Title: Trust as Individual Choice with Social Values, or Trust in the World
1Trust as Individual Choice with Social Values,
or Trust in the World
- Thomas B. Singh
- Feb. 5, 2008
- Research Seminar
- Dept. of Economics
- University of Kent
2Introduction
- Why is Trust studied, and How?
- Choice with Social Values - Social Interactions
(SI) - Individuals in a group tend to behave similarly
(Smoking, Crime, Educational Outcomes, etc.) - Individuals are more or less trusting as society
is more or less trusting - Contribution
- The Economics of Trust SI Framework
- The importance of Trust Norms
- Informal Institutions
3Informal Institutions
- Reference Group Country
- Expectations Interactions at country level
- How do informationally isolated agents
learn/form beliefs about the trust decisions of
others? How are beliefs transmitted/diffused? - Population Behaviour Trust in the World
- Beliefs about the trust stance others must
- be mutually consistent within countries and in
the world - explain/predict multiple (polarised stable
mean unstable) equilibria - SI Analytical NIE
4Informal Institutions
- Observation suggests that
- the propensity of an individual to behave in
some way varies with the prevalence of that
behaviour in some reference group containing the
individual (Manksi 1993) - But
- Can a researcher observing the distribution of
behaviour in a population infer whether the
average behaviour in some group influences the
behaviour of the individuals that comprise the
group? (Manski 1993) - To the extent that we can infer that Trust Norms
exert a statistically significant influence on
Trust Decisions, there is empirical evidence of
the importance of informal institutions.
5Baseline Model of SI
- The model assumes that
- where and
- y is the outcome variable
- x is a vector of group characteristics
- z, u are vectors of observable and unobservable
individual characteristics - Only Endogenous Effects produce a Social
Multiplier
Social Interactions Effects
6Baseline Model of SI
- Regression of y on x and z yields
- ß ? 0 gt endogenous effects due to average group
behaviour - ? ? 0 gt exogenous effects due to average group
characteristics - d ? 0 gt correlated effects due to unobservable
indiv. characteristics
(1)
7Equilibrium Identification
- Separating (ß, ?) from d ? and ß from ?
- Reflection problem due to E(yx) in (1)
- Group or social equilibrium is unique
- Only composite parameter identification
(2)
(3)
8Identification Possibilities
- Self-selection and identification
- Identification in a binary choice Manski model
- Random Assignment
- Brock Durlauf (2001) Random Utility Model with
Social Interactions (RUM-SI) - Mitigates the identification problem
- Accommodates multiple equilibria
9RUM-SI
- Individuals i 1, 2,, I get utility V(?i) from
their trust decisions ?i -1, 1, those of
all others ?-i, and a random utility term -
- u(.) and S(.) - private and social utility, resp.
- captures the social
interactions - are is expectations of others
trust decisions
(4)
10Strategic Complementarities
- Assume that ,
- is subjective belief about the average choice of
others - So
- Strategic Complementarities in trust decision
- Marginal returns to agent i of (say) trusting
increase as others are also believed to be
trusting
(5)
11A Stylised Trust Game
- NE Distrust, Distrust
- -1
- Suppose 1 emerges
- V(?i 1) now increases
12Repeated Trust Games
- Trusting becomes a regularity of behaviour due
to - Instrumental/informational conformity,
focal-ness, etc. - How does become a shared belief/common
knowledge among informationally isolated agents - How is this shared belief sustained?
- Off the path of play self-enforcing beliefs
- Deviation by i will induce country collapse in
trust
13Shared Beliefs Informal Institutions
- Institutions are systems of social variables
which conjointly generate a regularity of
behaviour (Greif 2004) - Shared values are embedded in the regularity of
behaviour cf. Granovetters embeddedness
Regularity of Behaviour
Behaviour
Social Variables/Shared Values/Expectations
Behavioural Internalised Beliefs
14Social Individual Utility
- Endogenous SI Effects Strategic
Complementarities among isolated agents - Facilitated by informal institutions
- Specific form of S(.) satisfying (5)
- (4) becomes
- Estimation the random utility term
(6)
15Trust as Individual Choice
- Assume that
- e(-1) - e(1) is logistically distributed
- The trust decision is made rationally
16Trust Decision Probabilities
- To motivate the empirical model
- Linearise u(?i) as h(?i) k
- u(1) h k and u(-1) -h k
- Assume that individuals have different private
utilities depending on some Xi and Yn(i) - Xi Observed individual-specific variables
- Yi Observed group-specific variables influencing
each group member
(7)
17Trust Decision Probabilities
- Thus hi b c'Xi d'Yn(i)
- And the individual trust decision probabilities
will become - given that u(?i) h(?i) k and
- assuming that ß 1
- Parameters b, c, d, J
(8)
18Model Closure
- Any set of observed choices can be reconciled
with the parameters - appropriate choice of subjective beliefs mie
- Beliefs should not only be shared, but mutually
consistent - In each country trust beliefs must be mutually
consistent - How is this possible?
19Informal Institutions
- Institution self-sustaining system of shared
beliefs about a salient way the game is played
repeatedly (Aoki, 2001) - Salience
- Boundedly rational players only observe the
regularity of behaviour - contains compressed information about the
equilibrium of the game - Self-Consistency
- Players hold beliefs consistent with that
equilibrium - Not just shared beliefs/common knowledge
20Model Closure Social Equilibrium
- Combining Salience and Self-Consistency and
Rational Expectations, the Social Equilibrium is - where m is the expected average choice level
- Multiple Equilibria can be accommodated
- To be expected given strategic complementarities
- We assume that this condition holds in the
empirical strategy
21The Empirical Strategy
- We have an expression for Pr(?ij Xij, Yij, mej)
- mej is not observable
- Our interest is in Pr(? X, Y, m) gt MLE
- Model Closure Self-Consistency, Salience, RE
22The Maximum Likelihood Estimator
- The Log Likelihood Function is
- ML gives us the parameter vector that finds for
each obs. a Pr(?ij Xij, Yij, mj) that
maximises - Pr(? X, Y, m)
23(No Transcript)
24Summary of Data
25Correlations
26Standard errors are given in parentheses. ()
indicates significance at the 5 (10) level.
27Concluding Comments
- The intial results show that
- Only Manski 3 RUM-SI have identical signs
- Formal Institutions are insignificant in RUM-SI
what matters is Confidence in Institutions - Endogenous Effects are significant
- Trust in the World
- High Trust Economies (Stable)
- Medium Trust Economies (Unstable)
- Low Trust Economies (Stable)
28Appendix Social Equilibrium
- From (7) the expected value of agent is trust
decision is - Where mei is is subjective belief about others
trust decisions.
29- By the assumption of common knowledge, all
individuals hold the same subjective beliefs me - By Self-consistency and Rational Expectations me
E(?i) E(?j) - By Salience E(?i) E(?j) m
- Hence the Social Equilibrium Condition