let’s (not) escalate this! intergroup leadership in a team ...€¦ · et al., 2009; potters et...
TRANSCRIPT
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Let’s (not) escalate this!Intergroup leadership in a team contest
Florian Heine 1 Arno Riedl 2
1Tilburg School of GovernanceTilburg School of Economics and Management
Tilburg University
2Department of EconomicsSchool of Business and Economics
Maastricht University
February 2017
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Introduction
I Research on leadership repeatedly finds an efficiencyenhancing effect of a leader in public goods games or groupproduction both in the lab and in the field (Kosfeld andRustagi, 2015; Arbak and Villeval, 2013; van der Heijdenet al., 2009; Potters et al., 2007)
I Large class of situations where groups are not isolated, but insome form of competition (cf. Dawes, 1980)
I In order to outperform competitors, considerable resources arespent without direct productive value
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Introduction
I Research on leadership repeatedly finds an efficiencyenhancing effect of a leader in public goods games or groupproduction both in the lab and in the field (Kosfeld andRustagi, 2015; Arbak and Villeval, 2013; van der Heijdenet al., 2009; Potters et al., 2007)
I Large class of situations where groups are not isolated, but insome form of competition (cf. Dawes, 1980)
I In order to outperform competitors, considerable resources arespent without direct productive value
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Introduction
I Research on leadership repeatedly finds an efficiencyenhancing effect of a leader in public goods games or groupproduction both in the lab and in the field (Kosfeld andRustagi, 2015; Arbak and Villeval, 2013; van der Heijdenet al., 2009; Potters et al., 2007)
I Large class of situations where groups are not isolated, but insome form of competition (cf. Dawes, 1980)
I In order to outperform competitors, considerable resources arespent without direct productive value
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Welfare enhancing effect of leaders: Public Good
Average team effort levels and earnings (van der Heijden, Potters, and Sefton,2009)
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Welfare enhancing effect of leaders
I In a team contest, things might very well look differentI ingroup / outgroup leadership trade-off, where strong
leadership comes to the detriment of outgroup relations(Pittinsky and Simon, 2007)
I Intergroup leadership: Encouraging contact between grouprepresentatives (Allport, 1954)
⇒ Can a leader curtail over-contribution and improve groupwelfare in a team competition for a public good?
⇒ Can intergroup leadership reduce between-group conflict andincrease social welfare?
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Welfare enhancing effect of leaders
I In a team contest, things might very well look differentI ingroup / outgroup leadership trade-off, where strong
leadership comes to the detriment of outgroup relations(Pittinsky and Simon, 2007)
I Intergroup leadership: Encouraging contact between grouprepresentatives (Allport, 1954)
⇒ Can a leader curtail over-contribution and improve groupwelfare in a team competition for a public good?
⇒ Can intergroup leadership reduce between-group conflict andincrease social welfare?
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Concepts
Contest game
I Two groups compete against each other
I One group gets the prize, the other one nothing
I Winning probability depends on tokens invested
Leadership
Leading-by-example One player moves before the others, setting abenchmark. No formal authority.
Transactional leadership Leader is in direct exchange relationshipwith other players. Leader in a hierarchicallysuperordinate position.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Concepts
Contest game
I Two groups compete against each other
I One group gets the prize, the other one nothing
I Winning probability depends on tokens invested
Leadership
Leading-by-example One player moves before the others, setting abenchmark. No formal authority.
Transactional leadership Leader is in direct exchange relationshipwith other players. Leader in a hierarchicallysuperordinate position.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Concepts
Contest game
I Two groups compete against each other
I One group gets the prize, the other one nothing
I Winning probability depends on tokens invested
Leadership
Leading-by-example One player moves before the others, setting abenchmark. No formal authority.
Transactional leadership Leader is in direct exchange relationshipwith other players. Leader in a hierarchicallysuperordinate position.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Baseline
I 15 periods in partnermatching
I Endowment E = 120tokens per period
I Individual prizez = 480 tokens forplayers in the winninggroup
I Individuals get toknow how much othergroup matescontribute
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Treatments
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Baseline
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup transactional leader
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup transactional leaderHeine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Treatments
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Baseline
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup transactional leader
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup transactional leaderHeine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Treatments
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Baseline
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup transactional leader
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup transactional leaderHeine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Treatments
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Baseline
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup transactional leader
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup transactional leaderHeine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Setup – Treatments
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Baseline
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup leading-by-example
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Ingroup transactional leader
1
2
3
4
1
2
3
4
∑k∈K
vk∑k∈K
vk+∑
m∈Mvm
Win
Leaders communicate
Intergroup transactional leaderHeine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Equilibrium Strategies
I Individual expected payoff πi(vi) = E +
vi+∑k 6=ik∈K
vK
vi+∑k 6=ik∈K
vK+∑
m∈MvM· z − vi
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Equilibrium Strategies
I Individual expected payoff πi(vi) = E +
vi+∑k 6=ik∈K
vK
vi+∑k 6=ik∈K
vK+∑
m∈MvM· z − vi
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Four stylised facts
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
First Interior solution (cf. public goods games)
Second Overall welfare maximising strategy is for nobody tocontribute and let chance decide
Third Under risk neutrality, equilibrium contribution staysthe same per contest party for every treatment
Fourth Either leader contributes zero, the followers fully shipin, or leader only contributor
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Four stylised facts
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
First Interior solution (cf. public goods games)
Second Overall welfare maximising strategy is for nobody tocontribute and let chance decide
Third Under risk neutrality, equilibrium contribution staysthe same per contest party for every treatment
Fourth Either leader contributes zero, the followers fully shipin, or leader only contributor
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Four stylised facts
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
First Interior solution (cf. public goods games)
Second Overall welfare maximising strategy is for nobody tocontribute and let chance decide
Third Under risk neutrality, equilibrium contribution staysthe same per contest party for every treatment
Fourth Either leader contributes zero, the followers fully shipin, or leader only contributor
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Four stylised facts
Equilibrium predictions
Leader Followers total Followers each if symmetric
Baseline 30 90 30Ingroup leading-by-example 0 120 40Ingrouptransactional leader 120 0 0Intergroup leading-by-example 0 120 40Intergrouptransactional leader 120 0 0
Nash equilibrium under risk neutrality.
First Interior solution (cf. public goods games)
Second Overall welfare maximising strategy is for nobody tocontribute and let chance decide
Third Under risk neutrality, equilibrium contribution staysthe same per contest party for every treatment
Fourth Either leader contributes zero, the followers fully shipin, or leader only contributor
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Alternative Hypotheses
I Leadership behaviour as benchmark for followersI Leader’s contribution as signal or sacrifice (Hermalin, 1998;
Potters et al., 2001; Meidinger and Villeval, 2002)I Transactional leader followed out of self-interest (De Cremer
and Van Vugt, 2002)
I Normative rules (Reuben and Riedl, 2013)I High level of efficiency for own groupI Maximising sum of payoffs of all players across groups
I Social preferences
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Alternative Hypotheses
I Leadership behaviour as benchmark for followersI Leader’s contribution as signal or sacrifice (Hermalin, 1998;
Potters et al., 2001; Meidinger and Villeval, 2002)I Transactional leader followed out of self-interest (De Cremer
and Van Vugt, 2002)
I Normative rules (Reuben and Riedl, 2013)I High level of efficiency for own groupI Maximising sum of payoffs of all players across groups
I Social preferences
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Alternative Hypotheses
I Leadership behaviour as benchmark for followersI Leader’s contribution as signal or sacrifice (Hermalin, 1998;
Potters et al., 2001; Meidinger and Villeval, 2002)I Transactional leader followed out of self-interest (De Cremer
and Van Vugt, 2002)
I Normative rules (Reuben and Riedl, 2013)I High level of efficiency for own groupI Maximising sum of payoffs of all players across groups
I Social preferences
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Procedures
I Recruited 360 participants using ORSEE (Greiner, 2004)(9 pairs of groups per treatment)
I Experiment programmed in z-Tree (Fischbacher, 2007) andconducted at BEElab, Maastricht NL
I Each session took about 80 minutes, including reading theinstructions, a trial period, a questionnaire and payment
I Average earnings e 11.75 (About $12.80 or £8.90 at the timeof the experiment)
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Contribution to the contest
100
200
300
400
Ave
rage
gro
up c
ontr
ibut
ion
Baseline Ingroup lbe Ingroup trans Intergroup lbe Intergroup trans
I Without between groupcommunication, leadersprompt an increase incontest expenditures.
I Allowing forcommunication betweengroup leaders cushionsthis effect and results ina contest expenditurelevel which is comparableto the baseline.
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Contribution over the Periods
4060
8010
0In
divi
dual
Con
tribu
tion
1 5 10 15Period
Leading by Example (lbe)
4060
8010
0
1 5 10 15Period
Transactional Leader
Ingroup (no communication) Intergroup (Leaders chat)Baseline Symmetric NE
Generally:
I Negative trend for lbetreatments
I Positive trend fortransactional treatments
I no trend for Baselinetreatment
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
First Round Contribution
020
4060
80F
irst m
ove
Lead
er a
nd F
ollo
wer
Baseline Ingroup lbe Ingroup trans Intergroup lbeIntergroup trans
Leader Follower
I Very small treatmentdifference on impact
I Only Intergroup lbetreatment significantlylower
I Leaders have noescalating effect on thegroup contest in round 1
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Scatterplot
020
4060
8010
012
0
0 20 40 60 80 100 120Leader Contribution
R-squared: 0.514
Leading by Example
020
4060
8010
012
0
0 20 40 60 80 100 120Leader Contribution
R-squared: 0.364
Transactional Leader
Ingroup
020
4060
8010
012
0
0 20 40 60 80 100 120Leader Contribution
R-squared: 0.372
Leading by Example
020
4060
8010
012
0
0 20 40 60 80 100 120Leader Contribution
R-squared: 0.531
Transactional Leader
Intergroup (with communication)
Fol
low
er C
ontr
ibut
ion
I A higher contribution bythe leader is paralleled bymore followercontributions across alltreatments
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Transactional Treatments – Sunflower Plot
0
200
400480
600
800
1000
Priz
e R
ecei
ved
from
Lea
der
-100 -50 0 50 100Contribution Relative to Leader
Reallocation 1 petal = 1 obs.1 petal = 23 obs.
Two fields of attrition:
I Contribute as Leaderdoes
I Reallocate 480
Also:
I North-western domainnearly empty
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Transactional Treatments – Sunflower Plot 2
-100
-50
0
50
100
Con
tribu
tion
Rel
ativ
e to
Lea
der
0 200 400 480 600 800 1000Prize Received from Leader previous period
single obs. 1 petal = 1 obs.1 petal = 20 obs.
Similar pattern:
I Contribute as Leaderdoes
I Reallocate 480
Now:
I South-eastern domainnearly empty
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
The chat contents
Leader Contribution asFunction of Chat Contents:
I Leaders’ chat contentshave predictive power
I i.e. Leaders coordinatingon contributing low(high) end up buying less(more) Lottery tickets
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
The chat contents
010
020
030
0
Low contribution
Medium
contribution
High contribution
Alternate
Bonding
Small talk
Understanding
Efficiency
Followers' behaviour
Give much to followers
Give little to followers
Other
Intergroup lbe Intergroup trans
I More mitigatingmessages sent inIntergroup lbe treatment(and vice versa)
I This contributes atexplaining the strongtreatment differences
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Conclusion
I This study provides the first (experimental) investigation onthe role of leaders in a team contest
I Our results substantiate a considerably pessimist view overallI Most leadership types prompt an escalation of the contestI Leaders tend to incentivise their teammates to ship in
resources to the competition
I Intergroup leadership (via free form text communication)between leaders of competing groups can contribute atabating the rat race
I significant improvement with respect to the baseline levelI Leaders who manage to coordinate through the chat on taking
turns at shipping in resources are most successful in guidingtheir groups towards more efficient play
Heine & Riedl Let’s (not) escalate this!
ConceptsSetup
Equilibrium Strategies / HypothesesResults
Conclusion
I This study provides the first (experimental) investigation onthe role of leaders in a team contest
I Our results substantiate a considerably pessimist view overallI Most leadership types prompt an escalation of the contestI Leaders tend to incentivise their teammates to ship in
resources to the competition
I Intergroup leadership (via free form text communication)between leaders of competing groups can contribute atabating the rat race
I significant improvement with respect to the baseline levelI Leaders who manage to coordinate through the chat on taking
turns at shipping in resources are most successful in guidingtheir groups towards more efficient play
Heine & Riedl Let’s (not) escalate this!
References
Literature
G. W. Allport. The Nature of Prejudice. Addison-Wesley, Reading,MA, 1954.
E. Arbak and M.-C. Villeval. Voluntary leadership: motivation andinfluence. Social Choice and Welfare, 40(3):635–662, 2013.ISSN 0176-1714. doi: 10.1007/s00355-011-0626-2. URLhttp://dx.doi.org/10.1007/s00355-011-0626-2.
R. M. Dawes. Social dilemmas. Annual Review of Psychology, 31(1):169–193, 1980. doi: 10.1146/annurev.ps.31.020180.001125.URL http://www.annualreviews.org/doi/abs/10.1146/
annurev.ps.31.020180.001125.
Heine & Riedl Let’s (not) escalate this!
References
Literature (cont.)
D. De Cremer and M. Van Vugt. Intergroup and intragroupaspects of leadership in social dilemmas: A relational model ofcooperation. Journal of Experimental Social Psychology, 38(2):126 – 136, 2002. ISSN 0022-1031. doi:10.1006/jesp.2001.1499. URL http://www.sciencedirect.
com/science/article/pii/S0022103101914994.
U. Fischbacher. z-tree: Zurich toolbox for ready-made economicexperiments. Experimental Economics, 10(2):171–178, 2007.
B. Greiner. The online recruitment system orsee 2.0. A guide forthe organization of experiments in economics. University ofCologne, 2004.
Heine & Riedl Let’s (not) escalate this!
References
Literature (cont.)
B. E. Hermalin. Toward an economic theory of leadership: Leadingby example. The American Economic Review, 88(5):pp.1188–1206, 1998. ISSN 00028282. URLhttp://www.jstor.org/stable/116866.
M. Kosfeld and D. Rustagi. Leader punishment and cooperation ingroups: Experimental field evidence from commons managementin ethiopia. American Economic Review, 105(2):747–83,February 2015. doi: 10.1257/aer.20120700. URL http:
//www.aeaweb.org/articles?id=10.1257/aer.20120700.
C. Meidinger and M.-C. Villeval. Leadership in Teams: Signaling orReciprocating ?, Dec. 2002. URLhttp://halshs.archives-ouvertes.fr/halshs-00178474.Working Paper du GATE 2002-13.
Heine & Riedl Let’s (not) escalate this!
References
Literature (cont.)
T. L. Pittinsky and S. Simon. Intergroup leadership. TheLeadership Quarterly, 18(6):586 – 605, 2007. ISSN 1048-9843.doi: http://dx.doi.org/10.1016/j.leaqua.2007.09.005. URLhttp://www.sciencedirect.com/science/article/pii/
S104898430700121X. The Leadership Quarterly Yearly Reviewof Leadership.
J. Potters, M. Sefton, and L. Vesterlund. Why announceleadership contributions?: An experimental study of thesignaling and reciprocity hypotheses. Tilburg University, 2001.
J. Potters, M. Sefton, and L. Vesterlund. Leading-by-example andsignaling in voluntary contribution games: an experimentalstudy. Economic Theory, 33(1):pp. 169–182, 2007. ISSN09382259. URL http://www.jstor.org/stable/27822588.
Heine & Riedl Let’s (not) escalate this!
References
Literature (cont.)
E. Reuben and A. Riedl. Enforcement of contribution norms inpublic good games with heterogeneous populations. Games andEconomic Behavior, 77(1):122 – 137, 2013. ISSN 0899-8256.doi: http://dx.doi.org/10.1016/j.geb.2012.10.001. URLhttp://www.sciencedirect.com/science/article/pii/
S0899825612001492.
E. van der Heijden, J. Potters, and M. Sefton. Hierarchy andopportunism in teams. Journal of Economic Behavior &Organization, 69(1):39 – 50, 2009. ISSN 0167-2681. doi:10.1016/j.jebo.2008.09.007. URLhttp://www.sciencedirect.com/science/article/pii/
S0167268108001844.
Heine & Riedl Let’s (not) escalate this!