an experimental comparison of incentive contracts in partnerships

Journal of Economic Psychology 34 (2013) 78–87

Contents lists available at SciVerse ScienceDirect

Journal of Economic Psychology

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

An experimental comparison of incentive contracts inpartnerships

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

⇑ Corresponding author. Tel.: +86 21 52301562; fax: +86 21 32231111.E-mail addresses: [email protected] (H. Chao), [email protected] (R.T.A. Croson).

1 Also known as individual output contracts, individual-based contracts, or independent performance pay, and team output contracts, team-basedor joint performance pay.

2 Note that individualized sharing schemes are slightly different than piece rates. In a piece rate compensation scheme, the firm takes a share of thin partnerships all profits are distributed to the partners.

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, Chinab Department of Economics, School of Economic, Political & Policy Sciences, University of Texas at Dallas, 800 W. Campbell Drive, GR31, Richardson, TX75080-3021, USA

a r t i c l e i n f o a b s t r a c t

Article history:Received 15 September 2011Received in revised form 21 November 2012Accepted 29 November 2012Available online 10 December 2012

PsycINFO classification:3660

Keywords:PartnershipIncentiveContractSynergy

Empirical work comparing individualized sharing and equal sharing schemes in partner-ships has produced mixed results. Some studies find individualized sharing schemes supe-rior, others find no difference, and still others find equal sharing schemes superior. Thispaper outlines a theory which reconciles these competing findings, and tests it with anexperiment. We find that in conditions of high synergy (when the teammate’s effort hasa proportionately larger impact on an agent’s output than the agent’s own effort), equalsharing schemes outperform individualized sharing schemes, while in conditions of lowsynergy, individualized sharing schemes outperform equal sharing schemes. These resultsare consistent with observations from the field. Our results have the potential to guidefirms choosing between competing compensation contracts by identifying situations underwhich each contract type is likely to yield increased productivity.

� 2012 Elsevier B.V. All rights reserved.

1. Introduction

Partnerships are commonly observed in firms, representing 100% of the top 100 law firms, 56% of the top 100 accountingfirms, and 18% of the top 100 architecture firms (Greenwood & Empson, 2003). Labor contracts within partnerships takevarying forms.

Two common forms involve individualized sharing schemes and equal sharing schemes.1 As the names suggest, in indi-vidualized sharing schemes the compensation of partnership members is their own individual output,2 while in equal sharingschemes, their compensation is a function of the team output. In practice, however, individual output is often a function notonly of one’s own efforts, but of others’ efforts as well (Alchian & Demsetz, 1972). This effect is referred to as team synergy (Law-ford, 2003).

Synergy effects can be frequently observed in partnerships. In a medical partnership, for example, the number of surgeriesa doctor can perform (and be paid for) may depend on their partners’ availability for collaborations. Although synergy is acommon feature of partnerships, its impact on the efficiencies of various incentive contracts is still vaguely understood andhas not been systematically examined in the literature (Alchian & Demsetz, 1972; Rose, 2002).

contracts

e profits;

http://crossmark.dyndns.org/dialog/?doi=10.1016/j.joep.2012.11.009&domain=pdf

http://dx.doi.org/10.1016/j.joep.2012.11.009

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.joep.2012.11.009

http://www.sciencedirect.com/science/journal/01674870

http://www.elsevier.com/locate/joep

H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 78–87 79

Observational and experimental evidence comparing the effectiveness of equal sharing and individualized sharing hasbeen mixed, with some studies finding individualized sharing superior (Encinosa, Gaynor, & Rebitzer, 2007; Gaynor & Ger-tler, 1995; Nalbantian & Schotter, 1997), others finding no difference (Dijk, Sonnemans, & Winden, 2001; Vandegrift & Yavas,2011), and still others finding 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 experimen-tally that when team synergy is high (characterized by a large degree of complementarity between one’s own effort and theeffort exerted by teammates), equal sharing schemes outperform individualized sharing schemes. When team synergy islow, the opposite is true.

The intuition for our result is straightforward. Equal sharing schemes internalize effort externalities but include incen-tives for free riding. When effort externalities are sufficiently high, the former effect outweighs the latter and equal sharingschemes outperform individualized sharing schemes. When effort externalities are not sufficiently high, the free-ridingincentives outweigh the benefit from internalizing the effort externality, and individualized sharing schemes outperformequal sharing schemes.

Our results are consistent with previous observations from the field, including the prevalence of equal sharing in high-synergy specialties (for example, emergency medicine) and the dominance of individualized sharing in low-synergy special-ties (for example, psychiatry) (Adams, 2006; Pauly, 1996; Pizzini, 2010). Finally, our results have the potential to guide part-nerships choosing between competing compensation contracts by identifying situations under which each contract type islikely 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 indi-vidualized sharing schemes. We argue that these types of partnerships have varying levels of synergies, and that this canexplain the differential use of the various incentive schemes.

Both observational and experimental data has compared the efficiency properties of these schemes, with mixed results.Some studies have found that equal sharing schemes underperform individualized sharing schemes. For example, Encinosaet al. (2007) find that for large medical groups, team productivity is lower under equal sharing, while Gaynor and Gertler(1995) find that equal sharing schemes generate fewer office visits for doctors. Nalbantian and Schotter (1997) run a labexperiment in which equal sharing schemes generate significant free-riding and lower levels of productivity than individu-alized sharing schemes. Erev, Bornstein, and Galili (1993) find 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 interdependenceamong participants, thus involve no synergies. We argue that the lack of synergies results in equal sharing underperformingindividualized sharing.4

In contrast, other studies have found that equal sharing schemes outperform individualized sharing schemes. Hamiltonet al. (2003) find a 14% increased productivity in a garment plant that switched from individual output to team output con-tracts. Chan et al. (2012) and Pizzini (2010) find similar effects in a field experiment run in department stores cosmetic coun-ters and in medical partnerships respectively. Hamilton et al. (2003) investigate manufacturing workplaces, which usuallyinvolve high synergy levels due to the extensive production interdependence in their operational lines (Adams, 2006). Chanet al. (2012) find that the advantage of equal sharing over individualized sharing increases if peer productivity spillovers in-crease. Pizzini (2010) finds that equal sharing schemes are more prevalent in the specialties with substantial productioninterdependence (e.g., anesthesiology, radiology, and general surgery). These results are thus consistent with our argumentthat 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 administeremergency care), depending on their specialties. Pizzini (2010) rates major medical specialties in terms of production interdependence (see Table A-1, Pizzini,2010). Using Pizzini’s rating, we find that only about 8% partnerships investigated in Encinosa et al. (2007) and 1 out of 4 specialties investigated in Gaynor andGertler (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 whichparticipants solve a two-variable optimization problem. Vandegrift and Yavas (2011) run an experiment using a forecasting task. However, in both studiesparticipants were grouped and provided with their teammates’ performance under equal sharing but not individualized sharing. It is possible that thisinformation 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 moreskillful workers can teach the less skillful how to execute tasks efficiently, yielding significant 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 generatedifferent results. Meidinger, Rullière, and Villeval (2003) report that equal sharing performs poorly in this setting, while Vandegrift and Yavas (2011), reportthat heterogeneity of ability increases the efficiency of equal sharing. This increased efficiency, 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 abilityheterogeneity and performance under different contractual terms.

80 H. Chao, R.T.A. Croson / Journal of Economic Psychology 34 (2013) 78–87

Previous theoretical work has attempted to explain these conflicting results. Itoh (1991) shows that a team output con-tract is optimal when own effort and helping effort are complementary, but not (always) when they are substitutes. Dem-ougin and Fluet (2006) show that when agents are envious of each other’s wages, team contracts can outperform individualcontracts, because under team contracts wages are equalized, employees do not suffer disutilities due to envy, and are thuswilling to work for less (also see Englmaier & Wambach, 2010). Kvaloy and Olsen (2006) use settings of incomplete infor-mation and allow for peer monitoring. They show that when there is little common noise, peer monitoring allows team con-tracts to outperform individual contracts. In contrast, when there is a lot of common noise, the cost of team contracts is highand thus individual contracts are preferred.6 This paper provides and tests a different organizing explanation for the competingempirical results, based on team synergy (Chao & Siqueira, 2013).

In our model (below), individuals choose a level of costly effort to exert. Effort influences both their own and their part-ner’s output. We define low synergy as a scenario where an agent’s output is determined primarily by the effort exerted bythat agent and the effort exerted by the agent’s teammate has little effect. High synergy, on the other hand, is characterized bya large degree of complementarity between one’s own effort and that exerted by one’s teammate; in fact in our high synergytreatment, the teammate’s effort has a proportionately larger impact on an agent’s output than the agent’s own effort.7 Wemake this characterization more specific and formal in Section 3, below, where we present our theoretical model. We use a sim-ple linear production technology and show experimentally that equal sharing will outperform individualized sharing under con-ditions 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 asymmetricmarginal per capita return (MPCR) and the addition of a stochastic shock to output (Vandegrift & Yavas, 2011).8 In theseexperiments, the benefit generated from the public account to the contributor (the internal return) is not equal to that tothe other group members (the external return). Goeree, Holt, and Laury (2002) and Laury and Taylor (2008) find that both inter-nal return and external return generate increased contributions to the public good although the former effect is stronger. Torelate our work to the VCM framework, one can imagine that an equal sharing scheme is a mechanism in which individualsinternalize 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 effortlevel ai. This effort generates output, yi according to the following function:

6 Othand Mo

7 Woothers madvice m2006; L

8 See

yi ¼ fownai þ fotheraj þ ei; j ¼ 1;2; and i – j ð1Þ

fown and fother (which are both nonnegative) describe the marginal product of effort on one’s own output and on the partner’soutput. When fown > fother, we consider the setting to exhibit low levels of synergy. When fown < fother, we consider the settingto exhibit high levels of synergy. The random term ei is uniformly distributed over the interval ½��e;þ�e� and assumed to bei.i.d. for both partners.

Efforts increase productivity, but are also costly. In particular, we assume that the cost of effort is quadratic:

CðaiÞ ¼ ca2i i ¼ 1;2 ð2Þ

3.2. Social optimum

We first calculate the socially optimal level of effort.

maxai ;aj

EUi þ EUj ¼ Eðyi þ yjÞ � CðaiÞ � CðajÞ ð3Þ

yields the socially optimal effort level aSO ¼ fownþfother2c and the socially optimal payoff level USO ¼ ðfownþfother Þ2

4c for each partner. Aswith any setting involving an externality, the first-best solution equates each partner’s marginal cost of effort with themarginal product of that partner’s effort on the total team output.

er 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 Masretti (2009).rkers in a team may possess various skills and perform different but highly interdependent tasks. The externalities generated by workers’ efforts on

ay thus be amplified in others’ production. For example, a physician spends 10 min to help a surgeon to make a plan for a gastric surgery. This 10-minay save the surgeon 10 h if she had done the job alone. Such high synergy situations are not uncommon in highly interdependent workplaces (Adams,

asker, Weiss, & Miller, 2001; Pizzini, 2010).Ledyard (1995) for an extensive survey on VCM literature.

Table 1Experimental parameters and equilibrium predictions.

a e {0, ..., 150}, e e [�40,40], c = 0.1, 2-subject teams, 15 rounds Individualized sharing Equal sharing

Low synergy (fown = 10, fother = 2)Equilibrium effort (NE) 50 30Socially-optimal effort (SO) 60 60

High Synergy (fown = 10, fother = 18)Equilibrium effort (NE) 50 70Socially-optimal effort (SO) 140 140


3.3. The individualized sharing scheme

The simplest contractual setting one can imagine is one in which each partner receives wages equivalent to their ownoutput. The partner’s expected payoff is then

9 In tindividu

maxai

EUindi ¼ EðyiÞ � CðaiÞ ð4Þ

which generates equilibrium effort and resulting payoff of aind ¼ fown2c and Uind ¼ 1

4c fownðfown þ 2f otherÞ.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. Noweach partner’s expected payoff is

maxai

EUequali ¼ 1

2Eðyi þ yjÞ � CðaiÞ ð5Þ

which generates equilibrium effort and resulting payoff of aequal ¼ 14c ðfown þ fotherÞ and Uequal ¼ 3

16c ðfown þ fotherÞ2. 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 find that (1) inlow-synergy settings (when fother < fown), the individualized sharing scheme generates higher effort than the equal sharingscheme, aind > aequal; (2) in high-synergy settings (when fother > fown), the equal sharing scheme generates higher effort thanthe 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 (highand low). Experimental participants were randomly assigned to one of the four treatments, thus both synergy level and shar-ing 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}. Themarginal product of effort on one’s own production, fown is set to be 10 in all treatments. The marginal product of effort onone’s partner’s production, fother, is 2 in the low synergy treatment and 18 in the high synergy treatment. The random term eis independently and uniformly distributed over the integer interval [�40,40], and we set c (the cost of effort) to 0.1. Theequilibrium 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 Univer-sity in China. The experiment was programmed using z-Tree (Fischbacher, 2007). Participants were recruited using the on-line recruiting system. We ran eight experimental sessions in total with two sessions for each treatment. One hundred andninety students from multiple disciplines participated in the experiment; 22 in one session of low-synergy/individualized-sharing 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 participantswere randomly matched with a different counterpart in each round (Andreoni & Croson, 2008). After each round, the com-puter randomly selected an integer between [�40,40] for each partner to represent e. Participants were reminded of theirown decision, told of the random number drawn for them, their own output and their counterpart’s output, their own earn-ings for the previous round and their cumulative earnings.9 A history table with this information for previous rounds was pro-vided during the session. Participants were not allowed to communicate other than via the program. An example of thetranslated version of instructions can be found in the Appendix.

he equal sharing treatment participants can infer their counterpart’ output from their own earnings, thus we provided this information in both thealized and equal sharing treatments to ensure parallelism.

Table 2Descriptive statistics of average efforts.a

Average efforts (standard deviation) [number of observations] Round 1 Rounds 1–15 Rounds 3–15 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 withintheir time limit.


In order to ensure that the experiments would be completed in a reasonable amount of time, participants were given afixed amount of time to make each decision (2 min for each of the first 2 rounds and 30 s for each of the remaining 13rounds). If they did not make a decision within the given time, the computer randomly chose an effort level for them froma uniform distribution between 0 and 150. This happened rarely (113 times out of 2850 decisions).10 In our main results re-ported below, we drop these auto-selected observations.

After the experimental decisions, participants completed a post-experimental survey. After the survey, participants werepaid anonymously based on their total earnings. Earnings were calculated in the fictitious currency eckels and were trans-lated into RMB using a fixed and commonly-known exchange rate. Average earnings were 39.72 RMB (including a show-upfee 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 illuminatelearning in this setting, we report effort levels for the first round, the entire 15 rounds, rounds 3–15 and the last round.Inspection of Table 2 demonstrates that efforts in all four treatments begin higher than the equilibrium predictions but de-crease 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 fourtreatments. Our dependent variable is the effort levels of each subject averaged over a certain number of rounds. The resultsof these tests are reported in Table 3.

As seen in Table 3, average effort levels are significantly different than equilibrium predictions for all comparisons,although the differences are reduced as the rounds progress. Figs. 1 and 2 demonstrate this convergence graphically. Averageefforts in low synergy treatments (Fig. 1) and high synergy treatments (Fig. 2) begin higher than their predicted levels, butgenerally 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. Wefirst use a two-tailed nonparametric Wilcoxon test to do the comparisons. Our dependent variable is the effort levels of eachsubject 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 everyone’s decision had been made, before moving to thenext round) meant that a participant could never know that their counterparts’ decisions were made automatically. Even if participants believed that they wereplaying 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 aspredicted here. While beliefs about auto-selection may matter in a repeated game, in our experiment participants played with different counterparts in eachround 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 thedecisions of their paired counterparts.

Table 3Comparison between observed and equilibrium efforts.a

Wilcoxon signed-rank test (two-tailed) P-values [number of observations]

Round 1 Rounds 1–15 Rounds 3–15 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]


0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

aver

age

effo

rt

round

Individualized Sharing

Equal Sharing

NE (Individualized Sharing)

NE (Equal Sharing)

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

0

20

40

60

80

100

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

aver

age

effo

rt

round

Individualized Sharing

Equal Sharing

NE (Individualized Sharing)

NE (Equal Sharing)

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


As hypothesized, we find that effort levels in low synergy conditions are significantly different under equal sharing thanunder individualized sharing. We also find that effort levels in high synergy conditions are significantly different under equalsharing than under individualized sharing.

To confirm the above findings, we run OLS regressions using individual efforts over the 3–15 rounds and including dum-mies for the treatment (our coefficient of interest), for each session and for each subject, and a control for the period. Resultsare reported in Table 5.11

This analysis confirms our previous results and supports both our hypotheses. We see a statistically significant andpositive coefficient 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 specification 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.

Table 4Nonparametric tests of sharing schemes.a

Wilcoxon test (two-tailed) Z-scores (individualized sharing – equal sharing)(P-values)[number of observations]

Round 1 Rounds 1–15 Rounds 3–15 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]


Table 5Individual efforts by sharing scheme.a

Individual effort Low synergy High synergy

Individualized sharing 5.66**(0.41) �8.77**(0.52)Period �0.92**(0.11) �0.71**(0.14)Session dummies YES YESSubject dummies YES YESConstant 56.15**(1.09) 75.21**(1.36)Number of observations 1189 1204Adjusted R2 0.383 0.492

** p < 0.01a 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.


low synergy conditions efforts under individualized sharing are higher than efforts under equal sharing (H1). We see a sta-tistically significant and negative coefficient on the individualized sharing scheme dummy in the high synergy treatments, aspredicted. Thus in high synergy conditions, efforts under individualized sharing are lower than efforts under equal sharing(H2). We further replicate our findings of decreasing efforts over time (significantly negative coefficient on period), as seen inFigs. 1 and 2.

6. Conclusion

This paper theoretically and experimentally compares worker behavior in partnerships under conditions of high and lowsynergy and team and individual incentives. We find, as predicted, that efforts are significantly higher under individualizedsharing 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 ofpartnerships. They help us to explain the puzzling frequency (and apparent success) of team incentive schemes in the pres-ence of the incentive to free-ride. They allow us to predict when a partnership will choose team incentives (in conditions ofhigh synergy) or individual incentives (in conditions of low synergy). Finally, our results make active recommendations forwhat types of contracts firms should choose, given their level of synergy.

Future work is needed to reinforce our understanding about the effect of synergy on contract selection. One possibledirection is to construct a measure of synergy level in practice. Previous studies have identified a few determinants of syn-ergy 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 jobattachment (Brown, Geddes, & Heywood, 2007; Heywood & Jirjahn, 2009; Heywood, Jirjahn, & Wei, 2008), and level of socialcapital (Bandiera, Barankay, & Rasul, 2010; Gant, Ichniowski, & Shaw, 2002). However, a comprehensive synergy evaluationmechanism 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 synergylevels. For example, Chao and Siqueira (2013) provide theoretical predictions for the comparison between team output con-tracts 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 thesetwo contractual types. In our experiment we hold monitoring constant (always present) to focus on the impact of synergy,but in the field both may be relevant. For example, equal sharing schemes are more likely to be adopted by partners who


have known each other for a long period of time. Two reasons suggest themselves. First, long-term relationships generate ahigh 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 inves-tigate 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 inpartnership settings. Ultimately this improved understanding has the potential to increase our ability to engage in contractdesign 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 theCenter 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 financial and logis-tical 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 EconomicDecision-Making Experiment.

There are 15 rounds of experiment in the following setting. In each round, you will be randomly and anonymouslymatched with a counterpart who is another participant in this room and then make one decision. After each round, you willbe matched with a different counterpart. Therefore, you will be matched with 15 different counterparts during the wholesession. Your earnings will depend on the decisions you make, the decisions your counterparts make, and chance. If you fol-low 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 withyour counterpart will take place through the computer program. If you have a question, please raise your hand and a monitorwill come to you to answer your question privately.

You will be making one decision in each of the 15 rounds. You will have a fixed amount of time to make each of yourdecisions which will be noted at the top of each screen; if you do not make a decision within that time, the computer willrandomly choose a decision for you. For the first two rounds you will have 2 min to make each decision. Starting in roundthree 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 finished with their decisions in one round, you will be matched with a different counter-part 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 fictitious currency called eckels. At the end of the session today, you will be paid oneRMB 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 numberbetween 0 and 150. You can choose any integer (any whole number, no decimals or fractions). After you and your counter-part enter your numbers, the computer will randomly select two numbers between�40 and 40; one for you and one for yourcounterpart. Each number between �40 and 40 is equally likely to be selected by the computer. Your payoffs will depend onthe numbers you and your counterpart choose, and the random numbers selected by the computer.

We will first use these numbers to calculate your and your counterpart’s output:

Your output ¼ 10 � ðyour choiceÞ þ 2½18� � ðtheir choiceÞ þ your random number:

Counterpart’s output ¼ 10 � ðtheir choiceÞ þ 2½18� � ðyour choiceÞ þ 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 counterpart’s) output. However, choosing higher numbers is morecostly. Your and your counterpart’s costs are:

Your cost ¼ 0:1 � ðyour choiceÞ2:

Counterpart’s cost ¼ 0:1 � ðtheir choiceÞ2:

Note that the number you choose will affect your costs but not your counterpart’s costs, and vice versa.Your payoffs will be calculated using a certain combination of your output, your counterpart’s output, and your own costs.

The specific form of combination will be described later.


To calculate your earnings, we add your earnings (in eckels) across all the decisions you made, then multiply by the ex-change 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. Specific 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 coun-

terpart each choose a number between 0 and 150. The computer will randomly select a number between �40 and 40 foreach of you. Your and your counterpart’s outputs are:



Your and your counterpart’s costs are:



Your earnings will be your output, minus your own costs. Those earnings are thus:

ðyour outputÞ � ðyour costÞ

This simplifies to:

10 � ðyour choiceÞ þ 2½18� � ðtheir choiceÞ þ your random number � 0:1 � ðyour choiceÞ2:

Please enter your choice from (0–150):________.

A.3. Specific 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 coun-

terpart each choose a number between 0 and 150. The computer will randomly select a number between �40 and 40 foreach of you. Your and your counterpart’s outputs are:



Your and your counterpart’s costs are:



Your earnings will be the average of your and your counterpart’s outputs, minus your own costs. Those earnings are thus:

1=2 � ðyour output þ counterpart’s outputÞ � ðyour costÞ

This simplifies to:

6½14� � ðyour choiceþ their choiceÞ þ ðaverage of the two random numbersÞ � 0:1 � ðyour choiceÞ2:

Please enter your choice from (0–150):________.

References

Adams, C. (2006). Optimal team incentives with CES production. Economics Letters, 92(1), 143–148.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), 777–795.Bandiera, O., Barankay, I., & Rasul, I. (2010). Social incentives in the workplace. Review of Economic Studies, 77(2), 417–458.Brown, M., Geddes, L., & Heywood, J. (2007). The determinants of employee involvement schemes: Private sector Australian evidence. Economic and

Industrial Democracy, 28(2), 259–291.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), 525–541.

http://dx.doi.org/10.2139/ssrn.1367441


Demougin, D., & Fluet, C. (2006). Group vs. individual performance pay when workers are envious. In D. Demougin & C. Schade (Eds.), Contributions toentrepreneurship and economics, first 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), 187–214.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), 187–214.Englmaier, F., & Wambach, A. (2010). Optimal incentive contracts under inequity aversion. Games and Economic Behavior, 69(2), 312–328.Erev, I., Bornstein, G., & Galili, R. (1993). Constructive intergroup competition as a solution to the free rider problem: A field experiment. Journal of

Experimental Social Psychology, 29(6), 463–478.Fischbacher, U. (2007). z-Tree: Zurich toolbox for ready-made economic experiments. Experimental Economics, 10(2), 171–178.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), 289–328.Gaynor, M., & Gertler, P. (1995). Moral hazard and risk spreading in partnerships. RAND Journal of Economics, 26(4), 591–613.Gneezy, U., Niederle, M., & Rustichini, A. (2003). Performance in competitive environments: Gender differences. Quarterly Journal of Economics, 118(3),

1049–1074.Goeree, J., Holt, C., & Laury, S. (2002). Private costs and public benefits: Unraveling the effects of altruism and noisy behavior. Journal of Public Economics,

83(2), 255–276.Greenwood, R., & Empson, L. (2003). The professional partnership: Relic or exemplary form of governance? Organization Studies, 24(6), 909–933.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), 465–497.Heywood, J., & Jirjahn, U. (2009). Profit sharing and firm size: The role of team production. Journal of Economic Behavior & Organization, 71(2), 246–258.Heywood, J., Jirjahn, U., & Wei, X. (2008). Teamwork, monitoring and absence. Journal of Economic Behavior & Organization, 68(3–4), 676–690.Itoh, H. (1991). Incentives to help in multi-agent situations. Econometrica, 59(3), 611–636.Kandel, E., & Lazear, E. (1992). Peer pressure and partnerships. Journal of Political Economy, 100(4), 801–817.Kvaloy, O., & Olsen, T. (2006). Team incentives in relational employment contracts. Journal of Labor Economics, 24(1), 139–169.Lasker, R., Weiss, E., & Miller, R. (2001). Partnership synergy: A practical framework for studying and strengthening the collaborative advantage. Milbank

Quarterly, 79(2), 179–205.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), 9–29.Lawford, G. (2003). Beyond success: Achieving synergy in teamwork. Journal for Quality & Participation, 26(3), 23–27.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), 112–145.Meidinger, C., Rullière, J., & Villeval, M. (2003). Does team-based compensation give rise to problems when agents vary in their ability? Experimental

Economics, 6(3), 253–272.Nalbantian, H., & Schotter, A. (1997). Productivity under group incentives: An experimental study. American Economic Review, 87(3), 314–341.Pauly, M. (1996). Economics of multispecialty group practice. Journal of Ambulatory Care Management, 19(3), 26–33.Pizzini, M. (2010). Group-based compensation in professional service firms: An examination of medical group practices. The Accounting Review, 85(1),

343–380.Rose, D. (2002). Marginal productivity analysis in teams. Journal of Economic Behavior & Organization, 48(4), 355–363.Vandegrift, D., & Yavas, A. (2011). An experimental test of behavior under team production. Managerial and Decision Economics, 32(1), 35–51.

an experimental comparison of incentive contracts in partnerships

Documents