Equity stakes and exit: An experimental approach to decomposing exit delay
| Date | 01 February 2017 |
| Published date | 01 February 2017 |
| Author | Daniel W. Elfenbein,Rachel Croson,Anne Marie Knott |
| DOI | http://doi.org/10.1002/smj.2493 |
Strategic Management Journal
Strat. Mgmt. J.,38: 278–299 (2017)
Published online EarlyView 29 February 2016 in WileyOnline Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2493
Received 7 February 2014;Finalrevision received 10 September 2015
EQUITY STAKES AND EXIT: AN EXPERIMENTAL
APPROACH TO DECOMPOSING EXIT DELAY
DANIEL W. ELFENBEIN,1*ANNE MARIE KNOTT,1and RACHEL CROSON2
1Olin Business School, Washington University in St. Louis, Missouri, U.S.A.
2College of Business, University of Texas, Arlington, Texas, U.S.A.
Research summary: Exit delay is an important problem for entrepreneurs and managers alike, yet
relativelylittle is known about its causes. Weconduct a laboratory experiment in which optimal exit
is well dened, and in which a treatment groupwith equity stakes —the actual cash ows of a rm
and decision rights over its continuation— is comparedto a control group whose compensation is
based solely on its assessment of the rm’s protability.While treatment group participants make
exit decisions that are nearly optimal given their beliefs, their beliefs are signicantly distorted
relative to the control group. The pattern of distortion is consistent with conrmatory bias and
motivated reasoning. A fundamental nding of our study is that incentives may not only affect
behavior, butbelief formation as well.
Managerial summary: Managers and entrepreneursfrequently destroy signicant value by failing
to shut down underperforming businesses in a timely manner. To address this problem, we must
understand the mechanisms causing exit delay. We examine behavioral mechanisms causing
delay through a laboratory experiment in which subjects make decisions about when to exit a
failing venture. We nd that “equity stakes” —receiving the rm’s cash ows and having decision
rights over exit— cause participants to discount negative performance information, retain overly
optimistic beliefs, and delay exit. By contrast, participants without these high-powered incentives
exit nearly optimally. Our ndings suggest ways to reduce exit delay in managerial settings,
including implementing automated decision rules, removing equity-based compensation, and
recruiting managers less susceptible to knowledge overcondence, a trait associated with exit
delay. Copyright © 2015 John Wiley& Sons, Ltd.
Man prefers to believe what he prefers to be
true. —Sir Francis Bacon
INTRODUCTION
Exit delay— persisting in a venture when evidence
indicates it will be unprotable— is a substantial
Keywords: behavioral strategy; escalation of commitment;
behavioral bias; real options; exit
*Correspondence to: Daniel W. Elfenbein, Campus Box 1156,
One Brookings Drive, St. Louis, MO 63130-4899, U.S.A. E-mail:
elfenbein@wustl.edu
[Correction added on 18 March 2016, after rst online publica-
tion: The note underneath Table 3 has been amended.]
Copyright © 2015 John Wiley & Sons, Ltd.
economic problem for entrepreneurs and managers
alike (Astebro, Jeffrey, and Adomdza, 2007; DeTi-
enne, Shepard, and Castro, 2008; Dranikoff, Koller,
and Schneider, 2002; Guler, 2007; Horn, Lovallo,
and Viguerie, 2006; Porter, 1976; Vieregger, 2014;
Wennberg etal., 2010). Prior work has examined
exit delay, or failure to exit entirely, as a func-
tion of nonpecuniary motives (Gimeno et al., 1997;
Meyer and Zucker, 1989), agency problems (e.g.,
Jensen, 1993), and decision-making biases, fre-
quently labeled escalation of commitment (Camerer
and Lovallo, 1999; Lowe and Ziedonis, 2006; Staw,
1976, 1981; Staw and Ross, 1978).
The study of exit delay is complicated by the fact
that, against the backdrop of uncertainty, waiting
Equity Stakes and Exit 279
may have option value (Dixit, 1989; Ryan and
Lippman, 2003). Efforts in empirical studies to
take this rational source of delay into account have
been hampered by an inability to precisely identify
an optimal exit point arising from the option value
(Elfenbein and Knott, 2015; O’Brien and Folta,
2009). Similarly, efforts to quantify the extent of
behavioral delay arising from decision-making
biases often ignore rational delay, and are fre-
quently limited in their ability to pinpoint the
specic mechanisms driving delay.
We attempt to address these issues by conduct-
ing a laboratory experiment in which optimal exit
(rational delay relative to the classical investment
theory benchmark) is theoretically well dened. In
it, participants in both treatment and control groups
must infer whether they have been assigned to a
high- or low-prot rm based on identical, noisy
signals of performance. We compensate partici-
pants in both groups for reporting beliefs about the
likelihood they have been assigned to a high-prot
rm using Holt’s (2007) proper scoring rule. Addi-
tionally, the treatment group receives the prots
from the rm and decision rights over exit, that
is, equity stakes, while the control group does
not receive these prots or decision rights, but
does offer recommendations about exit. For expo-
sitional purposes, we henceforth label the treatment
group “entrepreneurs,” and label the control group
“advisors.”1
The experiment employs the structure of infor-
mation and payoffs in the Ryan and Lippman (2003)
model of optimal exit from a project with noisy
returns. This allows us to characterize the opti-
mal exit point, and compare the behavior of both
groups to that baseline. We can, therefore, exam-
ine whether exit beyond the optimum (or behavioral
delay) occurs in the absence of sunk costs, agency
problems, or other organizational phenomena, and
the degree to which equity-like compensation itself
may affect the timing of exit.
While the experiment corroborates some prior
results regarding the existence of entrepreneurial
exit delay (e.g., Åstebro et al., 2007; Sandri et al.,
2010), we obtain intriguing new results that allow
us to peer into delay’s black box. Our rst nd-
ing is that our “entrepreneurs” exhibit delay relative
to the optimum when assigned to low-prot rms.
Entrepreneurs also exit later than advisors, whose
1We recognize, of course, that other interpretations of the treat-
ment and control groups are possible.
exit recommendations are, on average, nearly opti-
mal for low-prot rms.2The difference in exit out-
comes between entrepreneurs and advisors suggests
that equity stakes alone may be a sufcient condi-
tion for inducing behavioral delay.
To understand how equity stakes can induce
behavioral delay, we examine the evolution of
entrepreneurs’ and advisors’ beliefs. We nd
two signicant differences between them. First,
entrepreneurs exhibit lower exit thresholds than
advisors. Entrepreneurs require greater certainty
they are unprotable before exiting. Second, while
both groups exhibit asymmetric updating (modify-
ing their beliefs more in response to positive versus
negative signals), the problem is more pronounced
among entrepreneurs. Interestingly, the difference
stems exclusively from a reduced responsive-
ness in sensitivity to negative signals among
entrepreneurs— both groups had comparable
responses to positive signals. These ndings pro-
vide support for some biases that affect belief updat-
ing, such as motivated reasoning and conrmatory
bias, but are inconsistent with others, such as
anchoring or status quo bias. Further, because our
experimental design does not allow participants to
change their rm’s type (for example through hard
work or good choices), we can rule out ability over-
condence— proposed as a leading cause of both
excess entrepreneurial entry and exit delay (Åstebro
et al., 2014)— as a cause of delay in this setting.
Finally, we nd that extreme persistence, that is,
failure to exit entirely when assigned to low-prot
rms, is positively related to an individual’s
observed level of asymmetric updating. Failure to
exit among participants in our experiment is also
systematically related to some individual attributes
(e.g., internal locus of control, knowledge overcon-
dence), but not to others (e.g., math and statistics
ability, risk preferences), raising the possibility that
screening may be able to identify the entrepreneurs
or managers who are at greatest risk for exit delay.
The most important implication from the study,
perhaps, is one that extends beyond the question of
exit delay to organizational economics more gen-
erally. Typical models in organizational economics
assume that incentives shape behavior conditional
on beliefs. We nd that incentives may fundamen-
tally shape beliefs as well.
2Entrepreneurs’ exit decisions, then, are consistent with rational
and behavioral delay together, whereas advisors’ exit recommen-
dations are consistent with rational delay alone.
Copyright © 2015 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 278–299 (2017)
DOI: 10.1002/smj
To continue reading
Request your trial