Normative rationality in venture capital financing

Technovation 33 (2013) 255–264

Contents lists available at ScienceDirect

Technovation journal homepage: www.elsevier.com/locate/technovation

Normative rationality in venture capital financing Siri Terjesen a,n, Pankaj C. Patel b, James O. Fiet c, Rodney D’Souza d a

Department of Management and Entrepreneurship, Kelley School of Business, Indiana University, 1309 E. 10th St., Bloomington, IN 47405, USA Department of Marketing and Management, Miller College of Business, Ball State University, Muncie, IN 47306, USA College of Business, University of Louisville, Louisville, KY 40292, USA d Management and Entrepreneurship, Northern Kentucky University, Haile/US Bank College of Business, Highland Heights, KY 41099, USA b c

a r t i c l e i n f o

abstract

Available online 20 December 2012

We examine whether venture capitalists (VCs) make investments based on normative rationality, which is derived from habitual and embedded norms and traditions indicative of a macroculture. Syndication and social and professional relations facilitate the development of shared decision-making frameworks. Using a four step methodology and a unique dataset of 139VC decisions and 82 independent VC assessments of those decisions, we find that the VC industry exhibits collective investment decision-making preferences, reflecting normative rationality. We offer implications for theory, practice, and future research. & 2012 Elsevier Ltd. All rights reserved.

Keywords: Finite mixture regressions Macroculture Normative rationality Venture capital firms Venture capitalists

1. Introduction New technology firms contribute to job creation and economic growth and development (Kirchhoff, 1989; Kirchhoff and Phillips, 1988; Kirchhoff et al., 2007). According to Stouder and Kirchhoff (2004: 352), ‘‘One main critical task facing entrepreneurs is to acquire and manage the resources needed to starty [a venture], especially financial y resources,’’ and venture capital (VC) is one source of funding. Venture capitalists (VCs) also provide human capital and social capital—key resources for firm survival (National Venture Capital Association (NVCA), 2011). In the U.S., venture capitalist (VC)-backed firms account for 12 million jobs and $3.1 trillion in revenue (NVCA, 2011), approximately 11% of private sector employment, and 21% of gross domestic product. VC decisions ultimately affect industry innovation and economic growth (Lerner, 2002; Sorenson and Stuart, 2001), especially in critical sectors such as technology (Chorev and Anderson, 2006; Pandey and Jang, 1996) and life sciences (Platzer, 2009). VC firms frequently work together in syndicates with two or more firms investing in the same or in other investment rounds (Manigart et al., 2006; Tian, 2012), often developing repeated patterns of activities. Scholars have long argued that the pure neoclassical economic rationality perspective is insufficient to explain decision-making (e.g., Kirchhoff, 1994). A large body of theory and empirical research suggests the presence of institutional norms—that is

n

Corresponding author. Tel.: þ1 502 409 0634; fax: þ 1 765 285 5117. E-mail addresses: [email protected] (S. Terjesen), [email protected] (P.C. Patel), fi[email protected] (J.O. Fiet), [email protected] (R. D’Souza). 0166-4972/$ - see front matter & 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.technovation.2012.10.004

that decisions are based on what is considered acceptable or legitimate in a specific environment, as well as on technology and economic criteria (DiMaggio and Powell, 1983). In a decisionmaking context, normative rationality describes those decisions, which are embedded in norms and traditions (Oliver, 1997), and thus may result in almost homogeneous decisions. There is anecdotal evidence that suggests that the VC industry exhibits normative rationality, which dictates how funding decisions are and will be made; however, there are no known investigations of this contention. There is evidence in the finance literature on herding behavior in stock market investments (Kaplan and Schoar, 2005), which suggests some plausibility for normative rationality in highly uncertain decision-making contexts such as VC investments. This research attempts to answers the question: do individual VCs make homogeneous decisions regarding the funding of business plans? A better understanding of how VCs make decisions could guide entrepreneurs when soliciting financial support for their start-ups. If all VCs think and act alike with respect to investment decisions, then an entrepreneur’s time would not be well spent soliciting multiple VCs. Rather, an entrepreneur’s time and resources would be better spent incorporating VC feedback into a plan and then taking the revised plan to another VC. This article proceeds as follows. First, we outline the theoretical background for the research question and discuss how VC decisions reflect normative rationality. Next, we describe our unique primary dataset of 139 business plans that were presented to 82 VCs based on the East and West coasts of the U.S., and the four-step methodology. Following a presentation of the results, we conclude by discussing the limitations and the implications for theory, practice, and future research.

256

S. Terjesen et al. / Technovation 33 (2013) 255–264

2. Theoretical background: VC investment as a context for normative rationality A normative rationality perspective is consistent with the strategy literature on interorganizational macrocultures, which are described as ‘‘relatively idiosyncratic, organization-related beliefs that are shared among top managers across organizations’’ (Abrahamson and Fombrun, 1994: 730). A rich literature describes how managers can develop shared mental models and how decision-making can become routinized in groups and in the industry (Porac and Thomas, 1990). Embedded ties facilitate trust, fine-grained information transfer, and joint problem solving (Uzzi, 1997). Repeated interactions regarding specific decisions lead to collectively developed behavioral patterns. Strong and long-lasting ties foster the development of social rules and reciprocal trust, which, in turn, encourage communication among parties and the creation of routines, collective languages, and a collective culture (Coleman, 1990). Groups routinize their decision-making patterns over time (McClelland, 1984), especially through repeated interactions (Gersick and Hackman, 1990) and this aids sensemaking through continuity and coordination (Weick, 1979). There is anecdotal evidence to suggest that VCs may exhibit normatively rational decision-making, which consists of decisions that are embedded in historical and normative processes. The VC industry places a high value on historical interactions. VCs prefer to interact with individuals with whom they have a history and they know well, e.g., certain entrepreneurs, lawyers, or other VCs (Walske and Zacharakis, 2009). VCs are also more likely to support venture teams with whom they have experienced success in the past (Sorenson and Stuart, 2001). Furthermore, an embedded macroculture develops and maintains VC industry norms. The majority of VC investments often take place in syndicates, which are dense networks that are structurally embedded and enable information to diffuse across boundaries (Sorenson and Stuart, 2001). For example, of the estimated 31,000 firms that received U.S. venture capital from 1980 to 2005, 70% garnered funds from two or more VCs (Tian, 2012). Among VC-backed firms holding an initial public offering (IPO), two or more firms backed 88% of those that received funding (Tian, 2012). Through syndicate investing, VCs develop a web of relationships based on past and current investments (Lerner, 1994), which can lead to normative decision-making. Syndicates have high degrees of reciprocity (Lerner, 1994) and repeat investments (Bygrave, 1988), thus exposing participating VC firms to more deals. In a syndicate, individual VC firms may alternate between lead and non-lead roles over time (Bygrave, 1988), with the lead firm usually contributing the most resources and having larger equity stake (Wright and Lockett, 2003). Through syndication, VCs share knowledge, contacts, and other resources (Bygrave, 1988). Thus syndication allows individual VCs to combine their sector-specific and location-specific investment expertise to help diffuse information across sector boundaries and diversify their portfolios (Sorenson and Stuart, 2001). VC syndicate sanctions include the damaging effects of reputation, withheld deal flow in the future, and the threat of non-investment in subsequent rounds (Wright and Lockett, 2003). Embedded human capital structures facilitate the development of a normative rationality (Oliver, 1997). The embeddedness in the VC industry is also illustrated in the norms related to human capital. The majority of VC firm employees receive MBAs from a handful of premier institutions, namely Harvard, Stanford, MIT, and Wharton (Smart et al., 2000). Furthermore, instruction at these institutions comes from a limited set of experts, e.g., Georges Doriot at MIT (Bancroft, 2009; Roberts and Eesley, 2009).

Key VC employees can be hired away from other VC firms (Bancroft, 2009), facilitating direct knowledge spillover. VC firms hire entrepreneurs with experience working with VCs (Wetfeet, 2010). Also, key VC employees leave established firms to start new ones (Bancroft, 2009; Walske and Zacharakis, 2009). This human capital transfer is institutionalized outside the U.S. For example, U.S. VCs trained VC managers in Asia (Bruton et al., 2005) and established the early VC firms in Europe (Manigart, 1994). Worldwide, comparative studies indicate that the VC industry is increasingly homogeneous in terms of experiential background (Cornelius, 2005). Social and professional relations, such as friendship ties, business clubs, industry associations, and professional and occupational associations facilitate normative decision-making, which occurs by developing shared norms, embedding economic behavior, and facilitating trust (Oliver, 1997). VCs share extensive professional and social ties (e.g., Bancroft, 2009; Shane and Cable, 2002). VCs have high levels of relational embeddedness, which influence their partner selections in inter-firm collaborations (Meuleman et al., 2010). There are numerous professional VC associations at local (e.g., Silicon Valley) and national (e.g., National Venture Capital Association for the U.S., Canadian Venture Capital Association for Canada, European Venture Capital Association for Europe, and Australian Venture Capital Australia for Australia) levels, which enjoy widespread industry support (Bruton et al., 2005: 739) and participation, thus reinforcing norms. Industry homogeneity may also structure homogeneous, industry-level decisions. As examples, individuals working in the VC industry have high degrees of homogeneity in terms of gender (male) (Brush et al., 2004), education, and work experience (Wetfeet, 2010); and these homogenous groups tend to have higher levels of communication and lower levels of conflict (Ancona and Caldwell, 1992), thus reinforcing norms. Investment and hiring practices also reveal preferences for homogeneity: VCs are also less likely to pursue markets that are geographically distant to them (Dimov and De Holan, 2010). Also, VCs prefer entrepreneurial teams with training and professional experience similar to their own human capital (Franke et al., 2006). The above discussion highlights the high levels of interconnectivity in the VC industry (Bygrave, 1988) and suggests the presence of a macroculture and the strong likelihood of normative rationality in decision-making. The high levels of ambiguity and uncertainty in VC investment decisions are also likely to result in evolving cognitive frameworks that can become mutually constitutive (Wright and Lockett, 2003; Weick, 1979). Research indicates that VCs have a limited understanding of their own decisions (Zacharakis and Meyer, 1998); thus prompting the possibility that VCs are imitating other firms rather than making independent, rational decisions. Thus, we expect: Hypothesis: Individual venture capitalists will exhibit homogenous investment decisions when presented with different investment opportunities.

3. Data and analytical approach We gathered data using individual VC investment decisions because syndicate-level designs would have been confounded by syndicate-related factors. This design meets the requirements for a test of normative rationality that determines whether individuals independently make identical decisions after controlling for potential economic rationality (D’Andrade, 1995; Ross, 2004). We collected 70 funded business plans and 69 unfunded business plans as initially submitted to VCs for possible funding (this database is also used in Dos Santos et al., 2011). We asked VCs for unfunded business plans, which had been given due

S. Terjesen et al. / Technovation 33 (2013) 255–264

consideration and which had gone through the due diligence process. This helped assure that the unfunded business plans, which were analyzed, were thorough and complete. Later, we used floor/ceiling analysis to further ensure that unfunded business plans were of comparable quality as funded business plans. Some of the business plans we collected had been modified after an earlier rejection. If our sample would have included revised plans (based on early VC feedback), which had then been accepted by a different VC, we would not have known whether to attribute VC funding success to the business idea or to the VCs’ combined thinking. We eliminated these revised plans. The plans were for technology startups because technology-based industries receive the bulk of VC investment (NVCA, 2011). For each business plan, individual VCs, not a syndicate, made the funding decision. The 70 funded plans came from both the U.S. East coast (28 plans) and West coast (42 plans). Of the 69 unfunded plans, 38 came from East coast VCs and 31 came from West coast VCs. A total of 82 unique VCs (31 from the East coast, 51 from the West coast) provided funding decisions. Because there are various funding sources including banks, private VC firms, and corporate venture capital investors, we control for these differences in our sample by only using private VC firms. Our analysis followed the four steps depicted in Fig. 1: (1) freelisting, (2) focus groups, (3) expert evaluation, and (4) statistical analysis (see White et al., 2004 for a full description of the methodology for Steps 1 and 2).

257

recommended when little is known about a domain because it allows participants to provide information without researcher bias (Weller and Romney, 1988b). Because the free-listing technique is problematic if respondents use different definitions for the same term or use the same term with different meanings, we combined free-listing with a content analysis of existing literature of VC decision-making criteria (e.g., Macmillan et al., 1985; Tyebjee and Bruno, 1984a,b). In total, 22 criteria believed to affect VC funding decisions were identified. This list of terms and definitions was distributed to 120 VCs, angel investors, and commercial lenders at a venture club meeting in a Midwestern city in the US. Angel investors were included because they provide seed capital funding to start-ups that they believe will be successful in obtaining VC funding. Commercial bankers also provide early stage financing and their presence at the venture club meeting suggests an interest in start-up financing (Gonzalez and James, 2007). The 120 individuals were asked to identify which criteria they use to make a decision to invest in a new business and, if necessary, to add new criteria to the list and to define these. Thirty-eight respondents did not fill out the questionnaire and the remaining twenty four responses were incomplete. This resulted in fifty-eight usable responses. Six additional criteria were added, resulting in a total of 28 criteria. We eliminated seven of the 28 criteria because they appeared on few lists.

3.2. Step 2: focus groups 3.1. Step 1: free-listing Criteria used in VC funding decisions were identified by gathering data from 120 VCs and angel investors in the U.S. Midwest. We adopted a form of free-listing data collection, asking respondents for answers that represent pertinent elements about a particular domain (Romney et al., 1986). Free-listing is

We presented the remaining 21 criteria to a focus group (Focus Group A) of 12 individuals (none of whom were represented in Step 1) that included VCs and angel investors from a Midwest city, all of whom were lead investors in over fifty different businesses. We asked Focus Group A to (1) determine whether the terms and definitions of the criteria were consistent, (2)

Analytical Approach Mental Model: Assessing collective cognitive structures (macroculture )

Step 1: Free listing Sample: 120 VCs and angel investors in U.S. Midwest. Outcome: Together with a content analysis of existing literature, 21 criteria important for funding a business plan to be used in Step 2.

Behavioral Model: Assessing decision behavior using collective cognitive structures (macroculture)

Step 2: Focus Groups Sample: Focus Group A of 12 VCs and angel investors in U.S. Midwest, Focus Group B of 15 VCs and angel investors in U.S. Midwest. Outcome: Meaning, categorization, and importance of criteria identified in free?listing. Seven criteria eliminated due to limited role in decision?making, resulting in 14 criteria for decision?making used in Step 3: value added, market size, competition, timing, technology advantages, intellectual property, strategy, start? up experience, industry experience, leadership experience, revenue sales, strategic partners, customer adoption and margin analysis.

Step 3: Expert Evaluation of Business Plans Sample: 70 funded and 69 unfunded business plans of VCs from U.S. West and East coasts. All plans were first? time start?up investment decisions made by a venture capital firm (not as a part of syndicate). 9 experts with extensive VC experience and ~18 years of industry experience. Outcome: Blind evaluation of each plan by 3 experts using criteria and weights from Focus Groups, deriving rating of each plan’s criteria to be used in Steps 4a & 4b.

Fig. 1. Analytical approach.

Step 4: Statistical Analysis Structural Inferences (4a) Consensus Analysis - DV: VC Investment Decision - IV: Evaluations by experts Structural Heterogeneity Inferences (4b) 1) Residual Analysis - DV: VC Investment Decision - IV: Evaluations by experts 2) Finite Mixture regressions: heterogeneity in investment evaluations - DV: VC Investment Decision - IV: Evaluations by experts Industry Heterogeneity Inferences (4c) Heterogeneity in decisions based on VC firm characteristics - DV: VC Investment Decision - IV: VC Team Characteristics

258

S. Terjesen et al. / Technovation 33 (2013) 255–264

weigh each criterion in terms of importance in funding decisions, (3) identify scales to evaluate each criterion, and (4) group the 21 criteria into meaningful categories. We then presented the scales, weights, and definitions for each criterion to another focus group, Focus Group B, of 15 VCs not in group A. We asked Focus Group B to validate/change the criteria and scales and to determine how important each category was in funding decisions. Together with the Focus Group A data, the criteria were weighted for their importance in the VC decision-making process. Because we had access to both funded and unfunded business plans, we used only those criteria that could be verified externally and also modified during contract negotiations, eliminating criteria such as whether the business plans contained financial projections, milestones, and exit strategies (Weller and Romney, 1988a). This resulted in a set of the following 14 criteria used by experts in Step 3: value added, market size, competition, timing, technological advantage, intellectual property protection, strategy, start-up experience, industry experience, leadership experience, revenue sales, strategic partners, customer adoption, and margin analysis. 3.3. Step 3: expert evaluation of business plans Our hypothesis focused on identifying the extent to which criteria used for venture funding are similar across investors. The objective of Steps 1 and 2 was to identify espoused criteria that are commonly agreed upon by investors. In Steps 3 and 4, we assessed whether such agreed-upon criteria are used in decisionmaking. In our design choice, we could have VCs who actually invest in ventures evaluate business plans, or have experts use criteria identified in Steps 1 and 2 to evaluate business plans. Using expert evaluations is necessary in the current context. Our hypothesis posits VC decision-making would exert a ‘fixed’ effect. If VCs were driven by economic rationality, which was contingent on prior experiences and VC firm resources, then it would be a ‘random’ effect. Having an external expert evaluate a business plan indicates the extent to which the ‘fixed’ effect was uniform, based on criteria that are widely used in the industry. Expert evaluations are necessary as they help assess the extent to which decision outcomes are based on commonly held criteria. Although individual VC assessments for each criterion would be invariably different, expert evaluation helps assess the uniformity of the applicability of the criteria based on reaching the same decisions. In step 3, nine experts evaluated the business plans using the criteria and scales developed in the previous two steps. The nine experts worked in a Midwestern city in the US, had experience dealing with VCs and individual investors, and were well grounded in the technology industries. The experts had an average of 18 years of relevant industry experience in communication equipment, industrial electronics, semi-conductors, electromedical equipment, and/or computers and office equipment and 14 years of investment experience as a venture capitalist, angel investor, private investor, and/or serial entrepreneur. None of these experts were involved in the previous steps. See Table 1 for the experts’ profiles. This single blind study employed a balanced sample of funded and unfunded plans, thereby reducing the likelihood that evaluations could be correct by chance. Because most business plans submitted to VCs are not funded, someone with little knowledge of VC funding could be correct most of the time if he/she indicated that each plan fared poorly on all criteria and should not be funded. We specifically instructed experts to base the decision strictly on the business plan content (and not on an assumption about the likely distribution of funded versus unfunded plans) and not to respond if they had prior knowledge of a particular

venture. Furthermore, to limit information from external sources, we removed all identifying items from the business plans (e.g., names of company, management team, strategic partners, customers, and suppliers). We randomly assigned the 139 business plans to the nine experts, with each plan evaluated by three experts to limit individual bias. Each expert evaluated two plans per week over a 23-week period. In addition to rating each plan on the established criteria, we asked each expert to indicate if a plan should be funded. We used the expert rating of each plan’s criteria in steps 4a and 4b. We made sure that the experts did not have more information than what was available in the business plans because the experts were very familiar with the industries and we did not want them to be able to identify the plans from sources other than those that we provided to them. Inter-expert reliability was 0.93. The difference between funded and unfunded business plans based on expert ratings was significant (t-test: po0.01). Furthermore, a logistic regression of expert ratings explained 89% of the variance between funded and unfunded plans. To further explore these findings, we created a composite score using principal component analysis. The reliability was 0.92. Next, we determined whether the ratings for funded and unfunded plans were unevenly distributed to create artificial separation. We tested floor and ceiling effects to assess the uniformity of scale ranges (Nunnally, 1978), ensuring that the floor/ceiling effects were small ( o15% of the sample) and that skewness statistics were between 1 and þ 1. The maximum floor effect was 11.51% (intellectual property: 16/139) and the maximum ceiling effect was 12.94% (market size: 18/139). The skewness values ranged from 0.847 and 0.911, and were within the recommended bounds. 3.4. Step 4: statistical analysis We analyzed all data at the group level to assess the degree of agreement (consensus analysis). The heterogeneity tests (residual analysis and finite mixture regression) focused on the level of disagreement (D’Andrade, 1995). Following consensus and residual analyses, we used finite mixture regressions (FMR) to explore the extent of heterogeneity among VCs’ funding decisions. FMRs independently identify classes without external impositions (McLachlan and Peel, 2000). We assessed the latent classes in the VC evaluations. Despite the high inter-rater reliability and discriminatory power, the expert evaluations might have been unreliable if we had not tested all the VC criteria. We also examined a set of VC firm characteristics and network positions for possible correlation with funding decision patterns. 3.4.1. Consensus analysis and residual agreement The consensus analysis focused on three issues. First, it explored whether a particular ‘cultural model’ or shared knowledge of a particular domain exists among a group of informants (Borgatti, 1996, 1999). In survey research, homogeneity among informant responses indicates consensus. Consensus analysis related reliability testing applies to informants rather than to survey items. Second, it compares the relationship between each informant’s knowledge of a domain (his or her ‘cultural competence’) and the knowledge possessed by the aggregate (Ross, 2004). Cultural competence scores are estimated by factoring a matrix of person-by-person similarity coefficients and are represented as proportions. Thus, an informant with an estimated cultural competence of 0.7 commands 70% of that domain’s knowledge (Borgatti, 1996, 1999). Third, without assuming the correct answers a priori, it solicits the local or culturally correct

S. Terjesen et al. / Technovation 33 (2013) 255–264

259

Table 1 Expert profiles. Expert

Core investment industries

Involvement in investment process

Years of investment experience

Highest education

#1 #2

Communication Equipment (SIC: 3661, 3663, and 3669) Industrial Electronics (SIC: from 3821 to 3826 and 3829) and Semi-conductors (SIC: 3674) Electromedical Equipment (SIC: 3844 and 3845) Computers and Office Equipment (SIC: 3571, 3572, and 3575) Industrial Electronics (SIC: from 3821 to 3826 and 3829) and Semi-conductors (SIC: 3674) Communication Equipment (SIC: 3661, 3663, and 3669) Communication Equipment (SIC: 3661, 3663, and 3669) Electromedical Equipment (SIC: 3844 and 3845) Industrial Electronics (SIC: from 3821 to 3826 and 3829) and Semi-conductors (SIC: 3674)

Venture Capitalist Venture Capitalist

11 13

MBA MBA

Angel investment Venture Capitalist

19 16

PhD Biochemistry MBA

Private Investor/serial entrepreneur

14

PhD Mechanical Engineering

Serial Entrepreneur Venture Capitalist Private Investor/serial entrepreneur Private Investor/serial entrepreneur

12 10 18 15

MS Electrical Engineering PhD Electrical Engineering MBA MS Electrical Engineering

#3 #4 #5 #6 #7 #8 #9

answers to survey questions (Batchelder and Romney, 1988). An anthropologist can reach conclusions about the local understanding of VC investment criteria by interviewing a small number of VCs, observing participants, and analyzing printed material content. However, consensus analysis operationalizes the findings by estimating the social distribution of the knowledge, the average knowledge possessed by each respondent, and the culturally ‘correct’ answers to the questions. Expert informants are individuals with higher cultural competence scores. Experts tend to agree with each other more frequently, creating more homogeneous consensus (D’Andrade, 1995). Consensus analysis weights an informant’s responses according to individual competence (Romney et al., 1986). Using Bayesian posterior probability rather than a majority rule to rate answers ensured that the most competent informants received the most credit for their responses. To assess the level of homogeneity in decision criteria, we conducted a consensus analysis at both ‘mental’ and ‘behavioral’ levels. The mental level represents similarity in decision criteria among VCs from the criteria derived from Steps 1 (free-listing) and 2 (focus-group) in Fig. 1. Although the mental model focuses on preferences for funding, the behavioral level explains how effectively mental models are represented or used in actual funding scenarios. Overall, the mental level represents generally accepted knowledge in the field, whereas, the behavioral model assesses the extent to which the generally accepted decision criteria from the mental model are used. Thus, we are able to obtain robust consensus between the suggested criteria (the mental model) and the actual criteria (the behavioral model). Following Ross (2004), our consensus in mental and behavioral models met the following conditions: (a) the first factor’s eigenvalue is greater than the ratio of 3:1, (b) factor loadings are positive, and (c) the first factor explains significant amounts of variance (at least 80%). Because respondents had a high level of agreement between their mental and behavioral models, we treated the factor loadings as competence scores to be analyzed further (Borgatti, 1996; Ross, 2004). The notion of residual agreement is important for understanding consensus and for studying differences in cultural domains which could influence competence scores. For example, experienced VCs may accept different levels of market and agency risk compared to their less-experienced counterparts (Fiet, 1995). Consequently, VCs from different firms may exhibit varied competence loadings and be immersed in different cultural milieus. If a large number of VCs possessed different investment behaviors, the competence loadings on the second factor could also be explored.

Because patterns can spread across loadings, residual agreement must be calculated separately (Ross, 2004). The residual analysis procedure is based on each individual’s agreement with consensus responses (Nakao and Romney, 1984). We calculated the VC-by-VC residual agreement by subtracting the agreement predicted by the consensus (represented by the first factor loadings) from the observed agreement in the data. We calculated the predicted agreement matrix by multiplying the first factor loadings of each pair of participants, resulting in an index of agreement predicted by each participant’s knowledge of the consensus. Subtracting the predicted agreement matrix from the observed agreement matrix resulted in a matrix of agreement that was not accounted for by the consensus, as represented by the first factor. For example, if a VC agreed on 80% of the responses with the general model and another VC agreed on 70% of the responses for the same model, the predicted agreement between the two would be 56% (¼80%  70%). The predicted agreements for all the pairs of the respective mental and behavioral models produced a matrix of predicted VC agreement on financing decisions. We then standardized the residual agreement matrix, creating a matrix of values between zero and one. We used an OLS regression of each VC firm’s age, size, and portfolio size/fund managed between a dyad of VC firms in the residual matrix.

3.4.2. Finite mixture regressions Employed in fields as diverse as biology, medicine, physics, economics, and marketing, finite mixture regression (FMR) models are used when different groups and their membership cannot be observed. FMR can analyze inter-relationships such as macroculture when group memberships are simply unobservable (McLachlan and Peel, 2000). Another application of mixture models is market segmentation (Wedel and Kamakura, 2001) for which they are considered to be more state-of-the art than traditional cluster analysis and cluster-wise regression. Finite mixture models with a fixed number of components are usually estimated with the expectation-maximization algorithm within a maximum likelihood framework (Dempster et al., 1977) and with Markov Chain Monte Carlo sampling (Diebolt and Robert, 1994) within a Bayesian framework. We used the flexmix package in R for analysis. No inferential technique exists for identifying the number of segments in a whole, such as a macroculture; however a number of indicators may be used to draw inferences about the number of latent groups in a population. Information criteria are based on assessing the degree of improvement in explanatory power

260

S. Terjesen et al. / Technovation 33 (2013) 255–264

adjusted for the degrees of freedom required by additional parameters. We used the Akaike’s information criterion (AIC) and consistent Akaike’s information criterion (CAIC). The lesserknown criterion is Entropy which explains the degree to which segments are sufficiently distinctive and thus accounts for the separation in estimated posterior probabilities. Higher estimation implies a greater degree of separation. Normed Entropy Criteria (NEC(S)) adjusts for over-parameterization relative to a onesegment model: ES ½lnðLðSÞÞlnðLð1Þ N S   P P pns ln pns ES ¼ 1 n ¼ 1 s ¼ 1 N

NECðSÞ ¼

ES is the entropy of segment S, and indicates separation between segment S and the 1-segment solution; pns the posterior probability of subject n in segment s, and N the number of subjects; L(S) the log-likelihood for S-segment and L(1) the log-likelihood of a 1-segment solution. The model tests differences in entropy between different segments, L(S), and 1-segment, L(1). The number of segments retained, Sn in the finite mixture solution is where the entropy between segment S and segment 1 is minimized.

3.4.2.1. FMR for expert ratings. We conducted a FMR on plan evaluations to determine if there were independent groups of funded and unfunded plans. Because expert evaluations are highly discriminatory, we wanted to know if subgroups existed between the funded and unfunded groups identified by the FMR. These results presented a record of actual funding (mental model) rather than a retrospective rationalization of what the VCs said (behavior model) was important to them. 3.4.2.2. FMR for VC firms. We then compared the funding decision with VC firm characteristics and respective network positions using Abell and Nisar’s (2007) criteria. If economic rationality plays a critical role in decision-making, then VC firm characteristics should explain significant variance in differentiating funded and unfunded business plans. In other words, if VCs are driven by idiosyncratic criteria, then based on certain firm characteristics they would be more likely to fund a certain type of business plan over the other. For example, VCs with high IPO rates may be more inclined to fund high potential ventures. Similarly, VCs with larger fund size may not fund ventures with smaller initial funding, or VCs with smaller fund size may not be able to fund ventures with larger initial investments. More importantly, network positions could affect deal and information flow, and therefore, affect identification of investment opportunities and future growth of ventures. For example, VCs with centralized network positions are more likely to focus on ventures with strategic partners to create greater knowledge complementarity; whereas, VC firms with limited experience could rely more on venture team characteristics. We identified 82 VC firm characteristics based on performance data from 1994 to 2004 because the actual funding decisions were made in 2004. The VentureXpert database contains comprehensive information about ventures, buyouts, funds, private equity, firms, executives, portfolio companies, and limited partners and has been used in numerous studies (e.g., Sorenson and Stuart, 2001). We gathered the following VC firm characteristics from Abell and Nisar’s (2007) comprehensive review of past VC firm characteristics and additional network characteristics which were found to be important elsewhere in prior literature: (i) exit rate: the percentage of portfolio companies exited; (ii) IPO rate: the percentage of portfolio

companies sold via IPO; (iii) M&A rate: the percentage of portfolio companies sold via M&A; (iv) dollar exit rate: the percentage of invested $ exited; (v) dollar IPO rate: the percentage of invested $ exited via IPO; (vi) dollar M&A rate: the percentage of invested $ exited via M&A; (vii) book/market ratio: the book/market ratio of public companies in a sample fund’s industry of interest; (viii) VC fund size: the amount of committed capital reported by a VC fund; (ix) venture capital firm experience: the average number of years of VC firm experience in VC industry; (x) partner experience: the average number of years of VC firm partners’ experience in the VC industry; (xi) corporate board director: a dummy variable, which takes the value 1 if a VC firm has or had a seat on the board of directors of a company, 0 otherwise (we normalized each of the following network measures according to their theoretical maximum (e.g., the degree a VC firm could syndicate with other VC firms)); (xii) degree: number of unique VC firms with which a VC firm has syndicated with (regardless of syndicate role); (xiii) indegree: the number of unique VC firms that led syndicates in which a VC firm was a non-lead member; (xiv) outdegree: the number of unique VC firms that have taken part as non-lead investors in syndicates led by a VC firm; (xv) eigenvector: a VC firm’s ‘‘closeness’’ to other VC firms; (xvi) betweenness: the number of the shortest distance paths between other VC firms in a network with which a VC firm interacts. We controlled for factors that might promote heterogeneous investment preferences.

4. Results As described above, we utilized three methodologies to test our hypothesis that individual VCs exhibit homogenous investment decisions. Consensus analysis focuses on between mental models (agreement with espoused decision-making criteria) and behavioral models (the extent to which aspects of a behavioral model are manifested in actual decision-making). Steps 1 and 2 in Fig. 1 relate to mental models in the VC industry. If VCs draw on common decision-making criteria, they must share a mental model. If idiosyncratic VC resources drove VC decision criteria, then a mental model would likely be fragmented because each individual could draw on his or her unique decision-making templates. Based on consensus analysis, Table 2(a) demonstrates that there is a high level of agreement between both the mental and behavioral models. Specifically, among the criteria shared by investors in Step 1 and Step 2, there is a strong consensus in their decision-making criteria. Specifically, for the mental model, the ratio of the first to the second eigenvalue is 6.422, and the first factor explains 81% of the variance. Therefore, the espoused decision-making criteria are strongly shared among 147 investors (120 investors in Step 1 and 27 investors in Step 2). In other words, investors draw on similar decision-making criteria; however, Table 2(a) assesses whether these shared criteria are in fact used in actual decision-making. In the behavioral model, we used VC decisions as both outcome and expert evaluations based on the mental model criteria Table 2(a) Consensus analysis (Step 4(a) in Fig. 1). Model

Ratio of first and second Percentage variance eigenvalue explained by first eigenvalue (%)

Average competence

Mental

6.422:1

81

0.847

Behavioral Funded 5.872:1 Unfunded 6.271:1 Combined 5.556:1

85 87 84

0.885 0.893 0.847

S. Terjesen et al. / Technovation 33 (2013) 255–264

derived from Step 1 and Step 2. As listed in Table 2(a), the ratio of first and the second eigenvalue for the overall behavioral model is 5.556 to 1, which indicates that the first factor explains 84% of variance. This suggests that the mental model criteria are indeed used in actual decision-making. We further assessed whether these criteria were used when business plans were funded and when they were not funded. The ratio of the first and second eigenvalues for the funded business plans was 5.872 to 1; and for unfunded business plans, it was 6.271 to 1. In other words, the funding criteria were shared consistently across funded and unfunded business plans. Next, we assessed the average competence shared between the mental and behavioral model. Recall that average competence scores are based on the factor loadings of the criteria in the mental and behavioral models (funded, unfunded, and combined)—higher scores indicate higher levels of knowledge shared among the participants (Borgatti, 1996; Ross, 2004). High loadings and a significant difference with the other factors indicate that VCs share commonly accepted cultural knowledge about their investment domains. At the 99% confidence level, the average competence for the mental model was 84.7% and for the behavioral model it was 88.5% (funded plans), 89.3% (unfunded plans), and 84.7% (combined pool of funded and unfunded plans). Again, overall, the mental and behavioral models show that investors not only share common espoused criteria (mental model), but when faced with real life decisions they also use decision criteria as espoused (behavioral model). While Table 2(a) focuses on shared criteria, we further explored whether investor firm characteristics drove ‘unshared’ (or, residual) variance. If investor firm characteristics were significantly related to residual factors, in addition to shared criteria, idiosyncratic factors could drive decisions making. Our residual analysis (Table 2(b)) shows that VC firm characteristics do not significantly explain the residual, supporting the conclusion that there is no sub-group effect in VC decision-making. Because mental and behavioral models indicate that investors share espoused criteria and rely on these criteria when making decisions, FMR enables us to further test whether the weights on individual criteria for funding and not-funding a business plan are consistent. FMR allows us to start with no presumption about the underlying distribution of outcomes (funded/unfunded) or the distribution of criteria across funded and unfunded business plans. As shown in Table 3(b), without imposing criteria on underlying groups in the sample, we found that a two-segment solution provides the best fit. In other words, the two-segment solution had the lowest log-likelihood (  119.783), AIC (120.604), CAIC (129.627), and the highest entropy (0.920), NEC(S) (0.365), and R2 (0.614). Next, based on the two identified segments, we assessed business plan membership to each. Matching the segments with the actual funding decisions revealed a classification of 65 funded plans (compared to 70 actually funded by VCs), and 67 unfunded business plans (compared to 69 actually unfunded by VCs). In summary these results indicate a high degree of agreement (six misclassifications out of 139) with the actual VC decisions.

Table 2(b) Consensus analysis. Residual analysis on VC consensus model (Step 4(b), in Fig. 1). Mental model

Behavioral model

D Age (years) ln (D portfolio size [$ million]) ln (D portfolio size/fund manager [$ million])

0.154 0.232 0.178

0.110 0.085 0.119

R2

0.142

0.117

261

Table 3(a) Finite mixture regression for business plan evaluations (Step 4(b) in Fig. 1). Groups based on finite mixture models. Number of segments

Likelihood AIC CAIC Entropy NEC(S) R2

1

2

3

4

5

 141.659 137.188 154.888 – – 0.424

 119.783 120.604 129.627 0.920 0.365 0.614

 165.464 143.391 175.975 0.688 0.249 0.672

 156.821 140.282 163.236 0.785 0.227 0.687

 175.354 144.951 184.497 0.734 0.190 0.704

Note: classification based on the two segment solution for funded and unfunded business plans was 69 funded and 74 unfunded, thus indicating high discriminatory power of the segment.

Table 3(b) Finite mixture regression for business plan evaluations (Step 4(b) in Fig. 1). Finite mixture estimation. Two-segment solution Segment 1: funded Value added Market size Competition Timing Technology advantages Intellectual property Strategy Start-up experience Industry experience Leadership experience Revenue sales Strategic partners Customer adoption Margin analysis Log-likelihood AIC CAIC R2

n

0.17 0.81 0.65n 0.31 0.53 0.32 0.42n 0.36 0.32 0.70 0.25 0.37n 0.43n 0.14n

Segment 2: unfunded 0.53n 0.05n 0.12n 0.01 0.09 0.00 0.02n 0.65n 0.33nn 0.40n 0.04 0.07 0.03 0.04n

 122.62 119.37 128.52 0.61

Note: n

p o.01. po .05.

nn

Next, because only five funded and two unfunded business plans were misclassified (5.04% misclassification), we assessed the relevance of funding criteria for each segment in Table 3(b). We found that value added, competition, strategy, and margin analysis were central to both funded and unfunded business plans. However, the presence of strategic partners (b ¼0.37, po0.05) and customer adoption (b ¼0.43, p o0.05) were also critical to funding. However, limited market size (b ¼0.05, po0.05), lack of start-up capital (b ¼0.65, po0.05), industry (b ¼0.33, p o0.01), and leadership experiences (b ¼0.40, po0.05) could lower the possibility of funding. Overall, although funded and unfunded business plans shared some common criteria, lack of experience and limited market size reduced the likelihood of funding, whereas, the presence of strategic partners and customer adoption were central to receiving funding. In the next step, using the 16 VC firm characteristics identified previously, we ran a finite mixture model (Table 3(a)) and found no distinct groups. The log-likelihood ( 112.986), AIC (  21.374), and CAIC (258.450) were the lowest for one group solution. As shown in Table 4(a), the Entropy and NEC(S) coefficients were not so different as to be able to infer the presence of any particular number of segments. One group solution added additional credence to the relevance of normative rationality. Because

262

S. Terjesen et al. / Technovation 33 (2013) 255–264

Table 4(a) Finite mixture and logistic regression of VC firm characteristics (Step 4(c) in Fig. 1). Groups based on finite mixture models. Number of segments

Likelihood AIC CAIC Entropy NEC(S) R2

1

2

3

4

5

 112.986 219.374 258.450

 143.020 221.116 263.031 0.564 0.052 0.553

 151.533 221.426 281.359 0.519 0.068 0.622

 160.600 229.438 279.612 0.533 0.038 0.674

 165.227 231.691 292.141 0.543 0.043 0.755

0.524

Note: lower levels of AIC and CAIC are desirable; higher levels of entropy and NEC(S) indicate a greater degree of separation among groups.

Table 4(b) Finite mixture and logistic regression of VC firm characteristics (Step 4(c) in Fig. 1). Logistic regression.

b

(s.e.)

Exit rate IPO rate M&A rate Dollar exit rate Dollar IPO rate Dollar M&A rate Book/market ratio VC fund size VC experience Partner experience Corporate board director Degree Indegree Outdegree Eigenvector Betweenness

0.067 0.154 0.221 0.100 0.042 0.031 0.051 0.014 0.223 0.024 0.013 0.011 0.022 0.071 0.038 0.211

(0.073) (0.254) (0.182) (0.094) (0.076) (0.044) (0.062) (0.016) (0.348) (0.052) (0.016) (0.015) (0.016) (0.038) (0.049) (0.148)

Pseudo-R2

0.153

VC firm characteristics were not fundamentally distinct when comparing funded and unfunded business plans, firm specific characteristics did not seem to drive the funding and non-funding decisions in our sample. In other words, learning in VC firms driven by exit rate, IPO rate, M&A rates, and network positions did not distinguish actual funding criteria that were used. As an added caution, we ran a logistic regression to identify any differences in funding patterns and found no significant difference in funding based on sixteen VC firm characteristics (see Table 4(b)). We found that VC firm characteristics played insignificant roles in funding decisions. Past exit rate (b ¼0.067, p 40.10), IPO rate (b ¼0.154, p 40.10), and M&A rate (b ¼0.221, p 40.10) did not affect funding outcomes. Similarly, the dollar values of exit, IPO, and M&A rate did not affect funding outcomes either. VC firm fund size (b ¼0.014, p 40.10) and book to market ratio (b ¼0.051, p 40.10), indicative of VC firm value, did not affect funding criteria. Finally, the coefficients for VC experience (b ¼0.223, p 40.10), partner experience (b ¼0.024, p 40.10), and all the network positions were insignificant. Overall, consensus analysis indicated that investors shared a mental model and use similar funding criteria when making actual funding decisions. Furthermore, using an unsupervised approach under FMR, criteria differentiate funded from unfunded business plans. More importantly, although VCs used some criteria to differentiate funded from unfunded business plans, they focused on venture team experience when not funding a business plan. They placed greater weight on strategic partners

and customer adoption when funding a business plan. Based on unique firm resources, VC firm characteristics could drive funding decisions. However, neither FMR nor logistic regression showed that firm characteristics drove funding decisions.

5. Discussion Our finding that VCs exhibit homogeneous investment decisions implies that norms pertaining to business plan evaluation are readily diffused throughout the VC industry, leading to the standardization of knowledge and the unquestioned acceptance of strategic recipes among industry incumbents (Spender, 1989). Although we have not delved into differences, these strategic recipes may in time aggregate into macrocultures (Abrahamson and Fombrun, 1994), which may dictate how venture decisions across an industry come to resemble each other. To date, common understandings of the qualities of successful business plans have been mostly built on years of anecdotal evidence. In such a case, industry knowledge can serve incumbent firms well. However, strategic innovations may affect a firm’s viability: an incumbent’s deviation from accepted business plan evaluation practices could lead to a decline in performance. Another theoretical implication of this research relates to the likely importance of a shared VC learning capability. Specifically, if as expected a common understanding of the essential attributes of business plans will well serve VC firms (or at least those VC firms focused on similar investment opportunities), then being able to quickly and accurately acquire relevant industry knowledge is a key success requirement among VCs. Industry experience, both at the firm level and at the individual plan evaluator level, is critical as knowledge is sticky. The ability to acquire more knowledge is a function of the amount of knowledge already possessed (Cohen and Levinthal, 1990). A final theoretical implication of this research is the explicit nature of the industry knowledge that drives business plan evaluations within the VC industry. Several researchers suggest that the tacit knowledge embedded in a firm’s specific routines is the type of knowledge that is most likely to lead to long term success (Nelson and Winter, 1982; Nonaka and Takeuchi, 1995). However, the suggestion from the current study that plan evaluation best practices are easily known implies that tacit knowledge may not play a large role in determining why VC firms choose different investment opportunities. This is not to suggest that tacit knowledge will not or cannot be an important component of VC firm success; however, such knowledge probably creates value for VC firms through activities and in skill-based areas other than those pertaining to proficiency in evaluating business plans. Taken together, we expand the management/entrepreneurship branch of VC literature (Cornelius and Persson, 2006). Our study also offers important implications for practicing entrepreneurs and VCs. First, as our study indicates that VCs tend to rate new ventures similarly, an entrepreneur who fails to receive VC backing for his/her plan should incorporate VC feedback into a plan before bringing the plan to the next VC. From a VC perspective, almost homogeneous thinking suggests that some entrepreneurs’ ideas may be universally rejected. Second, the results from finite mixture estimation shows that funded and non-funded business plans do not share a common set of decision criteria. Lack of startup, industry, and leadership experience and limited market size lead to the rejection of business plans, but their presence does not necessarily increase the chances of funding. This finding suggests that human capital and market potential form the threshold criteria ‘‘to get a foot in the door.’’ In other words, an entrepreneur with the best idea may still not

S. Terjesen et al. / Technovation 33 (2013) 255–264

receive funding if he/she lacks the requisite human capital. Once the threshold criteria is met, factors that uniquely drive funding are strategic partners and product adoption by customers. The results suggest that VCs use funding criteria in a sequential pattern. Thus entrepreneurs must highlight their human capital to receive further consideration from the VCs. Third, for the VC industry, the findings suggest that shared decision criteria increases legitimacy and lowers uncertainty and ambiguity, but could limit the unique ‘random’ effects from VC firm resources that help VC firms to establish a distinct competitive advantage vis-a -vis portfolio investments. While prior research focused on cooperation among VC firms, our results indicate that VCs must manage tradeoffs between external learning (from the industry) with internal learning in order to develop unique knowledge that can be leveraged to ‘‘hit more home runs’’ than competitors. Fourth, in addition to sharing risk through syndicates, VCs could assign specialization roles among members to increase cognitive efficiency and develop more effective learning in uncertain investment environments. Fifth, there is increasing evidence of a ‘‘herd mentality’’ in VC investments. As normative rationality would further catalyze such herding behavior, building from the earlier discussion, focusing on internal firm resources and knowledge could help VCs make more informed investment decisions. The above implications should be considered with respect to four limitations. First, although free listing is a widely accepted technique in anthropology, it may be limited in eliciting the actual attributes used for VC decision-making. Second, our inferences on common normative frameworks are based on the technology industries and may not be generalizable outside this context. Third, the norms measured in this study are strictly based on investment preferences. Norms in due diligence, information seeking, and resource configuration may also be possible. Fourth, our study design did not enable us to examine normative rationality vis-a -vis another type of decision-making, for example neoclassical economic rationality.

6. Future research directions In addition to the directions stemming from the above limitations, the present study suggests a number of promising directions for future research. First, future studies could explore how knowledge sharing and social interactions facilitate knowledge convergence. Specifically, researchers could investigate whether certain experiences (such as working together in the same firm or studying on the same MBA program) are more likely to lead to homogeneous thinking. Along this line, researchers could investigate syndication networks in more detail, particularly the socialization and knowledge-sharing processes. Second, as the VC industry varies across countries (e.g., Pandey and Jang, 1996; Wonglimpiyarat, 2007) and there are vast differences in other aspects of the national institutional environment, future researchers could examine normative rationality in other national VC environments. This line of work would extend Kirchhoff’s significant interest in the phenomenon of entrepreneurship across countries (Davidsson et al., 2002). Third, future research could incorporate longitudinal exploration, examining and the evolution of macroculture over time. In doing so, researchers would further Kirchhoff’s calls to explore how thinking may evolve from pure neoclassical economic rationality (e.g. Kirchhoff, 1994). Fourth, we discussed how macroculture is driven at individual, firm, and industry levels. Over time, individual learning is embedded into firm related routines and processes which, in turn, could drive convergence toward shared knowledge, beliefs, and values. As firms in an industry become increasingly interdependent they are more likely to share markets and common resource bases, which

263

could drive common managerial schemas and frameworks. Although we cannot test these multilevel effects due to data limitations, future research could focus on an omnibus test of macroculture driven by shared belief and values at individual, firm, and industry levels (we thank an anonymous reviewer for this suggestion). Fifth, we observe that entrepreneurship research borrows theories from economics, sociology, and finance to explain the phenomenon. Kirchhoff (1994) was one of the first to question the underlying assumptions of mainstream economic theories and their relevance in entrepreneurship research. Continuing in the spirit of Kirchhoff’s (1994) call, we believe that studies in entrepreneurship could benefit by exploring both mainstream theories and more contextually relevant theories.

7. Conclusion Beyond the economic and social embeddedness arguments of venture finance, normative rationality influences venture financing criteria. If rational financing processes or social embeddedness processes affect venture finance, then one might expect considerable heterogeneity in values and behaviors due to the obvious variety in VC firm characteristics. However, in this study, we observe homogeneity in business plan values and VC firm characteristics. Thus, we find evidence that a macroculture prevails in the VC industry, which entails VCs sharing values and behaviors in the decision-making process. References Abell, P., Nisar, T.M., 2007. Performance effects of venture capital firm networks. Management Decision 45 (5), 923–936. Abrahamson, E., Fombrun, C.J., 1994. Macrocultures: determinants and consequences. Academy of Management Review 19 (4), 728–755. Ancona, D.G., Caldwell, D.F., 1992. Bridging the boundary, external activity and performance in organizational teams. Administrative Science Quarterly 37 (4), 634–666. Bancroft, 2009. Early Bay Area Venture Capitalists: Shaping the Business and Industrial Landscape. Oral History Project. Bancroft Library, Regents of the University of California. Available at: /http://bancroft.berkeley.edu/ROHO/ projects/vc/S. Batchelder, W.H., Romney, A.K., 1988. Test theory without an answer key. Psychometrika 53 (1), 71–92. Borgatti, S.P., 1996. ANTHROPAC 4.98 X. Analytic Technologies, Columbia, SC. Borgatti, S.P., 1999. Elicitation techniques for cultural domain analysis. In: Schensul, J., LeCompte, M. (Eds.), The Ethnographer’s Toolkit, vol. 3. Altamira, Walnut Creek, pp. 115–151. Brush, C.G., Carter, N.M., Gatewood, E.J., Greene, P.G., Hart, M., 2004. Gatekeepers of venture growth: a Diana project report on the role and participation of women in the venture capital industry, Kauffman Foundation, Kansas City, MO. Bruton, G., Fried, V., Manigart, S., 2005. Institutional influences on the worldwide expansion of venture capital. Entrepreneurship Theory and Practice 29 (6), 737–760. Bygrave, W.D., 1988. The structure of the investment networks of venture capital firms. Journal of Business Venturing 3 (2), 137–158. Chorev, S., Anderson, A.R., 2006. Success in Israeli high-tech start-ups: critical factors and process. Technovation 26, 162–174. Cohen, W., Levinthal, D., 1990. Absorptive capacity: a new perspective on learning and innovation. Administrative Science Quarterly 35 (1), 128–152. Coleman, J.S., 1990. Foundations of Social Theory. Harvard University Press, Cambridge. Cornelius, B., 2005. The institutionalization of venture capital. Technovation 25, 599–608. Cornelius, B., Persson, O., 2006. Who’s who in venture capital research. Technovation 26, 142–150. D’Andrade, R., 1995. The Development of Cognitive Anthropology. Cambridge University Press, Cambridge. Davidsson, P., Kirchhoff, B., Hatemi-J, A., Gustavsson, H., 2002. Empirical analysis of business growth factors using Swedish data. Journal of Small Business Management 40 (4), 332–349. Dempster, A.P., Laird, N.M., Rubin, D.B., 1977. Maximum likelihood from incomplete data via the EM algorithm. Journal of the Royal Statistical Society 39 (1), 1–38. Diebolt, J., Robert, C.P., 1994. Estimation of finite mixture distributions through Bayesian sampling. Journal of the Royal Statistical Society. Series B. Methodological 56 (2), 363–375.

264

S. Terjesen et al. / Technovation 33 (2013) 255–264

DiMaggio, P., Powell, W., 1983. The iron cage revisited: institutional isomorphism and collective rationality in organizational fields. American Sociological Review 48 (2), 147–160. Dimov, D., De Holan, P., 2010. Firm experience and market entry by venture capital firms (1962–2004). Journal of Management Studies 47 (1), 130–161. Dos Santos, B.L., Patel, P.C., D’Souza, R.R., 2011. Venture capital funding for information technology businesses. Journal of the Association for Information Systems 12 (1), 57–87. Fiet, J.O., 1995. Risk avoidance strategies in venture capital markets. Journal of Management Studies 32 (4), 551–573. Franke, N., Gruber, M., Harhoff, D., Henkel, J., 2006. What you are is what you like: Similarity biases in venture capitalists’ evaluations of start-up teams. Journal of Business Venturing 21 (6), 802–826. Gersick, C.J.G., Hackman, J.R., 1990. Habitual routines in task-performing groups. Organizational Behavior and Human Decision Processes 47 (1), 65–97. Gonzalez, L., James, C., 2007. Banks and bubbles: how good are bankers at spotting winners? Journal of Financial Economics 86 (1), 40–70. Kaplan, S.N., Schoar, A., 2005. Private equity performance: returns, persistence, and capital flows. Journal of Finance 60 (4), 1791–1823. Kirchhoff, B.A., 1989. Creative destruction among industrial firms in the United States. Small Business Economics 1 (3), 161–173. Kirchhoff, B.A., 1994. Entrepreneurship and Dynamic Capitalism: The Economics of Business Firm Formation and Growth. Praeger, Westport. Kirchhoff, B.A., Newbert, S.L., Hasan, I., Armington, C., 2007. The influence of university R&D expenditures on new business formations and employment growth. Entrepreneurship Theory and Practice 31 (4), 543–559. Kirchhoff, B.A., Phillips, B.D., 1988. The effect of firm formation and growth on job creation in the United States. Journal of Business Venturing 3 (4), 261–272. Lerner, J., 1994. The syndication of venture capital investments. Financial Management 23 (3), 16–27. Lerner, J., 2002. Boom and bust in the venture capital industry and the impact on innovation. Federal Reserve Bank of Atlanta Economic Review 4, 25–39. Macmillan, I.C., Siegel, R., Narasimha, P.N.S., 1985. Criteria used by venture capitalists to evaluate new venture proposals. Journal of Business Venturing 1 (1), 119–128. Manigart, S., 1994. The founding rate of venture capital firms in three European countries: 1970-1990. Journal of Business Venturing 9 (6), 525–541. ¨ Manigart, S., Lockett, A., Meuleman, M., Wright, M., Landstrom, H., Bruining, H., Desbrie res, P., Hommel, U., 2006. Venture capitalists’ decision to syndicate. Entrepreneurship Theory and Practice 30 (2), 131–153. McClelland, C., 1984. The Development and Persistence of Routinized Group Decision-Making. Unpublished Master’s Thesis. Purdue University, Lafayette. McLachlan, G., Peel, D., 2000. Finite Mixture Models. John Wiley and Sons, New York. Meuleman, M., Lockett, A., Manigart, S., Wright, M., 2010. Partner selection decisions in interfirm collaborations: the paradox of relational embeddedness. Journal of Management Studies 47 (6), 995–1019. Nakao, K., Romney, A.K., 1984. A method for testing alternative theories: an example from English kinship. American Anthropologist 86, 668–673. National Venture Capital Association (NVCA), 2011. Venture Industry Performance. Nelson, R.R., Winter, S.G., 1982. An Evolutionary Theory of Economic Change. Belknap Press of Harvard University Press, Cambridge. Nonaka, I., Takeuchi, H., 1995. The knowledge-creating company. Oxford University Press, New York. Nunnally, J.C., 1978. Psychometric theory second ed. McGraw-Hill, New York.

Oliver, C., 1997. Sustainable competitive advantage: combining institutional and resource-based views. Strategic Management Journal 18 (9), 697–713. Pandey, I.M., Jang, A., 1996. Venture capital for financing technology in Taiwan. Technovation 16 (9), 499–514. Platzer, M., 2009. Patient Capital: How Venture Capital Investment Drives Revolutionary Medical Innovation. White paper. Porac, J.F., Thomas, H., 1990. Taxonomic mental models in competitor definition. Academy of Management Review 15 (2), 224–240. Roberts, E.B., Eesley, C., 2009. Entrepreneurial Impact: The Role of MIT. MIT and Kauffman Foundation, Kansas City, MO. Romney, A.K., Weller, S.C., Batchelder, W.H., 1986. Culture as consensus: a theory of culture and informant accuracy. American Anthropologist 88 (2), 313–338. Ross, N., 2004. Culture and Cognition: Implications for Theory and Method. Sage, Thousand Oaks. Shane, S., Cable, D., 2002. Network ties, reputation, and the financing of new ventures. Management Science 48 (3), 364–381. Smart, G., Payne, W., Yuzaki, H., 2000. What makes a successful venture capitalist? Journal of Private Equity Fall 3 (4), 7–29. Sorenson, O., Stuart, T.E., 2001. Syndication networks and the spatial distribution of venture capital investments. American Journal of Sociology 106 (6), 1546–1588. Spender, J.C., 1989. Industry Recipes: An Enquiry Into the Nature and Sources of Managerial Judgment. Basil Blackwell, Oxford. Stouder, M., Kirchhoff, B., 2004. Funding the first year of business. In: Gartner, W., Shaver, K., Carter, N., Reynolds, P. (Eds.), Handbook of Entrepreneurial Dynamics. Sage, Thousand Oaks, pp. 71–352. Tian, X., 2012. The role of venture capital syndication in value creation for entrepreneurial firms. Review of Finance 16 (1), 245–283. Tyebjee, T.T., Bruno, A.V., 1984a. A model of venture capitalist investment activity. Management Science 30 (9), 1051–1066. Tyebjee, T.T., Bruno, A.V., 1984b. Venture capital: investor and investee perspectives. Technovation 2, 185–208. Uzzi, B., 1997. Social structure and competition in inter-firm networks: the paradox of embeddedness. Administrative Science Quarterly 42 (1), 36–67. Walske, J.M., Zacharakis, A., 2009. Genetically engineered: why some venture capital firms are more successful than others. Entrepreneurship Theory and Practice 33 (1), 297–318. Wedel, M., Kamakura, W.A., 2001. Factor analysis with (mixed) observed and latent variables in the exponential family. Psychometrika 66 (4), 515–530. Weick, K.E., 1979. The Social Psychology of Organizing. Addison, Reading, MA. Weller, S.C., Romney, A.K., 1988a. Systematic Data Collection. Sage, Beverly Hills. Weller, S.C., Romney, A.K., 1988b. Systematic Data Collection. Qualitative Research Methods. Sage, Newbury Park. Wetfeet, 2010. Guide to Venture Capital Careers. Wetfeet, Philadelphia. White, R.J., Hertz, G.T., D’Souza, R.R., 2004. Applying the stewards of place model: integrating teaching, research and service through entrepreneurship education. Journal of Entrepreneurship Education 7 (1), 95–105. Wonglimpiyarat, J., 2007. Management and governance of venture capital: a challenge for commercial bank. Technovation 27, 721–731. Wright, M., Lockett, A., 2003. The structure and management of alliances: syndication in the venture capital industry. Journal of Management Studies 40 (8), 2073–2102. Zacharakis, A.L., Meyer, G.D., 1998. A lack of insight: do venture capitalists really understand their own decision process? Journal of Business Venturing 13 (1), 57–76.