Incentive Redesign and Collaboration in Organizations: Evidence from a Natural Experiment

• Author: Sunkee Lee,Phanish Puranam
• DOI: http://doi.org/10.1002/smj.2685
• Published date: 01 December 2017
• Date: 01 December 2017

Strategic Management Journal

Strat. Mgmt. J.,38: 2333–2352 (2017)

Published online EarlyView 8 August 2017 in WileyOnline Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2685

Received 14 November 2015;Final revisionreceived 12 April 2017

Incentive Redesign and Collaboration

in Organizations: Evidence from a Natural Experiment

Sunkee Lee1*and Phanish Puranam2

1Tepper School of Business, Carnegie Mellon University, Pittsburgh, Pennsylvania

2Strategy Department, INSEAD, Singapore

Research summary: Separating the individual from the social effects of incentives has been

challenging because of the possibility of synergiesin team production. We observe a unique natural

experiment in a South Korean e-commerce company in whicha switch from pay-for-performance

to xed (but different) salaries took place in a staggered and effectively random manner across

employees. In this case, social and individual effects perspectives make opposing predictions,

enabling a critical test. We nd evidence consistent with social effects of incentives, particularly

as predicted by goal framing theory. The results have implications for the design of incentives to

foster collaboration, organizational learning, and organizational performance.

Managerial summary: Managers often neglectthe deeper hypothesis behind pay-for-performance

schemes— that people primarily care about how much they are individually paid. An opposing

school of thought contends that incentives have social effects too— that individuals careabout not

only what they receive but also what their peersreceive. It is difcult to say whether individual or

social effects would be more salient in a context,without a proper experiment with randomization.

We exploit a rare opportunity provided by a company that changed its incentive system in a

random order, thus unintentionally creating a natural experiment. The results strongly validate

the existence of social effects of incentives, but also make the general case for the opportunity to

learn from experimenting with organization design in a systematic manner. Copyright © 2017

John Wiley & Sons, Ltd.

Incentives continue to attract signicant interest in

the eld of strategy (e.g., Baumann & Stieglitz,

2014; Frank & Obloj, 2014; Lee & Meyer-Doyle,

2017). However, while there is a broad consensus

that incentives matter, precisely how they matter

remains a contested theoretical ground. In partic-

ular, a key fault line in the thinking about incen-

tives lies between the views of the individual versus

social effects of incentives.

By individual effects, we mean the ways in which

incentives alter an individual A’scompensation and

Keywords: incentives; organization design; collaboration;

goal framing theory; natural experiment

*Correspondence to: Sunkee Lee, Tepper School of Business,

Carnegie Mellon University,5000 Forbes Avenue, Pittsburgh, PA

15213. E-mail: sunkee@andrew.cmu.edu

Copyright © 2017 John Wiley & Sons, Ltd.

therefore change A’s behavior. The change in A’s

compensation can arise either because A’s own

incentive contract changes or that of others change.

In addition, the incentive contracts in question could

be either team or individual level (Kretschmer &

Puranam, 2008; Oxley & Pandher, 2016). As long

as a change in incentives causes a change in A’s

behavior via a change in A’scompensation, we refer

to it as an individual effect of incentives.

By the social effects of incentives, we mean the

ways in which changing B’s incentives (which may

be of an individual or a team level) may lead to

a change in A’s behavior or performance, even

though A’s compensation remains unchanged. The

perspective that focuses on the individual effects

of incentives is strongly rooted in the tradition of

2334 S. Lee and P. Puranam

agency theory in economics (Holmstrom & Mil-

grom, 1994; Jensen & Meckling, 1976), whereas

that on the social effects of incentives is behavioral

in nature, and draws its roots from the social psy-

chological research on fairness (e.g., Adams, 1963)

and on the salience of group versus individual inter-

ests (e.g., Lindenberg, 2001).

Testing the individual and social effectsof incen-

tives as competing explanationsfor observedbehav-

ior has proven to be challenging, for at least two

reasons. The rst is the usual problem of causal

inference from naturally occurring data. A num-

ber of creative efforts in recent years have tackled

this problem by employing experimental designs

in the eld, in which the treatment is a change in

incentive regime (e.g., Burks, Carpenter, & Goette,

2009; Shearer, 2004). The second is more subtle;

individual and social effects can be easily con-

founded under conditions of team production with

synergies (Shaw, Gupta, & Delery, 2002; Wage-

man & Baker, 1997). To appreciate this second

point, it is useful to think of incentives along two

well-established dimensions: depth— what fraction

of an individual’s pay depends on performance,

and breadth— how many others contribute to the

performance measure used to determine pay (e.g.,

Baker, 2002; Kretschmer & Puranam, 2008; Oxley

& Pandher, 2016; Rivkin & Siggelkow, 2003).

When one switches from xed salary to

pay-for-performance (e.g., Burks et al., 2009;

Lazear, 2000; Shearer, 2004) the depth of

incentives is increasing. If one switches from

individual pay-for-performance to group

pay-for-performance, the breadth is increasing

(e.g., Beersma et al., 2003; Petty, Singleton, &

Connell, 1992). A move from xed pay to group

pay-for-performance indicates an increase in both

depth and breadth. Assume we observe an increase

in collaborative behaviors by individuals under any

of these changes. However, from this we could not

infer that social effects exist, as these changes could

arise purely from individual effects as incentives

for individuals will change in each of these cases

in the presence of synergies from collaboration

(Milgrom & Roberts, 1995). For instance, a move

from individual pay-for-performance to group

pay-for-performance could trigger collaboration as

long as the gains from synergy are sufcient to deter

free-riding (Wageman & Baker, 1997), and a move

from xed salary to individual performance-based

pay could do the same if individuals are able to

coordinate to capture synergies (Camerer & Knez,

1996; Kretschmer & Puranam, 2008). Thus, what

look like social effects of incentives can in fact be

due purely to changes in individuals’ incentives.

One approach to cleanly separate the individual

from the social effects of incentives for the pur-

poses of establishing their independent existence

is to look for the kind of incentive regime change

that leads to sharply distinct predictions from the

two perspectives. A change that meets these require-

ments is that from individual pay-for-performance

to xed salary. In this case, the social effects per-

spective, as we explain in detail in our theoriz-

ing, predicts an increase in collaborative behaviors

under at least some conditions, but the individ-

ual effects perspective rules this out. A critical

test of these perspectives is thus possible (Lave

& March, 1975; Platt, 1964). We were fortu-

nate to observe exactly this transition in incentive

regimes in a South Korean e-commerce company

(that we name “Chimera”), in which a switch from

pay-for-performance to xed (but different) salaries

took place in a staggered and effectively random

manner across employees. Exploiting this setting,

we run a competing test of three theories— agency

theory (Eisenhardt, 1989), which represents an

individual effects perspective, and equity theory

(Adams, 1963) and goal framing theory (Linden-

berg & Foss, 2011), which both represent the social

effects perspectives—that each make different pre-

dictions about the individual versus social effect of

incentives, and thus allows us to separate the two

effects.

As with most eld experiments, what we gain

in terms of closer-to-causal inference comes at the

expense of generalizability (e.g., Burks et al., 2009;

Lazear, 2000; Shearer, 2004). We believe that off-

setting these limits to generalizability, our analysis

offers unique contributions in terms of being able

to cleanly separate social from individual effects of

incentives change in a eld setting. Our critical test

(with close-to-causal inference) of agency theory

versus equity theory versus goal framing theory is

among the rst efforts to provide eld evidence for

the latter.

We also believe that understanding the social

and individual effects of incentives and what

that means for collaboration in organizations is

of critical interest to students of strategy. The

competitive advantage of organizations rests not

only on their resource endowments (Barney, 1991),

but also on the capabilities embodied in patterns

of collaboration, knowledge sharing, and support

Copyright © 2017 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 2333–2352 (2017)

DOI: 10.1002/smj