviously
untrue) set of primitives can actually do a surprisingly good job of explaining
many phenomena. A crucial question for behavioral economists is identifying
which behavioral phenomena are quantitatively important for understanding
which questions. It would be great if this survey could try to do this in a more
evenhanded way
The level of detail and depth of analysis: I think that the article had two
problems with regard to the thoroughness of its analysis. First, I think that,
after 30 years or so of behavioral economics, an article such as this needs to
do more than just describe the existence of various behavioral phenomena, but
say something about how we model them, and the evidence for and against
these models. This seems to be pretty much entirely absent from this paper:
for example there is no discussion of inequality aversion as an explanation for
social preferences, quasi-hyperbolic discounting as a model of temptation and
self control or prospect theory as an explanation for reference dependence and
probability weighting. This is important with regard to the discussion above,
in the sense that these are models that potentially may work better that the
standard econ 101 model, and so a wider population of economists should be
using them.
Second, the article contains a number of unsubstantiated (and possibly untrue)
claims that at the very least need to be backed up. This is particularly the case
with regard to the cause of the financial crisis. This is a question that could
use a survey article in itself, but it is not obviously clear that the issues raised
in this paper offer a more convincing explanation of what went on that (for
example) the presence of financial frictions and credit multipliers. I would
therefore drop these claims (or, if I am wrong, do a better job of backing them
up)
Materiel covered, and organization. By and large, the article covers many of
the ‘big hits’ of behavioral economics. However, the organization seems to be
a little arbitrary. For example Intergenerational Justice seems to be related to
social preferences, yet is in a different section, while temptation and self
control issues turn up in section 5 and section 8. One way to organize the
paper that would (in my opinion) make sense would be
o Utility maximization, value construction and framing effects
o Social norms and preferences
o Complexity and computational costs
o Temptation and self control – including issues of discounting but also
preference for commitment, which is currently not discussed
o Choice under risk and uncertainty – including a discussion of
violations of expected utility and ambiguity aversion, which is
currently missing
o Reference dependent preferences – including the endowment effect,
but also status quo bias and the reflection effect (which are currently
not discussed)
o The effect of stakes, learning and experience: The authors discuss this
somewhat, but it seems so ubiquitous that I wonder if it is worth
getting its own section
o Happiness.
Specific Comments
In section 1, it is probably worth discussing some of the different models of other
regarding preferences, such as fairness [Rabin 1993] or inequality aversion
[Bolton and Ockenfels 2000] and the evidence for and against these models. Also
in the context of altruism, discussing the warm glow vs cold prickle [Andreoni
1995]
This is not my area, but I feel that section 2.3 is overstating the case. While there
is evidence that in some cases punishment and social norms are substitutes, is the
author really claiming that the amount of cheating is always inversely related to
the benefit of doing so? Surely this is not the case?
In section 2.4, the first thing that ‘standard economics doesn’t take into account’
about communities is, I think something that a lot of (completely standard)
economists think about - particularly in the network literature - at least as far back
as Rees [1966]
When discussing the limits of computational ability, it is probably worth
discussing the rational inattention literature started by Sims [2003] and the related
evidence. In the game theoretic context, the level K model and related evidence
should be discussed (e.g. Costa-Gomes et al 2001)
The claim at the end of section 3.2 is overly strong. There are at least some
occasions in which learning and experience do reduce ‘irrationality – for example
List [2003]
In order to serve its purpose as a survey article, I think it would be good if the
authors could give some more details about the biases listed by Barbaris and
Thaler [2003]
I’m not sure why the endowment effect is listed under evidence from cognitive
psychology. Wasn’t this first discussed in the economics literature?
The claim at the end of the first paragraph on page 25 seems to be another
example of overreach. There are many possible reasons that stadium owners
behave in this way, of which you have listed one.
The ‘three real world problems’ discussed on page 30 are not convincing reasons
to discard standard economic methodology. The first is something that could
easily be incorporated into standard models. The author makes a nice point that
this should be done, but there is nothing ‘behavioral’ about this. The second point
is wrong in the sense that these models are precisely trying to capture some
particular social norms. Of course, there may be others that are also worth
considering, but that does not mean that the exercise in this book is not
worthwhile. Third, the question of whether social planner solutions can be
implemented by the market is a large and well studied area in its own right
Section 6 ignores an enormous economic literature on consumers who have to
search in order to find out information on the goods that they may want to buy,
for example Stahl [1989]
For many reasons I found the section on the Polanyi critique unconvincing, and I
would drop it.
I don’t agree with the claim at the top of section 8 that “Ultimately, the subject of
economics is (or ought to be) about how materiel resources can be used for the
benefit of human beings”. This is one of its aims, but economics is a science
(whereby we want to understand how stuff works), not just a policy exercise.
The discussion of happiness misses out any discussion about the problems of
measuring happiness, and in particular what we learn from happiness surveys.
The claim about happiness and set points at the top of page 43 is very strong, and
needs to be either weakened or backed up.
References
Andreoni, James, 1995. "Warm-Glow versus Cold-Prickle: The Effects of Positive and
Negative Framing on Cooperation in Experiments," The Quarterly Journal of Economics,
MIT Press, vol. 110(1), pages 1-21, February.
Gary E. Bolton & Axel Ockenfels, 2000. "ERC: A Theory of Equity, Reciprocity, and
Competition," American Economic Review, American Economic Association, vol. 90(1),
pages 166-193, March.
Syngjoo Choi & Raymond Fisman & Douglas Gale & Shachar Kariv, 2007. "Consistency
and Heterogeneity of Individual Behavior under Uncertainty," American Economic
Review, American Economic Association, vol. 97(5), pages 1921-1938, December.
Costa-Gomes, Miguel & Crawford, Vincent P & Broseta, Bruno, 2001."Cognition and
Behavior in Normal-Form Games: An Experimental Study," Econometrica, Econometric
Society, vol. 69(5), pages 1193-1235, September.
Federico Echenique & Sangmok Lee & Matthew Shum, 2011. "The Money Pump as a
Measure of Revealed Preference Violations," Journal of Political Economy, University of
Chicago Press, vol. 119(6), pages 1201 - 1223.
John A. List “Does Market Experience Eliminate Market Anomalies?” The Quarterly
Journal of Economics Vol. 118, No. 1 (Feb., 2003), pp. 41-71
Rabin, Matthew. 1993. "Incorporating Fairness Into Game Theory and Economics." The
American Economic Review.83, 1281-1302.
Rees, A. (1966) “Information Networks in Labor Markets," American Economic Review,
56, 559-566.
Sims, C.A. 2003. Implications of Rational Inattention. Journal of Monetary Economics
50(3), 665-690.
Stahl, "Oligopolistic Pricing with Sequential Consumer Search," American Economic
Review, 79, September 1989, 700-712