Value-Added Activities in Venture Capital and the Effects of Stage: An Empirical Study

Bachelor Thesis
B.Sc. International Business Copenhagen Business School

May 2013

Written by
Martin Franqois Lucas Collignon Paul Joachim Brejnholt Satchwell Anders Philip Skovsgaard Valentin

Supervised by
Evis Sinani, Associate Professor, PhD Department of International Economics and Management

Character count: 167,327 | 73.5 standard pages

[PAGE INTENTIONALLY LEFT BLANK]

ii

iii

[PAGE INTENTIONALLY LEFT BLANK]

iv

Acknowledgments:
Special thanks to: Ted Zoller Director of the Centre for Entrepreneurial Studies at the University of North Carolina at Chapel Hill, Senior Fellow at the Ewing Marion Kauffman Foundation Mai-Britt Zocca Senior Program Manager at Copenhagen Bio Science Park Program Director of EL2 Knowledge Lab CEO & Founder of OncoNOx ApS Jesper Knudsen Senior Business Consultant & Partner at Accelerace Jesper Roested Investment Director at VF Venture Frank Knudsen Investment Director at SEED Capital Magnus Corfitzen Investment Director at Sunstone Capital Christian Wylonis Venture Capital Associate at Creandum Rasmus Toft-Kehler PhD Student at the Copenhagen Business School Co-founder and CEO at Synercure ApS Anders Hoffmann Deputy Director General at Danish Business Authority Nicolai Hjer Nielsen Serial entrepreneur Business angel External Associate Professor at the Copenhagen Business School Kristoffer Boye Astrup Chief Consultant at Ministry for Business and Growth Denmark Terttu Luukkonen Chief Research Scientist, Research Institute of Finnish Economy

Furthermore we are grateful for the contributions of: Leonora Bech, Fabio Bertoni, Joel Eriksson Enquist, Peter Thorlund Haahr, Martin Hauge, Gregers Kronborg, Lars Nordal Jensen, Kristine Leary, Anne Birgitte Lundholt, Simone Louise Petersen, Tine Thygesen, Peter Torstensen, as well as founders, CEOs, and entrepreneurs who took the time to respond to our survey.

[PAGE INTENTIONALLY LEFT BLANK]

vi

EXECUTIVE SUMMARY
This thesis explores how Venture Capital funds value-added activities differ between stage and if the Venture Capital funds specialize accordingly. Departing from the value-add perspective of Venture Capital, in which funds transfer knowledge through the relationships with their portfolio companies, this empirical study surveys 275 portfolio companies from Danish-based Venture Capital funds operating within ICT or Life Science. The survey consists of 28 questions on the perceived contribution of the VC funds in various value-added activities, as well as a maximum of 37 questions on the characteristics of the firm, the founders, the deal, and the level of interaction with the fund. With 69 usable responses, the effect of the stage on the portfolio companies perceptions is analysed through a factor analysis which constructs eight factors from the 28 questions on perceived value-add to be used as dependent variables in an ordinary least-squares regression. We construct two models using backward selection: First a model including independent variables for company stage, industry, age, founder experience, prior funding, the size of investment, and a factor for the degree of interaction between the firm and the fund. Thereafter, we construct an identical model which additionally includes nine dummy variables for the individual funds in our sample which assesses the specialization of funds by modelling whether the effect of stage from model one is removed by the funds. We find that the perceived importance of value-added activities differs across portfolio company stages for factors of strategy, market position, and credibility, while we find no difference across stages for factors of technology, professionalization, finance, internationalization, or exit orientation. We also find that Venture Capital funds specialize in stage with respect to valueadded activities within strategy, market position, and credibility. Finally, we find that increasing the amount of interaction between the Venture Capital fund and its portfolio company results in a higher perceived contribution of the fund, all else equal.

vii

[PAGE INTENTIONALLY LEFT BLANK]

viii

Table of Contents
1. Introduction............................................................................................................................................. 5 2. Literature Review .................................................................................................................................... 7 2.1 Perspectives on Venture Capital: Financial Intermediary versus Value-Add .......................... 8 2.1.1 Financial Intermediary Perspective ........................................................................................ 8 2.1.2 Value-Added Perspective ......................................................................................................... 9 2.2 Knowledge Transfer and Value-Add from the Venture Capital Fund................................... 10 2.3 Fund Level Determinants of Value-Added Activities .............................................................. 11 2.4 Specializations of Venture Capital Funds ................................................................................... 12 2.5 Effects of Stage Specialization ..................................................................................................... 13 2.6 Specific Value-Added Activities and Hypothesis Formation .................................................. 16 2.6.1 Strategic Value-Add ................................................................................................................ 17 2.6.2 Financial Value-Add ............................................................................................................... 18 2.6.3 Credibility Value-Add ............................................................................................................. 19 2.6.4 Professionalization Value-Add.............................................................................................. 20 2.6.5 Marketing Value-Add ............................................................................................................. 21 2.6.6 Technological Value-Add ...................................................................................................... 22 2.6.7 Internationalization Value-add .............................................................................................. 23 2.6.8 Exit Orientation Value-Add .................................................................................................. 24 2.6.9 Fund Stage Specialization ...................................................................................................... 25 3. Methodology ......................................................................................................................................... 26 3.1 Sample Selection ............................................................................................................................. 26 3.1.1 Geography ................................................................................................................................ 27 3.1.2 Institution Type ....................................................................................................................... 27 3.1.3 Industry..................................................................................................................................... 28 3.1.4 Sampling Error ........................................................................................................................ 29 3.2 Survey Construction....................................................................................................................... 30 3.2.1 Value-Added Activities........................................................................................................... 31 3.2.3 Survey Errors ........................................................................................................................... 36 3.3 Data Collection ............................................................................................................................... 37 3.3.1 Format of Survey and Email ................................................................................................. 37 3.3.2 Obtaining Contact information ............................................................................................ 37

Page 1 of 108

3.3.3 Data Collection Errors ........................................................................................................... 39 3.4 Data Correction .............................................................................................................................. 39 3.5 Methods for Data Analysis ........................................................................................................... 41 3.5.1 Variable Construction and Data Treatment ........................................................................ 41 3.5.2 Analysis Methods for Factor Analysis ................................................................................. 43 3.5.3 Analysis Methods for OLS Regression ................................................................................ 45 3.5.4 Including Fund Dummies ...................................................................................................... 47 3.5.5 Analysis Errors ........................................................................................................................ 47 4. Results .................................................................................................................................................... 49 4.1 Factor Analysis................................................................................................................................ 49 4.1.1 Factor Analysis of Dependent Variables ............................................................................. 49 4.1.2 Factor Analysis of Interaction Term .................................................................................... 52 4.2 Regression Analysis ........................................................................................................................ 52 4.2.1 Descriptive Statistics .............................................................................................................. 52 4.2.2 Model I: Factors Regressed on Stage .................................................................................. 57 4.2.3 Model II: Factors Regressed on Stage and Characteristics .............................................. 57 4.2.4 Model III: Factors Regressed on Stage, Characteristics, and Interaction ....................... 59 4.2.5 Model IV Backward Selection of Model of Stage on Value-Added Activities ........... 61 4.2.6 Model V Controlling for Individual Funds...................................................................... 64 4.3 Hypothesis Evaluation................................................................................................................... 67 5. Discussion .............................................................................................................................................. 69 5.1 Hypotheses 1-8: Value-Added Activities and Stage .................................................................. 69 5.2 Hypothesis 9: Venture Capital Stage Specialization .................................................................. 74 6. Conclusion: ............................................................................................................................................ 76 6.1 Implications for Practitioners and Researchers ......................................................................... 77 6.2 Limitations and potential weaknesses of our research.............................................................. 78 7. Recommendations for Future Research ............................................................................................ 80 8. Bibliography........................................................................................................................................... 81 Appendix .................................................................................................................................................... 89 I. Overview of Interviews .................................................................................................................... 89 II. Number of VC Investments by Industry in Denmark .............................................................. 90 III. Sample Survey ................................................................................................................................ 91 IV. Individual KMO Scores .............................................................................................................. 100

Page 2 of 108

V. Overall Principal Component Analysis with Screeplot ............................................................ 101 VI. Principal Component Analysis of Dependent Variables ....................................................... 102 VII. Principal Component Analysis of Interaction........................................................................ 104 VIII. Heteroskedasticity Tests .......................................................................................................... 105 IX. Theory of the Special Case of White Test................................................................................ 108

Page 3 of 108

List of Tables
Table 1: Overview of Funds Present in Our Sample........................................................................... 28 Table 2: Distribution of VC-backed firms in sample and in responses ............................................ 40 Table 3: Factorability Tests...................................................................................................................... 49 Table 4: Rotated Principal Component Analysis ................................................................................. 50 Table 5: Cronbach Alpha Scores ............................................................................................................ 51 Table 6: Factor Loadings Matrix............................................................................................................. 51 Table 7: Descriptive Statistics: Value-Add Importance ...................................................................... 52 Table 8: Descriptive Statistics: Independent Variables ....................................................................... 53 Table 9: Correlation Matrix (1/3) ........................................................................................................... 54 Table 10: Correlation Matrix (2/3) ......................................................................................................... 55 Table 11: Correlation Matrix (3/3) ......................................................................................................... 56 Table 12: Regression i Output ................................................................................................................ 57 Table 13: Regression ii Output ............................................................................................................... 58 Table 14 Regression iii Output ............................................................................................................... 60 Table 15: Regression iv Output .............................................................................................................. 63 Table 16: Regression v Output ............................................................................................................... 65

List of Figures
Figure 1: Outline of the Theoretical Causality of the Thesis................................................................ 7

Page 4 of 108

1. INTRODUCTION
How, as a fund, can I deliver the most value along with my money cause my moneys as green as someone elses?
- Josh Kopelman (2013), Partner at First Round Capital Venture Capital is the collective term for capital provided to early-stage, high risk growth companies. It operates in the undergrowth of the business formation forest by investing large sums of money in companies that have unobservable track records, but have the drive to get places, helping turn the ideas entrepreneurs of today into the companies of tomorrow. In recent years, this question of value-add has become ever more relevant from the perspective of both the fund and the entrepreneurs. The Venture Capital industry globally has seen only modest returns and a declining number of funds (EVCA 2012; NVCA 2012). This directly affects entrepreneurs, who list access to financing and bridging financing gaps as one of their largest challenges (EVCA 2012). Venture Capital is also ever-present in the minds of national governments and intra-governmental organizations such as the EUs, as it has a direct effect on innovation, job creation and growth in societies (Kortum and Lerner 1998). As a consequence, governments are pledging ever more money to innovation incubators, accelerators, direct investments, or funds-of-funds (E&Y 2010). Despite all this focus on Venture Capital, it still remains unclear how Venture Capital funds should overcome this challenge of adding value to portfolio companies and harvest solid returns. Academia research, although still a relatively young domain of research, is strongly warranted by the contemporaneousness of this issue. Two perspectives on role of Venture Capital are predominant: One stream investigates how funds can optimize their investments from a finance and economics perspective, utilizing concepts such as risk-reward and portfolio diversification. The other, based in management and entrepreneurship research focuses on the contribution of the Venture Capital fund in the interaction with its portfolio companies (Landstrm 2007). While the former theory might be utilized in practice, there is a clear observation of the latter with funds moving towards a pyramidal structure, attempting to provide more assistance, and ultimately value, to their portfolio companies (Wasserman 2005). Value-added activity is a branch of research within Venture Capital which examines these post-investment activities of Venture Capital funds towards their portfolio companies. It is engaged in the types of activities conducted by the Venture Capital funds, such as assistance and knowledge transfer, as well as the

Page 5 of 108

determinants of Venture Capital funds capability of these activities, such as what knowledge stock the Venture Capital fund holds and the mode of each individual activity. Overall, it focuses on the Venture Capital funds ability to conduct value-added activities successfully. In this paper, we shed light on one aspect of this. We investigate how the stage of the company relates to the need for value-added activities in portfolio companies, and whether or not Venture Capital funds actually act on the demands of their portfolio companies. This paper therefore investigates the relationship between portfolio companies and Venture Capitalists with the following research question:

How do Venture Capital funds value-added activities to their portfolio companies differ between stages, and do the funds specialize accordingly?

In the following section this thesis will first outline the literary framework for value-added activities within Venture Capital and construct nine hypotheses on the effect of stage on the perceived value of various value-added activities. This will be followed by section three, which contains the procedures used in this empirical study. This section contains the selection of our sample of all current ICT and Life Science portfolio companies in Danish-based Venture Capital funds, followed by an elaboration of our survey assessing entrepreneurs perception of VC funds contribution of value-added activities. This section concludes with an overview of our collected data and the methods used in this regard. Section four contains the results from our two analysis methods. First we conduct a Principal Component Analysis which groups answers within underlying factors allowing them to be used in a regression analysis. Second we model these factors against characteristics of the company and its founders in an Ordinary Least-Squares regression analysis to predict whether there is a significant difference between stages. Finally our hypotheses will be evaluated based on the regression outputs. Section five discusses our results in relation to our hypotheses and previous research, which leads to section six and the conclusion of our research question. This will be followed by an assessment of this thesis significance and implications for future research.

Page 6 of 108

2. LITERATURE REVIEW
This section places this thesis within the extensive literary framework on Venture Capital and provides a foundation for generating our hypotheses in relation to our research question. We first conduct a brief review of the two overarching perspectives on Venture Capital to highlight the differences between the two. A section on the impact of VC funds involvement in their ventures, which addresses the factor of interaction on value-added activities, will follow. Then, we will consider theories behind the impact of specialization of VC funds with respect to geography, industry, and stage of the portfolio company. This finally leads to our hypotheses formation, where we investigate the literature base for particular value-adding activities and how these activities relate to stage from the entrepreneurs perspective. The theoretical causality is highlighted in figure 1 (below).

FIGURE 1: OUTLINE OF THE THEORETICAL CAUSALITY OF THE THESIS

Perspectives on Venture Capital: Financial Intermediary versus Value-Added Knowledge Transfer and Value-Add from the Venture Capital Fund Fund Level Determinants of Value-Added Activities Specializations of Venture Capital Funds

Effects of Stage Specialization

Hypothesis Formation

Page 7 of 108

2.1 PERSPECTIVES ON VENTURE CAPITAL: FINANCIAL INTERMEDIARY VERSUS VALUE-ADD

Research on Venture Capital (VC) generally takes two perspectives. One is that Venture Capital funds (from here on referred to as VC funds) are purely financial intermediaries, reducing information asymmetries and thereby uncertainties. The other is that VC funds also act as valueadding investors who advise and help operate their portfolio companies, thus providing real value enhancement (Bertoni, Croce, and Quas 2010). 2.1.1 FINANCIAL INTERMEDIARY PERSPECTIVE The financial intermediary understanding of VC stems from the premise that VC funds operate in markets with information asymmetries. These asymmetries create a state where there is an excess demand of capital through credit rationing which results in profitable projects not being funded. In these situations, VC funds act as mechanisms to reduce the information asymmetries and resulting adverse selection problems (Amit, Brander, and Zott 1998). Additionally, the financial intermediary perspective is concerned with principal-agent issues in which moral hazard exist between the VC-fund and the entrepreneur from the premise that incentives are not always aligned (Kaplan and Strmberg 2001). In the financial intermediary understanding, the VC fund is oriented towards three methods of overcoming uncertainties, adverse selection and moral hazard. First, a key activity is selecting ventures correctly through a screening process which is particularly targeted high-technology startups with an unobservable track-record, whereby information that was previously hidden is understood and thus eliminates the effect of adverse selection (Ueda 2004). Second, the VC fund contractually eliminate moral hazard issues by virtue of control rights e.g. in the shape of cashflow rights, voting rights, redemption rights (Kaplan and Strmberg 2001). These contractual terms allow an element of control over the portfolio company, with which divergent incentives and intentions can be mitigated. Last, VC funds engage in extensive monitoring efforts, e.g. through reports on various key performance indicators in the portfolio ventures. This structure frequently revolves around a staging of financing within an investment: instead of a lump sum given to the entrepreneur, the total investment is divvied along multiple rounds, each of which is contingent on the portfolio companys progress. This allows a VC fund to control for moral hazard issues with respect to the entrepreneur along the lifetime of their principal-agent relationship (Gompers 1995).

Page 8 of 108

Indeed the financial intermediary perspective contributes to a substantial part of the understanding of Venture Capital and its raison d'tre. The following section will now focus on the perspective of value-add. 2.1.2 VALUE-ADDED PERSPECTIVE The value-added perspective is concerned with the VC funds efforts to improve the operations of their portfolio firms through an active involvement in their investments which exceeds solely monitoring efforts. The VC funds attempt to gain a competitive advantage through a costly effort to advise and control the entrepreneurial ventures. The value-added perspective is thereby only concerned with the post-investment activities of the Venture Capital process (for characterization of the phases, see Tyebjee and Bruno 1984). De Bettignies and Brander (2007) find that the existence of VC funds as purely financial intermediaries eventually discontinues them from an entrepreneurial choice perspective, as the survival of the VC fund necessitates value-added contributions when entrepreneur ownership is diluted in relation to bank financing. This evaluation serves, given an availability of bank financing, as an argumentation for the existence of value-added efforts. From the basis of performance post-Initial Public Offering (IPO), Arthurs and Busenitz (2006) find that VC-backed firms are better able to cope with product-related and management issues or as they put it, VC-backed firms exhibit better dynamic capabilities. This is due to VC fund involvement and engagement in value-added activities. This conclusion remains weak, in the sense that it seems hard to distinguish whether involvement actually caused this ability or if it is simply caused by superior selection efforts in the pre-investment phase. Baum and Silverman (2004) shed light on this question of whether Venture Capital funds select ventures which are inherently stronger performing or if they are able to build them. In their study, they find evidence of both occurrences taking place, i.e. that the selection efforts as well as the value-add measures result in better IPO performance of the portfolio company. The study finds significant evidence for the effect of value-added activities, but could not conclude on the isolated effect of value-added activities. This conclusion is in line with Frederiksen, Olofsson, and Wahlbin (1997), who find evidence that Venture Capitalists conduct value-added activities through active involvement. The economic development of the portfolio companies and its relationship to value-added is unclear, with Frederiksen, Olofsson, and Wahlbin (1997) interestingly implying that this results from VC funds dedicating their time to interacting with underperforming portfolio companies, fittingly coined firefighting in the article.

Page 9 of 108

These studies, which find evidence for the impact of value-add, are in conflict with the conclusions of Bertoni, Croce, and Quas (2010). Their study showed that value-add is not significant, in the sense that it seemed to be appropriated by non-VC-backed firms in the market, resulting in no significant differences being perceived. There thus seems to be disagreement with respect to the effect of value-added activities on portfolio companies in VC funds. In conclusion, it is not entirely agreed upon whether VC funds that conduct extensive valueadded activities actually exhibit better performance from a portfolio and, in extension, from a fund perspective. In this paper, however, we intend to apply the broader, value-added perspective rather than the financial intermediary perspective, as it seems likely that capability building as a result of value-added activities does occur. Thus the question becomes how these value-added activities transfer knowledge from the VC fund to the portfolio company. The following section sheds light on this subject.

2.2 KNOWLEDGE TRANSFER AND VALUE-ADD FROM THE VENTURE CAPITAL FUND
The premise of value-added activities is that entrepreneurial firms can learn from their investor. This entails a level of interaction between the two parties as well as permeability of knowledge between the VC fund and the portfolio company. The following section attempts to explain these two mechanisms: interaction and permeability. Interaction can be interpreted as communication between the VC fund and the portfolio company. Studies have shown that Venture Capitalists spend 45-70% of their time assisting their portfolio companies (Zider 1998; Gorman and Sahlman 1989). Interaction consequently signifies a key part of what Venture Capitalists do. In that regard, Sapienza (1992) finds that the frequency of interaction is a determinant of the perceived value-add from the VC fund, thus lending support to the notion that interaction plays a part in transferring knowledge. However, just as there is uncertainty regarding whether or not value-add actually increases performance, there is also uncertainty regarding the relationship between interaction and performance. Bottazzi, Da Rin, and Hellmann (2008) find that when controlling for reverse causality, i.e. VC funds may interact more with companies that perform well or poorly, which is similar to the proposition of Frederiksen, Olofsson, and Wahlbin (1997), the level of interaction is not a significant determinant of portfolio company performance. However, they do find that valueadded activities themselves are significant and positive. Activism in the broad sense on the part

Page 10 of 108

of VC funds thus helps guide the portfolio company towards success, but the study did not find that interaction, as a measure of communication, significantly contributed to the success of the portfolio companies. This lends to this thesis framework by forming a distinction between interaction and value-added activities. The second mechanism of knowledge transfer is the permeability of knowledge. This aspect of knowledge transfer is arguably governed by the nature of the relationship between the VC fund and the portfolio company. In this regard, Berglund, Hellstrm, and Sjlander (2007) suggest that merely through a VC funds presence in the processes of the entrepreneurial firm, VC funds can contribute to the entrepreneurs learning capacity, thus representing a form of indirect knowledge transfer under the right circumstances. More directly though, the high degrees of knowledge sharing, relationship-specific investments, goal congruence, and the level of trust between the two actors have been proposed as conducive to the quality of value-added activities (De Clercq and Sapienza 2001). This capacity to establish a solid relationship for permeability of knowledge further instigates better perceived performance (De Clercq and Sapienza 2006). Relationships and the corresponding permeability of knowledge thereby constitute a mechanism through which value-add is transferred, potentially increasing the performance of the portfolio companies and consequently the VC funds. This highlights the idea that the value-added perspective is of a softer nature than the financial intermediary perspective. The complexity of permeability of knowledge lends to this thesis omitting it as a further area of interest. While explanatory, it is still in a propositional stage of research and will therefore be disregarded for the remainder of this thesis. We assume value-added activities are only transferred through interaction when active efforts exist, thus taking the degree of permeability for granted.

2.3 FUND LEVEL DETERMINANTS OF VALUE-ADDED ACTIVITIES

We have given a review of the dichotomous research orientations with respect to Venture Capital, i.e. the financial intermediary vs. value-added perspective. Given this thesis focus on value-added activities, we highlighted the mechanisms through which knowledge is transferred between the VC fund and the portfolio company. In the following section we emphasize how value-added activities are related to VC fund level strategies, focusing on VC fund specialization as a method to enhance value-add to portfolio companies, beginning with the concept of limited attention.

Page 11 of 108

Given that interaction is the mechanism through which knowledge is transferred, there is a cost associated with value-added activities. This presents itself through a time-resource constraint in VC funds, implying an allocation of the VC funds efforts between portfolio companies (Fulghieri and Sevilir 2009; Gifford 1997). The Venture Capitalist in effect experiences a tradeoff between dedicated efforts to a few ventures and a low involvement in many ventures, risking by analogy scraping butter over too much bread. Portfolio size thereby becomes a determinant of the level of value-added activities given a set time resource. Nevertheless, portfolio size is not the only determinant given such a time constraint. Funds can also specialize in a particular type of firm to have the highest possible degree of adaptation to portfolio company needs, thereby making their value-added activities more constructive. The foundation of specialization is that VC funds are able to accumulate knowledge that can be transferred to the portfolio companies. Highlighting this concept, Yang, Narayanan, and Zahra (2009) conclude from an organizational learning perspective that the Venture Capitalist learns from a diversity of experiences, which contribute to an overall understanding of a concept, as well as intensity of the experiences that contribute to an in-depth expertise. This implies that under time and learning constraints a tradeoff ultimately exists between specialized and diversified learning from the VC funds point of view, where the in-depth expertise represents a specialized accumulated knowledge (Yang, Narayanan, and Zahra 2009). To obtain and accumulate this type of knowledge, VC funds can specialize in particular types of ventures, thus incurring a higher intensity of experiences and becoming experts in that field. This specialization results in a more particular type of expertise, reducing uncertainties associated with working with portfolio companies. Such a specialization can occur along multiple dimensions, namely geography, industry, and stage, which will be reviewed in the following section. As stage is the primary focus of this paper, the former two will be evaluated briefly, while the latter will be given a more extensive literary review.

2.4 SPECIALIZATIONS OF VENTURE CAPITAL FUNDS

We have thus far established that the amount of attention paid to a specific portfolio company, along with the expertise behind the attention, are determinants of value-added activities and that this accumulation of knowledge can be obtained through specialization. The question then becomes what implications specialization carries, which will be investigated here.

Page 12 of 108

Gompers, Kovner, and Lerner (2009) investigate the relationship between organizational structure with respect to industry specialization and the success of the VC funds investments, and whether this specialization results in improved or worsened performance. They find conclusive evidence that specialization is associated with a higher probability of IPO exit, a desired outcome of VC investments. Patzelt, Zu Knyphausen-Aufse, and Arnoldt (2006) moderate this conclusion by stating that diversification within an industry also effectively reflects a risk diversification because the within-industry risk profiles of portfolio companies can be substantially different. It has to be kept in mind that the sample used was biotechnology firms which encompass highly different risk profiles depending on their technological basis, thus making inference to other industries less given. The matter of specialization versus diversification is extensively researched by Knill (2009). Knills (2009) premise is that the choice between specialization and diversification is determined by the nature of the VC funds limits with respect to time and expertise. Through an investigation across industry, geography, and stage, the author finds that diversification is significantly and positively related to the growth of the VC fund over time, with stage representing the second strongest predictor. The analysis also concludes that the entrepreneurial outcome, measured in the study by time to exit via IPO, is negatively related to diversification of industry and stage, with industry having the most negative impact. In addition to conclusions that span geography, industry, and stage specialization, the study highlights a divergence between the VC fund and its portfolio company in their determinants of success. Knill (2009) thereby validates that stage specialization results in better performing firms because of increased levels of competency transfer. It also highlights that different interests exist in the relationship between VC fund and portfolio companies when it comes to stage specialization, lending to the necessity of research on the interaction between stage specialization and the portfolio firm perspectives. Thus, it serves as a fitting outset for a characterization of how stage specialization is viewed in the literature. The following section will attempt to shed light on this factor of VC fund strategy.

2.5 EFFECTS OF STAGE SPECIALIZATION

The premise for stage specialization is arguably best characterized through life-cycle theory. The concept of firm development through stages stems from an extensive body of literature commonly termed life-cycle growth models, in which a firms development consists of a sequence of challenges that have to be overcome to transition to the following stage (see Phelps, Adams, and Bessant 2007 for an exhaustive review of literature on life-cycle theories). These

Page 13 of 108

various stages characterizations help define and frame firm development, but seem to lack a stringent definition (Hanks et al. 1994). What furthermore complicates the understanding of stages with respect to VC funds and their relationship to the portfolio companies is that stage as a concept can be defined as one thing from the perspective of the VC fund, and another from that of the portfolio company. Either way, the underlying concept behind firm stages, i.e. that entrepreneurial ventures undergo a sequential development with corresponding changing nature of challenges, seems to be commonly accepted and connected with the sequencing of finance in the Venture Capital industry (e.g. Smith, Smith, and Bliss 2011; De Clercq et al. 2006; Sahlman 1990; Ruhnka and Young 1987). Given that entrepreneurial ventures undergo a transformation as they develop, particularly with respect to challenges and needs, this lends to a foundation for stage specialization of the VC fund from the premise of portfolio companies need for specific knowledge requirements and assistance in building expertise. This would explain why stage specialization is observed to be relatively common in VC funds. (Gejadze, Giot, and Schwienbacher 2012; Bartkus and Hassan 2009). Norton and Tenenbaum (1993) further add to the concept of stage specialization through an investigation of how systematic and unsystematic risk apply to Venture Capital, more specifically how specialization and information sharing act as risk control measures. They find that VC funds tend to specialize in particular stages. Furthermore, VC funds that are involved in early stages focus on particular industries because of the necessity of specialized industry knowledge. It is thereby possible to distinguish between VC funds that specialize and VC funds that do not. Schwienbacher (2012) depicts these options from an entrepreneurs point of view, where he deduces that when faced with a choice between a specialized versus a generalist VC fund, the entrepreneur faces a tradeoff. A specialist fund has a higher potential for value-add due to cumulated expertise, but lacks the ability or incentive to carry the venture across stages. Contrary to the specialist fund, the generalist fund has the ability to carry the portfolio company through multiple phases, but potentially lacks the specialized knowledge to guarantee a high value-add, often resulting in lower growth. Schwienbacher (2012) concludes that the potentially higher growth path for the startup matters more than the risk of premature startup liquidation stemming from a lack of ability to obtain follow-up financing. This proposition is supported by G. Smiths (1999) results regarding the selection criteria of entrepreneurs with respect to VC funds, where the stage specialization ranks in the upper quartile. Schwienbachers (2012)

Page 14 of 108

theoretical proposition is also in line with Clercq and Dimov (2006), who find stage specialization to be negatively associated with startup failure rates, thereby supporting Schwienbachers (2012) proposed notion that the value of specialized expertise outweighs pre mature cessation. This indicates that there is evidence of benefits to both entrepreneurs and VC funds when it comes to the accumulation of stage-specific knowledge. However, the evidence for stage specialization is not conclusive. Several studies find that it is stage diversification which increases fund performance, rather than stage specialization. Bartkus & Hassans (2009) empirical study finds a positive and significant relationship between stage diversification and VC fund performance with respect to successful exits. In relation to Schwienbacher (2012)s proposition regarding the upside of the generalist fund, Bartkus & Hassans (2009) dependent variable, exit success, could exhibit a positive relation to stage diversification because funds that span stages are better able to guarantee follow-up financing. Bartkus & Hassans (2009) conclusions are somewhat supported by Cumming, Fleming, and Schwienbachers (2009) study of style drift in Private Equity, i.e. deviations in stated stage focus by a Private Equity fund, keeping in mind that their stage definition is wider than that of other studies. Within the literature of stage specialization, researchers have found that there is a tendency to move towards later stage specialization, for example De Clercq et al. (2001) study of the Finnish Venture Capital funds. They also found, given a higher mean stage of investment, a tendency for diversification across stages, geography, and companies. This lends an interesting aspect to the concept of risk diversification from the resource-based view of specialization: the trend observed by De Clerq et al (2001) does not seem to stem from intent to specialize, but rather from a profit-oriented risk diversification. Justifications for this strategic move can be found in empirical evidence from Gompers, Kovner, & Lerner (2009) who find that funds focusing on later-stage investments exhibit higher exit success rates. This in turn may cause a shift towards the later stages. Another perspective in this regard is that the late-stage orientation is caused by a disproportional growth in assets under management in monetary terms, which given a lack of growth in the funds human capital resources implies a higher investment size in order to have a sufficient degree of interaction with the portfolio company (Bygrave and Timmons 1992 through Elango et al. 1995). Either way, this shift towards the later stages is a trend that carries implications for the structure of Venture Capital markets, because the VC funds in later stages are dependent on a flow of high

Page 15 of 108

quality ventures that have received prior funding. This shift has led to the much debated concept of the funding- or equity gap in various markets (e.g. Cressy 2012; Murray 1999), followed by governmental interventions in the market to avoid financial starvation of entrepreneurial ventures ( e.g. Da Rin, Nicodano, and Sembenelli 2006). This concludes our review of literature that sets the stage for our hypothesis formation. We find a relevant conjunction between stage specialization of VC funds and their value-added activities and that an investigation is appropriate and serves to contribute to existing literature. We emphasize that our understanding of the gap in the literature, with respect to the characterizations of stage specialization and value-add activities and the interactions between them, is shared with De Clercq and Manigart (2007). Furthermore, we believe the entrepreneurial perspective on this matter lends to interesting insights that add to the understanding of valueadded activities and Venture Capital in general.

2.6 SPECIFIC VALUE-ADDED ACTIVITIES AND HYPOTHESIS FORMATION

In the previous sections we have attempted to investigate the interactions between VC funds and their strategies with respect to portfolio companies. The following section represents a review on how value-add is perceived in the literature and how it interacts with stage. This section serves, based on the previous sections, to theoretically outline this thesis hypotheses. Large and Muegge (2008) provide an extensive review on the literature on non-financial valueadded activities, in which we also took a preliminary outset. It is important to note that singularizing a specific area of focus from articles investigating several dimensions of value-add would involve omitting other investigated variables. This would impoverish the value of the previous research. Therefore, several studies in the following section span multiple sub-sections of value-added components to avoid omitting relevant observations. Furthermore, we utilize the distinction between types of value-added provided by Luukkonen, Deschryvere, and Bertoni (2013). This distinction between types of value-add relatively closely resembles the classifications used in Large and Muegge (2008) with respect to horizontal levels (e.g. executive, marketing & sales), whereas the applied distinction excludes the differentiation on the vertical level (e.g. legitimization, mandating, strategizing). The vertical dimension is instead applied as singularized types of value-added activities.

Page 16 of 108

2.6.1 STRATEGIC VALUE-ADD The literature does not seem to agree on the definition of strategic contributions. Due to its loosely defined nature, it is frequently a topic of interest whether in the form of a distinct focus or in the shape of a more precisely defined variable that essentially is a subset of strategy. Strategic value-add from the VC fund to the entrepreneurial firm is typically understood as the VC fund guiding the entrepreneur in the right direction, primarily done through board representation (Fried, Bruton, and Hisrich 1998). Strategy is furthermore composed of an internal process and is related to context of the entrepreneurial firm. Wijbenga et al. (2003) propose that this context is best understood by the inclusion of an assessment of the degree of fit with regards to strategic involvement, the premise being a distinction between agency, entrepreneurial, positioning, and resource fit. This definition would capture in more detail the mechanisms which drive strategic importance. More studies have instead taken strategy as isolated from environmental concerns, i.e. from a more general perspective. Sapienza, Manigart, and Vermeir (1996) find that VC funds perceive the strategic role as being their most important contribution regardless of stage and that this type of involvement potentially increases firm performance, in line with the conclusions of previous studies (e.g. Ehrlich et al. 1994). This conclusion implies that entrepreneurial firms need strategic contributions. Macmillan, Kulow, and Khoylian (1988) also reach this conclusion, asking Venture Capitalists for the importance of their activities. Aside from the strategic role being the most important, their study also finds that strategic importance exhibits little variance across stages. Furthermore, Gabrielsson and Huse (2002), in a dyadic perception-based study, find the development of business concepts/strategies to have a middle-of-the-pack importance level, although being recognized as the boards main task. The conflicting results are possibly due to the different contexts of the studies and the various value-added components which were studied. Regardless of the various definitions, strategic value-add overall seems to be an important type of value-added activity, as it contributes to the ability of the portfolio company to orient itself in its environment. Pratch (2005) highlights that strategic value-add mainly lies in providing critical thinking, first and foremost defining the company with respect to its value proposition and business model. This defining orientation, Pratch (2005) argues, is effectively a continuous reevaluation of the entrepreneurial firm as it transitions through its life cycle. This implies that

Page 17 of 108

strategic value-add is an important feature of the fund-firm relationship at all stages of development. From the above review, we hypothesize that strategic value-add has a substantial value-add effect, irrespective of the company stage because it is an iterative and continuous process. Thus: H1: There is no significant effect of stage on strategic value-add 2.6.2 FINANCIAL VALUE-ADD Financial value-add is primarily concerned with the VC funds ability to actively contribute to raising additional capital for the portfolio firm. This additional capital can stem either from the VC fund themselves in the form of staged follow-on finance, from non-equity sources e.g. through banks, or from attracting third-party VC funds. Gomez-Mejia, Welbourne, and Balkin (1990) investigate financial value-add using a definition that includes financial management, i.e. building financial capabilities in the portfolio company, as well as the above-mentioned forms of capital. Based on interviews, their study finds that the importance of financial contributions as perceived by the entrepreneur is significantly important, both with respect to financial management and capabilities, as well as raising direct and indirect capital. This is in line with the dyadic perception-based findings of Gabrielsson and Huse (2002) with respect to building financial competencies and the acquisition of capital. This is also consistent with numerous other studies (Maula, Autio, and Murray 2005; Ehrlich et al. 1994; Gorman and Sahlman 1989; Macmillan, Kulow, and Khoylian 1988). The only inconsistency in the literature to the authors knowledge is Sapienza (1992, pt. through Large and Muegge 2008) which finds that from dyadic perspectives, the role of financier is only moderately important. This inconsistency can stem from a multitude of causes, examples being the context of questions asked, the definition of the financial value-add which remains unspecified in the article, or the sample. Nevertheless, there still seems to be a relatively high degree of agreement in the literature with respect to the significance of financial value-add. Given the premise that a firms survival is dependent on continuous financial resources to develop and expand the business, and that a startup firm intends to progress through all of its life-cycle stages, it can be argued that the startup has needs for external financial contributions regardless of stage (Smith, Smith, and Bliss 2011).

Page 18 of 108

We thus hypothesize that given the evidence from previous literature with respect to the importance of financial contributions and the continuous nature of this need, financial value-add has a significant effect irrespective of stage: H2: There is no significant effect of stage on financial value-add 2.6.3 CREDIBILITY VALUE-ADD In addition to active efforts in competency transfers from the VC fund to the portfolio company, the VC fund also contributes passive qualities that add value to the portfolio company. Of these passive elements, reputation or in effect credibility is particularly important. Startups are characterized by high degrees of uncertainty and unobservable track records. However, if the startup is affiliated with a third party that exhibits high degrees of observable credibility, part of this reputational benefit will effectively spill over to the startup. This lends to a lowering of the externally perceived uncertainty regarding the quality of the startup (Hochberg, Ljungqvist, and Lu 2007). This affiliation with a VC fund with a high prominence increases the likelihood that the VC is associated with other prominent investors and thus improves deal-flow (Hochberg, Ljungqvist, and Lu 2007). This in turn creates the effect that the reputation of a VC fund becomes a desirable quality from the entrepreneurs perspective. Therefore, as Hsu (2004) argues, a market effectively arises where startup firms pay for affiliation with high-reputation VC funds by accepting lower valuations, of which he also finds empirical support. Furthermore, Stuart, Hoang, and Hybels (1999) investigate this type of endorsement in relation to performance at the IPO stage. They find that, for the portfolio company, the prominence of an affiliated VC fund facilitates the association of other prominent partners, and has a significantly positive effect on the IPO. This conclusion is in line with the findings of Megginson and Weiss (1991), Dolvin (2005) and Nahata (2008) with respect to various normative IPO metrics, such as reduced underpricing at IPO, lowered underwriter costs, and issuance costs. In conclusion, there is evidence for the importance of credibility, warranting its investigation. We theorize that the effect of this value-add applies equally to through the various stages of the startups development i.e. pre-seed through expansion, but also in terms of exit. Thus we hypothesize: H3: There is no significant effect of stage on credibility value-add

Page 19 of 108

2.6.4 PROFESSIONALIZATION VALUE-ADD Professionalization is a commonly studied topic within the value-added perspective. Professionalizing the portfolio companies is primarily pursued because a) total dependency on the entrepreneur for management of the portfolio company is risky and inherently counterproductive and b) it makes the firm easier to finance, both for the VC fund and in follow-up rounds (Zingales 2000). Active efforts on part of the VC fund towards professionalizing the portfolio company thus constitute a value-add activity through efforts such as corporate governance, HR policies, board, management, and advisor setup. With respect to corporate governance, Hochberg (2011) conducted a normative study of venture-backed versus non-venture backed firms in the post-IPO phase. The study finds that VC-backed firms have better governance in the post-IPO phase when measured as cost control management, market reactions to shareholder rights agreements, and lastly a string of board, audit and chairman metrics. Meanwhile, Sapienza (1992, pt. through Large and Muegge 2008) investigates recruitment of management, and finds it to rank relatively low in importance. This finding is in line with Gabrielsson and Huse (2002), who find professionalization to rank moderate and recruitment low in importance, respectively, furthermore coinciding with the observations of CEOs in Gomez-Mejia, Welbourne, and Balkins (1990) dyadic study. From a human capital perspective, Hellmann and Puri (2002) investigate how VC funds professionalize their portfolio company from the firm perspective. Specifically, they investigate how VC funds build the internal organization, such as HR policies and the recruiting of management, along with an investigation of CEO turnover. Hellmann and Puri (2002) find that VC-backed firms professionalize along these dimensions faster, but also that the importance of the VC funds involvement is higher in the earlier stages of the firms development. This lends to the concept that professionalization is one of the first things VC funds engage in once an investment has been made. Even if there is not conclusive agreement as to the importance of professionalization measures, there is evidence in the literature that suggests that professionalization is more prominent in the early stages of portfolio companies. In conclusion, based on the literature we hypothesize that professionalization, and thereby the construction of professional processes and of human capital, will be more prevalent in the earlier

Page 20 of 108

stages. Thereby the professionalization value-added activities effect from the firm perspective will be higher the earlier the stage: H4: There is a negative effect of stage on professionalization value-add 2.6.5 MARKETING VALUE-ADD A significant part of the new ventures development is associated with the development of a defendable market position, i.e. the commercialization of the ventures core business concept. Building a solid market position represents an opportunity for the new venture to establish a revenue base and generate profits to fund future growth. Stre (2003)s case study on informal Venture Capital in Norway lends to the proposition that an important aspect of VC fund efforts is to develop a market orientation. However, this concept of market orientation can be carried out from a strategic rather than hands-on marketing base. This distinction is important because the understanding of marketing in the literature exhibits overlaps between the strategic and the operational side of marketing. Maula, Autio, and Murray (2005) apply a distinction between social capital-derived marketing, similar to an operational understanding, and a knowledge-based marketing, more congruent with a strategic understanding. In the study, they find that the two components measure comparably in relation to each other and that they exhibit moderate importance. However, that is not always the case. While the distinction is congruent with Macmillan, Kulow, and Khoylian (1988), they find that VC fund investment managers in the studys sample intended to be more engaged in marketing due to its importance in creating growth, but that the level of actual involvement with marketing was low, in line with other studies (Gabrielsson and Huse 2002, Ehrlich et al. 1994). Macmillan, Kulow, and Khoylian (1988) explain this conflict of importance versus intent through the relative costs associated with them: operational marketing efforts are more costly than strategic marketing efforts, with the VC fund investment manage thus preferring to engage in marketing at the strategic level. Meanwhile, Hellman and Puri (2000) study the time-to-market for venture-backed and non-venture-backed companies, and find evidence that VC fund involvement results in significantly faster time-tomarket, implying a value-added activity in the early stages. This implies that VC funds to some extent do exert operational marketing efforts. In that regard, operational marketing value-add represents a significant part of venture development. Nonetheless, the cost associated with it implies that it is not something that would be perceived as important by the portfolio company, due to the low involvement of the VC fund in this matter. However, we argue that as the marketing efforts move towards a more strategic

Page 21 of 108

nature with stage development, so does the VC fund investment managers involvement in the effort, implying that the value-added effect should grow as the firm moves along its lifecycle. In conclusion, we hypothesize that the effect of marketing value-add efforts will increase the later the portfolio company stage, due to a transition from an operational marketing focus to a strategic marketing focus. Thus: H5: There is a positive effect of stage on marketing value-add 2.6.6 TECHNOLOGICAL VALUE-ADD A commonly applied basis for innovation is the establishment and protection of new technology. Innovative solutions are a frequently associated with startups and Venture Capital, and a series of studies find Venture Capital to spur innovation in portfolio companies (e.g. Kortum and Lerner 1998; Popov and Roosenboom 2009). In line with these conclusions, Bertoni and Tykvov (2012) find that VC funds actively promote their portfolio firms effort towards patenting, and by that proxy spur innovation. Meanwhile, this connection between innovation and Venture Capital is found to only weakly exist in Engel and Keilbachs (2007) study of German startups. They explain this by the intensity of innovation happening prior to VC investment, wherein a technological position is an ex-ante condition for investment. This conclusion is supported by Caselli, Gatti, and Perrini (2009) who find that VC funds do not significantly promote innovation, but rather the pursuit of greater sales. It is thus important to recognize that VC funds contribution to innovation on a broad level might be visible due to VC funds picking innovative firms ex-ante, while not focusing on it ex-post, thereby lending to a conclusion that VC-backed firms are more innovative. This coincides with the findings of Gabrielsson and Huse (2002), Ehrlich et al. (1994), and Macmillan, Kulow, and Khoylian (1988) stating that building technical competence is perceived as being relatively unimportant from both the VC funds and entrepreneurs perspectives. VC funds furthermore add value to portfolio companies by leveraging their networks to establish strategic alliances that foster inter-firm technological and R&D partnerships (Lindsey 2008). This constitutes a value-added activity that exists outside of traditional mechanisms of transferring value-add, in that it is external in nature. However, we do not perceive it as a significant category of value-added activities, thus not particularly moderating the otherwise low importance evidenced in the literature. Due to the low importance associated with technological value-added activities, along with the general perception in the literature that VC funds generally select innovative companies rather

Page 22 of 108

than building them, we hypothesize that the effect of technological value-added activities will be more prominent in the earlier stages. H6: There is a negative effect of stage on technological value-add 2.6.7 INTERNATIONALIZATION VALUE-ADD Internationalization value-add represents the VC funds efforts to enhance the international aspects of the portfolio company. It represents the knowledge, abilities, and resources of the VC fund that are international in nature, which are transferred to the portfolio company. High-tech startups aim to internationalize because it represents an ability to scale up, leveraging costs already incurred, thus increasing profitability (Qian and Li 2003). Smolarski and Kut (2011) also find internationalization to be associated with the VC fund post-investment phase behavior with their investments and to be positively associated with performance. They conclude that syndication and monitoring both positively influence internationalization and performance when taken as separate effects, but that they moderate each other. This lends to the idea that VCs can influence the internationalization of their portfolio firms, and that it is in the firms interest. Fernhaber and Mcdougall-Covin (2009) also investigate this concept of internationalization of new ventures from the premise of the resource-based view. In their framework the VC funds transfer both their international knowledge and reputation and these two factors interaction terms. They furthermore find empirical and statistically significant positive relationships between knowledge and reputation of the VC fund, new venture internationalization, and their interaction effect. This implies that VC funds to some extent drive their portfolio companies abroad. One method through which this might occur is attracting customers from international markets. In this regard, Maula, Autio, and Murray (2005) find that internationalization efforts based on the VC funds ability to help the portfolio firm attract foreign customers exists, but ranks relatively low in importance. This conclusion can probably be ascribed to not all respondents having or intending to have international operations, which nonetheless means that the average importance will be low. In Lockett et al.s (2008) study, they find that VC-backed firms internationalization is significantly related to the VCs value-add contributions as measured along similar methods as Macmillan, Kulow, and Khoylian (1988). Although not specifically measuring internationalization value-added activities, but rather value-added activities broadly, the results imply that the effect of the VC funds efforts still significantly influences the portfolio companies internationalization process. Furthermore, Lockett et al. (2008) concludes that, based on the premise of resource

Page 23 of 108

availability, pre-seed and seed-stage portfolio companies have significantly higher benefits from VC fund value-add with respect to internationalization. We therefore hypothesize that the effect of international value-add of a VC fund will be higher in the earlier stages due to lower resource availability: H7: There is a negative effect of stage on internationalization value-add 2.6.8 EXIT ORIENTATION VALUE-ADD Exit orientation value-add consists of the VC funds ability to secure an exit opportunity for their investment, potentially generating returns to both VC fund and portfolio company owners. These exit opportunities consist of IPOs, company buybacks, trade-sales, liquidation, secondary sales, and reorganization, where an IPO usually is the most desirable outcome from a return and reputational perspective (Wang and Sim 2001). Achieving a desirable outcome is specifically strived towards from relatively early on, becoming an integrated part of the advice that the VC fund provides to its portfolio company. Isaksson (2006) found agreement across his sample of Swedish VC-backed firms that it was common to head towards a desirable exit as part of the strategic orientation of the portfolio firm. However, the intensity of exit-directed activities depends on whether or not an exit strategy is planned from the outset of the VC fund-startup relationship, and whether the type of VC exit is planned as well. Isaksson (2006) finds strong empirical support that exit strategy is a value-added activity particularly for IPOs and trade-sales as exit orientations. While both parties are interested in an exit eventually, the means to get there and the timing of the exit is a particularly frequent type of VC fund-startup conflict (Zacharakis, Erikson, and George 2010; DeTienne and Cardon 2010). These conflicts can moderate the effect of the VC funds with respect to exit orientation. By their very nature, exits still constitute a critical phase of the VC fund-startup relationship, particularly for funds engaged in stages where such exits are likely (Petty, Bygrave, and Shulman 1994). Thus positive outcomes, in line with Wang and Sims (2001) characterization, are more frequent for companies in later stages (Giot and Schwienbacher 2007). This indicates that exit orientation value-add is likely to increase in importance as the portfolio company moves towards the later stages, which in turn increases the likelihood of an exit.

Page 24 of 108

Therefore, we hypothesize that the effect of exit orientation value-add will increase the later the stage a portfolio company is in due to the increased likelihood of exit and the prospects of sizeable returns to the entrepreneur: H8: There is a positive effect of stage on exit value-add Through the distinction between eight types of value-add; strategic, financial, credibility, professionalization, marketing, technological, international and exit orientation value-add activities, we have established a set of hypotheses regarding the relative importance of the given value-add activity, and its relation to the stage of the firm. 2.6.9 FUND STAGE SPECIALIZATION Finally, referring to the previously reviewed literature on VC fund stage specialization and lifecycle theory, we argue that funds are interested in focusing their skillsets to be able to cope with the specific needs of their portfolio companies. We therefore theorize that when the value-added activities effect depends on stage, including the VC funds will remove this effect. This in turn means that, with regards to value-added activities for a given stage, that funds are specialized to a certain degree. We therefore hypothesize: H9: Given a significant effect of stage, the effect of individual funds removes the significance of stage This concludes our literature review. We have positioned this paper in the literature base by reviewing the financial intermediary perspective in relation to the value-added perspective that this paper applies. We furthermore highlighted how value-added activities transfer knowledge to the portfolio companies. We then characterized the fund level determinants of knowledge transfer, further reviewing the literature with respect to how VC fund specialization relates to these value-added activities. We then finally conducted an extensive review of the literature regarding specific value-added activities to create a foundation for our hypotheses formation. We hypothesize that portfolio company stage has varying effects on each of the eight types of valueadd and that VC funds specialize within these value-added activities when they depend on the company stage. In the following section we depict our approach to investigating hypotheses, thus explaining the methodology we apply to establish a solid foundation for our analysis.

Page 25 of 108

3. METHODOLOGY
This thesis utilizes a survey-based research design to collect primary data measuring the portfolio companies perceptions of the effect of value-added activities by their VC funds. This displays a methodological congruence with previous studies on value-added activities (e.g. Maula, Autio, and Murray 2005; Gabrielsson and Huse 2002; Ehrlich et al. 1994). We imposed a limitation of scope in restricting our focus to VC funds with a physical presence in Denmark. This was chosen because of the situational characteristics of a bachelor thesis. In the section below, we will elaborate on the methods used to determine the effect of stage on VC funds perceived contribution to their portfolio companies. First we provide a brief overview of the selection of our sample of all current ICT and Life Science portfolio companies in Danish-based Venture Capital funds. This will be followed by the construction of our survey on entrepreneurs perception of VC funds contribution of value-added activities based on selected questions from a validated survey. Thereafter we discuss how the data was collected via an internet-based survey and corrected for use in our analysis. Following this will be an elaboration of the principal component analysis, which groups the individual questions on perceived valueadded activities within factors. This is then followed by a characterisation of our applied Ordinary Least-Squares regression analysis, which models these perceptions against characteristics of the company and its founders to predict whether there is a significant difference between seed and early-stage ventures. The final section within methodology will elaborate on the model which, for those value-added activities which do differ between stage, assesses the specialization of funds by re-running the model including individual fund dummy variables and seeing whether the effect of stage is picked up by the fund dummies.

3.1 SAMPLE SELECTION

We define our population as all portfolio company investments from Danish Venture Capital funds. To test our hypotheses of whether VC funds value-added activities depend on company stage, we test statistics of a sample from the population. This following section will elaborate on the selection of our sample. The limited, dispersed, and highly private nature of the Venture Capital industry argues for the use of professional informants as they have privileged insights and hold archetypical portrayals of the industrys characteristics (Lee 1993). Considerable effort was used to complement our knowledge of Danish Venture Capital funds with those of a variety of professionals. We initially

Page 26 of 108

conducted open-ended stakeholder interviews1 in the Danish Venture Capital market. This was followed by five 45-minute semi-structured interviews with investment managers in the largest funds2 (Corfitzen 2013, F. Knudsen 2013, J. Knudsen 2013, Roested 2013, Wylonis 2013). This led us to conduct a cluster sampling of portfolio companies based on their geography, type of VC fund, and industry focus (Barnett 1991). 3.1.1 GEOGRAPHY The geographical restriction focused on portfolio companies who have received investments from Venture Capital institutions with a physical presence in Denmark. The Danish market has portfolio companies receiving Venture Capital from foreign funds, but this assignment is limited to the transfer of value-added competencies from Danish funds. This is not equivalent to a national focus, however, as portfolio companies are not subject to geographical limitations. Our sample includes firms based in the USA, Germany, Switzerland, United Kingdom, Finland, Hungary, Norway, and Sweden, as well as Denmark. 3.1.2 INSTITUTION TYPE The Danish Venture Capital and Private Equity Association (DVCA) is a national trade association, with over 200 members throughout the Venture Capital chain, from business angel to private equity fund (DVCA 2013). They categorize 18 of their members as Venture Capital funds, ranging across all industries. Of these, four are pure innovation incubators, two are corporate venture funds, one is an innovation branch of a university, and 11 are Venture Capital funds backed by either public or private investors. This thesis includes innovation incubators in its sample due to their centrality and strong presence, while disregarding the corporate venture funds and the university branch. While these institutions might be important actors in the Danish Venture Capital environment, notably Novo Fonden, which is the largest Life Science Venture Capital fund in the world, they operate quite differently as evergreen funds and are subject to different management regulations and structure of investments. Based on interviews (J. Knudsen 2013; Corfitzen 2013; Wylonis 2013; Appendix I), this thesis also includes the remaining two innovation incubators, the Swedish VC fund Creandum which invests in Denmark and is particularly prominent in the entrepreneurial environment, as well the dual accelerator and investment program Accelerace. Although Accelerace is different compared
Interviews were conducted with Vkstfonden, Erhvervs- og Vkstministeriet, Innovation incubator DTU Symbion, Professors of Entrepreneurship and Venturing, Venture Capital funds, individual investors, Business Angels, startup accelerators, and entrepreneurs. 2 Sunstone Life Science, SEED Capital, VF Ventures, Creandum, Accelerace
1

Page 27 of 108

to a VC fund, our interviews revealed it to be a significant institution in the Danish market for Venture Capital3. Its inclusion is warranted by a similar structure to other funds, such as limited time frame, competency transfer, and Limited Partner pressure (unlike, say, corporate funds). 3.1.3 INDUSTRY This thesis uses the industry definitions used by the Danish governmental investment fund: Vkstfonden. It classifies its investments within five areas: Information and Communication Technology (ICT), Trade and Services, Industrials, Life Science, and Energy. This thesis selects both Life Science and ICT as these two industries represented 83% of all venture investments in Denmark in 2011 (Appendix II). Furthermore, two industries were selected to avoid an industry bias and to ensure the differing competencies and characteristics of the two industries would be represented. Including both industries would also increase the likelihood of observing both large and small investments, as the size of an investment varies tremendously across these industries. All other industries were excluded to avoid small samples, since very few investments have been made in the past and will be made in the future in other areas (J. Knudsen 2013, Roested 2013). This in turn meant that specialized venture funds in other areas, such as Insero and NES Partners who focus on energy, were excluded from our sample. We thus have in our sample every major Life Science or ICT-focused Danish Venture Capital institutions in the Danish market. To make sure we had not missed an important institution, our list was confirmed through interviews with five investment managers from varying funds and with an employee at DVCA. An overview is visible in Table 1 (below).
TABLE 1: OVERVIEW OF FUNDS PRESENT IN OUR SAMPLE

For the remainder of this thesis, Accelerace will be considered interchangeably as a fund.

Page 28 of 108

The funds respective current industry focus was found through their websites with all companies not in Life Science and ICT excluded from the sample (Company Websites 20134). If the specialization was unclear, the portfolio company was initially included in the sample. In total, our sample includes 246 portfolio companies. Our sample is based on portfolio companies perception of the value-added activities from their funds, which implies one survey for each VC fund-portfolio company relationship. Certain individual portfolio companies were therefore represented twice in our sample, due to syndication and previous funding rounds. Our total sample constitutes 310 portfolio company investments within 246 portfolio companies from 14 ICT- or Life Science-focused VC funds. 3.1.4 SAMPLING ERROR Although we attempt to randomize our selection of observations, we suffer from several potential sampling errors. This section will elaborate on the three main sources of error: survivorship bias, self-selection bias, and industry bias. Survivorship bias indicates that while we are interested in all portfolio company investments from Danish Venture Capital funds, our sample is biased towards portfolio companies currently active. We thus have no chance of observing a company that is inactive in our sample, although we would ideally like to have the views of every company the fund has invested in. This is a significant limitation, but the uncertainty and inability to obtain contact information of all failed ventures means that this bias is inherently part of any survey assessing portfolio companies. Self-selection bias implies that we only obtain observations from individuals who volunteer to complete our survey. Respondents that are highly opinionated are more likely to respond to express this opinion. We argue that the prompts that govern the respondents opinionated nature will direct the answers in both directions, thus not exhibiting systematic error qualities. This error may however influence our regressions through an increased variance. This bias can sometimes be compensated for by using a Heckman selection model, but this requires that we are aware of a missing observation. This method is often used in longitudinal studies or time series, where a specific observation has reported answers some years, but not
4

(Accelerace 2013; CAT Science 2013; Creandum 2013; DTU Symbion Innovation 2013; Innovation Midtvest 2013; Northcap Partners 2013; Northzone Ventures 2013; NOVI Innovation 2013; SEED Capital 2013; Sunstone Capital 2013; Syddansk Teknologisk Innovation 2013; VIA Venture Partners 2013; Vkstfonden 2013; stjysk Innovation 2013)

Page 29 of 108

others. However, as we at no point have an observation of the self-selecting funds, such methods have no corrective effect on our sampling.

3.2 SURVEY CONSTRUCTION

Our survey is constructed around two overarching areas of focus. One section consists of the questions regarding the effect of value-added contributions by the VC fund from the perspective of the portfolio company which will be used as dependent variables in the regression. The other section consists of questions regarding the characteristics of the firm, the founders, and the deal, which will be used as independent variables. In the first part we chose to apply validated questions rather than constructing tailored questions, which would potentially be prone to non-sampling error. Hence, we selected questions from the VICO project5. This survey was chosen based on a) a high degree of observable validity, and b) content validity in which the survey questions matched the value-added activities we wished to investigate. With regards to validity, the VICO survey is significant and reliable. First and foremost, it has been applied on a large scale, having more than 8000 respondents across seven European countries. This sample size is the largest in the literature on value-added activities to the authors knowledge. Second, the survey was constructed by researchers with more than 45 prior and relevant studies in the field of entrepreneurship (VICO Project 2013). Third, the survey questions have been used multiple times in peer-reviewed research (e.g. Bertoni, F., Croce, A. and A. Quas 2010; Bertoni, F. and T. Tykvov 2011; Luukkonen, T., Deschryvere, M., Bertoni, F. and T. Nikulainen 2011). Finally, the survey instrument is congruent with previous methodologies (e.g. Ehrlich et al 1994; Macmillan et al 1988) with respect to the format of the questions, in that it utilizes a seven-point Likert scale for the importance of the respective valueadded activities. With respect to content validity, the VICO project survey contained a total of 28 questions regarding value-added activities within similar categories as our hypotheses, representing a sufficient level of breadth, depth, and scope. For the second part concerning firm, founder, and deal characteristics we constructed our own questions based on what previous studies on value-added activities had included as variables. We created questions for all descriptive information, such as the age, industry, stage, education of
The project is a 7 university collaboration, sponsored by the 7th Framework Programme of the European Commission (theme SSH-2007-1.2.3 Grant Agreement 217485), and a copy of the survey was obtained through first-hand contact with a VICO-participant, Terttu Luukkonen.
5

Page 30 of 108

cofounders, previous funding rounds, previous ventures attempted by cofounders, and investment size. This part of our survey might suffer from a lack of validation. However, the questions are factual, making answers less prone to error, and could additionally be verified by secondary sources. Furthermore, our questions were partially validated through previous studies inclusion. An example of the survey can be found in Appendix III. Although the questions in the VICO project are validated from a sampling point of view, their theoretical foundation within the literature has not been accessible6. The following two sections will therefore provide a theoretical validity for their usage by comparing the questions in our survey to previous survey methodology, thus implicitly elaborating on the content validity of the survey. The first section examines the theoretical foundation of the 28 questions addressing the perceived value-add of the fund, after which a second section will assess the relevance of our descriptive characteristics of the firm, the founders, and the deal. 3.2.1 VALUE-ADDED ACTIVITIES The 28 questions on perceived contribution to value-added activities are grouped within eight categories similar to our hypotheses.
S TRATEGIC V ALUE - ADD : With respect to strategic value-add, the survey asked respondents for

the perceived importance of contributions of the VC fund to the portfolio company in developing or changing business plan, in determining the strategic focus of the company, and in developing capabilities in weak areas. Previous studies assessing strategic value-add asked for the formulation and evaluation of strategic plans (Fried, Bruton, and Hisrich 1998; Ehrlich et al. 1994; Gorman and Sahlman 1989; Macmillan, Kulow, and Khoylian 1988) development of business concept/strategies (Gabrielsson and Huse 2002), and the role of business advisor (Sapienza, Manigart, and Vermeir 1996). These distinctions are therefore highly related to the conceptualization given by the survey questions.
F INANCIAL V ALUE - ADD : To measure financial value-add, the survey asked respondents for the

perceived importance of contributions from the VC fund in obtaining non-equity finance from banks, companies or other sources, in raising follow-on financing, and attracting new Venture Capital investors. We find this distinction to be in line with previous studies distinctions, e.g. being a source of capital and helping obtaining additional investors (Maula, Autio, and Murray 2005; Gabrielsson and

We contacted the authors of the survey, who were unable to provide us with their questionnaire design process and justification.
6

Page 31 of 108

Huse 2002; Gomez-Mejia, Welbourne, and Balkin 1990; Gorman and Sahlman 1989) and helping obtain debt financing (Ehrlich et al. 1994; Macmillan, Kulow, and Khoylian 1988). Other previous studies have highlighted a broader financial competency building, financier and financial management role (Sapienza 1992; Gomez-Mejia, Welbourne, and Balkin 1990), which the survey omits leading to our financial value-add questions being less operational in nature.
C REDIBILITY V ALUE - ADD : Credibility value-add was determined by asking respondents for the

importance of the VC fund contribution with respect to: providing credibility to the company for other investors, providing credibility to the company for initial or later customers, providing credibility to the company for new suppliers or cooperation partners, and providing continued credibility to the company for recruiting employees. To the authors knowledge, no research examines the value of credibility from a survey and perspective based basis. Several studies operationalize the impact of credibility through IPO measures such as IPO success rate, valuation, lower under-pricing, and lower cost of underwriting, in relation to reputational proxies such as cumulative market capitalization, syndication (Nahata 2008; Hochberg, Ljungqvist, and Lu 2007; Dolvin 2005). Hsu (2004) measures the perceived reputation from entrepreneur perspective, although without measuring impact or perceived contribution. There is thereby no close comparison, but we still consider the perspective valuable, with reference to the literature review.
P ROFESSIONALIZATION V ALUE - ADD : Professionalization value-add was composed of four

factors of the VC funds contribution towards: controlling the companys cost base, developing corporate governance systems, changing/expanding the management team, and finding Board/Advisory Board members. This grouping of factors into a category coined professionalization encompasses a wider selection of variables than most other previous studies, but includes an aspect of most of them. Hellmann and Puri (2002) included hiring an outside CEO, the creation of HR policies, and recruitment of personnel, i.e. the same parameters. Other studies merely include elements such as the role of management recruiter (Ehrlich et al. 1994; Sapienza 1992; Macmillan, Kulow, and Khoylian 1988), the building of managerial competence (Gabrielsson and Huse 2002), and building an HR incentive structure (Gomez-Mejia, Welbourne, and Balkin 1990). We thereby consider the survey questions to be inclusive of what has previously been asked with respect to professionalization.

Page 32 of 108

M ARKETING V ALUE - ADD : Marketing value-add contributions are measured through the role of

the VC fund in setting up a sales and marketing system, in putting pressure on the company to generate first sale, and in putting pressure on the company to accelerate growth. We find less congruence between the applied survey and previous literature with respect to marketing value-add. However, some studies have still investigated it from a similar position. Macmillan, Kulow, and Khoylian (1988) and Ehrlich et al. (1994) tested for formulating marketing plans, testing or evaluating market plans, while Gabrielsson and Huse (2002) questioned for marketing competence, both of which arguably are similar in nature to setting up a sales and marketing system. Secondly, Hellmann and Puri (2000) apply a time-to-market hazard test, effectively similar to generating sales pressure. Pressuring the company for growth however, has not, to the authors knowledge, been a specific point of investigation. We still believe it to be reasonable that VCs will exert a similar type of pressure. Maula et al. (2005) furthermore measure the transfer of market and competitor knowledge. This factor would have been interesting to include, but for the sake of validity through choosing a complete and tested survey part, it had to be omitted.
T ECHNOLOGICAL V ALUE - ADD : To measure the technological value-add of the VC fund, the

survey asked respondents how the VC fund had contributed towards improving the R&D function of the company, building a strong legal Intellectual Property base and building partnerships for technological development. A very common way of operationalizing technological elements is through patents (e.g. Bertoni and Tykvov 2012; Caselli, Gatti, and Perrini 2009), which could have been a way to measure value-add. However, we are interested in the entrepreneurial perspective. Gabrielsson and Huses (2002) measurement of the building of technical competence from the VC fund, and Macmillan et al. (1988) and Ehrlich et al.s (1994) measure of the VCs perception of valueadd with respect to developing actual product closely resembles improving the R&D function of the company. While previous studies on technological partnerships do not address the issue of valueadd effect, we still argue for its relevance based on it being a research topic of interest in the literature (Lindsey 2008). Furthermore, we argue IP development is important because the patent base of startups has been examined by other studies, albeit from different perspectives.
I NTERNATIONALIZATION V ALUE - ADD : Internationalization value-add was measured through

the survey asking respondents for the importance of the VC funds efforts in helping the portfolio ventures finding marketing and distribution channels abroad, seeking equity financing abroad, recruiting management team members abroad, recruiting other staff members abroad, and looking for international board members/advisory board members. While several studies have been engaged in measuring the

Page 33 of 108

normative impact of internationalization in start-ups and how it is promoted through VCs (e.g. Fernhaber and Mcdougall-Covin 2009; Smolarski and Kut 2011), only two studies have been involved with value-added activities in internationalization. One is Lockett et al. (2008), which did not specifically ask for internationalization value-add, but relied on Macmillan et al.s (1988) survey base. The other, Maula, Autio, and Murray (2005), was limited to investigating the ability of the VC fund to attract foreign customers. Therefore, we believe that the survey we applied collects a broader and more detailed perspective on internationalization value-add activities.
E XIT O RIENTATION V ALUE - ADD : Finally, exit orientation value-add was measured by the

survey asking respondents about the contribution of the VC fund with respect to: preparing the company for an IPO, finding potential acquirers for a trade sale, and preparing the company for other potential exit routes. In line with previous studies, we distinguish between generally accepted desirable outcomes, i.e. IPO and trade-sale, and other outcomes (Isaksson 2006; Wang and Sim 2001). While type of exit has frequently been a dependent variable as a performance indicator, it seems to be modestly researched from a value-add activity and perceived importance perspective. We however still find the survey questions validated by the noted importance in the literature. In conclusion we find the validated questions used in the VICO project to have a similar structure to questions found in other surveys used in the literature addressing value-add. The next section will address whether the same can be said for the measurement of the descriptive characteristics that will constitute our independent variables. 3.2.2 F IRM , FOUNDER , AND DEAL CHARACTERISTICS The descriptive characteristics of the firm, the founders, and the deal consist of a maximum of 34 questions within 11 overarching categories:
S TAGE : We acknowledge that the definition of stages is not stringent (Ruhnka and Young 1987),

which could result in comprehension errors. We asked respondents to choose the stage that best describes their company out of an option set of Pre-Seed, Seed, Early-Stage, Middle-Stage and Expansion, which is similar to the definitions provided by DVCA (2013). Using five categories is an attempt to capture a degree of the uncertainty stemming from a) entrepreneurs varying subjective definitions of stage and b) the uncertainty of categorizing the company within these stage definitions.
I NTERACTION : As a control variable, we measure three types of interaction between the VC

fund and the portfolio company; time spent per week, how often interaction occurs, and

Page 34 of 108

meetings face to face. We measure the former by weekly hours, while we measure the latter two on an eight point scale from once a day through once every year. This is a similar distinction to Sapienza, Manigart, and Vermeir (1996), although we add an additional option to distinguish between once a semester and once a year.
I NDUSTRY : Given that our sample is inherently limited to the ICT and Life Science industry, we

included these options, along with an option for Cleantech as we had experienced definition overlaps in our informal phase between Cleantech and Life Science. We further included an Other option to capture eventual sample selection errors. The variable is necessary to allow control for industry specialization as it has an evident effect (see e.g. Knill 2009; Gompers, Kovner, and Lerner 2009)
F IRM AGE : We asked respondents for the firm age in line with previous literature, thereby

controlling for acquired experience and capability effects.

C OFOUNDERS : We included a question on the number of cofounders. StartupGenome (2011)

found that solo founders take 3.6 times longer to reach scale stage compared to a founding team of two and are 2.3 times less likely to pivot. While not a typical control variable seen in the literature, we wished to include it for the potential effects of a higher aggregate experience in the founding team, or a founding team disparity effect.
E DUCATION LEVEL : We asked respondents to supply the average educational level of

cofounders, categorized by high school or less, bachelor, masters, PhD, along with an other category to capture categorization error. Education constitutes a formal knowledge set that provides a certain level of capabilities, and influences the need for value-add from the VC, along with performance (see e.g. Behrens, Patzelt, Schweizer, & Brger 2012).
P REVIOUS FUNDING ROUNDS : We included a question on the amount of previous funding

rounds. As Gompers (1995) argues, it is an indicator of reduced uncertainty regarding the venture, due to a typical performance and/or information contingency on achieving follow-up finance. We further argue that previous funding rounds act a proxy for experience with VC funds.
P REVIOUS VENTURES ATTEMPTED : We requested that the respondent indicates previous

ventures attempted. Serial entrepreneurs have increased success rates (Zhang 2009; Gompers,

Page 35 of 108

Kovner, Lerner, & Scharfstein 2006), likely due to increased ability to manage startups. This implies less importance of value-add contributions from the VC fund.
I NVESTMENT SIZE : The size of the investment received by the portfolio companies is a standard

measure of categorizing Venture Capital funds (DVCA 2013). We include it to control for an eventual financial commitment influencing the attention given to the portfolio company by the VC fund because of the amount of capital at stake.
S YNDICATION : We included a question asking for the number of syndicated partners in the

investment. Syndicated partners have been shown to provide value-add, by having complementary skillsets in relation to other syndicate partners (Hochberg, Ljungqvist, & Lu 2007)
C ONTRACT DESIGN : We included a set of questions regarding contractual elements. In line with

Hirsch & Walz (2006) we asked the respondents to supply information on investment structure, the voting rights, the operational veto rights, the structural veto rights, and the liquidation rights of the contract as reported by the portfolio companies. We argue that the degree of control imposed by the contractual terms could be a method for VCs to exercise power over entrepreneurs.
F UND : Lastly, we included the ability to differentiate answers based on which fund the

respondent was currently associated with, given that funds have inherent characteristics that influence the value-add (e.g. Elango et al. 1995) , along with the ability to check for specialization

In conclusion we argue that the chosen questions for our survey are founded in literature and validated by prior research, implying that we are confident of the questions relevance to our assignment and thereby that content validity is ensured. However, our survey still might contain sources of error which will be elaborated on in the following section. 3.2.3 SURVEY ERRORS Despite a relatively low non-sampling error due to the validity of survey construction, the method applied is still prone to certain error measures. Additionally, given our relatively extensive survey, the number of questions totals a maximum of 65. Therefore our results are prone to respondent attrition, which potentially distorts responses occurring late in our survey.

Page 36 of 108

Finally, our survey was constructed strictly in English to keep the validity of the survey. Although we argue that the English proficiency in our sample is likely to be high (English First Ltd. 2012), we cannot exclude the possibility of comprehension errors as a result from not translating to the native language of each respondent.

3.3 DATA COLLECTION

The following section will detail the process of data collection of our survey to the 246 ICT and Life Science portfolio companies within Denmark-based Venture Capital funds. The section below characterizes the format of our survey, which will be followed by an elaboration on how we obtained contact information for portfolio companies. Finally a section will describe the potential errors such a data collection method entails. 3.3.1 FORMAT OF SURVEY AND EMAIL We used the Google Docs platform to convey our survey. An electronic survey enabled a convenient, time- and cost-saving data collection and allowed the survey to be based on adaptive design. Such a design was adopted to ensure a higher number of completed responses by making the total number of responses contingent on the answers provided 7 (Schouten, Calinescu, & Luiten 2011). Thus our survey contains a minimum number of questions of 53 and a maximum of 65 questions for a maximum of nine pages. Each email contained a description of our project, the VC funds we were collaborating with, any personal affiliation that might be relevant, and a pitch as to why their answer mattered. We also included feedback in the form of an executive summary of our finding for participating firms. Such personalization would theoretically increase our response rate (Cook et al 2000). 3.3.2 OBTAINING CONTACT INFORMATION The authors five semi-structured interviews included questions to get the VC funds acceptance of a) sending out a survey asking for potentially confidential contract information, b) stating that we are collaborating with their fund, and c) the individual investment manager to use his / her name when we contact the funds portfolio companies. We believe this acceptance that this was a helpful factor for obtaining higher response rates because we could mention the affiliation when asking portfolio companies to fill out the survey.

e.g. if the person answered to No when asked if their portfolio company had received previous funding, they would not be asked questions about previous funding
7

Page 37 of 108

All five funds authorized our use of their name after reviewing additional information about our scope and the relevance of the funds participation, as well as a draft email we would send to the portfolio companies. We then contacted the other funds and innovation incubators to get their support. One fund collaborated actively by providing the contact information of their portfolio ventures and allowing us to use their name. Two of the later stage funds allowed us to use their name, but did not assist us in any other manner. All the innovation incubators and one of the later stage funds did not wish to collaborate actively, but would not hinder us from contacting their companies as long as we did not use their name. Lastly, one fund explicitly requested that we refrain from contacting their portfolio ventures. Contact information in the form of email addresses and telephone numbers for the portfolio companies was obtained from investment managers of collaborating firms, press releases, portfolio company websites, the Danish company database, the Yellow Pages, LinkedIn, and personal networks. 3.2.1.1 I NVALID CONTACT INFORMATION ACTUAL SAMPLE Among the 310 deals in our sample, it was not possible to contact the companies of 35 deals. Seven of these deals were in fund that did not allows to contact its portfolio companies, while the rest were due to companies which were dissolved, had gone bankrupt since our initial search, or simply due to the lack of contact information. Therefore 275 surveys were sent to 246 portfolio companies in our defined sample group. 3.2.1.2 C OLLECTION PERIOD The data collection started on April 12th when the first surveys were sent out by email. To increase our response rate, a follow-up email was sent to the non-respondents after a minimum of seven days containing a quick reminder of our earlier email, a new pitch to our survey, and a link to the web survey. Reminder emails were sent out by May 3 at the latest. If we had their contact information, non-respondents were contacted by telephone three days after receiving the follow-up email and once again asked to help by filling out the survey. If we were unable to reach them by telephone, the remaining non-respondents were instead sent a final email on the last day of the survey asking for their help. At the end of the data collection (May 6th), we had received 81 responses for our survey.

Page 38 of 108

3.3.3 DATA COLLECTION ERRORS Several errors in the collection of data might affect our results. First and foremost, although we explicitly chose not to ask about relative merits of funds, the individual companies have a degree of uncertainty of the potential consequences of participating. This is exacerbated by the information requested in the survey being secretive at best and potentially directly confidential, exemplified by one funds explicit desire to not participate. There is thus a risk of provoking a low response rate through such a survey. This can be mitigated by using certain techniques as funds have very heterarchical relationships with their portfolio companies, such as addressing the recipient of the email by his/her name and clearly stating the name of the personal contact within the respective fund (Cook et. Al 2000). Second, our collection method potentially exhibits recipient bias, i.e. the recipient might have no knowledge about the questions asked. Here contact information provided by fund managers is expected to be of much better quality, as they will provide contacts to individuals who are used to interaction with the VC fund and who might be influenced by the reference of an investment manager. Third, given that we are measuring a perception that in its nature is subject to transient personal factors, we also potentially are exposed to this error. Given that this survey is cross-sectional, and that we due to the time constraint of this thesis are unable to conduct a lagged test for intertemporal consistency in responses, we are unable to control for these factors. We argue, however, that they are not likely to cause variable error, given that the transient personal factors even out, thereby not creating bias, but merely variance.

3.4 DATA CORRECTION

Out of a total of 81 responses to our survey, several had to be discarded altogether due to severe incompletion. The following section reviews the data corrections that were performed on the remaining responses in order to ensure coherent data. The main correction was converting the values of the size of investment into Danish Kroner (DKK) using exchange rates at the time of investment. Due to the different years of investments, we corrected for inflation to ensure all investments were in 2013 prices. A second correction was to construct dates in similar formats. For dates where only month and year were provided, the date was set to the first of the month. For non-public investments lacking date and month of funding, June was selected as the month. Finally, we contacted the fund to verify

Page 39 of 108

portfolio companies self-reported stage if a portfolio company receiving a very small investment had placed itself in a later stage, and vice versa. This did not result in any relocation of firms between seed-stage and early-stage, and provides investment manager verification of the company stage. A significant number of responses selected the Do not know choice for contractual information. Although we believe contractual information is a second proxy for interaction (type of interaction), our limited clout with the portfolio companies and the uncertainty of potential repercussions for the firm means that it has been unobtainable. We are therefore forced to omit the dummy variable Contractual_veto in our regression analysis. While such an omission aligns our model more closely with previous research on non-financial value-add, this would likely result from the sensitivity of the information rather than a disregard for its importance. After the data correction, our sample contained 69 usable responses, representing 25.1% of all the surveys sent (see Table 2 below).

TABLE 2: DISTRIBUTION OF VC-BACKED FIRMS IN SAMPLE AND IN RESPONSES

This concludes the methodology used to collect our sample of 69 responses. We highlight the apparent difficulties of obtaining high response rates, as seen per other surveys (Bottazzi et al. 1998; Barney et al 1996; Luukkonen et al. 2013; Maula et al. 2005) and therefore believe our response rate of 25.1% is strong and above expectation. We attempt to overcome these obstacles to collecting data by personalizing our research through numerous interactions with professionals in the Venture Capital environment. This extensive preparation especially aimed at reducing the investment managers perceived level of risk, as they act as gatekeepers to our sample of portfolio companies (Lee 1993). However, by establishing a degree of personal trust with various stakeholders we were able to gain complete and relevant email lists, added clout in our email by referencing the fund and the investment manager, and

Page 40 of 108

snowball networking: gaining new contact by being referenced by our initial set of contacts (Lee 1993). We therefore believe this data collection method to be worthwhile and to be a significant determinant for our high response rate. Now that we have collected our data, the following section outlines the methods of analysis used to assess whether Venture Capital funds value-added activities differ between stages.

3.5 METHODS FOR DATA ANALYSIS

This section elaborates on the various methods and reasoning used for our analysis of whether perceived value-added activities differs between stage, and whether funds specialize accordingly. The first section details the construction of variables to be used in our regression analysis. This will be followed by an overview of the principal component analysis, which groups the individual questions of perceived value-added activities within underlying factors, and the subsequent tests which ensure that our sample of 69 responses in appropriate for this method of analysis. The third section will elaborate on the Ordinary Least-Squares regression analysis and the step-wise construction of the model that will be used to model the perceived value-add activities against characteristics of the company and its founders to predict whether there is a significant difference between seed and early-stage ventures. Fourth, a section will detail the model that includes individual fund dummies. For those value-added activities which do differ between stages, such a model assesses the specialization of VC funds by seeing whether the effect of stage is significantly picked up by the fund dummies. Lastly, we conclude our methodology section by describing sources of errors in these various methods. 3.5.1 VARIABLE CONSTRUCTION AND DATA TREATMENT This section will describe how we constructed our dependent and independent variables to be used in the regression analysis from the questions in our survey. The descriptive pieces of information need to be transformed to meaningful and uncorrelated variables to be used in our regression analysis. Including an individual variable for each distinct answer to every question makes the regression very complex, potentially invalid, and will not necessarily add to our understanding of whether perceived value-added activities differ between stages. Here follows a description of each variable we use in our regressions:
F ACTOR 1,2N : Each of the 28 validated survey question from the VICO-project will be used to

assess the perceived contribution of the VC fund. Since these questions are related to certain areas of value-added activities, a factor analysis will be performed to reveal whether these

Page 41 of 108

variables can be explained by underlying factors. These factors can then be named and used as dependent variables in our OLS regression and allow for fewer regressions and broader degrees of inference.
C OMPANY S TAGE : We included five categories of stage to minimize the uncertainty of

categorization following vague definitions. These five categories were grouped in seed stage (preseed and seed) and early-stage (early-stage, late-stage, expansion). Thereafter a dummy variable for company stage was created which takes a value of 1 if the firm is in early-stage and 0 if in seed stage.
I NDUSTRY : Only respondents who reported their firm within ICT and Life Science were of

interest. Thus industry was modelled as a dummy variable taking the value of 0 if the firm is in ICT and 1 if it is in Life Science.
A GE P RIOR : AgePrior is a continuous variable of the age of the portfolio company at the time in

was invested in by the current VC fund, which is used to control for the level of previous acquired knowledge.
A GE F UND : This continuous variable is the other component of the age of the fund, used to

partial out any effect from having had varying degrees of contact with the fund on the perceived contribution to value-added activities.
E DUCATION P H D: Since there is a distinction in literature between the perceptions of PhDs and

non-PhDs, a single dummy variable for education was created, taking the value 1 if the cofounder has a PhD and 0 if he/she does not.
P REV V ENTURES : As we are simply interested in a distinction between whether the founders

have previous venture experience or not, rather than the amount of previous venture experience, we create a single dummy variable taking the value 1 if a prior venture has been attempted and 0 otherwise.
P RIOR F UNDING : A single dummy variable, taking the value 0 if the current the firms current

investment round is its first and 1 for all other values, to incorporate any experience gained from other funds.

Page 42 of 108

S YNDICATION P ARTNERS : The number of syndication partners was retained as a continuous

variable, as the maximal number of syndication partners in our sample (11) could imply a much lower need for the specific VC fund.
L OG I NVESTMENT S IZE : When plotting investment size, it seems to follow an exponential

development. As a consequence, the values were log-transformed as the OLS regression assumes linearity of the parameters. Plotting an exponential trend to verify linearity of the transformed variables yielded a measure of fit of 0.90 (Appendix IV).

F ACTOR I NTERACT : Since many of the questions of involvement received few responses, either

due to uncertainty or a reluctance to reveal that information, these questions could not be transformed into variables. As a consequence, veto rights, voting rights, liquidation rights, and investment structure were discarded. Only three questions within this category were fully completed, all pertaining to the degree of interaction between VC fund and the portfolio company. A factor analysis of these questions will be conducted to model an underlying factor of Interaction.
F UND A,B I : Our sample contains responses from portfolio companies within 10 different VC

funds, for which we construct 9 dummy variables, taking the value of 1 if the company is within the fund and 0 if the fund is not. The names of the funds were replaced by a letter in a randomized order to anonymize the responses. This concludes the review of variables we will use in our regression analysis. The following sections will elaborate on the two main types of analysis used, principal component analysis and Ordinary Least-Squares and how they will be used to answer our research question of how VC funds perceived contributions of value-added activities differ between company stages and whether funds specialize accordingly.

3.5.2 ANALYSIS METHODS FOR FACTOR ANALYSIS We will conduct an exploratory factor analysis to construct our dependent variables and a factor for interaction between portfolio companies and VC funds. This allows us to validate that the eight groupings of the 28 questions on value-added activities from the VICO project are observed in our sample. It also allows us to conduct simpler OLS regressions and provide sample inferences on underlying factors rather than individual activities (Cramer, 2003). This

Page 43 of 108

section will elaborate on the principal component analysis that will be used as well as the subsequent tests which ensure such method of analysis is appropriate. The principal component analysis is a version of factor analysis, where the variability of observed random variables is explained in terms of fewer unobserved random variables called factors. In other words, the method explains the variability of the answers to the 28 questions on perceived value-add through fewer factors, such as internationalization importance, which cannot be directly observed. Using these factors in a regression analysis allows us to test whether the coefficient of CompanyStage has a statistically significant effect of on the perceived contribution of the underlying factors. We assume the question responses are interval data, implying equidistance between response options, which allows us to conduct parametric tests, hereunder a factor analysis (Williams and Monge 2001). We further assume that there is some degree collinearity among the 28 questions, that the relationships among the pairs of questions in each factor are linear, and that there is an absence of outliers. Finally we assume that our sample of 69 responses is large enough to yield reliable estimates of correlations among the individual questions. In order to legalise the use of a principal component analysis, the sample must meet be subject to Bartletts Test of Sphericity and Kaiser-Meyer-Olkins (KMO) Measure of Sampling Adequacy. Bartletts Test of Sphericity tests the null hypothesis that the intercorrelation matrix comes from a population in which the variables are noncollinear. Failure to reject this hypothesis implies that each question has a correlation with itself only, allowing the correlation matrix to be reduced to an identity matrix. KMO tests for the degree of common variance among the factors. Pett, Lackey, and Sullivan (2003) stated a value of 0.7 as a sufficient limit in order to conduct a factor analysis. These two tests imply that there is some observed correlation which warrants a factor analysis. To determine the number of factors a principal component analysis including all 28 questions will be conducted. Any individual questions with low degrees of communality will be discarded as they are not suitable for a factor analysis. The number of factors which should be constructed from remaining questions will be determined by eigenvalues and a screeplot. Eigenvalues above

Page 44 of 108

one signify that the factor explains more variance than a single variable, while a screeplot visualizes the added explanatory power from including an extra factor (Cramer 2003). An unrotated solution has the first factor describing most of the variability, and as we would rather spread the variability more evenly among the eight factors, the solution will be rotated if there is a large difference in the proportions of variance explained. Rotation doesnt improve the fit of the factors, but it makes the factor loadings as high or as low as possible in order to better interpret them. The factors can then be constructed based on how the individual questions load on the factors. Thereafter a factor analysis for each individual factor using a principal component analysis will be conducted to ensure that each factor has variables with low uniqueness and a single parameter with an eigenvalue over 1. Finally, we conduct a test of the factors reliability. Cronbach's alpha is an index of reliability associated with the variation accounted for by the true score of the underlying factor that is being measured. Coefficients of Cronbach alpha tests range from 0 to 1, with a higher score indicating a more reliable scale. Scores 0.7 and above are acceptable values (Kline 1994). This principle component analysis, with its validity founded in the related tests, allow the use of fewer, unobservable factors as regressands in an Ordinary Least-Squares regression, which will be elaborated on in the following section. 3.5.3 ANALYSIS METHODS FOR OLS REGRESSION This paper uses a stepwise multiple regression analysis, using the regression method of Ordinary Least-Squares (OLS) which fits a linear equation that minimizes the sum of squared residuals to build a model for predicting the role of stage on the factors of perceived contribution of the VC funds. Since we assume the question responses are interval data and we conduct a factor analysis, these variables can be modelled as dependent variables in a regression. The individual factors will be regressed on the nine independent variables to control for company and founder characteristics. Thus the framework of the model will be:

The aim of OLS is to estimate the true parameters (

which explain the perceived value-add by

the funds (dependent variables). OLS is the best, linear, unbiased estimator of the true parameters conditional on sample values of the explanatory variables if the Gauss-Markov

Page 45 of 108

assumptions of linearity of parameters, sample variation, random sampling, zero conditional mean, homoskedasticity, and no serial correlation of the error terms (Wooldridge 2009). Therefore all regressions will be tested for multicollinearity using variance inflation factors (VIF). The VIF states how much the variance of the coefficient is inflated. A value above 10 implies they are faulty, while values above 5 warrant further investigation. Furthermore, a Breusch-Pagan test and a White test will be conducted for each individual regression used to test for heteroskedasticity. Regression analysis allows us to draw ceteris paribus conclusions about how the independent variable affects the dependent variable that is; what is the effect of a one-unit increase in on when holding all other variables constant. By modelling the regression to include control variables we clean the sample relationship between stage and perceived value-add factor. Put in another way, the model measures the effect of stage on perceived value-add after the control variables have been partialled out. When interpreting the
-coefficients

predicted by the regression we use several statistical

measures of explanatory power. First we test the null hypothesis that all the independent coefficients are equal to zero using an F-test statistic at a 10% significance level. Rejecting this null hypothesis is a prerequisite in order to make any inference of the individual coefficients. Thereafter we can test the null hypothesis that each -coefficient is equal to zero, using a t-test statistic at a 10% significance level. A significant coefficient indicates that the modelled coefficient is the same as the true population parameter at the 90 % confidence level. As a measure of the models overall explanatory power on the individual factors we use adjusted R2. Although R2 shows that amount of observed variance explained by the model, we prefer to use the measure of adjusted R2 since adding an extra variable to the model will always explain at least as much of the observed variance as before. This also costs a degree of freedom, so the adjusted R2 takes the number of variables into account with a drop in the adjusted R2 value implying that the cost of a degree of freedom is greater than the benefit of the variables ability to explain the observed variance of our sample. Thus, very little explanatory power is lost with the variables omission. The model will be constructed over four stages. The first three stages progressively add control variables to the regression of on company stage to ensure that these variables explain a certain degree of the observed variance in our sample. The fourth stage will assess the effect of

Page 46 of 108

excluding non-significant variables to improve the adjusted R2 value of our model. This model will allow us to assess whether stage is a significant predictor of the factors when holding all the control variables constant. In other words, for what value-added activities do the perceived VC funds contributions differ between stages? 3.5.4 INCLUDING FUND DUMMIES Building on the model above, a fifth and final regression adding nine dummy variables for the individual funds will be constructed to assess whether funds specialize accordingly. First, if the variable CompanyStage was significant in the model without the fund dummies, but loses its significance when the fund dummies are included, then this indicates that the funds are relatively specialized in these areas of value-add as they account for the observed variance which was previously attributable to stage. In other words, the effect of stage is now picked up by the fund dummies. Second, if the individual fund dummies are statistically significant predictors of value-added activities where stage was previously a significant coefficient, then these funds are specialized in this value-added activity relative to the other funds in our sample. If the variable CompanyStage was insignificant in the model without the fund dummies, then the fund dummies have no ability to predict whether the funds are relatively specialized in these value-added activities. 3.5.5 ANALYSIS ERRORS Factor analysis is prone to several errors. The main issue is the subjective naming of the factors. Although several questions might appear to be predictors of an underlying factor, they might not be as interpretable as the namings suggest. Furthermore, a name implies that the factor contains all its predictors, and any inferences will be significantly invalid. Finally, a correlation matrix can be critiqued as a poor measurement of association between input variables (Kline 1994) For our regression analysis, we first and foremost might suffer from an omitted variable bias. If a variable which has a causal effect on the perceived contribution of VC funds value -adding activities, its omission from the model will cause the other variables to compensate for it, resulting in their over- or underestimation. If our model suffers from omitted variable bias, then the error term will no longer be uncorrelated with the regressors and the Gauss-Markov assumptions of OLS providing the best, linear, unbiased estimator will no longer be valid. (Wooldridge 2009)

Page 47 of 108

However, this does not warrant the inclusion of as many variables as possible. While omitting a relevant variable may cause the other regressors to be biased, the inclusion of an irrelevant variable will increase the variance of the other regressors, unless the correlation between the extra variable and the relevant variable is equal to zero (Wooldridge 2009). It is therefore imperative that the inclusion of an added variable has a theoretical foundation (Clarke 2005).

Page 48 of 108

4. RESULTS
To assess the relative importance of value-add and its relation to company stage, this thesis will use a principal component analysis to group the questions of perceived value-add into underlying factors. This analysis transforms the questions into continuous variables, allowing them to be modelled as dependent variables in an OLS regression. This determines whether company stage is significantly related to the different factors of value-added activities when the effects of the other independent variables have been partialled out. This analysis section will consist of three main sections: First we conduct the principal component analysis of the 28 questions on value-added activities, as well as the three questions of interaction. Thereafter we construct our multivariate linear regression model assessing the how value-added factors differ between stages through a four-step process of adding variables, assessing their explanatory impact, and omitting those with minimal value to the model. This section will also assess the specialization of VC funds in value-added activities that are significantly determined by stage by including dummy variables for each individual fund. Finally, the third section will evaluate the results in relation to our hypotheses.

4.1 FACTOR ANALYSIS

This section will first extensively elaborate on the results from the principal component analysis of the 28 questions to be used as dependent variables, after which the results from an identical analysis of the 3 questions on interaction will briefly be summarized. 4.1.1 FACTOR ANALYSIS OF DEPENDENT VARIABLES The results of our two tests of factorability, Bartletts Test of Sphericity and Kaiser-MeyerOlkins (KMO) Measure of Sampling Adequacy, are displayed in Table 3 below:
TABLE 3: FACTORABILITY TESTS

Bartletts Test of Sphericity for noncollinear variables returns a chi-squared value of 1569.5 and allows us to reject the null hypothesis that the correlation matrix is an identity matrix above the

Page 49 of 108

99% confidence level. KMO score of 0.794 indicates a value bordering on Meritorious (Kaiser 1974, referenced in Pett et al 2003) and implies that the factors extracted from a factor analysis will account for a substantial amount of the variance (for a table of each questions individual KMO score, see Appendix IV).
TABLE 4: ROTATED PRINCIPAL COMPONENT ANALYSIS

The factors from the rotated principal component analysis are displayed in Table 4 (above). The analysis implies that the 28 questions should be reduced to eight factors, based on the output of our principal component analysis, eigenvalues, and a screeplot (see Appendix V). Each of these eight factors explain more variance than an individual variable and the cumulative variance explained by these eight factors is 0.77350, implying 77.35% of the common variance shared by the 28 variables is accounted for by the eight factors. The rotated solution had uniqueness scores below 0.35 for all individual questions such that they all should be included in the factor analysis. From the factor loading matrix in Table 6 (below), certain variables cross-load on several factors, but we choose to retain the eight factor groupings proposed in our literature review for sake of interpretation. A factor analysis for each individual factor using a principal component analysis verified that each factor had a single factor with an eigenvalue above 1 and that all variables had low levels of uniqueness (see Appendix VI). Furthermore, the overall Cronbach Alpha score for each factor is greater than 0.7, implying these values are acceptable (Cramer 2003) and that our factors are reliable (see Table 5 below).

Page 50 of 108

TABLE 5: CRONBACH ALPHA SCORES

Variable FactorStrategy FactorTechnology FactorMarket FactorProfessionalization FactorFinance FactorCredibility FactorInternational FactorExit

Cronbach Alpha scores 0.8525 0.8778 0.8418 0.8518 0.7808 0.9091 0.9242 0.7611

TABLE 6: FACTOR LOADINGS MATRIX

Page 51 of 108

4.1.2 FACTOR ANALYSIS OF QUESTIONS ON INTERACTION The output from the principal component analysis of the three questions pertaining to interaction can be seen in Appendix VII. The analysis only found one factor to have an eigenvalue greater than one and having a Cronbach Alpha score of 0.738. Furthermore, the factor is deemed valid as it explained 67.62 % of the total observed variance.

4.2 REGRESSION ANALYSIS

The factors constructed in section 4.1 will be used as dependent variables in the following section to model the effect of stage on perceived value-add. First an overview of all the variables will be provided, after which the multivariate model will be constructed. The first three models assess the effect of stage on the factors of perceived value-added activities by step-wise adding additional variables. Model IV will omitting the variables with minimal value to the model through a backward selection process. Finally, model V will include dummy variables for each individual fund to assess the specialization of VC funds in value-added activities that are significantly determined by stage. 4.2.1 DESCRIPTIVE STATISTICS An overview of the variables constructed from our 69 sample responses can be seen in Table 7 and Table 8 (below)
TABLE 7: DESCRIPTIVE STATISTICS: VALUE-ADD IMPORTANCE

Page 52 of 108

TABLE 8: DESCRIPTIVE STATISTICS: INDEPENDENT VARIABLES

For our dependent variables, a positive (negative) factor score implies that company

will value

the funds contribution to the value-add factor relatively higher (lower) than other firms in the sample, since factor scores are standardized with a mean of 0 and a standard deviation of 1. Within our independent variables, the average portfolio company received a fund investment of 8.6 DKKm, but the range of investments is much larger, with the smallest investment at just 200,000 DKK and the largest investment 110.4 DKKm. Industry mean of 0.55 shows a slight majority of Life Science firms in our sample (38 versus 31). The age of the firms in our sample was varied: some firms are founded by a fund investment, as indicated by a minimum value of 0 in age_prior, while the portfolio company had a track record of over nine years before joining its current fund. When measuring the portfolio companys age within the fund, the results are just as varied, ranging between less than three months to over ten years. Our sample contains 24 cases where the founders on average have a PhD, 46 cases where the current firm was not the founders first venture, and 26 cases where the current firm had received at least one prior round of funding. These 69 investments contained a total of 96 syndication partners ranging from 0 to a maximum of six. The following pages display the correlation matrix containing for all the variables (see Table 911). We use a cut-off limit of 0.6, but none of our independent variables are correlated more than 0.54. This allows for the variables inclusion in a multivariate regression analysis.

Page 53 of 108

TABLE 9: CORRELATION MATRIX (1/3)

Page 54 of 108

TABLE 10: CORRELATION MATRIX (2/3)

Page 55 of 108

TABLE 11: CORRELATION MATRIX (3/3)

Page 56 of 108

We can thus include all our variables to model the effect of stage on perceived value-add, regressing each of the factors for value-add constructed in section 4.1 on the independent variables shown in Table 9-11 (above) 4.2.2 MODEL I: FACTORS REGRESSED ON STAGE This first section models the simplest relationship, with the results being displayed below:

TABLE 12: REGRESSION I OUTPUT

None of the eight regressions have a sufficient t-test statistic to reject the null hypothesis that CompanyStage is equal to 0 at the 90% confidence level (see Table 12). The R2 values are all below 0.03, but the model is consistent with the Gauss-Markov assumptions as the single variable has no VIF and both the Breusch-Pagan test and the White test for heteroskedasticity fail to reject the null-hypothesis of homoskedasticity (see Appendix VIII) 4.2.3 MODEL II: FACTORS REGRESSED ON STAGE AND CHARACTERISTICS This model includes the firm characteristics, with the regression output stated below:

Only regression 8 has a sufficient F-test statistic to reject the null hypothesis of all independent variables being equal to 0 at a 90% confidence level (see Table 13, next page) It is also the only regression that explains more than 10% of the variance when adjusting for the number of parameters with an

Page 57 of 108

TABLE 13: REGRESSION II OUTPUT

Page 58 of 108

adjusted R2 = 0.159. Variance Inflation Factors in all regressions are below 1.8, and BreuschPagan tests and the White tests for heteroskedasticity fail to reject the null-hypothesis of homoskedasticity in all eight regressions (see Appendix VIII). Regression 8 states that Industry and Age_prior are statistically significant predictors of perceived value-add at the 99% confidence level, with Life Science firms increasing the factor by 0.83, while Age_prior coefficient of 0.076 implies the impact of age on the factor is positive. Prior_funding_rounds is negatively significant at the 90% level, implying that having received previous funding rounds reduces the perceived value of Exit. 4.2.4 MODEL III: FACTORS REGRESSED ON STAGE, CHARACTERISTICS , AND INTERACTION This model includes the interaction factor as an independent variable. The results are displayed below:

The four regressions of Strategy, Finance, Credibility, and Exit all have F-test statistics which reject the null hypotheses that all independent variables are equal to zero at the 95% confidence level (see Table 14, next page). For Professionalization and Technology, respectively, the probability of getting a value larger than the F-test statistic is 5.5% and 5.6% allowing for a rejection of the null hypothesis for these regressions at the 90% significance level. These 6 regressions have adjusted R2 values of 0.25, 0.14, 0.21, 0.23, 0.12, and 0.12, respectively. Variance Inflation Factors, Breusch-Pagan tests, and the White tests all fail to violate the Gauss-Markov assumptions in all 8 regressions (see Appendix VIII). In these six regressions, the interaction variable is significant at the 95% level for Exit and at the 99% level for Strategy, Technology, Finance, Professionalization, and Credibility. Since the questions underlying FactorInteract have an inverse scale, with a high level of interaction denoted by a low number, a negative coefficient implies that more interaction has a positive impact on the perceived value for that factor. CompanyStage is only a significant predictor for Strategic at the 95% confidence level, implying later stage firms have a positive impact on the perceived strategic value-add.

Page 59 of 108

TABLE 14 REGRESSION III OUTPUT

Page 60 of 108

Industry is a significant predictor in five of the regressions, with Life Science firms having a positive impact on value-add perceptions. The factors of Technology and Professionalization find the impact significant at the 90% confidence level, Strategy and Finance find it significant at the 95% level, and Exit finds it significant at the 99% level. Credibility finds no significant impact on the perceived value when distinguishing between Life Science and ICT industries. AgePrior is statistically significant at the 95% level for Strategy, at the 90% level for Technology, at the 99% level for Exit, with a positive impact of an extra day on the perceived value of the factor. PrevVentures is statistically significant for Finance at the 95% level with a positive impact of previous venture experience on the perception of value. For Credibility, the coefficient has a ttest statistic of 1.65, implying significance at the 89.5% confidence level, while the coefficients in the other four regressions are far from significant. PriorFunding is a significant predictor of the perception of value of the factors Credibility and Exit at the 95% confidence level. The impact is negative as having received previous funding rounds significantly lowers the perceived value. LogInvestmentSize is a statistically significant negative predictor of the impact on the perceived value of Technology at the 90% confidence level. Increasing the size of the investment implies a lower perceived value of the funds contribution to Technological position. Finally, the number of syndication partners and the founders average education are statistically insignificant predictors of the impact on perceived value in all regressions. 4.2.5 MODEL IV BACKWARD SELECTION OF MODEL OF STAGE ON VALUE-ADDED ACTIVITIES AgeFund, SyndicationPartners, and EducationPhD have no statistically significant predictive effect on the factors of value-added activities, and this section assesses the case of their exclusion. The regression model has been tested for their individual exclusion, as well as their joint exclusion, with the highest adjusted R2 value being found from the model which excludes all three variables. The order they are excluded has no impact on this result, and thus the following section will assess their exclusion on a step-by-step basis. Removing AgeFund now allows us to reject the null hypothesis that all independent variables are equal to zero for Professionalization at the 95% level (see Table 15, below). Furthermore, three out of the six significant regressions have improved R2 values, and LogInvestmentSize is no longer a significant predictor of perceived value-add for Technology. This variable will not be assessed for exclusion, however, as its significant predictive effect in the model before implies

Page 61 of 108

that it explains a significant portion of the observed variance in our sample. This is highlighted by being significant at the 85% confidence level in this regression. Removing SyndicationPartners improves the R2 of seven out of eight regressions and allows us to now reject the null hypothesis that all the independent variables equal to 0 at the 95% significance level for all 6 regressions. Removing EducationPhD further allows us to reject the null hypothesis that all independent variables are equal to 0 for the Technology regression at the 90% significance level. This comes at a cost, however, as this regression only rejects the null hypothesis for Market at the 90% significance level. This model improves the adjusted R2 for four regressions, while worsening it for the other four. This is our final model for predicting the impact of company stage on the perceived value of each factor, with the output presented in Table 15 on the following page:

Page 62 of 108

TABLE 15: REGRESSION IV OUTPUT

Page 63 of 108

Seven out of eight regressions have F-test statistics that reject the null hypothesis that all independent variables are equal to zero at the 90% significance level. Variance Inflation Factors fall slightly to a mean value of 1.33, while the Breusch-Pagan tests and the White tests all fail to reject the null hypothesis of homoskedasticity (see Appendix VIII). CompanyStage has a statistically significant positive impact on perceived value-add for Strategy (95% confidence level), Market (90% confidence level), and Credibility (90% confidence level). Compared to the regression output in model IV, six out of the seven regressions report a higher adjusted R2 value. Industry is no longer a significant predictor of the value-add for Technology or Finance, while AgePrior is now has a significant positive impact on all value-add factors at the 90% confidence level, except for Finance. LogInvestmentSize has a negative significant impact on the perceived value of Technology at the 90% confidence level, while having attempted previous ventures positively and significantly impacts the perceived Finance value-add at the 95% confidence level. Finally, having received prior rounds of funding has a significant negative impact on the perceived value of the factors of Finance, Credibility, and Exit at the 95%, 99%, and 99% confidence level, respectively. 4.2.6 MODEL V CONTROLLING FOR INDIVIDUAL FUNDS This model assesses the specialization of VC funds in value-added activities of Strategy, Market, and Credibility, which are significantly determined by stage, by including dummy variables for each individual fund. The results are displayed in Table 16 on the following page. Including fund dummies affects the degrees of freedom, which only allows us to reject the null hypothesis that all the independent variables are equal to 0 at the 85% confidence level for seven regressions. Six out of seven regressions report higher adjusted R2 values and for all seven of the regressions the variables explain at least 31% of the total observed variances. The Variance Inflation Factors increase to a mean value of 1.90 with a maximum value of 3.25, which still implies that the regressions do not exhibit multicollinearity. The added variables render the general case of the White test useless, and instead a special case of the White test was conducted (see Appendix VIII for output and Appendix IX for theory), which preserves the spirit of the White test with squares and cross products, but simultaneously conserves degrees of freedom. Together with the Breusch-Pagan tests, the seven regressions that are of interest all fail to reject the null hypothesis of homoskedasticity.

Page 64 of 108

TABLE 16: REGRESSION V OUTPUT

Page 65 of 108

When controlling for the individual funds, the variable CompanyStage is no longer significant, suggesting that there is no significant impact of various stages within a given fund on perceived value of the seven factors. Instead, the funds themselves have a significant impact on the factors Strategy, Technology, Market, Finance, Credibility, and Exit. Only for the factor of Professionalization is there no significant impact of the various funds in the perceived value -add. When including fund dummies, FactorInteraction is no longer a significant predictor of perceived Market value-add. It is still, however, a significant predictor of the other 6 models that reject the null hypothesis of all variables equal to zero at the 95% confidence level for Exit and at the 99% at the others. Within the perceived value-add of the factors for Strategy and Market position, Funds A, F, and G are statistically significant predictors and have a positive impact at the 90% confidence interval. Fund A is also the only fund which has a statistically significant coefficient in the regressions for Credibility and Exit, with both regressions implying a negative impact on the respective factors of perceived value-add at a 90% confidence level. Fund A (95% confidence), C (95% confidence), D (90% confidence), and G (90% confidence) all have statistically significant negative impacts on the perceived value-add of Technological position at the 90% confidence level, while Funds A, B, C, D, and E all have a statistically significant negative impact on the perceived value-add of Finance. Fund A is significant at the 99% level of confidence, fund D and E are significant at the 95% level, while fund B and C are at the 90% level.

Page 66 of 108

4.3 HYPOTHESIS EVALUATION

Our hypotheses will be evaluated based on model iv with respect to hypotheses 1-8 and model v with respect to hypothesis 9. H1: There is no significant effect of stage on strategic value-add The coefficient of CompanyStage in the regression of the factor of perceived strategic value-add has a p-value below 0.05. We therefore reject H1 at the 95% confidence level, and conclude that stage has a significantly positive impact on the perceived level of strategic value-add. H2: There is no significant effect of stage on financial value-add The coefficient of CompanyStage in the regression of the factor of perceived financial value-add has a p-value above 0.1. We therefore fail to reject H2, and conclude that stage has no significant impact on the perceived level of financial value-add. H3: There is no significant effect of stage on credibility value-add The coefficient of CompanyStage in the regression of the factor of perceived credibility value-add has a p-value below 0.1. We therefore reject H1 at the 90% confidence level, and conclude that stage has a significantly positive impact on the perceived level of credibility value-add.

H4: There is a negative effect of stage on professionalization value-add The coefficient of CompanyStage in the regression of the factor of perceived professionalization value-add has a p-value above 0.1. We therefore find no support for H4 and conclude that stage has no significant impact on the perceived level of professionalization value-add. H5: There is a positive effect of stage on marketing value-add The coefficient of CompanyStage in the regression of the factor of perceived marketing value-add has a p-value below 0.1. We therefore fail to reject H5 at the 90% confidence level, and conclude that stage has a significantly positive impact on the perceived level of marketing value-add.

Page 67 of 108

H6: There is a negative effect of stage on technological value-add The coefficient of CompanyStage in the regression of the factor of perceived technological value-add has a p-value above 0.1. We therefore find no support for H6 and conclude that stage has no significant impact on the perceived level of technological value-add. H7: There is a negative effect of stage on internationalization value-add The regression of the factor of perceived internationalization value-add on selected portfolio company characteristics and interaction has a p-value of the F-test statistic above 0.1. We therefore find no support for H7 and conclude that stage has no significant impact on the perceived level of internationalization value-add. H8: There is a positive effect of stage on exit value-add The coefficient of CompanyStage in the regression of the factor of perceived exit valueadd has a p-value above 0.1. We therefore find no support for H8 and conclude that stage has no significant impact on the perceived level of exit value-add. H9: Given a significant effect of stage, the effect of individual funds removes the significance of stage In the regressions of the factors of perceived strategic, credibility, and market position value-add, the coefficients of CompanyStage have a p-value above 0.1. We therefore fail to reject H9 at the 90% confidence level and conclude that individual funds are specialized to a certain degree.

Page 68 of 108

5. DISCUSSION
In the previous section we presented our findings and found a) support for hypotheses 2 and 5, b) no support for hypotheses 4 and 6-8, and c) significant, but divergent evidence for hypotheses 1 and 3. Lastly, our results were in concordance with the prediction posited in hypothesis 9. In the following section we will discuss these findings in depth and place them in relation to existing evidence. We furthermore include relevant observations from interviews with investment managers from five of the 10 funds that were represented in our observed funds.

5.1 HYPOTHESES 1-8: VALUE-ADDED ACTIVITIES AND STAGE

Our results indicate that marketing value-add increases in effect the later the stage as predicted, with moderate support at the 90% level. We interpret this result as being due to marketing valueadd exhibiting more of a strategic nature, rather than an operational one, the later the stage of the company. We support this interpretation by the proposition in Macmillan, Kulow, and Khoylians (1988) study from the VC funds perspective, where the results of conflicting importance and involvement is explained by operational marketing efforts being too costly. These results are substantiated by the perceived importance of marketing value-added activity ranking 6th out of eight in our sample. This corresponds to a relatively low importance valueadded activity. We argue that this highlights a certain level of congruence with the normative results of Macmillan, Kulow, and Khoylian (1988) and other previous studies (e.g. Maula, Autio, and Murray, 2005; Gabrielsson and Huse, 2002; Ehrlich et al., 1994). Our results thus lend to the notion that stage specialization with respect to marketing value-added is in fact worthwhile, in that companies in the later stages value the contribution of these value-added efforts significantly more so than earlier stages. This notion was furthermore supported by F. Knudsen (2013) who highlighted that they, as an earlier stage oriented VC fund, hired outsiders to support the funds marketing value-added contributions. Furthermore, our results show that financial value-add is unrelated to stage, with moderate support at the 90% level. This signifies that the effect of financial value-add is valued irrespective of stage, in congruence with our hypothesis nr.5. We interpret this result as corresponding to a continuous need for financial value-add in the portfolio companies, to develop and expand the business and to progress to the later stages. This is in line with the characterizations of Smith, Smith, and Bliss (2011). Our data seems to be somewhat in accordance with previous studies with respect to the rank of opinion means. In our sample, the mean importance of financial value-added ranked of 3rd out of

Page 69 of 108

eight and is thereby similar to previous results (Maula, Autio, and Murray, 2005; Gabrielsson and Huse, 2002; Ehrlich et al., 1994; Gorman and Sahlman, 1989; Macmillan, Kulow, and Khoylian, 1988). These results highlight that the contribution of VC funds financial value -added activities is relatively highly valued, and is so irrespective of stage. We therefore argue that from a valueadded perspective, our results indicate that stage specialization, within this particular area of value-added activities, has little effect. In other words, all VC funds need to have competencies within financial value-added and the ability to transfer them, irrespective of the development stage of their portfolio companies. We find support for this notion by investment managers from funds focusing opposite ends of the development cycle, highlighting the ability to raise follow-up finance as one of their core activities (J.Knudsen 2013, Wylonis 2013). Our results found no significant evidence that professionalization value-add is perceived more important in the earlier stages of the portfolio companies development. We therefore conclude that this area of value-added activities seems to not depend on the stage of the portfolio company. This implies that VC funds engage in aiding their portfolio companies with corporate governance, HR policies, board, management, and advisor setup across all stages. These results conflict with those of Hellmann and Puri (2002), who found professionalization to be significantly more important at the earlier stages. We also find that the respondents in our sample on average ranked professionalization higher than other previous studies. While we acknowledge that interpretation of the ranking of mean importance is an imperfect basis for comparison across studies, we highlight that the respondents in our sample ranked professionalization 2nd out of our eight value-added categories. This is a considerably higher ranking of professionalization efforts than Gomez-Mejia, Welbourne, and Balkins (1990) dyadic interview-based study. It is also substantially higher than in the findings of Gabrielsson and Huse (2002) who framed their questionnaire of VC funds in a similar manner. This lends to a proposition that professionalization is a significant value-added activity, and that the VC funds contributions to the portfolio company are valued across stages of development. Thereby our results indicate that stage specialization within this area of value-added activities has little effect. It is a value-added activity that every VC fund should have the ability to assist their portfolio ventures with, regardless of a funds focus. This contrasts the view of Accelerace Partner Jesper Knudsen, who argues that earlier stage ventures would value VC funds human capital contribution more than later stage ventures, as these ventures are more open about their limitations and acknowledge that their team is incomplete (J.Knudsen 2013).

Page 70 of 108

Furthermore, technological value-added contributions were not found to be significantly more applicable in the earlier stages, unlike what we hypothesized. While some of this result can be arguably attributed to the overall low importance of VC fund contributions to this type of valueadded, we argue this result should primarily be attributed selection efforts. Specifically, we argue that when VC funds screen portfolio companies they are particularly focused on the prospective companies level of innovativeness and consequent technological. We therefore believe that the underlying mechanism at work here is that VC funds choose to invest in ventures that are already innovative and capable within R&D, as explained by Caselli, Gatti, and Perrini (2009) and supported by Corfitzen (2013). Furthermore, our data shows that technological value-added ranks 7th out of our eight valueadded categories which could arguably explain a part of why this value-added activity does not differ between stages. This relatively low importance is consistent with previous studies of technological value-added activities of Gabrielsson and Huse (2002), Ehrlich et al. (1994), and Macmillan, Kulow, and Khoylian (1988). These previous studies examine the importance of technological value-added activities from a dyadic, portfolio company, and VC fund perspective, respectively, which lends to an overall high degree of consistency of the results. We do not find support of the notion that contributions from the VC fund to the portfolio companies with respect to internationalization value-added activities is valued higher in the earlier stages. The results are therefore inconsistent with Lockett et al.s (2008) study. In their study, they find evidence that earlier stages express that they benefit more from value-added activities of the VC fund with respect to internationalization. Although we find no evidence that the perceived needs of portfolio companies differ between ICT and Life Science industries, a possible explanation is that the needs for internationalization are different in nature, as Lockett et al.s (2008) sample consists of a much broader array of industries. Another possibility is that Lockett et al. (2008) measure internationalization through export intensity, and value-add through a methodology that does not specifically ask for the importance of internationalization value-added activities, but a broad variety of value-added activities. When assessing the internationalization value-added activity and it contributory importance, this value-added activity ranked 8th out of the eight value-added activities. Thus, we find our results consistent with Maula, Autio, and Murray (2005). Our interpretation of the relatively low importance is that not all companies operate internationally, let alone intend to, while others are effectively born global. Consequently, our results indicate that stage specialization with respect to internationalization has little effect, whereby resources and competency building for the VC

Page 71 of 108

fund would have a higher impact elsewhere. We do not, however, argue that internationalization efforts generally are without effect. Internationalization can foster a better ability to select ventures (Corfitzen 2013), and foster syndication that adds competencies to the portfolio company from the syndicating partner (Corfitzen 2013). Furthermore, since international investors typically are very late stage (Roested 2013; Wylonis 2013) we cannot exclude the possibility that our sample was too focused on the earlier stages. If we had a broader sample, we might have seen that portfolio companies would judge their funds internationalization valueadded activities more positively in later stages. This thus lends to the possibility of an alternative hypothesis that stage is positively associated with internationalization, which could be tested on a different sample that includes more later-stage funds. Additionally, it would lend support to the Danish venture capital market being very nationally oriented and thus suffering from selfimposed growth constraints (Wylonis 2013). Our results regarding the contribution of exit orientation value-added activities signify no significant relation to stage. These results are somewhat surprising since we would argue that exit orientation value-added activities should be more relevant to the portfolio companies when their likelihood of exit increases at later stages. We therefore do not see results congruent with the general trends observed in Giot and Schwienbachers (2007) study. However, with reference to Isaksson (2006), we argue that an explanation could be that an intended exit strategy can be strived towards from the very beginning of the investment. This in turn could explain the lack of a significant effect of portfolio company stage. This notion is supported by Roested (2013) and Wylonis (2013). There are several other potential reasons for this surprising result: the interpretation of exit might have been misunderstood as Exit from the fund, i.e. a follow up investment from an external fund, as well as the likelihood of exit and the perceived ability of the fund to generate an exit might not necessarily have been interpreted the same way. Finally, the construction of the exit factor might have been flawed: since exit was the last factor, the questions included were the remnants after constructing the other factors. For example, the question obtaining non-equity financing loaded on factor 8 with a loading value of 0.69, but was included in the factor for finance as it cross-loaded on both factors. Furthermore, two of the questions included in the factor for exit, locating acquirers for a trade sale and preparing other exit routes, only had factor loadings of 0.14 and 0.11 for exit, respectively, but had scores for factor 7 of 0.67 and 0.69, respectively. However, since factor 7 also contained questions from strategy with high factor loadings, the inclusion of the questions from exit on factor 7 would impede the interpretation of

Page 72 of 108

the factor termed strategy. This exemplifies an inherent issue of factor analysis: the trade -off between constructing factors based on the questions which have the highest factor loadings and retaining the ability to interpret that factor. We find surprisingly strong evidence for the contribution of strategic value-added activities being dependent on stage, with later stages valuing the contribution more than earlier stages. Contrary to what we hypothesized, we conclude at the 95% confidence that strategy is perceived differently across stages, being perceived as more important in later stages. Given the relatively abundant literature on the importance of strategic value-added and how it depends on stage, we find this result somewhat surprising. Sapienza, Manigart, and Vermeir (1996) and Macmillan, Kulow, and Khoylian (1988) both find no significant relationship with stage. We argue that a possible reason for this difference is that these two studies are based on the VC funds perspective. VC funds might perceive their own contribution more important than they actually are, whereas the entrepreneur might consider his own abilities to produce a strategic plan in the earlier and thus simpler stages as sufficient, thereby reducing the perceived importance of the VC funds contribution. The converse argument, that later stages have strategic operations which are more complex and thus might not be within the competencies of the entrepreneur, also implies that entrepreneurs in these stages will value the contribution of a fund to a greater extent. Moreover, we find that our sample ranked strategic value-added lower, with a mean importance ranking strategy 4th. While this ranking is lower than previous studies (Sapienza, Manigart, and Vermeir, 1996; Macmillan, Kulow, and Khoylian, 1988), our sample exhibited little difference between the 2nd-4th ranks. We therefore argue that our results in this regard exhibit partial congruence with previous studies. Lastly, our results indicate that VC funds contribution of credibility value-added activities grows in importance the later the stage. With moderate support, this relationship thus differs from the hypothesized lack of effect of stage on credibility. We find it somewhat surprising that the certification effect from credibility of the VC fund changes the later the stage. It seems to indicate that later stage funds contribute more through credibility value-added. A possible cause is that reaching a later stage fund given that not all ventures reaches it signals exclusivity, thereby giving an increasing importance of credibility. Another explanation is that later stage portfolio companies need access to more stakeholders than earlier stage portfolio companies, using the fund as a signal. By that logic, portfolio companies can also use the VC funds signal value to increase their credibility with respect to

Page 73 of 108

already existing relationships. Interestingly, we find that credibility ranks 1st in perceived average importance. We therefore interpret that credibility will be highly important for the later stages, implying therefore that VC funds in the later stages would significantly benefit from specializing in signaling credibility with respect to their portfolio companies. We argue therefore that reputation but also the ability to convey this reputation to a broader environment can be a competitive advantage for VC funds. It can thereby be a contributor to sourcing better deals for the VC fund (Corfitzen 2013). We find the results interesting to the degree that both strategic, credibility, and marketing valueadded are significantly and positively related to stage. Thus, our results lend way to the proposition that there in effect is a demand from portfolio companies for stage specialization in the VC funds. We furthermore argue that credibility, marketing and strategic value-added activities as a group, are generally high-level. We posit that this indicates that later stage companies demand higher-level contributions more that earlier stage companies. While our results do not specify distinctively how the specialization is to be conducted, it nonetheless shows an overall direction VC funds can take.

5.2 HYPOTHESIS 9: VENTURE CAPITAL STAGE SPECIALIZATION

Our investigation of whether VC funds actually specialize accordingly yields results supporting our hypothesis. Because of the results and our analysis indicating that some funds do in fact remove the explanatory power of stage with respect to strategic, credibility, and marketing valueadded, this implies that stage specialization occurs. We therefore find that our results are in line with the conclusions from previous studies as to the presence of stage specialization (Gejadze, Giot, and Schwienbacher, 2012; Bartkus and Hassan, 2009; Knill, 2009). As Schwienbacher (2012) posits, the entrepreneur should choose the stage specialized fund over the generalist fund because specialist knowledge outweighs the chance of early cessation. An interpretation of results is therefore that the VC funds in a sense have listened to the entrepreneur, or in other words that early cessation is less of an issue to VC funds than stage specialized knowledge. Our results show a drop in the explanatory power of stage differs between strategic, marketing, and credibility value-added activities. We argue that the degree to which funds render company stage an insignificant predictor for each regression implies that stage specialization is apparent with respect to some value-added activities. Specifically, the p-value of the company stage on

Page 74 of 108

strategic value-added increases from 0.03 to 0.21 ( 0.18), marketing value-added from 0.08 to 0.36 ( 0.28) and credibility value-added from 0.09 to 0.13 ( 0.04) from regression iv to regression v respectively. Thereby, the effect of stage is explained away by the fund dummies to a much higher degree with marketing and strategic value-added. This thus lends to our interpretation that there is a higher stage specialization from VC funds within these two valueadded activities, compared to credibility value-added where the confidence level drops 4 percentage points. This does however not mean that we can explain whether each funds specialization on stage within a specific value-added activity is significant. In other words, our results going from regression iv to regression v cannot explain how the lost effect of stage on the three value-added activities is dispersed among our different fund dummies. Investigating this relationship warrants an alteration of our model, which will be explained in the limitations section. Nonetheless, stage specialization seems to be evident. While this is an interesting result, we cannot conclude as to the underlying reason for stage specialization based on our results. We do however see indications that funds are specialized for other reasons than the accumulation of specialized knowledge as proposed by Yang, Narayanan, and Zahra (2009). Rather, they seem to specialize in stages for financial and competitive reasons (Corfitzen 2013; F. Knudsen 2013; Wylonis 2013), but seem to acknowledge the side-effect of specialized accumulation of knowledge (J. Knudsen 2013). Thereby, it is stage specialization for the sake of providing stronger value-added activities can be second-order effect, but might not be a strategy in itself. We furthermore find strong support in our results that interaction is a strong determinant of the perceived contribution of value-added activities. Specifically, we find that interaction is a significant predictor for all but one value-added activity in regression v. Our results thus lend support to the notion that interaction is an important mechanism through which knowledge is transferred to the portfolio companies, in line with the findings of Sapienza (1992). We argue that these results under an assumption limited attention (Gifford 1997), specialization seems meaningful. As higher interaction is a significant predictor in almost all value-added activities, this implicitly limits the portfolio size of the VC fund, given a constant human capital base in the fund. However, since our thesis does not include normative measures, we cannot say whether the funds that portfolio companies that higher interaction, actually were strong or poor performers. Previous studies indicate that this might be the case (Bottazzi, Da Rin, and Hellmann 2008; Frederiksen, Olofsson, and Wahlbin 1997), which we also got an impression of in our interviews (Roested 2013).

Page 75 of 108

6. CONCLUSION:
Through our thesis we have investigated the following research question: How do Venture Capital funds value-added activities to their portfolio companies differ between stages, and do the funds specialize accordingly?

This thesis has shown that the perceived importance of value-added activities does in fact differ across portfolio company stages of development. Our multiple regression analysis revealed that stage is a significant predictor of marketing, strategic, and credibility value-added activities. Meanwhile, we found no significant relationship between stage and financial, internationalization, exit orientation, professionalization, and technological value-add. We thus find substantiation for value-added activities, as perceived from the entrepreneur, differing between stages. The result of this research supports the idea that stage specialization in these areas is meaningful. These findings furthermore suggest that the contribution of higher-level value-added activities is more valued by entrepreneurs in the later stages of the portfolio companies development. Our second major finding is that funds in our sample do in fact specialize by stage with respect to certain value-added activities. Through our multiple regression analysis, the explanatory power of stage was rendered insignificant when we introduced fund dummies. Taken together, these findings suggest that stage specialization is a valid strategy to increase the perceived contribution of VC funds value-added activities with portfolio companies in the areas of strategy, marketing and credibility. The results of this research therefore support the idea that VC funds do in fact specialize accordingly.

Page 76 of 108

6.1 IMPLICATIONS FOR PRACTITIONERS AND RESEARCHERS

F OR V ENTURE C APITALISTS : Our results are relevant to Venture Capitalists to the extent that

they engage in value-added activities. That stage specialization makes sense with regards to marketing, credibility, and strategic value-added activities, means that later stage firms will benefit from taking our results into consideration. Conversely our findings indicate that stage specialization in all value-added activities is unwarranted. Furthermore, the low importance of, and lack of significant with relation to stage of technical value-added activities posits that hiring managers with high degrees of technical expertise might not be ideal. We furthermore found evidence that interaction is a significant predictor of the perceived contribution of value-added activities. This has implications for VC funds that wish to maximize the impact of their value-added perspectives. This suggests that hiring a broader staff base increases the perceived value-added. It could additionally mean that a pyramidal structure is beneficial, at least from the perspective of the portfolio companies.
F OR E NTREPRENEURS : The key implication of our results to entrepreneurs is that there is a

difference between funds. Money is not fungible in this regard, and an entrepreneur would benefit from finding a fund that suits their needs for value-added activities. We furthermore posit that portfolio companies in later stage funds value credibility more than earlier stage portfolio companies. This suggests that paying for affiliation to a highly reputable fund might be worthwhile.
F OR R ESEARCHERS : Our methodological approach has implications for researchers. In

conducting our factor analysis, we found a high degree of congruence between the optimal factors, the factors we actually ended up choosing, and the original eight strategic groups of the survey we applied. This suggests that distinguishing between value-added activities in the eight groups we applied is meaningful. This therefore supports a higher consistency in the typology and applied categorization of value-added activities. Additionally, we argue that we through our sample and analysis have mapped a substantial part of the Danish venture capital industry. Specifically our sample closely resembles the population of investments within ICT and Life Science, categories that represent the vast majority of the market. Our data is to that extent useful for researchers, and could technically be applied to other studies.

Page 77 of 108

6.2 LIMITATIONS AND POTENTIAL WEAKNESSES OF OUR RESEARCH

A number of important limitations need to be considered. First, given that we have such a high proportion of the VC funds operating in Denmark in our sample (F. Knudsen 2013), we believe our results, at the conservative level, allow us to infer that stage specialization occurs in Denmark, with respect to ICT and Life Science companies that have received venture capital backing. However, we cannot validly infer beyond these limits. In itself, this poses a limitation, but we argue that our research design is replicable across borders. Inherent in our research design is that respondents answer to what extent they find the contribution of their VC funds value-added activities important. Our results are therefore limited to the extent that respondents both a) potentially are not aware of alternatives, in the case of a lack of previous funding rounds with other VC funds, and b) that we did not request an opinion on what value-added activities the portfolio companies desired more of. Therefore, our research design lacks the ability to gauge the portfolio companies opinions outside of the particular fund they are in. Our model was furthermore impeded by the lack of data quality with respect to contracts. We intended to include the option of controlling for contractual elements that might influence the ability of VC funds to exercise power over their portfolio companies. It is cause of conflict between the VC funds and their portfolio companies which might influence the relationship between the two. Thereby it might reduce the contribution of value-added activities through a lower permeability of knowledge, as posited by De Clercq and Sapienza (2006). Given that we did not attempt to measure the influence of this aspect of knowledge transfer mechanisms, our model is still consistent with what we set out to do. However, we argue that it warrants future research. We furthermore find our research design and analysis to be limited with respect to firm characteristics. We highlight that we have attempted to incorporate proxies as used in the literature, but at the same time argue that there are many more predictors of the needs of portfolio companies that could be examined. For example, as Wijbenga et al. (2003) propose, introducing the level of strategic fit whereby the value-added activities are placed in framework where agency fit, entrepreneurial fit, positioning fit, and resource fit, would allow for a more holistic multi-theoretical framework on which to evaluate the impact of value-added activities. That our mode is imperfect with respect to accurately predicting the perceived contribution of value-added activities is signified by our R2 in regression iv only explaining between 10-35% of

Page 78 of 108

our observed variance. The explanatory power of regression v is slightly higher, ranging between 26 and 52% of the observed variance, but there still remains a considerable fraction of our observed variance that we simply cannot explain. However, we argue that given our sample size the inclusion of additional variables would be troublesome. If we were to construct a more complex model, including a broader array of variables, we would be severely inhibited by a lack of degrees of freedom. Using the data available to us, we constructed a model taking into account the trade-off between the marginal explanatory effect on R2 and the loss of degrees of freedom. Omitting the variables EducationPhD, SyndicationPartners, and AgeFund was warranted through our backwards selection process as they had a detrimental effect on our adjusted R2 when viewed over all eight regressions. We therefore consider our model to provide the best level of simplicity and interpretability based on the sample. Another aspect which our sample size inhibited was the inclusion of interaction terms with respect to stage, the value-added activities and each fund dummy. We are with our model unable to accurately describe to what extent individual funds are stage specialized with respect to the particular value-added activities. As we mentioned, our results going from regression iv to regression v cannot explain how the lost effect of stage on the three value-added activities is dispersed among our different fund dummies. In this regard, our model would with the inclusion of interaction terms be able to explain this, but this would require a larger sample to gain degrees of freedom. The interpretation of individual fund specialization with respect to stage would also become much stronger if our model with the addition of interaction terms was applied to a much larger sample. This is due to our model inherently only explaining specialization from a relative point of view, i.e. that one VC fund is a relatively stronger predictor for a value-added activity, than others. The relativistic nature, we argue, could in a large sample approximate actual specialization more accurately.

Page 79 of 108

7. RECOMMENDATIONS FOR FUTURE RESEARCH

This research has resulted in several patterns and questions that warrant further investigation. First of all, we find indications of some funds exhibiting more stage specialization in value-added activities than others. Specifically, four VC funds seemed stage specialized in that when we ran regression v, the explanatory power of stage as a predictor for credibility, marketing and strategic value-add, we found the funds significant predictors. We cannot credibly state whether or not these funds are effectively stage specialized due to our lack of ability to include interaction terms in our regression model. With our sample size, such interaction terms will significantly restrict our degrees of freedom. This warrants a further investigation of stage specialization and valueadded where interaction terms are included with a sample that can carry such complexity. Furthermore, if richer data on fund characteristics was included in that model, the investigation of stage specialization would not only be limited to predicting whether stage specialization is present, but also how and where. Moreover, we find evidence that several strategic groups exist in our sample: Funds A, F, and G all have a significantly positive effect on the level of perceived contributions of the VC funds in strategy and marketing value add, suggesting that these funds are relatively specialized within these areas compared to other sampled funds. Furthermore, Funds C and D have a significantly negative impact on perceived contributions in finance and technology, suggesting that these funds are at a relative disadvantage compared to other funds. The coefficients for Fund A suggest an explanation for such disadvantages: funds might sacrifice certain areas of value-added activities in an attempt to specialize in certain key areas. Fund A seems to be relatively disadvantaged in value-added activities within technology, finance, legitimacy, and exit, but compensates this with a relative specialization in strategy and market position. These indications of strategic groups arguably warrant future research on this topic: to what extent do funds form strategic groups, and what are the determinants for these? An analysis of this sort could be done using a cluster analysis on a large sample of Venture Capital-backed companies and would potentially yield interesting results. With our models explaining between 15-50% of the observed variance, the underlying understanding of Venture Capital-backed firms needs for value-added activities is in our view not sufficient. Such a cluster analysis could potentially reveal connections that would warrant different types of Venture Capital fund specialization than the current three dimensions we observe today.

Page 80 of 108

8. BIBLIOGRAPHY
Amit, Raphael, James Brander, and Christoph Zott. 1998. Why Do Venture Capital Firms Exist? Theory and Canadian Evidence. Journal of Business Venturing 13 (6) (November): 441466. doi:10.1016/S08839026(97)00061-X. http://linkinghub.elsevier.com/retrieve/pii/S088390269700061X. Arthurs, Jonathan D., and Lowell W. Busenitz. 2006. Dynamic Capabilities and Venture Performance: The Effect s of Venture Capitalists. Journal of Business Venturing 21 (2) (March): 195215. doi:10.1016/j.jbusvent.2005.04.004. http://linkinghub.elsevier.com/retrieve/pii/S088390260500039X. Bartkus, James R., and M. Kabir Hassan. 2009. Specialization Versus Diversification in Venture Capital Investing. Journal of Financial Regulation and Compliance 17 (2): 134145. doi:10.1108/13581980910952577. http://www.emeraldinsight.com/10.1108/13581980910952577. Baum, Joel A.C., and Brian S. Silverman. 2004. Picking Winners or Building Them - Alliances, Patents, and Human Capital as Selection Criteria in Venture Fnancing and Performance of Biotechnology Startups. Journal of Business Venturing 19: 411436. Barnett, Vic. 1991. "Sample Survey: Principles and Methods", Second Edition. Edward Arnold, London: United Kingdom Behrens, J., Patzelt, H., Schweizer, L., & Brger, R. 2012. "Specific managerial human capital, firm age, and Venture Capital financing of biopharmaceutical ventures: A contingency approach". The Journal of High Technology Management Research, 23(2), 112121. doi:10.1016/j.hitech.2012.06.003 Berglund, Henrik, Tomas Hellstrm, and Sren Sjlander. 2007. Entrepreneurial Learning and the Role of Venture Capitalists. Venture Capital 9 (3) (July): 165181. doi:10.1080/13691060701324445. http://www.tandfonline.com/doi/abs/10.1080/13691060701324445. Bertoni, Fabio, Annalisa Croce, and Anita Quas. 2010. Real Vs. Financial Value Enhancement by Venture Capital. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1714585&download=yes. Bertoni, Fabio, and Tereza Tykvov. 2012. Which Form of Venture Capital Is Most Supportive of Innovation? De Bettignies, Jean-Etienne, and James a. Brander. 2007. Financing Entrepreneurship: Bank Finance Versus Venture Capital. Journal of Business Venturing 22 (6) (November): 808832. doi:10.1016/j.jbusvent.2006.07.005. http://linkinghub.elsevier.com/retrieve/pii/S0883902606000863. Bottazzi, Laura, Marco Da Rin, and Thomas Hellmann. 2008. Who Are the Active Investors? Evidence from Venture Capital. Journal of Financial Economics 89 (3) (September): 488512. doi:10.1016/j.jfineco.2007.09.003. http://linkinghub.elsevier.com/retrieve/pii/S0304405X08000974. Bygrave, W.D., and J.A. Timmons. 1992. Venture Capital at the Crossroads. Boston, MA: Harvard Business School Press. Caselli, Stefano, Stefano Gatti, and Francesco Perrini. 2009. Are Venture Capitalists a Catalyst for Innovation? European Financial Management 15 (1) (January): 92111. doi:10.1111/j.1468-036X.2008.00445.x. http://doi.wiley.com/10.1111/j.1468-036X.2008.00445.x. CAT Science. 2013. CAT Innovation. Retrieved May 20, 2013, from http://www.catscience.dk/ Clarke, Kevin A. 2005. The Phantom Menace: Omitted Variable Bias in Econometric Research Conflict Management and Peace Science, 22: 341-352

Page 81 of 108

Clercq, Dirk De, and Dimo Dimov. 2006. Venture Capital Investment Strategy and Portfolio Failure Rate: A Longitudinal Study. Entrepreneurship: Theory & Practice 30 (2): 207224. Corfitzen, Magnus. Interview by authors. Audio recording. Hellerup, Denmark, April 5. 2013. Transcript available in Appendix I. De Clercq, Dirk, Vance H. Fried, Oskari Lehtonen, and Harry J. Sapienza. 2006. An Entrepreneurs Guid e to the Venture Capital Galaxy. Academy of Management Perspectives 20 (3) (August 1): 90112. doi:10.5465/AMP.2006.21903483. http://amp.aom.org/cgi/doi/10.5465/AMP.2006.21903483. De Clercq, Dirk, Philip K Goulet, Mikko Kumpulainen, and Manu Mkel. 2001. Portfolio Investment Strategies in the Finnish Venture Capital Industry: a Longitudinal Study. Venture Capital 3 (1): 4162. doi:10.1080/13691060010017348. De Clercq, Dirk, and Sophie Manigart. 2007. The Venture Capital Post-investment Phase: Opening the Black Box of Involvement. In Handbook of Research on Venture Capital, ed. Hans Landstrm, 454. 1st ed. Northhampton, Massachusetts: Edward Elgar Publishing Ltd. De Clercq, Dirk, and Harry J. Sapienza. 2001. The Creation of Relational Rents in Venture Capitalist-entrepreneur Dyads. Venture Capital 3 (2) (April): 107127. doi:10.1080/13691060110045661. http://www.tandfonline.com/doi/abs/10.1080/13691060110045661. De Clercq, Dirk, and Harry J. Sapienza. 2006. Effects of Relational Capital and Commitment on Venture Capitalists Perception of Portfolio Company Performance. Journal of Business Venturing 21 (3) (May): 326347. doi:10.1016/j.jbusvent.2005.04.007. http://linkinghub.elsevier.com/retrieve/pii/S0883902605000455. Cramer, Duncan. 2003. "Advanced Quantitative Data Analysis" Open University Press, Glasglow: United Kingdom Creandum. 2013. Creandum. Retrieved May 20, 2013, from http://www.creandum.com/ Cressy, Robert. 2012. Funding Gaps. In The Oxford Handbook of Entrepreneurial Finance, ed. Douglas Cumming, 1 73. Oxford, Uk: Oxford University Press. doi:10.1093/oxfordhb/9780195391244.001.0001. http://www.oxfordhandbooks.com/view/10.1093/oxfordhb/9780195391244.001.0001/oxfordhb9780195391244. Cumming, Douglas, Grant Fleming, and Armin Schwienbacher. 2009. Style Drift in Private Equity. Journal of Business Finance & Accounting 36 (5-6) (June): 645678. doi:10.1111/j.1468-5957.2009.02137.x. http://doi.wiley.com/10.1111/j.1468-5957.2009.02137.x. Danish Venture Capital and Private Equity Association. 2013. Danish Venture Capital and Private Equity Association. Retrieved May 20, 2013, from www.dvca.dk DeTienne, Dawn R., and Melissa S. Cardon. 2010. Impact of Founder Experience on Exit Intentions. Small Business Economics 38 (4) (April 22): 351374. doi:10.1007/s11187-010-9284-5. http://link.springer.com/10.1007/s11187-010-9284-5. Dolvin, Steven. 2005. Venture Capitalist Certification of IPOs. Venture Capital 7 (2) (April): 131148. doi:10.1080/1369106042000335601. http://www.tandfonline.com/doi/abs/10.1080/1369106042000335601. DTU Symbion Innovation. 2013. DTU Symbion Innovation. Retrieved May 20, 2013, from http://dtusymbioninnovation.dk/

Page 82 of 108

Ehrlich, Sanford B, Alex F. De Noble, Richard R Weaver, and Tracy Moore. 1994. After the Cash Arrives: A Comparative Study of Venture Capital and Private Investor Involvement in Entrepreneurial Firms. Journal of Business Venturing 9: 6782. Elango, B., Vance H. Fried, Robert D. Hisrich, and Amy Polonchek. 1995. How Venture Capital Firms Differ. Journal of Business Venturing 10 (2) (March): 157179. doi:10.1016/0883-9026(94)00019-Q. http://linkinghub.elsevier.com/retrieve/pii/088390269400019Q. Engel, Dirk, and Max Keilbach. 2007. Firm-level Implications of Early Stage Venture Capital Investment An Empirical Investigation. Journal of Empirical Finance 14 (2) (March): 150167. doi:10.1016/j.jempfin.2006.03.004. http://linkinghub.elsevier.com/retrieve/pii/S0927539806000442. English First Ltd. 2012. EF English Proficiency Index 2012 (p. 43). Retrieved from http://www.efdanmark.dk/__/~/media/efcom/epi/2012/full_reports/ef-epi-2012-report-master-lr-2 Ernst & Young. 2010. Evalureing af Vkstfonden, Copenhagen, Denmark.124p Fernhaber, Stephanie A., and Patricia P. Mcdougall-Covin. 2009. Venture Capitalists as Catalysts to New Venture Internationalization: The Impact of Their Knowledge and Reputation Resources. Entrepreneurship: Theory & Practice: 277296. Frederiksen, ystein, Christer Olofsson, and Clas Wahlbin. 1997. Are Venture Capitalists Firefighters? A Study of the Influence and Impact of Venture Capital Firms. Technovation 17 (9): 503511. Fried, Vance H., Garry D. Bruton, and Robert D. Hisrich. 1998. Strategy and the Board of Directors in Venture Capital-backed Firms. Journal of Business Venturing 13 (6) (November): 493503. doi:10.1016/S08839026(97)00062-1. http://linkinghub.elsevier.com/retrieve/pii/S0883902697000621. Fulghieri, P., and M. Sevilir. 2009. Size and Focus of a Venture Capitalists Portfolio. Review of Financial Studies 22 (11) (March 27): 46434680. doi:10.1093/rfs/hhp012. http://rfs.oxfordjournals.org/cgi/doi/10.1093/rfs/hhp012. Gabrielsson, Jonas, and Morten Huse. 2002. The Venture Capitalist and the Board of Directors in SMEs: Roles and Processes. Venture Capital 4 (2) (April): 125146. doi:10.1080/13691060110094397. http://www.tandfonline.com/doi/abs/10.1080/13691060110094397. Gejadze, Maia, Pierre Giot, and Armin Schwienbacher. 2012. Private Equity Fundraising , Fund Performance and Firm Specialization: 141. Gifford, Sharon. 1997. Limited Attention and the Role of the Venture Capitalist. Journal of Business Venturing 12: 459482. Giot, Pierre, and Armin Schwienbacher. 2007. IPOs, Trade Sales and Liquidations: Modelling Venture Capital Exits Using Survival Analysis. Journal of Banking & Finance 31 (3) (March): 679702. doi:10.1016/j.jbankfin.2006.06.010. http://linkinghub.elsevier.com/retrieve/pii/S037842660600152X. Gomez-Mejia, Luis R., Theresa M. Welbourne, and David B. Balkin. 1990. Influence of Venture Capitalists on High Tech Management. Journal of High Tech Management Research 1 (1): 103118. Gompers, Paul. 1995. Optimal Investment, Monitoring, and the Staging of Venture Capital. The Journal of Finance 50 (5) (December): 1461. doi:10.2307/2329323. http://www.jstor.org/stable/2329323?origin=crossref. Gompers, P., Kovner, A., Lerner, J., & Scharfstein, D. 2006. Skill vs. Luck in Entrepreneurship and Venture Capital: Evidence from Serial Entrepreneurs. Cambridge, MA.

Page 83 of 108

Gompers, Paul, Anna Kovner, and Josh Lerner. 2009. Specialization and Success: Evidence from Venture Capital. Journal of Economics & Management Strategy 18 (3): 817844. Gorman, Michael, and William A. Sahlman. 1989. What Do Venture Capitalists Do? Journal of Business Venturing 4 (4): 231248. doi:10.1038/4371087a. Hanks, Steven H., Collin J. Watson, Erik Jansen, and Gaylen N. Chandler. 1994. Tightening the Life -Cycle Construct: A Taxonomic Study of Growth Stage Configurations in Organizations. Entrepreneurship: Theory & Practice 18 (2): 530. Hirsch, J., & Walz, U. 2006. Why Do Contracts Differ between VC Types? Market segmentation versus Corporate Governance Varieties. Hellmann, Thomas, and Manju Puri. 2000. The Interaction Between Product Market and Financing Strategy: The Role of Venture Capital. The Review of Financial Studies 13 (4): 959984. Hellmann, Thomas, and Manju Puri. 2002. Venture Capital and the Professionalization of Startup Firms: LVII (1). Hochberg, Yael V. 2011. Venture Capital and Corporate Governance in the Newly Public Firm. Review of Finance 16 (2) (December 22): 429480. doi:10.1093/rof/rfr035. http://rof.oxfordjournals.org/cgi/doi/10.1093/rof/rfr035. Hochberg, Yael V., Alexander Ljungqvist, and Yang Lu. 2007. Whom You Know Matters : Venture Capital Networks and Investment Performance. The Journal of Finance LXII (1): 251301. Hsu, David H. 2004. What Do Entrepreneurs Pay for Venture Capital Affiliation? The Journal of Finance LIX (4): 18051844. Innovation Midtvest. 2013. Innovation Midtvest. Retrieved May 20, 2013, from http://www.innovationmidtvest.dk/ Isaksson, Anders. 2006. Exit Strategy and the Intensity of Exit-directed Activities Among Venture Capital-backed Entrepreneurs in Sweden. In Venture Capital in Europe, ed. Greg N. Kolli Gregoriou and Roman Maher Kraeussi, 143156. 1st ed. Jordan Hill, GBR: Butterworth-Heinemann. Kaplan, Steven N, and Per Strmberg. 2001. Venture Capitalists as Principals: Contracting, Screening, and Monitoring. Cambridge, MA. Kline, Paul 1994 An Easy Guide to Factor Analysis Routledge, London: United Kingdom Knill, April. 2009. Should Venture Capitalists Put All Their Eggs in One Basket? Diversification Versus Pure-Play Strategies in Venture Capital. Financial Management 38 (3) (September): 441486. doi:10.1111/j.1755053X.2009.01044.x. http://doi.wiley.com/10.1111/j.1755-053X.2009.01044.x. Knudsen, Frank. Interview by authors. Audio recording. DTU, Lyngby, Denmark. April 10, 2013. Transcript is available in Appendix I. Knudsen, Jesper. Interview by authors. Audio recording. Hellerup, Denmark. April 11. 2013. Transcript is available in Appendix I. Kortum, Samuel, and Josh Lerner. 1998. Does Venture Capital Spur Innovation? Cambridge, Mass.

Page 84 of 108

Kopelman, Josh. Interview with Sarah Lacy. Fireside Chat with First Round Capitals Josh Kopelman. Pandodaily, February 8, 2013. Accessible at http://video.pandodaily.com/E52/fireside-chat-with-first-round-capitalsjosh-kopelman/ Large, David, and Steven Muegge. 2008. Venture Capitalists Non-financial Value-added: An Evaluation of the Evidence and Implications for Research. Venture Capital 10 (1) (January): 2153. doi:10.1080/13691060701605488. http://www.tandfonline.com/doi/abs/10.1080/13691060701605488. Lee, Raymond. 1993. "Doing Research on Sensitive Topics" SAGE Publications, London: United Kingdom Lerner, Josh. 1996. The Government as Venture Capitalist, The Long-Run Effects of the SBIR Program. National Burea of Economic Researcg. Cambridge, Mass. Lindsey, Laura. 2008. Blurring Firm Boundaries: The Role of Venture Capital in Strategic Alliances. The Journal of Finance LXIII (3): 11371169. Lockett, Andy, Mike Wright, Andrew Burrows, Louise Scholes, and Dave Paton. 2008. The Export Intensity of Venture Capital Backed Companies. Small Business Economics 31 (1) (April 2): 3958. doi:10.1007/s11187-0089109-y. http://www.springerlink.com/index/10.1007/s11187-008-9109-y. Luukkonen, Terttu, Matthias Deschryvere, and Fabio Bertoni. 2013. The Value Added by Government Venture Capital Funds Compared with Independent Venture Capital Funds. Technovation 33 (4-5) (April): 154162. doi:10.1016/j.technovation.2012.11.007. http://linkinghub.elsevier.com/retrieve/pii/S0166497212001411. Macmillan, Ian C., David M. Kulow, and Roubina Khoylian. 1988. Venture Capitalists Involvement in Their Investments: Extent and Performance. Journal of Business Venturing 4: 2747. Maula, Markku, Erkko Autio, and Gordon Murray. 2005. Corporate Venture Capitalists and Independent Venture Capitalists: What Do They Know, Who Do They Know and Should Entrepreneurs Care? Venture Capital 7 (1) (January): 321. doi:10.1080/1369106042000316332. http://www.tandfonline.com/doi/abs/10.1080/1369106042000316332. Marmer, M., Herrmann, B., Berman, R., & Easley, C. (2011). Startup Genome Report. A new framework for , 1(March). Retrieved from http://scholar.google.com/scholar?hl=en&btnG=Search&q=intitle:Startup+Genome+Report#0 Megginson, William L, and Kathleen A Weiss. 1991. Venture Capitalist Certification in Initial Public Offerings. Journal of Finance XLVI (3): 879904. Murray, Gordon. 1999. Early-stage Venture Capital Funds , Scale Economies. Venture Capital 1 (4): 351384. Motzfeldt, Christian. "Nye finansieringsmuligheder for vkstvirksomheder" ("New Financing Opportunities for Growth Companies". Conference Talk. Vkstfonden Annual Meeting from Vkstfonden. Copenhagen, Denmark, May 2, 2013. Transcript is available at Vkstfonden, Hellerup. Nahata, R. 2008. Venture Capital Reputation and Investment Performance. Journal of Financial Economics 90 (2) (November): 127151. doi:10.1016/j.jfineco.2007.11.008. http://linkinghub.elsevier.com/retrieve/pii/S0304405X08001414. Northcap Partners. 2013. Northcap Partners. Retrieved May 20, 2013, from http://www.northcap.vc/ Northzone Ventures. 2013. Northzone Ventures. Retrieved May 20, 2013, from http://www.northzone.com/

Page 85 of 108

Norton, Edgar, and Bernard H. Tenenbaum. 1993. Specialization Versus Diversification as a Venture Capital Investment Stratey. Journal of Business Venturing 8: 431442. NOVI. 2013. NOVI Innovation. Retrieved May 20, 2013, from http://novi.dk/da/innovation/noviinnovation.html Patzelt, Holger, Dodo Zu Knyphausen-Aufse, and Ilona Arnoldt. 2006. How Do Venture Capitalists Spread Risk by Diversification Within Specialised Life Science Portfolios? International Journal of Technology Management 34 (1/2): 105125. Patzelt, Holger, Dodo Zu Knyphausen-Aufse, and Heiko T. Fischer. 2009. Upper Echelons and Portfolio Strategies of Venture Capital Firms. Journal of Business Venturing 24: 558572. Pett, Marjorie, Nancy Lackey, and John Sullivan. 2003. "Making Sense of Factor Analysis: The Use of Factor Analysis for Instrument Development in Health Care Research", SAGE Petty, J. William, William D. Bygrave, and Joel M. Shulman. 1994. Harvesting the Entrepreneurial Venture: A Time for Creating Value. Journal of Applied Corporate Finance 7 (1): 4759. Phelps, Robert, Richard Adams, and John Bessant. 2007. Life Cycles of Growing Organizations: A Review with Implications for Knowledge and Learning. International Journal of Management Reviews 9 (1) (March): 130. doi:10.1111/j.1468-2370.2007.00200.x. http://doi.wiley.com/10.1111/j.1468-2370.2007.00200.x. Popov, Alexander, and Peter Roosenboom. 2009. Does Private Equity Spur Innovation? Evidence from Europe. Frankfurt Am Main, Germany. Pratch, Leslie. 2005. Value-Added Investing: A Framework for Early Stage Venture Capital Firms. The Journal of Private Equity: 1329. Qian, Gongming, and Lee Li. 2003. Profitability of Small- and Medium-sized Enterprises in High-tech Industries: The Case of the Biotechnology Industry. Strategic Management Journal 24 (9) (September): 881887. doi:10.1002/smj.344. http://doi.wiley.com/10.1002/smj.344. Da Rin, Marco, Giovanna Nicodano, and Alessandro Sembenelli. 2006. Public Policy and the Creation of Active Venture Capital Markets. Journal of Public Economics 90 (8-9) (September): 16991723. doi:10.1016/j.jpubeco.2005.09.013. http://linkinghub.elsevier.com/retrieve/pii/S0047272705001702. Roested, Jesper. Interview by authors. Audio recording. Hellerup, Denmark, April 5. 2013. Transcript available in Appendix I. Ruhnka, John C., and John E. Young. 1987. A Venture Capital Model of the Development Process for New Ventures. Journal of Business Venturing 2 (2) (March): 167184. doi:10.1016/0883-9026(87)90006-1. http://linkinghub.elsevier.com/retrieve/pii/0883902687900061. Sahlman, William a. 1990. The Structure and Governance of Venture -capital Organizations. Journal of Financial Economics 27 (2) (October): 473521. doi:10.1016/0304-405X(90)90065-8. http://linkinghub.elsevier.com/retrieve/pii/0304405X90900658. Sapienza, Harry J. 1992. When Do Venture Capitalists Add Value? Journal of Business Venturing 7: 927. Sapienza, Harry J., Sophie Manigart, and Wim Vermeir. 1996. Venture Capitalist Governance and Value Added in Four Countries. Journal of Business Venturing 11: 439469.

Page 86 of 108

Schouten, B., Calinescu, M., and Luiten, A. 2011. Optimizing quality of response through adaptive survey designs. Statistics Netherlands. Retrieved from http://www.cbs.nl/NR/rdonlyres/2D62BF4A-6783-4AC4-8E4512EF20C6675C/0/2011x1018.pdf Schwienbacher, Armin. 2012. The Entrepreneurs Investor Choice: The Impact o n Later-stage Firm Development. Journal of Business Venturing (December). doi:10.1016/j.jbusvent.2012.09.002. http://linkinghub.elsevier.com/retrieve/pii/S0883902612000912. SEED Capital. 2013. SEED Capital Denmark. Retrieved May 20, 2013, from http://www.seedcapital.dk/ Smith, Gordon. 1999. How Early Stage Entrepreneurs Evaluate Venture Capitalists. SSRN Electronic Journal: 20. doi:10.2139/ssrn.183981. http://www.ssrn.com/abstract=183981. Smith, Janet K., Richard L. Smith, and Richard T. Bliss. 2011. Entrepreneurial Finance - Strategy, Valuation, and Deal Structure. 1st ed. Stanford, California: Stanford Economics and Finance, Stanford University Press. Smolarski, Jan, and Can Kut. 2011. The Impact of Venture Capital Financing Method on SME Performance and Internationalization. International Entrepreneurship and Management Journal 7 (1) (November 18): 3955. doi:10.1007/s11365-009-0128-1. http://www.springerlink.com/index/10.1007/s11365-009-0128-1. Stuart, Toby E., Ha Hoang, and Ralph C. Hybels. 1999. Interorganizational Endorsements and the Performance of Entrepreneurial Ventures. Administrative Science Quarterly 44 (2) (June): 315. doi:10.2307/2666998. http://www.jstor.org/stable/2666998?origin=crossref. Sunstone Capital. 2012. Sunstone Capital. Retrieved May 20, 2013, from www.sunstone.eu Syddansk Teknologisk Innovation. 2013. Syddansk Teknologisk Innovation. Retrieved May 20, 2013, from http://www.sdti.dk/ Symbion. 2013. Accelerace. Retrieved May 20, 2013, from http://symbion.dk/forretningsudvikling/accelerace/ Stre, Alf. 2003. Entrepreneurial Perspectives on Informal Venture Capital. Venture Capital 5 (1) (January): 7194. doi:10.1080/1369106032000062731. http://www.tandfonline.com/doi/abs/10.1080/1369106032000062731. Tyebjee, Tyzoon T., and Albert V. Bruno. 1984. A Model of Venture Capitalist Investment Activity. Management Science 30 (9) (September 1): 10511066. doi:10.1287/mnsc.30.9.1051. http://mansci.journal.informs.org/cgi/doi/10.1287/mnsc.30.9.1051. Ueda, Masako. 2004. Banks Versus Venture Capital: Project Evaluation , Screening , and Expropriation LIX (2). VIA Venture Partners. 2013. VIA Venture Partners. Retrieved May 20, 2013, from http://www.viaventurepartners.com/ VICO Project. 2013. VICO - Financing entrepreneurial ventures in Europe. European Commisson Seventh Framework Programme. Retrieved May 18, 2013, from http://www.vicoproject.org/ Vkstfonden. 2013. Vkstfonden - VF Venture. Retrieved May 20, 2013, from http://www.vf.dk/For virksomheder/VF Venture Wang, Clement K., and Valerie Y.L. Sim. 2001. Exit Strategies of Venture Capital-backed Companies in Singapore. Venture Capital 3 (4) (October): 337358. doi:10.1080/13691060110060664. http://www.tandfonline.com/doi/abs/10.1080/13691060110060664. Williams, Frederick, and Peter Monge. 2001. "Reasoning with Statistics: How to Read Quantitative Research" Harcourt College Publisher, Orlando: United States of America

Page 87 of 108

Wijbenga, Frits, Theo Postma, Arjen Van Witteloostuijn, and Peter Zwart. 2003. Strategy and Performance of New Ventures: A Contingency Model of the Role and Influence of the Venture Capitalist. Venture Capital 5 (3) (July):231250. doi:10.1080/1369106032000118171. http://www.tandfonline.com/doi/abs/10.1080/1369106032000118171. Wooldridge, Jeffrey M. 2009. Introductory Econometrics: A Modern Approach South -Western Cengage Learning, Mason, OH: United States of America Wylonis, Christian. Interview by authors. Audio recording. Hellerup, Denmark, April 16. 2013. Transcript available in Appendix I. Yang, Yi, V.K. Narayanan, and Shaker Zahra. 2009. Developing the Selection and Valuation Capabilities Through Learning: The Case of Corporate Venture Capital. Journal of Business Venturing 24 (3) (May): 261273. doi:10.1016/j.jbusvent.2008.05.001. http://linkinghub.elsevier.com/retrieve/pii/S0883902608000554. Zacharakis, Andrew, Truls Erikson, and Bradley George. 2010. Conflict Between the VC and Entrepreneur: T he Entrepreneurs Perspective. Venture Capital 12 (2) (April): 109126. doi:10.1080/13691061003771663. http://www.tandfonline.com/doi/abs/10.1080/13691061003771663. Zhang, J. (2009). The advantage of experienced startup founders in Venture Capital acquisition: evidence from serial entrepreneurs. Small Business Economics, 36(2), 187208. doi:10.1007/s11187-009-9216-4 Zider, Bob. 1998. How Venture Capital Works. Harvard Business Review 76 (6): 131140. Zingales, Luigi. 2000. In Search of New Foundations. The Journal of Finance LV (4): 16231653. stjysk Innovation. 2013. stjysk Innovation. Retrieved May 20, 2013, from www.oei.dk/

Page 88 of 108

APPENDIX
I. OVERVIEW OF INTERVIEWS

Page 89 of 108

II. NUMBER OF VC INVESTMENTS BY INDUSTRY IN DENMARK

Page 90 of 108

III. SAMPLE SURVEY

Page 91 of 108

Page 92 of 108

Page 93 of 108

Page 94 of 108

Page 95 of 108

Page 96 of 108

Page 97 of 108

Page 98 of 108

Page 99 of 108

IV. INDIVIDUAL KMO SCORES

Kaiser-Meyer-Olkin Panel A
Question Score (1) Developing~n 0.7944 (2) Strategicf~s 0.759 (3) Developcap~s 0.8027 (4) ImprovingRD 0.8015 (5) BuildingIP~e 0.8406 (6) Technologi~s 0.7698 (7) Marketings~m 0.7807 (8) Salespress~e 0.8087 (9) Growthpres~e 0.6559 (10) Controllin~s 0.7604 (11) Developing~e 0.8322 (12) Alteringma~m 0.8483 (13) Findingboa~s 0.7418 (14 ) Obtainingn~g 0.772

Panel B
Question Score (15) Raisingfol~s 0.6301 (16) Attracting~s 0.794 (17) Credibil~ors 0.7997 (18) Credibi~mers 0.6993 (19) Credibi~iers 0.7862 (20) Credibili~es 0.8382 (21) Foreigndis~s 0.8298 (22) Foreignequ~y 0.7628 (23) Foreignman~t 0.892 (24) Foreignemp~s 0.8549 (25) Foreignboa~s 0.8429 (26) IPOprepara~n 0.7932 (27) Locatingin~s 0.8546 (28) Preparingo~s 0.8173

Page 100 of 108

V. OVERALL PRINCIPAL COMPONENT ANALYSIS WITH SCREEPLOT

Page 101 of 108

VI. PRINCIPAL COMPONENT ANALYSIS OF DEPENDENT VARIABLES

STRATEGY

TECHNOLOGY

MARKET POSTION

PROFESSIONALIZATION

Page 102 of 108

FINANCE

CREDIBILITY

INTERNATIONALIZATION

EXIT

Page 103 of 108

VII. PRINCIPAL COMPONENT ANALYSIS OF INTERACTION

Page 104 of 108

VIII. HETEROSKEDASTICITY TESTS

Heteroskedasticity Tests for Regression i

Panel A
Breusch-Pagan test *Chi-sq (1) chi2 Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit 0.53 0.03 1.29 0.44 0.00 1.24 0.27 0.47 Prob > chi2 0.47 0.87 0.26 0.51 0.96 0.27 0.60 0.49

Panel B
White test* *Chi-sq (1) Whites General Test Statistic 1.14 0.60 2.69 0.76 0.00 1.78 0.38 0.82 Prob > chi2 0.29 0.81 0.10 0.38 0.95 0.18 0.54 0.36

Heteroskedasticity Tests for Regression ii

Panel A
Breusch-Pagan test* *Chi-sq (1) Factor Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit chi2 0.35 0.01 1.43 0.17 0.04 3.66 0.78 0.01 Prob > chi2 0.55 0.94 0.23 0.68 0.85 0.06 0.38 0.93

Panel B
White test* *Chi-sq (49) Whites General Test Statistic 41.18 59.27 54.83 46.63 51.88 43.02 58.50 53.41 Prob > chi2 0.78 0.15 0.26 0.57 0.36 0.71 0.17 0.31

Page 105 of 108

Heteroskedasticity Tests for Regression iii

Panel A
Breusch-Pagan test* *Chi-sq (1) chi2 Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit 1.14 0.50 3.37 0.49 0.32 0.21 3.06 0.15 Prob > chi2 0.29 0.48 0.07 0.49 0.57 0.64 0.08 0.69

Panel B
White test* *Chi-sq (60) Whites General Test Statistic 61.05 65.88 66.96 58.29 55.29 49.02 63.53 58.74 Prob > chi2 0.44 0.28 0.25 0.54 0.65 0.84 0.35 0.52

Heteroskedasticity Tests for Regression iv

Panel A
Breusch-Pagan test *Chi-sq (1) chi2 Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit 1.75 0.51 3.63 0.24 0.20 0.36 4.48 0.31 Prob > chi2 0.19 0.48 0.06 0.63 0.66 0.55 0.03 0.58

Panel B
White test* *Chi-sq (31) Whites Prob > chi2 General Test Statistic 25.05 0.77 35.51 0.26 30.83 0.47 23.81 0.82 41.34 0.10 21.78 0.89 31.01 0.47 31.84 0.42

Page 106 of 108

Heteroskedasticity Tests for Regression v (1/2)

Panel A
Breusch-Pagan test* *Chi-sq (1)

Panel B
White test* *Chi-sq (68) Whites General Test Statistic 69.00 69.00 69.00 69.00 69.00 69.00 69.00 69.00

chi2

Prob > chi2

Prob > chi2

Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit

0.45 2.91 0.01 0.27 3.81 0.05 4.20 0.18

0.50 0.09 0.93 0.60 0.05 0.83 0.04 0.67

0.44 0.44 0.44 0.44 0.44 0.44 0.44 0.44

Heteroskedasticity Tests for Regression v (2/2)

Panel C
Special case of the white test N Strategy Technology Market Professionalization Finance Credibility Internatioanlization Exit 69 69 69 69 69 69 69 69 R^2 of squared residuals 0.06 0.07 0.03 0.08 0.06 0.06 0.13 0.04 LM 4.27 4.74 1.89 5.71 4.25 4.13 8.99 3.06 P > Chi sq (2 df) 0.12 0.09 0.39 0.06 0.12 0.13 0.01 0.22

Page 107 of 108

IX. THEORY OF THE SPECIAL CASE OF WHITE TEST

The special case of the White test, the formula is shown below:

This combines the features of the Breusch-Pagan and White tests. Both tests regress the OLS squared residuals, , but they regress them on different values. The Breusch-Pagan test uses the independent variables from the linear model, while the White test uses the OLS fitted values for all the independent variables, the squares of the independent variables, and all the cross products. Since the fitted values are just linear function of the independent variables, preserves the spirit of the Breusch-Pagan test, while is a particular function of all the squares and cross products of the independent variables (Wooldridge 2009).

Page 108 of 108