Social Norms and Naïve Beliefs Amrish Patel and Edward Cartwright Thursday 4 th September, IAREP/SABE World Meeting 2008, Rome, Italy.
Dec 31, 2015
Social Norms andNaïve Beliefs
Amrish Patel and Edward Cartwright
Thursday 4th September,IAREP/SABE World Meeting 2008,
Rome, Italy.
Why conform? Exogenous taste(Matsuyama, 1991)
Genetic fitness(Fudenberg and Maskin, 1990; Carmichael and MacLeod, 1997)
Informational conformity(Bikhchandani, Hirshleifer and Welch, 1998; Chamley, 2003)
Coordination benefits(Sugden, 1986; Young, 1996; Young, 2001)
Normative conformity(Akerlof, 1980; Bernheim, 1994; Nyborg and Rege, 2003)
Social Norms and Naïve Beliefs Patel IAREP/SABE 2008
Signalling models of conformity
Social Norms and Naïve Beliefs
SenderType is private information
e.g. generosity, innateproductivity, discount rate.
ReceiverObserves action and infers type;
some types are more valuedthan others.
Receiver offers sender a ‘reward’ or ‘punishment’ based on his inference,e.g. esteem or future favours.
Sender undertakes some observable action (signals)
Patel IAREP/SABE 2008
Equilibria and beliefs
Separating equilibria → no social norm Pooling equilibria → social norm existence Receivers inference of senders type critical Literature assumes Bayesian rationality Is this realistic? Lack of desire and/or ability to be Bayesian(e.g. cost-benefit, trust, information etc) Clearly not all receivers are Bayesian…
Social Norms and Naïve Beliefs Patel IAREP/SABE 2008
Naïve beliefs
The Fundamental Attribution ErrorOver-attributing observed behaviours to personal rather than situational
causes (Jones and Harris, 1967).
A Belief in the Law of Small NumbersExaggerating the extent to which a small sample represents the
population (Tversky and Kahneman, 1971).
Naïve beliefs: taking actions at ‘face value’. Compare equilibria where all agents have Bayes
rational beliefs with equilibria where a subset are Naïve.
Social Norms and Naïve Beliefs Patel IAREP/SABE 2008
A model of working hours Ingroup (e.g. colleagues), , and outgroup (e.g.
friends), , are sets with a continuum of agents. Each worker in chooses his working hours ,
from the set . Agent’s type, , his solution to textbook
labour supply problem is private information. Intrinsic U, ; max at . Esteem U, ; max at . Action function, .
Social Norms and Naïve Beliefs
IO
I x 0,20X
t T X
20.01g x t x t x t
20.01 10h b b 10b:T X
Patel IAREP/SABE 2008
A model of working hours Payoff function of a worker of type ,
are relative weights on esteem utilities.
A Signalling Equilibrium- Actions optimal given beliefs- All agents have Bayes rational beliefs. A Naïve Equilibrium- Actions optimal given beliefs- Ingroup have Bayes rational beliefs, Outgroup are naïve.
Social Norms and Naïve Beliefs
t
, , , , , .I O I OT TU t x g x t h b b x db h b b x db
&
Patel IAREP/SABE 2008
Result 1: Naïve beliefs weakennorm existence
Social Norms and Naïve Beliefs
Action functions when lambda = 0.1 and theta = 1
A pooling signalling equilibrium A separating naive equilibrium
Patel IAREP/SABE 2008
0 2 4 6 8 10 12 14 16 18 200
2
4
6
8
10
12
14
16
18
20
Type
Tim
e in
wor
k
Result 2: Naïve beliefs reduceset of possible norms
Exists a unique central norm Symmetric model so just
consider max norm Difference surface weakly
positive therefore,
Norm closer to social optimum
Social Norms and Naïve Beliefs
Difference between max forsignalling versus naïve equilibrium
0 2010 128Hours
px
max signalling eqm
max naive eqm .
p
p
x
x
Patel IAREP/SABE 2008
ExplanationResult 1 Naïve beliefs → shade actions towards middle. Bayes rational beliefs → imitate more desirable types. Naïve beliefs do still have a positive effect on conformity in
conjunction with Bayes rational beliefs.
Result 2 Esteem: Bayes rational (believed type); naïve (action). Only the former can give high esteem at a norm ≠10. Loss in intrinsic U from norm compliance is less likely to be
compensated at a norm ≠10 with naïve agents → deviate.
Social Norms and Naïve Beliefs Patel IAREP/SABE 2008
Conclusion Naïve beliefs are important: affect the existence of social
norms and encourage norms closer to the social optimum.
Further work: Naïve beliefs in evolutionary models of norms. Model a continuous degree of naivety, endogenise belief
type and let agents learn to be rational. Test for existence and effects of naïve beliefs
experimentally. Identify the class of signalling models for which naïve beliefs
act as an equilibrium selection device. Analyse a more complex esteem function.
Social Norms and Naïve Beliefs
Thank you
Patel IAREP/SABE 2008