Status Quo Effects in Fairness Games: Reciprocal Responses to Acts of Commission vs. Acts of Omission

James C. Cox, ExCEN Georgia State

Maroš Servátka, NZEEL University of Canterbury

Radovan Vadovič, ITAM & UT Dallas

BEM 2012

Acts of commission vs. acts of omission

• Intensity of human reactions often depends on the perceived intent

• But intent could be hard to read out from actions

• In law and culture intent inferred from acts of commission and omission

– Status quo allows to distinguish whether an outcome achieved by taking a deliberate action (acts of commission) or by failing to do so (acts of omission)

• We study the extend of reciprocal reaction to actions that reveal kind or hostile intent versus those that do not

Reciprocal responses & status quo

• Do acts of commission which overturn the status quo, generate a stronger reciprocal response than acts of omission which uphold it?

• Scenario 1: Your initial wealth is $100K and John’s $100K

(a) John had an opportunity to give you $10K and did not do so. Would you want to punish him?

(b) Suppose now that John does give you $10K. Would you want to reward him?

• Scenario 2: Your initial wealth is $110K and John’s $90K

(c) John had an opportunity to take $10K from you but did not do so. Would you want to reward him?

(d) Suppose now that John does take $10K from you. Would you want to punish him?

Compare 1a with 2d and 1b with 2c

Status quo treatments

4

15,5T 10,10T

What does theory tell us?

• Most social preference models predict no difference in reciprocal responses between T15,5 and T10,10

– Economic man model, nonreciprocal distributional models (FS 1999, BO 2000), belief-dependent reciprocity models (DK 2004, FF 2006)

• Revealed Altruism (Cox, Friedman & Sadiraj, 2008)

– Axiom R: If FM provides a more generous opportunity set to SM then SM’s preferences will become more altruistic towards FM

Player B’s preferences will be more altruistic if Player A moves “right” than if he moves “left” in either treatment

– Axiom S: If the decision made by FM overturns the status quo then the reciprocal response, for individuals with preferences consistent with Axiom R, will be stronger than when the status quo is upheld

Axiom S implies that our treatments are not isomorphic

Testable hypotheses

• H0:The distribution of play across the four terminal nodes is the same in both treatments

• HA: Frequency of observation of nodes with payoffs (15,5) and (12,9) is greater in treatment T15,5 than in treatment T10,10

More “rewarding” and “punishment” following an act of commission than an act of omission

• Two ways of testing Axiom S:

– Using all data – Axioms R and S are defined as weak relations

– Nuanced test – only for subjects who have revealed consistency with the strict-preference version of Axiom R

Implementation of status quo

• The key to testing of Axiom S is a successful implementation of status quo

– In the field the status quo arises naturally

– In the lab – stylized decision problems in which subjects often lack clear ex-ante expectations

• We use three design features to induce status quo

1. Initial endowments – Discuss house money effect

2. Labeling of actions

• No change or give 5 in T15,5 vs. take 5 or no change in T10,10

3. Entitlements

• We have strengthened entitlements to initial endowments by making the subjects earn the money in Day 1 laborious task

• Discuss earning, time, possession

Experiment 1: EARNED Endowments

• Day 1: VB computerized multiple-choice quiz: 40 math GMAT questions– Score = # correct answers –1/4 (# incorrect answers)

– Scores ranked from highest to lowest and ties resolved randomly• Top 25% received an IOU certificate for $15• Middle 25-75%: $10 IOU certificate• Bottom 25%: $5 IOU certificate

• Day 2: Subjects who earned $15 or $5 invited to a T15,5 session, subjects who earned $10 to a T10,10 session

– Strategy method

– Double blind

Experiment 1 results – All Player Bs

No Decrease Decrease by 6 No Increase Increase by 2

All Player Bs

EARNED T15,5 (n=35)

77.1% 22.9% 54.3% 45.7%

EARNED T10,10

(n=35)

68.6% 31.4% 82.9% 17.1%

Fisher’s test 0.061 (two-sided for strategies)

Support for Axiom S

Experiment 1 results – subgames

No Decrease Decrease by 6 No Increase Increase by 2

All Player Bs

EARNED T15,5 (n=35)

77.1% 22.9% 54.3% 45.7%

EARNED T10,10

(n=35)

68.6% 31.4% 82.9% 17.1%

Fisher’s test 0.061 (two-sided)

Fisher’s testfor subgames

0.296 0.01

• A tougher test compares behavior between individual subgames

• Status quo effect might be stronger for positive reciprocity than for negative reciprocity.

Experiment 1 results – Reciprocal Player Bs

No Decrease Decrease by 6 No Increase Increase by 2

Reciprocal Player Bs

EARNED T15,5 (n=21)

61.9% 38.1% 23.8% 76.2%

EARNED T10,10

(n=15)

26.7% 73.3% 60% 40%

Fisher’s testfor subgames

0.039 0.032

• Strong test of the status quo effect on individuals who demonstrated that they have reciprocal preferences by selecting at least one decision to punish or reward

Strong support for Axiom S in both subgames

Experiment 2 – RANDOM Endowments

• Does the support for Axiom S hinge on strong entitlements?– Robustness check

• Control for math skills

• One-day experiment with RANDOM allocation of endowments

Experiment 2 results – All Player Bs

No Decrease Decrease by 6 No Increase Increase by 2

All Player Bs

RANDOM T15,5 (n=33)

78.8% 21.2% 63.6% 36.4%

RANDOM T10,10

(n=34)

68.6% 41.4% 94.1% 5.9%

Fisher’s test 0.000

Again strong support for Axiom S

Experiment 2 results – Subgames

•

No Decrease Decrease by 6 No Increase Increase by 2

All Player Bs

RANDOM T15,5 (n=33)

78.8% 21.2% 63.6% 36.4%

RANDOM T10,10

(n=34)

68.6% 41.4% 94.1% 5.9%

Fisher’s test 0.000

Fisher’s testfor subgames

0.067 0.002

Experiment 2 results – Reciprocal Player Bs

• We observe the same patterns as in Experiment 1

Robust support for Axiom S

Earning endowments not necessary

No Decrease Decrease by 6 No Increase Increase by 2

All Player Bs

RANDOM T15,5 (n=17)

58.8% 41.2% 29.4% 70.6%

RANDOM T10,10

(n=15)

26.7% 73.3% 60% 40%

Fisher’s testfor subgames

0.002 0.001

Discussion

• Subjects with reciprocal preferences are quite sensitive to acts of commission that overturn SQ

• In revealed altruism – SQ imposed exogenously

• Establishing SQ is the key to the empirical bite of Axiom S and the intensity of reciprocal reactions towards acts of commission

• Our experiment develops a procedure that identifies SQ naturally

– A combination of initial endowments and appropriate labeling of actions that framed the game in terms of monetary transfers.

– Actions of active behavior, such as giving or taking, stand in stark contrast with inaction

• Is this all it takes to establish SQ in general environments?

– Probably not

– Experience, habits, customs and norms likely to play an important role

– Future research