Team formation with complementary skills

• DOI: http://doi.org/10.1111/jems.12296
• Author: Mürüvvet Büyükboyaci,Andrea Robbett
• Published date: 01 November 2019
• Date: 01 November 2019

Received: 17 January 2018

|

Revised: 2 September 2018

|

Accepted: 6 November 2018

DOI: 10.1111/jems.12296

ORIGINAL ARTICLE

Team formation with complementary skills

Mürüvvet Büyükboyaci

1

|

Andrea Robbett

2

1

Department of Economics, Middle East

Technical University, Ankara, Turkey

2

Department of Economics, Middlebury

College, Middlebury University,

Middlebury, Vermont

Correspondence

Büyükboyaci Mürüvvet, Department of

Economics, Middle East Technical

University, Universities District,

Dumlupinar Bulvari No: 1, 06800

Cankaya, Ankara, Turkey.

Email:muruvvet@metu.edu.tr

Funding information

European Union’s Horizon 2020 research

and innovation programme under the

Marie Skłodowska‐Curie, Grant/Award

Number: 2015ABH67840007

Abstract

One explanation for the prevalence of self‐managed work teams is that they

enable workers with complementary skills to specialize in the tasks they do

best, a benefit that may be enhanced if workers can sort themselves into teams.

To assess this explanation, we design a real‐effort experiment to study the

endogenous formation of teams, and its effect on productivity, when

specialization either is or is not feasible. We find a strong positive interaction

between endogenous team formation and the ability to specialize, indicating

that endogenous team formation is a particularly effective mechanism for

promoting team output in production environments that enable the exploitation

of skill complementarities.

KEYWORDS

comparative advantage, contest, endogenous team formation, matching, team performance,

weakest‐link technology

JEL CLASSIFICATION

C91, C92, C71, C72, D72, H41

1

|

INTRODUCTION

The use of team‐based incentives in the workplace has become increasingly common, with the majority of large firms

employing self‐managed work teams in recent years (Lazear & Shaw, 2007). One explanation for the increasing

prevalence of work teams, despite the inherent free‐riding incentives, is that they allow firms to exploit skill

complementarities among workers (Lazear, 1998, 2000). Such teams are often formed endogenously, particularly in

manufacturing industries, law, academia, and other professional organizations (Bandiera, Barankay, & Rasul, 2013).

Self‐organization has the potential to further amplify this benefit of teams—provided that agents are able to successfully

partner with those individuals whose skills best complement their own. Both laboratory and field studies, along with

anecdotal evidence, indicate that enabling workers to voluntarily join teams can have a positive impact on overall

productivity (Hamilton, Nickerson, & Owan, 2003; Herbst, Konrad, & Morath, 2015). However, this study has thus far

focused on how much effort teammates exert, rather than how they allocate effort across distinct tasks or whether

endogenous team formation can enable agents to form teams that exploit skill complementarities. Despite the

prevalence of self‐managed work teams, there is little direct evidence that teammates are able to successfully coordinate

on an allocation of tasks that maximize team productivity; nor is there evidence that they anticipate such coordination

and sort themselves into teams that enable them to take advantage of skill complementarities.

1

To address these questions, we design an experiment to study the endogenous formation of teams and its effect on

production, in team contests when specialization either is or is not feasible. By denying some teams the ability to

© 2018 The Authors. Journal of Economics & Management Strategy Published by Wiley Periodicals, Inc.

J Econ Manage Strat. 2019;28:713–733. wileyonlinelibrary.com/journal/jems

|

713

-----------------------------------------------------------------------------------------------

This is an open access article under the terms of the Creative Commons Attribution‐NonCommercial‐NoDerivs License, which permits use and distribution in any

medium, provided the original work is properly cited, the use is non‐commercial and no modifications or adaptations are made.

specialize, we are able to assess whether teammates seek out different partners in different production environments.

Additionally, we can isolate the channel through which team formation is likely to be most effective: Facilitating the

formation of teams with complementary skills. Specifically, we consider a simple production environment, in which

two team members must complete two distinct tasks to produce a unit of output. The introduction of multiple tasks,

which can be potentially completed by either team member, incorporates the potential gains from specialization that is

a natural feature of most team production environments (Lazear, 1998). Our experimental design allows us to induce

and control participants’abilities in each task, by manipulating how much they can accomplish within the time period.

Among the population of potential team members, all agents have the same overall capacity to complete tasks, but

differ in the distribution of their induced “skills”across the two tasks and which task they are “best”at. Specifically, our

population is equally divided among agents who are equally skilled at each task, agents who are slightly better at one

task and weaker at the other, and agents who are very good at one task and very poor at the other.

Within this setting, we conduct a 2 × 2 team contest experiment, in which we independently vary: (a) Whether

teammates have some ability to choose their partner or they instead rotate through the different possible match

types in the population; (b) whether each worker must completebothtaskstoproduceoritispossibletospecialize.

In the endogenously formed team contests, participants first rank each possible teammate on the basis of the agent’s

observed productivity in each task and agents are sorted into teams on the basis of these rankings. We predict that

the attractiveness of teammates with different skill sets will depend on whether the agents anticipate working

independently on both tasks or specializing in one task. To assess these predictions, we exogenously vary whether or

not specialization is possible by changing how teammates’efforts are aggregated in a perfect‐complement’s

production function. In half the sessions, each teammate must work on both tasks to produce and we expect all

productivity types to strictly prefer partners with more balanced skills. In the other sessions, participants have

flexibility in how to allocate the division of tasks across teammates. If they anticipate that they will be able to

coordinate on an outcome in which team members specialize, then their preferences over teammates should flip,

with agents no longer valuing balanced skill sets and instead preferring those who are best at the task at which they

themselves are weakest.

We find that endogenous team formation has a positive effect on productivity, primarily through the channel of

producing more efficient matches, with little effect on effort or coordination. Our results indicate that team formation

promotes productivity by facilitating the formation of teams of complementary skills that enable agents to focus on the

task they do best. Teammates who cannot specialize spend most of their time on their weaker task and divide their time

similarly regardless of how teams form. When specialization is possible, in contrast, participants spend significantly

more time on their stronger task when they are able to form their own teams. Consistent with our predictions,

participants prefer to be matched with more balanced types when specialization is not possible, but rank those with

more extreme, but complementary, skills higher when the teammates can specialize. As a result, the realized teams in

both treatments are characterized by balanced types matched with other balanced types and extreme types matched

with other extreme types. Surprisingly, in situations when agents are predicted to be indifferent between two types,

participants rank those with complementary skills as more preferred teammates, even when specialization is not

possible. Our results highlight the remarkable ability of agents to successfully sort into teams in which they can be most

productive and indicate that, as a result, endogenous team formation is particularly effective as a means of exploiting

gains from complementarities in worker skills. The ability to seek out and partner with other agents whose skills are

complementary is likely to be particularly valuable in situations where workers’“on the ground”knowledge of the

relative strengths of their coworkers is superior to the information held by management.

Our paper is most closely related to the strands of the experimental literature concerning behavior in team contests

and endogenous group formation in strategic interactions. As discussed above, the primary focus of the team contest

literature is on the provision of costly effort and an overarching conclusion is that teams regularly overprovided chosen

effort relative to the Nash equilibrium.

2

With respect to the optimal formation of teams under different production

technologies, Brookins, Lightle, and Ryvkin (2015, 2017) investigate how a principal should organize agents who differ

in their overall abilities when teammates’efforts are perfect substitutes or perfect complements. These studies differ

from the current paper in that they study teammates engaged in an identical task with heterogeneous values for the

prize or costs of effort, whereas we consider teammates each engaged in two distinct tasks, with heterogeneous skills at

each, while holding constant their overall costs and benefits.

The experiment most similar to the current design is Büyükboyaci and Robbett (2017), which used the same two‐

task production environment.

3

That study considered the effect of team contests by varying whether participants

competed in teams or individually, as well as whether team members could specialize and communicate. The results

714

|

BÜYÜKBOYACI AND ROBBETT