state or nature: formal vs. informal sanctioning in a voluntary contribution game kenju kamei* louis...

45
State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University ** University of Copenhagen University of Vienna

Post on 19-Dec-2015

215 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

State or Nature:Formal vs. Informal

Sanctioning in a Voluntary Contribution Game

Kenju Kamei*

Louis Putterman*

Jean-Robert Tyran**

* Brown University ** University of Copenhagen

University of Vienna

Page 2: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Why State or Nature?• Hobbes (1588-1679) and Locke (1632-

1704) contrasted a state of nature ruled only by self-interest, with civilized life under a commonwealth, state, or civil government.

• Hobbes described the state’s relation to individuals in terms of “the introduction of that restraint upon themselves …[to] tie them by fear of punishment to the performance of their covenants.”

Page 3: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Locke argued that the state is necessary due to the unreliability of informal sanctions and the problem of counter-punishment:

… the irregular and uncertain exercise of the power every man has of punishing the transgressions of others, make them take sanctuary under the established laws of government … It is this makes them so willingly give up every one his single power of punishing, to be exercised by such alone, as shall be appointed to it amongst them; and by such rules as the community, or those authorized by them to that purpose, shall agree on. (§. 127.)

Page 4: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

If putting the power to punish in the hands of a central authority is the hallmark of civilization, why have we experimental economists been so focused on informal (horizontal, decentralized) rather than formal (vertical, centralized) sanctioning systems in our research of the past dozen years?

We asked ourselves:

Page 5: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• In Putterman, Tyran and Kamei (PTK, 2010) we began to study how groups would structure formal sanction schemes if given the opportunity.

• In the present paper, KPT (and also in Markussen, Putterman and Tyran), we continue this research program by conducting experiments in which subjects can not only construct formal sanction schemes but also decide by voting whether to use informal sanctions or formal ones.

• By “State or Nature” we mean “Formal or Informal sanctions.”

Page 6: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Conceptual Comparison of FS vs. IS

Formal Informal

PredictabilityPredictable:

Contributions to the “wrong” account are punished at a uniform rate.

Unpredictable:

Punishment depends on the decisions of individual group members.

TargetingEfficiently Targeted:

Punishment is only enacted on those who violate established rules.

Not Efficiently Targeted:

Perverse (or anti-social) punishment may occur, or subjects may choose not to punish.

CostsFixed Cost:

Little variable cost, since predictability reduces violations. But likely “up front” or fixed costs.

Variable Costs:

No fixed cost, but variable costs to convince subjects not to free ride are likely to be greater.

Kenju Kamei
You could improve on this table's format and look, Kenju, but I also now have an undergraduate summer RA whom I can ask to do that.--> I am going to ask Peter to improve the table' format
Page 7: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Conceptual Comparison of FS vs. IS (cont.)

Formal Informal

RationalityRational

Both purely self-interested individuals and those with pro-social preferences will support.

‘Irrational’

A self-interested agent wouldn’t incur the cost.

2nd Order Free Riding

Eliminated

Sharing in the burden of punishment is enforced.

Present

While some incur cost to punish, others contribute to group account but free ride on punishing.

FairnessFair

Easy to assure that the costs of the system are equally born.

Unfair

Possible for some to free ride on punishing done by others.

Page 8: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Real World Applications• There are “local public goods” problems on

the borderline of where state action can reach in a cost-effective fashion (for example, maintenance of a village woodlot in a developing country)—see Ostrom Nobel address (AER 2010).

• Organizations can choose between more formal rules vs. more voluntary forms of cooperation (example: adopt a strict dues system or ask supporters to contribute “whatever they feel they can”).

• Other applications in firms, etc.

Page 9: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Note: some real world settings mix formal and informal sanctions.

By treating formal and informal sanctions as substitutes rather than complements, we abstract from this.

Kube and Traxler (2010) study the combination of a non-deterrent formal sanction and informal sanctions (in exogenously imposed treatments).

Page 10: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Experimental framework• Our basic framework is the voluntary

contribution mechanism (VCM) or public goods game (PGG).

• For simplicity, we use a linear VCM played by partner groups in which endowments are equal.

• There are a known finite number of repetitions, for clarity of theoretical predictions.

• Formal or informal sanction schemes are added mainly endogenously (by vote), but sometimes exogenously.

Page 11: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• Our main focus is on whether subjects display a preference for formal or for informal sanction schemes when given a choice.

• We let them make these choices repeatedly so that we can observe any effects of experience and/or learning.

• In 6-Vote treatments, voting begins at the outset; in 3-Vote treatments, subjects undergo play with exogenous rules first, so that each vote (once votes begin) is by informed / experienced subjects. We compare results to test the robustness of our findings.

Page 12: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• For treatments in which subjects are first exposed exogenously to sanctioning schemes, we study both orders (formalinformal; informalformal) to check for robustness.

• The costliness of using a formal sanction scheme (“having a state”) varies across treatments.

• We have a baseline no sanctions treatment and we added as a later check a fully exogenous IS treatment to investigate whether endogenous choice of scheme has an effect in its own right.

Page 13: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Informal Sanctions (IS)• An informal sanction scheme in a VCM (PGG)

means each period has two stages: (1) a contribution stage, (2) after learning others’ contributions, each subject can punish others at a cost to him/herself.

• We have partner groups, finite repetition, scrambling of IDs (no indiv. reputation), and a fixed ratio of cost to punisher/versus cost to person targeted, as in Sefton, Shupp and Walker (2007), Carpenter (2007), Page, Putterman and Unel (2005), Nikiforakis and Normann (2008) and other papers.

• We use cost ratio 1:4, giving a relatively high punishment effectiveness.

Page 14: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Formal Sanctions (FS)• Under a formal sanction scheme, an individual

suffers an automatic earnings deduction for one or the other allocation decision. (In future work, we may add some uncertainty about punishment and also allow for erroneous sanctioning.)

• If contributing to one’s private account is what is penalized, and at a high enough rate, it becomes privately optimal to contribute only to the group account, hence the social dilemma vanishes (it’s 1st best to contribute full endowment).

Page 15: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Formal Sanctions (FS)• However, we let subjects vote on (a) whether allocating

to private accounts or to the group account is subject to sanction, (b) what the sanction rate is (it could be set at a low, “non-deterrent” level).

• Why? FS with endogenous terms is more comparable to IS in that subjects might choose an inefficient sanction scheme (punish the wrong action, or choose a non-deterrent sanction rate) due to confusion, ideological opposition, cost-cutting, or other factors. As in PTK (2010), we wanted to study how subjects vote and what explains any variation in voting.

• We give them no clue what they should do. The instructions are as neutral and non-interpretive as possible.

Page 16: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• Money lost to penalties is not redistributed (as is also true of informal sanctions in our own and most experiments ).

• All group members pay a small variable cost whenever a penalty is imposed (we set the ratio of var. cost of punishing, shared among the non-punished members, versus cost to the punished member to 1:4, paralleling our IS condition).

• Half the treatments impose a fixed cost to use the FS scheme equal to ¼ of the per period endowment (i.e., 5 points fixed cost vs. per period endowment of 20 points).

Formal Sanctions (FS)

Page 17: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• The fixed cost represents a fixed administrative cost of establishing and sustaining “a state” or formal structure, such as a police force, courts, prisons, etc., and is described in the instructions as “an administrative cost of having a fine scheme in place.”

• In the other half of treatments, there is no fixed cost of using the formal sanctions scheme.

Formal Sanctions (FS)

Page 18: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

A Few More Details…• Groups have 5 members• MPCR = 0.4

• There are 6 phases of 4 periods per phase• Each phase is preceded by a vote between IS and FS in

6-vote treatments.• In 3-vote treatments, phase 1 is played with no

sanctions, phases 2 and 3 with IS and FS in one or the other order, then subjects vote for either IS or FS for each of phases 4, 5 and 6.

Earning from Public AccountEarning from Private Account

Page 19: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

4 8 12 16 20 24

FS (IS) CS CS CSNS

Periods

IQ test and questionnaire about political preference

additional round 1.5

IS (FS)Timeline:

3-Vote Treatment

4 8 12 16 20 24

Periods

IQ test and questionnaire about political preference

additional round 0.5

0

CS CS CS CSCS CSNSTimeline:

6-Vote Treatment

NS: No Sanctions

FS: Formal Sanctions

IS: Informal Sanctions

CS: Chosen sanctions scheme

: Indicates a vote on method of sanction

Page 20: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• Subjects were drawn from the general undergraduate population at Brown University and interacted via computer in a computer lab

• The experiment was programmed in Z-Tree• An incentivized conditional contribution task (as in

Fischbacher-Gächter-Fehr and Fischbacher-Gächter) preceded the main part of the experiment.

• An incentivized intelligence test and survey questions on political orientation followed the main part of the experiment.

• The overall experiment took just under 2 hours; subjects earned an average of $31.58 including a $5 show-up fee.

• There were a total of 305 subjects in 8 treatments including a baseline treatment with 6 phases, 24 periods, no sanctions, and no voting.

Page 21: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Notes: NS = no sanctions, FS = formal sanctions, IS = informal sanctions. The experiment as a whole consisted of 16 sessions in which 61 groups consisting of 305 individual subjects participated. There were a total of 183 group votes on the use of FS versus IS, with 915 individual votes cast.

Louis_Putterman
Table pasting in slide can be improved.
Page 22: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Theoretical Predictions• Predictions are straightforward if one

assumes common knowledge of own-payoff maximizing behavior and unbounded rationality.

• Contributions to the group account, and informal punishment when available, are always zero unless there is a sanction of sufficient strength for allocating endowment to one’s private account, in which case all allocations go to the group accounts.

Page 23: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• Since the symmetric equilibrium with full contributions entails higher earnings for each than the symmetric equilibrium with zero contributions, voting to sanction contributing to the private accounts and to set a high penalty level is always a weakly dominant strategy (if fixed cost modest).

• We can assume subjects vote accordingly, for example by invoking trembling hand perfection.

(Theoretical Predictions, cont.)

Page 24: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

(Theoretical Predictions, cont.)

Having three voting stages* may complicate choice-theoretic predictions, so Markussen, Putterman & Tyran are studying treatments with only one voting stage, whereas the present paper allows testing predictions about what formal sanctions schemes subjects construct, at the cost of this added complexity.

* when FS chosen, i.e. 1. Formal/informal; 2. Public/private; 3. Penalty rate.

Page 25: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Behavioral Predictions• Informal sanctions will work better than standard

theory predicts because people do punish, due to negative reciprocity, inequity aversion, etc.

• Given this, subjects will choose between formal and informal sanctions on the basis of cost (e.g., vote against using formal sanctions when there’s a large fixed cost).

• The effectiveness of informal sanctions and of non-deterrent formal sanctions may be enhanced by the fact of being chosen by vote (cf. Tyran & Feld, 2006; Dal Bó, Foster, Putterman, 2010; Sutter, Haigner & Kocher, 2010).

Page 26: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

However, beware of perverse punishment …

• We know high contributors get punished also (Ostrom, Walker & Gardiner, 1992; Cinyabuguma, Page & Putterman, 2006; Herrmann, Thöni & Gächter, 2008) – a phenomenon probably related to counter-punishment (Nikiforakis, 2008; Denant-Boemont, Masclet & Noussair, 2007).

Page 27: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Main Results1. Which sanctions scheme do groups

prefer?Most groups choose informal over

formal sanctions when the latter entail a fixed cost, and most groups choose formal over informal sanctions when they do not.

Note: This and other results hold in both 6-Vote and 3-Vote treatments. (6-Vote subjects act in vote 1, Phase 1, like their more experienced 3-Vote counterparts in vote 1, Phase 4…as if they predict well!)

Page 28: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Choice of informal or formal sanctions (vote outcomes) in 3-vote and 6-vote treatments with and without administrative cost of FS

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

IS

FS

(n.a.) (n.a.)

3-Vote treatment without administrative cost

3-Vote treatment with administrative cost

6-Vote treatment without administrative cost

6-Vote treatment with administrative cost

Page 29: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Main Results2. What is sanctioned and at what rate,

when FS is chosen?Groups almost always select to

penalize contributing to the private (not public) accounts, and the sanction rate is usually but not always “deterrent” (1.2 or 0.8) rather than non-deterrent (0.4 or 0.0).

Page 30: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1.2 (priv.) 0.8 (priv.) 0.4 (priv.) 0.0 (priv.) 0.0 (pub.)

(n.a.) (n.a.)

3-Vote treatment without administrative

3-Vote treatment with administrative cost

6-Vote treatment without administrative cost

6-Vote treatment with administrative cost

Page 31: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Note: There are indications of a tendency to choose a higher fine rate in later periods of the two 6-Vote treatments (learning?) (based on regression analysis of the groups that always chose FS in those treatments).

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

1.2 (priv.) 0.8 (priv.) 0.4 (priv.) 0.0 (priv.) 0.0 (pub.)

(n.a.) (n.a.)

3-Vote treatment without administrative

3-Vote treatment with administrative cost

6-Vote treatment without administrative cost

6-Vote treatment with administrative cost

Page 32: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Main Results3. How does IS perform when chosen?

IS works much better than it “should” according to standard theory, usually achieving contributions and efficiencies not significantly different from those with formal sanctions and almost always signicantly higher than with NS.

Page 33: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Average contribution, all treatments

Page 34: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Average earnings under IS and FS in the two 6-Vote treatments (N = no administrative cost, C = administrative cost)

(a) 6-N treatment (b) 6-C treatment

Page 35: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Main Results

4. How much and what punishing occurs when IS is chosen?

The effectiveness of IS is associated with the fact that many subjects do incur the cost of punishing and most punishment is given to low contributors.

Page 36: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

0.0

0.4

0.8

1.2

1.6

2.0

0 2 4 6 8 10 12 14 16 18 20 22 24Period

Total Perverse

Informal Sanctions Given, Endogenous Condition

0.0

0.4

0.8

1.2

1.6

2.0

0 2 4 6 8 10 12 14 16 18 20 22 24Period

Total Perverse

Informal Sanctions Given, Exogenous Condition

We identify as “perverse” any punishment that is given to a group member who contributed more than the median amount during the period in question.

Page 37: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Main Results

5. Do contributions under FS depend on the sanction rate in the manner predicted by standard theory?

Not entirely. Average contribution under FS varies with the sanction level in a semi-continuous fashion, including:

Page 38: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

- more at fine of 1.2 than at fine of 0.8 although both should induce contribution of full endowment

- more at fine of 0.4 than at fine of 0.0 although 0.4 is a “non-deterrent sanction” that should not affect contribution for uniformly rational, self-interested subjects with common knowledge

- effectiveness of voted non-deterrent sanction of 0.4 is the most important departure from standard prediction

(contributions under formal sanctions)

Page 39: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Contributions under formal scheme by sanction rate

0

4

8

12

16

20

phase 1 phase 2 phase 3 phase 4 phase 5 phase 6

0.0 0.4 0.8 1.2 BASELINESanction rateContribution

Page 40: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

6. Does IS perform better when chosen than when assigned endogenously?

Yes. The average efficiency associated with informal sanctions is higher in Phase 2 of Treatment 6-C groups that chose IS than in “exogenous IS comparison treatment” in which IS is assigned to the subjects by the experimenters. The same applies in a parallel comparison with an exogenous IS treatment in Markussen et al.

Page 41: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

0

2

4

6

8

10

12

14

16

18

20

1 3 5 7 9 11 13 15 17 19 21 23

IS (endogenous)IS (Exogenous)

Period

Contribution

0

5

10

15

20

25

30

35

40

1 3 5 7 9 11 13 15 17 19 21 23Period

Earnings

Endogenous IS (4 groups) vs. Exogenous IS (6 groups)

Page 42: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

• However, the differences just shown are not statistically significant (based on M-W tests with group-level observations) in our study, whereas they are in the Markussen et al. study.

Page 43: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Brief, tentative conclusions• As in PTK, subjects can construct efficient

formal sanctions schemes.• But standard theory incorrectly predicts that

deterrent formal sanctions are always preferred to informal sanctions.

• The prediction errs because subjects in fact have the capacity to achieve informal cooperation if offered the power to punish one another.

• That power is mainly used to punish free riders.• Given actual behaviors under each scheme,

subjects sensibly choose between formal and informal sanctions based on their relative costs.

Page 44: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Caveat

• We do not claim to have fully addressed IS’s potential “Achilles heal”: retaliation.* This is because our IS treatment does not offer maximum opportunities for counter-punishment. In future research we hope to investigate how IS compares with FS when individuals are able to learn who punished them and to punish back. * See Nikiforakis (2008), “Punishment & Counter-Punishment: Can we still govern ourselves?”; also Denant-Boemont, Masclet, Noussair

Page 45: State or Nature: Formal vs. Informal Sanctioning in a Voluntary Contribution Game Kenju Kamei* Louis Putterman* Jean-Robert Tyran** * Brown University

Thank you.