View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Vol. 7, No. 10 • SEPTEMBER 2012­­


Cost of Decisionmaking Influences
Individual Selections
by Anton Cheremukhin and Antonella Tutino

A rational person
may be willing to err
if the cost of making
a mistake is less than
the cost of a precise
and correct evaluation
of each option.


arket prices are often driven
by choices later viewed as
mistakes. Waves of optimism
or pessimism sometimes dramatically move prices; a burst bubble of
euphoria can bring significant macroeconomic consequences.
A sudden change of sentiment may
occur when a large number of stock
market professionals consistently err by
holding on to stocks for too long when
they should sell, or by selling equities too
quickly when they should be holding on
to them. Yet, these individuals are specialists with every incentive to evaluate
stocks correctly.
Behavioral experiments show that
in laboratory conditions, people behave
like market participants.1 When faced
with the same question repeatedly within
any single experiment, they frequently
change their minds.
Why are people so inconsistent? Do
rational people blunder? Theories on
how individuals and groups reach decisions don’t provide a satisfactory answer.
By and large, the mystery of costly human
errors remains unsolved. Understanding
why such mistakes occur can help
researchers interpret change in observed
behavior and carries implications for the
behavior of financial markets.

Weighing the cost of making decisions may provide an answer.2 A rational
person may be willing to err if the cost of
making a mistake is less than the cost of
a precise and correct evaluation of each
option. A rational person balances the
gain from a consistently beneficial choice
with the cost of paying attention—that is,
the cost of being precise.
Though information is abundant,
not all of it is necessary to make a
well-informed choice. Inattention to
some information is a perfectly rational
response in these circumstances.

A Theory of Error
We can illustrate the implications of
rational inattention. Suppose a person is
asked to choose between option A, which
provides a guaranteed payout of $10, and
a risky option B. Classical rational choice
theory holds that if the value of option B
is higher than that of option A, the person
should choose option B. It makes no difference how much better option B is; if
the person is asked to choose 100 times,
option B should be selected 100 times.
Experimental data testing the theory
show that people do not behave in the
predicted manner. Chart 1 compares the
prediction of rational choice theory with
observed responses to an experiment in

Economic Letter

which the A and B choices are offered.
The data suggest that it is very common
for a person to pick the better option
only 70 to 80 times out of 100, even if that
option is unquestionably superior.
Rational inattention theory can
explain this seemingly erratic behavior.3 It
postulates that rational people are prone
to mistakes if they believe that the effort
of making an informed choice is greater
than the benefit of a correct choice.
Further, this theory not only helps clarify
individual preferences but also plays a

pivotal role in discerning what a person
considers a costly mistake. For instance,
a risk-averse individual may favor a safe
and informed choice over a riskier and
more profitable bet. By contrast, a riskloving person may lean toward more risky
gambles, even without fully analyzing the
properties of such an option, if the stakes
are high enough.
While rational choice theory predicts
deterministic outcomes—sure events—
the thrust of rational inattention theory
stems from its focus on expressing the

Measuring Responses in the Lab
Forty individuals were repeatedly asked to compare 20 pairs of propositions—a choice between
proposition A and proposition B, each with two potential payouts—in an experiment Michel Regenwetter conducted at the University of Illinois at Urbana–Champaign in 2009. Each individual faced each
proposition 60 times. The set of propositions per individual was shuffled to ensure that the experiment
was unaffected by memory effects—when people recall the answer they gave to the same question
The table below gives an example of the propositions participants faced. Each question contains
two propositions. Participants were asked to choose one of them.
Sample of Gamble Payouts and Probabilities
Proposition A


Proposition B































































The diagram below is derived from a sample screen from the experiment representing question
1 from the table. The pie chart illustrating proposition A offers $29.38 at a 65 percent probability and
$1.19 otherwise. The pie chart illustrating proposition B offers $18 at a 68 percent probability and $3.21


Proposition A

Proposition B







likelihood of people’s choices in terms
of probability. A rationally inattentive
person behaves in terms of “odds,” flipping a coin as each choice is made.
While the outcome of a particular
choice is unpredictable, rational inattention theory predicts the probability
of each occurrence.

Laboratory Tests Behavior
The natural place to study probabilistic choices is in the laboratory.
We focus on laboratory experiments
where subjects are repeatedly asked the
same sets of questions to see whether
and under which circumstances they
change their answers. Behavioral patterns emerging from the laboratory
show that our model based on rational
inattention can reconcile theory and
evidence. The lab experiments test the
ability of people to make a consistent
choice over pairs of propositions that
promise different expected payoffs. The
actual gambles played in one experiment are shown in the box, “Measuring
Responses in the Lab.”
In Chart 1, experimental data are
depicted together with the predictions
of rational choice theory4 (dashed line)
and our rational inattention model
(solid line). Our model predicts that
the greater the difference in values
between options A and B, the more
consistent people will be in their
responses. The slope of the curve predicted by our model reflects the cost
of attention. The more costly attention
is, the greater the difference between
two options must be for a person to pay
close attention.
The actual experimental data seem to
agree well with the theory’s predictions,
allowing quantification of the effort associated with paying attention and giving it
a monetary value.
To illustrate this point, consider a person who wants to reduce uncertainty by
asking simple binary (yes/no) questions.
Each answer to a binary question would
reduce uncertainty, but it costs effort to
process that can be quantified in terms of
lower expected payoff.
We estimate the cost of attention to be
on the order of six questions per 1 cent

Economic Letter • Federal Reserve Bank of Dallas • September 2012

Economic Letter

Risk Aversion and Inattention
Drawing on rational inattention theory, individual attitudes toward risk and
inattention can be measured. While the
cost of attention determines the slope of
the curve, attitude toward risk determines
perceived values of options A and B.
Thus, people’s risk aversion determines
the sorting of options along the horizontal axis of Chart 1.
The attitude toward risk and the cost
of attention for the 40 participants in the
experiment are estimated using experimental data. In Chart 2, we use two indicators: the risk aversion characteristic on
the vertical axis and the cost of attention
on the horizontal axis.
Side-by-side consideration of two
propositions illustrates risk aversion. Offer
A yields a $20 winner 50 percent of the
time; a losing wager is worth nothing. Offer
B yields a sure-fire $10 winner 100 percent
of the time. A risk-neutral person will be
indifferent toward the two propositions.
The risk-neutral person is denoted by zero
on Chart 2. By contrast, a risk-averse person will prefer offer B, and the value of his
risk-aversion characteristic would be in the
positive portion of the vertical axis. Finally,
a risk-loving person would prefer offer A,
and his risk-aversion characteristic would
be in the negative range of values on the
vertical axis. Likewise, a less-attentive
person—someone more concerned about
effort than an expected return—would be
characterized by low values on the horizontal axis of Chart 2. A more-attentive
person—someone more concerned about
an expected return than effort—would
take on high values.
Chart 2 shows an overwhelming
difference in these two characteristics across a sample of fairly similar
people. Although the majority of participants were drawn from a relatively
homogeneous student population at
the University of Illinois campus, their
responses demonstrate remarkably
different behavior.5 Some participants
are very averse to risk—corresponding



Rational Inattention Theory Guides Betting Choices

Times option B chosen


Experimental data


Rational choice theory


Rational inattention theory




Value of option B (in dollars)



SOURCES: Michel Regenwetter, University of Illinois at Urbana–Champaign; authors’ calculations.



Group Reaction Differs from Individual Decisions

Attitude toward risk



Average individual
Mixed response


Risk neutrality


Love Aversion

of expected payoff. This implies that an
average person values the effort associated with answering six simple yes/no
questions at 1 cent.


Less attentive

More attentive
Cost of attention (questions/cent)



SOURCES: Michel Regenwetter, University of Illinois at Urbana–Champaign; authors’ calculations.

to a factor of risk aversion equal to +4.
Others are willing to take on a substantial
amount of risk if they think the stakes are
good enough—corresponding to a factor
of risk aversion equal to –2.
The cost of attention characteristic
takes on an even wider range of values.
Individual participants are motivated to
pay attention and give answers in a consistent way to between one and 12 questions for each cent paid.

People Aren’t Equally Attentive
Our estimates show that people
are extremely different in both their
costs of paying attention and attitudes
toward risk. One stark implication of
these differences involves the tendency
of group behavior to diverge from that
of the individual. Evidence presented
in Chart 2 suggests that the average
individual response (depicted by the
square) is characterized by a cost of

Economic Letter • Federal Reserve Bank of Dallas • September 2012


Economic Letter

attention twice that of the group aggregate (depicted by the diamond) for participants in the experiment.
To understand the implications of this
finding, consider the following example.
After a market-price bubble bursts, standard theory would predict that once the
average market participant realizes an
asset is overvalued, the overall market
would quickly reprice the asset to its fundamental value. However, we find that it
will take more than half of market participants to realize that the asset is overvalued for its price to be corrected.
While rational choice theory cannot
easily account for this result, it fits comfortably within rational inattention theory
postulating a mixed response as an outcome of participants’ choices.

Tempering Market Bubbles
This has important implications for
comprehending the behavior of stock
markets, which similarly aggregate choices of market participants. Each individual
market participant must be somewhat
inattentive, like every single participant
of our experiment. If differences in levels
of attention among market participants
are at least as large as those observed in
the laboratory, then aggregation bias (the
difference between individual decisions
and implied group decisions) should
have a substantial impact on the way
market-price fluctuations are interpreted.
The prediction of rational inattention
theory is that the behavior of the market


as a whole should look as if produced by
an extremely inattentive “representative”
market participant.
This result may aid recognition of how
market fluctuations take form and whether they can be tempered by some entity
or mechanism. Maybe bubbles cannot
be avoided, but they might be lessened
by appropriate action and aggregation of
information, such as making more and
better information available to market
participants on traded assets.

rise to curves used to describe the data. For an axiomatic
approach to formulating such a relationship, see Individual
Choice Behavior: A Theoretical Analysis, by R.D. Luce, New
York: John Wiley, 1959.
We thank, without implicating, Michel Regenwetter for
kindly providing access to the data. Anna Popova’s work
and data collection were supported by National Science
Foundation grant SES # 08-20009 (PI: M. Regenwetter,
University of Illinois at Urbana–Champaign), titled “A
Quantitative Behavioral Framework for Individual and Social
Choice,” awarded by the Decision, Risk and Management
Science Program.

Cheremukhin and Tutino are research
economists in the Research Department at
the Federal Reserve Bank of Dallas.

“An Experimental Measurement of Utility,” by Frederick
Mosteller and Philip Nogee, Journal of Political Economy,
vol. LIX, no. 5, 1951, pp. 371–404.
This article is based on a joint paper with Anna Popova:
“Experimental Evidence on Rational Inattention,” by Anton
Cheremukhin, Anna Popova and Antonella Tutino, Federal
Reserve Bank of Dallas, Working Paper no. 1112, December
See “Implications of Rational Inattention,” by Christopher
A. Sims, Journal of Monetary Economics, vol. 50, no.
3, 2003, pp. 665–90, and “ ‘Rational Inattention’ Guides
Overloaded Brains, Helps Economists Understand Market
Behavior,” by Antonella Tutino, Federal Reserve Bank of
Dallas Economic Letter, vol. 6, no. 3, March 2011.
Rational choice theory postulates that individuals behave
quickly and precisely in any environment. To account for
observed variations in responses, a relationship between
the error probability and the values of the options is often
exogenously postulated. This assumed relationship gives

Economic Letter

is published by the Federal Reserve Bank of Dallas. The
views expressed are those of the authors and should not
be attributed to the Federal Reserve Bank of Dallas or the
Federal Reserve System.
Articles may be reprinted on the condition that the
source is credited and a copy is provided to the Research
Department of the Federal Reserve Bank of Dallas.
Economic Letter is available free of charge by writing
the Public Affairs Department, Federal Reserve Bank of
Dallas, P.O. Box 655906, Dallas, TX 75265-5906; by fax
at 214-922-5268; or by telephone at 214-922-5254. This
publication is available on the Dallas Fed website,

Richard W. Fisher, President and Chief Executive Officer
Helen E. Holcomb, First Vice President and Chief Operating Officer
Harvey Rosenblum, Executive Vice President and Director of Research
E. Ann Worthy, Senior Vice President, Banking Supervision
Mine Yücel, Director of Research Publications
Jim Dolmas, Executive Editor
Michael Weiss, Editor
Jennifer Afflerbach, Associate Editor
Ellah Piña, Graphic Designer

Federal Reserve Bank of Dallas
2200 N. Pearl St., Dallas, TX 75201