Journal of Economic Psychology 34 (2013) 7887

Contents lists available at SciVerse ScienceDirect

Journal of Economic Psychology

journal homepage: www.elsevier.com/locate/joep

An experimental comparison of incentive contracts in

partnerships
Hong Chao a,, Rachel T.A. Croson b
a

Department of Economics, Antai College of Economics & Management, Shanghai Jiao Tong University, 535 Fahua Zhen Rd., Shanghai 200052, China
Department of Economics, School of Economic, Political & Policy Sciences, University of Texas at Dallas, 800 W. Campbell Drive, GR31, Richardson, TX
75080-3021, USA

a r t i c l e

i n f o

Article history:
Received 15 September 2011
Received in revised form 21 November 2012
Accepted 29 November 2012
Available online 10 December 2012
PsycINFO classication:
3660
Keywords:
Partnership
Incentive
Contract
Synergy

a b s t r a c t
Empirical work comparing individualized sharing and equal sharing schemes in partnerships has produced mixed results. Some studies nd individualized sharing schemes superior, others nd no difference, and still others nd equal sharing schemes superior. This
paper outlines a theory which reconciles these competing ndings, and tests it with an
experiment. We nd that in conditions of high synergy (when the teammates effort has
a proportionately larger impact on an agents output than the agents own effort), equal
sharing schemes outperform individualized sharing schemes, while in conditions of low
synergy, individualized sharing schemes outperform equal sharing schemes. These results
are consistent with observations from the eld. Our results have the potential to guide
rms choosing between competing compensation contracts by identifying situations under
which each contract type is likely to yield increased productivity.
2012 Elsevier B.V. All rights reserved.

1. Introduction
Partnerships are commonly observed in rms, representing 100% of the top 100 law rms, 56% of the top 100 accounting
rms, and 18% of the top 100 architecture rms (Greenwood & Empson, 2003). Labor contracts within partnerships take
varying forms.
Two common forms involve individualized sharing schemes and equal sharing schemes.1 As the names suggest, in individualized sharing schemes the compensation of partnership members is their own individual output,2 while in equal sharing
schemes, their compensation is a function of the team output. In practice, however, individual output is often a function not
only of ones own efforts, but of others efforts as well (Alchian & Demsetz, 1972). This effect is referred to as team synergy (Lawford, 2003).
Synergy effects can be frequently observed in partnerships. In a medical partnership, for example, the number of surgeries
a doctor can perform (and be paid for) may depend on their partners availability for collaborations. Although synergy is a
common feature of partnerships, its impact on the efciencies of various incentive contracts is still vaguely understood and
has not been systematically examined in the literature (Alchian & Demsetz, 1972; Rose, 2002).
Corresponding author. Tel.: +86 21 52301562; fax: +86 21 32231111.
E-mail addresses: chaohong@sjtu.edu.cn (H. Chao), crosonr@utdallas.edu (R.T.A. Croson).
Also known as individual output contracts, individual-based contracts, or independent performance pay, and team output contracts, team-based contracts
or joint performance pay.
2
Note that individualized sharing schemes are slightly different than piece rates. In a piece rate compensation scheme, the rm takes a share of the prots;
in partnerships all prots are distributed to the partners.
1

0167-4870/$ - see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.joep.2012.11.009

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

79

Observational and experimental evidence comparing the effectiveness of equal sharing and individualized sharing has
been mixed, with some studies nding individualized sharing superior (Encinosa, Gaynor, & Rebitzer, 2007; Gaynor & Gertler, 1995; Nalbantian & Schotter, 1997), others nding no difference (Dijk, Sonnemans, & Winden, 2001; Vandegrift & Yavas,
2011), and still others nding equal sharing superior (Chan, Li, & Pierce, 2012; Hamilton, Nickerson, & Owan, 2003; Pizzini,
2010).
In this paper we provide an organizing explanation for these mixed results. We show both theoretically and experimentally that when team synergy is high (characterized by a large degree of complementarity between ones own effort and the
effort exerted by teammates), equal sharing schemes outperform individualized sharing schemes. When team synergy is
low, the opposite is true.
The intuition for our result is straightforward. Equal sharing schemes internalize effort externalities but include incentives for free riding. When effort externalities are sufciently high, the former effect outweighs the latter and equal sharing
schemes outperform individualized sharing schemes. When effort externalities are not sufciently high, the free-riding
incentives outweigh the benet from internalizing the effort externality, and individualized sharing schemes outperform
equal sharing schemes.
Our results are consistent with previous observations from the eld, including the prevalence of equal sharing in highsynergy specialties (for example, emergency medicine) and the dominance of individualized sharing in low-synergy specialties (for example, psychiatry) (Adams, 2006; Pauly, 1996; Pizzini, 2010). Finally, our results have the potential to guide partnerships choosing between competing compensation contracts by identifying situations under which each contract type is
likely to yield increased productivity.
2. Previous work
Observational work has documented the prevalence of both individualized and equal sharing schemes in partnerships.
For example, Encinosa et al. (2007) report that 38.3% of medical groups use equal sharing schemes, while the rest use individualized sharing schemes. We argue that these types of partnerships have varying levels of synergies, and that this can
explain the differential use of the various incentive schemes.
Both observational and experimental data has compared the efciency properties of these schemes, with mixed results.
Some studies have found that equal sharing schemes underperform individualized sharing schemes. For example, Encinosa
et al. (2007) nd that for large medical groups, team productivity is lower under equal sharing, while Gaynor and Gertler
(1995) nd that equal sharing schemes generate fewer ofce visits for doctors. Nalbantian and Schotter (1997) run a lab
experiment in which equal sharing schemes generate signicant free-riding and lower levels of productivity than individualized sharing schemes. Erev, Bornstein, and Galili (1993) nd similar results by using a real-world task of picking oranges.
The samples investigated in the former two studies on medical partnerships mainly consist of specialties of low synergy.3
The experimental settings of the latter two studies similarly do not allow for the possibility of production interdependence
among participants, thus involve no synergies. We argue that the lack of synergies results in equal sharing underperforming
individualized sharing.4
In contrast, other studies have found that equal sharing schemes outperform individualized sharing schemes. Hamilton
et al. (2003) nd a 14% increased productivity in a garment plant that switched from individual output to team output contracts. Chan et al. (2012) and Pizzini (2010) nd similar effects in a eld experiment run in department stores cosmetic counters and in medical partnerships respectively. Hamilton et al. (2003) investigate manufacturing workplaces, which usually
involve high synergy levels due to the extensive production interdependence in their operational lines (Adams, 2006). Chan
et al. (2012) nd that the advantage of equal sharing over individualized sharing increases if peer productivity spillovers increase. Pizzini (2010) nds that equal sharing schemes are more prevalent in the specialties with substantial production
interdependence (e.g., anesthesiology, radiology, and general surgery). These results are thus consistent with our argument
that equal sharing is superior under high synergy.5

3
Pauly (1996) and Pizzini (2010) argue that doctors sometimes engage in teamwork (for example, collaborate to perform complex surgeries or administer
emergency care), depending on their specialties. Pizzini (2010) rates major medical specialties in terms of production interdependence (see Table A-1, Pizzini,
2010). Using Pizzinis rating, we nd that only about 8% partnerships investigated in Encinosa et al. (2007) and 1 out of 4 specialties investigated in Gaynor and
Gertler (1995) involve high synergy, while the rest involve low synergy.
4
Two previous experiments demonstrate no differences between contractual types in settings of no synergies. Dijk et al. (2001) run an experiment in which
participants solve a two-variable optimization problem. Vandegrift and Yavas (2011) run an experiment using a forecasting task. However, in both studies
participants were grouped and provided with their teammates performance under equal sharing but not individualized sharing. It is possible that this
information induced competitive preferences, raising efforts in equal sharing but not individualized sharing settings (Dijk et al., 2001; Gneezy, Niederle, &
Rustichini, 2003; Vandegrift & Yavas, 2011).
5
What might cause conditions of high (or low) synergy? Hamilton et al. (2003) suggest that heterogeneity of ability within a team is one such cause, as more
skillful workers can teach the less skillful how to execute tasks efciently, yielding signicant long-term increases in productivity. Evidence on this question,
however, is mixed. Two experiments involving no producting interdependence (and thus no synergies) but with heterogeneity of worker ability generate
different results. Meidinger, Rullire, and Villeval (2003) report that equal sharing performs poorly in this setting, while Vandegrift and Yavas (2011), report
that heterogeneity of ability increases the efciency of equal sharing. This increased efciency, however, is only found in men, and (as mentioned in footnote 4)
might be driven by competitive preferences. While not the focus of this paper, future work is needed to understand the relationship between ability
heterogeneity and performance under different contractual terms.

80

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

Previous theoretical work has attempted to explain these conicting results. Itoh (1991) shows that a team output contract is optimal when own effort and helping effort are complementary, but not (always) when they are substitutes. Demougin and Fluet (2006) show that when agents are envious of each others wages, team contracts can outperform individual
contracts, because under team contracts wages are equalized, employees do not suffer disutilities due to envy, and are thus
willing to work for less (also see Englmaier & Wambach, 2010). Kvaloy and Olsen (2006) use settings of incomplete information and allow for peer monitoring. They show that when there is little common noise, peer monitoring allows team contracts to outperform individual contracts. In contrast, when there is a lot of common noise, the cost of team contracts is high
and thus individual contracts are preferred.6 This paper provides and tests a different organizing explanation for the competing
empirical results, based on team synergy (Chao & Siqueira, 2013).
In our model (below), individuals choose a level of costly effort to exert. Effort inuences both their own and their partners output. We dene low synergy as a scenario where an agents output is determined primarily by the effort exerted by
that agent and the effort exerted by the agents teammate has little effect. High synergy, on the other hand, is characterized by
a large degree of complementarity between ones own effort and that exerted by ones teammate; in fact in our high synergy
treatment, the teammates effort has a proportionately larger impact on an agents output than the agents own effort.7 We
make this characterization more specic and formal in Section 3, below, where we present our theoretical model. We use a simple linear production technology and show experimentally that equal sharing will outperform individualized sharing under conditions of high synergy but not under conditions of low synergy.
Our team production setting is also related to the Voluntary Contributions Mechanism (VCM) game with asymmetric
marginal per capita return (MPCR) and the addition of a stochastic shock to output (Vandegrift & Yavas, 2011).8 In these
experiments, the benet generated from the public account to the contributor (the internal return) is not equal to that to
the other group members (the external return). Goeree, Holt, and Laury (2002) and Laury and Taylor (2008) nd that both internal return and external return generate increased contributions to the public good although the former effect is stronger. To
relate our work to the VCM framework, one can imagine that an equal sharing scheme is a mechanism in which individuals
internalize part of the external return of their contribution but demands that they share part of their internal return with others.
3. A simple model and hypotheses
3.1. Model of production
Consider the case of two homogeneous risk-neutral partners working in a partnership. Each partner i chooses an effort
level ai. This effort generates output, yi according to the following function:

yi fown ai fother aj ei ;

j 1; 2; and i j

fown and fother (which are both nonnegative) describe the marginal product of effort on ones own output and on the partners
output. When fown > fother, we consider the setting to exhibit low levels of synergy. When fown < fother, we consider the setting
to exhibit high levels of synergy. The random term ei is uniformly distributed over the interval e; e and assumed to be
i.i.d. for both partners.
Efforts increase productivity, but are also costly. In particular, we assume that the cost of effort is quadratic:

Cai ca2i

i 1; 2

3.2. Social optimum

We rst calculate the socially optimal level of effort.

maxEU i EU j Eyi yj  Cai  Caj

ai ;aj

other
other
yields the socially optimal effort level aSO fown f
and the socially optimal payoff level U SO fown f
for each partner. As
2c
4c
with any setting involving an externality, the rst-best solution equates each partners marginal cost of effort with the
marginal product of that partners effort on the total team output.

6
Other studies examining the effect of peer monitoring in teams include, but are not limited to, Kandel and Lazear (1992)and Che and Yoo (2001), and Mas
and Moretti (2009).
7
Workers in a team may possess various skills and perform different but highly interdependent tasks. The externalities generated by workers efforts on
others may thus be amplied in others production. For example, a physician spends 10 min to help a surgeon to make a plan for a gastric surgery. This 10-min
advice may save the surgeon 10 h if she had done the job alone. Such high synergy situations are not uncommon in highly interdependent workplaces (Adams,
2006; Lasker, Weiss, & Miller, 2001; Pizzini, 2010).
8
See Ledyard (1995) for an extensive survey on VCM literature.

81

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

Table 1
Experimental parameters and equilibrium predictions.
a e {0, ..., 150}, e e [40, 40], c = 0.1, 2-subject teams, 15 rounds
Low synergy (fown = 10, fother = 2)
Equilibrium effort (NE)
Socially-optimal effort (SO)
High Synergy (fown = 10, fother = 18)
Equilibrium effort (NE)
Socially-optimal effort (SO)

Individualized sharing

Equal sharing

50
60

30
60

50
140

70
140

3.3. The individualized sharing scheme

The simplest contractual setting one can imagine is one in which each partner receives wages equivalent to their own
output. The partners expected payoff is then

maxEU ind
Eyi  Cai
i
ai

1
which generates equilibrium effort and resulting payoff of aind fown
and U ind 4c
fown fown 2f other .
2c
As expected in settings with positive externalities, the equilibrium level of effort is smaller than the socially optimal level.

3.4. The equal sharing scheme

A more complex but often-observed compensation scheme involves partners sharing the total team production. Now
each partners expected payoff is

maxEU equal

i
ai

1
Eyi yj  Cai
2

1
3
which generates equilibrium effort and resulting payoff of aequal 4c
fown fother and U equal 16c
fown fother 2 . As can be seen,
the comparison between equilibrium efforts in these two schemes will vary depending on the size of the synergy.
This theory provides an organizing explanation for competing results from the previous literature. We nd that (1) in
low-synergy settings (when fother < fown), the individualized sharing scheme generates higher effort than the equal sharing
scheme, aind > aequal; (2) in high-synergy settings (when fother > fown), the equal sharing scheme generates higher effort than
the individualized sharing scheme, aequal > aind. These two comparisons generate our hypotheses H1 and H2.

4. Experimental design and parameters

Our experiment involves a 2  2 design: we vary sharing scheme type (individualized and equal) and synergy levels (high
and low). Experimental participants were randomly assigned to one of the four treatments, thus both synergy level and sharing scheme type were between-subject factors.
We consider teams of two participants who simultaneously choose effort levels from the set of integers {0, 1, . . ., 150}. The
marginal product of effort on ones own production, fown is set to be 10 in all treatments. The marginal product of effort on
ones partners production, fother, is 2 in the low synergy treatment and 18 in the high synergy treatment. The random term e
is independently and uniformly distributed over the integer interval [40, 40], and we set c (the cost of effort) to 0.1. The
equilibrium predictions generated by these parameters are depicted in Table 1.
The experiment was conducted in the Smith Experimental Economics Research Center at the Shanghai Jiao Tong University in China. The experiment was programmed using z-Tree (Fischbacher, 2007). Participants were recruited using the online recruiting system. We ran eight experimental sessions in total with two sessions for each treatment. One hundred and
ninety students from multiple disciplines participated in the experiment; 22 in one session of low-synergy/individualizedsharing and 24 in each of the remaining seven sessions.
Participants played 15 rounds in each session under a single treatment. We used a stranger design in which participants
were randomly matched with a different counterpart in each round (Andreoni & Croson, 2008). After each round, the computer randomly selected an integer between [40, 40] for each partner to represent e. Participants were reminded of their
own decision, told of the random number drawn for them, their own output and their counterparts output, their own earnings for the previous round and their cumulative earnings.9 A history table with this information for previous rounds was provided during the session. Participants were not allowed to communicate other than via the program. An example of the
translated version of instructions can be found in the Appendix.

9
In the equal sharing treatment participants can infer their counterpart output from their own earnings, thus we provided this information in both the
individualized and equal sharing treatments to ensure parallelism.

82

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

Table 2
Descriptive statistics of average efforts.a
Average efforts (standard deviation) [number of observations]

Round 1

Rounds 115

Rounds 315

Round 15

Low synergy individualized sharing (NE effort: 50; SO effort: 60)

75.21
(28.14)
[n = 43]

55.39
(15.78)
[n = 668]

53.39
(12.23)
[n = 581]

51.09
(7.18)
[n = 46]

Low synergy equal sharing (NE effort:30; SO effort: 60)

77.16
(36.39)
[n = 43]

44.82
(23.80)
[n = 691]

42.11
(20.55)
[n = 608]

34.38
(12.41)
[n = 48]

High synergy individualized sharing (NE effort: 50; SO effort: 140)

77.80
(35.08)
[n = 40]

61.62
(26.18)
[n = 690]

59.96
(24.27)
[n = 604]

58.50
(26.80)
[n = 48]

High synergy equal sharing (NE effort: 70; SO effort: 140)

84
(36.46)
[n = 43]

77.82
(24.14)
[n = 688]

77.39
(22.77)
[n = 600]

75.56
(20.14)
[n = 48]

a
We have excluded as observations any effort levels that were randomly chosen by the computer because participants did not make decisions within
their time limit.

In order to ensure that the experiments would be completed in a reasonable amount of time, participants were given a
xed amount of time to make each decision (2 min for each of the rst 2 rounds and 30 s for each of the remaining 13
rounds). If they did not make a decision within the given time, the computer randomly chose an effort level for them from
a uniform distribution between 0 and 150. This happened rarely (113 times out of 2850 decisions).10 In our main results reported below, we drop these auto-selected observations.
After the experimental decisions, participants completed a post-experimental survey. After the survey, participants were
paid anonymously based on their total earnings. Earnings were calculated in the ctitious currency eckels and were translated into RMB using a xed and commonly-known exchange rate. Average earnings were 39.72 RMB (including a show-up
fee of 5 RMB), and sessions lasted on average 50 min.
5. Results
5.1. Descriptive results
Average effort levels, standard deviations and number of observations are reported in Table 2. In order to help illuminate
learning in this setting, we report effort levels for the rst round, the entire 15 rounds, rounds 315 and the last round.
Inspection of Table 2 demonstrates that efforts in all four treatments begin higher than the equilibrium predictions but decrease over time.
5.2. Comparing behavior and equilibria
We use a two-tailed Wilcoxon signed-rank test to statistically compare observed and predicted effort levels in the four
treatments. Our dependent variable is the effort levels of each subject averaged over a certain number of rounds. The results
of these tests are reported in Table 3.
As seen in Table 3, average effort levels are signicantly different than equilibrium predictions for all comparisons,
although the differences are reduced as the rounds progress. Figs. 1 and 2 demonstrate this convergence graphically. Average
efforts in low synergy treatments (Fig. 1) and high synergy treatments (Fig. 2) begin higher than their predicted levels, but
generally decrease over the course of the 15 rounds.
5.3. Comparing treatments
Our primary concern, however, involves the effort levels chosen within a synergy setting between contractual types. We
rst use a two-tailed nonparametric Wilcoxon test to do the comparisons. Our dependent variable is the effort levels of each
subject averaged over a certain number of rounds. Results are reported in Table 4.
10
One concern raised by reviewers is that experimental participants might have been aware that the decisions of their matched counterpart were, in fact,
auto-selected. The lock-step nature of the experimental implementation (all participants waited until everyones decision had been made, before moving to the
next round) meant that a participant could never know that their counterparts decisions were made automatically. Even if participants believed that they were
playing with a partner whose action was auto-selected, their optimal effort level (the best response to an effort choice uniform [0-150]) would be the same as
predicted here. While beliefs about auto-selection may matter in a repeated game, in our experiment participants played with different counterparts in each
round and thus, a belief about auto-selection should not affect their behavior. As a result, while we drop the auto-selected decisions, we do not drop the
decisions of their paired counterparts.

83

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

Table 3
Comparison between observed and equilibrium efforts.a
Wilcoxon signed-rank test (two-tailed)

P-values [number of observations]

Round 1

Rounds 115

Rounds 315

Round 15

Low synergy individualized sharing vs. NE (50)

<0.001
[n = 43]

<0.001
[n = 46]

<0.001
[n = 46]

0.023
[n = 46]

Low synergy equal sharing vs. NE (30)

<0.001
[n = 43]

<0.001
[n = 48]

<0.001
[n = 48]

0.010
[n = 48]

High synergy individualized sharing vs. NE (50)

<0.001
[n = 40]

<0.001
[n = 48]

<0.001
[n = 48]

0.027
[n = 48]

High synergy equal sharing vs. NE (70)

0.019
[n = 43]

0.007
[n = 48]

0.001
[n = 48]

0.008
[n = 48]

a
We have excluded as observations any effort levels that were randomly chosen by the computer because participants did not make decisions within
their time limit.

average effort

100
80

Individualized
Sharing

60

Equal Sharing

40

NE
(Individualized
Sharing)

20

NE (Equal
Sharing)

0
1

9 10 11 12 13 14 15

round
Fig. 1. Average efforts in the low synergy condition.

average effort

100
80

Individualized
Sharing

60

Equal Sharing

40

NE
(Individualized
Sharing)

20

NE (Equal
Sharing)

0
1

9 10 11 12 13 14 15

round
Fig. 2. Average efforts in the high synergy condition.

As hypothesized, we nd that effort levels in low synergy conditions are signicantly different under equal sharing than
under individualized sharing. We also nd that effort levels in high synergy conditions are signicantly different under equal
sharing than under individualized sharing.
To conrm the above ndings, we run OLS regressions using individual efforts over the 315 rounds and including dummies for the treatment (our coefcient of interest), for each session and for each subject, and a control for the period. Results
are reported in Table 5.11
This analysis conrms our previous results and supports both our hypotheses. We see a statistically signicant and
positive coefcient on the individualized sharing scheme dummy in the low synergy treatments, as predicted. Thus in

11
The OLS regressions on the data of 15 rounds for the specication described in Table 4 return similar results. These observations are censored at 0 and 150
(the minimum and maximum possible efforts). However, the numbers of observations at these boundaries are small in all treatments (between 0.6% and 2.6%).
Tobit regressions return similar results.

84

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

Table 4
Nonparametric tests of sharing schemes.a
Wilcoxon test (two-tailed)

Z-scores (individualized sharing equal sharing)

(P-values)
[number of observations]
Round 1

Rounds 115

Rounds 315

Round 15

Low synergy:
Equal sharing < individualized sharing

0.219
(0.827)
[n1 = n2 = 43]

4.570
(<0.001)
[n1 = 46, n2 = 48]

4.686
(<0.001)
[n1 = 46, n2 = 48]

6.660
(<0.001)
[n1 = 46, n2 = 48]

High synergy:
Equal sharing > individualized sharing

0.886
(0.376)
[n1 = 40, n2 = 43]

5.163
(<0.001)
[n1 = n2 = 48]

5.194
(<0.001)
[n1 = n2 = 48]

5.739
(<0.001)
[n1 = n2 = 48]

a
We have excluded as observations any effort levels that were randomly chosen by the computer because participants did not make decisions within
their time limit.

Table 5
Individual efforts by sharing scheme.a
Individual effort

Low synergy

High synergy

Individualized sharing
Period
Session dummies
Subject dummies
Constant
Number of observations
Adjusted R2

5.66**(0.41)
0.92**(0.11)
YES
YES
56.15**(1.09)
1189
0.383

8.77**(0.52)
0.71**(0.14)
YES
YES
75.21**(1.36)
1204
0.492

**

p < 0.01
We have excluded as observations any effort levels that were randomly chosen by the
computer because participants did not make decisions within their time limit.
a

low synergy conditions efforts under individualized sharing are higher than efforts under equal sharing (H1). We see a statistically signicant and negative coefcient on the individualized sharing scheme dummy in the high synergy treatments, as
predicted. Thus in high synergy conditions, efforts under individualized sharing are lower than efforts under equal sharing
(H2). We further replicate our ndings of decreasing efforts over time (signicantly negative coefcient on period), as seen in
Figs. 1 and 2.
6. Conclusion
This paper theoretically and experimentally compares worker behavior in partnerships under conditions of high and low
synergy and team and individual incentives. We nd, as predicted, that efforts are signicantly higher under individualized
sharing in conditions of low synergy, and under equal sharing in conditions of high synergy.
These results have important implications for our understanding of labor contracts and the compensation structures of
partnerships. They help us to explain the puzzling frequency (and apparent success) of team incentive schemes in the presence of the incentive to free-ride. They allow us to predict when a partnership will choose team incentives (in conditions of
high synergy) or individual incentives (in conditions of low synergy). Finally, our results make active recommendations for
what types of contracts rms should choose, given their level of synergy.
Future work is needed to reinforce our understanding about the effect of synergy on contract selection. One possible
direction is to construct a measure of synergy level in practice. Previous studies have identied a few determinants of synergy including heterogeneity of worker ability (Chan et al., 2012; Hamilton et al., 2003), heterogeneity of worker specialty
(Lasker et al., 2001; Pizzini, 2010), presence of assembly line work, need to cross-train workers, high absence costs, high job
attachment (Brown, Geddes, & Heywood, 2007; Heywood & Jirjahn, 2009; Heywood, Jirjahn, & Wei, 2008), and level of social
capital (Bandiera, Barankay, & Rasul, 2010; Gant, Ichniowski, & Shaw, 2002). However, a comprehensive synergy evaluation
mechanism is still needed, and may require involvement of both academia and industry.
Another possible extension of our experiment is to include comparisons of other contract types under various synergy
levels. For example, Chao and Siqueira (2013) provide theoretical predictions for the comparison between team output contracts and mixed contracts, which could be experimentally tested.
Finally, as discussed in Section 2, peer monitoring is a competing explanation for the differential performance of these
two contractual types. In our experiment we hold monitoring constant (always present) to focus on the impact of synergy,
but in the eld both may be relevant. For example, equal sharing schemes are more likely to be adopted by partners who

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

85

have known each other for a long period of time. Two reasons suggest themselves. First, long-term relationships generate a
high level of synergy due to the establishment of social capital (Bandiera et al., 2010; Gant et al., 2002). But alternatively,
long-term relationships make monitoring less costly (Kandel & Lazear, 1992; Mas & Moretti, 2009). Future work might investigate the relative importance of these competing explanations.
In conclusion, this study adds to our understanding of conditions under which individuals will exert effort or shirk in
partnership settings. Ultimately this improved understanding has the potential to increase our ability to engage in contract
design and elicit optimal effort levels from partners in a variety of settings.
Acknowledgements
We gratefully acknowledge an Associate Editor and two anonymous referees for helpful comments. We are grateful to the
Center for Behavioral and Experimental Economic Science and The Negotiations Center at the University of Texas at Dallas,
and the Smith Experimental Economics Research Center at Shanghai Jiao Tong University for providing nancial and logistical support.
Appendix A. Example of translated version of Chinese instructions
A.1. General instructions for low [high] synergy
Welcome to the Smith Experimental Economics Research Center. You are now participating in a session of an Economic
Decision-Making Experiment.
There are 15 rounds of experiment in the following setting. In each round, you will be randomly and anonymously
matched with a counterpart who is another participant in this room and then make one decision. After each round, you will
be matched with a different counterpart. Therefore, you will be matched with 15 different counterparts during the whole
session. Your earnings will depend on the decisions you make, the decisions your counterparts make, and chance. If you follow the instructions carefully and make good decisions, you could earn a considerable amount of money.
Please do not talk, exclaim, or otherwise communicate with the other participants during the session. Interactions with
your counterpart will take place through the computer program. If you have a question, please raise your hand and a monitor
will come to you to answer your question privately.
You will be making one decision in each of the 15 rounds. You will have a xed amount of time to make each of your
decisions which will be noted at the top of each screen; if you do not make a decision within that time, the computer will
randomly choose a decision for you. For the rst two rounds you will have 2 min to make each decision. Starting in round
three you will have 30 s to make each decision. After each decision, you will learn how much you earned.
When everyone in the room is nished with their decisions in one round, you will be matched with a different counterpart for the next round. Remember, you will have a new counterpart in each of the 15 rounds.
The decisions you make are anonymous; no other participant in the room will know your decisions (or your earnings)
unless you choose to tell them. You will never learn the identity of your counterpart, nor will they ever learn who you are.
Your earnings will be calculated in a ctitious currency called eckels. At the end of the session today, you will be paid one
RMB for each 300 [1500] eckels earned, in addition to your show-up fee.
In each round of the following setting you and your counterpart will simultaneously and independently choose a number
between 0 and 150. You can choose any integer (any whole number, no decimals or fractions). After you and your counterpart enter your numbers, the computer will randomly select two numbers between 40 and 40; one for you and one for your
counterpart. Each number between 40 and 40 is equally likely to be selected by the computer. Your payoffs will depend on
the numbers you and your counterpart choose, and the random numbers selected by the computer.
We will rst use these numbers to calculate your and your counterparts output:

Your output 10  your choice 218  their choice your random number:
Counterparts output 10  their choice 218  your choice their random number:
Note that the number you choose will affect not only your own output, but also the output of your counterpart. Similarly,
the number they choose will affect their output as well as your own.
Choosing higher numbers increases your (and your counterparts) output. However, choosing higher numbers is more
costly. Your and your counterparts costs are:

Your cost 0:1  your choice2 :

Counterparts cost 0:1  their choice2 :
Note that the number you choose will affect your costs but not your counterparts costs, and vice versa.
Your payoffs will be calculated using a certain combination of your output, your counterparts output, and your own costs.
The specic form of combination will be described later.

86

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

To calculate your earnings, we add your earnings (in eckels) across all the decisions you made, then multiply by the exchange rate of 300 [1500] eckels = 1 RMB. You will be paid this amount, plus your show-up fee, at the end of the session.
Are there any questions before we continue? If so, please raise your hand.
A.2. Specic instructions for low [high] synergy, individualized sharing scheme
Reminder:
You have been matched with a counterpart in this round that you have never been matched before. You and your counterpart each choose a number between 0 and 150. The computer will randomly select a number between 40 and 40 for
each of you. Your and your counterparts outputs are:

Your output 10  your choice 218  their choice your random number:
Counterparts output 10  their choice 218  your choice their random number:
Your and your counterparts costs are:

Your cost 0:1  your choice2 :

Counterparts cost 0:1  their choice2 :
Your earnings will be your output, minus your own costs. Those earnings are thus:

your output  your cost

This simplies to:

10  your choice 218  their choice your random number  0:1  your choice2 :
Please enter your choice from (0150):________.
A.3. Specic instructions for low [high] synergy, equal sharing scheme
Reminder:
You have been matched with a counterpart in this round that you have never been matched before. You and your counterpart each choose a number between 0 and 150. The computer will randomly select a number between 40 and 40 for
each of you. Your and your counterparts outputs are:

Your output 10  your choice 218  their choice your random number:
Counterparts output 10  their choice 218  your choice their random number:
Your and your counterparts costs are:

Your cost 0:1  your choice2 :

Counterparts cost 0:1  their choice2 :
Your earnings will be the average of your and your counterparts outputs, minus your own costs. Those earnings are thus:

1=2  your output counterparts output  your cost

This simplies to:

614  your choice their choice av erage of the two random numbers  0:1  your choice2 :
Please enter your choice from (0150):________.
References
Adams, C. (2006). Optimal team incentives with CES production. Economics Letters, 92(1), 143148.
Andreoni, J., & Croson, R. (2008). Partners versus strangers: Random rematching in public goods experiments. In C. Plott & V. Smith (Eds.), Handbook of
experimental economics results. Vol. 1. Amsterdam: North-Holland.
Alchian, A., & Demsetz, H. (1972). Production, information costs, and economic organization. American Economic Review, 62(5), 777795.
Bandiera, O., Barankay, I., & Rasul, I. (2010). Social incentives in the workplace. Review of Economic Studies, 77(2), 417458.
Brown, M., Geddes, L., & Heywood, J. (2007). The determinants of employee involvement schemes: Private sector Australian evidence. Economic and
Industrial Democracy, 28(2), 259291.
Chan, T. Y., Li, J., & Pierce, L. (2012). Compensation and peer effects in competing sales teams (working paper). Olin Business School, Washington University,
St. Louis, MO http://dx.doi.org/10.2139/ssrn.1367441.
Chao, H., & Siqueira, K. (2013). The Importance of Mixed Incentive Contracts in Partnerships. The International Journal of Economic Theory, 9.
Che, Y., & Yoo, S. (2001). Optimal incentives for teams. The American Economic Review, 91(3), 525541.

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 7887

87

Demougin, D., & Fluet, C. (2006). Group vs. individual performance pay when workers are envious. In D. Demougin & C. Schade (Eds.), Contributions to
entrepreneurship and economics, rst Haniel-Kreis meeting on entrepreneurial research. Berlin: Duncker & Humblot.
Dijk, F., Sonnemans, J., & Winden, F. (2001). Incentive systems in a real effort experiment. European Economic Review, 45(2), 187214.
Encinosa, W., III, Gaynor, M., & Rebitzer, J. (2007). The sociology of groups and the economics of incentives: Theory and evidence on compensation systems.
Journal of Economic Behavior & Organization, 62(2), 187214.
Englmaier, F., & Wambach, A. (2010). Optimal incentive contracts under inequity aversion. Games and Economic Behavior, 69(2), 312328.
Erev, I., Bornstein, G., & Galili, R. (1993). Constructive intergroup competition as a solution to the free rider problem: A eld experiment. Journal of
Experimental Social Psychology, 29(6), 463478.
Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171178.
Gant, J., Ichniowski, C., & Shaw, K. (2002). Social capital and organizational change in high-involvement and traditional work organizations. Journal of
Economics & Management Strategy, 11(2), 289328.
Gaynor, M., & Gertler, P. (1995). Moral hazard and risk spreading in partnerships. RAND Journal of Economics, 26(4), 591613.
Gneezy, U., Niederle, M., & Rustichini, A. (2003). Performance in competitive environments: Gender differences. Quarterly Journal of Economics, 118(3),
10491074.
Goeree, J., Holt, C., & Laury, S. (2002). Private costs and public benets: Unraveling the effects of altruism and noisy behavior. Journal of Public Economics,
83(2), 255276.
Greenwood, R., & Empson, L. (2003). The professional partnership: Relic or exemplary form of governance? Organization Studies, 24(6), 909933.
Hamilton, B., Nickerson, J., & Owan, H. (2003). Team incentives and worker heterogeneity: An empirical analysis of the impact of teams on productivity and
participation. Journal of Political Economy, 111(3), 465497.
Heywood, J., & Jirjahn, U. (2009). Prot sharing and rm size: The role of team production. Journal of Economic Behavior & Organization, 71(2), 246258.
Heywood, J., Jirjahn, U., & Wei, X. (2008). Teamwork, monitoring and absence. Journal of Economic Behavior & Organization, 68(34), 676690.
Itoh, H. (1991). Incentives to help in multi-agent situations. Econometrica, 59(3), 611636.
Kandel, E., & Lazear, E. (1992). Peer pressure and partnerships. Journal of Political Economy, 100(4), 801817.
Kvaloy, O., & Olsen, T. (2006). Team incentives in relational employment contracts. Journal of Labor Economics, 24(1), 139169.
Lasker, R., Weiss, E., & Miller, R. (2001). Partnership synergy: A practical framework for studying and strengthening the collaborative advantage. Milbank
Quarterly, 79(2), 179205.
Laury, S., & Taylor, L. (2008). Altruism spillovers: Are behaviors in context-free experiments predictive of altruism toward a naturally occurring public good?
Journal of Economic Behavior & Organization, 65(1), 929.
Lawford, G. (2003). Beyond success: Achieving synergy in teamwork. Journal for Quality & Participation, 26(3), 2327.
Ledyard, J. (1995). Public goods: A survey of experimental research. In J. Kagel & A. Roth (Eds.), The handbook of experimental economics. Princeton, NJ:
Princeton University Press.
Mas, A., & Moretti, E. (2009). Peers at work. American Economic Review, 99(1), 112145.
Meidinger, C., Rullire, J., & Villeval, M. (2003). Does team-based compensation give rise to problems when agents vary in their ability? Experimental
Economics, 6(3), 253272.
Nalbantian, H., & Schotter, A. (1997). Productivity under group incentives: An experimental study. American Economic Review, 87(3), 314341.
Pauly, M. (1996). Economics of multispecialty group practice. Journal of Ambulatory Care Management, 19(3), 2633.
Pizzini, M. (2010). Group-based compensation in professional service rms: An examination of medical group practices. The Accounting Review, 85(1),
343380.
Rose, D. (2002). Marginal productivity analysis in teams. Journal of Economic Behavior & Organization, 48(4), 355363.
Vandegrift, D., & Yavas, A. (2011). An experimental test of behavior under team production. Managerial and Decision Economics, 32(1), 3551.