Geographical preference in venture capital investments

(1)

Geographical preference in venture capital investments

‘An empirical study of geographical preference in the Dutch venture capital market’

By

Klaes Wester

University of Groningen

International Business & Management

Uppsala University

Business & Economics

MSc Thesis for IFM

MSc International Financial Management

Supervisor: Dr. Hein Vrolijk Co-assessor: Dr. Wim Westerman

(2)

Geographical preference in venture capital investments

‘An empirical study of geographical preference in the Dutch venture capital market’

Abstract

This thesis expands the literature on geographical preference by examining the investments of venture capital with a sample of Dutch venture capital investments for the period 2000 – 2010, showing that even in a small country as the Netherlands, venture capital investors prefer as a result of advantageous information and cultural distance geographical local ventures. By using a methodology developed by Coval & Moskowitz (1999), this paper proves that the average distance of venture capital investments is 23 percent smaller than expected. I find that larger venture capital funds and venture capital funds located in the capital Amsterdam exhibit a smaller geographical preference. Furthermore, this paper demonstrates that the geographical preference is larger for ventures operating in the ICT sector and as most venture capital funds are located within the Randstad, this geographical preference has impact on the availability of venture capital in the rest of the Netherlands.

(3)

I INTRODUCTION

Venture capital (VC) has long been considered as a crucial driver of new firm creation. Providing the funds and expertise that make startups and new firms grow faster and create more value compared to other firms. Therefore, venture capital can be seen as a critical driver of economic growth and value creation (Schwienbacher, 2006). Venture capitalists make a careful selection of which firms they will invest in. Apart from criteria related to innovation, growth and finance they also seem to pay attention to the geographical distance. Some venture capitalists seem to maintain criteria based on geographical distance, such as firms which are a 20-minute driving range will not be funded1.

This leads to a geographical preference where venture capital funds prefer ventures that are proximate. The existence of a geographical preference has been proven for a large country as the United States (Douglas & Dai, 2009). However, it has never been investigated for a small country as the Netherlands. This represents a serious gap in the knowledge required to establish an efficient venture capital market. For instance, in the Netherlands most venture capital funds are located in an area called the Randstad, which consists of the four largest Dutch cities (Amsterdam, Rotterdam, The Hague and Utrecht), and the surrounding areas. If venture capital funds has a geographical preference, ventures which are geographical distant from the Randstad can find it hard to attract venture capital.

There are two main arguments for the existence of a geographical preference. First increasing information asymmetries with increasing distance make distant investments costlier (Douglas & Dai, 2009). Second are cultural reasons, e.g. a preference to invest in the familiar and a discomfort with the alien and unknown (Huberman, 2002). Again, no research has been conducted in order to investigate if these aspects play a role in a small country as the Netherlands, which represents a gap in the knowledge on the functioning of venture capital.

1

(5)

5

The aim of this thesis is to find out whether a geographical preference in the Netherlands exists and if so, to investigate the causes and effects of a geographical preference.

This thesis adds to the existing research on the geographical preference of venture capital funds in two aspects. First and foremost, it investigates the existence of a geographical preference of venture capital funds for the Netherlands for the period 2000-2010, investigating whether a geographical preference for venture capital can exist for a small country as the Netherlands. By deepening the knowledge on the effect of distance on venture capital investment behavior, this thesis adds value to those who are involved with the venture capital market. For instance, the results can help ventures aiming for venture capital to support their selection process of the venture capital funds which they address. It can also aid government agencies who incorporate venture capital in their development policies by recommending taking the geographical distance and cultural distance into account.

Second, by analyzing characteristics that influence this preference, this thesis elaborates on the existence and causes of a geographical preference for a small country as the Netherlands. Research has shown that the working of venture capital is influenced by country specific effects such as cultural, legal, economic and institutional factors (Wells, 2000). Taking into account that these aspects differ from the Netherlands compared to the United States, this thesis will add to the understanding of the working of venture capital outside the United States.

(6)

6

II LITERATURE REVIEW

This chapter starts with discussing the problems that young high-tech ventures face due to information asymmetries, when trying to attract capital, and why for the same reason venture capital funds prefer to invest in high-tech ventures. Then, it is discussed how the existence of information asymmetry can result in a geographical bias, which can increase as a result of cultural factors. The next section describes the Dutch venture capital market. The chapter ends with research questions.

2.1 Problems for startups with attracting capital

I start the discussion on venture capital with the following question, what are the problems young firms, so called startups, face with attracting capital? This question is answered by describing the growth path of ventures and the resulting need for capital, and how information asymmetry, which leads to agency conflicts, making it hard for a venture to attract that capital. Every venture starts with an idea or vision, but it is a long way from an idea to a successful venture. A startup goes through several development stages before it becomes a mature company. The capital needs and characteristics of the company vary greatly during these stages. As a result, the investor which is most suitable for the company also differs per stage.

These development stages are described by Ruhnka & Young (1987). They analyzed the perceptions of the CEO or managing partner of 73 U.S. venture capital firms about the key features of new ventures throughout the development process. The results of this research were used to develop the; ‘venture capital model of the development process for new ventures’. This model consists of five stages and describes for each stage the distinguishing characteristics, goals and risks. The model is created using the experience of venture capital firms; therefore it does not necessarily portray the development process of all ventures (for instance ventures that are not meant to follow a high growth path and then go public or to be sold to another company). According to Ruhnka & Young (1987) the five stages of development are;

(7)

7

identified by Ruhnka and Young (1987); the production of the working prototype, making a market assessment and creating the structure of the company. Major risks are that the concept cannot be transformed into a workable prototype or that the potential market is too small. Other risks are that the development is delayed, funds running out or not being able to produce the product for competitive costs.

 Start-up/first stage – In this stage the business plan and market analysis are completed. The prototype is under evaluation or is ready for the market. However, the management team is sometimes incomplete and market studies must look promising enough to proceed with small scale production. Major goals mainly are, having a product ready to market, make some initial sales and verify demand, establish manufacturing feasibility and to have a complete management team. Major risks are similar to the seed stage with the additional concern that the founder cannot attract key management.

 Second stage –In this stage it is assumed that the management team is completed, the market is receptive, there are some orders and a marketing push is needed. Major goals at this stage are achieving market penetration and sales. Other goals are reaching breakeven or profitability, and increasing production capacity or reducing unit cost. A major risk at this stage is that the management team proves to be inadequate. Other risks are, products being not competitive in the marketplace with costs that are too high and there is no sufficient profit margin, or the size and growth of the market is lower than projected.  Third stage – Company pushes for expansion. There are significant sales and orders and

the company is starting to generate positive profit margins. However, the rapid expansion requires more working capital than can be generated from the internal cash flow. Major goals are meeting targets related to sales, growth, market share, cash flow breakeven or profitability. At this stage the preparation for the exit of an investor in the form of an Initial Public Offering (IPO), buyout or merger begins. A major risk is the management, who may not be able to manage the different demands of a larger more formal company. Also the sales level or market share can be below expectation. Now the company grows larger, there is also the increasing risk that unexpected competition appears.

(8)

8

IPO, merger or buyout. A major risk in this stage is not being solvent to increase the market share. Other minor ranked risks are related to problems with the exit of the venture capital firm from the company and management being inadequate.

During the development from an idea to a mature company there is a high need for capital. A startup following a high growth path requires large investments in working capital, however it is mostly not up till the Mezzanine stage that the venture becomes profitable. Therefore, in order to grow the startup has to search for external capital. Nonetheless, it proves to be a tough challenge for a new venture to attract that capital (Kanniainen & Keuschnigg, 2003)

Agency problems resulting in agency conflicts form the reason why it is hard for a startup to attract capital. Agency problems arise when ownership and control of an organization are separated. The investor receives a share of equity in return for providing capital and thus becomes owner of the company. Management, on the other hand, does not own the company but is responsible for controlling it (Fama & Jensen, 1983). Agency conflicts arise when the investor and management have conflicting interests; this leads to actions undertaken by management that do not maximize the value of the investors (Jensen, 1986).

These conflicting interests are described by Hamburg (2004), who calls it ‘goal incongruence’. In this situation, rational action undertaken by management will lead to different behavior than desired by the investor due to different goals of the investor and management. The investor demands maximization of the value of equity as this maximizes his return (shareholder wealth maximization). However, management strives for the maximization of their wealth (management wealth maximization). Because management does not own the equity, maximizing its wealth can lead to different goals than that of the investor, leading to goal incongruence. This incongruence is magnified by information asymmetry as it is reduces the ability of the investor to monitor the actions taken by the entrepreneur (Raphael, Brander, & Zott, 1998).

The observation that management of the firm might not always interact in the best interest of its stakeholders was already made by the classic economist Adam Smith;

(9)

9

The value of the difference in interests between shareholders and managers are called agency costs. In theory these costs can be measured as the value of the firm if it would have been run completely in the interests of the investors minus the actual value (Hamberg, 2004). Investors can take actions to reduce agency costs, although it must be kept in mind that actions taken by investors to reduce information asymmetry and align the efforts of management are costly. Therefore, their use should be weighed against the possible benefits. Two different categories of actions can in general be distinguished. First, by designing contracts that align management their interests with shareholders. These contracts specify the rights, payoff, and performance criteria of management and the investor (Fama & Jensen, 1983; Sahlman, 1990). Second, are monitoring activities, which have as goal to reduce information asymmetry by observing the actions of management. Examples are, for instance, meetings between the investor and the entrepreneur, or the production of annual reports (Hamburg, 2004; Raphael, 1998).

(10)

10

stage the venture possesses some assets and banks might be willing to provide small amount of debt. But these may not be enough to finance the working capital which is required for rapid growth (Sahlman, 1990). The venture now has the option to finance growth by earnings, and thus to rely on a slow growth path, or it must search for other forms of capital (Powell et al. 2002) Concluding, and to answer the question at the beginning of the paragraph, what are the problems young firms face with attracting capital? Young firms find it difficult to attract capital as a result of agency conflicts which stern form information asymmetry. Especially debt is difficult because the young firm lacks assets which can be used as collateral.

2.2 For which ventures is venture capital appropriate

This paragraph continues with the theoretical framework and focuses on the following question, what is the influence of information asymmetry on venture capital investments? To answer this question I first introduce a framework that uses the development stages of Ruhnka & Young (1987) to describe active investors and the venture they prefer to invest in. Followed by a definition of venture capital, the advantages venture capital brings and its peculiar relation with information asymmetry.

Active investors

(11)

11 Figure 1: capital market

€ x 1000 stock market

>20.000 private equity

20.000

10.000 combination of investors and venture capital

5.000 government support 2.560 venturecapitalists 1.280 640 informal investors 320 160 80 friends / family 40 20

Seed stage Start-up phase Second stage Third stage Mezzanine stage

Maturity

Source: (Witte, 2008)

The lines in figure one describe common grow paths that firms take. Line A; illustrates a high growth company. Line B illustrates a company with high research and development costs. Line C illustrates an average innovative company. Line D illustrates a successful entrepreneur that relies mostly on self financing. This study focuses on the high growth companies described with the lines A, B and C as these require capital support.

Figure one describes six different investors (Witte, 2008). For the seed stage, startups typically rely on Friends and family. These investments consist out of small amounts of funding which usually do not exceed 20.000 euro and are used for market assessment and working out ideas.

Informal investors, also known as business angels form a broad category of investors,

commonly consist of individuals with an entrepreneurial background which is used to provide assistance to companies (Wester-Koopmans, 2010). It is estimated that there are more than 1000 informal investors in the Netherlands (Witte, 2008). Informal investors are good financers for the start-up phase and second stage. However, as they rarely invest more than half a million euro, they are less suitable for providing the amounts required in later development stages. Venture

Capitalists are private investors who mostly invest with a minimum of two million and up till

fifty million euro. They seem equal to informal investors but differ in their strong focus on returns and the larger capital provisions. Venture capitalists, in general, do not possess much

Line B

Line A Line C

(12)

12

time for guiding their companies because of the severe amounts of capital they control. Private

equity has a similar approach to venture capital, which is sensible as venture capital refers to a

form of private equity. Private equity, excluding venture capital is described by Wells and Leslie (2000) as; “investments made by institutions or wealthy individuals in both publicly quoted and privately held companies, mainly focusing on management and leveraged buyouts.” In general by private equity, excluding venture capital, are meant investments in more mature companies. The stock market is a feasible option for more established companies that require vast amounts of capital. The disadvantage of going public are the high costs involved, as it is estimated that costs associated with bringing a company public in the Netherlands are around fifteen million euro.

Venture capital

Venture capital, as shown in figure one, is the preferred investor for companies that are between

the startup stage and the mezzanine bridge. Venture capital prefers to invest in high-tech companies that possess vast growth prospects. Cumming (2002) gives the following definition:

“Investments in small private growth companies that typically do not have cash flows to pay interest on debts or dividends on equity. VC invests in companies over period that generally stretches between 5 and 7 years. After this period the main exit strategies are; going public (IPO), acquisition, merger buyback, and liquidation.”

(13)

13

Advantages of venture capital

Compared to traditional forms of capital there are significant advantages for a company having a venture capital funds as investor. First, the funds do not just posses and distribute capital, but also get active involved in managing their portfolio companies. By taking on a seat on the board of directors, they, using their expertise, help in recruiting key personnel, work with suppliers and customers and help with drafting strategy and marketing (Sahlman, 1990; Engel, 2006; Powel, 2001). For example, Sapienza (1996) found that venture capital perceive strategic involvement as their most important role, i.e. providing financial and business advice and functioning as a sounding board.

A second advantage is the signaling effect that goes out from being associated with venture capital. As noted earlier, it is hard to evaluate a startup, especially a high-tech startup, without prior track record on which to base an evaluation. A stream of research has suggested that external actors rely in this case on the quality of the startup’s affiliates as a signal of its quality. Venture capital often invests in less than one percent of the business plans that they receive and their due diligence process requires a detailed analysis of the company (Sahlman, 1990). Passing this screening sends a powerful signal to external parties increasing the reputation of the startup (Hsu, 2004; Davila et al. 2003). Engel & Keilbach (2006) found that these two advantages led to higher company growth.

Venture capital and information asymmetry

(14)

14

An appropriate quote here is of Raphael et al. (1998): “We view their (venture capital) “raison d’être” as their ability to reduce the cost of informational asymmetries.”

A high return is required to compensate for the risk that is involved with investing in these companies (30 percent per year is one benchmark). The companies are required to show tremendous growth during this investment period to provide these returns. This makes companies that show fast growth prospects particular interesting for venture capital investments (Davila et al. 2003).

Concluding, and to answer the question of this paragraph, what is the influence of information asymmetry on venture capital investments? venture capital has a preference for early stage high-tech ventures with vast growth prospects. These ventures require severe amounts of capital to realize their growth potential and have evident information asymmetry problems making them unattractive for other investors. Due to its active involvement, in terms of selecting and monitoring venture capital is especially equipped to deal with information asymmetry.

2.3 Information asymmetry and the geographical bias.

This section explains the tools that venture capital posses for reducing the agency conflict resulting from information asymmetry and how these tools lead to a geographical bias. This section answers the following question, how does venture capital cope with information asymmetry and how does this lead to a geographical bias?

(15)

15

variability in return (Cumming, 2002). The due diligence undertaken by venture capital is essential in overcoming the adverse selection problem. The moral hazard problem occurs after the contract has been signed, when management undertakes different actions than it is supposed to under terms of the contract, for example adopting ‘pet’ projects. Venture capital responds to the moral hazard problem by staging capital investments and providing intensive monitoring (Sahlman, 1989; Lerner, 1995).

Being geographical proximate helps with handling information asymmetry and the resulting agency problems as it provides an information advantage and easier access to information (e.g., Lerner, 1995; Martin, 2001; Douglas & Dai, 2009; Cumming, 2002). Accordingly, information asymmetry and the problems of hidden information and action are reduced, and as a result local venture capital funds are expected to hold an advantage over distant funds. This preference for local investments holds as well for other forms of equity investments, as the two notable studies of Coval & Moskowitz (2001) and Hubermann (2002) demonstrate.

Coval & Moskowitz (2001) prove that a geographical proximate can have a positive impact on earnings. They investigate the geographical proximate of U.S. investment managers, and found that fund managers earns a substantial abnormal return on geographical close investments. In addition, this study states that the extent to which the firm is held by nearby investors has a positive influence on future performance. The average fund manager seemed to be able to generate a 1.84% more return, on investments which were located less than 100 miles from their office. The Coval & Moskowitz (2001) study suggested that fund managers are in the possession of informational advantages about local companies, which might be a result of improved monitoring or access to private information.

(16)

16

to actual informational asymmetries also perceived information asymmetries have an influence on the geographical bias.

I expect the geographical preference to be more severe for venture capital investments because high-tech ventures are characterized by high information asymmetry. There are number of ways in which a geographical preference helps in reducing information asymmetry and agency problems. First, is the due diligence process that a venture capitalist undertakes in order to reduce the risk of adverse selection. Second, is the monitoring role venture capital undertakes to reduce the moral hazard problem.

Ex ante: due diligence

Venture Capitalists typically consider an enormous amount of proposals, whereas only a small amount are seriously considered and even less are being carried out. As already noted Sahlman (1990) found that a venture capitalist typically receives more than a 1000 proposals each year for which the rejection rate is more than 99 percent. Indeed it can be said ‘many are called but few are chosen’ (Matthew 22:14). The process of conducting due diligence and the screening of all these companies can be carried out more convenient when ventures are geographical proximate. For instance, it is easier to collect important soft and private information (Wright et al. 1998). There are also the costs of traveling. Douglas & Dai (2009) found in a study of US based venture capital that there are on average three to eight meetings between the venture capitalists and the company before funding was given. This makes travel time a substantial cost factor with increasing distances taken into account.

Ex post: monitoring

(17)

17

2009). Powell et al. (2002) argues that these advantages are augmented for life science and other high-tech startups. Venture capital can bring for these ventures the much needed management advice as many of the firms founders are research scientists lacking management experience.

Research conducted in the US has confirmed the existence of a geographical bias. Lerner (1995) found in a study on US based venture capital that more than half of the companies are located within 60 miles from their venture capital firm. Furthermore, did he prove that venture capitalists have a less active role in firms which are geographical distant. Lerner (1995) investigates the relation between geographical distance and board representation and concludes that geographic proximity is an important determinant for the amount of venture capital board members. Companies located within five miles from their venture capital firms are twice as likely to be represented on the board2. Cumming (2002) finds a geographical bias of 60.4 percent for venture capital investments in the United States. The relation between venture capital and its preference for local ventures is summarized in figure two.

Figure 2: Causes for a geographical bias in venture capital investments.

In order to conclude this paragraph, and to answer the question raised, how does venture capital cope with information asymmetry and how does this lead to a geographical bias? The problems related to information asymmetry can be analyzed in two ways; before or after the contract is

2

Taking into account the many advantages a geographical proximate brings it might seem unlikely for distant relations to even occur. However, there are ways in which the potential drawback of a distant relation can be alleviated. For instance, information asymmetry can be reduced by syndication or the creation of networks. Syndication facilitates sharing of information and members in different locations can rely on each other for providing monitoring and important local information. Therefore it can be expected that better networked VCs exhibit a smaller geographical bias. (Douglas & Dai, 2009; Hochberg, 2007)

Geographical bias Due diligence – hidden

information problem

Monitoring – moral hazard problem

(18)

18

signed, leading to respectively adverse selection or the moral hazard problem. Due diligence is used to prevent adverse selection and monitoring to prevent the moral hazard problem. Both tasks can be performed more convenient by venture capital investors that are proximate to their ventures.

2.4 Characteristics of the Dutch venture capital market

This paragraph describes the Dutch venture capital market by answering the following question: “What are the distinguishing characteristics of the Dutch venture capital market?” The major issues regarding this question are the structural differences between venture capital markets, and I can elaborate on this question by comparing the UK and the US venture capital markets. Especially the US has an older more developed venture capital market which is often seen as a result of a more liberal economy. Thereafter, two distinguishing characteristics of the Dutch venture capital market are described.

(19)

19

Netherlands requires a lower rate of return compared to the UK, US and France. Manigart et al. (2000) states that venture capital in those countries have a knowledge advantage which is translatable into a strategic advantage and higher margins. Furthermore, this research found that the attained strategic advantage and higher margins are hard to transport to other countries and that there are important differences between the valuation approaches of venture capital investment and the relative importance placed upon financial and accounting information during this valuation. Venture capital funds in network based system such as the Netherlands may place a greater importance on information relating to the individual entrepreneur than in market based systems as the UK and the US. Manigart et al. (2000) concludes that if venture capital wants to exploit their perceived competitive advantage in another country, need to devote considerable time and effort adapting to these differences. A major result of venture capital crossing borders is mainly that these venture capitalists usually participate in larger investments, since then the need for management support in these investments is lower (Schertler, 2009).

The government traditionally holds a strong presence in the Dutch venture capital market, although this presence is decreasing during the last decennium (Martin, 2001). Currently the Dutch government has two important instruments; the ‘Regionale Ontwikkelingmaatschappijen’ (ROM) and the ‘seed funds’.

(20)

20

back by a percentage of the profit made (as a result the government bears the default risk). The conditions of the Seed funds are simple; investments must be aimed at innovative starters with a maximum investment of 2.500.000 and with a maximum average investment over the fund its portfolio of 800.000 euro. This makes seed investments on average a lot smaller than the average venture capital investment which is 6.9 million (NVP-Ondernemend Vermogen, 2009). 29 funds are making use of the seeds funds rule since the start in 2005. These funds have financed 114 investments during this period.

In order to conclude this paragraph, and to answer the raised question concerning the characteristics of the Dutch venture capital market. The Dutch venture capital spends less time monitoring and providing value adding efforts to their portfolio companies compared to the US and the UK and therefore their required rate of return is also lower. Further, has the Dutch government a presence on the market via the ROM and the Seed funds.

2.5 Research hypothesis

The analysis in this thesis follows a two-step approach. The first step is to investigate the Dutch preference of venture capital funds for geographical local ventures. In order to do so, an argument for expecting a geographical bias in the Netherlands is established which is presented as a research hypothesis. In the second step is to investigate how, as a result of information asymmetry, attributes of venture capital firms, ventures and deal characteristics influence the geographical bias. By analyzing the effect of information asymmetry on venture capital investments I will be able to shed some light on the reason of a local bias.

Step one: the preference of venture capital for geographical local firms.

(21)

21

assumption of the CAPM do not hold for venture capital (Douglas & Dai 2009; Stuart & Sorensen, 2001; Tian, 2009). However, all these articles were using a sample of US based venture capital, a country which is a lot larger than a small country as the Netherlands which is from north to south no longer than 300 km. In comparison, Douglas & Dai (2009) found an average distance of 453 kilometer between a venture and the closest branch office of its venture capital funds (although the median distance was only 38 kilometers suggesting a highly skewed dataset). These findings show that because of agency problems, venture capital funds prefer local ventures over distant ventures. These can be a result of information advantages as suggested by Coval and Moskowitz (1999) or a preference of the human psychology to invest in the familiar as advocated by Huberman (2002). This implies that a geographical bias would even exist for a small country as the Netherlands.

Therefore, I use the hypothesis that also in the Netherlands there is a geographical bias; ventures capital investors do not follow the predictions of the CAPM model when making investments but prefer local ventures.

Hypothesis 1: Dutch Venture capital funds prefer geographical local ventures.

Step two: the effect of information asymmetry on geographical preference of venture capital investments.

Having established the reason to expect a pattern of geographical preference I will now develop specific predictions which are related to attributes of venture capital firms and characteristics of the ventures that they invest in. These predictions are presented as research hypothesis in figure three.

Figure 3: characteristics influencing the geographical preference

Geographical bias

Venture capital characteristics Hyp2: Large venture capital funds

Hyp3: Location venture capital

(22)

22

Hypothesis 2: The geographical preference is smaller for larger venture capital funds.

Hsu (2004) argues that the investment experience of a venture capital funds reduces the geographical preference. First, experience can reduce the time spent on monitoring as venture capital becomes more adept at this task. This will have as result that venture capital should be more prepared to invest in geographical more distant companies as the time costs of monitoring these investments decreases (Stuart & Sorensen, 2001). Further, creates a venture capital funds during the investment process relationships with entrepreneurs, experts and other venture capital ventures. Through this network information of a better quality and with more ease is provided, which increases the capacity of the venture capital funds to reduce the information asymmetry associated with distant investments. Therefore, I expect larger venture capital funds that have made more investments and have consequently more experience to exhibit a smaller geographical preference.

Hypothesis 3: The geographical preference is smaller in the capital ‘Amsterdam’.

It can be expected that competition among venture capital funds decreases a geographical preference (Cumming, 2002; Sorensen & Stuart, 2001). According to the law of supply and demand; a high supply lowers the price. Therefore too much capital in a local area is expected to lower margins with as result that venture capital starts searching for investments further away. Because when the return of local investments decreases, distant investments become more attractive despite the higher information asymmetry. In the Netherlands around fifty percent of the venture capital funds are located in the capital ‘Amsterdam’, and as a result I expect these venture capital funds to have a lower geographical preference.

Hypothesis 4: The geographical preference is larger for seed stage financing.

(23)

23

equity gap is a result of small firms becoming overly expensive in terms of valuation and monitoring. First, it is harder to evaluate the investment proposal of a venture in an early development stage as it lacks a track record that can be used for the due diligence process of judging an investment proposal (Kanniainen & Keuschnigg, 2003). In contrast, for a venture in a later development stage there is more information available on the company, and importantly, also on the performance of its management. Further, might later stage companies also require less monitoring as organizational routines and management become more established (Stuart & Sorensen, 2001). Because of the higher information asymmetry and resulting agency conflict I expect investments in seed ventures, which are in the earliest development stage, to exhibit a larger geographical preference compared to the rest of the research sample.

Hypothesis 5: The geographical preference is larger for high technology firms.

(24)

24

III METHODOLOGY

The set of research hypothesis is tested using a two-step approach. In the first step, a methodology developed by Coval & Moskowitz (1999) is used to calculate the geographical preference. The preference is measured as a bias showing the percentage difference between the actual distance and a hypothetical benchmark distance. In the second step a multiple regression analysis is used to find out which venture and venture capital characteristics have an influence on the geographical bias.

3.1 Model description and justification

Step one: description of methodology to calculate the geographical bias

The methodology developed by Coval & Moskowitz (1999) compares the average distance of an investment portfolio to a hypothetical benchmark portfolio. The distance of the hypothetical benchmark portfolio is measured as the average distance between the venture capital funds and all ventures it could have invested in. This is consistent with the prediction of the Capital Asset Pricing Model (CAPM) that an investor considers all possible investments possibilities and distance poses no restriction. The CAPM functions under several assumptions of which two are especially of interest; there are no transaction costs and all information is publicly available. As argued in the literature review, are there large transaction costs and high information asymmetries in the venture capital market. If consequently, venture capital prefers local ventures, then this will show up as a geographical bias measured as the percentage difference between the actual and the benchmark distance of the venture capital funds and its investment. This method is also used for investigating the geographical bias of shares by Coval & Moskowitz (1999, 2001) and Zhu (2002) and the geographical bias of venture capital by Cumming (2002), Douglas & Dai (2009). A detailed description of how the geographical bias has been calculated is included in Appendix 8.1.3

3

(25)

25

Gini-coefficient

A geographical preference in the Netherlands has influence on the supply of venture capital. A major reason is that a geographical preference makes it harder to attract venture capital in areas geographical distant from a venture capital fund. Anecdotal evidence suggests that in the Netherlands the regional distribution of venture capital is quite uneven as most venture capital funds are located in the capital Amsterdam. To investigate the relation between the distribution of venture capital funds and its effect with respect to a geographical preference, the ‘Gini-coefficient’ is used. This coefficient is a measure of inequality of distribution, where a value of 0 signals total equality and 1 total inequality. The Gini-coefficient is mathematically based on the Lorentz curve, where the cumulative proportion of venture capital activity in percent of the provinces is measured. A detailed explanation of the Gini-coefficient is included in the appendix.

Step two: regression of factors influencing the geographical bias

In the second step the factors that contribute to the geographical bias in venture capital investments are analyzed. This analysis is conducted using a multiple regression, which is the same approach as Cumming (2002) uses in a study of the geographical bias of venture capital in the United States. The Multiple regressions has the following form:

= + +

is a vector of the geographical bias of venture capital trade4

and a matrix of venture capital and venture characteristics. is the random error term with mean 0 and variance of that measures the effect of unidentified factors.

4

Normally, takes on values between 0 and 1 which suggest using a logit transformation. However, can also

(26)

26 3.2 Operationalization of the independent variables

The characteristics used in the second step include, sector of a venture, the size of a venture capital funds, venture capital funds located in the area Amsterdam, and whether or not it is a seed fund. The operationalizations of these variables are shown in table 1.

Sector venture: three dummy variables measure if the geographical bias differs between

industries, by taking on the value 1 if a venture operates in a specific industry. These dummy variables are; life science, ICT, other. The ventures are classified using US SIC (United States Standard Industrial Classification). 70 investments (10 percent) for which no business description is available are classified as ‘other’.

Amsterdam: I use a dummy that takes on a 1 if the fund is located in the region Amsterdam to

measure if the bias of venture capital funds operating from this region differs. Amsterdam is defined as all postcodes ranging from 1000 up till 1500.

Seed funds: to measure if the size of the venture has influence on the geographical bias I include

a dummy that takes on a 1 if the investment is made by a seed funds. A venture is characterized as a seed venture when falls under the ‘seed instrument’ as described in chapter 2.4.

Size: to determine if the geographical bias differs for large venture capital funds a dummy

(27)

27

Table 1

Operationalization of the variables

Variables Measurement

Dependent Geographical bias Percentage difference between the actual distance and a hypothetical benchmark.

Independent Sector venture Dummy taking on a 1 if “ICT”.

Dummy taking on a 1 if “Life Science”. Dummy taking on a 1 if “other”.

Amsterdam Dummy taking on a 1 if the VC is located in the postcode area “Amsterdam”.

Seed funds Dummy taking on a 1 if the VC is a “seed funds”. Size Dummy taking on a 1 if the VC belongs to the 25%

(28)

28

IV DATA DESCRIPTION

The first section of this chapter describes the sources of the data used in this thesis and the second section the sampling frame used to construct the research sample. The third section contains descriptive statistics that show the sector, stage preferences and geographical distribution of venture capital investments across the Netherlands. The fourth section contains an analysis of the correlation between the independent variables.

4.1 Data sources

For my investigation I needed a research sample that is representative for the venture capital market of the Netherlands and, at least, contains data on the location of the venture capital funds as well as the location of its portfolio of companies. There is no dataset that readily meets these criteria in the Netherlands and as result I had to construct my own dataset. This is done by using three different sources with information on venture capital investments.

Table 2

Description databases used to construct the research sample

Database Number

investments

Description

NVP 2657 Contains the data of private equity investments of its 60 members and has coverage of 95% of the Netherlands measured in monetary value.

Zephyr 276 Gives a good representation of investment made by venture capital funds that are not member of the NVP. Contains data on many investments made by foreign venture capital funds.

(29)

29

NVP

The NVP or “Nederlandse Vereniging van Participatiemaatschappijen” is the Dutch association of ‘participatiemaatschappijen’ which literately translated into English means; “every organization that takes a controlling stake in a company”. By this definition are meant all forms of private equity. The NVP has around 60 members which are, measured for the value of the investment, responsible for around 95 percent of all private equity investments made in Dutch companies.5 This percentage tends to vary from year to year because foreign private equity funds tend to make incidental very large investments.

The database of the NVP contains information on 2657 investments made by its members for the period 2000 – 2010. These investments include both venture capital and private equity. For each deal the NVP gives the year of the investment, divestment and a business description of the venture which contains a short summary of the main activity of the venture.

Zephyr

Zephyr is a database from Bureau van Dijk which has a worldwide coverage for various deal types. When searching for venture capital investments and selecting only investments that are described as venture capital, 276 venture capital investments are found for the period 2000 – 2010. These investments are made by 105 different venture capital funds of which 73 are from outside of the Netherlands. A big advantage of Zephyr is the additional information it provides, it has for venture capital funds the postal codes and for ventures the postal codes and sector. Furthermore, contains it the monetary value for around 50 percent of the investments, which is on average 12.44 million euro with a median of 6.5 million euro. This means that Zephyr does a better job at capturing larger scale investments as the average size of a Dutch venture capital investment is 6.9 million (NVP-Ondernemend Vermogen, 2009).

(30)

30

TechnoPartner

TechnoPartner is a governmental organization which supports high-tech entrepreneurs. One of its goals is to keep track of the progress of the seed funds instrument, a governmental instrument aiming at reducing the equity gap for innovative ventures that require funding between 100.000 and 2.500.000 euro. This is done by providing a loan to venture capital funds that invest in seed ventures. Data on these investments are retrieved from the internet site of Technopartner (www.TechnoPartner.nl). The TechnoPartner database contains data on 29 venture capital funds and 114 investments for the period 2005 – 2010. The advantage of the TechnoPartner database is that it contains information on a set of small venture capital investments of which the majority is not recorded by the two other data sets. The disadvantage is that it only contains the name of the venture capital funds and its portfolio ventures and no data on the sector and location of the venture or venture capital funds.

Postcodes

(31)

31 4.2 Sampling frame

The next section describes step by step how I constructed the research sample used in this thesis.

1. First, I combine the data from the NVP, Zephyr and the Seed database into one data file and delete 213 investments that are now double counted.

2. Using the member description on the NVP site (www.nvp.nl) every member that does not explicitly state they are investing in venture capital is deleted. In this step 942 transactions are deleted.

3. After that the business description is used to control if an investment is of venture capital nature. In this step 192 are not considered to be venture capital investments.

4. The research sample also contains the information on 413 investments made by ROM (regional development and investment companies). They are financed by the Dutch government and their goal is to improve the economic infrastructure of the less developed areas of the Netherlands. These agencies are bound to a certain area of the Netherlands and therefore show per definition a highly geographical bias. These investments are excluded in order to prevent them from dominating the results.

Table 3

Construction of research sample

Step taken Excluded

investments

Remaining investments

1. Combining NVP, Zephyr, TechnoPartner 3047

Excluding double items 213 2834

2. Excluding funds that invest in private equity 942 1892 3. Excluded private equity investments

4. Excluding investments made by ROM

192 413

1700 1308 5. Seperating foreign venture capital investments 198 895 6. Excluding venture capital on which no data is

available

(32)

32

5. The research sample contains information on 198 international venture capital investments (investments made in the Netherlands by foreign venture capital). The same selection criteria as for Dutch venture capital are applied after which 48 funds and 72 venture capital investments remain. These international funds are separated from the Dutch venture capital funds. This is done because for international funds the distance can be that much higher that they act as strong outliers clouding the results. For example, an US based venture capital funds that is located 5000 kilometer away will have a strong influence on the average distance and also on the average distance of the benchmark portfolio6.

6. Missing postal codes of the venture capital funds and their portfolio ventures are found using the database Reach. Ventures on which Reach there has no data available are excluded from the research sample. After these steps 123 venture capital funds remains which have made investments during the period 2000-2010 in 697 companies.

6

(33)

33 4.3 Descriptive statistics

Pluijm (2010) estimates that there are around 350 venture capital funds in the Netherlands. The research sample used in this thesis contains the investments made by 123 Dutch venture capital funds and covers therefore the activities of slightly more than one third of all Dutch venture capital funds. However, I believe that the research sample contains data on more than one third of all investments made during the last 10 years. This is because the main source of my data is the NVP database which has mainly larger venture capital funds as a member. The venture capital market is skewed in the sense that large venture capital funds stand for much more investments than smaller ones. For example, the eleven largest venture capital funds in the research sample account for 25 percent of the investments. The average portfolio size of a fund in the research sample is 5.6 ventures compared to fifteen for the eleven largest funds. A list of the ten largest venture capital funds in the Netherlands is presented in table 4. When looking at fund characteristics, 32 funds (26 percent) of the research sample have a preference for a sector of industry (ICT, life science, other), and 36 funds (29 percent) have made only one investment during the sample period.

Table 4

Ten largest venture capital funds measured for the number of transactions, 2000-2010.

Name venture capital funds VC

transactions

Province

Friesland bank investments 51 Friesland

Gimv 31 Zuid Holland

Gilde 31 Utrecht

Synergia Capital Partners 22 Utrecht

Newion Investments 22 Friesland

Ecart Invest 17 Zuid Holland

Doen 17 Noord Holland

Wadinko 17 Overijssel

Greenfield Capital Partners 16 Noord Holland

(34)

Geographical preference in venture capital investments

34

Table 4

Distribution of venture capital across the Netherlands

P r o v in c e V e n tu re c a p it a l fu n d s P e rc e n ta g e in v e st m e n ts fr o m p r o v in c e p e rc e n ta g e in v e st m e n ts in p r o v in ce p e rc e n ta g e F o re ig n v e n tu re c a p it a l in v e st m e n ts p e rc e n ta g e Noord Holland 62 50,4% 274 39,3% 207 29,7% 31 43,1% Zuid Holland 22 19,5% 106 15,2% 135 19,4% 13 18,1% Utrecht 14 11,3% 127 18,2% 96 13,8% 14 19,4% Overijssel 3 2,4% 25 3,6% 34 4,9% 8 11,1% Noord Brabant 10 8,1% 56 8% 105 15,0% 5 6,9% Limburg 2 1,6% 6 0,9% 12 1,7% 0 0 Groningen 1 0,8% 1 0,1% 25 3,6% 0 0 Gelderland 4 3,3% 29 4,2% 52 7,5% 0 0 Friesland 2 1,6% 72 10,3% 14 2,0% 1 1,4% Flevopolder 1 0,8% 1 0,1% 9 1,3% 0 0 drenthe 0 0,0% 0 0,0% 6 0,9% 0 0 Zeeland 0 0,0% 0 0,0% 2 0,3% 0 0 Total 123 100% 697 100% 697 100% 72 100%

(35)

35

Table 5 describes the distribution of venture capital investments in the Netherlands. The first column contains the location of venture capital, what seizes the attention is the high clustering of venture capital in the province Noord Holland which is a result of the large number of venture capital funds located in the capital Amsterdam. The second column presents the number of investments made from a province, here Friesland stands out with 10 percent of the investments while only 2 percent of the ventures is located in that province. This discrepancy between the amount of ventures and venture capital investments is caused by “Friesland Bank Investments” a large venture capital firm which is located in Friesland’s capital Leeuwarden.The third column describes per province in how many ventures has been invested. The venture capital market in the Netherlands is highly skewed, 80 percent of all venture capital funds are located in three provinces while, in comparison, these provinces stand for around 45 percent of the gross domestic product (GDP) of the Netherlands. Table 6 shows the distribution of the ventures towards sector and development stage. Almost half of the ventures are operating in the ICT or the life science sector and 16 percent of the ventures are in the seed stage. A graphic overview of the venture capital investments in the Netherlands is given by figure four and five in the appendix. These figures clearly illustrate the clustering of venture capital in some areas.

Table 5

Distribution of ventures by sector and development stage

Variables Ventures percentage

sector Life science 120 17%

Other 382 55% ICT 192 27% Development stage Seed stage 114 16% Later stage 580 84%

Size Large venture capital funds 174 25%

Other venture capital funds 523 75%

Amsterdam Located in Amsterdam 201 29%

(36)

36 4.4 Analysis of the correlation

I analyze the research sample for correlation between theindependent variables using a Pearson correlation. The resulting matrix of correlation coefficients for the independent variables is presented in table 7. The correlation coefficients are analyzed to test for a multicollinearity problem between the independent variables in the regression model. Multicollinearity can be problematic as it makes it harder to assess the impact of one variable on , because a higher correlation between independent variables decreases the change that they give an accurate estimation of the partial slope (Blalock, 1963). A number of significant correlations are found which are made bold and presented in table seven. To test if these significant correlations lead to multicollinearity, variance inflation factors (VIF) are used. The VIF looks at effects of the proportion variance that independent variable shares with other independent variables on the estimated regression coefficient of the th variable. As a rule of thumb; if any of the VIF’s is greater than five there is a multicollinearity problem (O’Brien, 2007). Because no VIF exceeds the value two I conclude there is no multicollinearity problem.

(37)

37

Table 6

Correlation matrix of the independent variables

Pearson Correlation Coefficients S ee d V e n tu re C a p ita l IC T V e n tu re s L if e S c ie n c e V e n tu re s L a rg e V C VC A m st e rd a m

Seed venture capital 1 ,289** ,106** -,216** ,032

Sig. (2-tailed) ,000 ,005 ,000 ,405

ICT ventures 1 -,283** -,127** ,075*

Sig. (2-tailed) ,000 ,001 ,047

Life Science ventures 1 -,143** -,045

Sig. (2-tailed) ,000 ,238

Large VC 1 -,006

Sig. (2-tailed) ,866

VC Amsterdam 1

Sig. (2-tailed)

(38)

38

V RESULTS

In section 5.1 the absolute distances between venture capital and ventures are presented. Section 5.2 shows the geographical bias and how this bias changes in different periods. The results for answering research question three, ‘the geographical bias is lower for venture capital located in the capital Amsterdam’, are presented in section 5.3. Section 5.4 analyses and compares international venture capital investments with Dutch venture capital. The last section presents the results of the multivariate regression.

5.1 Geographical distance

In table 8 the absolute distances between venture capital and its ventures are presented. The table also contains results on how this distance changes according to the size of a venture capital funds and the development stage and sector of a venture.

The results suggest that venture capital investments in the Netherlands are located geographical proximate to their ventures; the average distance between a venture capital funds and its portfolio venture is 54 kilometers. Furthermore, the distance does not exceed 78 kilometers for 75 percent of the investments and is for 25 percent even less than 21 kilometer. In comparison, the distance from the northern city of Groningen to Amsterdam is around 150 kilometers.

Venture characteristics seem to affect the distance of venture capital investments. The distance is less for ventures operating in the technology intensive sectors ICT and Life Sciences. 75 percent of the investments in ICT ventures do not exceed a distance of 63 kilometers. The average investment distance for seed ventures is also less, although this seems to be caused by outliers as the median distance does not differ from the full sample.

(39)

39 5.2 Geographical bias

In table 9 the geographical bias of venture capital investments is presented. This is measured as the percentage difference between the average distance of venture capital investments and what would be expected under the assumption of the CAPM model. Further, is shown how various venture capital and venture characteristics affect the geographical bias, for all characteristics the F-test show a consistent significant geographical bias. Venture capital investments in the Netherlands are highly localized; the average geographical bias is 23 percent. This means that the actual distance between a Dutch venture capital funds and its ventures is 23 percent smaller than which would be expected if distance would pose no restriction.

Table 7

Geographical distance of venture capital investments

Variables Mean distance Percentiles (kilometers) 25 % 50 % 75 % Full sample 54,4 20,6 50,1 77,6 Sector Other 59,5 23,1 52,7 84,6 ICT 45,6 15 41,3 62,8 Life Science 52,2 16,8 47,7 73,4 Stage Seed 49,4 15 50,1 68,9 No seed 55,3 20,6 50,1 80,9 Size Large 73,2 29,8 66,4 112,8 Other 48,7 17,55 45,3 68,6

(40)

40

The geographical bias is also clearly affected by venture characteristics. The bias is with 33 percent the largest for ventures which operate in the ICT sector and is also a little higher, although less profound, for the life science sector. The bias is three percent higher for seed venture capital, which means that the development stage has influence on the geographical bias. Venture capital characteristics also have a clear influence on the geographical bias. With 13 percent the geographical bias is the smallest amongst all categories for large venture capital funds.

Table 8

Geographical bias of venture capital investments Variables Distance Benchmark

distance

Geographical bias

F-test

Full sample 54,4 72,6 23 %

Sector Life science 52,2 67,7 22 % 0.000

ICT 45,6 69,3 33 % 0.000 Other 59,5 75,77 19 % 0.000 Stage Seed 49,4 70,5 26 % 0.000 Non seed 55,3 73 23 % Size VC Large 73,2 86,3 13% 0.000 Other 48,7 68,45 27% 0.000

(41)

41

Geographical bias per year

The geographical preference of venture capital investments is consistent over time and the absolute distance for venture capital investments increases a little7. Furthermore, increases the benchmark distance, becomes the Gini-coefficient8 lower and decreases the share of investments conducted in the Randstad. This is the result of an interesting trend; venture capital investments become during the sample period more evenly distributed across the Netherlands. In the beginning 46 percent of all investments are conducted in the Randstad in the last period the share has decreased to 27 percent.

7_{Regression only finds a small, but significant, effect (} _{= 0,014 at the 0.00 level).}

8_{The Gini-coefficient can be influenced by differences in absolute number of investments between periods as it only}

looks at relative changes and not changes in absolute values. Therefore, the number of investments per period is presented in column 6.

Table 9

Geographical bias by year

(42)

42 5.3 Geographical bias of Amsterdam

In this section I compare venture capital funds located in the area Amsterdam with the rest of the Netherlands. To make the comparison I use a set of variables which describe the density of the venture capital market in Amsterdam and compare it to the rest of the Netherlands. Amsterdam is defined as all postcodes between 1000 and 1500.

There is a significant difference in the geographical bias; funds located in Amsterdam have a bias of 20 percent compared to 31 percent for funds located in the rest of the Netherlands. Interestingly, the funds located in Amsterdam are smaller when measured for the number of investments they make. A venture capital funds located in Amsterdam made over the period 2000 – 2010 on average only 4.4 investments compared to 7 for the rest of the Netherlands. The variable ‘ventures per VC’ compares the ratio of number of new ventures available for investment to the number of venture capital investors. This gives an indication of how competitive the local venture capital market is. Table eleven shows that the local market for venture capital is much more competitive in Amsterdam, for every funds there were over the period 2000 – 2010 3,3 ventures in Amsterdam compared to 8,3 for the rest of the Netherlands.

Table 10

Geographical bias of Amsterdam

Variables Full sample Amsterdam Outside

Amsterdam Local bias 23% 20% 31% Number VC 100% 51,2% 48,7% Number of investments 5,7 4,4 7,0 Ventures per VC 5,7 3,3 8,3 Cluster index 100% 29,7% 70,3% Percentage economy 100% 17,9% 82,1%

(43)

43

Cluster Index measures the amount of ventures located in an area. Almost 30 percent of all ventures are located in Amsterdam although the area contributes for only 18 percent to the GDP of the Netherlands. This shows that next to venture capital funds also ventures tend to cluster in the capital Amsterdam.

5.4 Foreign venture capital

In this section the investments made by foreign venture capital are analyzed. First the characteristics of foreign venture capital are presented, including the country of origin and the sector in which investments are made. After that, using the Gini-coefficient a comparison is made of the regional distribution of Dutch and foreign venture capital across the Netherlands. The distribution of venture capital is analyzed because most funds are located in the Randstad which can have as result that ventures outside the Randstad will have difficulties attracting venture capital. For obvious reasons the distances within the Netherlands are not expected to pose much concern towards foreign venture capital, which would result in an more even regional distribution of foreign venture capital investments.

Characteristics of foreign venture capital

Pankaj Ghemawat concludes in ‘World 3.0’ that only 20 percent of venture capital is deployed outside the fund’s home country (Schumpeter, 2011). This seems also to be the case for international venture capital investing in the Netherlands. Out of a total of 772 investments only 9.3 percent are international venture capital investments of which the average value is 36.8 million euro9. As the average investments size of a Dutch venture capital investment is 6.9 million euro this shows that international deals are on average (a lot) larger than Dutch venture capital investments (NVP-Ondernemend Vermogen, 2009).

Table 13 contains variables which describe characteristics of international venture capital investments. Distance seems to have a large influence on foreign venture capital as around half of the international venture capital investments in the Netherlands are made by neighboring countries (the UK, Belgium and Germany). The United States is the country which houses most of the foreign venture capital funds, one third of all the venture capital investments are done by

(44)

44

funds that are located in this country. It seems that investing in the Netherlands is most often a one-off issue as only nine funds made more than one investment and the average number of investments is 1.5.

Foreign venture capital funds prefer to invest in ventures that already have international operations. 85 percent of the companies are either expanding internationally or are part of a multinational operating enterprise. They also prefer to cooperate with Dutch funds as 78 percent of the investments are conducted together with a Dutch venture capital funds.

The sector preference is clearly different for international venture capital. 22 percent of the international investments are made in life science ventures and 24 percent in other ventures. Interestingly, 55 percent of the international venture capital investments are made in ventures operating in the ICT sector compared to 27 percent for Dutch venture capital investments.

Gini-coefficient

Geographical preference in venture capital investments

Geographical preference in venture capital investments

Geographical preference in venture capital investments

Table of contents

I INTRODUCTION

II LITERATURE REVIEW

III METHODOLOGY

IV DATA DESCRIPTION

V RESULTS