Entrepreneurs Under Uncertainty: An Economic Experiment in ...people.soc.cornell.edu/nee/pubs/EntrepreneursUnderUncertainty.pdf · Published online in Articles in Advance January

This article was downloaded by: [128.84.124.81] On: 25 September 2017, At: 22:23Publisher: Institute for Operations Research and the Management Sciences (INFORMS)INFORMS is located in Maryland, USA

Management Science

Publication details, including instructions for authors and subscription information:http://pubsonline.informs.org

Entrepreneurs Under Uncertainty: An EconomicExperiment in ChinaHakan J. Holm, Sonja Opper, Victor Nee,

To cite this article:Hakan J. Holm, Sonja Opper, Victor Nee, (2013) Entrepreneurs Under Uncertainty: An Economic Experiment in China.Management Science 59(7):1671-1687. https://doi.org/10.1287/mnsc.1120.1670

Full terms and conditions of use: http://pubsonline.informs.org/page/terms-and-conditions

This article may be used only for the purposes of research, teaching, and/or private study. Commercial useor systematic downloading (by robots or other automatic processes) is prohibited without explicit Publisherapproval, unless otherwise noted. For more information, contact [email protected].

The Publisher does not warrant or guarantee the article’s accuracy, completeness, merchantability, fitnessfor a particular purpose, or non-infringement. Descriptions of, or references to, products or publications, orinclusion of an advertisement in this article, neither constitutes nor implies a guarantee, endorsement, orsupport of claims made of that product, publication, or service.

Copyright © 2013, INFORMS

Please scroll down for article—it is on subsequent pages

INFORMS is the largest professional society in the world for professionals in the fields of operations research, managementscience, and analytics.For more information on INFORMS, its publications, membership, or meetings visit http://www.informs.org

http://pubsonline.informs.org

https://doi.org/10.1287/mnsc.1120.1670

http://pubsonline.informs.org/page/terms-and-conditions

http://www.informs.org

MANAGEMENT SCIENCEVol. 59, No. 7, July 2013, pp. 1671–1687ISSN 0025-1909 (print) � ISSN 1526-5501 (online) http://dx.doi.org/10.1287/mnsc.1120.1670

© 2013 INFORMS

Entrepreneurs Under Uncertainty:An Economic Experiment in China

Hakan J. Holm, Sonja OpperDepartment of Economics, Lund University, 22007 Lund, Sweden

{[email protected], [email protected]}

Victor NeeDepartment of Sociology, Cornell University, Ithaca, New York 14850, [email protected]

This study reports findings from the first large-scale experiment investigating whether entrepreneurs differfrom other people in their willingness to expose themselves to various forms of uncertainty. A stratified

random sample of 700 chief executive officers from the Yangzi delta region in China is compared to 200 controlgroup members. Our findings suggest that in economic decisions, entrepreneurs are more willing to acceptstrategic uncertainty related to multilateral competition and trust. However, entrepreneurs do not differ fromordinary people when it comes to nonstrategic forms of uncertainty, such as risk and ambiguity.

Key words : economics; behavior and behavioral decision making; decision analysis; risk; microeconomicbehavior

History : Received November 28, 2011; accepted September 14, 2012, by John List, behavioral economics.Published online in Articles in Advance January 8, 2013.

1. IntroductionWhy entrepreneurs routinely accept the high uncer-tainties associated with entrepreneurial activities hasfascinated social scientists over centuries. Not onlydo payoffs of investment take place over time; butprospective payoffs are highly uncertain because ofthe novelty of entrepreneurial activities (Say 1855,Kirzner 1973, Kihlstrom and Laffont 1979, Knight2006). What accounts for the entrepreneur’s capacityfor making highly concentrated equity investmentsin spite of the considerable risks entailed (Hamilton2000, Moskowitz and Vissing-Jorgensen 2002)? Andwhy do entrepreneurs accept the uncertainties of mar-ket cooperation, if contracting and governance remainincomplete (Williamson 1987)? Given the nature ofentrepreneurial action, most theories of entrepreneur-ship and entrepreneurial decision making emphasizeuncertainty as a conceptual cornerstone. Two broadexplanations can be identified. One focuses on theability to develop strategic responses in the pres-ence of uncertainty embedded in the environment(Pich et al. 2002, Bernstein and Federgruen 2005); theother emphasizes behavioral traits that distinguishentrepreneurs from nonentrepreneurs in their willing-ness to accept uncertainty.

The emphasis on identifying distinct “entrepre-neurial types” reflects the popularity of the latterapproach; however, the evidence confirming higheruncertainty tolerance of entrepreneurs is largely

inconclusive, in spite of intensifying research effortsover the last three decades. In reviewing some ofthe most influential studies published across thesocial science disciplines, a few observations standout (see the appendix for a literature review): Firstof all, the empirical research has focused on stan-dard risk situations with known probabilities of dif-ferent outcomes. Standard risk situations, however,do not represent typical scenarios in entrepreneurialdecision making. More often than not, entrepreneursact under nonstrategic uncertainty with unknownprobabilities of outcomes. Similarly, prior researchhas almost completely neglected study of behav-ioral responses under strategic forms of uncertainty(where the outcome is contingent on responsesof another individual as typical in any bilateralbusiness contract). Second, methodological limita-tions of sampling strategy and subject pool donot invite general inferences on behavioral traits ofentrepreneurs. Studies confirming higher risk tol-erance of entrepreneurs have typically employedrelatively small samples of non-randomly selectedpopulations (Begley 1995, Koh 1996, Sarasvathy et al.1998, Stewart et al. 1999). Confirmatory evidencefrom randomly sampled survey populations, in con-trast, almost exclusively involves studies compar-ing self-employed individuals with those who reportthat they have never been self-employed, but theyoverlook the more narrowly defined category of

1671

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Holm, Opper, and Nee: Entrepreneurs Under Uncertainty1672 Management Science 59(7), pp. 1671–1687, © 2013 INFORMS

entrepreneurs whose firms actually employ wage-labor (Van Praag and Cramer 2001, Uusitalo 2001,Cramer et al. 2002, Caliendo et al. 2010). Moreover,recent attempts to compare randomly sampled pop-ulations of entrepreneurs and nonentrepreneurs haveproduced mixed results (Djankov et al. 2006, 2007;both studies use a comparable survey instrument).Third, experimental studies using incentivized tasksshed further doubt on the assumed behavioral dif-ferences between entrepreneur and nonentrepreneurs.Evidence based on nonrandomly selected samplessuggests that entrepreneurs do not behave differ-ently than nonentrepreneurs when exposed to situa-tions of standard risk (Elston et al. 2006, Macko andTyszka 2009, Sandri et al. 2010, List and Mason 2011),although they may be better able to cope with uncer-tainty (Macko and Tyszka 2009).

Our approach builds on the research of Elston et al.(2006) and Macko and Tyszka (2009) using controlledincentivized tasks (instead of psychometric surveymeasures) to elicit behavioral differences betweenentrepreneurs and nonentrepreneurs. In particular, weexploit well-established methods from experimentaleconomics, namely, the multiple price list method(see Binswanger 1980, Holt and Laury 2002, Andersenet al. 2006).1 To take research a step forward, we intro-duce two methodological innovations: First, insteadof relying on convenience samples, our study is thefirst that utilizes large-scale samples of entrepreneursand nonentrepreneurs randomly selected from firmand household registers. Second, we employ a mul-tidimensional analysis incorporating four types ofuncertainty rather than just one behavioral trait.On one hand, we employ standard risk and ambiguityas common forms of nonstrategic uncertainty salientin entrepreneurial decision making. On the other,we use competition (where uncertainty concerns anindividual’s performance relative to others) and trust(where there is a “social” risk that another party doesnot act favorably toward the trustee) to exemplify sit-uations involving strategic forms of uncertainty.

Experimental research involving entrepreneursfaces the practical question of whether monetaryincentives are high enough to offer sufficiently attrac-tive rewards. Our research site in the Yangzi deltaregion in China—an emerging economy known tobe one of the most entrepreneurial regions—partly

1 Harrison and List (2004) used the term “artefactual” to describefield experiments (like this one) that expose uncommon subject cat-egories to experimental procedures normally used in the economiclaboratory. We use an “abstract” rather than a “natural” framewhen eliciting attitudes to uncertainty. The strategy to provideboth entrepreneurs and nonentrepreneurs with a neutral scenariois also likely to reduce the impact of so called background risk, aproblem proven to be associated with “natural” frames (Harrisonet al. 2007b).

alleviates this concern in light of its lower percapita income relative to a developed economy (Neeand Opper 2012). Our experimental group includes700 entrepreneurs randomly sampled from local firmregisters. All entrepreneurs have been in business forat least three years and employ at least 10 salariedworkers. The focus on established business seemsmost appropriate to capture the “entrepreneurialfunction” discussed in entrepreneurship theories.Two hundred randomly sampled nonentrepreneurslocated in the same region serve as control group.Our motivating question is do entrepreneurs displayunique behavioral traits that distinguish them fromnonentrepreneurs. Our study does not address anequally important question of who actually becomesan entrepreneur, nor do we have clear evidencewhether entrepreneurial traits are acquired by learn-ing or stem from self-selection (Evans and Leighton1989, Shane 2003, Lazear 2004).

2. Theory and HypothesesEntrepreneurs face multiple dimensions of uncer-tainty, both nonstrategic and strategic. We focus onrisk and ambiguity to examine nonstrategic formsof uncertainty. Risk involves situations in which thedecision maker has information about the probabilityof different outcomes and can choose between differ-ent alternatives. According to neoclassical utility the-ory, such decisions are affected by the curvature ofthe individual’s utility function for money. We defineambiguity, as did Ellsberg (1961), as situations whereeconomic actors have information about conceivableoutcomes, but not about their probabilities.

The characteristic feature of strategic forms of un-certainty is that the outcome of decision making iscontingent upon other individuals’ actions. Here wefocus on competition and trust. In both cases, beliefsabout other people are likely to influence individualdecisions. To assess tolerance for uncertainty stem-ming from competition, we ask the participant tochoose between a competitive situation and a non-competitive one. Clearly, the choice will depend onthe decision maker’s beliefs about his or her own per-formance relative to others and also to some extent onher preference for competition per se. To assess trust,we let participants choose between delegating a dis-tribution task to a stranger or to a random device. Evi-dently, decisions will reflect a decision maker’s beliefswhether or not a stranger is more likely to take afavorable decision than a random device.

2.1. RiskIn Kihlstrom and Laffont (1979), economic agents canchoose between supplying their labor on the labormarket to secure relatively risk-free wage employ-ment or becoming an entrepreneur facing risky

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Holm, Opper, and Nee: Entrepreneurs Under UncertaintyManagement Science 59(7), pp. 1671–1687, © 2013 INFORMS 1673

income prospects. In the equilibrium, less risk-averseagents become entrepreneurs, and more risk-averseagents become wage earners. From this theory, wederive our first hypothesis.2

Hypothesis 1A. Entrepreneurs differ from others withrespect to risk taking.

Prospect theory suggests that people use weight-ing functions instead of probabilities when choos-ing between different alternatives (Kahneman andTversky 1979). Except for certain outcomes (with prob-ability one), actors put too little weight on outcomesnot obtained with certainty. This phenomenon—denoted as “subcertainty” or as “certainty effect”—explains risk aversion in choices involving sure gainsand drives results like the Allais paradox. Because itis plausible that entrepreneurs differ from others intheir certainty preference, we compare entrepreneurswith ordinary people when choosing between a cer-tain and a risky alternative (where weights do notsum to one). Potentially, this allows us to separatebetween differences relating to the shape of the so-called “value function” and differences with respectto the “weighting function.”3 Thus we specify the fol-lowing hypothesis:

Hypothesis 1B. When facing a risky alternative,entrepreneurs have a different preference for guaranteedoutcomes compared to ordinary people.

2.2. AmbiguityKnight (2006, p. 231) was one of the first to empha-size that business decisions typically involve unmea-surable risk, because they “deal with situations whichare far too unique, generally speaking, for any sort ofstatistical tabulation to have any value for guidance.The conception of an objectively measurable proba-bility or chance is simply inapplicable.” Such uncer-tainty poses a dilemma for firms, in that economicactors must make investment and production deci-sions that shape long-term business strategy and per-formance despite not being able to assess downstreamrisks. Knight (2006) does not claim that entrepreneurshave a higher or lower aversion to uncertainty thanothers, but asserts that entrepreneurs may have ahigh “capacity for forming correct judgments,” imply-ing that entrepreneurs behave differently from otherswhen acting under uncertainty (Knight 2006, p. 43).

2 Although the crucial assumption of Kihlstrom and Laffont (1979)implies a direction of the difference in risk taking, we prefer tokeep our hypotheses open to allow for two-sided tests, because thispaper does not aim to test isolated theories. The theories are merelyused to motivate the investigation of different forms of uncertainty.3 This is the more exact test of Kihlstrom and Laffont’s (1979) the-ory, because here individuals chose between a certain and a riskyoutcome.

To reserve “uncertainty” as the more general con-cept, we use in the following the term ambiguity forsituations where the probability distributions of theoutcomes are completely or partially unknown. Ourhypothesis is therefore as follows:

Hypothesis 2. Entrepreneurs have a different degree ofambiguity aversion than others.

2.3. Willingness to CompeteAn essential feature of entrepreneurship is exposureto competition, which involves a risk contingent onthe entrepreneur’s performance in comparison to thecompetitors. There are at least two conceivable mech-anisms behind differences in the entrepreneur’s will-ingness to compete with others. The first concernsbeliefs about one’s own performance relative to oth-ers. It is possible that entrepreneurs are more opti-mistic than others or may even be overconfident.The theoretical literature draws on this to explainexcessive market entry (Bernardo and Welch 2001,Hayward et al. 2006, Wu and Knott 2006).4 The otherpossible mechanism is that entrepreneurs may havea preference for competition per se. For instance,Marshall (1920, p. 23) claimed that “a manufactureror a trader is often stimulated much more by thehope of victory over his rivals than by the desire toadd something to his fortune.” Similarly, Schumpeter(1983, p. 93) interpreted the wish to innovate, to suc-ceed, and to prove superiority as an exogenous fac-tor having its roots in a person’s Unternehmergeist, theentrepreneurial spirit. It is “the will to conquer: theimpulse to fight 0 0 0 to succeed for the sake, not ofthe fruits of success, but of success itself” that setsentrepreneurs apart from others. This leads us to ourthird hypothesis:

Hypothesis 3. Entrepreneurs have a different willing-ness to compete than others.

2.4. TrustIn the words of Coleman (1990, p. 91), “Situationsinvolving trust constitute a subclass of those involv-ing risk. They are situations in which the risk onetakes depends on the performance of another actor.”Trust is the willingness to expose oneself to suchuncertainty. Entrepreneurs are constantly exposed tothese “social risks” when interacting with suppli-ers, customers, employees, debtors, and their com-petitors. A delicate task here is to strike a balancebetween trust and control, a task entrepreneurs maybe particularly good at (Say 1855, Sections III, VII,p. 30). Thus, whether or not trust in general is higheror lower among entrepreneurs compared to averagepeople, entrepreneurs may in a given situation have a

4 Overconfidence and optimism are general human tendencies thatare not restricted to competitive situations and should be distin-guished from the more narrowly defined willingness to compete.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


different level of trust than others (Knight 2006, p. 43).Our fourth hypothesis states:

Hypothesis 4. Trust behavior among entrepreneurs isdifferent from that among others.

3. Research Strategy and DesignThis study combines five features previously notjointly applied to study entrepreneurial behaviorunder uncertainty: (1) reliance on incentivized behav-ioral tasks, (2) multidimensional specification ofuncertainty, (3) a focus on entrepreneurs as foundersand chief executive officers (CEOs) of industrial firms,(4) generation of a large-scale sample, and (5) relianceon random-sampling of participants.

3.1. DesignMost of the research to date is based on surveysexploring psychological attitudes toward uncertainty(see the appendix). This makes the economic inter-pretation of these earlier results somewhat diffi-cult. In this experiment, the behavioral tasks weuse allow for a relatively straightforward interpre-tation. For instance, the behavioral task measuringrisk can be directly related to the shape of theutility function. Furthermore, we agree with manyexperimental economists that the use of monetaryincentives in behavioral tasks increases the credibil-ity of the observed data.5 After all, entrepreneurialdecisions involve real stakes. The introduction ofmonetary incentives therefore should increase theexternal validity of our observations. Another impor-tant feature of our design is the parallel study ofdifferent facets of uncertainty. Testing hypotheses foreach of these dimensions of uncertainty with thesame research design allows us to better distinguishbetween different behavioral characteristics. Considerthe following example: The decision to trust gener-ates a distribution of uncertain outcomes, and theexpected utility of this distribution is affected by thecurvature of the decision maker’s utility function inthe same way as for standard risk decisions. In addi-tion, beliefs about others’ behavior affect trust. Evi-dently, the observation that entrepreneurs are moreinclined to trust than others is open to two competingexplanations. By also testing for risk preferences it ispossible to distinguish between the two.

In our study, we formulated each behavioral taskin a multiple price list format, where option Awas expected to be the most attractive for thefirst decisions, whereas the relative attractiveness ofoption B grew for decisions further down the list. This

5 For a theory of incentives in experiments, see Smith (1982). Thereis also evidence that subjects put a higher degree of cognitive effortin decisions when these are incentivized (Camerer and Hogarth1999). Attitudinal responses may even be unrelated to incentivizeddecisions (Glaeser et al. 2000).

was done to minimize the cognitive load of the sub-jects and to make comparisons between tasks easier.After a number of pilot tests with undergraduates inSweden and a pretest with 70 entrepreneurs locatedin the Yangzi delta region, we decided to presenteach decision separately, which seemed to make thetasks easier to grasp for the entrepreneurs. Belowwe describe the different tasks and how they weredesigned.6

3.1.1. Risk. The experiment utilized a multipleprice list format to elicit risk aversion (see Binswanger1980, Holt and Laury 2002), which has become moreor less the standard design in experimental studies ofrisk aversion (see Andersen et al. 2006, 2008). We usedtwo different lists, one with two lotteries of differentrelative risk (called task R1) and one with a certainoutcome and a risky lottery (called task R2). This wasdone to investigate both the standard risk attitudesand the certainty effect (discussed above). The list forthe R1 task is presented in Table 1 (with the deci-sions in the middle omitted for reasons of space). TheR2 task has the same specifications for option A asthe A1 task and the same specifications for option Bas the R1 task.

3.1.2. Ambiguity Aversion. Ambiguity aversionwas elicited by letting outcomes for option A be cer-tain and the probabilities for the outcomes in option Bbe fully or partially unknown, denoted as the A1and A2 tasks, respectively (see Table 1).7 This meansthat further information about the risky lottery isremoved, which may increase the feeling of uncer-tainty. To investigate the degree of ambiguity aver-sion, we need some “price” of the ambiguous lotteryreflecting the aversion. This method is similar to theone used by Fox and Tversky (1995), who elicited thewillingness to pay for some of their ambiguous lot-teries. When the subject in our design switches fromthe certain (option A in the A1 task) to the ambiguousalternative (option B in the A1 task), we take this asan indication that the subject has reached her reser-vation price. In some business decisions one mightknow a lower or upper bound on the probabilityfor an event but not the exact probability. To elicitthe aversion to such situations, we also includeda treatment with partially ambiguous lotteries (theA2 task).8

6 The instructions and forms used are provided in the onlinesupplement (available at http://www.nek.lu.se/NEKJHO/MS_SUPPLMATERIAL.pdf).7 The difference between A1 and A2 was that both uncertain out-comes (i.e., in option B) in the latter task had a probability ofat least 25%.8 This is also motivated by the finding that ambiguity aversionmay differ in various groups depending on the level of ambiguity(Borghans et al. 2009).

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Table 1 Decision Tasks

Task R1 Task A1

Option A Option B Option B(probabilities (probabilities (probabilities

Decision of payoffs) of payoffs) Option A of payoffs)

1 10% of ¥300 10% of ¥580 ¥360 ?% of ¥58090% of ¥240 90% of ¥15 ?% of ¥15

2 20% of ¥300 20% of ¥580 ¥330 ?% of ¥58080% of ¥240 80% of ¥15 ?% of ¥15

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

9 90% of ¥300 90% of ¥580 ¥120 ?% of ¥58010% of ¥240 10% of ¥15 ?% of ¥15

10 100% of ¥300 100% of ¥580 ¥90 ?% of ¥580?% of ¥15

Task C1

Option B Task T(amount earned per

correct answer if Option AOption A you/your (X decides Option B

(amount earned coparticipant between (probabilitiesper correct have/has the highest payments of payments

Decision answer) number of points) I and II) I and II)

1 ¥50 ¥50/¥5 X decides 100% of II2 ¥45 ¥50/¥5 X decides 10% of I

90% of II0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

9 ¥10 ¥50/¥5 X decides 80% of I20% of II

10 ¥5 ¥50/¥5 X decides 90% of I10% of II

3.1.3. Willingness to Compete. The willingnessto compete may be triggered by many mechanisms,and depending on which the subject considers impor-tant, there are different methods to elicit it. Onemethod related to the one we used is to let subjectschoose between a fixed fee and entering a contestwith an unknown endogenously determined numberof entrants, in which the winner gets a larger fixedprize (see Camerer and Lovallo 1999, Elston et al.2006). The subjective expected utility of the uncer-tain alternative will depend on many things, includ-ing beliefs about the number of entrants, the shapeof the utility function, the belief about the subject’sown performance relative to others, and possibly alsopreferences for competition per se (as suggested byMarshall 1920).9 Because the shape of the utility func-tion is investigated in the standard risk task, thepurpose of our elicitation method is to get a moredistinct measure of beliefs about relative performance

9 Such beliefs may be linked to the decision to become anentrepreneur. Koellinger et al. (2007) demonstrated in a cross-country study that the decision to start a firm is correlated withbeliefs about business skills.

and preferences for competition per se. To dampenthe risk element, we compare two uncertain alter-natives, where one involves a competitive element.10

We also believe that in many situations, the alter-native to an entrepreneurial income (at least in thelong run) is not a fixed salary, which is the same forall, but rather a performance-based salary, which (likethe entrepreneurial income) has an expected valueassociated with beliefs about absolute performance.The payoff for the winner in the competitive alter-native is therefore scaled up from the individual one(see Table 1), which means that beliefs about absoluteperformance are more or less controlled for. Hence,the measure emphasizes the subject’s beliefs abouther relative performance and potential preferences forcompetition per se. To get an indication of subjects’belief about relative performance, we asked subjectsto guess their performance in the task compared totheir coparticipant(s). Because the performance taskwas a trivia quiz, we asked “If you do the quiz, whatpercentage of the other participants do you think willhave more correct answers than you?”

In task C1 (see the Table 1) subjects could choosebetween being paid a certain sum (from CNY 5 toCNY 50) per correctly answered question (option A)in the quiz or entering a competition with anothersubject and being paid CNY 50 per correctly answeredquestion if the subject had the highest number ofcorrect answers and CNY 5 if the competitor hadthe highest number.11 We also included a task C2where competition was multilateral instead of bilat-eral. In this case the subject could choose between thesame option A as in C1, but option B concerned acompetition with three other participants, thus mak-ing the competition fiercer.

3.1.4. Trust. An individual’s aversion to exposingherself to the discretion of another person (the trustee)involves both a component of risk (in the sense thatmore than one outcome is possible) and a belief com-ponent (the trustor’s subjective belief that the trustee’saction will be advantageous to him, or not). Althoughthese components are seldom separated in the litera-ture on trust behavior, we believe it is important notto confound trust with low levels of risk aversion.We have therefore designed an elicitation methodwhere the trustor chooses between one socially riskyoption, in which the outcome is conditional upon thetrustee’s action (option A), and a risky option (B) inthe standard “lottery” sense, with different paymentprobabilities. The trustee/lottery will then decideabout two different payment distributions: Payment

10 This design was partly inspired by Gneezy et al. (2003) andCharness and Villeval (2009).11 In the trivia quiz, we used the quiz performance from a pretestgroup of entrepreneurs as competitors.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


I gives the trustor CNY 580 and the trustee CNY 50.Payment II gives the trustor CNY 15 and the trusteeCNY 55. The T task is formulated in a multiple pricelist format of 10 individual choices, where option A,i.e., reliance on the trustee’s decision, is expected tobe the most attractive for the first decision, whereasthe relative attractiveness of option B, i.e., relianceon the payment decision via a random lottery out-come, increases for decisions further down the list(see Table 1). The level of trust in others is therebyrevealed by the switching point at which the trustorrather relies on a lottery decision than on another per-son. Note that the trustor’s decision cannot in anyobvious way be driven by social preferences.12

A small group from the staff at the ShanghaiAcademy of Social Sciences served as trustees. Deci-sions from this group were recorded for all con-ceivable contingencies (i.e., the strategy method wasused). To participants in our experiment, the otherperson was described in general terms (as an anony-mous person who was born and lives in China), tocapture what Putnam (2001) described as “thin” trust.

The general design of the task simulates situationswhere one person at a low personal cost can make abig difference to another person. For instance, at theend of the workday, an employee might detect that amachine needs lubrication. The employee can choosebetween lubrication, which costs a few minutes extrawork, or ignore it, which will damage the machineand incur costs to the owner of the firm.

3.1.5. The Distribution of Tasks. Each subjectwas exposed to four different tasks, one for each typeof uncertainty (see Table 2). To have some observa-tions uncontaminated by potential order effects and tocheck for certain treatment effects, the entrepreneurswere divided into six different treatment groups. Thedistribution of tasks and their order were decidedafter several pilots and one pretest. Different issueslike cognitive difficulties were taken into account.Each task was placed first in at least one treatment.The exception was task T, because this was the mostdifficult to grasp. However, it seemed to be easierto understand if the subject had first understood themultiple price list format through some of the othertasks. To minimize the potential influence of ordereffects in between subjects’ comparisons, we madesure to have the control group in their treatmentsgetting mostly the same sequences of tasks as theentrepreneurs.

12 For instance, all standard parameterizations of the Fehr andSchmidt (1999) model would suggest that the trustor prefers thefirst to the second payment. However, it is quite possible that thetrustor’s behavior can be guided by beliefs about the trustee’s socialpreferences, but this is something different and a reasonable com-ponent in trust.

Table 2 Treatment Design and Mean Switching Points

Entrepreneurs Control group

Treatment 1 2 3 4 5 6 Treatment 7 8 9 10

1st task A1 A2 R1 R2 C1 C2 1st task A1 R2 R1 C12nd task R2 R1 A2 A1 T T 2nd task R2 A1 A2 T3rd task T T T T A1 A2 3rd task T T T A14th task C1 C2 C1 C2 R2 R1 4th task C1 C2 C1 R1# subjects 117 117 117 117 116 116 # subjects 50 50 50 50

Mean switching pointsa Entrepreneurs Control group

R1 task 5098 (1.69; 111) 5077 (2.36; 48)R2 task 5086 (1.80; 234) 5030∗∗ (1.86; 98)A1 task 6025 (2.05, 347) 6011 (2.45; 147)A2 task 6022 (2.10; 115) 6028 (2.43; 49)C1 task 6021 (1.69; 344) 6039 (1.91; 147)C2 task 6010 (1.71; 116) 6076∗ (1.97; 50)T task 5010 (2.00; 464) 4079∗∗ (1.95; 197)

aMeans and tests are based on treatments, where the entrepreneurs andthe control group share the same history (or histories) of tasks up to the taskconsidered (e.g., the R1 task is based on T1 and T9, R2 is based on T1, T7,T4, and T8). Standard deviations and number of observations in parentheses.

∗Statistically significant at the 10% level; ∗∗statistically significant at the5% level (Wilcoxon–Mann–Whitney test).

3.2. Sampling and Descriptive StatisticsA large-scale experiment involving 700 randomlyselected entrepreneurs would be a demanding under-taking anywhere in the world. Owners and CEOsof established firms are rarely willing to devotetheir scarce time to time-consuming academic studies.Many studies of entrepreneurial characteristics havefor this reason focused on self-employed individuals.This may be perfectly reasonable for studying char-acteristics related to some specific forms of activitiessuch as start-up strategies of very small businesses.However, because some individuals are pushed toself-employment for lack of alternatives, such subjectsobviously do not reflect the entrepreneurial functionsthat Say (1855), Marshall (1920), Schumpeter (1983),Knight (2006), or Kirzner (1973) studied. Other stud-ies secure participation of entrepreneurs by relying onsmall-scale “convenience samples.”13 In these studies,researchers visit public events attended by the sur-vey population and let subjects then self-select intothe subject pool. Although understandable in practi-cal terms, the use of convenience samples naturallyraises concerns over the external validity of findings(Levitt and List 2007).

We were able to mitigate some of these prob-lems by appending our experiment to a multiyearresearch project already in progress since 2006 in theYangzi delta region utilizing a random sample of

13 This is not necessarily a problem, because large differencesbetween entrepreneurs and others can also be detected in smallsamples. What is more problematic is that studies with few obser-vations that do not detect any differences may have a limited value,because small samples may mask moderate differences. This in turnmay lead to a publication bias that favors studies reporting signif-icant differences.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


entrepreneurs running sizeable industrial firms (seeNee and Opper 2012, pp. 52–71). The economy ofscale allowed us to offer credible monetary rewardsin our experiments. Our study thus uses a strati-fied random sample of firms, stratified according tocity, industry, and firm size and selected from 7 ofthe region’s 16 municipalities (Nanjing, Nantong, andChangzhou in Jiangsu province; Wenzhou, Ningbo,and Hangzhou in Zhejiang province; and Shanghaimunicipality) from which 100 entrepreneurs wererecruited for each city (for a total of 700). The sam-ple of private enterprise excludes newly foundedand small-scale household firms from the samplingpool, and only includes those firms that survivedthe first three years of firm operation (with a sam-ple median of eight years in business). To narrowdown industrial diversity, the sampling procedureselected five industries reflecting strong local produc-tion lines. These industries range from labor-intensiveto technology-intensive productions covering textiles,ordinary machinery, vehicle and auto parts, medi-cal and pharmaceutical products, and computer andcommunication equipment. To reach sizable establish-ments, the sample oversampled “large” (more than300 employees) and “medium-size” (100–300) firms.

The recruitment of participants followed a two-stage procedure. In a first step, firms were ran-domly sampled from local firm registers of China’sBureau of Industry and Commerce. The control groupconsists of individuals randomly selected from localhousehold registers in the same region that resem-bles the entrepreneurs with respect to gender and age.In a second step, the survey team contacted the ran-domly selected firms and households by mail andphone to arrange for interview appointments. In total,2,842 entrepreneurs were invited to participate yield-ing an overall response rate of 25%, an acceptableresult for studies targeting CEOs and top man-agers.14 On average, the participating firms employ130 employees falling into the category of medium-size private firms. Eighty-six percent of these CEOswere owners, and 78% were founders or cofoundersof the firm, thereby qualifying as “entrepreneurs” inthe most literal meaning. Given the fairly large sam-ple of 700 entrepreneurs and 200 individuals sam-pled as a control group, observations of potential “nodifferences” are of great interest, because one can berelatively sure not to overlook substantive differencesin the underlying distributions.

Clearly, the ideal control group would be one thatis identical to the entrepreneurs except that they are

14 Based on a survey including 175 different studies published in theyears 1975, 1985, and 1995 in top-tier academic journals in manage-ment and behavioral studies, Baruch (1999) identified a norm valueof 3505% ± 1303 for studies involving top management, whereasmean values in non-Western societies tend to be even lower.

not entrepreneurs. But such a sampling frame doesnot exist. Our strategy instead was to employ a richset of controls including the standard control vari-ables gender, age, years of education, and income.We also add four multidimensional measures to cap-ture less standard variables like career experienceand parental characteristics, both aspects that notonly may well influence the decision to become anentrepreneur (Shane 2003, Chap. 4; Lindquist et al.2012), but may also have a continuing effect on behav-ior under uncertainty (Dohmen et al. 2011). First,we include 10 dummy variables controlling for thelast position individuals held before the current occu-pation (i.e., becoming an entrepreneur). The occu-pational categories include technical personnel, salesand marketing staff, accounting and finance staff,administrative officer, enterprise director, ordinaryworker, retail service staff, farmer, military personnel,and unemployment. Second, we include six dummyvariables capturing the father’s highest level of educa-tion and the same 10 dummy variables to capture thefather’s occupation before retirement as used to clas-sify the individual’s prior occupation. Furthermore,we include 10 dummy variables to classify the orga-nizational unit of the father’s last occupation. Theseinclude research institution, higher education institu-tion, party and government organization, state-ownedenterprise, red-hat firm, collective enterprise, privateenterprise, individual household business, rural col-lective enterprise, and foreign invested firm.15

Table 3 summarizes the descriptive statistics ofthe 900 participants in our study. The matching ofgender and age in the control group worked well,with an almost identical composition of subjects withregard to these variables. As to educational attain-ment, entrepreneurs have a higher level of educa-tion than the average control group member. Anotherfactor to control for is personal income, which isexpected to differ between both groups. According toself-reports of annual personal income, entrepreneursearn substantially more than the control group mem-bers. The median annual personal income of theentrepreneur was CNY 175,000 per year, whereascontrol group members earned only CNY 21,600. Inline with previous research, entrepreneurs have a par-tially different career background from those in thecontrol group, which is indicated by the differencesin jobs before the current one (in Table 3). Also, theparental background differs in some characteristics,mainly with respect to the father’s organizational unit.

15 Jointly these four measures provide a fine-grained account of therespondent’s personal background. For instance, information thata respondent was brought up by a college graduate working as atechnician in a state-owned firm provides a precise picture of thesocial and economic background of a member of the urban middleclass. Compare this with the lower end of the social strata, the childof a farmer without formal education.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Table 3 Descriptive Statistics

Entrepreneurs Control groupVariable (mean/SD) (mean/SD) t-value

Male 00833 00855 007494003735 4003535

Age 430056 420435 −009274802835 4805765

Years of formal education 12093 11076 −40990∗∗∗

4209185 4209475Log of income 50120 20939 −310485∗∗∗

4007445 4100205Job before current job

Technical personnel 0012 002 20905∗∗∗

4003255 4004015Sales/marketing 00064 0011 20175∗∗

4002455 4003145Accounting/finance 00017 00065 30655∗∗∗

4001305 4002475Administrative officer 00569 00255 −80093∗∗∗

4004965 4004375Enterprise director 00167 0007 −30454∗∗∗

4003735 4002565Ordinary worker 00023 00205 90803∗∗∗

4001505 4004055Retail service staff 00006 0002 10900∗

4000755 4001405Farmer 00003 0 −00756

4000535 (0)Military personnel 00001 0 −00534

4000385 (0)Unemployed 0 00005 10873∗

(0) 4000715Other 00024 00065 20834∗∗∗

4001545 4002475Father’s highest education

No formal education 00144 00075 −20586∗∗∗

4003525 4002645Primary school 00256 00305 10390

4004375 4004625Junior high school 00279 0027 −00239

4004495 4004455Vocational school/high school 00209 00245 10102

4004075 4004315Junior college 0009 00085 −00219

4002865 4002805Undergraduate education 00021 0002 −00124

4001455 4001405Master degree and above 00001 0 −00534

4000385 (0)Father‘s last occupation

Technical personnel 00161 00145 −005624003685 4003535

Sales/marketing 00013 0002 007464001135 4001405

Accounting/finance 0005 0003 −101964002185 4001715

Administrative officer 00147 0019 104714003555 4003935

Enterprise director 0004 00045 003144001965 4002085

Table 3 (Continued)

Entrepreneurs Control groupVariable (mean/SD) (mean/SD) t-value

Father‘s last occupationOrdinary worker 0034 0032 −00528

4004745 4004685Retail service staff 00006 0001 −00656

4000755 4001005Farmer 00219 0019 −00871

4004145 4003935Military personnel 00001 00005 00945

4000385 4000715Unemployed 0 0 —

(0) (0)Other 00023 00045 10681∗

4001505 4002085Organizational unit/father’s job

Research institute 00009 0001 001904000925 4000105

Higher education institute 00007 00005 −003284000845 4000715

Party/government organization 00004 00045 40471∗∗∗

4000655 4002085State-owned enterprise 00364 00445 20074∗∗

4004825 4004985Red-hat firm 0001 0 −10420

4000105 (0)Collective firm 00206 001 −30434∗∗∗

4004055 4003015Private firm 00126 00125 −00027

4003325 4003325Individual business 00044 0004 −00262

4002065 4001965Rural collective enterprise 00087 00145 20410∗∗

4002825 4003535Foreign invested firm 00001 0001 10856∗

4000385 4001005Other 00141 00075 −20497∗∗

4003495 4002645

Notes. Diff = mean(0) − mean(1). Ha: diff ! = 0. Null hypothesis: Diff is notequal to 0.

∗p < 001; ∗∗p < 0005; ∗∗∗p < 0001.

3.3. Research SiteSkeptics may still question whether China’s economicand political system provides conditions to derivegeneral insights on behavioral differences betweenentrepreneurs and nonentrepreneurs. Generally, how-ever, there is little reason to believe that China’sprivate entrepreneurs differ much from their coun-terparts in other countries. Entrepreneurs have toorganize resources, take decisions under variousforms of uncertainty, negotiate, and compete in ahighly competitive market economy.16 These arecommon qualities generally noted by the classical lit-erature on entrepreneurship, and they are not less

16 All industries in our study are highly competitive with low con-centration ratios. The highest concentration ratios are in the vehicleand electronics sectors, with 20% market share of the top 10 pro-ducers in the country.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


valid in China’s emerging market economy. Zhejiangprovince, the center of the private enterprise economyin the Yangzi delta region already in 2005, ranked ascomparable with the United States in enterprise com-petitiveness (IMD 2005, p. 498). Forbes magazine ranksthe capital of Zhejiang repeatedly as the city with thebest business climate in China.

Historical features of the recent rise of privateentrepreneurship in China even control for someconfounding effects usually present in firm sam-ples in developed economies. At any given pointin time, the structure of a private firm popula-tion naturally depends on a complex set of differ-ent determinants. Some firm owners were simplypushed into entrepreneurship because they lost theirjob, others became second-generation entrepreneursbecause they took over their parents’ firms. Clearly,these firm owners bear little resemblance to theentrepreneurial types or pioneers envisioned by Say(1855), Marshall (1920), or Schumpeter (1983). Ide-ally, one would like to explore a population ofentrepreneurs where individuals actively self-selectto become entrepreneurs. We claim that China pro-vides an appropriate research site to study such pop-ulations of true start-up entrepreneurs. First of all,the history of private entrepreneurship is relativelybrief. Before 1988, private entrepreneurship was noteven legalized, and full constitutional recognition wasnot granted before 2004. With this brief history, thecurrent generation of entrepreneurs is truly a genera-tion of founders, not yet diluted by owners of inher-ited productive assets. Second, China’s governmenthas not implemented policies aimed at actively pro-moting private start-up firms. To the contrary, privateentrepreneurs are disadvantaged relative to the state-owned enterprises, which benefit from governmentpolicies and loans from state-owned banks. Briefly,China’s current generation of private industrialistsrepresent a generation of start-up entrepreneurs whofit fairly well the original idea of entrepreneurship asobserved in the rise of modern capitalism in the West(Schumpeter 1947). Third, a specific concern in usinga high-income group like entrepreneurs in experi-ments is whether incentives can be considered salient(see Smith 1982). However, even if the entrepreneursbelong to the high-income group, their median annualpersonal income is only around USD 25,000 (accord-ing to the exchange rate 6.83 per USD 1 in July2009). This suggests substantial incentives can still begiven at a reasonable cost in China. The average sub-ject in our experiment earned CNY 289 (or USD 42)on the behavioral tasks that took only 25 minutes.The median of the entrepreneur’s self-reported dailypersonal income (if he works 300 days a year) isCNY 583. The average experimental earning wouldthus equate to half a day’s earnings (realized in

25 minutes).17 Notwithstanding these advantages ofconducting our study in China, we naturally do notclaim perfect external validity. Obviously, all experi-mental studies have to acknowledge cultural and his-torical idiosyncracies that may to some extent shapebehavioral responses.

3.4. Training and ExecutionAll research assistants selected for the implementationof the experiment were familiar with the local dialectand had worked as professional interviewers for sev-eral years. Starting on December 7, 2008, in Shanghai,all research assistants participated in a three-daytraining program designed and led by the authorsof this paper to standardize the implementation ofthe experiment. Each assistant was trained to do theinterviews and all tasks. They also received detailedwritten instructions and questionnaires for each task.At the end of the training, the authors accompa-nied teams of research assistants and supervisors tothe field to conduct a series of trial experiments tocheck and test the design and implementation by ourresearch assistants.

To minimize the time and inconvenience for theparticipating entrepreneurs, the behavioral tasks wereconducted at the firm site, usually in a confer-ence room or the entrepreneur’s private office withno other people attending the experiment. The ses-sion proceeded as follows: The entrepreneurs wereasked questions about their background (education,demographics) and the firm (start-up capital, firmrevenues, etc.). Subjects in the control group, alsovisited by two research assistants at their private res-idences, received the same set of questions, exceptfor those about the firm and business. Each subjectwas then presented with four different behavioraltasks, designed to measure risk aversion, ambiguityaversion, trust, and willingness to cooperate. After-ward, one task was randomly selected as the money-earning task. The earning was calculated, and theentrepreneur received the payment.

4. ResultsFor each task, our analysis first presents the meanswitching points in the different treatment groups(see Table 2).18 To mitigate order effects, we only com-pare groups that received the same treatments before

17 The average hourly experimental earning (which was CNY 763)corrected for purchasing power (which was 1.95 according to theBig Mac index in July 2009), would translate to an average experi-mental pay of approximately USD 200.18 We have omitted a few observations due to incorrectly filledout forms, such as “multiple switching points.” For no task is theshare of omitted observations above 5%. Research assistants werecarefully trained and given instructions, for example, to repeat theinformation when the subjects seemed to not fully understand thetask.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


the task under review.19 A consequence of this is thatthe number of observations will vary between thetasks, and we pool only those treatment groups ofentrepreneurs that have a matching control group.20

In a second step we conduct a regression analysis(using ordered probit estimations) to check the robust-ness of results to the inclusion of all variables listedin Table 3. We also include experimental variables tocontrol for experimenter and order effects. Even if theinterviewers were carefully instructed to act in a stan-dardized way, it is well known that the experimentermay affect the subject’s choices (Rosenthal 1966).21

Definitions of “entrepreneurs” vary. Up to thispoint we defined entrepreneurs relatively broadlyas a firm’s CEO (denoted as “Entrepreneur”). Thisdefinition is widely accepted because CEOs are incharge of entrepreneurial decisions in everyday busi-ness operations. Fifty-five percent of the firms inthe sample implemented more than three types ofinnovations—i.e., new products, new production pro-cesses, new management techniques, new qualitycontrol—consistent with Schumpeter’s definition ofinnovation as “new combinations” (Nee and Opper2012, pp. 195–198). Notwithstanding, a narrower def-inition of the entrepreneur as the owner or actualfounder of a firm may come closer to the classical con-cept of entrepreneurs as individuals who undertake aventure instead of relying on wage employment. Tocheck whether our regression results depend on thespecific definition of the entrepreneurial role, we raneach regression also for the subsample of 604 owner-entrepreneurs (denoted as “Owners” in the regres-sions) and the group of 546 founders, who actuallystarted their company (denoted as “Founders”).

4.1. RiskTable 2 summarizes the mean and standard devia-tions of the individual switching points. Both groupsexhibit a certain degree of risk aversion in theR1 task, because a risk-neutral subject should switchat decision 5 and a risk lover at an earlier decision.This result is consistent with studies of risk aversionin other groups (for college students, see Holt andLaury 2002; for the general population, see Harrisonet al. 2007a, von Gaudecker et al. 2011). Contrary towhat many would expect, the entrepreneurs are not

19 In a study of the multiple price list method, Andersen et al. (2006)detected that responding to a given task in a list might effect howthe subject responded in later tasks.20 Thus, if one treatment group of entrepreneurs received task Xbefore the task to be studied and another group received tasks YZ,we can pool these groups if and only we can find one treatmentgroup in the control that received X and another that received YZ.21 For each subject, we record the interviewer code. All regressionsinclude interviewer codes as dummy variables.

more risk taking than other people. In fact, their aver-age switching point (5.98) is slightly higher than theone for the control group (5.77), although the differ-ence is not significant in a Wilcoxon–Mann–Whitneytest (henceforth, WMW test). Because entrepreneurshave, on average, substantively higher incomes thanthe control group, this would reinforce risk taking inthe former group.22

When the decision is between a riskless outcomeand a risky one (R2 task), the entrepreneurs show asignificantly lower inclination to take risk than thecontrol group (WMW, p = 00021). However, this resultis not robust under inclusion of control variables(see Table 4). Hence, we do not find robust evidenceagainst Hypotheses 1A and 1B. In the regressions,the signs of the various definitions of entrepreneursare negative, suggesting the reverse impact (i.e., lessrisk aversion), but the coefficients are far from signif-icant. Few of the control variables predict the switch-ing point well. Age is negatively correlated to riskaversion, whereas education has the reverse sign andis significant in some equations. We also observe someindications of parental background effects as well asexperimenter effects. Thus, we do not find robust pre-dictors of standard risk taking, nor do entrepreneursseem to differ from ordinary people.

4.2. AmbiguityThe results on the A1 and A2 tasks do not suggestthat there are any notable differences on the level ofambiguity aversion between the groups (see Table 2).This is confirmed by regression analysis (see Table 4).Age is negatively correlated with ambiguity aversionand is marginally significant in the A2 task. Highereducation is significantly associated with lower ambi-guity aversion in the A1 task. This correlation isslightly weaker in the A2 tasks. As for risk aversion,there exist significant parental background and exper-imenter effects. The subject’s career experience (previ-ous job) also has a relatively robust association withambiguity aversion.

4.3. Willingness to CompeteFor the C1 and C2 tasks, willingness to compete isgenerally slightly higher among entrepreneurs (seeTable 2). The difference in switching point levels issignificant for the C2 task, although only at the 10%level (WMW, p = 00082). There is also a tendencythat the entrepreneurs’ distribution of switchingpoints has a slightly lower variance. A Kolmogorov–Smirnov test rejects equality of distributions at the

22 Based on previous studies there is no evidence that incomehas a substantial impact on the subjects’ risk-taking behavior(see Donkers et al. 2001, Harrison et al. 2007a, von Gaudeckeret al. 2011).

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Table 4 Nonstrategic Uncertainty: Ordered Probit Regressions of Switching Points

Risk Ambiguity

R1 task R2 task A1 task A2 taskcoeff. (SE) coeff. (SE) coeff. (SE) coeff. (SE)

Entrepreneur −00138 00045 00053 −001824002575 4002705 4002475 4003055

Owner −00137 00048 00091 −001364002635 4002825 4002605 4003105

Founder −00171 −00023 00105 −001764002685 4003045 4002785 4003185

Male −00128 −00057 −00041 −00009 −00058 00002 −00009 −00046 00005 −00093 −00069 −000414001385 4001525 4001575 4001555 4001715 4001855 4001475 4001615 4001735 4001465 4001615 4001675

Age −00008 −00008 −00006 00001 −00000 −00005 00011 00009 00000 −00012∗ −00015∗ −00015∗

4000075 4000075 4000085 4000085 4000085 4000095 4000075 4000085 4000095 4000075 4000085 4000085Education 00042∗ 00044∗ 00030 00032 00042∗ 00055∗∗ 00048∗∗ 00059∗∗ 00061∗∗ 00047∗ 00049∗ 00043

4000235 4000255 4000265 4000225 4000255 4000265 4000215 4000235 4000245 4000245 4000265 4000285Log income −00005 −00010 −00008 −00025 −00011 −00013 −00021 −00012 00012 00003 −00014 −00016

4000645 4000675 4000695 4000805 4000845 4000905 4000745 4000775 4000825 4000695 4000725 4000745Task 2 −00085 −00157 −00082 00089 −00046 −00045 00123 00246∗ 00225 00041 00035 −00001

4001395 4001485 4001535 4001265 4001365 4001445 4001265 4001355 4001425 4001445 4001545 4001595Task 3 00029 00063 00082 00012 −00004 −00030

4001265 4001345 4001345 4001435 4001555 4001625Task 4 −00197 −00222∗ −00203 00120 −00005 00077

4001245 4001325 4001385 4001415 4001585 4001695Job before Yes Yes Yes Yes Yes∗∗ Yes Yes∗∗∗ Yes∗∗∗ Yes∗∗ Yes∗ Yes∗∗ Yes

current jobFather’s Yes Yes Yes Yes Yes Yes Yes∗∗∗ Yes∗∗ Yes∗∗ Yes∗ Yes Yes

educationFather’s job Yes∗∗ Yes∗∗∗ Yes∗∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗∗ Yes∗∗∗ Yes∗∗∗ Yes∗∗∗ Yes∗∗∗

at retirementOrganizational unit Yes Yes Yes Yes∗ Yes Yes Yes∗ Yes Yes∗ Yes∗∗ Yes Yes∗∗

of father’s last jobInterviewer a Yes Yes∗ Yes∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗ Yes∗ Yes∗

Prob. > �2 00009 00010 00012 00017 00048 00019 00000 00000 00000 00000 00000 00002Pseudo-R2 00054 00060 00035 00056 00058 00038 00064 00069 00071 00068 00072 00072Observations 428 381 362 426 377 344 462 414 377 390 345 325

aDummy variables indicate the main interviewer conducting the experiment. The level of significance reflects highest level of significance found for at leastone dummy variable. Coefficient estimates for dummy variables are available from the authors.

∗Statistically significant at the 10% level; ∗∗statistically significant at the 5% level; ∗∗∗statistically significant at the 1% level.

10% level of significance (p = 00078) for task C1. Oneexplanation might be that entrepreneurs have a moredistinct evaluation of this type of uncertainty. We alsocontrolled whether the higher willingness to competeamong entrepreneurs is associated with the belief thatthey will have a higher score on the performance taskthan others. This is not the case, because the averagerelative performance beliefs of the entrepreneurs andthe control subjects are almost identical.23 Higher will-ingness to compete among entrepreneurs thereforeseems to be driven by a preference for competitionper se.

Our results are confirmed under inclusion of con-trol variables (see Table 5). In the case of multilat-eral competition (task C2), the entrepreneur effect is

23 In the C1 task, both entrepreneurs and control subjects believedthat 44.7% would have a higher score than themselves. In theC2 task, entrepreneurs believed that 48.1% would have a higherscore, and the corresponding figure among the control groupwas 48.9%.

robustly significant at the 5% level. In all three sam-ples, entrepreneurs are less averse to multilateral com-petition than ordinary people. Two specifications alsoshow a significant gender effect. Male participants aremore willing to participate in multilateral competitionthan females, which is consistent with results reportedby Gneezy et al. (2003).24

4.4. TrustEntrepreneurs are more willing to expose themselvesto social risks than the control group (see Table 2).The switching points of the entrepreneurs are signif-icantly higher than for the control group (WMW, p =

00012), although the effect is moderate.25 The average

24 Gneezy et al. (2003) used student subjects. The study’s resultssuggest that gender differences may also hold in multilateral com-petition among more experienced subjects, like CEOs who own orfounded their firms.25 Here the switching point is increasing in the subject’s willingnessto expose herself to trust.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Table 5 Strategic Uncertainty: Ordered Probit Regressions of Switching points

Competitiveness Trust

C1 task C2 task T taskcoeff. (SE) coeff. (SE) coeff. (SE)

Entrepreneur 00014 −00877∗∗∗ 00564∗∗∗

4002525 4003005 4001785Owner −00105 −00771∗∗ 00551∗∗∗

4002615 4003145 4001775Founder −00006 −00750∗∗ 00565∗∗∗

4002825 4003215 4001915Male 00028 00098 00085 −00230 −00309∗ −00350∗ 00056 00007 00101

4001425 4001535 4001645 4001505 4001695 4001795 4000995 4001085 4001145Age −00003 −00007 −00009 −00007 −00005 −00010 00002 00002 −00002

4000075 4000075 4000085 4000075 4000085 4000085 4000055 4000055 4000065Education 00025 00024 00045∗ −00057∗∗ −00058∗∗ −00065∗∗ 00015 00023 00012

4000215 4000235 4000245 4000245 4000275 4000285 4000155 4000175 4000175Log income −00012 00012 −00016 00093 00097 00095 −00070 −00065 −00042

4000655 4000675 4000705 4000795 4000835 4000885 4000485 4000505 4000525Task 3 00157∗∗ 00195∗∗ 00209∗∗

4000775 4000845 4000885Task 4 −00255∗∗ −00221∗ −00237∗ −00335∗∗∗ −00176 −00093

4001095 4001155 4001235 4001265 4001315 4001465Job before Yes∗ Yes∗ Yes Yes∗ Yes∗ Yes Yes Yes Yes

current jobFather’s Yes Yes Yes Yes∗∗ Yes Yes Yes∗ Yes∗ Yes∗

educationFather’s job Yes∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes Yes Yes∗∗ Yes∗ Yes∗

at retirementOrganizational unit Yes∗∗ Yes∗∗ Yes Yes∗∗ Yes Yes Yes∗ Yes Yes

of father’s last jobInterviewer a Yes∗∗ Yes∗∗ Yes∗∗∗ Yes∗∗ Yes∗∗ Yes∗∗ Yes∗ Yes∗ Yes∗

Prob. > �2 00000 00000 00000 00000 00001 00000 00000 00000 00000Pseudo-R2 00066 00068 00074 00082 00047 00052 00040 00042 00042Observations 465 423 388 388 337 317 857 761 703

aDummy variables indicate the main interviewer conducting the experiment. The level of significance reflects highest level of significance found for at leastone dummy variable. Coefficient estimates for dummy variables are available from the authors.

∗Statistically significant at the 10% level; ∗∗statistically significant at the 5% level; ∗∗∗statistically significant at the 1% level.

difference between the switching points is 19% of thestandard deviation. Because we compare two riskyoptions, it is reasonable to interpret this difference asa higher confidence in another person’s willingness toact favorably toward the subject.26 Entrepreneurs rou-tinely rely on ongoing personal ties in upstream anddownstream transactions with their suppliers and dis-tributors. Moreover, in industrial districts the capac-ity for on-the-fly cooperation between producers isan important source of competitive advantage (Uzzi1996). Because business networks provide a readyconduit of fine-grained information on reputation andtrustworthiness, trust in relational contracting may bea self-reinforcing feature of entrepreneurial activity.The regression analysis confirms the robustness of theresult (see Table 5). The dummy for entrepreneur is

26 Because we did not detect any significant difference in risk aver-sion when two risky alternatives were present, it is unlikely thatthis result is driven by risk aversion.

significant with the expected sign at the 1% level inall three equations. Hence, the nonparametric resultand the regression analysis suggest that entrepreneurshave a higher tolerance of social risks than otherpeople.27

4.5. Robustness ConcernsWe have scrutinized our findings along three dif-ferent dimensions, exploring the presence of sam-ple selection problems, overspecification, and omittedvariable concerns.

Although our baseline models have controlled forindividual income and education, skeptics might stillclaim that both groups are too different with respectto their economic and educational status to be com-pared. One statistical method to deal with this issue

27 Fehr and List (2004) made congruent observations when theycompared CEOs in the Costa Rican coffee industry with undergrad-uate students. The CEOs were more trusting than the students.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


is the application of propensity score analysis (PSA).However, the extreme size requirements of such bal-ancing strategies rarely allow application of PSA inexperimental research.28 However, running the regres-sions on a subsample of the entrepreneurs and thecontrol group that overlap in some critical dimensionsmay alleviate some of these concerns. With respectto income, we therefore exclude control group mem-bers who do not reach the income of the poorestentrepreneur represented in the sample. With respectto education, we use the spread of education observedin the control group (5 to 19 years) to exclude all thoseentrepreneurs who either have less or more years offormal education. Regression estimates are consistentwith the findings of our baseline models, confirmingsignificant entrepreneur effects in the case of strategicuncertainty.29

A second concern may be the risk of overspecify-ing our model. Our estimation model includes a largenumber of variables (26 dummy variables controllingfor parental background). To confirm that our resultsare not driven by the numerous control variables, wedid a robustness check by running a reduced modelexcluding the dummy variables for career experiencesand parental background.30

Others may still fear an omitted variable bias. Forexample, members of the Chinese Communist Partytypically have better access to government and partynetworks, and may enjoy a higher social and politi-cal status (Nee and Opper 2010). This could theoret-ically affect behavioral choices, particularly when itcomes to decisions under uncertainty. Statistical meancomparison tests, however, do not suggest signifi-cant behavioral differences between groups. Inclusionof party membership as an additional control vari-able also does not affect our baseline estimations.31

Finally, we have explored the possibility that the esti-mated entrepreneur effect could simply reflect on-the-job learning of entrepreneurs stemming from morefrequent exposure to situations involving trust andcompetition. A closer review of the mean responsesof entrepreneurs, however, does not hint at any obvi-ous learning experience driving their choices. Meanresponses of experienced entrepreneurs who havebeen in business for more than 15 years do not

28 To be successful, these matching methods typically require con-trol groups that are multiple times larger than the treatment group,which in our case is the entrepreneurs (Dehejia and Wahba 2002).Control groups of that size, however, are almost impossible to real-ize given the obvious financial constraints of conducting such anexperiment.29 Tables S2 and S3 in the online supplement present our regressionestimates.30 Regression estimates are given in Tables S4 and S5 in the onlinesupplement.31 Results are available upon request.

differ significantly from mean responses of thoseentrepreneurs in charge of newly founded firms withless than five years of experience. Years in businessand individual behavioral responses generally seemonly weakly correlated. Inclusion of a continuousvariable capturing “years in business” into our base-line estimations confirms this. Coefficient estimatesare never even close to being statistically significant.32

This suggests that selection mechanisms explainingthe entry into entrepreneurship (rather than learningeffects) are more likely to explain the behavioral dif-ferences between entrepreneurs and ordinary people.

We do not claim that our results are robust to allconceivable modifications of our estimation strategy.As with any empirical design, we cannot exclude thatthere are unobservable covariates to both uncertaintybehavior and becoming an entrepreneur, which mayexplain our results. What we can claim, however, isthat our results hold for a number of alternate specifi-cations of observables, which cover central predictorsof both behavioral choices as well as individual careerchoices.

5. Concluding RemarksThis paper reports results from the first large-scaleexperimental study on a randomly sampled popula-tion to investigate whether entrepreneurs are moreprone than nonentrepreneurs to expose themselvesto various forms of uncertainty. Our multidimen-sional design of four experimental tasks builds onthe observation that uncertainty has both strategicand nonstrategic dimensions, which require separatetreatments to better understand the nature of dis-tinct entrepreneurial traits. Results from our strati-fied random sample of 700 established entrepreneursand 200 control group members located in the Yangzidelta region in China suggest that entrepreneursdo not generally differ from other people when itcomes to behavior under uncertainty. When exposedto nonstrategic forms of uncertainty such as situationsinvolving standard risk or ambiguity, entrepreneursact similarly to ordinary people. This observation con-tradicts the theory by Kihlstrom and Laffont (1979)that selection into entrepreneurship is based on riskaversion.33 It does also not support risk tolerance asa key causal mechanism explaining entrepreneurs’undiversified portfolios (Moskowitz and Vissing-Jorgensen 2002).

At the same time, entrepreneurs seem to be morewilling to bear uncertainties involving a strategic

32 Results are available upon request.33 It should be noted that our focus on established entrepreneursmay reflect that the market punishes and weeds out less prudentindividuals who are willing to take unnecessary risks (see Caliendoet al. 2010).

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


element. First of all, entrepreneurs are significantlymore willing to enter situations involving multilat-eral competition than members of the control group.This result is not unexpected. An essential featureof entrepreneurship is to compete with other firmsin various respects and also to expose others (e.g.,upstream firms and employees) to competitive pres-sure. Established entrepreneurs cannot shy away fromcompetition. It is therefore likely that this group, assuggested by Marshall (1920) and Schumpeter (1983),has a preference for such situations. As a conse-quence, our results correspond with Hamilton’s (2000)observation that entrepreneurs are willing to enterand persist in business even if the pecuniary com-pensation is low. Second, our findings suggest thatentrepreneurs are more willing to accept uncertaintiesrelated to trusting an anonymous other. Due to ourdesign and results on nonstrategic uncertainty, we canrule out lower risk aversion among entrepreneurs asa causal explanation. Still, several alternative expla-nations remain, and further research is needed todisentangle the deeper mechanisms involved. Stud-ies of comparison groups that match entrepreneurs’properties more closely are likely to give importantinsights. However, at this stage, a preliminary con-clusion is that entrepreneurship requires a certain

Appendix. Entrepreneurship and Uncertainty

FG moreprone

to acceptuncertainty

Focus group Comparison group Region Method Sampling than CG

RiskHull et al.

(1980)Owner-managers

(n = 57)Business school

alumni (n = 250)United States S NR Yes

Brockhaus(1980)

Start-upentrepreneurs(n = 31)

Business managers(n = 62)

United States S R No

Caird(1991)

Owner managers(n = 73)

Professional groups(n = 189)

United States S NR Yes

Begley(1995)

Firm founders(n = 114)



Koh (1996) Entrepreneuriallyinclined MBAstudents (n = 22)

Not entrepreneuriallyinclined MBAstudents (n = 32)

Hong Kong S NR Yes

Sarasvathyet al.(1998)

Entrepreneursparticipating ineducation program(n = 4)

Bankers participatingin same educationprogram (n = 4)


Stewartet al.(1999)

Entrepreneur(n = 428)

Corporate manager(n = 239)


amount of trust, and that more trusting individu-als are selected into the group of entrepreneurs. Thisview is consistent with the important role of pri-vate order and relational contracting in everydaybusiness situations (Williamson 1987, Macaulay 1963).Clearly entrepreneurship is not a purely individualis-tic endeavor and requires the willingness and abilityto cooperate even though complete contracts are typ-ically missing. Trust may therefore provide the gluethat makes business networks actually work.

AcknowledgmentsA preliminary version of this paper was presented at theResearch Institute of Industrial Economics in Stockholm,the Department of Economics at Umeå School of Busi-ness and Economics, the Department of Economics at LundUniversity, the Mercator School of Management and EssenLaboratory for Experimental Economics at University ofDuisburg-Essen, the Department of Sociology at CornellUniversity, the Economic Science Association (ESA) WorldCongress in Copenhagen, the Asia-Pacific ESA meeting inKuala Lumpur, and the Entrepreneurial Finance Conferenceat Tilburg University. We are grateful for valuable com-ments from the audience on these presentations. This paperalso benefitted from insightful comments from two anony-mous referees and editors of this journal. Financial supportof the Swedish Research Council and the John TempletonFoundation is gratefully acknowledged.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Appendix. (Continued)

FG moreprone

to acceptuncertainty

Focus group Comparison group Region Method Sampling than CG

RiskVan Praag

and Cramer(2001)

Currently or beforeself-employed(n = 258)

Never beforeself-employed(n = 1,505)

Netherlands S R Yes

Uusitalo(2001)

Self-employed(n = 2,418)

Not self-employed(n = 23,551)

Finland S R Yes

Cramer et al.(2002)

Self-employed atsome point in life(n = 330)

Neverself-employed(n = 1,567)

Netherlands S R Yes

Elston et al.(2006)

Entrepreneurs(n = 42)

Nonentrepreneurs(n = 38)

United States E NR No

Djankov et al.(2006)



China S Stratified RS Yes

Djankov et al.(2007)



Russia S Stratified RS No

Caliendoet al. (2010)

Individuals whotransfer intoself-employment(n = 147)

Individuals whoremain employed(n = 8,561)

Germany S RS Yes

Macko andTyszka(2009)


Students(n = 86)

Poland E NR No

Sandri et al.(2010)


Students/nonstudents(n = 84)

Germany E NR No

List andMason(2011)


Students(n = 101)

Costa Rica E NR No

Burmeister-Lamp et al.(2012)


Students(n = 29)

Germany E NR No

AmbiguitySchere (1982) Firm founders

(n = 52)Managers (n = 65) United States S NR Yes

Koh (1996) Entrepreneuriallyinclined students(n = 22)

Notentrepreneuriallyinclined students(n = 32)

Hong Kong S NR Yes

Macko andTyszka(2009)


Students(n = 86)

Poland E NR Yes

Willingness to competeBegley and

Boyd (1987)Business founders

(n = 268)Nonfounding chief

executives(n = 203)


Begley (1995) Firm founders(n = 114)


United States S NR No

Elston et al.(2006)



United States E NR No

TrustFehr and List

(2004)CEOs (n = 76) Students (n = 126) Costa Rica E NR Yes

Notes. FG, focus group; CG, comparison group; S, survey; E, experiment; R, random; NR, nonrandom.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


ReferencesAndersen S, Harrison GW, Lau MI, Rustström EE (2006) Elici-

tation using multiple price list formats. Experiment. Econom.9(4):365–366.

Andersen S, Harrison GW, Lau MI, Ruström EE (2008) Eliciting riskand time preferences. Econometrica 76(3):583–618.

Baruch Y (1999) Response rate in academic studies—A comparativeanalysis. Human Relations 52(4):421–438.

Begley TM (1995) Using founder status, age of firm, and companygrowth rate as the basis for distinguishing entrepreneurs frommanagers of smaller businesses. J. Bus. Venturing 10(3):249–263.

Begley TM, Boyd DP (1987) A comparison of entrepreneurs andmanagers of small business firms. J. Management 13(1):99–108.

Bernardo A, Welch I (2001) On the evolution of overconfidence andentrepreneurs. Econom. Management Strategy 10(3):301–331.

Bernstein F, Federgruen A (2005) Decentralized supply chains withcompeting retailers under demand uncertainty. ManagementSci. 51(1):18–29.

Binswanger HP (1980) Attitudes toward risk: Experimental meas-urement in rural India. Amer. J. Agricultural Econom. 62(3):395–407.

Borghans L, Heckman J, Golsteyn B, Meijers H (2009) Gender differ-ences in risk aversion and ambiguity aversion. J. Eur. Econom.Assoc. 7(2–3):649–658.

Brockhaus RH (1980) Risk taking propensity of entrepreneurs. Acad.Management J. 23(3):509–520.

Burmeister-Lamp K, Levesque M, Schade C (2012) Are entrepre-neurs influenced by risk attitude, regulatory focus or both? Anexperiment on entrepreneurs’ time allocation. J. Bus. Venturing27(4):456–476.

Caird S (1991) The enterprise tendency of occupational groups.Internat. Small Bus. J. 9(4):75–91.

Caliendo M, Fossen F, Kritikos A (2010) The impact of risk atti-tudes on entrepreneurial survival. J. Econom. Behav. Organ.76(1):45–63.

Camerer C, Hogarth RM (1999) The effects of financial incentives ineconomic experiments: A review and capital-labor-productionframework. J. Risk Uncertainty 19(1–3):7–42.

Camerer C, Lovallo D (1999) Overconfidence and excess entry: Anexperimental approach. Amer. Econom. Rev. 89(1):306–318.

Charness G, Villeval MC (2009) Cooperation and competition inintergenerational experiments in the field and the laboratory.Amer. Econom. Rev. 99(3):956–978.

Coleman JS (1990) Foundations of Social Theory (The Belknap Pressof Harvard University, Cambridge, MA).

Cramer JS, Hartog J, Jonker N, van Praag M (2002) Low risk aver-sion encourages the choice for entrepreneurship: An empiricaltest of a truism. J. Econom. Behav. Organ. 48(1):29–36.

Dehejia RH, Wahba S (2002) Propensity score-matching meth-ods for nonexperimental causal studies. Rev. Econom. Statist.84(1):151–161.

Djankov S, Qian Y, Roland G, Zhuravskaya E (2006) Whoare China’s entrepreneurs? Amer. Econom. Rev. Papers Proc.96(2):348–352.

Djankov S, Qian Y, Roland G, Zhuravskaya E (2007) What makes asuccessful entrepreneur? Evidence from Brazil. Working Paper0104, Center for Economic and Financial Research, Moscow.

Dohmen T, Falk A, Huffman D, Sunde U, Schupp J, WagnerGG (2011) Individual risk attitudes: Measurement, determi-nants, and behavioral consequences. J. Eur. Econom. Assoc.9(3):522–550.

Donkers B, Melenberg B, van Soest A (2001) Estimating risk atti-tudes using lotteries: A large sample approach. J. Risk Uncer-tainty 22(2):165–195.

Ellsberg D (1961) Risk, ambiguity, and the savage axioms. Quart. J.Econom. 75(4):643–669.

Elston JA, Harrison GW, Rutström EE (2006) Experimentaleconomics, entrepreneurs and the entry decision. WorkingPaper 6, Economics Department, University of Central Florida,Orlando.

Evans DS, Leighton LS (1989) Some empirical aspects ofentrepreneurship. Amer. Econom. Rev. 79(2):519–535.

Fehr E, List J (2004) The hidden costs and returns of incentives:Trust and trustworthiness among CEOs. J. Eur. Econom. Assoc.2(5):743–771.

Fehr E, Schmidt KM (1999) A theory of fairness, competition, andcooperation. Quart. J. Econom. 114(3):817–868.

Fox CR, Tversky A (1995) Ambiguity aversion and comparativeignorance. Quart. J. Econom. 110(3):585–603.

Glaeser E, Laibson D, Scheinkman J, Soutter Ch (2000) Measuringtrust. Quart. J. Econom. 115(3):811–846.

Gneezy U, Niederle M, Rustichini A (2003) Performance in com-petitive environments: Gender differences. Quart. J. Econom.118(3):1049–1074.

Hamilton B (2000) Does entrepreneurship pay? An empirical anal-ysis of the returns to self-employment. J. Political Econom.108(3):604–631.

Harrison GW, List J (2004) Field experiments. J. Econom. Literature42(4):1009–1055.

Harrison GW, Lau MI, Rutstrom EE (2007a) Estimating risk atti-tudes in Denmark: A field experiment. Scandinavian J. Econom.109(2):341–368.

Harrison GW, List J, Towe Ch (2007b) Naturally occurring prefer-ences and exogenous laboratory experiments: A case study ofrisk aversion. Econometrica 75(2):433–458.

Hayward MLA, Shepherd DA, Griffin D (2006) A hubris theory ofentrepreneurship. Management Sci. 52(2):160–172.

Holt ChA, Laury SK (2002) Risk aversion and incentive effects.Amer. Econom. Rev. 92(5):1644–1655.

Hull D, Bosley J, Udell G (1980) Renewing the hunt for heffalump:Identifying potential entrepreneurs by personality characteris-tics. J. Small Bus. 18(1):11–18.

IMD (2005) IMD World Competitiveness Yearbook 2005 (IMD,Lausanne, Switzerland).

Kahneman D, Tversky A (1979) Prospect theory: An analysis ofdecision under risk. Econometrica 47(2):263–291.

Kihlstrom RE, Laffont JJ (1979) A general equilibrium entrepre-neurial theory of firm formation based on risk aversion. J. Polit-ical Econom. 87(4):710–748.

Kirzner I (1973) Competition and Entrepreneurship (Chicago Univer-sity Press, Chicago).

Knight F (2006) Risk, Uncertainty and Profit (Dover Publications,Mineola, NY).

Koellinger Ph, Minniti M, Schade Ch (2007) “I think I can”: Over-confidence and entrepreneurial behavior. J. Econom. Psych.28(4):502–527.

Koh HC (1996) Testing hypotheses of entrepreneurial characteris-tics: A study of Hong Kong MBA students. J. Managerial Psych.11(3):12–25.

Lazear EP (2004) Balanced skills and entrepreneurship. Amer.Econom. Rev.: Papers Proc. 94(2):208–211.

Levitt SD, List JA (2007) What do laboratory experiments measur-ing social preferences reveal about the real world? J. Econom.Perspect. 21(2):153–174.

Lindquist MJ, Sol J, Van Praag M (2012) Why do entrepreneurialparents have entrepreneurial children? IZA DiscussionPaper 6740, Institute for the Study of Labor, Bonn, Germany.

List JA, Mason CF (2011) Are CEOs expected utility maximizers?J. Econometrics 162(1):114–123.

Macaulay S (1963) Non-contractual relations in business: A prelim-inary study. Amer. Sociol. Rev. 28(1):55–67.

Macko A, Tyszka T (2009) Entrepreneurship and risk taking. Appl.Psych. 58(3):469–487.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.


Marshall A (1920) Principles of Economics, 8th ed. (Macmillan & Co.,London).

Moskowitz T, Vissing-Jorgensen A (2002) The returns toentrepreneurial investment: A private equity premium puzzle?Amer. Econom. Rev. 92(4):745–778.

Nee V, Opper S (2010) Political capital in a market economy. Soc.Forces 88(5):2105–2133.

Nee V, Opper S (2012) Capitalism from Below: Markets and InstitutionalChange in China (Harvard University Press, Cambridge, MA).

Pich MT, Loch ChH, de Meyer A (2002) On uncertainty, ambigu-ity, and complexity in project management. Management Sci.48(8):1008–1023.

Putnam RD (2001) Bowling Alone. The Collapse and Revival of theAmerican Community (Simon & Schuster, New York).

Rosenthal R (1966) Experimenter Effects in Behavioral Research(Appleton-Century-Crofts, East Norwalk, CT).

Sandri S, Schade Ch, Musshoff O, Odening M (2010) Hold-ing on for too long? An experimental study on inertia inentrepreneurs’ and non-entrepreneurs’ disinvestment choices.J. Econom. Behav. Organ. 76(1):30–44.

Sarasvathy DK, Simon H, Lave L (1998) Perceiving and managingbusiness risks: Differences between entrepreneurs and bankers.J. Econom. Behav. Organ. 33(2):207–222.

Say JB (1855) A Treatise on Political Economy (Lippincott, Grambo &Co, Philadelphia).

Schere J (1982) Tolerance of ambiguity as a discriminating variablebetween entrepreneurs and managers. Acad. Management BestPaper Proc. 42:404–408.

Schumpeter JA (1947) Capitalism, Socialism and Democracy, 2nd ed.(Allen and Unwin, London).

Schumpeter JA (1983) The Theory of Economic Development: AnInquiry Into Profits, Capital, Credit, Interest, and the Business Cycle(Transaction Publishers, New Brunswick, NJ).

Shane SA (2003) A General Theory of Entrepreneurship: The Individual–Opportunity Nexus (Edward Elgar, Northampton, MA).

Smith VL (1982) Microeconomic systems as an experimental sci-ence. Amer. Econom. Rev. 72(5):923–955.

Stewart WH, Watson WE, Carland JC, Carland JW (1999) A Pro-clivity for entrepreneurship: A comparison of entrepreneurs,small business owners, and cooperate managers. J. Bus. Ven-turing 14(2):189–214.

Uusitalo R (2001) Homo entreprenaurus? Appl. Econom. 33(13):1631–1638.

Uzzi B (1996) The sources and consequences of embeddednessfor the economic performance of organizations: The networkeffect. Amer. Sociol. Rev. 61(4):674–698.

Van Praag M, Cramer JS (2001) The roots of entrepreneurshipand labour demand: Individual ability and low risk aversion.Economica 68(269):45–69.

von Gaudecker HM, van Soest A, Wengström E (2011) Hetero-geneity in risky choice behavior in a broad population. Amer.Econom. Rev. 101(2):664–694.

Williamson OE (1987) The Economic Institutions of Capitalism, Firms,Markets, Relational Contracting (Free Press, New York).

Wu B, Knott AM (2006) Entrepreneurial risk and market entry. Man-agement Sci. 52(9):1315–1330.

CORRECTION

In this article, “Entrepreneurs Under Uncertainty: An Economic Experiment in China” by Hakan J. Holm,Sonja Opper, and Victor Nee (first published in Articles in Advance, January 8, 2013, Management Science,DOI:10.1287/mnsc.1120.1670), Option B of Task T in Table 1 has been corrected to read: Decision 1, 100% ofII; Decision 2, 10% of I, 90% of II; Decision 9, 80% of I, 20% of II; Decision 10, 90% of I, 10% of II.

Dow

nloa

ded

from

info

rms.

org

by [

128.

84.1

24.8

1] o

n 25

Sep

tem

ber

2017

, at 2

2:23

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Documents

Entrepreneurs Under Uncertainty: An Economic Experiment in ...people.soc.cornell.edu/nee/pubs/EntrepreneursUnderUncertainty.pdf · Published online in Articles in Advance January