Behavioural Economics

Sujoy Chakravarty
Centre for Economic Studies and
Planning

What is behavioural economics?

Behavioural economics is a new branch of economics that uses
theories, tools and techniques of analysis from social sciences such
as psychology or anthropology that are used to enrich the model of
human behaviour that has been used traditionally in neo-classical
economics.
As it stresses observed agent behaviour rather than logical
optimization algorithms, behavioural economists are often also
experimental economists, who conduct lab and field experiments in
the vein of psychologists.
Observations of deviation from theory predictions may provide
insights into building new theories, that improve upon existing
ones. Eg.- herd behaviour, satisficingbehaviour, ecological rationality,
other-regarding preferences and prospect theory

Why do we need this?

Though economic theory predicts behaviour in a large number
of economic problems, the predictions are quite narrow in
scope.
In the real world however we see a vast number of
behavioural realizations:
The amount of contribution made to charities.
The proclivity towards acquiring resources through corrupt
means.
Cooperation displayed towards teammates and co-workers
The wherewithal to take on monetary risks.
Consumption and saving behaviour.

Economic Theory
Over the last century has concerned itself more with rational
benchmarks and less with trying to model empirically observed
behavior.
According to Smith (1989) most economists feel that
economics is an a-priori science rather than an observational
one.
According to Milgrom and Roberts (1987, p. 185) no mere fact
was a match in economics for a consistent theory. Thus most
economic theorists believe that economic problems and agent
behavior therein can be fully conceptualized by thinking about
them.
Accordingly after the thinking has produced sufficient
technical rigour, internal coherence and interpersonal
agreement, economists can then apply this to the world of
data. (Smith, 1989, p. 152)

The bottom line

The rational agent thus modeled is a selfinterested maximizer always taking a decision
that maximizes his expected wealth.

But from casual empiricism

if we look around us we see serious violations
of this behavioural norm.
Human beings are prone to voluntary acts of
kindness sometimes motivated by altruism.
Most people cooperate with their neighbours
in the upkeep of their neighbourhoods.
Most would trust another human being
unconditionally often suffering negative
consequences from this trusting behavior.

Predictably or systemically irrational

Are most people then irrational?

If that were the case shouldnt theory reflect this

divergence from the norm especially as this divergence
is often systematic and not random?
Ariely (2008) refers to these divergences as predictably
irrational ones.
Basu (1983), Sugden (1986) and Coleman (1990) have
attributed social norms (especially those that inculcate
values and a sense of morality) as modifiers on our
desire to maximize our material wealth.

Early glimmers of behaviourism

Adam Smith in The Theory of Moral Sentiments
(1759) builds on the work of David Hume (A
Treatise of Human Nature, 1739)
Here, Smith writes about an important
psychological motivation that governs the way
that economic agents conduct exchange.
The particular moral sentiment that Smith
describes at length is sympathy, i.e. the feeling
of compassion or concern for others.
This tempers self-interest in socio-economic
transactions and leads us to sacrifice narrow
profiteering in order to maintain ties of affection
with our fellow human beings.

So then why do we need economic

theory at all?
Economists have traditionally avoided
explaining behavior as less than rational for
fear of developing many fragmented theories
of mistakes.Erev and Roth (1998, p. 848)
Economic theory provides a behavioural
(rational) benchmark of great generality
against which we may compare observed
behaviour.

The role of experiments

The field of experimental economics (not to mention
experimental social psychology) stands at the centre of this
debate on observed behavioural deviations from rational
theoretical prediction.
The method of controlled experiments from the fifties
onwards has allowed us to test game theoretic models as well
as individual choice models with human subjects in the
laboratory.
Observed economic behaviors are compared to rational
theoretical benchmarks, divergences from the rational norm
are noted and in certain cases theories are advanced to
explain these deviations.

A difference in approach
A fundamental point of divergence between
psychology/cognitive science and economics is related to
the theoretical underpinnings behind various results.

Most economists dont really need a precise and accurate

theory at the individual level, just as long as it is general
enough to explain some of the observed behaviour
accurately and generate an aggregate prediction that is
more or less accurate.
So elegance and generality are generally desirable in
economic theory, whereas psychologists and cognitive
scientists are interested in modeling the precise nuances of
behavior displayed by individuals.

An example
A typical economics experiment on attitudes towards risk
would have the researcher make an assumption about the
form of the utility function (say constant relative risk
aversion or CRRA) that the agents purportedly follow.
Using this function and the choice response in the
experiment, one can calculate some measure (maybe
Arrow-Pratt) of risk aversion and then compare this across
agents, over time, cross-culturally etc.
If anyone questions the validity of using this functional
form over another one, most of the time the answer that a
theorist or an experimental economist would give you
would be that it doesnt matter as long as everyones
attitude to risk is measured using the same CRRA
specification.

The as-if approach to behaviour

Most psychologists and cognitive scientists
would be quite unhappy with this as if way
of evaluation and would be more interested in
the cognitive processes that govern the choice
made by the decision maker.
Economists on the other hand, some measure
which has good internal validity but
potentially scanty external support.

One-size fits all

According to Camerer (1995), most economists have a onesize-fits-all approach to studying economic problems vis-vis psychologists.
if a task involves elicitation of a probability, most
psychology experiments would frame the problem in a
natural setting using a vignette. This would anchor the tasks
to certain specific stimuli.
Most economists would go ahead and attempt to elicit the
same probability using a more decontextualised device
such as a pair of dice or a bingo cage.
This is in keeping with the institution free pedagogy of
neoclassical economics where elicitation of a probability is
coming up with a specific statistical measure rather than an
assessment of a contextualized measure of uncertainty.

Bounded rationality
The origin of this boundedly rational approach is from Simons
(1955) idea of procedural rationality whereby agents follow
reasonable heuristics and on average achieve close to optimal
outcomes.
This is distinct from substantive rationality, where the agent
considers the entire set of variables to make her decision.

The idea of bounded rationality was one of failed optimization.

Agents are unable to compute the optimum and settle for a second
best satisficing outcome.
The more modern idea of ecological rationality (Gigerenzer and
Brighton, 2009) strongly opposes this idea that all heuristics are
examples of failed optimization.

Heuristics: A new look at bounded rationality

Notice that the idea of bounded rationality implies that heuristics, which
do not attempt to optimize an objective function with respect to all the
relevant variables is a result of cognitive limitations which force the
decision maker to perform shoddy second best computations.
In contrast Gigerenzer and Brighton (2009) find that heuristics are efficient
cognitive processes that ignore information.
In contrast to the widely held view that less processing reduces accuracy,
their study of heuristics shows that less information, computation, and
time can in fact improve accuracy. Gigerenzer and Brighton specifically
show that there is a trade-off between bias and variance. More variables
decrease bias but increase variance.
According to this fast and frugal or ecologically rational approach to
computation, Homo Heuristicus has a biased mind and ignores part of
the available information, yet a biased mind can handle uncertainty more
efficiently and robustly than an unbiased mind relying on more resourceintensive and general-purpose processing strategies.

What exactly is a controlled experiment in econ ?

It is a way of obtaining data on how economic agents behave (i.e.- the
actions they take) in a game, decision problem or a market.

Subjects are recruited in a randomized way to be agents in a

stylized environment that studies a particular economic problem.
The interaction is described to the participating subjects (i.e.- the
economic agents) with the help of laboratory instructions that present in
detail the tasks that he or she needs to perform.
Monetary incentives are provided in a salient way, so that agents can map
the action space to the payoffs.
In other words, the laboratory interface is a real market or game that
has agents motivated by financial incentives.

Examples include a laboratory double auction market, an oligopoly game

or a two-player Prisoners Dilemma.

Findings from experimental studies

Game theory people are not as hyper-rational as predicted by
theory. Though self interest is an important motivator for decisions
made, it is by no means the only one. People display preferences
for altruism, reciprocity, spite and mutualism. E.g.- cooperation in
prisoners dilemma, egalitarian shares in bargaining games. Culture
and norms play a role in shaping agent behaviour.
Individual decision making Humans are prone to systemic biases
making their behaviour diverge considerably from the model of the
expected utility maximizer. E.g.- endowment effect, reflection
effect. These are often to do with the procedures humans use to
aggregate payoffs.
Since most real world problems involve both individual decision
making as well as interactive components we see procedural as well
as preferential divergences from theory.

Pre-1960 institution free theory

OUTCOMES
Prices, allocations

Post 1960 institutions matter

Post 1980 culture and demographics
matter.

CHOICE
BEHAVIOUR

ENVIRONMENT
Agent values, costs,
endowment,
technology

CULTURE and
DEMOGRAPHICS
Social Norms

INSTITUTIONS
Language of the market
Rules of communication and
contract
Extensive form structure

Figure 1: Institutions, culture, environment and behaviour in economics (extended from Smith, 1989)

Prisoners and tragedies

The meta analysed cooperation in 2 X 2 prisoners dilemma
experiments show cooperation rates of greater than 50 %
Public goods games show cooperation rates of 20-40 %
Contribution to public goods is dependent on cultural norms
with some populations showing very high cooperation and
others close to Nash equilibrium behaviour (Henrich, et al.,
2001, 2005)
Chakravarty et al. (2013) find very high rates of cooperation
(40-50 %) among high altitude village communities in Kumaon.

Altruism and warm glow

Altruism Increase in utility from giving to an other with no
explicit or implicit reciprocity.

Warm glow Increase in utility from the act of giving with no

concern regarding the recipient.
In dictator games a meta give rate of 27 % is seen (Engel, 2011)
Dictator giving is sensitive to contexts, framing, reference
points and social environments.
Banerjee and Chakravarty (2014) find that framing and implied
property rights over the endowment determine dictator
outcomes.

Trust and reciprocity

A trust or investment game is one where a proposer chooses
whether to keep his endowment E or send x, keeping (E-x). The
respondent obtains 3x (original investment trebled) and decides the
amount [0, 3x] to send back to the proposer.
An ultimatum bargaining game is one where a proposer decides what
part (x) of his endowment (E) he will give to the respondent. If the
respondent accepts then the split is (E-x, x) if not then neither get
anything (0, 0)
What are the Nash equilibria of the above games?
For TG, in giving x the proposer displays altruism and forward looking
reciprocity and in returning y, the respondent displays strong
reciprocity.
For UBG, the respondent vetoes only at the cost of making zero
payoff.
How did experimental subjects behave ?

Trust and reciprocity

Berg at al. (1986) find that a substantial number of subjects do not send zero,
i.e.- are trusting and a significant amount of trust is reciprocated.

Ultimatum bargaining [Bowles, 2004]

Game

Results

Interpretation

Citation

Standard

Modal offer =
Offers < 20 % rejected

Reciprocity by
respondent

Guth, Schmittberger and

Schwarze (1982)

Randomized Offers

Few rejections of low

offers.

Proposer not
responsible

Blount (1995)

Roles chosen by
quiz

Many low offers, very few

rejections

Proposer is
deserving

Hoffman, McCabe, Shachat and

Smith (1994)

Exchange game

Many low offers, very few

rejections.

Situational
framing

Hoffman, McCabe, Shachat and

Smith (1994)

No fair offers
possible, only
[(8,2), (10,0)]

Low offers not rejected

Proposers
intentions
matter

Falk, Fehr and Fischbacher

(2003)

Punishment by
third party

Most observers punish

low offers by proposer

Generalized
fairness norms

Fehr and Fischbacher (2001)

Standard: Au/Gnau

Offers > are common

high and low offers are
rejected w/ equal freq.

Endogenous
situation
dependent prefs

Henrich, Bowles, Boyd,

Camerer, Fehr, Gintis and
McElreath (2001)

Standard:Machigue
nga

Many low offers, very few

rejections

Endogenous
situation
dependent prefs

Henrich (2000)

A theory of other-regarding or social

preferences
Says that an agent is not merely interested in increasing his
own payoff from a game, but also concerned about the other
agents payoff, i.e.- he is inequality averse.
Thus Ui = Ui(own, | own other|), increasing in own and
decreasing in | own other|)
Fehr and Schmidt 1999, Falk and Fischbacher, 1998; Fehr and
Schmidt, 1999; Bolton and Ockenfels, 1999; Rabin, 1993;
Charness and Rabin, 1999; Levine, 1998 all have models of
social preferences.
Some of these models use reciprocity/reputation and are not
static.

Norms and culture

A very important intellectual direction that emerged out of the
anomalies observed in laboratory experiments was the study of
the effect of culture and demographics on economic behaviour
through the formation and enactment of social norms.
Sen (1973) alludes to social norms when he discusses the
prevalence of cooperative action in the Prisoners Dilemma
game.
Arrow (1982) clearly states that The model of laissez-faire
world of total self-interest would not survive for ten minutes;
its actual working depends on an intricate network of
reciprocal obligations, even among competing firms and
individuals.

Preference reversals
Problem 1: Choose Between the following two risky bets, A or B:
A. 2,500 with probability of .33,
2,400 with probability of .66,
0 with a probability of .01
B. 2,400 with certainty

--------------------------------------------------------------- Problem 2: Choose between the following risky bets C or D:

C: 2,500 with probability .33, 0 with probability .67

D: 2,400 with probability .34, 0 with probability .66.

N = 72 A [18%] and B [82 %]

Majority display preferences of this type:

U(2,400) > .33U(2,500) + .66U(2,400) --- (I)

N = 72 C [83%] and D [17%]

Thus most people displayed preferences of the form:

.33U(2,500) > .34U(2,400) --- (II)
(I) and (II) are opposite to each other. This was first documented by
Maurice Allais (1953)

Loss domain
Problem 3:
A: 4000 with prob = 0.8
0
with prob = 0.2
B: 3000 with certainty
Problem 4:
C: - 4000 with prob = 0.8
0
with prob = 0.2

D: - 3000 with certainty

The reflection effect

Preferences are risk averse over gains and risk preferring over loss
domains. We are also loss averse, i.e.- attach a greater weight to a
loss as compared to a gain.
Chakravarty and Roy (2009) documents this over both risky and
ambiguous preferences.

Prospect theory
Formulated by Kahneman and Tversky(1979). Its salient points:
People overweight the importance of unlikely events and
correspondingly overweight near certain events.
People respond to framing, i.e.-equivalent outcomes are treated
differently depending on the manner in which the outcomes or the
decision setting are described.
Provides a conceptual framework for dealing with situationdependence. If the utility function is to explain behaviour its
arguments should be changes in states or events rather than the
states themselves. Thus, the value individuals place on states
depends on the relationship of the state to the status quo (initial
wealth, state enjoyed by peers, etc).

Equity Premium Puzzle

Given the return of stocks and bonds over the last century, an unreasonably
high level of risk aversion would be necessary to explain why investors are
willing to hold bonds at all (Mehra and Prescott (1985)).
Benartzi and Thaler (1995) combined two behavioral conceptsloss aversion
(Kahneman and Tversky (1979)) and mental accounting (Thaler (1985))to
provide a theoretical foundation for the observed equity premium puzzle.
Thaler et al. (1997), Gneezy and Potters (1997), and Gneezy, Kapteyn, and
Potters (2003) have all observed individual behavior consistent with the
Myopic Loss Aversion (MLA) conjecture.
Individuals are loss averse and often have myopic (short term) ways in which
they evaluate their portfolios. Both of these traits make them choose bonds
over equities.
It is assumed that with an increase in evaluation periods (i.e.- less myopic
decision making) may make people hold less of their wealth in bonds.

Equity-premium puzzle experiments

Haigh and List (JoF, 2005) conduct an experiment in lottery
choices and show that MLA is important in decisions made not
just by students but also by 54 professional futures and options
pit traders from the Chicago Board of Trade.
The traders display more MLA than the students.

Anchoring and reference points

Anchoring effects-Initial impressions become reference points
that anchor subsequent thoughts and judgments.
Salesperson has three items for sale-expensive, medium high
priced, and cheap. Show the customer the expensive item first,
which acts as an anchor. Makes it easier to sell the medium
high priced item.
Dramatic or easy-to-recall events often become strong anchors.
For example, the vividness of the horrible events of September
11 caused many to view airline travel as too risky, but many
experts believe that travel has never been safer.
When NYSE trading begins traders are more careful but over
the day their behaviour becomes more risk preferring and by
closing time they are almost risk neutral. Why is this
happening?

Framing and preferences

Tversky and Kahneman (1981) told people to

assume there was disease affecting 600 people
and they had two choices:
Program A, where 200 of the 600 people will be
saved .
Program B, where there is 33% chance that all
600 people will be saved, and 66% chance that
nobody will be saved.
They then offered them another two choices:
Program C, where 400 people will die, 200
people live.
Program D, where there is a 33% chance that
nobody will die, and 66% chance that all 600
people will die.

Asian Disease Problem

The majority of people selected A, showing a preference
for certainty or risk aversion.
Most people now selected D, seeking to avoid the loss of
400 people.
Notice how the framing makes the difference. Prospects
A and C are the same, and B and D are the same.

Framing the prospect as a gain makes people risk averse.

Framing the prospect as a loss makes people risk takers.

Status quo bias and endowment effect

The endowment effect is peoples tendency to value
something more highly when they own it than when they
dont
Example: experiment in which median owner value for mugs
was roughly twice the median non-owner valuation
Some economists think this reflects something fundamental
about the nature of preferences
Incorporating the endowment effect into standard theory
implies an indifference curve kinked at the consumers initial
consumption bundle
Smooth changes in price yield abrupt changes in consumption

Endowment Effect
Half the participants were given mugs available at the campus bookstore
for $6
The other half were allowed to examine the mugs
Each student who had a mug was asked to name the lowest sale price
Each student who did not have a mug was asked to name the highest
purchase price
Supply and demand curves were constructed and the equilibrium price
was obtained
Trade followed
There were four rounds of this

Bias Toward the Status Quo:

Default Effect
When confronted with many alternatives, people sometimes
avoid making a choice and end up with the option that is
assigned as a default
Example: Experiment showing that more subjects kept $1.50
participation fee rather than trading it for a more valuable
prize when the list of prizes to choose from was lengthened
Possible explanation is that psychological costs of decisionmaking rise as number of alternatives rises, increasing
number of people who accept the default
Retirement saving example illustrates the default effect when
the stakes are high

13-40

Default effect: retirement

Prior to April 1, 1998, the default option was
nonparticipation in the retirement plan
After April 1, 1998, all employees were by default enrolled
in a plan that invested 3% of salary in money market
mutual funds
Only the default option changed

Dynamic inconsistency
Hyperbolic discounting-people generally prefer smaller,
sooner payoffs to larger, later payoffs when the smaller
payoffs would be imminent; but when the same payoffs
are distant in time, people tend to prefer the larger, even
though the time lag from the smaller to the larger would
be the same as before.
When given a choice, some people would prefer $50
today to $100 one year from now, but would choose $100
six years from now versus $50 five years from now.

Examples

Lots of people want the IT dept. to withhold more

than they owe in taxes so they get a big refund check.
This behavior amounts to giving the IRS an interest
free loan.
School teachers who work 9 months are given the
option of receiving their salary over 9 months or over
12 months. Many choose the 12 monthly checks
because they dont trust themselves. They lose
interest income.
Before you choose a college think of the reputation
the college has: is it a diploma mill or does it require
hard work?
Most people prefer the college to have a good
reputation, but once they arrive, they often prefer
easy classes.

Poverty and cognition

Recent work by Shafir and Mullainathan (2013) Poverty and all
its related concerns require so much mental energy that the
poor have less remaining brainpower to devote to other areas
of life.
As a result, people of limited means are more likely to make
mistakes and bad decisions that may be amplified by and
perpetuate their financial woes.
Their work could explain a conundrum of public policy: If you
give an individual Rs. 1000, why does he not invest it (his
marginal utility of a Rupee is high) and instead buy alcohol?
Chakravarty and Warglien (unpublished, 2013) demonstrate
that poor people may be much more myopic in their evaluation
of time.

Minor temptations
Ariely and his colleagues gave thousands of people 20 number
problems. When they tackled the problems and handed in the answer
sheet, people got an average of four correct responses.
When they tackled the problems, shredded their answers sheets and
self-reported the scores, they told the researches they got six correct
responses. They cheated a little, but not a lot.

He put cans of Coke and plates with dollar bills in the kitchens of
college dorms. People walked away with the Cokes, but not the dollar
bills, which would have felt more like stealing.
He had one blind colleague and one sighted colleague take taxi rides.
The drivers cheated the sighted colleague by taking long routes much
more often than they cheated the blind one, even though she would
have been easier to mislead. They would have felt guilty cheating a
blind woman.

So
Given that these minor moral transgressions are in of
themselves largely innocuous there is no real reason for us to
correct them.
However if everyone in the population performs them, can we
end up with harmful social outcomes?
How can we prevent these ?

Dan Ariely (2012) The (Honest) Truth About Dishonesty.