MANAGEMENT SCIENCEArticles in Advance, pp. 1–19ISSN 0025-1909 (print) � ISSN 1526-5501 (online) https://doi.org/10.1287/mnsc.2016.2566

© 2016 INFORMS

The Impact of Idea Generation andPotential Appropriation on Entrepreneurship:

An Experimental StudySoheil Hooshangi

Carnegie Mellon University, Pittsburgh, Pennsylvania 15213; and Católica Lisbon School of Business and Economics,1649-023 Lisbon, Portugal, hooshangi@cmu.edu

George LoewensteinCarnegie Mellon University, Pittsburgh, Pennsylvania 15213, gl20@andrew.cmu.edu

Using a novel experimental paradigm, we explore how the experience of generating an idea and the possi-bility that another investor might adopt a rejected investment opportunity, bias the investment decisions

of innovator and imitator entrepreneurs. We find that individuals who generate a business idea form biasedevaluations of the economic potential of ideas, be it their own idea or somebody else’s idea. On the one hand,they are overconfident about the value of, and overly likely to invest in, their own idea. On the other hand,when investing in another person’s idea, even if it is not competing with their own idea, they are underconfi-dent about the value of, and insufficiently likely to invest in, the idea. Surprisingly, we find that entrepreneurialexperience exacerbates this pattern of over- and underconfidence. In addition, we find that the threat thatanother investor can appropriate a declined investment opportunity increases willingness to invest. We proposea theoretical account to explain the observed pattern of over- and underconfidence in imitative and innovativeentrepreneurship. Our findings challenge the traditional account that lowering the cost of imitation has a dis-incentive effect on the investment decisions of pioneer entrepreneurs and provide evidence that a more lenientappropriability regime may, unexpectedly, have positive effects on entrepreneurship. Our findings also identifynew psychological mechanisms that can play a role in important phenomena such as the emergence of spin-offsand rush to market entry.

Data, as supplemental material, are available at https://doi.org/10.1287/mnsc.2016.2566.

Keywords : entrepreneurship; investment; imitation; overconfidence; takeover aversionHistory : Received March 10, 2013; accepted May 11, 2016, by Uri Gneezy, behavioral economics. Published

online in Articles in Advance October 27, 2016.

IntroductionEconomic theories of entrepreneurship assume thatentrepreneurs’ investment decisions are guided bymaximization of expected returns. However, empir-ical evidence in the literature on entrepreneurship(e.g., Cooper et al. 1988, Busenitz and Barney 1997,Camerer and Lovallo 1999, Keh et al. 2002, Åstebro2003) and business strategy (Hodgkinson et al. 1999)suggests that individuals’ valuations of opportunitiesmay be subject to systematic biases and influencesthat fall outside of the range of the standard eco-nomic perspective. Besides the fact that entrepreneursmay be subject to the same heuristics and biases asdocumented among other decision makers, as high-lighted in the behavioral literature on entrepreneur-ship, we provide evidence that entrepreneurs havedeeply personal connections to the ideas they caninvest in. A classic paper by Abelson (1986) drewattention to, as the title expressed it, the observationthat “beliefs are like possessions.” In this paper, we

discuss, and demonstrate empirically, that the sameis true for the ideas generated and invested in byentrepreneurs. Entrepreneurs who generate an ideaget disproportionately attached to their own idea andare dismissive of ideas proposed by others, and whenentrepreneurs have an opportunity to invest in anidea, they are more likely to do so if failing to investmeans that the opportunity may go to somebody else.

We explore the impact of two psychological factorsthat have not received attention in prior psychologi-cally oriented research on entrepreneurial investmentdecisions. The first has to do with the question ofwhether an individual has generated her own idea.Most entrepreneurs follow neither a purely innovativenor a purely imitative investment strategy. Innovators,by definition, pursue their own ideas, but most inno-vators also act as imitators who monitor the emerg-ing ideas of others to potentially exploit promisingones (Ziegler 1985, Gans and Stern 2003). Conven-tional economic theories of entrepreneurship do notdraw a strong distinction between these two types of

1

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship2 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

investment decisions, and they assume that both arevalued purely on the basis of their expected returns.We propose, in contrast, that the experience of gener-ating an idea affects potential entrepreneurs’ evalua-tions not only of that idea but also of others’ ideas.We find that people who generate ideas tend to beexcessively enthusiastic about, overconfident in, andwilling to invest in those ideas. More surprisingly,we predict and find that developing an entrepreneur-ial idea makes people overly critical of, undercon-fident in, and hence unwilling to invest in others’ideas, even if those other ideas do not compete againstthe individual’s own idea. Potential entrepreneurswho have not created an idea, in contrast to thosewho have, show appropriate levels of confidence, andinvest appropriately, in ideas proposed by other peo-ple. We also find that entrepreneurial experience doesnot reduce but exacerbates these patterns of over-and underconfidence. Prior experimental research onthe valuation of entrepreneurial opportunities has notaddressed the distinction between innovative and imi-tative investment decisions—i.e., investing in ideasgenerated by the investor or by someone else—but haseither focused only on opportunities that were basedon the decision maker’s idea (e.g., Åstebro 2003) or notconsidered the source of idea generation as a relevantfactor (e.g., Keh et al. 2002).

The second factor our study addresses is whetheran idea not invested in could potentially be investedin by another individual. Following up on prior re-search on the role of “social takeover” in betting deci-sions (Hoelzl and Loewenstein 2005), we hypothesizeand find that potential entrepreneurs are more likelyto invest if there is a chance that someone else canappropriate their opportunity if they fail to invest,even though the other person’s decision has no mate-rial consequences for them.

We present results from a novel incentive-compat-ible experiment that investigates the joint impact ofidea generation and the possibility that someone elsecan appropriate one’s idea on the valuation of, andtendency to invest in, entrepreneurial opportunities.We also present results from a follow-up experimentthat provides further supportive evidence for ourpredictions and proposed theoretical account of theimpact of idea generation. Overall, we find that theexperience of creating an idea results in overconfi-dence in the promise of one’s own idea but undercon-fidence in the promise of somebody else’s idea, andthat both overconfidence and underconfidence trans-late into differential willingness to invest. We alsofind that investors may be driven in part by takeoveranxiety—a fear that if they fail to exploit an opportu-nity, someone else will do so. By providing evidencethat a decrease in the stringency of enforcement ofintellectual property rights may, surprisingly, havepositive effects on entrepreneurship, our findings

challenge the long-held assumption that lowering thecost of imitation has a disincentive effect on theinvestment decisions of pioneer entrepreneurs. Fur-thermore, our findings indicate that entrepreneurialexperience is not a remedy for, and may even exacer-bate, biased investment valuations and decisions. Weelaborate more on our findings and their implicationsin the conclusion of the paper.

BackgroundOverconfidence, Underconfidence, andIdea GenerationCamerer and Lovallo (1999) study the role of overcon-fidence in entrepreneurship using a stylized experi-ment in which individuals have the choice of whetheror not to enter into a competition. Finding that indi-viduals were likely to overenter and lose moneyin skill-based competitions (but not in chance-basedtasks), and that this pattern persisted when individ-uals could self-select into a skill-based competition,they suggest that overestimation of one’s own skillrelative to others and neglect of the fact that oth-ers will self-select into competitions in which theyare themselves skilled, help to explain excess entryin skill-based competitions. Malmendier and Tate(2005), using field data, find that managers investexcessively when using internal funds but underuti-lize external financing. They attribute both resultsto overoptimism. Managers overuse internal fundsbecause they are overly optimistic about their invest-ments, but they underutilize external funds becausethey believe that equity markets undervalue theirfirms and raise their cost of capital above what itshould be.

Moore and Cain (2007) argue, however, that over-confidence is not necessarily pervasive. Running astudy modeled closely on Camerer and Lovallo’s(1999) experimental design, they find that overcon-fidence and excess entry are common when peoplecompete on easy skill-based tasks but that the oppo-site pattern emerges when tasks are perceived asdifficult.

Prior research has not addressed whether businessidea generation is usually perceived as an easy taskor a difficult task. Interviews and writings publishedin magazines and websites whose main focus is busi-ness or entrepreneurship provide evidence for bothperspectives. While some entrepreneurs express theview that “ideas are easy, execution is difficult” (e.g.,Masnick 2008), others believe that the difficult partof entrepreneurship is coming up with a promisingbusiness idea (e.g., Andrews 2011).

Moore and Healy (2008), building on the the-oretical account initially proposed by Moore andCain (2007) and Moore and Small (2007), proposean information-theoretic account of the pattern of

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 3

over- and underconfidence for easy and difficult tasks.According to their account people are uncertain abouttheir own performance but even more uncertain aboutothers’ performance. To the extent that informationabout performance is lacking, a good initial assump-tion about performance level (a good “prior”) wouldbe the group average. People usually possess bet-ter information about themselves than about others;thus, they update from their prior beliefs more forthemselves than for others. Because of uncertainty,therefore, judgments about both one’s own and others’performance tend to be regressive. However, becauseof inferior information about others, judgments aboutothers tend to be more regressive than judgmentsabout oneself. This leads people to underestimate theirabsolute performance and overestimate their relativeperformance on easy tasks, but to display the oppositepattern for difficult tasks.

Kruger (1999) proposes an alternative account ofover- and underconfidence based on egocentrism.According to this account, when forming judgmentsof relative confidence, people focus first, and dispro-portionately, on the self, because information aboutthe self is more readily and automatically availableand people have a greater database of informationabout themselves than about others. On easy tasks,people recognize that they are performing well andtake insufficient account of the fact that the task islikely to be easy for others as well, leading to relativeoverconfidence. On difficult tasks, the reverse patternoccurs, leading to relative underconfidence.

Norton et al. (2012) introduce the notion of an“Ikea effect,” which, among other implications, sug-gests that underestimation of absolute performanceon easy tasks is not universal. (Note that Moore andHealy’s 2008 theory predicts overestimation of relativeperformance and underestimation of absolute perfor-mance, on easy tasks. Kruger and Dunning (1999) alsofind that people tend to underestimate their absoluteperformance on easy tasks.) Norton et al. find thatexerting personal effort to generate something, be it ahedonic or utilitarian product, leads to increased valu-ation of the product. Their findings hold true for tasksthat are as easy as assembling the simplest box soldby Ikea. They also find that people are unaware oftheir bias, as they expect the products of their personaleffort to be similarly overly valued by other individu-als. In a related study, Kruger et al. (2004) show thatwhen people lack information about an item’s quality,they assume that the quality must be higher if greatereffort was put into its creation.

The main focus of our study is not on whether peo-ple are, on average, over- or underconfident, but onthe impact of the act of generating an idea on relativeconfidence (i.e., on confidence in how an idea one caninvest in stacks up against other ideas), an issue that is

particularly relevant to entrepreneurial decision mak-ing. Our findings suggest that those who did not gen-erate an idea do not exhibit over- or underconfidencewhen investing in an idea, but those who generatedan idea show either overconfidence or underconfi-dence depending on whether they have the option toinvest in their own idea or in somebody else’s idea(that does not compete with one’s own idea).

Overall, our findings suggest that business ideageneration was perceived as an easy task by our studyparticipants. Kruger (1999), Moore and Cain (2007),and Moore and Healy (2008) all predict that if ideageneration is perceived as an easy task, then peo-ple will tend to overestimate the quality of their ownknown ideas relative to others’ unknown ideas, asituation characteristic of much entrepreneurial deci-sion making, as well as of one experimental treat-ment in our study. Overinvestment in one’s ownidea would also be consistent with people’s well-documented tendency to perceive themselves in apositive light (Tesser 1988; Bénabou and Tirole 2002,2003) and to overweight information derived frompersonal experience (Simonsohn et al. 2008), as well aswith Malmendier and Tate’s (2005) research on man-agerial overconfidence.

In addition to examining people’s investments intheir own ideas, we studied another key conditionin which individuals who had generated their ownidea had the opportunity to invest in another person’s(known) idea. We designed the experiment so thatthe idea that can be invested in would not, in fact,compete with one’s own idea. Hence, from a rationaleconomic perspective, whether an individual gener-ated her own idea should not be relevant to the deci-sion of whether and how much to invest in the otherperson’s idea. We included this condition because wedid expect it to matter. Specifically, we anticipatedthat people who had developed their own idea wouldevaluate specific other ideas they encountered morenegatively than would people who had not developedtheir own idea.

Our explanation for these predictions draws on fea-tures of each of the lines of research just discussed andis summarized in Figure 1, which presents the per-ceived promise of different business ideas by peoplewho are in the situations corresponding to the threeexperimental conditions in our main study. Depictedin the left-hand panel of the figure is the case, corre-sponding to the NOT condition in our experiment, ofa potential entrepreneur who has not generated herown idea but is deciding whether to invest in a spe-cific idea she is presented with (designated by thebold black circle containing an X). To decide whetherto invest in this idea, the individual must make aguess about its merit relative to other, unknown ideasthat the idea she is presented with may end up

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship4 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Figure 1 Illustration of Proposed Theoretical Account

Pro

mis

e of

bus

ines

s id

ea a

spe

rcei

ved

by in

divi

dual

s

NOT

: Own idea that may not be invested in

: Own idea that may be invested in

: Other people’s unknown ideas that idea one can invest in would compete against

: Somebody else’s known idea that may be invested in

: Other ideas that do not compete against the idea that can be invested in

OWN

OTH

Notes. NOTs are individuals who did not generate an idea but have the option to invest in an idea generated by somebody else. OWNs are individuals whogenerated an idea and may invest in their own idea. OTHs are individuals who generated an idea but may only invest in somebody else’s idea.

competing with (designated by the open circles). Weassume that the individual in this situation assessesboth the focal idea and the distribution of compet-ing ideas in an unbiased fashion. This means that, asdepicted in the figure, on average, the promise of thefocal idea will be rated as being in the middle of thedistribution of the perceived promise of the unknownideas.

The middle panel of Figure 1 depicts the situation,corresponding to the OWN condition in our experi-ment, of an individual who has generated her ownidea and has the opportunity to invest in it. Consistentwith prior research on overconfidence, we assumethat people tend to have an especially positive viewof ideas they generate themselves and are unawarethat their appraisals are biased. We represent suchoverconfidence by depicting, in the middle panel, anindividual who rates the promise of their own idea ashigh.1

1 One could argue that this assumption is inconsistent with the priorexperimental results showing that people tend to underestimate

A second key assumption, whose examination isthe central focus of a follow-up experiment wepresent after the main experiment, is that individualswho have generated an idea estimate the quality ofothers’ unknown ideas quite differently than others’known ideas, but individuals who have not gener-ated an idea do not manifest such a difference in theirestimates. Consistent with Moore and Healy (2008),Kruger (1999), and Kruger et al. (2004), we assumethat people estimate the quality of ideas that theydo not have direct exposure to based on their pri-ors. These priors are likely to be inflated when peoplehave proposed their own idea. In judging the qual-ity of other people’s unknown ideas (the open circles

their absolute performance on easy knowledge quizzes. However, itis not obvious that business ideas are seen as comparable in diffi-culty to easy quiz items. Perhaps more importantly, business ideasare the fruit of personal effort, and prior research (Norton et al. 2012)shows that outcomes of personal effort, even on easy tasks, tend tobe overly valued by people and, critically, that people expect othersto value those outcomes highly as well.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 5

in all three panels), when people have proposed theirown idea, it is perfectly rational for them to use theirappraisal of their own idea as a data point in judg-ing the likely quality of ideas they have not seen (see,e.g., Dawes 1989, Hoch 1987). However, because theirappraisal of their own idea is inflated, this results ina similar, but smaller, inflation in their guess aboutthe quality of other people’s unknown ideas. In Fig-ure 1, therefore, the perceived distribution of the qual-ity of unknown ideas is elevated in both the OWNand OTH conditions in which the focal individual hasgenerated her own idea, but not in the NOT condi-tion, in which the focal individual has not generatedan idea.

In the OWN condition, because the upward biasin the guess about the quality of unknown ideas isnot as great as the bias in judging their own idea,we expect that people will still overestimate the rel-ative value of their own ideas, which is consistentwith their tendency to maintain a positive view ofideas they generated. However, this is not the casein the OTH condition, in which people evaluate oth-ers’ known idea compared to competing unknownideas. We assume that people who have generatedan idea tend to evaluate others’ known ideas impar-tially,2 but the act of having generated their own ideabiases upward their guesses about the quality of oth-ers’ unknown ideas. As a result, people in the OTHcondition tend to view the known idea they are eval-uating as less favorable compared to unknown com-peting ideas, and they are less likely to invest in theknown idea.

Putting all these lines of reasoning together gen-erates the predictions that, first, as illustrated in themiddle panel of Figure 1, people will be overcon-fident about the quality of their own idea relativeto other people’s ideas. This prediction is consistentwith a wide range of positive self-biases as well aswith informational and egocentric accounts of over-confidence. Second, as illustrated by the right-handpanel of Figure 1, people who have generated an ideaof their own will tend to be underconfident aboutthe quality of somebody else’s known idea relativeto other people’s unknown ideas. Because they basetheir judgment of the quality of unknown idea inpart on their inflated appraisal of their own idea,they underestimate the relative quality of somebodyelse’s known idea. As illustrated by the left-handpanel of Figure 1, however, we do not expect peopleto exhibit over- or underconfidence when evaluatingothers’ ideas if they did not generate their own idea.

As a more nuanced test for our proposed theoreti-cal framework, we compare the pattern of over- and

2 It is possible that people denigrate others’ known ideas, whichwould only strengthen our predicted results.

underconfidence of participants who do and do nothave experience with entrepreneurship. Our theoreti-cal perspective predicts that those with greater expe-rience are more likely to have a high view of theirown abilities, which should lead to raising their over-confidence when they bet on their own idea but, crit-ically, also increasing underconfidence when bettingon another person’s idea.

Aversion to Potential Appropriation by OthersIn addition to examining the impact of idea genera-tion on investment decisions, we examine the impactof whether an idea not invested in could potentiallybe invested in by another individual. On the basis ofeconomic rationality, the presence of other investorswho compete with a focal investor might well affectthe focal investor’s valuation of an investment oppor-tunity. For example, if the opportunity attracts a lotof competition, this might decrease the prospects forfuture returns, making it less attractive. If, on theother hand, there were first-mover advantages, thenthe threat of another person investing might make itmore attractive to invest early. It would also be possi-ble that learning of others’ interest in investing mightindicate (or be interpreted as a kind of social proof)that the investment is good.

By contrast, there is no logical reason why thepresence of another investor who may only considerwhether to invest in the opportunity after the focalinvestor decides not to invest should affect the in-vestment decision of the focal investor. In this sce-nario, which characterizes our experimental design,the focal investor’s return on investment would notbe affected by the second investor’s decision, andfocal investors would not receive any informationabout whether the other investor was interested ininvestment until they had made their own irreversibleinvestment decisions.

The “possibility of appropriation by another per-son” aspect of the experiment was thus designedto isolate the impact of social factors on investmentdecisions. Prior research on and writing about entre-preneurship has emphasized the social dimension ofentrepreneurial decisions (Granovetter 1995, Sarasonet al. 2006). The presence of another person whomight invest in an unexploited opportunity adds tothe salience of losing the opportunity and its asso-ciated anticipation of regret and envy. Research onregret aversion (e.g., Loomes and Sugden 1982, Zee-lenberg et al. 1996, Zeelenberg 1999) and social com-parison envy (e.g., Salovey and Rodin 1984) finds thatdecisions are often affected by people’s aversion toanticipated regret or envy, which is likely to be farmore salient if another investor adopts an investmentone declined to make. As a result, and consistent with

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship6 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

the concept of takeover aversion introduced and doc-umented by Hoelzl and Loewenstein (2005), we pre-dict that potential entrepreneurs will invest more inan idea when there is a potential that the declinedopportunity could be appropriated by somebody else.

Main ExperimentExperimental DesignFigure 2 presents a narrative depiction of the flowof the experiment, a three by two factorial designwith two stages. The procedure of the experimentwas carefully explained to participants at differentpoints during the experiment, and participants wererequired to answer a number of quiz-style questionsat several points throughout the experiment to ensurethat they had correctly understood how the exper-iment and the investment process worked. If theyanswered questions incorrectly, they were providedwith clarifying information and then asked to answerthe question again. Interested readers can try out eachsegment of the experiment by going to the followingweb page: http://cmusds.qualtrics.com/SE/?SID=SV_2ml2AXLwd0YiHgp. The experiment was imple-mented online using the Qualtrics survey interface.

In the first stage of the experiment, some partici-pants were asked to develop a business idea—specif-ically, to propose a business to be opened in a loca-tion in Pittsburgh, Pennsylvania, that was familiar tothem. Although the location was familiar, participantswere given a detailed description as well as photosand a map of the location. In part so that all partici-pants would engage in both stages of the experiment(a feature that, if absent, could have produced selec-tion effects), all participants, whether they were askedto develop a business idea or not, participated in thefirst stage by, at a minimum, completing a question-naire that included questions about their experiencewith and interest in entrepreneurship. All participantswere entitled to receive a $12 payment for completingthe first stage of the experiment.

Figure 2 Flow of the Main Experiment

From the large pool of business ideas proposedby participants, ideas were clustered randomly intogroups of six. Each group of six ideas was presentedto an anonymous judge who ranked the ideas from 1to 6 based on their business promise and feasibility.Judges were recruited from Carnegie Mellon Univer-sity Ph.D. students in entrepreneurship, economics,and management, and public policy.

In the second stage, participants who had com-pleted the first stage were given an opportunity toinvest in a business idea. Out of each group of sixideas that had been judged against one another, twowere randomly selected. To test our predictions re-garding the impact of idea generation on confidenceand investment, three conditions were included in theexperimental design. In the “OWN” condition, eachof the two selected ideas was given to the personwho had developed it in the first stage. In the “OTH”condition, the same two selected ideas were given totwo other participants who had developed an idea inthe first stage but not one of the two selected ones(nor any of the other four ideas that were clusteredtogether in the group of six). Each participant in theOTH condition was given one of the selected ideasto invest in. To ensure that in all conditions no par-ticipant had, or believed that she had, informationabout ideas in the ranking other than the one theycould invest in, participants in the OTH conditionwere informed that their idea was not among the sixideas that had been ranked. Finally, in the “NOT” con-dition, the same two selected ideas were given to twoparticipants who had not developed a business ideain the first stage. Again, each participant in the NOTcondition was given one of the two ideas to investin. We gave the same ideas to participants in eachof the three conditions so we could conduct a fixedeffects analysis holding idea constant and focusing onthe impact of condition (reported in Table A.3 of theappendix).

Each participant in the OWN, OTH, and NOT con-ditions decided, independently, how much to investin the idea he or she was presented with. As shown

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 7

in Figure 2, in half of the groups (those in the ap-propriation-by-others condition, which is abbreviatedas APPR), OWNs, OTHs, and NOTs were informed,truthfully, that if they failed to invest in the idea, asecond investor would be given the opportunity toinvest in that idea. In the other half of groups, therewas no second investor. This design was implementedto test our predictions regarding the effects of the pos-sibility of appropriation.

Because the design was complicated, we also showhow subjects and ideas were grouped together, focus-ing only on the idea generation aspect of the study,and not on the threat of appropriation conditions. Fig-ure 3 depicts the procedure for a representative groupof 10 subjects, numbered 1 to 10 (top row). Eight ofthese subjects (numbered 1–8 in the diagram) wererandomly assigned to generate ideas, and two (num-bered 9 and 10) were not assigned to generate ideas.Six of the eight ideas were randomly selected (clus-tered in a rectangular box in the second row of thefigure) and ranked against each other by a judge,from best to worst (ranking not shown in the figure).Two of the six ideas (numbered 5 and 6 in the fig-ure) were then randomly selected. In the OWN con-dition (subjects numbered 5 and 6 in the third row ofthe figure), each of the two randomly selected ideaswas presented to the subject who developed that idea,and each subject had the opportunity to invest in herown idea. In the OTH condition (subjects numbered 1and 2 in the fourth row of the figure), each of therandomly selected two ideas, numbers 5 and 6, waspresented to a subject who had generated an idea,but not ideas 5 or 6. Each of the OTHs then decided

Figure 3 Assignment of Ideas to Subjects

.

.

8 2

how much to invest in the idea they were presented.Finally, in the NOT condition (subjects numbered 9and 10), each of the two subjects who had not gen-erated ideas were presented the randomly selectedideas (numbers 5 and 6) and decided how much toinvest.

We used the Becker–DeGroot–Marschak (Beckeret al. 1964) incentive-compatible method to elicit en-trepreneurs’ valuation of the opportunity to invest.Each participant decided, for a series of sequentiallyincreasing investment amounts, whether to invest ateach amount. Investment amounts started from $0.50and could go as high as $12 (increasing in 50-centincrements). This process continued until the par-ticipant decided either not to invest at a specifiedamount or reached the highest possible investmentamount ($12). Once a participant decided not to investat a specified amount, it was automatically checkedoff that she did not wish to invest for all amountsgreater than that. It was explained to participants thatone of the investment decisions (including those auto-matically checked off) would be randomly selected tocount for real. For example, if the investment deci-sion for $5 was randomly selected and the decisionfor $5 showed “Invest,” then the participant investedin the idea and paid $5 for making the investment.In another example, if the investment decision for $11was randomly selected and the participant’s decisionhad been to not invest at that amount, then she failedto invest in the idea and did not need to pay any-thing. Two of the quiz questions subjects were givenensured that they understood this procedure.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship8 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Table 1 Rank-Dependent Payoffs

Rank 1 (best) 2 3 4 5 6

Money reward ($) 12 8 4 0 0 0

To ensure that subjects were happy with theirinvestment decision, all of the investment choicesincluding those automatically checked off werepresented to them in a final page before submissionand they were instructed as follows: “If you are satis-fied with your investment decisions, submit them byproceeding to the next page. Otherwise use the backbutton to modify your decisions. Once you proceedto the next page, you will NOT be able to return tochange your decisions.”

Investment payoff was determined based on thejudge’s ranking of the idea (as described in Table 1). Ifan investor invested in an idea and it was ranked firstamong the six clustered ideas, the investor received$12. For a second-place ranking, the investor received$8, and for a third-place ranking, she received $4. Ifthe idea was ranked from 4 to 6, the investor receivedno payment. In each of these cases, the investor had topay the amount she had invested in the idea, with netlosses being deducted from the $12 fee they earnedfor participating in the first stage of the experiment.

All participants were informed that, whether or notthey invested in the idea, they would learn about thejudges’ ranking of the idea at the end of the exper-iment. This was done to prevent participants frominvesting for the purpose of learning the rank of theiridea, i.e., to satisfy curiosity. Additionally, those in theAPPR condition were informed that they would learnwhether the second investor invested in the idea, and,if the second investor had invested, how much he orshe had paid.

At the end of the experiment, after participants hadsubmitted their investment decisions, but before theylearned about their payoffs or the judge’s rankingof the idea, all participants completed a brief ques-tionnaire in which, among other questions, they wereasked to guess the rank of the idea they had an optionto invest in. Keeping all decisions within the experi-ment incentive compatible, we informed participantsthat correct guesses would be rewarded with an addi-tional $1.

Participants and ProcedureParticipants were recruited from alumni of CarnegieMellon University (CMU) as well as from a pool ofuniversity students and alumni who had lived orwere living in Pittsburgh, Pennsylvania. CMU alumnias well as students who had registered on a web-site indicating their interest in participating in exper-iments were sent invitation emails. We restrictivelyinvited university students and alumni who had lived

in Pittsburgh (more than 7,000 individuals) to partic-ipate in “an experiment on entrepreneurial decisionmaking” for a participation fee of $12 and a chanceto win money rewards up to a total of $24.5. No fur-ther information was given to them at this point. Mostparticipants who completed the experiment did sowithin days of receiving the invitation email. We didnot receive any completed response after the end offirst month, and web links to participate in the sec-ond stage of the experiment were removed after threemonths.

Three hundred participants completed both stagesof the experiment; however, 63 of them were secondinvestors (in the APPR condition) who were includedto avoid deception but are not central to the issuesaddressed by the study. Thus, the reported results arebased on 237 participants, of whom 136 were univer-sity alumni and 101 were students; 71 had entrepre-neurial experience (an average of 4.44 years for thosewho did). Overall, our sample provides a good mixof subjects with and without entrepreneurial experi-ence, which resembles the mix of experience (and lackthereof) in the early stages of entrepreneurship in thereal world. Table A.1 in the appendix presents a sum-mary of participant demographics, which shows thatthere were no differences in demographics betweenexperimental groups beyond what one would expectby chance.3

Measures and Variables

Dependent Variable. Participants’ willingness toinvest in the business opportunity is the main depen-dent variable of the study. The amount of invest-ment (or equivalently, individuals’ valuation of theopportunity) is a discrete cardinal (ratio-scale) vari-able that ranges from $0 to $12 (increasing in 50-centincrements).

Independent Variables. Idea generation status isdesignated by three binary variables (OWN, OTH, and

3 Our experimental procedures conform to those proposed by Sim-mons et al. (2011). On the basis of considerations of statisticalpower, we selected an initial population to whom participationinvitation emails were sent. After sending the last email, we waitedfor three months until web links to participate in the second stageof the experiment expired. Most participants who completed theexperiment did so within days of receiving the invitation email.We did not receive any completed response after the end of thefirst month. We started data analysis after web links to partici-pate expired at the end of third month. We included all completedresponses in our analysis (i.e., no observation was omitted). There-fore, we did not collect further data after initiating analysis, andour sample size was determined exogenously and was equal to thenumber of participants who completed the experiment. In addition,consistent with the suggestions of Simmons et al. (2011), all exper-imental conditions in this study include at least 20 observations,all variables and experimental conditions collected in the study arereported, no observation was rejected, and all statistical results arereported without covariates.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 9

NOT), each of which was set equal to 1 if a sub-ject was in that condition and to 0 otherwise. Thepossibility of appropriation by somebody else is indi-cated by a binary variable, APPR, which was set equalto 1 if participants belonged to the group facing theprospect of another investor investing in their idea ifthey did not.

Cognitive and Emotions Variables. A survey wasincluded at the end of experiment, after partici-pants finalized their investment decisions but beforethey were given information about their payoff andranking of the idea. It included items that mea-sured passion and anticipated regret, using five-pointLikert-type scales (ranging from strongly agree tostrongly disagree). Passion was measured by askingeach participant whether she felt very enthusiasticabout the idea. Anticipated regret was measured byasking each participant whether she would feel both-ered if she ended up not investing in the idea and itended up being highly ranked.4 Overconfidence wasmeasured by subtracting each participant’s guess ofthe ranking of the idea he or she could invest in fromthe judge’s ranking. In this sense, zero would repre-sent a well-calibrated judgment, while negative andpositive amounts are suggestive of underconfidenceand overconfidence, respectively.

Control Variables. Control variables include Fe-male, Rank (reflecting the judge’s ranking of the idea,ranging from 1 to 6), Alum (equal to 1 if the partic-ipant is an alumnus and 0 if a student), Experience(indicating the total number of months the partici-pant worked as self-employed or in a business com-pletely or collaboratively started by the participant),and Field of study (coded with dummy variables forengineering, basic sciences, medical/health, art, socialscience/humanities, and business and economics).Because of an oversight, we failed to collect age infor-mation for all participants. Including experience andstudent/alumnus status as control variables shouldpartly control for the effect of age.

ResultsIn our study, participants’ valuation of opportunitywas measured by the amount they chose to invest.Participants could not make negative investments andwere not permitted to invest more than their $12 par-ticipation fee. Because our measure of valuation is

4 We also elicited anticipated envy by asking participants if theywould feel bothered if they ended up not investing in the idea anda second investor made a profit on it. However, we only elicitedthis variable from subjects in the APPR group, since there was nosecond investor whom subjects could envy in the NO-APPR group.Given the lack of data on this variable in the NO-APPR condi-tion, we were unable to compare it across conditions or determinewhether it mediated the impact of takeover.

censored, while its underlying behavioral tendency isnot, we used a Tobit model (Tobin 1958) to analyzethe effect of idea generation and potential appropria-tion on investment amounts.

Figure 4 presents the means for investments madeby participants in the six different experimental con-ditions. The overall pattern of investment levels pro-vides clear evidence for our expected effects. OWNsinvested the most, OTHs invested the least, and NOTslay in between. On average, OWNs invested 93%more than OTHs ($5.18 versus $2.68). Participants inthe APPR condition invested 19% more than those inthe NO-APPR condition ($4.13 versus $3.48).

Table 2 reports the results of the hierarchical Tobitregressions analyzing the effects of the experimen-tal treatments on investment. The first specification(Model 1) includes only the effects of idea gener-ation (OWN and OTH, with NOT as the referencegroup) and the possibility of appropriation as inde-pendent variables. Both variables have the predictedsigns. The effect of the possibility of appropriation issignificant at the 0.1 level, and the effect of idea gen-eration is significant at the 0.01 level for OWN andborderline insignificant (p = 0015) for OTH. Model 2adds interactions between the possibility of appropri-ation and idea generation to test whether the effect ofthe possibility of appropriation depends on idea gen-eration status. Comparison of Models 1 and 2 usingthe F -test for joint effect of interaction terms indi-cates that adding the interaction terms did not sig-nificantly improve the overall fit of the regression(Wald = 1068; p < 0043). In Model 3 we add controlvariables to Model 1; these have little impact on theprimary effects. Model 4 is the same as Model 3 butcontrols for a participant’s guess of the rank instead ofthe judge’s ranking as in Model 3. Model 4 is includedto show that the pattern of results does not change ifwe control for the subjective ranking by participants

Figure 4 Investment Levels for Each Experimental Group

Notes. Numbers in parentheses are standard deviations. Investment wasmeasured using a discrete cardinal scale ranging from $0 to $12 and increas-ing in 50-cent increments.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship10 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Table 2 Tobit Regressions Examining the Effects of ExperimentalTreatments on Investment Levels

Model 1 Model 2 Model 3 Model 4

Intercept 2078∗∗∗ 3020∗∗∗ 3030∗∗∗ 7024∗∗∗

400445 400555 400845 400695APPR 0070∗ −0010 0073∗ 0072∗∗

400425 400765 400415 400355OWN 2011∗∗∗ 1059∗∗ 2018∗∗∗ 1061∗∗∗

00525 400745 400525 400445OTH −0076 −1044∗ −0075 −0015

400535 400765 400535 400455APPR × OWN 0098

410035APPR × OTH 1032

410055Rank −0016

400155Rank guess −1049∗∗∗

400155Female 0073∗ 0038

400435 400365Alum −0013 0059

400455 400365Experience 0001 <0001

400015 4<00015Engineeringa −0041 −0051

400535 400445Basic sciencesa −0056 −0075

400775 400655Medical/Healtha 1077 0070

410265 410055Art a 0029 0012

410115 400925Social science/Humanitiesa −1028 −1013∗∗

400615 400515Sigma 3018∗∗∗ 3017∗∗∗ 3010∗∗∗ 2058∗∗∗

400175 400175 400165 400135AIC 1,125 1,127 1,130 1,044N 237 237 237 237

Note. AIC, Akaike information criterion.aEngineering, Basic sciences, Medical/Health, Art, and Social science/

Humanities are dummy variables for field of study of participants. The refer-ence group is Business and economics.

∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p =

0001.

instead of judges’ rankings. The experimental manip-ulations, therefore, had their anticipated effects anddid not show any indication of interactions.

Figures 5a, 5b, and 5c present means for Passion,Overconfidence, and Regret across the six experimen-tal conditions, respectively, and Table 3 presents threelinear regression models in which these variablesare dependent variables. Both the idea generationand the possibility of appropriation manipulationssignificantly influence passion, although whether pas-sion should be considered a mediator or a variablethat simply reflects participants’ tendency to act onthe opportunity is unclear (Adams et al. 2006). Over-confidence, also, was significantly affected by theidea generation manipulation, but not the possibilityof appropriation manipulation. On average, OWNsshow overconfidence in their ranking of idea theycould invest in; they rank the idea on average better

Figure 5a Level of Passion for Each Experimental Group

Notes. Numbers within parentheses are standard deviations. Passion wasmeasured using a five-point Likert-type scale ranging from 1 to 5.

Figure 5b Overconfidence Levels for Each Experimental Group

Notes. Numbers within parentheses are standard deviations. The measureof overconfidence ranges from −5 to +5.

Figure 5c Level of Anticipated Regret for Each Experimental Group

Notes. Numbers within parentheses are standard deviations. Anticipatedregret was measured using a five-point Likert-type scale ranging from 1 to 5.

than judge’s ranking. More interestingly, OTHs showunderconfidence in their ranking of the idea theycould invest in. NOTs lie in between, and their guessof the ranking of the idea in the NO-APPR group isparticularly close to the judge’s ranking. In general,

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 11

Table 3 Passion, Overconfidence, and Regret as DependentVariables (Ordinary Least Squares)

Passion Overconfidence Regret

Intercept 3028∗∗∗ −2054∗∗∗ 3056∗∗∗

400275 400325 400295APPR 0024∗ <0001 −0014

400135 400165 400155OWN 0090∗∗∗ 0038∗ 0029

400175 400205 400185OTH −0048∗∗∗ −0042∗∗ 0005

400175 400205 400185Control variables Included Included IncludedR2 0032 0057 0015Adjusted R2 0028 0055 0011F -statistics 8073∗∗∗ 24093∗∗∗ 3032∗∗∗

N 237 237 237

∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p =

0001.

the pattern of investment levels coupled with the pat-tern of overconfidence across conditions supports ourexplanation that the difference in confidence (over-and underconfidence) resulting from idea generationcan lead to overvaluation or undervaluation of eco-nomic opportunities. Regression results (reported inTable 3) confirm that, compared with NOTs, over-confidence is significantly higher for OWNs (p < 001)and significantly lower and negative (i.e., underconfi-dence) for OTHs (p < 0005).

As seen in Figure 5c and Table 3, Regret is not signif-icantly affected by either experimental manipulation,which suggests that anticipated regret is not a maindriver of the aversion to appropriation by others inour experiment. This finding runs counter to the find-ing from Hoelzl and Loewenstein (2005) that accen-tuated anticipation of regret contributes to takeoveraversion. Regret is not always consciously accessedand is therefore difficult to measure with a subjectiveresponse item. In addition, regret was measured with-out any anchor points or points of comparison (seeHsee et al. 1999). Thus, the measure may not havebeen defined precisely enough to pick up subtleties ofdifferences across conditions. Participants in all con-ditions reported, on average, that they would experi-ence substantial regret if they failed to invest but theirinvestment would have paid off.

We also investigated whether passion, overcon-fidence, and regret mediate the effect of experi-mental manipulations on participants’ investment inentrepreneurial opportunities. Table 4 reports Tobitmodels in which Passion, Overconfidence, and Regretare included as independent variables. Model 0 isthe base model, which is the same as Model 3 inTable 2. Adding Passion as an independent variable(see Model 1 in Table 4) renders all focal main ef-fects insignificant. Nevertheless, as stated previously,

Table 4 Tobit Regressions Including Passion, Overconfidence, andRegret as Independent Variables

Model 0 Model 1 Model 2 Model 3 Model 4

Intercept 3030∗∗∗ −3020∗∗∗ 7010∗∗∗ 0067 0007400845 400885 400805 410065 410245

APPR 0073∗ 0021 0072∗∗ 0084∗∗ 0038400415 400335 400355 400405 400325

OWN 2018∗∗∗ 0042 1061∗∗∗ 1098∗∗∗ 0060400525 400435 400445 400515 400415

OTH −0075 0024 −0015 −0080 0026400535 400435 400455 400515 400415

Passion 1098∗∗∗ 1039∗∗∗

400175 400205Overconfidence 1049∗∗∗ 0078∗∗∗

400155 400165Regret 0073∗∗∗ 0018

400195 400155Control variables Included Included Included Included IncludedSigma 3010∗∗∗ 2044∗∗∗ 2058∗∗∗ 3002∗∗∗ 2031∗∗∗

400165 400135 400135 400165 400125AIC 1,130 1,021 1,046 1,117 999.72N 237 237 237 237 237

Note. AIC, Akaike information criterion.∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p =

0001.

whether passion should be considered a mediator ora variable that simply reflects participants’ tendencyto act on the opportunity is unclear. Model 2, whichincludes Overconfidence as an independent variable,shows a significant impact of overconfidence and sub-stantially reduced coefficients for OWN and OTH,findings that are consistent with partial mediation ofthe idea generation effect by Overconfidence. Model 3,which includes Regret as an independent variable,results in a significant coefficient for Regret. It does not,however, support a mediation effect for Regret, giventhat inclusion of Regret has little impact on the coeffi-cients for the experimental treatments. Note also thatRegret is unaffected by experimental manipulation aspresented in Table 3 and Figure 5c, a result that is alsoincompatible with Regret playing a mediating role.

Finally, we compared the pattern of results betweenparticipants who did or did not have experience withentrepreneurship. We compared the 166 participantswho had no experience with entrepreneurship to the71 people who had at least some experience withentrepreneurship (this is as close as we can come toa median split, since the median individual had noexperience). Figures 6a and 6b present a comparativeillustration of means for Overconfidence and Investmentfor participants who have and do not have entrepre-neurial experience. The pattern of results shows, con-sistent with the account of the effects proposed in thetheory section, that people who have entrepreneurial

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship12 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Figure 6a Overconfidence Levels by Entrepreneurial Experience

Notes. Numbers in parentheses are standard deviations. The measure ofoverconfidence ranges from −5 to +5.

Figure 6b Investment Levels by Entrepreneurial Experience

Notes. Numbers in parentheses are standard deviations. Investment wasmeasured using a discrete cardinal scale ranging from $0 to $12 and increas-ing in 50-cent increments.

experience show accentuated levels of both over- andunderconfidence.5

We also compared the pattern of results betweenstudents and nonstudents. Student status is looselycorrelated with experience status (Corr = 0024), asseveral student participants were doing their grad-uate studies and did, in fact, have entrepreneurialexperience, and several nonstudents were employeeswith no entrepreneurial experience. We did not finda difference for mean investment levels between stu-dents and nonstudents, although students exhibitedslightly more overconfidence.

5 For those who reported experience with entrepreneurship, we alsoasked them whether they had ever experienced failure, and weconducted a post hoc analysis of the experience or nonexperienceof failure on overconfidence (see the appendix, Figure A.1). Whilethe N values are small, the results are strikingly in accordancewith what one would expect. Both groups with experience showeda greater impact of the main experimental manipulation on theirlevel of confidence, but those who had not experienced failure weresubstantially more overconfident than those who had experiencedfailure.

Follow-up ExperimentThe follow-up experiment was designed to focus onthe OTH condition—specifically, to test the less obvi-ous key assumption underlying our predictions of theresults of the first study: that those who generatedan idea themselves evaluate others’ unknown ideasmore favorably and others’ known ideas less favor-ably compared with those who did not generate anidea themselves. This is the assumption that drivesour most surprising prediction: that people who gen-erate their own ideas will underestimate the valueof another person’s known idea that they have theopportunity to invest in.

The follow-up experiment had a two by two fac-torial design with two stages. In the first stage, halfof the participants were randomly assigned to theOTH condition and were asked to propose an ideafor facility development in a specific location in afamiliar neighborhood adjacent to the CMU campusin Pittsburgh, Pennsylvania. The other half of partici-pants were randomly assigned to the NOT conditionand were not asked to propose a facility developmentidea. To keep the exerted effort levels about the sameacross all participants, individuals in the NOT condi-tion were asked to propose an idea for another unre-lated matter—namely, improving pedestrian safety inthe neighborhood adjacent to the CMU campus. Allparticipants were given $10 as a reward for proposingtheir idea, which then served as money they couldinvest if they chose to. The facility development ideaswere rated by a group of judges, similar to those inthe main study, who had no financial stake in the out-come of the experiment. Ideas were rated on a scalefrom 0 to 10, where 0 meant that the idea was deemedto be totally worthless, and 10 meant that the ideawas financially viable, seen as creating superior ben-efits for the neighborhood and CMU community, andalmost certainly worthy of implementation.

In the second stage, participants were randomlyassigned to either the KNOWN or UNKNOWN con-dition and were asked to invest in one of the facilitydevelopment ideas. In the KNOWN condition, par-ticipants could invest in an idea that was known tothem—i.e., presented in the same detail with whichjudges could see it. In the UNKNOWN condition,participants were simply told that the idea they hadan option to invest in was randomly selected fromthe pool of ideas that were rated by judges, but theywould not be shown the specific idea until after theymade their investment decision.

Before making their investment decisions, all par-ticipants were informed that the idea they wouldhave an option to invest in had been rated by agroup of expert judges, and the method of ratingwas explained to them. Participants could invest anyamount from $0 to $10. If the judges’ rating of the

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 13

idea they had an option to invest in ended up beingless than 7, they would lose the amount they invested,but if the judges’ rating ended up being 7 or more,the amount of their investment was doubled andreturned to them. After completing the investmenttask, participants guessed the judges’ rating of theidea they had an option to invest in, as well as theidea they generated (if they had generated one), andwere informed that they would be given an extra $1for each correct guess. The difference between the par-ticipants’ guesses and the mean of the judges’ ratings(participant’s guess minus judges’ rating) of the ideasthat could be invested in was used to create a measureof overconfidence in the quality of the idea.

The experiment was implemented online using theQualtrics survey interface. Participants were recruitedfrom alumni and students of CMU and other universi-ties in Pittsburgh. Ninety-four participants completedthe experiment (22 in OTH-KNOWN, 21 in OTH-UNKNOWN, 25 in NOT-KNOWN, and 26 in NOT-UNKNOWN conditions). Of 94 participants, 40 wereuniversity alumni and 54 were students. Thirty-threeof the 94 participants had entrepreneurial experience,and of these, their average duration of entrepreneurialexperience was 5.09 years. Data for demographics andother control variables were collected using a surveybefore the experiment.6

ResultsFigure 7, panels (a) and (b) present the mean invest-ment levels across four experimental conditions. Thepattern of investment levels made by OTHs supportsour theoretical assumption that individuals who haveproposed an idea respond to others’ unknown ideasquite differently than other’s known ideas, and theytend to form an inflated appraisal of the quality ofother people’s unknown ideas. OTHs’ investment inunknown ideas is 98.4% more than their investmentin known ideas, while NOTs investments in unknownand known ideas are about the same (only a 9.8% dif-ference). Figure 8, panels (a) and (b) show that OTHsare overconfident in the quality of others’ unknownideas but underconfident in the quality of others’known ideas, but NOTs’ confidence in the quality ofothers’ ideas does not differ by whether the idea isknown.

Table 5 reports the results of hierarchical Tobit re-gressions analyzing the effect of the experimentaltreatments of the follow-up experiment on investmentlevels. Model 5 in Table 5 reports results for partic-ipants in the OTH condition and shows that OTHs’average investment on unknown ideas is significantly

6 Our procedures for the follow-up experiment were exactly thesame as those for the main experiment (see Footnote 3) and con-formed to those proposed by Simmons et al. (2011).

more than that for known ideas (p < 0001). Model 6in Table 5 reports results for participants in the NOTcondition and shows that NOTs’ average investmentlevels do not significantly differ between known andunknown ideas. Model 2 in Table 5 reports resultsfor all participants and shows that, on average, OTHsinvest less than NOTs, and investments made onUNKNOWN ideas are more than those on KNOWNideas. Model 3 in Table 5 adds the interaction betweenUNKNOWN and OTH to Model 2, and it shows thatthe interaction term is marginally significant (p < 001),and its addition to the model renders the main effect ofUNKNOWN insignificant. This result provides furthersupportive evidence that the impact of UNKNOWNon investment is attributable to participants in theOTH condition, because once belonging to OTH-UNKNOWN cell is controlled for by addition ofthe interaction term, UNKNOWN has no significantimpact on participants’ tendency to invest. Resultsfrom Model 8 in Table 5 shows that OTHs evaluateothers’ known ideas significantly lower than do NOTs(p < 0001). This result provides strong evidence for thepredicted impact of idea generation on individuals’valuation of others’ (known) ideas, specifically for thedifference between OTHs and NOTs, a finding thatwas marginal in the main experiment.

To test whether OTHs’ tendency to value unknownideas more than known ideas is mediated by theiroverconfidence in the quality of unknown ideas, weran two regressions using experimental data fromparticipants in the OTH condition. Table A.4 (in theappendix) reports the results of these regressions.Model 1, in which Overconfidence is the dependentvariable, shows the significant positive impact ofUNKNOWN on Overconfidence (p < 0005). Model 2 issimilar to Model 4 from Table 5 except for additionof Overconfidence as an independent variable. Resultsreported in Model 2 show that addition of Overcon-fidence resulted in a decrease in value of the coeffi-cient of UNKNOWN, from 2.82 to 1.61. These resultsare consistent with partial mediation of the impact ofUNKNOWN on investment levels by Overconfidence,which in turn is supportive of the theoretical assump-tion that idea generators tendency to value unknownideas higher than known ideas is driven by their over-confidence in the quality of unknown ideas.

Concluding RemarksIndividuals who have developed an entrepreneurialidea are more likely to form biased evaluations ofeconomic opportunities and to make distorted invest-ment decisions as a result. On the one hand, whenthey have the opportunity to invest in an idea theygenerated, their judgment of the value of economicopportunities is characterized by overconfidence, and

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship14 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Figure 7 Investment Levels for Known and Unknown Ideas in the Follow-up Experiment: (a) OTH, (b) NOT

Notes. Numbers in parentheses are standard deviations. Investment was measured using a continuous scale ranging from $0 to $10.

Figure 8 Overconfidence Levels for Known and Unknown Ideas in the Follow-up Experiment: (a) OTH, (b) NOT

Notes. Numbers in parentheses are standard deviations. The measure of overconfidence ranges from −10 to +10.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 15

Table 5 Tobit Regressions Examining the Effect of Experimental Treatments on Investment Levels in the Follow-up Experiment

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9(All participants) (All participants) (All participants) (Only OTHs) (Only OTHs) (Only NOTs) (Only NOTs) (Only KNOWNs) (Only KNOWNs)

Intercept 4014∗∗∗ 6059∗∗ 7014∗∗ 1096∗∗∗ 3035 4081∗∗∗ 13066∗∗ 4082∗∗∗ 5032400715 430105 430085 400705 430095 400935 460055 400965 440705

OTH −1062∗ −1096∗∗ −3025∗∗∗ −3019∗∗ −3073∗∗∗

400845 400885 410175 410425 410305UNKNOWN 1078∗∗ 1075∗∗ 0050 3005∗∗∗ 2082∗∗∗ 0056 0081

400845 400835 410105 410005 410015 410285 410265OTH × 2075∗

UNKNOWN 410645Judges’ rating −0010 −0006 −0047 0005 0095∗∗

400285 400275 400385 400385 400445Female −0069 0053 1004 −2007∗ 0087

400865 400855 410095 410265 410325Alum 1082 1072 1077 −1021 2002

410135 410125 410375 410735 410715Experience −0029 −0038 1016 −0090 −0023

400315 400315 410325 400695 400485Collaboration −1011 −1032 0003 −4078∗ −3037∗

410395 410375 410485 420815 410965Age 0000 0001 0000 0003 −0004

400045 400045 400045 400075 400065Sigma 3095∗∗∗ 3082∗∗∗ 3076∗∗∗ 3019∗∗∗ 3005∗∗∗ 4046∗∗∗ 4008∗∗∗ 4061∗∗∗ 4001∗∗∗

AIC 458086 464064 463086 200021 208017 256056 259003 220099 222068N 94 94 94 43 43 51 51 47 47

Notes. Alum is a dummy variable that equals 1 if the participant is an alumnus and 0 if a student. Collaboration is a dummy variable that is 1 if the participanthas ever collaborated with others in starting a venture and 0 otherwise. Experience is the participant’s level of entrepreneurial experience measured by thelogarithm of the number of months. AIC, Akaike information criterion.

∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p = 0001.

they are prone to overinvest. On the other hand, whenconfronted with ideas developed by others, those whohave generated ideas themselves exhibit underconfi-dence and are insufficiently likely to invest. In addi-tion, the threat that others may invest in an idea ifa potential investor forgoes the investment increaseswillingness to invest. Finally, entrepreneurial experi-ence not only does not help mitigate the observedover- and underconfidence but seems to aggravateit. These findings contribute to our understanding ofthe psychological factors that influence entrepreneur-ial judgments and investment behaviors, over andabove the effect of expected return on investment.

At a policy level, our findings have implicationsfor debates over adjusting the fees and stringencyof enforcement of intellectual property rights (IPRs).While traditional accounts of intellectual propertyrights have assumed that strengthened IPR protec-tion would lead to increased motivation to innovate,later research (e.g., Spence 1984; Cohen and Levinthal1989a, b) drew attention to limitations of this per-spective by suggesting that decreased IPR protectionmight result in increased spillover, which in turn couldincentivize follower innovators who would benefitfrom available knowledge. Despite recognizing thepossibility that follower entrepreneurs could benefitfrom weakened IPRs, neither Spence (1984) nor Cohenand Levinthal (1989b, a) argued that lowering the costof imitation in some market settings might, in the net,

increase pioneer entrepreneurs’ willingness to investand enter the market (though they did not rule outsuch a possibility). Our analysis suggests three newinsights supporting the possibility that such an effectmight occur. First, the current research suggests thatlowering the cost of imitation, e.g., by decreasing thestringency of enforcement of IPRs, may increase entre-preneurship by pioneer entrepreneurs who are afraidthat if they do not exploit an opportunity, someoneelse may do so. This may particularly be the casewhen, subsequent to commercialization of a break-through idea, entrepreneurial profit can be appro-priated through mechanisms that do not depend onenforcement of IPRs, such as scale economies, leadtime, first mover, or learning curve advantages.

Second, we find that imitators may be less threat-ening for entrepreneurship than what is generallyassumed. Imitator entrepreneurs, except of the mostrote variety, are likely to have generated their ownideas at one point. When faced with the opportu-nity to exploit somebody else’s idea, such imita-tors may undervalue the opportunity and act lessopportunistically than what pure profit maximizationwould predict.

Finally, our findings reveal that the investmentdecisions of pure imitators who have never developedtheir own ideas may be better (less biased) than thoseof innovators, who have experience with inventingan idea. Such imitators are, therefore, more likely to

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship16 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

invest, or desist from investing, when it makes eco-nomic sense. If relaxed enforcement of IPRs bringsmore of such investors into play, it could lead to animprovement in the efficiency of investment. Takentogether, this trio of possible effects suggests that amore lenient appropriability regime could, surpris-ingly, have positive effects on entrepreneurship.

Our findings regarding the impact of idea genera-tion on the evaluation of economic opportunities canalso potentially shed light on one factor sometimescited as a cause of spin-offs, as well as the indus-trial clusters that often emerge from spin-offs: thefrequency of serious disagreements over the promiseof ideas. Klepper (2007) and Klepper and Thompson(2010) proposed that disagreements about pursuingideas led to the spawning of spin-offs by inventorsof the ideas in the laser and automobile industries,among others. Nevertheless, mechanisms throughwhich such disagreements emerge have received lit-tle attention from researchers. Our research pointsto a mechanism that may play an important rolewhen multiple decision makers within a firm decidewhether to invest in an idea, such as whether tolaunch or discontinue a product. Specifically, individ-ual employees’ judgments of the promise of the ideaare likely to be affected by their self-perceived con-tribution to the development of the idea. Individualswho believe they deserve credit for generating theidea are likely to value it more highly than would oth-ers in the firm, especially those who have proposedcompeting ideas. These differences in idea valuation,derived from differences in self-perceived responsibil-ity for generating ideas, can result in conflicts overwhether to pursue ideas. Self-perceived inventors ofthe idea may then decide to leave the parent firm topursue the idea through a spin-off. We also find thatexperienced people, and specifically those who havean idea of their own, are more likely to exhibit over-or underconfidence, which might help explain one ofthe key findings of Klepper (2007)—that successfulfirms spawn more spin-offs. This might occur becausesuccessful firms tend to have more experienced (orotherwise knowledgeable) employees who, based onour findings, excessively overvalue their own ideaand undervalue others’ ideas and, hence, are morelikely to find themselves in disagreement with otherson the promise of ideas and to decide to leave thefirm to pursue a seemingly promising idea that theybelieve is undervalued by the firm.

Identifying the importance of the possibility ofappropriation by others as a motive underlying entre-preneurship can also, potentially, help to explain thephenomenon of “rush to market entry.” Famous exam-ples of rush to market entry occurred during the boomin Internet-related businesses in 1990s, when firms,many of which consisted of one or a few individuals,

rushed to enter the market with similar products thatwere insufficiently refined or tested. For instance, inthe early days of online search engines, many compa-nies entered the market one after the other (such asAltavista, Infoseek, Yahoo!, and WebCrawler, amongmany others), using basically similar algorithms to dothe online search tasks. Rush to market entry in theInternet industry went beyond what could be econom-ically justified by first-mover advantage for pioneerentrants (Lieberman 2007). Our research, and specifi-cally the notion of aversion to appropriation by oth-ers, could help to explain such rush to market entry asbeing a function, in part, of entrepreneurs’ fear that ifthey fail to exploit an opportunity, someone else willdo so (Lieberman 2007, Lieberman and Asaba 2006).Such an effect would be over and above any moti-vation arising from the perfectly rational fear of get-ting scooped. We find that both innovator and imitatorinvestors are affected by the threat that someone elsemay invest, even when the second potential investor’spresence has no impact on the returns of the initialinvestor. Therefore, the possibility of appropriation ofidea by others in economic markets may motivatenot only initial entry but also subsequent “copycat”entrants, resulting in the phenomenon of rush to mar-ket entry.

Our findings provoke questions whose answerswould require further investigation. For instance, itwould be interesting to explore whether entrepre-neurs’ increased motivation to invest is due to thesimple presence of second investors or the fact thatentrepreneurs will hear about the decisions and out-comes of second investors. The latter explanation isnot supported by our finding that the level of antic-ipated regret among entrepreneurs is unaffected bythe presence of a second investor, but this null find-ing might also be due to noise in our measure ofanticipated regret. An experiment in which entrepre-neurs would either receive or not receive informationabout investment decisions of second investors wouldprovide more evidence on the psychological mecha-nism that underlies the aversion to the possibility ofappropriation by others. It would also be interestingto explicitly examine, via either experiments or fielddata, the impact of the biases introduced by idea gen-eration on spin-offs, as well as the impact of the possi-bility of appropriation by others on the phenomenonof rush to market.

The impact of the two mechanisms—idea genera-tion and the possibility of appropriation by others—discussed here in connection with entrepreneurshipvery likely play an important role in other contexts.For example, academics seem prone to overrate thepromise of their own ideas while being overly criti-cal of others’ findings and theories. Cognizant of sucheffects, editors of journals should be (and many, if not

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 17

most, probably are) wary of soliciting reviews from,or at least placing much weight on the judgments of,researchers who have proposed alternative ideas. Bythe same token, many authors’ protests of unfair treat-ment are doubtlessly colored by overblown appraisalsof the value of their own ideas. Likewise, upon reflec-tion on the influence of the possibility of appropria-tion, it is easy to imagine that academics would bemore likely to pursue research ideas if they are afraidthat if they fail to pursue the ideas, someone else maydo so.

AppendixTable A.1 Demographics

Overall OWN NOT OTH p-valueb NO-APPR APPR p-valueb

Female 112 44 33 35 0065 59 53 0028847026%9 851016%9 845083%9 844030%9 850086%9 843080%9

Alum 136 51 41 44 0089 69 67 0052857038%9 859030%9 856094%9 855070%9 859048%9 855037%9

Experience 15097 20037 8070 17080 0021 14002 17083 00496420607 6470977 6300207 6450617 6380937 6450947

Engineeringa 80 29 22 29 0073 36 44 0038833076%9 833072%9 830056%9 836071%9 831003%9 836036%9

Basic sciencesa 23 14 6 3 0002 13 10 004589070%9 816028%9 88033%9 83080%9 811021%9 88026%9

Medical/Healtha 7 2 1 4 0038 4 3 006682095%9 82039%9 81033%9 85006%9 83045%9 82048%9

Art a 9 3 4 2 0062 3 6 003483080%9 83049%9 85056%9 82053%9 82059%9 84096%9

Social science/Humanitiesa 47 20 10 17 0031 25 22 0052819083%9 823026%9 813089%9 821052%9 821055%9 818018%9

Business and economicsa 71 18 29 24 0003 35 36 0094829096%9 820093%9 840028%9 830038%9 830017%9 8290759

Rank 3025 3023 3026 3025 0099 3014 3036 0024610437 610447 610417 610457 610447 610427

Rank 3045 3023 — 3068 0007 3045 3045 0098(developed idea) 610627 610447 610787 610597 610677

N 237 86 72 79 126 121

Notes. Numbers inside brackets denote standard deviations. Numbers inside braces denote percentages.aDummy variables for participants’ field of study.bReported p-values are based on analysis of variance results.

Table A.2 Summary Statistics for Outcome Variables

OWN, NOT, OTH, OWN, NOT, OTH,Overall OWN NOT OTH p-value NO-APPR APPR p-value NO-APPR NO-APPR NO-APPR APPR APPR APPR p-value

Investment 3082 5018 3042 2069 <0001 3049 4013 0009 4075 3044 2020 5059 3040 3019 <0001620987 630197 620697 620397 620937 630017 630257 620707 620157 630127 620727 620557

Passion 3022 3095 3004 2057 <0001 3009 3033 0012 3079 3003 2043 4011 3005 2072 <0001610187 600887 610127 610097 610167 610207 600927 610067 610067 600817 610187 610127

Regret 3038 3056 3028 3028 0021 3044 3031 0038 3050 3047 3038 3061 3011 3018 0035610717 610087 610197 610247 610117 610227 610027 610167 610197 610157 610207 610297

Envy 3006a 3020a 3000a 2095a 0057 3006 3020 3000 2095 0057610717 610137 610237 610167 — 610177 — — — — 610137 610237 610167

Overconfidence 0012 0050 0011 −0028 0002 0002 0022 0038 0040 −0005 −0033 0059 0026 −0023 0012610797 610817 610677 610827 610817 610777 610827 610617 610927 610827 610717 610727

Rank guess 3013 2073 3015 3053 <0001 3012 3013 0094 2071 3017 3050 2075 3013 3056 <0001610277 610117 610277 610307 610217 610327 600997 610117 610387 610227 610427 610237

N 237 86 72 79 126 121 42 34 40 44 38 39

Note. Numbers inside brackets denote standard deviations.aFor Overall, OWN, NOT, and OTH columns, N is equal to 121, 44, 38, and 39, respectively.

Supplemental MaterialSupplemental material to this paper is available at https://doi.org/10.1287/mnsc.2016.2566.

AcknowledgmentsThe authors thank David Hounshell, Irina Cojuharenco,and all members of the Department of Social and DecisionSciences at Carnegie Mellon University for their thought-ful and helpful comments during multiple stages of thisresearch. Support for this research was provided by the Fun-dação para a Ciência e a Tecnologia (Portuguese Foundationfor Science and Technology) through the Carnegie MellonPortugal Program [Grant SFRH/BD/51159/2010].

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on Entrepreneurship18 Management Science, Articles in Advance, pp. 1–19, © 2016 INFORMS

Figure A.1 Overconfidence Levels for Entrepreneurs Who Did andDid Not Experience Failure

Table A.3 Fixed Effects Analysis of Effect of Experimental Treatmentson Investment Levels

Fixed effect Tobit

Intercept 3057 3030∗∗∗

420665 400845APPR 0073∗

400415OWN 1079∗∗∗ 2018∗∗∗

400445 400525OTH −0065 −0075

400455 400535Rank −0016

400155Female 0076∗ 0073∗

400455 400435Alum −0092∗ −0013

400555 400455Experience 0001 0001

400015 400015Engineering −0032 −0041

400555 400535Basic sciences 0003 −0056

400825 400775Medical/Health 2025∗ 1077

410295 410265Art 0060 0029

410125 410115Social science/Humanities −0001 −1028

400605 400615Sigma — 3010∗∗∗

400165AIC — 1,130N 237 237

Notes. The table presents a fixed effects analysis examining the impact of theexperimental treatment on investment level (the dependent variable). Tobitestimates from the main regression analysis are included for ease of compar-ison. Rank is not included in the fixed effects analysis because it is constantfor the same idea (and hence is controlled for via the fixed effects). The fixedeffect analysis was conducted with a linear regression that included dummyvariables for ideas. AIC, Akaike information criterion.

∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p = 0001.

Table A.4 Examining the Mediating Role of Overconfidence on theImpact of UNKNOWN on Investment Levels (Within theOTH Condition in the Follow-up Experiment)

Model 1a Model 2b

(Only OTHs) (Only OTHs)(DV: Overconfidence) (DV: Investment)

Intercept 4006∗∗ −0015410875 420875

OTHUNKNOWN 1043∗∗ 1061

400625 400945∗

Overconfidence 0082∗∗∗

400245Judges’ rating −0078∗∗∗ 0020

400235 400395Female 0015 0085

400665 400955Alum 0002 1065

400825 410195Experience −0002 −0013

400195 400285Collaboration −0036 0036

400875 410295Age 0001 −0001

400035 400045

R2 0040 —Adjusted R2 0028 —F -statistics 3028∗∗∗ —Sigma — 2066∗∗∗

AIC — 199.63N 43 43

Notes. AIC, Akaike information criterion; DV, dependent variable.aThis is an ordinary least squares regression. The dependent variable is

overconfidence. Only participants in the OTH condition are included in theanalysis.

bThis is a Tobit regression. The dependent variable is investment level.Only participants in the OTH condition are included in the analysis.

∗Significant at p = 0010; ∗∗significant at p = 0005; ∗∗∗significant at p =

0001.

ReferencesAbelson R (1986) Beliefs are like possessions. J. Theory Soc. Behav.

16(3):223–250.Adams RB Jr, Nalini AC, Neil M, Kleck RE (2006) Emotional expres-

sions forecast approach-avoidance behavior. Motivation Emotion30(2):179–188.

Andrews D (2011) 2 reasons you can’t come up with a businessidea. Tropical MBA (blog), August 13, http://www.tropicalmba.com/how-to-find-a-business-idea.

Åstebro T (2003) The return to independent invention: Evidence ofunrealistic optimism, risk seeking or skewness loving? Econom.J. 113(484):226–239.

Becker GM, DeGroot MH, Marschak J (1964) Measuring utility bya single-response sequential method. Behav. Sci. 9(3):226–232.

Bénabou R, Tirole J (2002) Self-confidence and personal motivation.Quart. J. Econom. 117(3):871–915.

Bénabou R, Tirole J (2003) Intrinsic and extrinsic motivation. Rev.Econom. Stud. 70(3):489–520.

Busenitz LW, Barney JB (1997) Differences between entrepreneursand managers in large organizations: Biases and heuristics instrategic decision-making. J. Bus. Venturing 12(1):9–30.

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

Cohen WM, Levinthal DA (1989a) Innovation and learning: Thetwo faces of R&D. Econom. J. 99(397):569–596.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.

Hooshangi and Loewenstein: Impact of Idea Generation and Potential Appropriation on EntrepreneurshipManagement Science, Articles in Advance, pp. 1–19, © 2016 INFORMS 19

Cohen WM, Levinthal DA (1989b) The implications of spilloversfor R&D investment and welfare: A new perspective. Al Link,Smith K, eds. Advances in Applied Microeconomics: The FactorsAffecting Technological Change, Vol. 5 (JAI Press, Greenwich, CT),29–45.

Cooper AC, Woo CY, Dunkelberg WC (1988) Entrepreneurs’ per-ceived chances for success. J. Bus. Venturing 3(2):97–108.

Dawes RM (1989) Statistical criteria for establishing a truly falseconsensus effect. J. Experiment. Soc. Psych. 25(1):1–17.

Gans JS, Stern S (2003) The product market and the market forideas: Commercialization strategies for technology entrepre-neurs. Res. Policy 32(2):333–350.

Granovetter M (1995) The economic sociology of firms and entre-preneurs. Portes A ed. The Economic Sociology of Immigration(Russell Sage Foundation, New York), 128–165.

Hoch SJ (1987)Perceived consensus and predictive accuracy: Thepros and cons of projection. J. Personality Soc. Psych. 53(2):221–234.

Hodgkinson GP, Bown NJ, Maule AJ, Glaister KW, Pearman AD(1999) Breaking the frame: An analysis of strategic cognitionand decision making under uncertainty. Strategic Management J.20(10):977–985.

Hoelzl E, Loewenstein G (2005) Wearing out your shoes to preventsomeone else from stepping into them: Anticipated regret andsocial takeover in sequential decisions. Organ. Behav. HumanDecision Processes 98(1):15–27.

Hsee CK, Loewenstein G, Blount S, Bazerman MH (1999) Pref-erence reversals between joint and separate evaluations ofoptions: A review and theoretical analysis. Psych. Bull. 125(5):576–590.

Keh HT, Foo MD, Lim BC (2002) Opportunity evaluation underrisky conditions: The cognitive processes of entrepreneurs.Entrepreneurship Theory Practice 27(2):125–148.

Klepper S (2007) Disagreements, spinoffs, and the evolution ofDetroit as the capital of the U.S. automobile industry. Manage-ment Sci. 53(4):616–631.

Klepper S, Thompson P (2010) Disagreements and intra-industryspinoffs. Internat. J. Indust. Organ. 28(5):526–538.

Kruger J (1999) Lake Wobegon be gone! The “below-average effect”and the egocentric nature of comparative ability judgments.J. Personality Soc. Psych. 77(2):221–232.

Kruger J, Dunning D (1999) Unskilled and unaware of it: How diffi-culties in recognizing one’s own incompetence lead to inflatedself-assessment. J. Personality Soc. Psych. 77(6):1121–1134.

Kruger J, Wirtz D, Van Boven L, Altermatt TW (2004) The effortheuristic. J. Experiment. Soc. Psych. 40(1):91–98.

Lieberman MB (2007) Did first-mover advantage survive the dot-com crash? Working paper, Anderson School of Management,University of California, Los Angeles, Los Angeles.

Lieberman MB, Asaba S (2006) Why do firms imitate each other?Acad. Management Rev. 31(2):366–385.

Loomes G, Sugden R (1982) Regret theory: An alternative theory ofrational choice under uncertainty. Econom. J. 92(386):805–824.

Malmendier U, Tate G (2005) CEO overconfidence and corporateinvestment. J. Finance 60(6):2661–2700.

Masnick M (2008) Ideas are easy, execution is difficult. Techdirt(May 15), https://www.techdirt.com/articles/20080514/0336421112.shtml.

Moore DA, Cain DM (2007) Overconfidence and under-confidence:When and why people underestimate (and overestimate) thecompetition. Organ. Behav. Human Decision Processes 103(2):197–213.

Moore DA, Healy PJ (2008) The trouble with overconfidence. Psy-chological Rev. 115(2):502–517.

Moore DA, Small DA (2007) Error and bias in comparative judg-ment: On being both better and worse than we think we are.J. Personality Soc. Psych. 92(6):972–989.

Norton MI, Mochon D, Ariely D (2012) The IKEA effect: Whenlabor leads to love. J. Consumer Psych. 22(3):453–460.

Salovey P, Rodin J (1984) Some antecedents and consequencesof social-comparison jealousy. J. Personality Soc. Psych. 47(4):780–792.

Sarason Y, Dean T, Dillard JF (2006) Entrepreneurship as the nexusof individual and opportunity: A structuration view. J. Bus.Venturing 21(2):285–305.

Simmons JP, Nelson LD, Simonsohn U (2011) False-positive psy-chology: Undisclosed flexibility in data collection and analy-sis allows presenting anything as significant. Psych. Sci. 22(11):1359–1366.

Simonsohn U, Karlsson N, Loewenstein G, Ariely D (2008) Thetree of experience in the forest of information: Overweightingexperienced relative to observed information. Games Econom.Behav. 62(1):263–286.

Spence MA (1984) Cost reduction, competition, and industry per-formance. Econometrica 52(1):101–121.

Tesser A (1988) Toward a self-evaluation maintenance model ofsocial behavior. Berkowitz L, ed. Advances in Experimental SocialPsychology, Vol. 21 (Academic Press, San Diego), 181–227.

Tobin J (1958) Estimation of relationships for limited dependentvariables. Econometrica 26(1):24–36.

Zeelenberg M (1999) Anticipated regret, expected feedback, andbehavioral decision making. J. Behav. Decision Making 12(2):93–106.

Zeelenberg M, Beattie J, van der Pligt J, de Vries NK (1996) Con-sequences of regret aversion: Effects of expected feedback onrisky decision making. Organ. Behav. Human Decision Processes65(2):148–158.

Ziegler CA (1985) Innovation and the imitative entrepreneur.J. Econom. Behav. Organ. 6(2):103–121.

Dow

nloa

ded

from

info

rms.

org

by [

128.

237.

144.

175]

on

21 J

anua

ry 2

017,

at 1

0:08

. Fo

r pe

rson

al u

se o

nly,

all

righ

ts r

eser

ved.