Identity and economics
06/05/2021
1
Outline
Questions:
1. Is group identity relevant for economics?
2. Is identity good or bad?
Outline:
1. Part I: theory.
2. Let us play!
3. Part II: experiments on identity and economics
2
Part I
If we consider the standard economic agent of
microeconomics, social identity should not matter. However,
people perceive to belong to social groups and this gives
them identity:
1. Incentives related to membership to groups
2. Prejudices or cultural stereotypes
3. Something more?
3
Incentives
• Material incentives associated to cultural/language
difference
• Payoff structure → group conflict
• Policies that favour/disfavour certain groups →
foreigners, positive discrimination, etc.
Being part of a group can change the budget constraint of
individuals. Is this type of influence really related to identity?
4
Prejudices
Prejudices or cultural steoretypes can change how we interact
with other people:
• in cooperative interactions
• in principal-agent problems
Being part of a group and categorizing other people as part of
groups can change our beliefs on what the other person will
do.
5
Something more?
Simply being part of a group can affect behaviour by shifting
the utility function:
• Preferences
• Social norms
This often leads to the so-called ”ingroup bias”, where one
favours a subject of his own group over someone from
another group.
6
Social identity theory (Akerlof and Kranton, 2000, 2010)
General idea: individuals belong to groups, they internalize
context-related norms attached to their affiliations, and they
derive utility from compliance with these norms.
• Person One and Two belong to the same Blue group, and the norm of
the Blue group is to engage in Activity One.
• Persone One gets material utility V by engaging in Activity One and
Person Two gets material utility V by engaging in Activity Two.
• Person Two’s internalization of the norm causes her to suffer a loss in
utility Is if she does not comply with the norm of its group - i.e. if she
chooses Activity Two - where s stands for ”self”.
• Person One loses Io if Person Two choses Activity Two, where o stands
for ”other”. We can interpret this loss as an identity externality.
• Person One can stop Activity Two at a cost c, and Person Two loses L
from this action.
7
8
Four possible subgame perfect equilibrium outcomes:
• when c < Io and Is < V < Is + L, Person One will deter
Person Two from engaging in Activity Two.
• when c < Io and V > Is + L, Person One responds but will
not deter Person Two from engaging in Activity Two.
• when c > Io and V > Is, Person One does not respond and
Person Two engages in Activity Two.
• when V < Is, Person Two does not engage in Activity Two
regardless of Person One’s response.
9
A simple framework
One economic agent has to decide how much to contribute to
a public good. His utility U: Ui = Ui(NC, C)
Now, let us introduce group membership and distinguish
between C(own), C(other) and C(nogroup).
C(own) > C(other) → in-group bias or discrimination
10
A simple framework, ctd.
C(own) > C(other) = C(nogroup) → In-group love
C(own) = C(nogroup) > C(other) → Out-group hate
Conclusion I: welfare effects are positive if in-group love is
larger in magnitude than out-group hate.
11
A simple framework, ctd.
Hidden assumption was that interactions are random: there is
an equal chance of playing with an in-group as with an
out-group. Let us relax this assumption. z is now the chance
of playing with an in-group.
zC(own) − (1 − z)C(other) > C(nogroup) ⇒
z(C(own) − C(nogroup)) > (1 − z)(C(nogroup) − C(other))
Conclusion II: under homophily (z > 1/2), it is sufficient that
in-group love is equal to out-group hate that welfare effects
of group identity are positive.
12
Part II
Three lab experiments on social identity where we assess if
identity is actually relevant and if it is a force for good or for
bad:
1. Contest game with artificial and natural identities
(Chowdhury et al., 2016)
2. Public good game on prejudices (Castillo and Petrie, 2010)
3. Trust game with minimal group identities (Hargreaves
Heap and Zizzo, 2009)
13
The minimal group paradigm
1. By the group assignment rule, subjects are randomly
assigned to non-overlapping groups on the basis of some
trivial tasks.
2. No social interaction takes place between subjects.
3. Group membership is anonymous, no info is given on
group members
4. No link between the subject’s payoff and her choice.
Criterion 4 is violated in most economic experiments.
14
Chowdhury et al., 2016
An intergroup contest game with group identity. Each subject
i decides a level of ”effort” xgi over an endowment e where g
is the group. Let Xg =
∑
i xgi and X =
∑
g Xg. pg is the
probability of winning a prize V and πgi the payoff.
pg =



Xg/X if X ̸= 0
1/n otherwise
πg = pgV + e − xgi
What we already know (Abbink et al., 2010): in contests, subjects already
exhert more effort than predicted by the Nash Equilibrium (2x as high). In
intergroup contests, conflict expenditures are even larger than among
individuals (2x as high).
15
Chowdhury et al., 2016, ctd.
What if another layer of identity is added to the game?
• Color treatment: subjects are randomly assigned to Green
and Blue groups.
• Race treatment: subjects are assigned to groups based on
their ethnicity (Asian or Anglo-Saxon).
16
Chowdhury et al., 2016, ctd.
17
Castillo and Petrie, 2010
Subjects play 3 stages of a public good game of 10 rounds
each.
πi = e − Ci + α
∑
i Ci
where πi is the payoff at each round of player i, e her/his
endowment and Ci her/his contribution to the public good.
Nash equilibrium Ci = 0. Socially efficient outcome: Ci = e.
Who would you want to play this game with? Is identity
relevant? Subjects express their preferences on this before
Stage 3.
18
Castillo and Petrie, 2010, ctd.
Treatments manipulate the information:
• Contribution only: subjects are informed about the
contributions of each potential partner in the previous
stages.
• Photo only: subjects see a photo of each potential
partner.
• Photo and contribution: subjects see both.
19
Castillo and Petrie, 2010, ctd.
This is evidence of statistical discrimination (Arrow, 1973).
20
Hargreaves Heap and Zizzo, 2009
Only one subgame perfect equilibrium: the trustee gives back
zero to the truster, who in turn keeps everything for himself.
The socially efficient outcome: the truster gives every token
to the trustee.
21
Treatments
In experiments, in each treatment a choice parameter is
manipulated. In this way, the change in behaviour invoked by
specific manipulations will have causal interpretation.
1. C. color group assignment (Red vs. Blue)
2. SG. group segregation (2x with in-group than out-group)
3. SF. group segregation with reduced framing (Blue vs. Not
Blue)
4. SM. group segregation with reduced market
5. SI. group segregation with group incentives
22
Market phase
The experiment had 4 phases of 6 rounds each. After each
phase, subjects entered a market for buying or selling their
group membership.
1. They were given an endowment of 48 points and asked if they
hypothetically wanted to stay in their group or switch to the other.
2. Depending on the choice, they would become potential sellers or
buyers.
3. If they were switchers, they became potential buyers and had to state
their WTP between 0 and 48.
4. If they were stayers, they became potential sellers and had to state
their WTA between 0 and 48.
5. The market then operated as a Walrasian market (no. buyers = no.
sellers).
23
Results on discrimination - figures
24
25
The psychological value of the groups
26
Is identity relevant?
YES. Based on the evidence, the more groups are natural and
provided with meaning the more the influence of identity is
strong (e.g., Ben-Ner et al., 2009, Goette et al., 2012).
• Identity increases conflict (Chowdhury, 2020)
• It determines in-group love in public good games (Balliet
et al., 2014)
• It determines out-group hate in trust games (Balliet and
Van Lange, 2014)
27
Possible explanations
• We have seen Social Identity Theory. It originates from
social psychology (Tajfel and Turner, 1979). Individuals
categorize other individuals in groups, identify with them
and derive a sense of self-esteem from group affiliation.
• Individuals have generalized expectations of reciprocity
when they interact with individuals from their own group
(Brewer, 1999, Yamagishi and Mifune, 2000).
28
Is identity good or bad?
IT DEPENDS.
• From the interaction: 1) on matters of trust, identity
seems mostly bad and at best not very relevant, 2) on
pro-sociality, it seems to mostly increase aggregate
pro-sociality, 3) on conflict, it exacerbates it. Most of all,
it carries forward discrimination.
• However, also discrimination may be good sometimes.
Positive discrimination for example may be a way for
disadvantaged minorities to re-balance the score with
advantaged majorities.
29
Bibliography
Akerlof, G. A., & Kranton, R. E. (2010). Identity Economics: How
Our Identities Shape Our Work, Wages, and Well-Being.
Princeton University Press.
Charness, G., & Chen, Y. (2020). Social identity, group behavior,
and teams. Annual Review of Economics, 12, 691-713
Chen, Y., & Li, S. X. (2009). Group identity and social
preferences. American Economic Review, 99(1), 431-57
Chowdhury, S. M. (2021). The Economics of Identity and
Conflict. In the Oxford Research Encyclopedia of Economics
and Finance. Oxford University Press
30