Title: Trust and Betrayal: Evidence Across Nations Iris Bohnet, Fiona Greig, Benedikt Herrmann and Richard Zeckhauser Harvard University
1Trust and BetrayalEvidence Across NationsIris
Bohnet, Fiona Greig, Benedikt Herrmann and
Richard ZeckhauserHarvard University
- Conflict and Cooperation, 2007
- Northwestern University
- November 16, 2007
2Motivation
- Generalized trust in others associated with
good outcomes (social capital, economic
performance, democracy,) - But in most countries, fewer than 50 percent of
the respondents indicate that they generally
trust others (WVS 2004, Inglehart 2005). E.g.,
trust rates in English speaking countries
38 Protestant European countries
46 Islamic countries 28 - How can trust be fosteredin Gulf and Western
countries? - Internet
3MechanismChanging the expected benefits from
trusting
- Elasticity of trust
- How does the willingness to trust respond to
changes in the likelihood of trustworthiness? - How does the willingness to trust respond to
changes in the material costs of betrayal? - Aversion to the lottery of trust (TL-aversion)
- Risk aversion
- Inequality aversion (social preferences)
- Betrayal aversion
4Willingness to trust for given likelihoods of
trustworthiness (cumulative distribution) in US
and UAE
5Willingness to take risk
- Traditional decision analysis/economic
theoryAttitudes to risk - But what if the agent of uncertainty are people
rather than nature?Behavioral economics on
social versus natural risksAttitudes to
betrayal - Social risks include
- Speculative bubbles, HIV, terrorism
- Principal-agent relationships, asymmetric
information, incomplete contractsTrust
interactions
6Risks of betrayal versus natural risks
- Payoffs to other(s)Social preferences
- Agent of uncertaintyCausal attributions
(intentions) beyond outcome-based preferences - HypothesisHolding payoffs to others constant,
if people have an aversion to betrayal, they will
be less willing to take risks in a trust
situation than in an equivalent situation where
chance determines the outcome.
7Context
- Differences in societal, economic, political
organization and law suggest that trust is
produced differently in Western than in Gulf
countries. - Western countries Contractual arrangements.
- Gulf countries Loyalty.
81. Contracts
- Western law Damages for betrayal part of most
contractual arrangements. US contract law -
efficient breach - perfect expectation
damages victim of breach as well off
financially whether there is performance or
breach. ? Goal Decrease cost of betrayal. - Islamic law Damages unheard of. Gharar Risk
taking is prohibited under most circumstances.
Trust is risky. Damages would encourage risk
taking (gharar contracts). ? Goal
Eliminate likelihood of betrayal.
9Trust Game
Principal
- Principal confronted with a sure outcome and a
lottery, A and B.Principal does not know p, the
probability that Agent chooses 1 if Principal
chooses B.Probability that makes the lottery
actuarially fairp.29
B
A
Do Not Trust
Trust
Agent
1
2
p
1-p
P10 A10
P15 A15
P8 A22
10Experimental Design
- Three treatment conditions
- Binary-choice trust game (TG) Principal chooses
between sure thing and trusting an anonymous
agent. Agents decides whether to reward trust. - Binary-choice risky dictator game (RDG)
Principal chooses between sure thing and lottery
with identical payoffs to principal and inert
agent (recipient). - Binary-choice decision problem (DP) Principal
chooses between sure thing and lottery with
identical payoffs to principal. No second person
is affected.
11Risky Dictator Game (RDG) and Decision Problem
(DP)
Principal
- Principal chooses between sure thing and
lottery. The agent of risk is
nature.Principal does not know the probability
of getting the higher outcome in the lottery,
p.Principals decision affects Agents payoffs
in RDG but no second person is involved in DP.
B
A
Nature
p
1-p
P 10 A 10
P 15 A 15
P 8 A 22
12Measuring betrayal aversion
- Compare behavior in TG and RDG Attitudes to
betrayal. - Compare behavior in RDG and DP Social
preferences (incl. efficiency concerns). - Compare behavior in DP and probability that makes
lottery actuarially fair Risk preferences.
13MAP
- Introduce new methodology to measure willingness
to take risk under different circumstancesWith
MAPs (minimal acceptable probabilities) we
measure how high the likelihood of getting the
better outcome (15,15) minimally has to be for
the principal to choose the risky over the sure
outcome. - For example, in a binary-choice trust game, we
ask principals (in neutral language) Out of
100 people, how many would minimally have to be
trustworthy for you to be willing to trust??
MAP Minimal Acceptable Probability - Compare principals MAPs to p, the fraction of
trustworthy agentsWe simultaneously ask agents
(in neutral language)If your principal trusted
you, would you be trustworthy? (strategy method)
14Determination of payoffs
- If Principals MAP is higher than the actual
probability of trustworthiness, p, (gaining the
higher outcome in the gamble), then both players
get the sure option. - If Principals MAP is equal to or lower than the
actual probability of trustworthiness, p,
(gaining the higher outcome in the gamble), then
the agents decision (the lottery) determines
final payoffs. - Agents are not informed on MAP procedure.
- Incentive compatible mechanism people should
reveal their preferences truthfully. - p constant across treatment conditions.
Determined by the average p in the trust games,
which were run first. Subjects in RDG and DP
informed that p was determined beforehand.
15What we learn Net effects
0.29 is the MAP that makes a risk-neutral player
indifferent between the sure thing and the risky
option
16Experimental Details
- Design Between-subjects, one-shot, anonymous.
- Subjects To establish generality of betrayal
aversion, experiments run in very different
places. 614 students at universities in Brazil
(Rio), Switzerland (Zürich), United Arab Emirates
(Al-Ain), and United States (Boston). - Incentives Preserving PPP (practically
opportunity cost of time). - Instructions translated (and back-translated)
into respective languages.
17Result 1 Subjects are betrayal averse. MAPs in
TG and RDG MAPTGgtMAPRDG
18Betrayal Aversion by Country(no cross-country
differences in p0.33)
19Result 2 No sig. evidence of social
preferences.Result 3 Subjects are risk averse.
MAPDPgt0.29
20Social and risk preferences by country
21Bootstrap results for the three preference
phenomenaPercent of randomly drawn sub-samples
smaller than the experimentally observed country
sample
22Determinants of MAPs (OLS)
23Result 1 Willingness to trust (MAPs in )no
cross-country diff. in trustworthiness approx.
37
24Result 1 TGgenderMen MAPGulfgtMAPWest,
plt0.05, MW-test Women MAPUAE, OmangtMAPSWI, US,
Kuwait
25Result 2 Elasticity of trust (1)(dt/(1-t)/dw/
(1-w) -- percent reduction in those not trusting
over the percent reduction in those not being
trustworthy)
26Result 3 Responsiveness to Costs of Betrayal
MAPTGhigh gt MAPTGlow in USn.s. in Oman
27Table 3 Significance of betrayal aversion,
social preferences and risk aversion, by gender
and country, based on rank order tests.
significant at 5 significant at 1.
282. Loyalty(Arab Human Development Report 2004)
- 2. Societal organization Clannism/tribalism
obedience and loyalty are offered in return
for protection, sponsorship, and a share of the
spoils. (p. 145) - 3. Political system Authoritarian
- Clannism flourishes wherever civil or
political institutions that protect rights and
freedoms are weak or absent. Without
institutional supports, individuals are driven to
seek refuge in narrowly based loyalties that
provide security and protection (AHDR 2004
146). - 4. Economic structure (Oil) Rent-based
economygovernment as a generous provider that
demands no taxes or duties in return but is
entitled to require loyalty from its citizens
invoking the mentality of the clan. (p. 152)
Taxes account for approx. 5 of GDP in three
Gulf countries.
29Question
- Do these differences affect subjects
TL-aversion and the elasticity of trust? - ? Contract law (damages) increases trust by
decreasing the material costs of betrayal - ? Loyalty increases trust by decreasing/eliminat
ing the likelihood of betrayal
30Explanations for Betrayal Aversion
- Betrayal Costs
- Losses When p is Less than Reference Probability
31Betrayal Costs
Solving for MAP, we have
When a Decision Maker gets the bad outcome in the
Trust game (when his trust is betrayed), he
incurs an additional negative element. We call
this a betrayal cost. This implies that,
even when the monetary payoffs are the same in
the two games. A lower UB implies that MAP in
trust game is higher.
32(No Transcript)
33Conclusions
- People are less willing to take risk when the
agent of uncertainty is another person rather
than nature betrayal aversion. ? Behavioral
economics People do not only care about
outcomes but also about how outcomes come to be
(e.g., Rabin 1993). - People are slightly more willing to take risk
when another person benefits from risk taking
positive social preferences and generally are
risk averse. - ? Behavioral economics People care about
payoffs to others and/or about efficiency
(e.g., Fehr and Schmidt 2002 for a survey). - The RDG and the MAP elicitation procedure may
prove to be useful innovations to examine
willingness to take risk under different
conditions.
34Conclusions, continued
- Cross-regional differences in TL-aversion and
elasticity of trust. ? Changes in the
likelihood or the material cost of trust
strongly affect behavior in the Western but
hardly in the Gulf countries. - There are plausible methodological explanations
for betrayal aversion.