should i quit my day job?: a hybrid path to entrepreneurship
TRANSCRIPT
SHOULD I QUIT MY DAY JOB?: A HYBRID PATH TOENTREPRENEURSHIP
JOSEPH RAFFIEEUniversity of Wisconsin–Madison
JIE FENGUniversity of Wisconsin–Madison
Research suggests that the risk and uncertainty associated with entrepreneurial activ-ity deters entry and contributes to the high rates of new business failure. In this study,we examine how the ability to reduce these factors by means of hybrid entrepreneur-ship—the process of starting a business while retaining a “day job” in an existingorganization—influences entrepreneurial entry and survival. Integrating insights fromreal options theory with logic from the individual differences literature, we hypothe-size and find that individuals who are risk averse and have low core self-evaluationare more likely to enter hybrid entrepreneurship relative to full-time self-employment.In turn, we argue and find that hybrid entrepreneurs who subsequently enter full-timeself-employment (i.e., quit their day job) have much higher rates of survival relative toindividuals who enter full-time self-employment directly from paid employment. Add-ing support to our theory that the survival advantage is driven by a learning effect thattakes place during hybrid entrepreneurship, we find that the decrease in exit hazardis stronger for individuals with prior entrepreneurial experience. Taken together, ourfindings suggest that individual characteristics may play a greater role in determiningthe process of how (rather than if) entrepreneurial entry occurs, and that the processof how entrepreneurial entry transpires has important implications for new businesssurvival.
Research indicates that entrepreneurial activityis a key driver of economic growth, but only ifentrepreneurial entrants are able to avoid early ex-odus (Santarelli & Vivarelli, 2007). Not surpris-ingly, therefore, understanding the determinants ofentrepreneurial entry and dynamics of venture sur-vival has attracted the interest of numerous organ-izational scholars (e.g., Elfenbein, Hamilton, &Zenger, 2010; Evans & Leighton, 1989; Geroski,Mata, & Portugal, 2010; Ozcan & Reichstein, 2009;Patel & Thatcher, 2012). As evidenced by the highfrequency of new business failure (Shane, 2003),understanding entry implies explaining why some
individuals opt to start businesses despite the riskyand uncertain returns associated with doing so(Kihlstrom & Laffont, 1979). Likewise, understand-ing survival entails identifying how and why someentrepreneurs are able to overcome these risks tosurvive (Santarelli & Vivarelli, 2007). As such, pin-pointing ways in which the risk and uncertaintyassociated with entrepreneurship can be managedor reduced should offer further insight regardinghow these processes unfold (see Folta, 2007).
One such way is through hybrid entrepreneur-ship—the process of initiating a business whilesimultaneously remaining employed for wages(Folta, Delmar, & Wennberg, 2010). By launching abusiness while retaining their “day job,” hybridentrepreneurs implicitly reduce (or eliminate) theopportunity cost (i.e., earnings from paid employ-ment) associated with starting the venture (Folta,2007; Folta et al., 2010). As such, by reducing whatis put “at risk,” starting a business via hybrid en-trepreneurship is inherently less risky than doingso full time. Recognizing this, scholars have re-cently noted that hybrid entrepreneurs represent a
We are deeply grateful to our editor Gerard George andthree anonymous AMJ reviewers for their insightful com-ments and guidance throughout the review process. Wealso wish to thank Russ Coff, Barry Gerhart, Randy Dun-ham, Jon Eckhardt, Ken Kavajecz, Phil Kim, and HartPosen for their helpful feedback and suggestions. Thefirst author graciously acknowledges financial supportprovided by the Robert W. Pricer PhD student fellowshipat the University of Wisconsin-Madison. Of course, allerrors remain our own.
936
� Academy of Management Journal2014, Vol. 57, No. 4, 936–963.http://dx.doi.org/10.5465/amj.2012.0522
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s expresswritten permission. Users may print, download, or email articles for individual use only.
significant and growing component of total entre-preneurial activity (Burke, FitzRoy, & Nolan, 2008;Folta et al., 2010; Petrova, 2012). According to theU.S. Bureau of Labor Statistics, in 2011, roughly10 percent of self-employed workers were also em-ployed by existing firms. Going a step further, re-search shows that many hybrid entrepreneurs ulti-mately decide to commit to their ventures full time(Folta et al., 2010). Indeed, anecdotal evidence in-dicates that some of the world’s most innovativeand successful entrepreneurs started their compa-nies as hybrid entrepreneurs. For example, SteveWozniak remained an employee at Hewlett–Pack-ard long after co-founding Apple (Wozniak &Smith, 2006), Pierre Omidyar launched eBay whileworking for the software development companyGeneral Magic (Cohen, 2002), and, with the help ofinvestors, Henry Ford founded the Detroit Automo-bile Group while employed by the Edison Illumi-nating Company (Ford & Crowther, 1922). In 1997,20 percent of CEOs on Inc. magazine’s 500 fastest-growing private companies list indicated that theycontinued to work a paying job long after foundingtheir organization (Inc. staff, 1997). Yet, despitethese observations, extant entrepreneurship theorylargely assumes that entrepreneurial entry is anall-or-nothing phenomenon (Folta et al., 2010),and, because empirical testing is driven by theory,the treatment of labor force status as a mutuallyexclusive dichotomy is generally considered “un-controversial” in the literature (Sørensen & Fas-siotto, 2011: 1,323).1
In this article, we depart from this trend and addto an emerging stream of literature (e.g., Burke etal., 2008; Folta et al., 2010; Petrova, 2012) by con-sidering the theoretical implications of hybrid en-trepreneurship for theories of entrepreneurial entryand survival. For example, extant theory suggestsentrepreneurs have high tolerances for risk (Kihl-strom & Laffont, 1979) and/or perceive less risk dueto greater confidence in their abilities (Moore,
Oesch, & Zietsma, 2007). However, considering hy-brid entrepreneurship reduces (or eliminates) theneed for risk-bearing when starting a business,these theories may not adequately explain entry.Likewise, once an individual enters hybrid entre-preneurship, the uncertainty surrounding the fu-ture returns and viability of the business lessens(Folta et al., 2010). As a result, hybrids that opt toenter full-time self-employment do so under con-ditions of greater certainty (relative to those whoenter directly from paid employment). However,despite numerous theoretical explanations (e.g.,Geroski et al., 2010; Santarelli & Vivarelli, 2007),scholars have yet to consider how staged entry intofull-time self-employment via the pathway of hy-brid entrepreneurship influences venture survival.
To reconcile these issues, we turn to logic fromreal options theory (Trigeorgis, 1996). Conceptual-izing hybrid entrepreneurship as analogous to theestablishment of a real option—a small initial com-mitment that creates the right, but not the obliga-tion, to subsequently commit full time to the ven-ture (Folta et al., 2010; Wennberg, Folta, & Delmar,2006)—we integrate insights from real options the-ory with logic from the individual differences lit-erature (Funder, 2001) to theorize that risk-averseand less confident individuals will be more likelyto enter hybrid entrepreneurship relative to full-time self-employment. In turn, using real optionslogic that emphasizes the benefits of learning fromsmall investments (Roberts & Weitzman, 1981), weargue staged entry into full-time self-employmentthrough hybrid entrepreneurship will relate posi-tively to venture survival. Finally, emphasizing theinherent heterogeneity among real options decisionmakers (e.g., Barnett, 2008), we posit that individ-ual characteristics (cognitive ability and entrepre-neurial experience) that influence the hybrid entre-preneur’s ability to evaluate the prospects of theirventure during hybrid entrepreneurship will mod-erate this relationship.
The present study makes several contributions.First, acknowledging that hybrid entrepreneurshipimplicitly reduces the risk associated with startinga business (Folta et al., 2010), our study suggests forthe first time that individual characteristics per-taining to risk preferences and risk perception mayinfluence how rather than if entrepreneurial entryoccurs. Second, despite a substantial literature onventure survival (Santarelli & Vivarelli, 2007) andmounting evidence that entry into full-time self-employment is endogenous to hybrid entrepreneur-ship (Folta et al., 2010), the current study is the first
1 Hybrid entrepreneurship is common in academic en-trepreneurship (e.g., Jain, George, & Maltarich, 2009),where academics often form firms to commercialize re-search while retaining their academic position (as op-posed to exiting the institution to pursue the venture fulltime) (Nicolaou & Birley, 2003a). However, perhaps dueto the uniqueness of the university setting, few studieshave accounted for this distinction theoretically or em-pirically (see Nicolaou & Birley, 2003a, 2003b for notableexceptions). Ultimately, this broadens the contributionsof our study.
2014 937Raffiee and Feng
to argue and show that entering full-time self-em-ployment incrementally via hybrid entrepreneur-ship increases the odds of survival. Third, we ex-tend recent research (Autio & Acs, 2010; Folta et al.,2010; O’Brien, Folta, & Johnson, 2003) by testingpredictions from real options theory using individ-ual characteristics as explanatory and moderatorvariables. By doing so, our work contributes to thereal options literature by theoretically arguing andempirically demonstrating that individual charac-teristics influence both the likelihood and efficacyof real options reasoning.
THEORY AND HYPOTHESES
Motivated by the importance of the phenome-non, scholars from a variety of backgrounds,including sociology (Sørensen, 2007), economics(Hamilton, 2000), and psychology (Hmieleski &Baron, 2009), have sought to explain entrepreneur-ial activity. While the multidisciplinary approachhas greatly contributed to our understanding of theentrepreneurial process (Shane, 2003), it has alsoled to areas of disagreement, particularly with re-gards to how entrepreneurship should be concep-tualized and defined (see Sørensen & Fassiotto,2011). In this article, our interest is in understand-ing entrepreneurship in terms of labor force status(i.e., self-employment versus paid employment).Defined this way, entrepreneurship encompassesthe entire array of entrepreneurial activity, rangingfrom the small self-employed sole proprietor to thelarge venture-backed start-up, making the implica-tions of our theory generalize to all entrepreneurialentrants, not just those with a specific type or via-ble entity (Yang & Aldrich, 2012). Accordingly, toremain consistent with prior hybrid entrepreneur-ship studies (e.g., Folta et al., 2010), we use theterms “entrepreneurship” and “self-employment”interchangeably.
Despite the definitional divergence, there exists arelative consensus within the literature that entre-preneurial activity involves risk and uncertainty(Folta, 2007; McMullen & Shepherd, 2006). For ex-ample, even if entrepreneurs are able to shift allfinancial risk to other actors (e.g., investors), “inevery case, the entrepreneur risks the opportunitycosts associated with starting the venture” (Folta,2007: 98). Thus, by entering self-employment, it istypically assumed that individuals transform theirsource of income from a relatively safe asset (i.e.,earnings in paid employment) into a more riskyasset (i.e., returns to self-employment), as reflected
by the high dispersion in self-employed earnings(Hamilton, 2000) and likelihood of business failure(Shane, 2003). In turn, the risk associated with thistransformation has been argued to be a key deter-rent of entrepreneurial entry (Amit, Muller, & Cock-burn, 1995). However, this logic overlooks the factthat individuals can circumvent this trade-off bymeans of hybrid entrepreneurship (Folta et al.,2010). Hence, by starting a business without quit-ting one’s day job, hybrid entrepreneurs need notput their “certain” earnings from paid employmentat risk. Accordingly, recognizing that hybrid entre-preneurship represents a smaller-scale and lessrisky (i.e., less sunk commitment) entrepreneurialentry path, scholars at the forefront of the hybridentrepreneurship literature have argued that realoptions theory is a theoretical perspective well-suited to provide insights into the hybrid phe-nomenon (Folta et al., 2010; Wennberg et al.,2006). In the next section we briefly reviewreal options theory and its link to hybridentrepreneurship.
Real Options Theory and HybridEntrepreneurship
Real options theory is a framework for makinginvestments in risky and uncertain contexts (Dixit& Pindyck, 1994). In real options theory, the “op-tion” typically refers to a small initial investmentthat creates the possibility, but not the obligation,to make subsequent larger investments (McGrath,1997). A key benefit of investing in real options isthat it allows decision makers to gather informationand learn, thereby reducing the uncertainty sur-rounding the investment, prior to making largercommitments (Majd & Pindyck, 1987; Roberts& Weitzman, 1981; Weitzman, Newey, & Rabin,1981). Should the information generated from theoption appear favorable (unfavorable), subsequentcommitments can be made (ceased). Thus, becausethe potential upside gain has no ceiling, but thedownside loss (i.e., risk) is limited to the cost of theoption, real options become more valuable in situ-ations characterized by high uncertainty (i.e., highvariance in returns) (McGrath, 1997). Accordingly,while real options theory predicts that high uncer-tainty dissuades large commitments, it also sug-gests that it can encourage small commitments inthe form of real options. For instance, O’Brien et al.(2003) find the likelihood of full-time entrepre-neurial entry is lower in industries characterizedby greater uncertainty. Ziedonis (2007) concludes
938 AugustAcademy of Management Journal
that firms interested in university technology aremore likely to purchase an option contract (smallcommitment) prior to committing to a licensingagreement (large commitment) when the uncer-tainty embedded in the technology is high.
The commonalities between real options and hy-brid entrepreneurship are twofold. First, as notedby O’Brien et al. (2003), it is difficult to surmise acontext in which risk and uncertainty are moresalient than entrepreneurship. Second, much like areal option, hybrid entrepreneurship allows indi-viduals to start a business on a smaller scale withless sunk costs and downside risk (Folta et al.,2010). Empirical findings support this connection.For example, in accord with real options theory,Wennberg et al. (2006) find that individuals aremore likely to enter hybrid entrepreneurship asopposed to full-time self-employment in more un-certain industries. Similarly, noting that hybrid en-trepreneurship is akin to a real option to invest,Folta et al. conclude that many hybrid entrepre-neurs ultimately enter full-time self-employment,indicating that hybrid entrepreneurship is oftenused as a means to “test the entrepreneurial waters”prior to committing to the venture full time (Foltaet al., 2010: 253).2 Hence, as these studies demon-strate, real options theory has the power to explainboth entrepreneurial entry and entrepreneurial out-comes. On the one hand, real options theory sug-gests that the ability to reduce the risk associatedwith starting a business can entice entrepreneurialentry (Lee, Peng, & Barney, 2007). On the otherhand, the learning benefits associated with hybridentrepreneurship (Roberts & Weitzman, 1981) im-ply that starting a business through a pathway ofpaid employment ¡ hybrid ¡ full-time self-em-ployment should be associated with positive out-comes. As a result, real options theory provides asingle unifying framework suitable to explain theentire entrepreneurial process. However, given itsemphasis on the effects of investment risk and un-certainty, real options theory is built on the as-sumption of a risk-neutral and preference-free de-cision maker (Dixit & Pindyck, 1994). Yet, in mostreal-world scenarios, these assumptions are unre-
alistic, leading to a scholarly push for researchers toincorporate decision maker characteristics and pref-erences into real options theory (e.g., Barnett, 2008).For instance, O’Brien et al. argue it is time to “shift thefocus of research away from macroeconomic mea-sures and towards using firm-specific (or even indi-vidual-specific) determinants of entry thresholds”(O’Brien et al., 2003: 515, emphasis added). We do soin this study, beginning by using individual charac-teristics that influence individual thresholds for riskand uncertainty to generate real options predictionsregarding entrepreneurial entry.
Risk Aversion, Core Self-Evaluation, and Entryinto Hybrid Entrepreneurship
Given that entrepreneurial activity is generallyassumed to include the bearing of risk and uncer-tainty, the notion that entrepreneurs are comfort-able with risk has a long theoretical tradition in theacademic literature (Kihlstrom & Laffont, 1979;Knight, 1921). Nevertheless, empirical evidence re-garding risk preferences and entrepreneurial entryis largely mixed (see Brockhaus, 1980; Cramer, Har-tog, Jonker, & Van Praag, 2002; Miner & Raju, 2004).The majority of studies, however, do not theoreti-cally or empirically account for the fact that entryinto hybrid entrepreneurship inherently involvesless downside risk. As a result, it is possible thatrisk aversion influences the process of how, ratherthan if, an individual decides to start a new busi-ness. Along these lines, operating on the logic thatestablished businesses are less risky than start-upventures, Block, Thurik, van der Zwan, and Walter(2013) find that risk-averse individuals are morelikely to purchase an existing business rather thanstart a new venture from scratch. Likewise, by ac-knowledging the fact that hybrid entrepreneurshipallows individuals to reduce what is put at risk(i.e., earnings from paid employment) when start-ing a new venture (Folta et al., 2010), logic fromreal options theory can provide a more nuancedpicture regarding the relationship between riskaversion and entrepreneurial entry.
A central prediction of real options theory is thathigh levels of risk and uncertainty dissuade largecommitments (O’Brien et al., 2003). However, the-ories of risk aversion highlight heterogeneity withregards to individual comfort thresholds for riskand uncertainty. As a result, we expect that invest-ment behavior of a risk-averse individual should besimilar to the behavior of a risk-neutral decisionmaker facing high levels of exogenous uncertainty.
2 Folta et al. (2010) also conclude that individuals mayenter hybrid entrepreneurship to generate non-monetarybenefits, but found no indication that people becomehybrids to earn supplemental income. Similar findingswere reported by Petrova (2012), who concluded thatpart-time entrepreneurs are not impacted by financialconstraints.
2014 939Raffiee and Feng
Thus, extending insights from real options theoryto incorporate risk preferences would suggest that,in accordance with traditional equilibrium modelsof risk aversion and self-employment (Kihlstrom &Laffont, 1979), risk-averse individuals should beless willing to make the large commitments associ-ated with entry into full-time self-employment. Inother words, risk aversion effectively increases thevalue associated with deferring full-time entry (Hu-gonnier & Morellec, 2007). However, as uncertaintyrises, so too does the value of holding a real option(McGrath, 1997). Thus, while real options theorysuggests individuals with high risk aversion shouldbe less likely to make the large commitments asso-ciated with direct entry into full-time self-employ-ment, they should be more willing to make smallercommitments associated with entry into hybrid en-trepreneurship. Indeed, by entering hybrid entre-preneurship, risk averse individuals can start abusiness and reduce what is at risk/the amount ofrisk-bearing. Thus, we suggest the following:
Hypothesis 1. Individuals with higher riskaversion are more likely to enter hybridentrepreneurship in comparison to full-timeself-employment.
In addition to risk aversion, a number of otherindividual characteristics have been argued to in-fluence entrepreneurial entry. For instance, build-ing on the notion that entrepreneurs are highlyconfident (Knight, 1921), researchers have shownthat internal locus of control (Evans & Leighton,1989), self-efficacy (Zhao, Seibert, & Hills, 2005),and emotional stability (Zhao, Seibert, & Lumpkin,2010) relate positively to entrepreneurial entry.However, various elements of personality are oftentreated as entirely separate constructs, with little (ifany) discussion regarding the interrelationshipsamong traits or acknowledgement that related ele-ments of personality may all be tapping the samehigher-order construct (Judge, Erez, Bono, & Thore-sen, 2003; Judge, Locke, & Durham, 1997). To thatend, our study focuses on core self-evaluation(CSE), a broad dispositional trait manifested byfour elements of personality: self-esteem, general-ized self-efficacy, locus of control, and emotionalstability (Judge et al., 1997).
Independent of context and time, CSE is theo-rized to reflect the fundamental appraisals individ-uals make about themselves, their capabilities, andtheir competence (Judge et al., 1997). For example,Chang, Ferris, Johnson, Rosen, and Tan write thatCSE is “proposed to be the most fundamental eval-
uations people hold, reflecting a baseline appraisalthat is implicit in all other beliefs and evaluations”(Chang et al., 2012: 83, emphasis added). Thus, CSErepresents an individual’s overarching generalevaluations, not specific evaluations regarding anyparticular context (e.g., organizational, entrepre-neurial, etc.). Accordingly, research has demon-strated CSE to have predictive validity regarding awide variety of work- and life-related outcomes(Chang et al., 2012). Anchored in a real optionsframework, we add to this literature by developingtheory to explain how CSE influences real optionsreasoning and the process of entrepreneurial entry.
Given that individuals high in CSE are confidentin their ability to successfully complete tasks andcontrol their environment (Judge et al., 2003), theseindividuals should be less deterred by the risk anduncertainty associated with starting a business.Stated differently, because individuals high in CSEare confident in their capabilities, they should per-ceive entering self-employment as less risky anduncertain (i.e., perceive less variance in outcomes).In contrast, individuals with low CSE tend to beunsure of themselves and their capabilities, makingthem more likely to perceive entering self-employ-ment as a high-risk endeavor (i.e., perceive morevariance in outcomes). Accordingly, a predisposi-tion to perceive entry into self-employment as morerisky and uncertain would, in effect, raise the valueassociated with using an option approach. Thus,consistent with Caves (1998), who argued that lessconfident entrepreneurs will tend to start theirbusinesses on a smaller scale, our logic suggeststhat individuals with low CSE who enter self-em-ployment will be more likely to do so incremen-tally via hybrid entrepreneurship.
The upper echelons literature provides somesupport for our reasoning. For example, Hiller andHambrick (2005) argue that CEOs high in CSE aremore likely to launch large-scale, quantum strategicinitiatives, while CEOs low in CSE favor a smaller,incremental approach. Chatterjee and Hambrick(2007) found support for a positive relationshipbetween the CSE of a firm’s CEO and the likelihoodthat the firm pursues entrepreneurial opportuni-ties. Simsek, Heavey, and Veiga (2010) concludethat CEO CSE is positively related to a firm’sentrepreneurial orientation. As we detailed above,we expect a similar relationship regarding entre-preneurial entry. Accordingly, we suggest thefollowing:
940 AugustAcademy of Management Journal
Hypothesis 2. Individuals with low core self-evaluation are more likely to enter hybridentrepreneurship in comparison to full-timeself-employment.
Staged Entry into Full-Time Self-Employmentand Survival
The previous section uses insights from real op-tions theory and the individual differences litera-ture to predict hybrid entry. However, for manyindividuals, entry into hybrid entrepreneurshiprepresents just the first step on the path to full-timeself-employment. For example, Folta et al. (2010)argue that entry into full-time self-employment isendogenous to hybrid entrepreneurship, conclud-ing that hybrid entrepreneurs are thirty-eight timesmore likely than wage earners to enter full-timeself-employment. A prediction from real optionstheory is that hybrid entrepreneurs will enter full-time self-employment only when they perceive theoption to do so to be “in the money” (Trigeorgis,1996). Indeed, a key benefit of real options is theability to postpone decision making until the un-certainty surrounding the investment has been re-solved (Dixit & Pindyck, 1994). In the context ofentrepreneurship, the uncertainty resolved duringthe option period (i.e., hybrid entrepreneurship) istypically endogenous—meaning that it can be re-duced by actions of the entrepreneur (see Folta,1998, for a detailed discussion). In other words, byentering hybrid entrepreneurship, individuals areable to learn about their venture, thereby reducingthe uncertainty surrounding its prospects, prior todeciding if they want to increase commitment(Roberts & Weitzman, 1981). As such, because in-dividuals enter self-employment after consideringthe relative costs and benefits (Muller & Arum,2004), absent positive information, hybrid entre-preneurs should see no reason to forego the benefitsassociated with their job in paid employment(Becker, 1960). The underlying logic is driven bythe fact that real options entail less sunk cost. As aresult, options that do not yield favorable informa-tion can be quickly abandoned while those thatappear promising can be exercised (O’Brien &Folta, 2009). Specifically, we focus on informationhybrid entrepreneurs accrue that reduces the un-certainty surrounding two components of a sustain-able business: (1) the quality of the business idea
and (2) the entrepreneur’s skills, capabilities, andfit within the entrepreneurial context.3
First, hybrid entrepreneurs benefit from the abilityto learn about the quality, potential, and feasibility oftheir business idea. Indeed, prior to the introductionof a new product (or service), it is difficult to knowwith certainty if one will be able to physically pro-duce the product or if the product will meet thecharacteristics of market demand (Autio, Dahlander,& Frederiksen, 2013). Over time, however, the uncer-tainty surrounding the value and feasibility of theventure lessens, making the prospects of the businessmore salient (Sorenson & Stuart, 2001). In other cases,business ideas may be difficult to fully understandwithout actually “starting the commercialization pro-cess” (George & Bock, 2012: 69). As such, the onlyway to determine the value and feasibility of theseideas is to go forth and attempt to exploit them.
Second, hybrid entrepreneurs benefit from theability to learn about their entrepreneurial skills, ca-pabilities, and fit within the entrepreneurial context(Folta et al., 2010). Indeed, a lack of fit betweenfounder and company is a major reason for new busi-ness failures (Holmes & Schmitz, 1995). Much likedetermining the prospects of a business idea, only bystarting the business are individuals able to fully eval-uate if they have the necessary skills required to runthe business (Jovanovic, 1982). However, even if thehybrid entrepreneur does not possess these skills exante, hybrid entrepreneurship provides a low-risksetting where the necessary capabilities can belearned prior to committing to the venture full time.Furthermore, hybrid entrepreneurship provides a re-alistic preview of life as an entrepreneur, illuminatingthat many of the glamorous portrayals of entrepre-neurship are largely myths (Shane, 2008) and thatbeing self-employed is a time-consuming and chal-lenging process.
Given that hybrid entrepreneurs have no obliga-tion to enter full-time self-employment (McGrath,1997), that the cost of abandoning the venture hasless sunk cost (O’Brien & Folta, 2009), and thathybrid entrepreneurs learn about the merits of theirventure idea, skills, and entrepreneurial fit prior to
3 As an aside, it is crucial to note that the learningbenefits associated with hybrid entrepreneurship applyto all hybrid entrepreneurs, including those with no ex-ante intention to enter full-time self-employment (Foltaet al., 2010). For example, when Omidyar founded eBay,he had no intention of ever quitting his day job. However,after a positive market reaction, he felt he had no choicebut to focus on eBay full time (Cohen, 2002).
2014 941Raffiee and Feng
committing to the business full time (Roberts &Weitzman, 1981), real options theory suggests thathybrid entrepreneurs who “exercise” the optionand enter full-time self-employment have reason tobelieve their business is sustainable and holdspromise (i.e., the option is in the money). Accord-ingly, we suggest the following:
Hypothesis 3. Individuals who transition intofull-time self-employment in a staged entryprocess via hybrid entrepreneurship will sur-vive longer than individuals who transitioninto full-time self-employment directly frompaid employment.
Moderators of the Staged Entry–SurvivalRelationship
In the previous section, we use logic from realoptions theory to argue that hybrid entrepreneurs ex-ercise the option to enter full-time self-employmentwhen they believe the option to be in the money.However, unlike financial options where proper ex-ercise thresholds are intuitive, determining if realoptions are in the money is subjective, less straight-forward, and heavily reliant on decision-maker in-sight (e.g., Barnett, 2008). As such, we expect that thesurvival benefit associated with staged entry into full-time self-employment will vary with the hybrid en-trepreneur’s ability to make effective assessmentsregarding the venture’s potential. Specifically,we focus on two characteristics that influencethis ability: (1) cognitive ability (i.e., general in-telligence) (Schmidt & Hunter, 2004) and (2) spe-cific knowledge accumulated through prior (en-trepreneurial) experience (Cohen & Levinthal,1990; Zahra & George, 2002).
Cognitive Ability
Cognitive ability is conceptualized as the generalability to think abstractly, learn from experiences,comprehend surroundings, and “figure things out”(Lubinski, 2004). Indeed, not all individuals areable to learn, process, and apply new knowledgeequally (Hunter, 1986). Empirical research hasdemonstrated that individuals with high general in-telligence are better able to assimilate information toapply it in new situations (Jensen, 1998) and acquirenew skills (Gottfredson, 1997). As a result, Schmidtand Hunter (2004) argue that general mental ability isthe primary factor responsible for turning experienceinto knowledge and the single most important attri-bute explaining variance in job performance.
Surprisingly, however, studies regarding the rela-tionship between intelligence and entrepreneurial ac-tivity are rare (Baum & Bird, 2010; Vinogradov &Kolvereid, 2010). Nevertheless, scholars have longhinted that cognitive ability plays an important rolein entrepreneurial process. For example, Knight(1921) argued that intellectual ability would lead tothe identification of more valuable opportunities.Similarly, Vinogradov and Kolvereid (2010: 153) sug-gest that, since intelligence represents a “broader anddeeper capability for comprehending surroundings,”it should be particularly useful when evaluating newopportunities. Along these lines, while scholars haveargued that creativity is important for generating busi-ness ideas, analytical intellectual ability is most im-portant when assessing an idea’s merits and potential(Baum & Bird, 2010). For example, entrepreneurialresearchers have noted that intelligence increases theability to see value embedded within new informa-tion (Shane, 2003), and that analytical ability is par-ticularly helpful when interpreting and making senseof complex information in an entrepreneurial setting(Baum & Bird, 2010). As a result, the benefit of cog-nitive ability should be particularly salient duringhybrid entrepreneurship, where hybrid entrepre-neurs accrue a wealth of information about their busi-ness that can be used to determine if the business isworth pursuing full time (i.e., if they should exercisethe option). Stated differently, because intelligenceincreases a hybrid entrepreneur’s ability to “analyzeand evaluate multiple and complex courses of ac-tion” (Baum & Bird, 2010: 399), we expect intelligenthybrid entrepreneurs to make better exercise deci-sions, being more likely to exercise the option when itis in the money and abandon it when it is not. Thus,we suggest the following:
Hypothesis 4. Cognitive ability moderates thepositive relationship between staged entry andfull-time self-employment survival such thatthe relationship is stronger for individuals withhigh cognitive ability.
Entrepreneurial Experience
The ability to assess, assimilate, and make senseof new information is also a function of an organi-zation or individual’s prior experience and stock ofknowledge (Cohen & Levinthal, 1990; Zahra &George, 2002). Given that repeat entrepreneurs areable to draw on their prior experiences foundingventures, scholars have posited that entrepreneur-ial experience should be particularly useful whenevaluating the prospects of a new business (Wright,
942 AugustAcademy of Management Journal
Westhead, & Sohl, 1998). For instance, Toft-Kehler,Wennberg, and Kim (2014) argue that experiencedentrepreneurs are able to use their knowledge re-garding past entrepreneurial ventures to make moreeffective connections and deduce inferences re-garding the prospects of a new venture. Similarly,the literature on entrepreneurial cognition suggestsexperienced entrepreneurs develop expert scriptsand knowledge structures allowing them to useinformation more effectively than inexperiencedentrepreneurs (Mitchell et al., 2007). Indeed, re-search suggests that experienced entrepreneursthink differently than novice entrepreneurs whenevaluating and assessing opportunities (Baron &Ensley, 2006; Ucbasaran, Westhead, & Wright,2009). For example, Baron and Ensley (2006) findthat experienced entrepreneurs emphasize moremundane characteristics indicating venture feasi-bility and the likelihood of positive financial re-turns, while novice entrepreneurs focus on charac-teristics reflecting a greater degree of novelty andexcitement.
Despite this evidence, empirical studies linkingentrepreneurial experience and venture survivalhave generally reported mixed findings (Delmar &Shane, 2006; Gimeno, Folta, Cooper, & Woo, 1997;Jørgensen, 2005). One explanation is that the rela-tionship is more complex than a simple main ef-fect. In other words, ex ante, the outcomes of thefounding process remain highly uncertain even forrepeat entrepreneurs (Aldrich, 1999). However,given the chance to amass information about theirventure through hybrid entrepreneurship, repeatentrepreneurs can utilize their knowledge regard-ing what worked and what did not when assessingthe new venture’s prospects (Toft-Kehler et al.,2014). Moreover, the opportunity characteristicsexperienced entrepreneurs look for when assessingthe quality of business ideas, such as positive cashflow, high margins, and the ability to quickly gen-erate revenue (Baron & Ensley, 2006), are moresalient and quantifiable once the business has ac-tually been started. Accordingly, by drawing onprior business experiences and focusing on quanti-fiable metrics that indicate business feasibility, weexpect experienced entrepreneurs to make moreeffective exercise decisions, exercising the optionto commit full time to the business when the optionis in the money and abandoning the option when itis not. Accordingly, we suggest the following:
Hypothesis 5. Entrepreneurial experiencemoderates the positive relationship between
staged entry and full-time self-employmentsurvival such that the relationship is strongerfor experienced entrepreneurs.
METHODS
Data
We use data from the National Longitudinal Sur-vey of Youth, 1979 cohort (NLSY79). The NLSY79survey is sponsored and directed by the U.S. Bu-reau of Labor Statistics and conducted by the Cen-ter for Human Resource Research at The Ohio StateUniversity. Interviews are conducted by the Na-tional Opinion Research Center at the University ofChicago. The data has been used by managementscholars to study issues such as self-employment(Schiller & Crewson, 1997), employee turnover(Lee, Gerhart, Weller, & Trevor, 2008), and careeroutcomes (Judge & Hurst, 2007, 2008). The NLSY79consists of a nationally representative sample of12,686 men and women aged between 14 and 22years when first surveyed in 1979. The cohort wasinterviewed annually until 1994, and bienniallythereafter.
Several features of the NLSY79 make it particu-larly attractive to test our hypotheses. First, it con-tains rich information on individual preferences,attitudes, and socioeconomic status. Second, thedata contain comprehensive employment historiesfor each participant. During each survey, partici-pants are allowed to report up to five jobs. For eachjob, the date (month/day/year) when the job beganas well as the date if/when the job ended is re-corded. Hence, by comparing job start and stopdates, we can determine if any participant held twojobs simultaneously (i.e., if a new job begins beforean existing job ends). This allows us to overcome amajor challenge when studying hybrid entrepre-neurship—the ability to identify true hybrids (Foltaet al., 2010).4 The NLSY79 codes jobs into the fol-
4 In many datasets used to study self-employment, laborforce status or income is reported on an annual basis.Therefore, in a given year, if the data indicate that a personwas employed in both paid and self-employment, it is un-clear if the person entered hybrid entrepreneurship or ifthey transitioned from paid employment to self-employ-ment sequentially. To overcome this issue, Folta et al.(2010) identified individuals as hybrids only if they re-ported the same paid job and same self-employed job fortwo consecutive years. Although conservative, a limitationof this approach is that individuals with short stints inhybrid entrepreneurship are potentially excluded.
2014 943Raffiee and Feng
lowing employment categories: government, pri-vate company, self-employed in own business,non-profit, and family business. We treat partici-pants who report being self-employed in their ownbusiness as entrepreneurs. We consider all partici-pants holding jobs not classified as self-employedto be in paid employment. Figure 1 provides anexample of the data structure and how we treatlabor status transitions.
Sample Construction
Since we hypothesize both the determinants ofentrepreneurial entry and factors influencingsurvival, testing our hypotheses required the con-struction of two samples. For each sample, weeliminated non-respondents to key questions: par-ticipants who worked fewer than 30 hours perweek and participants with cognitive ability belowthe 10th percentile. For all hypothesis tests, weused data from 1994 to 2008, representing a 14-yearsample window. We started analysis in 1994 be-cause one of our key predictor variables (risk aver-sion) was not available until 1993. To avoid com-plications due to left censoring, we followed therecommendations of Allison (1984) and droppedspells where participants began jobs (paid or self-employed) prior to 1994. Thus, we focus our anal-ysis on newly employed (self-employed) partici-pants, observing each participant as soon as they
become at risk to enter hybrid or full-time self-employment (Hypothesis 1 and Hypothesis 2) or atrisk to exit full-time self-employment (Hypothe-ses 3 to 5).
Since Hypotheses 1 and 2 are related to entrepre-neurial entry, we constructed a sample of partici-pants who were employed in a paid job and did nothold any self-employed jobs. The sample to testHypotheses 1 and 2 consisted of 5,299 unique par-ticipants, representing 31,919 paid job spells. Hy-potheses 3 to 5 relate to full-time self-employmentsurvival. To test these hypotheses, we constructeda sample consisting of self-employed participantswho held no additional paid jobs (i.e., fully self-employed). The sample to test Hypotheses 3 to 5consisted of 1,093 unique participants, represent-ing 2,198 full-time self-employed job spells.
Estimation Strategy
We use continuous survival analysis to test ourhypotheses. Survival analysis models the amountof time one “survives” before an event occurs. Akey advantage of survival analysis is the ability tohandle issues of right censoring, which occurswhen a study window ends prior to an event oc-curring or if a participant exits the sample for al-ternative reasons (Allison, 1984). Unless thesecases occur randomly, failure to statistically ac-count for them can threaten internal validity and
FIGURE 1Examples of Entry Paths into Self-Employment from Paid-Employmenta
a As indicated by the dashed line in Path 2, given the continuous nature of the data, few participants report starting a self-employedbusiness on the exact day they quit their job in paid employment (i.e., there is usually a gap between jobs). As such, in our primary analysis,we treated participants who started a self-employed business within a three-month window of exiting their paid job as entering full-timeself-employment. Our results are robust to shorter and longer windows.
944 AugustAcademy of Management Journal
biases estimates (Cook & Campbell, 1979). Survivalanalysis considers information up to the point ofcensoring, thereby minimizing such concerns (Al-lison, 1984).
To test Hypotheses 1 and 2, we used a compet-ing-risks framework. In a competing-risks model,individuals are assumed to be at risk to experiencea number of potential events. In the context of thisstudy, participants who hold paid jobs are at risk toenter hybrid entrepreneurship or full-time self-em-ployment. Therefore, we model the amount of timea participant survives at a given paid job prior toentering hybrid entrepreneurship (Path 1 in Figure1) or full-time self-employment (Path 2 in Figure 1).Participants who remained at their paid job at theend of the survey window, exited their paid job totake another paid job, or entered unemploymentwere treated as right censored (Path 3 in Figure 1).Once a participant entered self-employment (orwas censored), he or she was removed from the riskset until they began a new paid job, at which timethey again became at risk to enter self-employment.We estimate separate event-specific survival mod-els (hybrid versus full-time entry), thereby allowingus to test the equality of parameters across modelsvia the test statistic developed by Narendranathanand Stewart (1991).
To test Hypotheses 3 to 5, we used a single-riskframework to model the amount of time a partici-pant survives in a full-time self-employed job. Forparticipants who transitioned into full-time self-employment from hybrid entrepreneurship (i.e.,staged entry), survival time does not include thetime spent as a hybrid entrepreneur. Moreover,since research has demonstrated that entrepreneurswho hold a secondary paid job are able to persistfor longer periods in self-employment (Gimeno etal., 1997), we treated participants who began a sec-ondary paid job as if they exited full-time self-employment.5 Participants who remained in theirself-employed job at the end of the survey windowwere treated as right censored. Once a participantexited their full-time self-employed job, he or shewas removed from the risk set until they began anew full-time self-employed job, at which timethey re-entered the risk set.
We use Cox semiparametric proportional hazardsmodels to test our hypotheses. Since participantscan experience multiple employment (self-employ-
ment) spells, we used robust estimators to calculatestandard errors clustered by each participant (Lin &Wei, 1989) and the Efron method in cases of eventties. Cox proportional hazards models produceboth hazard ratios and regression coefficients. Ex-ponentiating the unstandardized regression coeffi-cient (using the formula: 100 * [е� � 1]) from a Coxmodel eases interpretation by producing thepercent change in risk of experiencing an eventassociated with a one-unit change of the predictorvariable.
When testing Hypotheses 3 to 5, we addressedthe possibility of selection effects (Shaver, 1998) byestimating a shared frailty model (Gutierrez, 2002),specifying the frailty to be shared among partici-pants who entered full-time self-employment fromhybrid entrepreneurship (i.e., staged entry). Theshared frailty captures the effects of unobservedcharacteristics common among individuals whotransitioned from hybrid entrepreneurship intofull-time self-employment, which would also influ-ence survival time (Allison, 2009; Song, 2010). Anadvantage of the shared frailty model is that itdoes not rely on the use of instruments (as Heck-man’s (1979) selection correction does), therebyavoiding the problem of identifying instrumentsthat properly satisfy theoretical assumptions (Pu-hani, 2000). The likelihood ratio test from theshared frailty model failed to reach statistical sig-nificance (p � .5), suggesting that unobserved het-erogeneity was not present (Gutierrez, 2002). Ac-cordingly, we report results without the frailty.
Measures
Independent variables. We followed Barsky,Juster, Kimball, and Shapiro (1997) and measuredrisk aversion using an index based on responses tothree hypothetical occupational income gambles(see Appendix). Participants were asked the in-come gamble questions in the 1993, 2002, 2004,and 2006 surveys. We constructed a time-varyingmeasure using the responses from each of the foursurveys. Specifically, we used responses from the1993 survey from 1994 to 2002, the 2002 responsesuntil 2004, the 2004 responses until 2006, and the2006 responses until 2008. Results were unchangedwhen we used the 1993 measure as a time-invarianttrait and when we limited our sample to the 2002–2006 surveys where risk aversion is updatedbiennially.
For core self-evaluation, we followed Judge andHurst (2007, 2008) and used 12 questions from the
5 Results were unchanged when the stop date of thefull-time self-employed job was used.
2014 945Raffiee and Feng
NLSY79 to measure CSE. Since the NLSY79does not include a measure of CSE, these questionswere selected because they reflect the 12 items onthe CSE scale developed by Judge et al. (2003). Themeasure demonstrates high construct validity, con-tent validity, discriminant validity, and reliability(see Judge & Hurst, 2007, 2008, for extensive scalevalidation procedures). CSE is time invariant.
We measured staged entry two ways. First, weused a dummy variable to indicate whether a par-ticipant transitioned into full-time self-employ-ment from hybrid entrepreneurship (“1” � transi-tion occurred). We call this measure the stagedentry dummy. Second, for those who entered full-time self-employment from hybrid entrepreneur-ship (i.e., staged entry � “1”), we calculated theamount of time, measured in years, each partici-pant spent in hybrid entrepreneurship immediatelyprior to transitioning into full-time self-employ-ment. Participants who transitioned to full-timeself-employment directly from paid work (i.e., di-rect entry) were coded as “0.” We call this measurethe staged entry duration.
We measured cognitive ability with the ArmedForces Qualifications Test (AFQT), which mea-sures quantitative and verbal skills. Prior studieshave demonstrated the AFQT to be a reliable mea-sure (� � .9) (Bock & Moore, 1986), correlatinghighly (.95 or higher) with the g factor, an alterna-tive measure of cognitive ability (Stauffer, Ree, &Carretta, 1996), and stable over time (Gottfredson,1986). Cognitive ability is time invariant.
For entrepreneurial experience, we followed re-cent research (e.g., Eesley & Roberts, 2012; Gregoire& Shepherd, 2012; Hmieleski & Baron, 2009; Toft-Kehler et al., 2014) and measured it as the cumu-lative number of businesses started. In addition, wefollowed Folta et al. (2010) and differentiated be-tween full-time self-employment experience andexperience as a hybrid entrepreneur. We call thesevariables no. full SE experience and no. hybrid SEexperience. Additionally, we used the duration ofthe participant’s most recent prior entrepreneurialspell: duration full SE experience and duration hy-brid SE experience.
Controls. Guided by existing research, we in-cluded a series of control variables theorized toinfluence entrepreneurial entry and survival. Toaccount for socioeconomic and demographic fac-tors (Kim, Aldrich, & Keister, 2006), we includedcontrols for gender (male � “1,” female � “0”); age,measured in years; education, measured as the totalyears of schooling; family net income, measured as
the logged value of total family income; and region(urban � “1,” rural � “0”). We also included loggedhourly rate of pay, logged number of years of in-dustry experience, and a count of the total no. ofprevious jobs to account for opportunity costs, abil-ity, and labor market experience (Shane, 2003).Firm size was included as the logged number ofemployees to control for the small firm effect,which may influence entry decisions (Elfenbein etal., 2010), and for the fact that larger ventures mayhave better chances of survival (Geroski et al.,2010). Industry and occupation differences werecontrolled for with fixed effects based on the U.S.Census Bureau’s three-digit industry codes and theone-digit occupational codes, respectively. Yearfixed effects were included to account for macro-economic conditions.
RESULTS
Table 1 shows descriptive statistics and correla-tions for all variables.
Models 1 to 4 (M1–M4) in Table 2 display theunstandardized regression coefficients from thecompeting-risks Cox proportional hazards model.Models 5 and 6 (M5, M6) display the unstandard-ized regression coefficients from a single-risk Coxmodel (i.e., pooled model) where we treat hybridentry as synonymous with entry into full-time self-employment. Column 7 (M7) shows Wald chi-square tests of coefficient equality (using boot-strapped standard errors) between Models 2 and 4.
Hypothesis 1 predicts that individuals withhigher risk aversion are more likely to enter hybridentrepreneurship relative to full-time self-employ-ment. Results from Models 2 and 4 support thishypothesis. The coefficient for risk aversion pre-dicting full-time self-employment entry was nega-tive and statistically significant (� � �.178;p � .001). In terms of percentage change, a one-unitincrease in risk aversion is associated with a 16.3%decrease in the hazard of entering full-time self-employment. In contrast, the coefficient for riskaversion predicting hybrid entry was not signifi-cant (� � �.005; n.s.). Column 7 of Table 2 con-firms the statistical difference between coefficients(p � .05), providing further support for Hypothe-sis 1.
Hypothesis 2 predicts individuals with low CSEare more likely to enter hybrid entrepreneurshiprelative to full-time self-employment. Results fromModels 2 and 4 provide support for this hypothesis.The coefficient for CSE predicting entry into full-
946 AugustAcademy of Management Journal
TA
BL
E1
Des
crip
tive
Sta
tist
ics
and
Cor
rela
tion
s
Var
iabl
eM
ean
SD
12
34
56
78
910
1112
1314
1516
1718
19
1S
urv
ival
ofw
age
job
(Hyp
oth
eses
1,2)
a18
2.79
017
8.56
0—
2S
urv
ival
ofS
Ejo
b(H
ypot
hes
es3–
5)b,c
277.
880
274.
050
n/a
—
3R
isk
aver
sion
2.86
01.
240
.106
.055
—4
CS
Ed
3.25
0.3
40.0
42.0
69�
.003
—5
Cog
nit
ive
abil
ity
.440
.260
.048
.093
�.0
52.3
89—
6G
end
er(m
ale
�“1
”).5
30.5
00�
.008
.165
.100
.056
�.0
79—
7A
ge39
.340
4.98
0.3
55.1
35.1
06.0
68.0
92.0
33—
8E
du
cati
on13
.560
2.31
0.0
47�
.071
�.0
39.3
39.3
39.0
10.0
54—
9F
irm
size
(log
ged
)3.
920
2.07
0.1
13�
.056
.043
.042
.055
.015
.079
.134
—10
Pay
7.15
0.7
90.1
63.0
57.0
10.2
05.2
75�
.201
.189
.264
.147
—11
No.
ofp
rior
jobs
2.05
01.
140
�.3
78�
.284
�.0
63�
.032
�.0
38�
.027
�.1
42�
.045
�.0
81�
.114
—12
Fam
ily
net
inco
me
(log
ged
)1.
900
1.02
5.1
67.1
22.0
42.1
88.2
56�
.036
.240
.272
.113
.348
�.1
10—
13N
o.of
full
SE
exp
erie
nce
1.29
12.
571
�.0
45�
.170
.017
.003
.029
.010
.160
.089
.004
.018
.220
.054
—
14N
o.of
hyb
rid
SE
exp
erie
nce
.231
1.03
1�
.020
�.0
76.0
33.0
01.0
60.0
38.0
96.0
26.1
35.0
71.1
02.1
02.2
93—
15H
ybri
dtr
ansi
tion
du
mm
y.1
99.3
99n
/a.0
99.0
86.0
29.0
62.0
44.1
64.0
62.2
56.0
70.1
77.0
88�
.080
.080
—
16H
ybri
dtr
ansi
tion
du
rati
on(y
ear)
2.06
25.
181
n/a
.085
.080
.062
.053
.026
.181
.051
.207
.064
.123
.073
�.0
60.0
08.7
79—
17D
ura
tion
full
SE
exp
erie
nce
(yea
r)1.
956
2.32
3n
/a�
.113
.013
.048
.073
.091
.118
.034
�.0
25.0
33.1
03.0
56.7
23.2
95�
.067
�.0
23—
18D
ura
tion
hyb
rid
SE
exp
erie
nce
(yea
r)1.
131
1.47
2n
/a�
.072
�.0
04�
.007
.079
.041
.114
.043
.125
.087
.124
.112
.352
.804
.063
.000
.385
—
19In
du
stry
ten
ure
(yea
r)5.
259
11.4
73�
.052
.263
.053
.017
.055
.123
.231
�.0
22�
.023
.074
�.1
72.1
14.0
33.0
12.0
70.0
63.0
34�
.010
—
an
�31
,919
.b
n�
2,19
8.c
SE
�se
lf-e
mp
loym
ent.
dC
SE
�co
rese
lf-e
valu
atio
n.
time self-employment is positive and statisticallysignificant (� � .398; p � .05). In terms of percent-age change, a one-unit increase in CSE increasesthe hazard of entering full-time self-employment byroughly 32.8%. In contrast, the coefficient for CSEpredicting entry into hybrid entrepreneurship isnegative and statistically significant (� � �.473;p � .05). Column 7 of Table 2 confirms the differ-
ence between coefficients (p � .01), providing fur-ther support for Hypothesis 2.
Next, we compared the competing-risks modelwith a single-risk (i.e., pooled) model in terms ofvariance explained. To do so, we computed thepseudo-R2 m, a measure of variance explainedused in Cox modeling similar to R2 in multiplelinear regression (Maddala, 1983). Comparing the
TABLE 2Results of Competing- and Single-Risk Cox Regression Analysis: Self-Employment Entry
Variables
Full-Time SE Entry Hybrid SE Entry Pooled SE EntryaCoefficient
Comparisonb
M1(Baseline)
M2(Main Effect)
M3(Baseline)
M4(Main Effect)
M5(Baseline)
M6(Main Effect)
M7(M2 vs. M4)
Hypothesized EffectsRisk aversion �.178*** �.005 �.078� *
(.050) (.062) (.041)CSE .398* �.473* �.078 **
(.199) (.233) (.159)Individual ControlsCognitive ability �.000 �.142 .489 .609 .256 .274 �
(.284) (.296) (.422) (.426) (.280) (.285)Gender �.064 �.093 �.230 �.215 �.144 �.158
(.149) (.150) (.205) (.206) (.136) (.135)Age .011 .007 .034 .042 .023 .024
(.028) (.028) (.041) (.041) (.027) (.027)Education .141*** .126*** .063 .078 .101** .101**
(.033) (.033) (.049) (.050) (.031) (.031)Family net income �.032 �.011 �.051 �.028 �.042 �.019
(.049) (.049) (.063) (.062) (.043) (.044)Employment ControlsFirm size �.088* �.086* �.026 �.026 �.052� �.053�
(.035) (.034) (.037) (.037) (.027) (.027)Pay �.107 �.107 .041 .055 �.015 �.009
(.076) (.076) (.077) (.076) (.056) (.056)No. of previous jobs .524*** .519*** .636*** .637*** .586*** .583*** *
(.042) (.041) (.046) (.042) (.036) (.036)Industry tenure �.035 �.038 .014 .018 �.013 �.013 �
(.023) (.023) (.024) (.025) (.023) (.023)No. of full SE
experience.120� .100 .232** .271** .185** .183** �
(.072) (.072) (.075) (.075) (.057) (.057)No. of hybrid SE
experience�.026 �.015 �.020 �.021 �.021 �.018(.049) (.049) (.061) (.062) (.046) (.046)
Model FitPseudo-R2
m .128 .132 .126 .127 .105 .106 Equality of allparametersc
***BIC 7155.41 7719.572 9623.884 9634.812 16751.020 16764.034Pseudo log likelihood �3396.213 �3211.601 �4137.830 �4132.923 7540.648 �7536.784Model (�2) 3497.269 1810.321 1480.724 1454.056 1833.835 1936.741Wald test (�2) �2 (2) � 7.151* �2 (2) � 53.872*** �2 (2) � 4.170
Note: n � 31,191. All models include 3-digit industry, 1-digit occupation, region, and year fixed effects. CSE � core self-evaluation.SE � self-employment. Robust standard errors clustered by participant in parenthesis.
a Single-risk model where hybrid entry is treated as synonymous with full-time self-employment entry (“pooled”).b Wald tests for coefficient equality. Because, in competing risks models, it is possible that two censoring situations may not be
independent, we compute the standard errors using the bootstrapping method (Preacher & Hayes, 2008), a simulation approach (re-sampling 200 times from the original data).
c �2 (13). Test statistic from Narendranathan and Stewart (1991).Two-tailed tests for significance for all effects.
� p � .10.* p � .05.
** p � .01.*** p � .001.
948 AugustAcademy of Management Journal
pseudo-R2 m from the competing-risks model withthe pseudo-R2 m from the single-risk model showsthat the competing-risks model explains almosttwo and a half the variance (i.e., .259/.106) as thepooled model, providing additional support thatthe determinants of hybrid entrepreneurship andfull-time self-employment entry are distinct.
Tables 3a and 3b display the unstandardized re-gression coefficients from single-risk Cox modelsestimating full-time self-employment survival. Hy-pothesis 3 predicts individuals who enter full-timeself-employment in a staged entry process via hy-brid entrepreneurship will survive longer than in-dividuals who enter full-time self-employment di-rectly from paid employment. Results fromModel 2 (Table 3a) and Model 2 (Table 3b) providesupport for this hypothesis. The coefficient forstaged entry dummy in Model 2 (M2) in Table 3a isnegative and statistically significant (� � �.405;p � .001), implying that the hazard of exit is 33.3%lower for individuals who enter full-time self-em-ployment in a staged process relative to those whoenter directly from paid work. Likewise, the coef-ficient for staged entry duration in Model 2 (M2) inTable 3b is negative and statistically significant(� � �.025; p � .001), meaning that a one-unitchange in staged entry duration is associated with a2.5% reduction in the hazard of exit.
Hypothesis 4 predicts that the positive effect ofstaged entry on full-time self-employment survivalis stronger for individuals with high cognitive abil-ity. Model 7 (M7) in Table 3a shows that the inter-action between staged entry dummy and cognitiveability is not statistically significant (� � .153; n.s.).In Model 7 (M7) of Table 3b, the interaction be-tween staged entry duration and cognitive ability ispositive and statistically significant (� � .053; p �.01), which is the opposite of Hypothesis 4. Asdemonstrated by a comparison of the slopes inFigure 2, the decrease in hazard of exit (i.e., sur-vival benefit) associated with longer stays in hybridentrepreneurship is stronger for individuals withlow cognitive ability. Thus, we do not find supportfor Hypothesis 4.
Hypothesis 5 predicts that the positive effect ofstaged entry on full-time self-employment survivalis stronger for individuals with entrepreneurial ex-perience. Models 3 to 6 (M3–M6) in Table 3a pro-vide the coefficients for the interactions betweenour measures of entrepreneurial experience andstaged entry dummy. Models 3 to 6 (M3–M6) inTable 3b display the coefficients for the interac-tions between entrepreneurial experience and
staged entry duration. All interactions are negative,and, in the majority of models, reach statisticalsignificance. Thus, as demonstrated by the slopesin Figure 3, the decrease in hazard of exit (i.e.,survival benefit) associated with staged entry isstronger for experienced entrepreneurs. Thus, over-all, we find support for Hypothesis 5.
Robustness Checks and Supplementary Analysis
We conducted several robustness checks. To be-gin, we took steps to determine if our results arerobust when using measures of entrepreneurshipother than self-employment (Carter, 2011). Thus,we re-tested our hypotheses using two narrowermeasures of entrepreneurship.
First, we focused our analysis on participantswho started a business and reported having em-ployees (i.e., multi-person firms). By doing so, wetreated entrepreneurship as the creation of organi-zations, classically defined as the “coordinated ac-tivities of two or more people” (Barnard, 1938: 73).Specifically, we treated participants as entrepre-neurs only if they reported being self-employed intheir own business and a firm size greater than two.To test Hypotheses 1 and 2, we treated participantswho entered self-employment but reported a firmsize less than two as right censored. Since the sam-ple used to test Hypotheses 3 to 5 is comprised ofparticipants already in full-time self-employment,individuals who reported being full-time self-em-ployed but did not report their self-employed firmsize to be greater than two were excluded fromanalysis.
Second, we categorized participants as entrepre-neurs only if they reported being self-employedand that their business was incorporated. Incorpo-rating a business results in a distinct legal entityseparate from the founder, requires the founder topay various legal fees, comply with governmentmandates, and often signifies entry into the formaleconomy (Kim & Li, 2014). Consequently, focusingon incorporated business is similar to other com-mon measures of entrepreneurship, such as identi-fying new firms through the Dun & Bradstreet(D&B) database (e.g., Batjargal, Hitt, Tsui, Arregle,Webb, & Miller, 2013; Hmieleski & Baron, 2009),which rely on signals that the business intends toengage in commercial activity. For Hypotheses 1and 2, participants who entered unincorporatedself-employment were treated as right censored. Totest Hypotheses 3 to 5, participants with unincor-porated businesses were excluded from analysis.
2014 949Raffiee and Feng
TA
BL
E3a
Res
ult
sof
Sin
gle-
Ris
kC
oxR
egre
ssio
nA
nal
ysis
:F
ull
-Tim
eS
elf-
Em
plo
ymen
tS
urv
ival
Var
iabl
eM
1M
2M
3M
4M
5M
6M
7
Sta
ged
entr
yd
um
my
�.4
05**
*�
.327
***
�.3
88**
*�
.272
***
�.3
92**
*�
.487
***
(.07
3)(.
078)
(.07
4)(.
080)
(.07
6)(.
143)
No.
offu
llS
Eex
per
ien
ce.0
90**
*(.
014)
Sta
ged
entr
yd
um
my
�N
o.of
full
SE
exp
erie
nce
�.0
35�
(.02
5)N
o.of
hyb
rid
SE
exp
erie
nce
.196
***
(.02
7)S
tage
den
try
du
mm
y�
No.
ofh
ybri
dS
Eex
per
ien
ce�
.130
***
(.03
2)D
ura
tion
full
SE
exp
erie
nce
.241
***
(.04
3)S
tage
den
try
du
mm
y�
Du
rati
onfu
llS
Eex
per
ien
ce�
.199
*(.
094)
Du
rati
onh
ybri
dS
Eex
per
ien
ce.3
43**
*(.
089)
Sta
ged
entr
yd
um
my
�D
ura
tion
hyb
rid
SE
exp
erie
nce
�.1
44�
(.10
9)S
tage
den
try
du
mm
y�
Cog
nit
ive
abil
ity
.153
(.25
0)C
ontr
ols
Ris
kav
ersi
on�
.006
.003
.001
.001
�.0
03.0
07.0
02(.
024)
(.02
4)(.
025)
(.02
4)(.
024)
(.02
4)(.
020)
CS
E�
.127
�.1
10�
.110
�.1
04�
.119
�.0
99�
.110
(.10
0)(.
100)
(.10
0)(.
100)
(.10
0)(.
100)
(.10
0)C
ogn
itiv
eab
ilit
y�
.434
**�
.431
**�
.426
**�
.462
**�
.436
**�
.441
**�
.466
**(.
152)
(.15
3)(.
156)
(.15
3)(.
152)
(.15
2)(.
174)
Gen
der
�.3
02**
*�
.283
***
�.2
85**
*�
.290
***
�.3
04**
*�
.281
***
�.2
84**
*(.
074)
(.07
3)(.
074)
(.07
3)(.
073)
(.07
3)(.
073)
Age
�.0
44**
�.0
41**
�.0
41**
�.0
40**
�.0
46**
�.0
41**
�.0
41**
(.01
4)(.
014)
(.01
4)(.
014)
(.01
4)(.
014)
(.01
4)E
du
cati
on.0
60**
*.0
63**
*.0
58**
*.0
67**
*.0
63**
*.0
64**
*.0
63**
*(.
017)
(.01
6)(.
017)
(.01
7)(.
016)
(.01
7)(.
017)
Fam
ily
net
inco
me
�.0
95**
�.0
95**
�.1
09**
*�
.103
***
�.1
02**
*�
.102
***
�.0
93**
(.03
1)(.
031)
(.03
1)(.
030)
(.03
0)(.
030)
(.03
1)F
irm
size
�.0
06.0
04.0
10.0
00.0
10.0
02.0
05(.
014)
(.01
4)(.
014)
(.01
5)(.
014)
(.01
4)(.
014)
Pay
.004
.005
.02
.003
.013
.002
.005
(.03
2)(.
031)
(.03
4)(.
031)
(.03
2)(.
032)
(.03
2)N
o.of
pre
viou
sjo
bs.3
03**
*.3
44**
*.3
16**
*.3
41**
*.3
28**
*.3
39**
*.3
43**
*(.
034)
(.03
3)(.
033)
(.03
3)(.
033)
(.03
3)(.
033)
Ind
ust
ryte
nu
re�
.073
***
�.0
68**
*�
.065
***
�.0
67**
*�
.067
***
�.0
66**
*�
.068
***
(.01
5)(.
016)
(.01
6)(.
016)
(.01
6)(.
016)
(.01
6)M
odel
Fit
Pse
ud
o-R
2m
.034
.036
.038
.037
.038
.036
.036
BIC
3026
3.12
230
216.
670
3016
6.08
430
194.
473
3016
4.72
130
214.
389
3023
1.73
6P
seu
do
log
like
lih
ood
�14
223.
515
�14
196.
441
�14
395.
770
�14
404.
050
�14
398.
210
�14
411.
730
�14
422.
750
Mod
el( �
2)
2873
0.63
125
507.
110
2133
3.88
722
265.
879
2149
3.03
122
292.
473
2556
0.56
0W
ald
test
(�2)a
�2
(1)
�36
.471
***
�2
(2)
�53
.872
***
�2
(2)
�50
.304
***
�2
(2)
�71
.333
***
�2
(2)
�19
.384
***
�2
(2)
�19
.779
***
Not
e:n
�2,
198.
All
mod
els
incl
ud
e3-
dig
itin
du
stry
,1-d
igit
occu
pat
ion
,reg
ion
,an
dye
arfi
xed
effe
cts.
CS
E�
core
self
-eva
luat
ion
.SE
�se
lf-e
mp
loym
ent.
Rob
ust
stan
dar
der
rors
clu
ster
edby
par
tici
pan
tin
par
enth
esis
.a
Wal
dte
stco
mp
ared
toba
seli
ne
mod
el(M
odel
1).
On
e-ta
iled
test
for
hyp
oth
esiz
edef
fect
s,tw
o-ta
iled
test
sfo
rco
ntr
ols.
�p
�.1
0.*
p�
.05.
**p
�.0
1.**
*p
�.0
01.
TA
BL
E3b
Res
ult
sof
Sin
gle-
Ris
kC
oxR
egre
ssio
nA
nal
ysis
:F
ull
-Tim
eS
elf-
Em
plo
ymen
tS
urv
ival
Var
iabl
eM
1M
2M
3M
4M
5M
6M
7
Sta
ged
entr
yd
ura
tion
�.0
25**
*�
.021
**�
.024
***
�.0
25**
*�
.024
***
�.0
56**
*(.
006)
(.00
7)(.
007)
(.00
8)(.
007)
(.01
6)N
o.of
full
SE
exp
erie
nce
.091
***
(.01
3)S
tage
den
try
du
rati
on�
No.
offu
llS
Eex
per
ien
ce�
.001
(.00
3)N
o.of
hyb
rid
SE
exp
erie
nce
.115
***
(.02
7)S
tage
den
try
du
rati
on�
No.
ofh
ybri
dS
Eex
per
ien
ce�
.005
(.00
4)D
ura
tion
full
SE
exp
erie
nce
.272
***
(.04
1)S
tage
den
try
du
rati
on�
Du
rati
onfu
llS
Eex
per
ien
ce�
.022
**(.
008)
Du
rati
onh
ybri
dS
Eex
per
ien
ce.2
88**
*(.
077)
Sta
ged
entr
yd
ura
tion
�D
ura
tion
hyb
rid
SE
exp
erie
nce
�.0
10(.
013)
Sta
ged
entr
yd
ura
tion
�C
ogn
itiv
eab
ilit
y.0
53**
(.02
0)C
ontr
ols
Ris
kav
ersi
on�
.006
.011
.005
.009
�.0
02.0
12.0
09(.
024)
(.02
6)(.
027)
(.02
6)(.
026)
(.02
6)(.
026)
CS
E�
.127
�.1
38�
.138
�.1
25�
.139
�.1
22�
.139
(.10
0)(.
099)
(.10
1)(.
101)
(.10
0)(.
100)
(.09
8)C
ogn
itiv
eab
ilit
y�
.434
**�
.658
***
�.6
21**
*�
.681
***
�.6
37**
*�
.672
***
�.7
73**
*(.
152)
(.15
9)(.
163)
(.16
0)(.
159)
(.15
9)(.
170)
Gen
der
�.3
02**
*�
.245
**�
.249
**�
.234
**�
.280
***
�.2
31**
�.2
44**
(.07
4)(.
082)
(.08
3)(.
084)
(.08
1)(.
084)
(.08
2)A
ge�
.044
**�
.044
**�
.045
**�
.044
**�
.054
***
�.0
45**
�.0
43**
(.01
4)(.
014)
(.01
4)(.
014)
(.01
4)(.
014)
(.01
4)E
du
cati
on.0
60**
*.0
87**
*.0
79**
*.0
91**
*.0
88**
*.0
88**
*.0
88**
*(.
017)
(.01
8)(.
018)
(.01
8)(.
018)
(.01
8)(.
018)
Fam
ily
net
inco
me
�.0
95**
�.1
19**
*�
.129
***
�.1
26**
*�
.125
***
�.1
24**
*�
.118
***
(.03
1)(.
028)
(.03
0)(.
028)
(.03
0)(.
028)
(.02
9)F
irm
size
�.0
06.0
17.0
21.0
11.0
19.0
14.0
18(.
014)
(.01
6)(.
016)
(.01
6)(.
016)
(.01
6)(.
016)
Pay
.004
�.0
09.0
02�
.012
�.0
12�
.013
�.0
05(.
032)
(.03
7)(.
037)
(.03
7)(.
037)
(.03
7)(.
037)
No.
ofp
revi
ous
jobs
.303
***
.386
***
.356
***
.386
***
.370
***
.381
***
.385
***
(.03
4)(.
032)
(.03
3)(.
032)
(.03
2)(.
032)
(.03
2)In
du
stry
ten
ure
�.0
73**
*�
.067
***
�.0
63**
*�
.069
***
�.0
67**
*�
.067
***
�.0
67**
*(.
015)
(.01
8)(.
018)
(.01
8)(.
018)
(.01
8)(.
018)
Mod
elF
itP
seu
do-
R2m
.034
.035
.037
.036
.038
.035
.035
BIC
3026
3.12
230
24.9
3330
186.
020
3023
2.22
230
171.
022
3024
4.42
130
241.
135
Pse
ud
olo
gli
keli
hoo
d�
1422
3.51
5�
1442
3.03
0�
1439
5.77
0�
1440
4.05
0�
1439
8.21
0�
1441
1.73
0�
1420
4.82
7M
odel
( �2)
2873
0.63
124
300.
935
2098
6.93
819
910.
391
1822
0.52
821
632.
841
2475
7.39
4W
ald
test
( �2)a
�2
(1)
�28
.024
***
�2
(2)
�81
.452
***
�2
(2)
�28
.450
***
�2
(2)
�90
.530
***
�2
(2)
�12
.707
***
�2
(2)
�37
.501
***
Not
e:n
�2,
198.
All
mod
els
incl
ud
e3-
dig
itin
du
stry
,1-d
igit
occu
pat
ion
,reg
ion
,an
dye
arfi
xed
effe
cts.
CS
E�
core
self
-eva
luat
ion
.SE
�se
lf-e
mp
loym
ent.
Rob
ust
stan
dar
der
rors
clu
ster
edby
par
tici
pan
tin
par
enth
esis
.aW
ald
test
com
par
edto
base
lin
em
odel
(Mod
el1)
.O
ne-
tail
edte
stfo
rh
ypot
hes
ized
effe
cts,
two-
tail
edte
sts
for
con
trol
s.�
p�
.10.
*p
�.0
5.**
p�
.01.
***
p�
.001
.
Results from both robustness checks were consis-tent with the findings of our main analysis. In bothtests, Hypotheses 1 to 3 were fully supported. Themoderation effects hypothesized in Hypotheses 4and 5 displayed the same directional signs, buttypically exhibited less statistical significance.Given the large reduction in sample size (i.e., fromn � 2,198 to n � 877 and n � 1,028, respectively),the decrease in significance is expected.
Next, we checked the robustness of our resultsusing different sample exclusion criteria and statis-tical specifications. First, although we controlledfor gender, we checked the robustness of our resultsusing gender-specific samples (Folta et al., 2010).Second, we included individuals who workedfewer than 30 hours per week as this may representa strategic decision in anticipation of reallocatingtime between paid and self-employment. Third,we employed a number of alternative event his-tory models and hazard specifications (Allison,1984), including a fixed effects Cox model inwhich we specified a unique baseline hazard foreach participant, thereby allowing us to removeany unobserved sources of variation constantwithin individuals (Allison, 2009). We also usedthe “stcrreg” command in STATA 12.1 to testHypotheses 1 and 2 via the competing risksmethod of Fine and Gray (1999), which focuseson the cumulative incidence function as opposedto cause-specific hazards. No material differencesmanifested from these tests; our results remainedrobust and unchanged.
Finally, to demonstrate how our study buildsupon and extends prior hybrid entrepreneurshipresearch, we conducted scientific replications oftwo studies that explore the differences betweenhybrid and full-time entrepreneurs: Folta et al.(2010) and Petrova (2012). Using the NLSY79, wereplicated each study by using similar exclusion/inclusion criteria for sample construction, left-/right-hand side variables, and statistical methods.Despite the innate differences between datasources, the broad nature of the NLSY79 allowed usto reconstruct or use a similar proxy for the vastmajority of variables used in each study.
Table 4 reports the results of our replications.We, first, replicated each study using only the vari-ables in the original studies (Models 1a, 1b, 3a, 3b).We then added our hypothesized variables and ad-ditional controls (Models 2a, 2b, 4a, 4b). For thesake of brevity, Table 4 is abbreviated and reportsonly the regression coefficients for a subset of vari-ables used in Folta et al. (2010) and Petrova (2012)(i.e., some regression coefficients are omitted).Given the inherent differences in data, measures/proxies, and sample size, our results deviate somefrom the findings of Folta et al. (2010) and Petrova(2012). Likewise, given the additional controls,changes in sample size/sample window, and statis-tical estimation, some results in Table 4 differ fromthe results reported in Table 2. Nevertheless, inboth replications, Hypotheses 1 and 2 receive gen-eral support (Models 2a, 2b, 4a, 4b): risk-averse andlow CSE individuals prefer hybrid entry over entry
FIGURE 2Interaction between Staged Entry Duration and Cognitive Ability
952 AugustAcademy of Management Journal
FIGURE 3Entrepreneurial Experience and Staged Entry Interactions(3a) Staged entry dummy � No. of hybrid SE experience(3b) Staged entry dummy � Duration full SE experience(3c) Staged entry duration � Duration full SE experience
2014 953Raffiee and Feng
TA
BL
E4
Rep
lica
tion
ofP
rior
Hyb
rid
En
trep
ren
eurs
hip
Stu
die
s
Var
iabl
esa
Rep
lica
tion
ofP
etro
va(2
012)
Rep
lica
tion
ofP
etro
va(2
012)
�O
ur
Hyp
oth
esiz
ed/c
ontr
olva
riab
les
Rep
lica
tion
ofF
olta
etal
.(2
010)
Rep
lica
tion
ofF
olta
etal
.(2
010)
�O
ur
Hyp
oth
esiz
ed/c
ontr
olva
riab
les
Mod
el1a
Fu
ll-T
ime
En
try
Mod
el1b
Hyb
rid
En
try
sigb
Mod
el2a
Fu
ll-T
ime
En
try
Mod
el2b
Hyb
rid
En
try
sigb
Mod
el3a
Fu
ll-T
ime
En
try
Mod
el3b
Hyb
rid
En
try
sigb
Mod
el4a
Fu
ll-T
ime
En
try
Mod
el4b
Hyb
rid
En
try
sigb
Ris
kav
ersi
on�
.001
.021
**
�.0
86**
*.0
40**
(.03
2)(.
009)
(.02
3)(.
035)
CS
E.4
87*
�.0
40.3
29**
*�
.039
*(.
209)
(.51
4)(.
095)
(.13
9)C
ogn
itiv
eab
ilit
y�
.001
.019
****
�.0
02�
.023
***
**(.
003)
(.00
8)(.
001)
(.00
2)A
ge�
.023
�.0
40�
.017
�.0
32�
.049
***
.025
***
.008
�.0
01(.
020)
(.06
0)(.
026)
(.06
4)(.
010)
(.01
8)(.
013)
(.01
9)E
du
cati
on�
.044
��
.100
�.1
51**
*�
.259
**.8
25**
*�
.214
**.4
68�
2.97
1***
***
(.02
5)(.
071)
(.03
7)(.
084)
(.22
4)(.
359)
(.31
7)(.
441)
Mar
ital
stat
us
.140
�.0
37�
.011
�.1
99�
.074
.627
***
***
�.0
54.5
26**
***
*(.
122)
(.38
0)(.
163)
(.42
1)(.
057)
(.10
7)(.
074)
(.11
6)F
amil
yto
tal
net
inco
me
.243
*2.
748
.129
8.09
6**
n/a
.049
.146
�(.
117)
(3.6
78)
(.15
3)(4
.058
)(.
055)
(.08
4)F
amil
yto
tal
net
wea
lth
�.0
35�
.192
.035
�.1
52�
.013
�.0
60**
**.0
11�
.060
****
(.04
6)(.
147)
(.05
9)(.
152)
(.00
9)(.
021)
(.01
0)(.
021)
Fir
msi
zen
/a�
.207
***
.072
***
�.0
66**
*.0
72**
***
*�
.041
*.0
69**
***
(.03
5)(.
077)
(.01
4)(.
021)
(.01
6)(.
022)
Pay
n/a
�.1
16.3
92�
.108
*�
.087
.005
�.2
04**
*(.
129)
(.38
3)(.
043)
(.06
2)(.
059)
(.06
7)N
o.of
pre
viou
sjo
bsn
/a.0
75.5
20**
***
.157
***
.451
***
***
.010
.373
***
***
(.05
9)(.
126)
(.02
5)(.
041)
(.03
2)(.
044)
Ind
ust
ryte
nu
ren
/a.0
00�
.022
.050
�.0
57.1
47**
*�
.072
(.00
0)(.
016)
(.04
0)(.
087)
(.04
2)(.
083)
Wor
kte
nu
re.0
02�
.186
�.1
57*
.752
�.0
51*
.131
****
*.0
47�
.247
**�
(.05
1)(.
194)
(.07
1)(.
898)
(.02
1)(.
044)
(.02
7)(.
065)
Nof
spel
ls4,
106
4106
20,0
5920
,059
Pse
ud
o-R
2.0
10.3
62.1
40.3
53B
IC35
42.3
8025
06.3
2019
146.
425
1568
0.47
9P
seu
do
log
like
lih
ood
�16
54.7
14�
8879
.762
�66
76.2
33M
odel
( �2)
48.5
0111
51.1
9728
83.1
9972
90.2
57W
ald
test
( �2)
�2
(16)
�75
9.61
2***
�2
(12)
�26
23.4
86**
*M
eth
odM
ult
inom
ial
Pro
bit
Mu
ltin
omia
lL
ogit
Yea
rsu
sed
1998
–200
019
94–2
001
Ori
gin
ald
ata
sou
rce
PS
ED
(pan
elst
ud
yof
entr
epre
neu
rial
dyn
amic
s)T
hre
em
atch
edlo
ngi
tud
inal
dat
aso
urc
eson
the
Sw
edis
hla
bor
mar
ket
(LO
UIS
E,
LO
UIS
E,
SR
U)
Sam
ple
Res
pon
den
tsw
ith
ap
osit
ive
cash
-flo
wfo
rm
ore
than
3m
onth
sw
ere
excl
ud
ed;
busi
nes
ssp
onso
red
star
t-u
ps
are
rem
oved
Hig
h-t
ech
nol
ogy
man
ufa
ctu
rer
orkn
owle
dge
-in
ten
sive
serv
ice
firm
aM
odel
sin
clu
de
all
re-c
reat
edco
vari
ates
from
Pet
rova
(201
2)an
dF
olta
etal
.(2
010)
,re
spec
tive
ly.
For
brev
ity
and
illu
stra
tion
pu
rpos
es,
we
only
rep
ort
asu
bset
ofre
gres
sion
coef
fici
ents
.C
omp
lete
resu
lts
avai
labl
efr
omth
eau
thor
su
pon
requ
est.
bW
ald
test
sfo
rco
effi
cien
teq
ual
ity.
Tw
o-ta
iled
test
sfo
rsi
gnif
ican
cefo
ral
lef
fect
s�
p�
.10.
*p
�.0
5.**
p�
.01.
***
p�
.001
.
into full-time self-employment. Moreover, a com-parison of the pseudo-R2 demonstrates that the ad-dition of our hypothesized and control variablessignificantly increases the variance explained ineach replication (i.e., from .01 to .36 and .14 to .35).
DISCUSSION
In this study, we examined the implications ofhybrid entrepreneurship for theories of entrepre-neurial entry and survival. Drawing from real op-tions theory and the individual differences litera-ture, we hypothesized and found that risk-averseand less confident individuals are more likely toenter hybrid entrepreneurship (as it entails lessdownside risk) relative to full-time self-employ-ment. In turn, we argued and found evidence thatindividuals who enter full-time self-employment ina staged entry process by means of hybrid entrepre-neurship survive significantly longer than thosewho enter full-time self-employment directly froma paying job. Adding support to our theory that thesurvival advantage is driven by a learning effectthat takes place during hybrid entrepreneurship,our findings suggest that factors that influence anindividual’s capacity to process new informationmoderate this relationship.
Hybrid Entrepreneurship and Entry
This study makes several contributions. First, weadd to the entrepreneurship literature by providinga more nuanced understanding of the entrepreneur-ial entry process. Indeed, the findings from ourstudy suggest that the classical assumption thatentrepreneurs are comfortable with risk (e.g., Kihl-strom & Laffont, 1979) comes with an importantcaveat. On the one hand, individuals who jumpdirectly into full-time self-employment are less riskaverse than non-entrepreneurs, just as extant the-ory predicts. In fact, a one standard deviation in-crease in risk aversion is associated with a 20.21%decrease in the hazard of entry into full-time self-employment. On the other hand, individuals whoenter hybrid entrepreneurship appear to have riskpreferences that are indistinguishable from thosewho remain in paid employment. Thus, our find-ings suggest that risk aversion influences the pro-cess of how an individual decides to start a busi-ness (i.e., full-time versus hybrid), not necessarilywhether the individual decides to start a businessor not. A similar pattern of results was detectedwith regards to CSE. Supporting traditional logic
that entrepreneurs are highly confident (e.g.,Knight, 1921; Moore et al., 2007), we find that lowCSE decreases the likelihood of direct entry intofull-time self-employment. Specifically, a one stan-dard deviation decrease in CSE is associated withan 11.2% reduction in the hazard of full-time self-employment entry. However, low CSE does notdecrease the likelihood of entry into hybrid entre-preneurship. Thus, consistent with our finding re-garding risk aversion, our results suggest that CSEinfluences how rather than if entrepreneurial entryoccurs.
Taken together, the implications of these find-ings extend to a number of research streams withinthe entrepreneurship literature. For example, ourfindings suggest that traits-based researchers maybe better served by seeking to understand how anindividual’s predispositions influence the form ofentrepreneurial entry rather than broadly categoriz-ing all entrepreneurs as individuals who systemat-ically exhibit certain traits. Similarly, theories ofentrepreneurial cognition, which were developedin part to address the inconsistent findings from thetrait-based approaches (Mitchell et al., 2007), havethe potential to become more nuanced by incorpo-rating hybrid entrepreneurship into their concep-tual models. To illustrate, cognition research hasshown that entrepreneurs are willing to make gen-eralizations from limited information (Busenitz &Barney, 1997). However, it seems possible that hy-brid entrepreneurs, who, by the very nature of en-tering hybrid entrepreneurship, collect additionalinformation about their venture prior to commit-ting to it full time, may be less comfortable withdoing so (hence their choice of hybrid entry). Like-wise, theories of entrepreneurship that emphasizethe importance of context in shaping entrepreneur-ial decisions (e.g., Sørensen & Fassiotto, 2011) maybenefit from theoretically accounting for the differ-ences between hybrid entrepreneurship and full-time self-employment. For example, consistentwith prior research (e.g., Elfenbein et al., 2010;Sørensen, 2007), in the current study, we find em-ployer size to decrease the likelihood of entry intofull-time self-employment (i.e., small firm effect).However, in accord with the findings of Folta et al.(2010), we find no such relationship with regards tohybrid entry. Indeed, inconsistencies such as thesehighlight how the integration of hybrid entrepre-neurship into existing entrepreneurship theoryopens up multiple new lines of scholarly inquiry.
2014 955Raffiee and Feng
Hybrid Entrepreneurship and Survival
In addition, prior research has focused on theantecedents of hybrid entrepreneurship withoutconsidering its implications for entrepreneurialoutcomes. Building on the work of Folta et al.(2010), who argue that full-time self-employmententry is endogenous to hybrid entrepreneurship,we investigate how entering full-time self-employ-ment in a staged entry process (i.e., paid employ-ment ¡ hybrid entrepreneurship ¡ full-time self-employment) impacts venture survival. This isimportant for several reasons. First, we find that anon-trivial portion of hybrid entrepreneurs ulti-mately transition into full-time self-employment,further substantiating the argument that entry intofull-time self-employment is endogenous to hybridentrepreneurship (Folta et al., 2010). Second, inaccord with real options logic, our findings dem-onstrate that there is a significant survival benefit(i.e., a 33.3% decrease in the hazard of exit) asso-ciated with staged entry, even when controlling forother factors theorized to influence survival. On thesurface, this finding may be viewed as being some-what at odds with the commonly held belief that, inorder to be successful, entrepreneurs must devotetheir full attention to their business. While this maybe the case, our results suggest that it is worthwhileto take steps to determine if the business idea war-rants large-scale commitment prior to doing so. Inother words, our findings suggest that, given theuncertainty associated with new businesses, entre-preneurs are best served by making small initialcommitments early on, giving themselves the op-tion to commit fully to their business after theyhave had a chance to accumulate information andassess its potential/prospects.
Real Options and Entrepreneurship
Our study offers empirical evidence that individ-ual characteristics influence real options decisions.While several studies have sought to relax the as-sumption of risk neutrality in real options theory,they have typically done so using formal mathe-matical models (e.g., Hugonnier & Morellec, 2007).Our study adds to this literature by offering empir-ical evidence (i.e., non-mathematical model) thatrisk aversion and low CSE have effects that mirrorthose of high exogenous uncertainty in traditionalreal options models. As such, we proffer empiricalevidence that risk aversion and CSE influence in-dividual-specific entry thresholds (O’Brien et al.,
2003), which, in turn, influence the process of en-trepreneurial entry (i.e., full-time versus hybrid).Likewise, while we focused on the establishment ofreal options and the survivorship implications con-ditional on option execution, our consistent patternof results suggest that individual characteristics arelikely to play important roles in other aspects ofreal options decisions (e.g., speed to option adjudi-cation) (Majd & Pindyck, 1987).
Similarly, our finding that staged entry into full-time self-employment is associated with increasedchances of survival proffers evidence that, on aver-age, individuals tend to exercise real options whenthey are “in the money” (Dixit & Pindyck, 1994),despite the inherent difficulties of doing so (e.g.,Adner & Levinthal, 2004). Along these lines, wefurther argued and demonstrated that individualcharacteristics moderate the strength of this rela-tionship. By doing so, we add to the real optionsliterature by demonstrating that individual factorsinfluence the efficacy of real option exercise deci-sions. Interestingly, we did not find support forHypothesis 4, which predicted that the positiverelationship between staged entry and survivalwould be stronger for individuals with high cogni-tive ability. Rather, our results suggest the marginalbenefit of extended stays (staged entry duration) inhybrid entrepreneurship are stronger for individu-als with low cognitive ability (the opposite of Hy-pothesis 4). More specifically, a one-year stay inhybrid entrepreneurship prior to entry into full-time self-employment reduces the hazard of exit by4.20% for individuals with low cognitive ability(25th percentile) but only 0.44% for individualswith high cognitive ability (75th percentile). Al-though this ran counter to our expectation, oneexplanation for this finding is that individuals withlow cognitive ability learn at a slower pace (Lubin-ski, 2004).
As expected, our results provided general sup-port for Hypothesis 5, which predicted the positiverelationship between staged entry and survivalwould be stronger for experienced entrepreneurs.This finding carries several implications worthy ofdiscussion. First, our study reinforces that the re-lationships between entrepreneurial experienceand entrepreneurial outcomes may be contingenton other factors (e.g., Hmieleski & Baron, 2009).Second, although not hypothesized, the main effectof entrepreneurial experience was associated withan increased likelihood of entrepreneurial exit,similar to the findings of Jørgensen (2005). Oneexplanation for this finding is that experienced
956 AugustAcademy of Management Journal
entrepreneurs are quicker to pull the plug on po-tentially struggling ventures. This explanation isreinforced by our finding that experienced entre-preneurs who start a business as a hybrid, and thenopt to commit to it full time, experience longersurvival (Hypothesis 5). Specifically, our resultsindicate that the decrease in exit hazard associatedwith staged entry can be up to three times larger forexperienced entrepreneurs (75th percentile of ex-perience) relative to inexperienced entrepreneurs(25th percentile of experience).
Our study also has implications for the currentdialogue within the entrepreneurship literature re-garding entrepreneurial opportunities (e.g., Shane,2012). An emerging view in this literature is thatsome opportunities are enacted via a highly uncer-tain and often myopic process, where the feasibilityand/or value of ideas are illuminated through aniterative progression of action and reaction (Alva-rez, Barney, & Anderson, 2013). As such, given theuncertain outcomes associated with opportunityenactment, the limited sunk cost and risk exposureassociated with hybrid entrepreneurship would ap-pear to make an attractive exploitation strategy.Indeed, the strong survival advantage associatedwith hybrid entrepreneurship is in accord with Al-varez et al.’s (2013) assertion that the enactmentprocess should be linked to sustained advantages.Relatedly, our findings compliment recent researchindicating that the reduction of demand uncer-tainty is a key driver of entrepreneurial action (Au-tio et al., 2013). In some situations, however, reduc-ing demand uncertainty may only be possiblethrough the process of commercialization (George& Bock, 2012). Thus, individuals may intentionallyuse hybrid entrepreneurship as a means to reducedemand uncertainty prior to committing to theirbusiness. Future research capturing the specificmotivations of hybrid entrepreneurs would bevaluable.
Finally, the practical importance of our findingsis underscored by the fact that hybrid entrepreneur-ship has experienced a recent explosion in growth(Grant, 2011). A key driver of this trend has beenadvances in technology, which have reduced thecost and time commitments necessary when start-ing a business. For example, instead of opening abrick-and-mortar location, hybrid entrepreneurscan use online marketplaces such as eBay (coinci-dentally, started by a hybrid entrepreneur) in theirspare time. Likewise, advancements in social me-dia marketing tools offer low-cost and efficientways to reach target consumers. Indeed, trends
such as these suggest the occurrence of hybrid en-trepreneurship is likely to continue to grow, em-phasizing why more research on hybrid entrepre-neurship is needed from an academic and policy-making standpoint.
Limitations
Our study is not without limitations. First, al-though we engaged in considerable efforts todemonstrate that our findings are robust to multi-ple measures of entrepreneurship, our measuresare not perfect. In particular, we do not differenti-ate ventures based on the degree of novelty/inno-vativeness (e.g., Schumpeterian entrepreneurship).While we included industry controls at the three-digit level and demonstrated that our results arerobust to a fixed effects specification of the Coxproportional hazards model, we acknowledge thatwe are unable to fully account for venture-specificcharacteristics. Thus, while we contribute to realoptions theory by using individual characteristicsto predict a real options argument, our study lacksa measure of venture-specific risk/uncertainty. Fu-ture research should examine these dimensions intandem.
Second, we use a measure of CSE that is timeinvariant. Although CSE is a broad, latent person-ality trait proposed to be independent of contextand time (Judge et al., 1997), there have been callsfor scholars to investigate intra-individual changesin CSE over time (Chang et al., 2012). We echothese calls, noting that an interesting direction forfuture research would be to study how entrepre-neurial experiences influence changes (or lackthereof) in CSE.
Third, we followed extant research (Delmar &Shane, 2006; Dencker, Gruber, & Shah, 2009a) anddefined survival as the continuation of entrepre-neurial effort (Shane, 2003). Although the realiza-tion of economic gains from entrepreneurial effortsusually requires sustainment in self-employment(Patel & Thatcher, 2012), longer survival is not nec-essarily equivalent to higher performance (Gimenoet al., 1997). Therefore, although survival is animportant dimension of entrepreneurship, futureresearch could examine the effects of hybrid entre-preneurship on other outcomes such as financialperformance (Hmieleski & Baron, 2009), job cre-ation (Dencker, Gruber, & Shah, 2009b), resourceacquisition (Kotha & George, 2012), or the likeli-hood of being favorably acquired (Wennberg, Wik-lund, DeTienne, & Cardon, 2010).
2014 957Raffiee and Feng
Fourth, we postulated that hybrid entrepreneur-ship facilitates learning regarding two key dimen-sions that impact venture survival: (1) the qualityor viability of the venture idea and (2) the individ-ual’s capabilities/fit in the entrepreneurial context.However, given the structure of our data, we areunable to isolate the relative importance of eachdimension. We encourage future studies to unpackthese relationships further. In addition, it is possi-ble that the survival advantage is not driven bylearning, but, rather, by the fact that individualswho choose to enter hybrid entrepreneurship sys-tematically select better ideas ex ante. However, theresults from our shared frailty model indicate thatthis is unlikely.
Fifth, we followed recent research and measuredentrepreneurial experience as the cumulative num-ber of prior businesses started (e.g., Eesley & Rob-erts, 2012; Toft-Kehler et al., 2014) and the durationof recent self-employment spells. As such, a limi-tation of our study, and one that is endemic innearly all studies linking entrepreneurial experi-ence to outcomes (Ucbasaran, Shepherd, Lockett, &Lyon, 2013), is that we are unable to determinewhether an individual’s prior entrepreneurial en-deavors were successes or failures. Indeed, giventhat research has linked entrepreneurial failurewith re-entry into entrepreneurship (Ucbasaran etal., 2013), and that entrepreneurial failure has thepotential to influence both learning (Cope, 2011;Shepherd, 2003) and opportunity identification(Ucbasaran et al., 2009), understanding how priorfailure (or multiple/sequences of failure) influencethe antecedents and outcomes of entrepreneurship/hybrid entrepreneurship represents an importantarea for future research.
Finally, although we used a nationally represen-tative sample and controlled for both industry andoccupation, there might be concerns as to the gen-eralizability of our findings. In some industries, itmay be unrealistic to launch a venture as a hybridentrepreneur. Thus, while we did not find indica-tion of differing rates of hybrid entrepreneurship byindustry, our findings should be interpreted withcaution, particularly with respect to capital-inten-sive industries.
CONCLUSION
This study adds to the entrepreneurship litera-ture by using a real options approach to examinethe implications of hybrid entrepreneurship for en-trepreneurial entry and survival. Using a large lon-
gitudinal sample, our results highlight that the pro-cesses of entrepreneurial entry and survival aremore complex than typically assumed. With re-gards to entrepreneurial entry, we find support forour predictions that risk-averse and less-confidentindividuals who enter self-employment are morelikely to do so via hybrid entrepreneurship. Withrespect to entrepreneurial survival, we find evi-dence that individuals who enter full-time self-employment in a staged entry process through hy-brid entrepreneurship survive significantly longerthan individuals who enter directly from paid em-ployment. This relationship is stronger for experi-enced entrepreneurs and individuals with lowerintelligence. Given our findings, we emphasize thatfuture research aimed at furthering our knowledgeof the nuances associated with hybrid entrepre-neurship is critical for the field to develop a morecomprehensive and complete understanding of theentrepreneurial process.
REFERENCES
Adner, R., & Levinthal, D. A. 2004. What is not a realoption: Identifying boundaries for the application ofreal options to business strategy. Academy of Man-agement Review, 29: 74–85.
Aldrich, H. 1999. Organizations evolving. ThousandOaks, CA: Sage.
Allison, P. D. 1984. Event history analysis: Regressionfor longitudinal event data. Beverly Hills, CA: Sage.
Allison, P. D. 2009. Fixed effects regression models. LosAngeles, CA: Sage.
Alvarez, S. A., Barney, J. B., & Anderson, P. 2013. Form-ing and exploiting opportunities: The implicationsof discovery and creation processes for entrepre-neurial and organizational research. OrganizationScience, 24: 301–317.
Amit, R., Muller, E., & Cockburn, I. 1995. Opportunitycosts and entrepreneurial activity. Journal of Busi-ness Venturing, 10: 95–106.
Autio, E., & Acs, Z. 2010. Intellectual property protectionand the formation of entrepreneurial growth aspira-tions. Strategic Entrepreneurship Journal, 4: 234–251.
Autio, E., Dahlander, L., & Frederiksen, L. 2013. Informa-tion exposure, opportunity evaluation and entrepre-neurial action: An empirical investigation of an on-line user community. Academy of ManagementJournal, 56: 1348–1371.
Barnard, C. 1938. The functions of the executive. Cam-bridge, MA: Harvard University Press.
958 AugustAcademy of Management Journal
Barnett, M. L. 2008. An attention-based view of realoptions reasoning. Academy of Management Re-view, 33: 606–628.
Baron, R. A., & Ensley, M. D. 2006. Opportunity recogni-tion as the detection of meaningful patterns: Evi-dence from comparisons of novice and experiencedentrepreneurs. Management Science, 52: 1331–1344.
Barsky, R. B., Juster, F. T., Kimball, M. S., & Shapiro,M. D. 1997. Preference parameters and behavioralheterogeneity: An experimental approach in thehealth and retirement study. Quarterly Journal ofEconomics, 112: 537–579.
Batjargal, B., Hitt, M., Tsui, A., Arregle, J.-L., Webb, J., &Miller, T. 2013. Institutional polycentrism, entrepre-neur’s social networks and new venture growth.Academy of Management Journal, 56: 1024–1049.
Baum, J. R., & Bird, B. J. 2010. The successful intelligenceof high-growth entrepreneurs: Links to new venturegrowth. Organization Science, 21: 397–412.
Becker, H. S. 1960. Notes on the concept of commitment.American Journal of Sociology, 66: 32–40.
Block, J., Thurik, R., van der Zwan, P., & Walter, S. 2013.Business takeover or new venture? Individual andenvironmental determinants from a cross-countrystudy. Entrepreneurship Theory and Practice, 37:1099–1121.
Bock, R. D., & Moore, E. G. J. 1986. Advantage anddisadvantage: A profile of American youth. Hills-dale, NJ: L. Erlbaum Associates.
Brockhaus, R. H. 1980. Risk taking propensity of entre-preneurs. Academy of Management Journal, 23:509–520.
Burke, A. E., FitzRoy, F. R., & Nolan, M. A. 2008. Whatmakes a die-hard entrepreneur? Beyond the “em-ployee or entrepreneur” dichotomy. Small BusinessEconomics, 31: 93–115.
Busenitz, L. W., & Barney, J. B. 1997. Differences betweenentrepreneurs and managers in large organizations:Biases and heuristics in strategic decision-making.Journal of Business Venturing, 12: 9–30.
Carter, S. 2011. The rewards of entrepreneurship: Explor-ing the incomes, wealth, and economic well-being ofentrepreneurial households. Entrepreneurship The-ory and Practice, 35: 39–55.
Caves, R. E. 1998. Industrial organization and new find-ings on the turnover and mobility of firms. Journalof Economic Literature, 36: 1947–1982.
Chang, C. H., Ferris, D. L., Johnson, R. E., Rosen, C. C., &Tan, J. A. 2012. Core self-evaluations: A review andevaluation of the literature. Journal of Manage-ment, 38: 81–128.
Chatterjee, A., & Hambrick, D. C. 2007. It’s all about me:Narcissistic chief executive officers and their effectson company strategy and performance. Administra-tive Science Quarterly, 52: 351–386.
Cohen, A. 2002. The perfect store: Inside eBay. Boston,MA: Little, Brown and Company.
Cohen, W. M., & Levinthal, D. A. 1990. Absorptive ca-pacity: A new perspective on learning and innova-tion. Administrative Science Quarterly, 35: 128–152.
Cook, T. D., & Campbell, D. T. 1979. Quasi-experimen-tation: Design & analysis issues for field settings.Boston, MA: Houghton Mifflin.
Cope, J. 2011. Entrepreneurial learning from failure: Aninterpretative phenomenological analysis. Journalof Business Venturing, 26: 604–623.
Cramer, J. S., Hartog, J., Jonker, N., & Van Praag, C. M.2002. Low risk aversion encourages the choice forentrepreneurship: An empirical test of a truism.Journal of Economic Behavior and Organization,48: 29–36.
Delmar, F., & Shane, S. 2006. Does experience matter?The effect of founding team experience on the sur-vival and sales of newly founded ventures. StrategicOrganization, 4: 215–247.
Dencker, J. C., Gruber, M., & Shah, S. K. 2009a. Pre-entryknowledge, learning, and the survival of new firms.Organization Science, 20: 516–537.
Dencker, J. C., Gruber, M., & Shah, S. K. 2009b. Individ-ual and opportunity factors influencing job creationin new firms. Academy of Management Journal,52: 1125–1147.
Dixit, A. K., & Pindyck, R. S. 1994. Investment underuncertainty. Princeton, NJ: Princeton UniversityPress.
Eesley, C. E., & Roberts, E. B. 2012. Are you experiencedor are you talented?: When does innate talent versusexperience explain entrepreneurial performance?Strategic Entrepreneurship Journal, 6: 207–219.
Elfenbein, D. W., Hamilton, B. H., & Zenger, T. R. 2010.The small firm effect and the entrepreneurial spawn-ing of scientists and engineers. Management Sci-ence, 56: 659–681.
Evans, D. S., & Leighton, L. S. 1989. Some empiricalaspects of entrepreneurship. American EconomicReview, 79: 519–535.
Fine, J. P., & Gray, R. J. 1999. A proportional hazardsmodel for the subdistribution of a competing risk.Journal of the American Statistical Association,94: 496–509.
Folta, T. B. 1998. Governance and uncertainty: The trade-
2014 959Raffiee and Feng
off between administrative control and commitment.Strategic Management Journal, 19: 1007–1028.
Folta, T. B. 2007. Uncertainty rules the day. StrategicEntrepreneurship Journal, 1: 97–99.
Folta, T. B., Delmar, F., & Wennberg, K. 2010. Hybridentrepreneurship. Management Science, 56: 253–269.
Ford, H., & Crowther, S. 1922. My life and work. GardenCity, NY: Doubleday, Page & Co.
Funder, D. C. 2001. Personality. Annual Review of Psy-chology, 52: 197–221.
George, G., & Bock, A. J. 2012. Models of opportunity:How entrepreneurs design firms to achieve theunexpected. Cambridge, England: Cambridge Uni-versity Press.
Geroski, P. A., Mata, J., & Portugal, P. 2010. Foundingconditions and the survival of new firms. StrategicManagement Journal, 31: 510–529.
Gimeno, J., Folta, T. B., Cooper, A. C., & Woo, C. Y. 1997.Survival of the fittest? Entrepreneurial human capi-tal and the persistence of underperforming firms.Administrative Science Quarterly, 42: 750–783.
Gottfredson, L. S. 1986. Societal consequences of the gfactor in employment. Journal of Vocational Behav-ior, 29: 379–410.
Gottfredson, L. S. 1997. Why g matters: The complexityof everyday life. Intelligence, 24: 79–132.
Grant, A. 2011. Careers on the side, Chicago Tribune,April 8, 2011. http://www.chicagotribune.com/features/tribu/ct-tribu-free-time-entrepreneur,0,4964041.story.
Gregoire, D. A., & Shepherd, D. A. 2012. Technology-market combinations and the identification ofentrepreneurial opportunities: An investigation ofthe opportunity-individual nexus. Academy ofManagement Journal, 55: 753–785.
Gutierrez, R. G. 2002. Parametric frailty and sharedfrailty survival models. STATA Journal, 2: 22–44.
Hamilton, B. H. 2000. Does entrepreneurship pay? Anempirical analysis of the returns of self-employment.Journal of Political Economy, 108: 604–631.
Heckman, J. 1979. Sample selection bias as a specifica-tion error. Econometrica, 47: 153–161.
Hiller, N. J., & Hambrick, D. C. 2005. Conceptualizingexecutive hubris: The role of (hyper-)core self-eval-uations in strategic decision-making. Strategic Man-agement Journal, 26: 297–319.
Hmieleski, K. M., & Baron, R. A. 2009. Entrepreneurs’optimism and new venture performance: A socialcognitive perspective. Academy of ManagementJournal, 52: 473–488.
Holmes, T. J., & Schmitz, J. A. 1995. On the turnover ofbusiness firms and business managers. Journal ofPolitical Economy, 103: 1005–1038.
Hugonnier, J. N., & Morellec, E. 2007, Real options andrisk aversion. (Swiss Finance Institute ResearchPaper Series). Geneva, Switzerland: Swiss FinanceInstitute.
Hunter, J. E. 1986. Cognitive-ability, cognitive aptitudes,job knowledge, and job-performance. Journal of Vo-cational Behavior, 29: 340–362.
Inc. staff. 1997. Inc. 500 almanac: A statistical look at thisyear’s inc. 500 companies and the CEOs that runthem. http://www.inc.com/magazine/19971015/1463.html. October 17.
Jain, S., George, G., & Maltarich, M. 2009. Academics orentrepreneurs? Investigating role identity modifica-tion of university scientists involved in commercial-ization activity. Research Policy, 38: 922–935.
Jensen, A. R. 1998. The g factor: The science of mentalability. Westport, CT: Praeger.
Jørgensen, R. 2005. A duration analysis of Danish start-ups. Copenhagen, Denmark: Centre for Economicand Business Research.
Jovanovic, B. 1982. Selection and the evolution of indus-try. Econometrica, 50: 649–671.
Judge, T. A., Locke, E. A., & Durham, C. C. 1997. Thedispositional causes of job satisfaction: A core eval-uations approach. Research in Organizational Be-havior, 19: 151–188.
Judge, T. A., Erez, A., Bono, J. E., & Thoresen, C. J. 2003.The core self-evaluations scale: Development of ameasure. Personnel Psychology, 56: 303–331.
Judge, T. A., & Hurst, C. 2007. Capitalizing on one’sadvantages: Role of core self-evaluations. Journal ofApplied Psychology, 92: 1212–1227.
Judge, T. A., & Hurst, C. 2008. How the rich (and happy)get richer (and happier): Relationship of core self-evaluations to trajectories in attaining work success.Journal of Applied Psychology, 93: 849–863.
Kihlstrom, R. E., & Laffont, J. J. 1979. A general equilib-rium entrepreneurial theory of firm formation basedon risk aversion. Journal of Political Economy, 87:719–748.
Kim, P. H., Aldrich, H. E., & Keister, L. A. 2006. Access(not) denied: The impact of financial, human, andcultural capital on entrepreneurial entry in theUnited States. Small Business Economics, 27: 5–22.
Kim, P. H., & Li, M. 2014. Injecting demand throughspillovers: Foreign direct investment, domestic so-cio-political conditions, and host-country entrepre-neurial activity. Journal of Business Venturing, 35:359–391.
960 AugustAcademy of Management Journal
Knight, F. H. 1921. Risk, uncertainty and profit. NewYork, NY: Houghton Mifflin Company.
Kotha, R., & George, G. 2012. Friends, family, or fools:Entrepreneur experience and its implications for eq-uity distribution and resource mobilization. Journalof Business Venturing, 27: 525–543.
Lee, S.-H., Peng, M. W., & Barney, J. B. 2007. Bankruptcylaw and entrepreneurship development: A real op-tions perspective. Academy of Management Re-view, 32: 257–272.
Lee, T. H., Gerhart, B., Weller, I., & Trevor, C. O. 2008.Understanding voluntary turnover: Path-specific jobsatisfaction effects and the importance of unsolicitedjob offers. Academy of Management Journal, 51:651–671.
Lin, D. Y., & Wei, L. J. 1989. The robust inference for thecox proportional hazards model. Journal of theAmerican Statistical Association, 84: 1074–1078.
Lubinski, D. 2004. Introduction to the special section oncognitive abilities: 100 years after Spearman’s (1904)“‘General intelligence’ objectively determined andmeasured.” Journal of Personality and Social Psy-chology, 86: 96–111.
Maddala, G. S. 1983. Limited-dependent and qualita-tive variables in econometrics. Cambridge, Eng-land; New York, NY: Cambridge University Press.
Majd, S., & Pindyck, R. S. 1987. Time to build, optionvalue, and investment decisions. Journal of Finan-cial Economics, 18: 7–27.
McGrath, R. G. 1997. A real options logic for initiatingtechnology positioning investments. Academy ofManagement Review, 22: 974–996.
McMullen, J. S., & Shepherd, D. A. 2006. Entrepreneurialaction and the role of uncertainty in the theory of theentrepreneur. Academy of Management Review,31: 132–152.
Miner, J. B., & Raju, N. S. 2004. Risk propensity differ-ences between managers and entrepreneurs and be-tween low- and high-growth entrepreneurs: A replyin a more conservative vein. Journal of AppliedPsychology, 89: 3–13.
Mitchell, R. K., Busenitz, L. W., Bird, B., Gaglio, C. M.,McMullen, J. S., Morse, E. A., & Smith, J. B. 2007.The central question in entrepreneurial cognitionresearch 2007. Entrepreneurship Theory and Prac-tice, 31: 1–27.
Moore, D. A., Oesch, J. M., & Zietsma, C. 2007. Whatcompetition? Myopic self-focus in market-entry de-cisions. Organization Science, 18: 440–454.
Muller, W., & Arum, R. 2004. Self-employment dynamicsin advanced economies. In W. Muller & R. Arum(Eds.), The re-emergence of self-employment: Acomparative study of self-employment dynamics
and social inequality: 1–35. Princeton, NJ: Prince-ton University Press.
Narendranathan, W., & Stewart, M. B. 1991. Simplemethods for testing for the proportionality of cause-specific hazards in competing risks models. OxfordBulletin of Economics and Statistics, 53: 331–340.
Nicolaou, N., & Birley, S. 2003a. Academic networks in atrichotomous categorisation of university spinouts.Journal of Business Venturing, 18: 333–359.
Nicolaou, N., & Birley, S. 2003b. Social networks in or-ganizational emergence: The university spinout phe-nomenon. Management Science, 49: 1702–1725.
O’Brien, J., & Folta, T. 2009. Sunk costs, uncertainty andmarket exit: A real options perspective. Industrialand Corporate Change, 18: 807–833.
O’Brien, J. P., Folta, T. B., & Johnson, D. R. 2003. A realoptions perspective on entrepreneurial entry in theface of uncertainty. Managerial and Decision Eco-nomics, 24: 515–533.
Ozcan, S., & Reichstein, T. 2009. Transition to entrepre-neurship from the public sector: Predispositionaland contextual effects. Management Science, 55:604–618.
Patel, P. C., & Thatcher, S. M. B. 2012. Sticking it out:Individual attributes and persistence in self-employ-ment. Journal of Management. Advance onlinepublication.
Petrova, K. 2012. Part-time entrepreneurship and finan-cial constraints: Evidence from the panel study ofentrepreneurial dynamics. Small Business Econom-ics, 39: 473–493.
Preacher, K. J., & Hayes, A. F. 2008. Asymptotic andresampling strategies for assessing and comparingindirect effects in multiple mediator models. Behav-ior Research Methods, 40: 879–891.
Puhani, P. A. 2000. The Heckman correction for sampleselection and its critique. Journal of Economic Sur-veys, 14: 53–68.
Roberts, K., & Weitzman, M. L. 1981. Funding criteria forresearch, development, and exploration projects.Econometrica, 49: 1261–1288.
Santarelli, E., & Vivarelli, M. 2007. Entrepreneurship andthe process of firms’ entry, survival and growth.Industrial and Corporate Change, 16: 455–488.
Schiller, B. R., & Crewson, P. E. 1997. Entrepreneurialorigins: A longitudinal inquiry. Economic Inquiry,35: 523–531.
Schmidt, F. L., & Hunter, J. 2004. General mental abilityin the world of work: Occupational attainment andjob performance. Journal of Personality and SocialPsychology, 86: 162–173.
Shane, S. 2003. A general theory of entrepreneurship:
2014 961Raffiee and Feng
The individual-opportunity nexus. Cheltenham,England; Northampton, MA: E. Elgar.
Shane, S. 2008. The illusions of entrepreneurship: Thecostly myths that entrepreneurs, investors, andpolicy makers live by. New Haven, CT: Yale Uni-versity Press.
Shane, S. 2012. Reflections on the 2010 AMR decadeaward: Delivering on the promise of entrepreneur-ship as a field of research. Academy of Manage-ment Review, 37: 10–20.
Shaver, J. M. 1998. Accounting for endogeneity whenassessing strategy performance: Does entry modechoice affect FDI survival? Management Science,44: 571–585.
Shepherd, D. A. 2003. Learning from business failure:Propositions of grief recovery for the self-employed.Academy of Management Review, 28: 318–328.
Simsek, Z., Heavey, C., & Veiga, J. F. 2010. The impact ofCEO core self-evaluation on the firm’s entrepreneur-ial orientation. Strategic Management Journal, 31:110–119.
Song, S. G. 2010. Mortality consequences of the 1959–1961 Great Leap Forward famine in China: Debilita-tion, selection, and mortality crossovers. Social Sci-ence and Medicine, 71: 551–558.
Sørensen, J. B. 2007. Bureaucracy and entrepreneurship:Workplace effects on entrepreneurial entry. Admin-istrative Science Quarterly, 52: 387–412.
Sørensen, J. B., & Fassiotto, M. A. 2011. Organizations asfonts of entrepreneurship. Organization Science,22: 1322–1331.
Sorenson, O., & Stuart, T. E. 2001. Syndication networksand the spatial distribution of venture capital invest-ments. American Journal of Sociology, 106: 1546–1588.
Stauffer, J. M., Ree, M. J., & Carretta, T. R. 1996. Cogni-tive-components tests are not much more than g: Anextension of Kyllonen’s analyses. Journal of Gen-eral Psychology, 123: 193–205.
Toft-Kehler, R., Wennberg, K., & Kim, P. H. 2014. Practicemakes perfect: Entrepreneurial-experience curvesand venture performance. Journal of Business Ven-turing.
Trigeorgis, L. 1996. Real options: Managerial flexibilityand strategy in resource allocation. Cambridge,MA: MIT Press.
Ucbasaran, D., Westhead, P., & Wright, M. 2009. Theextent and nature of opportunity identification byexperienced entrepreneurs. Journal of BusinessVenturing, 24: 99–115.
Ucbasaran, D., Shepherd, D. A., Lockett, A., & Lyon, S. J.2013. Life after business failure: The process and
consequences of business failure for entrepreneurs.Journal of Management, 39: 163–202.
Vinogradov, E., & Kolvereid, L. 2010. Home country na-tional intelligence and self-employment rates amongimmigrants in Norway. Intelligence, 38: 151–159.
Weitzman, M. L., Newey, W., & Rabin, M. 1981. Sequen-tial R&D strategy for synfuels. Bell Journal of Eco-nomics, 12: 574–590.
Wennberg, K., Folta, T. B., & Delmar, F. 2006. A realoptions model of stepwise entry into self-employ-ment. In A. Zacharakis (Ed.), Frontiers in entrepre-neurship research: 119–132. Babson Park, MA:Babson College.
Wennberg, K., Wiklund, J., DeTienne, D. R., & Cardon,M. S. 2010. Reconceptualizing entrepreneurial exit:Divergent exit routes and their drivers. Journal ofBusiness Venturing, 25: 361–375.
Wozniak, S., & Smith, G. 2006. IWoz: Computer geek tocult icon: How I invented the personal computer,co-founded Apple, and had fun doing it. New York,NY: W. W. Norton & Co.
Wright, M., Westhead, P., & Sohl, J. 1998. Editors’ intro-duction: Habitual entrepreneurs and angel investors.Entrepreneurship Theory and Practice, 22: 5–21.
Yang, T. T., & Aldrich, H. E. 2012. Out of sight but not outof mind: Why failure to account for left truncationbiases research on failure rates. Journal of BusinessVenturing, 27: 477–492.
Zahra, S. A., & George, G. 2002. Absorptive capacity: Areview, reconceptualization, and extension. Acad-emy of Management Review, 27: 185.
Zhao, H., Seibert, S. E., & Hills, G. E. 2005. The mediatingrole of self-efficacy in the development of entrepre-neurial intentions. Journal of Applied Psychology,90: 1265–1272.
Zhao, H., Seibert, S. E., & Lumpkin, G. T. 2010. Therelationship of personality to entrepreneurial inten-tions and performance: A meta-analytic review.Journal of Management, 36: 381–404.
Ziedonis, A. A. 2007. Real options in technology licens-ing. Management Science, 53: 1618–1633.
APPENDIX
Risk Aversion Index Protocol
Income gamble scenarios. First, all participants wereasked the following question (scenario 1):
“Suppose that you are the only income earner in thefamily, and you have a good job guaranteed to giveyou your current (family) income every year for life.You are given the opportunity to take a new andequally good job, with a 50–50 chance that it will
962 AugustAcademy of Management Journal
double your (family) income and a 50–50 chancethat it will cut your (family) income by a third.Would you take the new job?”
If the participant responded “no” to the first question,they were presented with the following (less risky) ques-tion (scenario 2):
“Suppose the chances were 50–50 that it woulddouble your (family) income and 50–50 that itwould cut it by 20 percent. Would you take the newjob?”
If the participant responded “yes” to the first question,they were presented with the following (more risky)question (scenario 3):
“Suppose the chances were 50–50 that it woulddouble your (family) income and 50–50 that itwould cut it in half. Would you still take the newjob?”
Risk aversion index calculation. The index is calcu-lated assuming that risk aversion is a function of ex-pected utility where an expected utility-maximizing in-dividual will choose a 50–50 gamble of doubling lifetimeincome as opposed to having it fall by the fraction 1-� if:
.5U (2c) � .5U(�c) � U(c)
where U represents an individual’s utility function and crepresents permanent consumption. That is, the partici-pant will accept the gamble if the expected utility of thepayoff offered by the gamble exceeds the utility of havingthe current payoff with absolute certainty. We can calcu-late the expected utility that is implicit in each of thehypothetical occupational income gamble questions:
E[scenario 1] � .5(2c) � .5 (.67c) � 1.33c
E[scenario 2] � .5(2c) � .5 (.8c) � 1.4c
E[scenario 3] � .5(2c) � .5 (.5c) � 1.25c
As demonstrated, all three scenarios have an expectedvalue that is greater than the payoff with certainty (i.e.,they are actuarially fair). Thus, we calculate our riskaversion index as the degree to which the expected valueof the payoff must be greater than value of the absolutecertain payoff in order for the respondent to accept thegamble:
Risk Aversion Scenario 1 Scenario 2 Scenario 3 Premium
(4) Strong Reject Reject n/a n/a(3) Moderate Reject Accept n/a .4c(2) Mild Accept n/a Reject .33c(1) Weak Accept n/a Accept .25c
Joseph Raffiee ([email protected]) is a doc-toral candidate in the Management and Human Re-sources Department at the University of Wisconsin–Mad-ison. He received his MBA from the University ofWisconsin–Madison. His research interests lie at the in-tersection of strategy and entrepreneurship, with a par-ticular interest in microfoundations.
Jie Feng ([email protected]) is a doctoral candidate at theWisconsin School of Business, University of Wisconsin–Madison. Her research interests include strategic humanresources and human capital, compensation, and em-ployee turnover.
2014 963Raffiee and Feng