Master thesis
Business Economics:
Finance
European Venture Capital investment decisions
and liquidity risk
By: Wouter Floris, 10616004
Date: 3 June 2015
Supervisor: Patrick Tuijp
Contents
I. Introduction ... 1
II. Related Literature ... 4
A. Why VC investment decisions matter ... 4
B. VC investments and stock market liquidity ... 5
III. Methodology ... 8
A. Theory of thesis ... 8
B. Liquidity risk ... 9
C. Overall investments and investment stages ... 9
I.
Hypothesis overall investments and stages ... 9
II.
Regressions overall investments and stages ... 11
D. Extended trade-‐off theory ... 13
E. Combination investment stage and industries ... 15
I.
Identification and classification industries ... 15
II.
Hypothesis combination industries and investment stages ... 17
III.
Regressions combination industries and investment stages ... 18
IV. Data ... 20
A. Independent variables ... 20
B. Dependent variables ... 21
V. Results ... 23
A. Technological Risk: Investment Stage ... 24
B. Combination investment stages and industry ... 28
VI. Conclusion and limitations ... 33
VII. Bibliography ... 35
Appendix I: Industry regressions ... 38
Appendix II: VEIC ... 40
Appendix III: Industry regressions ... 47
I.
Introduction
Entrepreneurs often find it difficult to find financing through the traditional channels such as a bank loan and/or personal savings. Venture Capital firms (VCs) provide financing to these entrepreneurs and small privately-‐held firms in return, usually for a minority share of equity. VCs require a high return on their investments and are therefore quite an expensive provider of financing to an entrepreneur. This source of finance also has its benefits. Zu Knyphausen-Aufsess (2005) identifies in the literature five types of contributions beyond the funds provided: (1) reputation effects from co-‐operation with an established VC firm, (2) stimulation of initial orders, (3) access to distribution channels, (4) Research and development support and (5) industry relationships, which is affirmed by Sapienza et al. (1996).
VC as a source of capital is believed to contribute to the economic growth and competitive strength of the US by promoting the development of innovative start-‐ups. This role of US VCs on technological innovation has been well documented (Hellmann, 2000; Kortum, 2000). In addition there is a consensus among economists, business leaders and policy-‐makers that the vibrant capital industry is one of the cornerstones of the leadership in the commercialization of technological innovation in the US (Bottazzi & Da Rin, 2002). European politics recognize that an improved European VC market is necessary for increasing the EU performance in terms of innovation (Kortum, 2000) and economic growth (Samila & Sorenson, 2011). Policymakers in Brussels and within nations currently try to bridge this gap by stimulating VC investments and its environment.
One of the main characteristics of VCs is that they need to disinvest their position after a limited period of time, generally two to seven years, with the intention to collect a return on their investment. There are several exit strategies available such as a leveraged buyout, acquisition by a strategic buyer or through an Initial Public Offering (IPO) on a stock exchange. IPOs generally provide the highest return (Cumming et al., 2005). The foundation of this thesis lies in this relationship between VC investments and the exit transaction through an IPO.
There is relatively little known in the financial literature about the cyclicality of early-‐stage investment and what the external drivers of these investments are. Understanding more about the choices and drivers behind these investments may contribute to the decision-‐making of policy makers around the globe.
This thesis focuses on the relation between different investment decisions by VCs and stock market conditions, liquidity in particular, in Europe over the period 2000-‐2014. The possibility and profitability of an IPO depends on stock market conditions. Therefore it is natural to expect a
relationship between VC investment decisions and stock market liquidity. Cumming et al. (2005) provide a trade-‐off theory in which VCs alter their investment decisions to stock market liquidity conditions in the United States. This trade-‐off is based on the liquidity risk VCs face in the market and the technological risk of the projects invested in. Liquidity risk refers to the risk of not being able to effectively exit an investment. Technological risk refers to all other types of risk investing in a project of uncertain quality. In this study technological risk refers to the development stage of the project, earlier stages versus later stages. They find evidence that during a year of low liquidity and with an expected low liquidity in the year thereafter (high liquidity risk), VCs tend to invest in projects with low technological risk (later-‐stage projects) and vice versa.
This thesis builds upon this trade-‐off between liquidity risk and technological risk and provides an extended empirical implementation of the theory in the European market. Using a sample of 18 European countries with well-‐developed stock markets over the period 2000 to 2014, I will examine whether this trade-‐off theory holds for Europe as well. This is interesting to test because of the substantial differences of the capital markets in Europe and the US. The ratio between venture capital and private equity investments were estimated in 2009 to be 17% in Europe and 67% in the US and the overall value of the venture capital investments over the GDP is nearly three times higher in the US than in Europe (Grilli, 2014). Besides the difference in size and ratios of the capital markets, Schwienbacher (2005) argues that the main difference lies in the exit market liquidity relevant for VCs. European exit markets tend to be less liquid compared to the US, therefore VCs may respond differently to different levels of liquidity risk. Aside from the geographical difference between their empirical study and this thesis, there are mayor differences in the interpretation of the theory and its scope. This thesis examines whether the level of liquidity risk has a significant effect on the choice of investment projects with varying technological risk profiles in the quarter thereafter. Thereby using different liquidity risk and technological risk measures. I have narrowed the scope of technological risk to a more detailed segmentation of the development stage of the project. I have also extended the definition of technological risk to the particular industry of the project. I argue that different industries have a different risk profile and therefore could be included in the trade-‐off theory as well.
In the first part of my analysis I find that low (high) liquidity risk results in increased investments in seed-‐stage (later-‐stage) projects in the quarter thereafter. These findings are in line with the trade-‐ off theory. However the middle two stages (early-‐stage and expansion-‐stage) behave exactly opposite to what would be expected. These findings required an adjustment in the trade-‐off theory for the second part of the analysis. I found that VCs tend to specialize in either earlier stages (seed-‐
stage and early-‐stage) or later stages (expansion-‐stage and later-‐stage). The extended trade-‐off theory states that VCs opt for low (high) technological risk projects after a period of high (low) liquidity risk within their specialization. Seed-‐stage and expansion-‐stage are considered to be more risky because they have a longer time horizon within the specialization. Therefore VCs which specialize in earlier stages opt for seed-‐stage (early-‐stage) projects after a period of low (high) liquidity risk in high-‐risk (low-‐risk) industries. VCs specialized in later stages choose rationally for expansion-‐stage (later-‐stage) projects after a period of low (high) liquidity risk in high-‐risk (low-‐risk) industries. I find that investments in all investment stages behave according to the extended trade-‐ off theory. However I cannot find any evidence that VCs choose industries with different risk profiles after a period of changed liquidity risk.
The remainder of this thesis proceeds as follows. In Section II I will discuss the related literature of VC investments in general and the determinants of VC investments decisions. Section III is divided in to two parts. The first part provides the basic trade-‐off theory and the methodology of the investment stages. The second part provides the extended trade-‐off theory and the methodology of the combination of investment stages and industries. Section IV provides the data and descriptive statistics. The results are given in Section V. Section VI provides the interpretation of the results, conclusion and limitations.
II.
Related Literature
The decision-‐making of VCs is a relatively unexplored topic in the scientific literature. Most papers examine the determinants of VC investments in an economy. This thesis tries to identify liquidity as a determinant of overall VC investments but the scope is mainly on explaining VC investment choices based on different levels of stock market liquidity. I will first discuss the literature on the importance of understanding VC investments itself before going into more depth. The second part of the literature review discusses the relation between VC investments and stock market conditions/liquidity.
A. Why VC investment decisions matter
First of all I discuss the importance of VC investments for the economy as a whole. The effects of VC on macro-‐economic factors such as innovation, employment and economic growth is well covered in the literature. Kortum and Lerner (2000) studied the influence of VC on patented inventions in the US. They show that increases in VC activity in a particular industry are associated with significantly higher patenting rates. These findings are confirmed by Ueda and Hirukawa (2008) and Tykvova (2000) who find a similar positive relationship in the US and Germany respectively. Rahman et al. (working paper, year unknown) state in their working paper several plausible reasons to explain this relationship. First, aside from the supply of finance VCs often bring in other essential resources, such as legal and marketing expertise, VCs are capable of forging linkages among other networks and organizations such as banks, corporations and other start-‐ups. Second, the reduction of asymmetric information because VCs tend to specialize in a specific sector and therefore may have an advantage in evaluating the business accurately. Besides the relationship between VC activity and innovation in general, Hellman et al. (2000) find that VC-‐backed firms apply more innovative strategies compared to their non-‐VC-‐backed peers. There is however one note to these findings. Some argue that this relationship covers only one side. Hirukawa and Ueda (2011) find in a later study a reversed causality between VC activity and innovation. They find that total production growth is often positively and significantly related with future VC investments and therefore argue that an arrival of new technology increases the demand for VC. It is possible that this relationship works both ways. Increased innovation causes an increase in VC activity and increased VC activity causes more possibilities for innovation within a country.
Through, for instance, innovation it is likely that VC investments increase economic growth and employment. Samila and Sorenson (2011) find that increases in the supply of VC positively affect
firm starts, employment and aggregate income in US metropolitan areas. Their findings imply that VC stimulates the creation of more firms through two mechanisms: First, when the supply of capital expands, would-‐be entrepreneurs more commonly start firms. Second, funded companies may transfer know-‐how to their employees, thereby enabling spin-‐offs and encourage other to become entrepreneurs. Engel (2002) find similar evidence on the positive correlation between VC finance and firm performance.
B. VC investments and stock market liquidity
VC clearly has an impact on different economical aspects but what are its key determinants? The literature has identified several possible determinants of VC activity. Black and Gilson’s (1998) were the first to link stock market conditions to VC markets. Their empirical study shows the importance of a well-‐developed stock market for a strong VC market. Stock market-‐centered capital markets, such as the US, tend to have a much stronger VC industry in comparison to bank-‐centered capital markets, because firms that need financing must, by nature, obtain this capital with equity rather than debt. Different papers went further on these findings and examined effects of stock market conditions on VC activity and investment decisions, specifically the effect of stock market liquidity. The rationale behind this relationship lies in the nature of VC strategies. VCs generally have an investment time horizon of 2 to 7 years. In order to derive their returns on investments they need to exit the investment after this period through a so called exit transaction. There are different exit transactions possible for a VC such as selling the equity to a strategic buyer, an initial public offering (IPO) or leveraged buy-‐out. Cumming et al (2005) argue that an IPO provides the largest return. Therefore VCs face the risk of not being able to effectively exit the investment and are either forced to sell their equity at a high discount or hold their position in the firm. Lerner and Schoar (2004) find that this exit risk is one of the key determinants of the high return on investment required by VCs. The ability to exit effectively is therefore dependent on market conditions, hence it is natural to expect that stock market liquidity, thus exit market liquidity, affects VC investments.
In order to determine whether there is a relationship between stock market liquidity and different investment stages and VC activity in general, the literature uses different proxies for stock market liquidity because it is not possible to observe stock market liquidity directly. Most proxies used in the literature are based on the concept of liquidity of traded assets provided by Kyle (1985). Harris (1990) expanded this definition and identifies four different dimensions:
• Width is based on the bid-‐ask spreads. Stock market is liquid when trading small amounts costs little.
• Immediacy refers to the time needed to accomplish a trade.
• Depth refers to the corresponding volumes of the different bid-‐ask spreads which define the depth of the order book. In liquid markets large trades have little impact on the price. • Resiliency refers to the speed with which prices recover after a large transaction. In liquid
markets prices correct quickly and deviate only little from true value.
There are numerous of proxies stock market liquidity such as bid-‐ask spreads, the number or value of IPOs, traded volume on stock markets and market capitalizations. Most commonly used proxy in the VC investment literature are the IPOs. The rationale behind this is that IPO markets are highly cyclical and are characterized by “hot” and “cold” issue phases (Brailsford et al., 2000). The possibility to exit the investment therefore refers to the immediacy dimension of the liquidity concept.
There are several papers which examine the relationship between stock market liquidity and investment activity in general and differences within these investments. Balboa and Martí (2003) examine the determinants that are directly related to the private equity volume of funds raised. They examine market liquidity as one of these determinants because of the disinvestment requirements of VCs. They find that an increase in market liquidity in the previous year has a positive and significant effect on annual future fundraising activity.
Besides the effect of market liquidity on overall private equity it is interesting to examine the effects on different investment characteristics such as investment stages. In general papers distinguish four different VC investment stages, as provided by Thomson One database: seed-‐stage,
early-‐stage, expansion-‐stage and later-‐stage. Jeng and Wells (2000) examined the relationship
between liquidity and VC investments stages using IPOs as a liquidity proxy. They find evidence that the market value of IPOs is the most important determinant of European VC investments in later stages of the firms. Shertler (2003) used a European panel to examine the relationship between capitalization of stock markets and VC investments. In contrast to Jeng and Well (2000) he finds that investments in earlier stages positively depend on the capitalization of stock markets but not on later stages. Therefore it is interesting to see whether my proxy for liquidity has a significant effect on the different VC investment stages in Europe in the period thereafter.
The most important work and the main motivation for this thesis is based on the paper of Cumming et al. (2005). They provide a theory which links VC investment decisions to stock market liquidity. Their theory is based on the risk VCs face when investing. They distinguish two types of risk: liquidity
all other types of risk VC face when investing in a project of uncertain quality. They introduce a theoretical model that shows that VCs will rationally trade-‐off liquidity risk against technological risk by investing more in early-‐stage projects when the liquidity of exit markets is low and thus the exit risk is high. The rationale behind this theory is that VCs alter their portfolio of investments according to the exposure to the liquidity risk. An early-‐stage investment has a greater uncertainty compared to a later-‐stage investment, thus a greater exposure to liquidity risk. Therefore the theory states that VCs reduce their exposure during times of high liquidity risk. Using a US annual dataset they explain differences in VC investment choices in high or low technological risk projects by changes in liquidity risk in the current year and the year thereafter. In particular whether liquidity risk affects the amount of VC investments in early-‐stage firms [high technological risk]. They use the number of IPOs as a proxy for exit market liquidity. Their empirical analysis finds evidence for the theory. US VCs will rationally trade-‐off liquidity risk against technological risk. VCs tend to invest proportionally more in early-‐stage, thus riskier, projects in order to postpone exit requirements in times of high (expected) liquidity risk. In contrast, with low liquidity risk VCs tend to choose projects with a shorter time horizon, hence they rush for an exit by investing in later-‐stage projects.
This thesis builds on the same trade-‐off principle as proposed by Cumming et al. (2005) for the European VC market. However it is important to note that the European and United States capital markets differ substantially. Schwienbacher (2005) argues that although there are numerous similarities between the US and European capital markets, there are also important differences. They show that much of these differences have one common denominator: stock market liquidity. The exit markets relevant for VCs are less liquid in Europe. This forces European venture capitalists to “shop around” for longer periods when trying to sell their shares. Because both capital markets differ substantially it is interesting to see whether European VC investment decisions also change with different levels of market liquidity. The fact that European VCs have to “shop around” longer in order to exit their investments could lead to large differences in investment decisions compared to the US. This could go both ways. Because European exit markets are less liquid one could expect that European VCs tend to attach greater value to the exit market liquidity and “seize the moment” when possible. On the other hand one could argue that VCs in Europe attach less value to the exit market liquidity because it is possible that they depend less on IPOs as an exit strategy. Lerner and Schoar (2003) show that the choice of exit strategies in countries that lack liquid capital markets differ from the US. Therefore it could be reasonable to expect that exit market liquidity is of less influence on VC investment decisions within Europe.
III.
Methodology
This part discusses the theory and methods used in this empirical study. Firstly, I will discuss the theory behind the analysis in part A. Part B provides the method used to estimate the liquidity risk. Part C discusses the hypothesis and regressions of the effect of liquidity risk on overall VC investments and on the different investment stages. The regressions discussed in Part C require an extension of the trade-‐off theory. Part D provides an explanation for this extended trade-‐off theory. Part E discusses the second component of technological risk: the industry of the investments. First I provide the method of identification and classification of the different industries before discussing the hypothesis and regressions concerning both technological risk components.
A.
Theory of thesis
The rationale behind this thesis is based on the same basic trade-‐off theory introduced by Cumming et al. (2005), however the interpretation of this theory and the empirical model differ substantially. The sample period and geography differ and I have extended the trade-‐off theory by including an additional technological risk component besides the investment stage: the industry of the project. Each industry has its own risk level, which is measured as the standard deviation of returns. I will use quarterly data for all variables in order to examine whether liquidity risk affects the technological risk taken by VCs in the period thereafter. The trade-‐off between liquidity risk and technological risk is explained below.
1 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑟𝑖𝑠𝑘 VS 𝑇𝑒𝑐ℎ𝑛𝑜𝑙𝑜𝑔𝑖𝑐𝑎𝑙 𝑟𝑖𝑠𝑘
Liquidity risk is measured using an illiquidity ratio. High illiquidity on the stock markets results in a large liquidity risk for VC investments. The theory states that VCs prefer investments with smaller technological risk after a period of large liquidity risk. The technological risk consists out of two components: (A) the investment stage and (B) industry of the firm invested in. The theory states that VCs alter their investment decision after a period of high liquidity risk towards investments with less technological risk. In order to empirically test this theory one needs to identify and measure the different aspects of this trade-‐off. First, the liquidity risk is determined using a proxy for illiquidity. Second, the technological risk is determined by identifying different investment stages and by calculating the industry returns over the 14-‐year period as a proxy for industry risk.
B.
Liquidity risk
Unfortunately it is not possible to observe stock market liquidity directly. As shown in the literature review there are numerous proxies to choose from. This thesis examines the effect of quarterly liquidity on VC investment decisions based on daily data. Fong et al. (2014) compare different liquidity proxies and find that the daily version of Amihud (2002) liquidity proxy is the best. I will use this measure and is calculated as follows:
2 𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!"=𝐷1 !" |𝑅!"#| 𝑉𝑂𝐿𝐷!"#$ !!" !!!
This measure refers to the liquidity concepts tightness and depth provided by Kyle (1985) and Harris (1990), where 𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!" measures the illiquidity of a stock i in year y, |𝑅!"#| is the absolute value of the return on stock i on day d of year y and 𝑉𝑂𝐿𝐷!"#$ is the respective daily volume traded in EUR millions. 𝐷!" is the number of days for which data are available for stock i in year y.
C.
Overall investments and investment stages
The methods to obtain and measure both components of technological risk are explained below. Thomson One identifies four different stages in VC investments; seed-‐stage, early-‐stage, expansion-‐ stage and later-‐stage. This thesis uses these stages as a proxy for the first component of technological risk. In the seed-‐stage the firm sets up its initial concept, builds prototypes and explores the market. The early-‐stage is one step further in the progress where the production and marketing begins. In the expansion-‐stage and later-‐stage firms sell their services or products but require funding because the internal cash flow is not sufficient to finance the expenses necessary to expand. In the earlier stages the firms generally do not generate cash nor experienced feedback from the market on their services/products. Therefore there is more uncertainty in the earlier development stages of the projects. Pintado et al. (2007) find that a significant higher percentage of Spanish VCs required a higher return on investments in the earlier stages than the later stages of development and vice versa. This is in line with the financial theory that riskier investments should be associated with higher (expected) returns. Therefore the earlier stages are considered to be risky.
I. Hypothesis overall investments and stages
In order to examine whether the trade-‐off theory holds in the European VC market different hypothesis are tested. The hypothesis can be divided into three segments: The first estimates the
effect of liquidity risk (increased stock market illiquidity) on total investments. The second segment focusses on estimating the effect of liquidity risk on different investment stages, the first component of technological risk, and the third segment of the hypothesis estimates the effect of liquidity risk on investments in high-‐risk and low-‐risk industries.
Overall VC investments
The first step in this analysis is to examine whether the VC investments change with different levels of liquidity risk. Balboa and Martí (2003) find a significant increase in overall annual fundraising volume after an increase in stock market liquidity in the previous year. Therefore one might expect that overall VC investments decrease with high liquidity risk in the stock market in the previous period as well. Cumming et al. (2005) find in their full sample of early-‐stage and expansion-‐stage investments a positive effect of IPO volume on the overall propensity to invest in new projects, because of the reduced liquidity risk. Aside from this it is important to analyze the overall effects of liquidity risk on VC investments before analyzing the effects on differences within VC investments [technological risk].
Hypothesis 1: High liquidity risk will decrease the total number and total money amount of VC investments in the period thereafter.
Investment stages
In order to test whether the trade-‐off theory holds I will examine whether liquidity risk affects the first component of technological risk. The effect on the different investment stages is tested with hypothesis 2 to 5. I expect that high liquidity risk in the previous period results in the preference for low technological risk projects in the period thereafter, hence more investments in expansion-‐stage and later-‐stage projects. Conversely I expect that low liquidity risk in the previous period results in the preference for low technological risk projects in the period thereafter, hence more investments in seed-‐stage and early-‐stage projects. Cumming et al. (2005) show that during illiquid stock markets VCs tend to prefer early-‐stage investments. In times of liquid stock markets VCs tend to opt for later-‐ stage investments. The rationale behind this is that in times of illiquid stock markets VCs choose to postpone liquidity requirements by investing in a project with a longer time horizon and vice versa. Jeng and Wells (2000) found evidence that the market value of IPOs explains differences in VC investments in later stages, not the earlier stages. Schoar (2003), on the contrary, finds a positive impact of liquidity of stock markets on early-‐stage investments. These different studies all use different locations, time horizons and proxies. It is therefore interesting to estimate the effect of the
Amihud (2002) illiquidity measure on the different investment stages and test if the trade-‐off theory holds in the European VC market.
Hypothesis 2: Low liquidity risk increases seed-‐stage investments in the period thereafter (increase technological risk).
Hypothesis 3: Low liquidity risk increases early-‐stage investments in the period thereafter (increase technological risk).
Hypothesis 4: High liquidity risk increases expansion-‐stage investments in the period thereafter (reduce technological risk).
Hypothesis 5: High liquidity risk increases later-‐stage investments in the period thereafter (reduce technological risk).
II. Regressions overall investments and stages
The five different hypotheses are tested using OLS regressions. Table A on the next page provides a description of the tested variables and the independent variables. Each dependent variable has four different testable versions and thus four different regressions. The first version is the logarithm of one plus the absolute number of investments in that period. This is a conventional method of testing the number and amount of VC investments (Gompers et al. 2006). The second version is the
proportion of all investments made in the given period. The third version is the logarithm of one plus
the absolute money amount invested in the given period. The fourth version is the money proportion invested of all investments in that period. The logarithm is used to measure the overall effect. The proportion variable is the proportion of the particular investment stage to all stages investments. This is used to control for overall effects of the independent variables and as a robustness check. Appendix I: Descriptive Statistics provides an overview and descriptive statistics of each tested and independent variable. The dependent variables each have four different versions thus four different regressions as given below:
3 𝐿𝑛_𝑁𝑢𝑚_𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡!= 𝛼 + 𝛽!𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!!! + 𝛽!𝑊𝐺𝐷𝑃!!! + 𝛽!𝑆𝑇𝑂𝑋𝑋!!! + 𝜀!
(4) 𝑃𝑟𝑜𝑝_𝑁𝑢𝑚_𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡!= 𝛼 + 𝛽!𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!!! + 𝛽!𝑊𝐺𝐷𝑃!!! + 𝛽!𝑆𝑇𝑂𝑋𝑋!!! + 𝜀!
5 𝐿𝑛_𝐸𝑈𝑅_𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡!= 𝛼 + 𝛽!𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!!! + 𝛽!𝑊𝐺𝐷𝑃!!! + 𝛽!𝑆𝑇𝑂𝑋𝑋!!! + 𝜀!
(6) 𝑃𝑟𝑜𝑝_𝐸𝑢𝑟_𝐷𝑒𝑝𝑒𝑛𝑑𝑒𝑛𝑡!= 𝛼 + 𝛽!𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!!! + 𝛽!𝑊𝐺𝐷𝑃!!! + 𝛽!𝑆𝑇𝑂𝑋𝑋!!! + 𝜀!
𝐴𝑚_𝑖𝑙𝑙𝑖𝑞!!! is the illiquidity ratio in the quarter before, 𝑊𝐺𝐷𝑃!!! is the weighted average GDP growth
in percentages of the European countries included in the sample in the quarter before. This variable is included to control for general economic conditions. 𝑆𝑇𝑂𝑋𝑋!!! is the percentages growth of the
STOXX Europe 600 in the period before. This variable is included in order to control for general market effects, 𝛼 measures the intercept and 𝜀! is the error term.
There are two regressions on the total investments: total absolute number of investments and total money amount of investments and four versions on each of the four different investment stages: absolute number of investments, proportion of investments, money amount invested and money proportion invested, thus 16 regressions. An overview of the variables is given in Table A below.
Table A. Overview variables overall investments and investment stages
Independent
Amihud The Amihud illiquidity ratio as a measure for liquidity risk WGDP Weighted average European GDP growth in percentages STOXX The STOXX Europe 600 market index growth in percentages
Dependent
Ln_Num_Investments Total number of VC investments in that quarter
EUR_Num_Investments Total money amount invested in a given quarter in Euros
Ln_Num_[Stage] The logarithm of the number of investments in the particular investment stage
Prop_Num_[Stage] The number of investments in the particular investment stage as a proportion of total number of investments in that period
Ln_EUR_[Stage] The logarithm of the money amount of investments in that particular investment stage Prop_Eur_[Stage] The money amount of investments in that particular investment stage as a proportion of
the whole money amount invested in that period
D.
Extended trade-‐off theory
After the first part of my analysis I found that the outermost two stages (seed-‐stage and later-‐stage) behaved according to the trade-‐off theory. The middle two stages (early-‐stage and expansion-‐stage) behaved exactly opposite. These findings had a significant effect on my thesis because half of the found effects were without explanation. The other studies mentioned in the literature review such as Cumming et al. (2005) and Schertler (2003) did not include all four investment stages in their analysis and therefore I had to go back to the literature to find an explanation for these anomalies.
After some research I found that a possible explanation of the effects may lie in the fact that VCs tend to specialize. Robinson (1987) was one of the first to find that VC firms differ significantly in the stage in which they invest. These findings are consistent with other papers such as Elango et al. (1995). They show that VCs specialize in earlier stages or in later stages, where VC firms in earlier stages tend to be smaller compared to the larger VC firms specialized in later stages. They show that earlier stages VCs are interested in unique, proprietary products with high growth potential. Later stages VCs require investments with market-‐proven products/services.
I extended the trade-‐off theory according to these findings. It is possible that specialized VCs, in either earlier stages (seed-‐stage and early-‐stage) or later stages (expansion-‐stage and later-‐ stage) alter their investment decisions according to liquidity risk, but only within their specialization. VCs specialized in earlier stages will rush for liquidity after a period of high liquidity risk by investing in the less risky option of the two; the early-‐stage. After a period of low liquidity risk they will choose rationally for the riskier option of the two; the seed-‐stage. The VCs specialized in later stages will do the same within their two investment stage options and rationally opt for projects in the later-‐stage after a period of high liquidity risk. After a period of low liquidity risk they will opt for the option with a higher risk: the expansion-‐stage. To summarize: the specialized VCs increase (decrease) the exposure to technological risk after a period of low (high) liquidity risk. This extended trade-‐off theory explains all the findings on investment stages. The next section also includes the second component of technological risk in the theory. The extended theory of specialized VCs expects the following trade-‐off between liquidity risk and technological risk:
7 High 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑟𝑖𝑠𝑘!= Decrease investments in 𝑆𝑒𝑒𝑑 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐻𝑖𝑔ℎ − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
Incease investments in 𝐸𝑎𝑟𝑙𝑦 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐿𝑜𝑤 − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
8 Low 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑟𝑖𝑠𝑘!= Increase investments in 𝑆𝑒𝑒𝑑 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐻𝑖𝑔ℎ − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
Decrease investments in 𝐸𝑎𝑟𝑙𝑦 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐿𝑜𝑤 − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
A similar trade-‐off is expected for VCs specialized in the later stages:
9 High 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑟𝑖𝑠𝑘!= Decrease investments in 𝐸𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐻𝑖𝑔ℎ − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
Increase investments in 𝐿𝑎𝑡𝑒𝑟 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐿𝑜𝑤 − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
10 Low 𝐿𝑖𝑞𝑢𝑖𝑑𝑖𝑡𝑦 𝑟𝑖𝑠𝑘!= Increase investments in 𝐸𝑥𝑝𝑎𝑛𝑠𝑖𝑜𝑛 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐻𝑖𝑔ℎ − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
Decrease investments in 𝐿𝑎𝑡𝑒𝑟 − 𝑠𝑡𝑎𝑔𝑒!!! in 𝐿𝑜𝑤 − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠!!!
Where 𝐻𝑖𝑔ℎ − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠 consist of the identified industries with an above average standard deviation in returns and 𝐿𝑜𝑤 − 𝑟𝑖𝑠𝑘 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠 below average standard deviation in returns over the
sample period 2000 – 2014, as shown in Table B in the following section.
E
.
Combination investment stage and industries
The effects of the specialized VCs on investment stages are significant. In the second part of my analysis I examine whether this extended theory holds for the combination of both technological risk components: investment stage and industry. First I shall explain the method used in identifying and classifying the different industries before providing the hypothesis and regressions in section II and III respectively.
I. Identification and classification industries
The second component of technological risk is the industry of the particular project invested in. The VC investments dataset used is obtained from Thomson One, which uses the Venture Economics Identification Codes (VEIC) as a sector identifier. VEIC identifies firms on three levels: 9 broad industries (Level 1), 86 subsectors (Level 2) and 271 specific branches (Level 3). For a detailed overview please refer to Appendix II: VEIC. This thesis distinguishes the investments on the following level 1 industries:
1 Communications 2 Computer related 3 Other electronics related
4 Genetic engineering/Molecular biology 5 Medical/Health related
6 Energy
7 Consumer related 8 Industrial products
9 Other (transport, financial sector etc.)
It is likely that each industry has its own risk profile. This thesis calculates and measures the technological risk for each industry with the volatility in returns. Each of the nine industries is classified with high or low-‐risk using the quarterly standard deviation of daily returns of the industry indices. The STOXX Europe Industry indices are used to calculate the return of each VEIC industry. However the VC investments dataset consists of VEIC codes and stock market industry indices, such as STOXX Europe Industry indices, are based on the Industry Classification Benchmark (ICB) codes instead. ICB identifies firms on four levels; 10 industries (Level 1), 19 supersectors (Level 2), 41 sectors (Level 3) and 114 subsectors (Level 4). There are no perfect identical matches between ICB and VEIC codes. Therefore I had to match them by hand, as shown in Table B. In some cases the VEIC industry (Level 1) corresponds roughly to an ICB industry (Level 1) code. This does not hold for four industries. In these cases the ICB (Level 1) code was broader than the VEIC (Level 1). Therefore I have merged these VEIC industries to correspond to the ICB industries. ICB industry Technology includes
VEIC industries Computer Related and Other Electronics Related. VEIC industries Genetic Engineering/Molecular Biology and Medical/Health related are merged to correspond to ICB industry Health Care.
Table B: Composition and risk measure of each identified industry
Identified Industry: VEIC (Level 1): ICB Industry (Level 1): Average Std. Dev. 1 1. Communications 1. Telecommunications (6000) 2.22
2 2. Computer related 3. Other electronics related
2. Technology (9000) 1.43
3 4. Genetic engineering/Molecular biology
5. Medical/Health related
3. Health Care (4000) 2.79
4 6. Energy 4. Oil & Gas (0001) 4.80
5 7. Consumer related 5. Consumer Services (5000) 6. Consumer Goods (3000)
4.46
6 8. Industrial products 7. Basic materials (1000) 8. Industrials (2000)
6.60
7 9. Other 9. Utilities (7000) 10. Financials (9000)
4.17 Note: VEIC are the identifiers used in Venture Capital database retrieved from Thomson One. ICB are the codes used in the industry stock market indices. These industries do not correspond in some cases. This table provides the composition of the identified industries used in the thesis. The average standard deviation is calculated as the average quarterly standard deviation from each STOXX Europe industry index based on daily stock price deviations. In the cases of identified industry 5, 6 and 7 the industry consists of two STOXX Europe industry indices and are therefore calculated as the average of both average standard deviations.
The risk profile per ICB industry is calculated using the average quarterly standard deviations of the daily STOXX Europe 600 industry (STOXX industry) return index. Industry 1, 2, 3 and 4 correspond directly to a STOXX industry index. Industry 5, 6 and 7 require an extra step in the calculation because they contain two STOXX industry indices. For these industries I have calculated the average of the quarterly standard deviations of both STOXX industry indices. These averages in standard deviation of a particular identified industry is used as a proxy for the risk as shown in the fourth column of Table B. The industries with above average standard deviation are considered to be high-‐
risk. Industries with below average standard deviations are considered to be low-‐risk. The identified
industries are grouped in the low-‐risk and high-‐risk categories. The low-‐risk industry group consists of identified industry 1, 2, 3 and 4. The high-‐risk industry group consist out of identified industry 5, 6 and 7.
II. Hypothesis combination industries and investment stages
In the trade-‐off theory a high liquidity risk will result in investments in low technological risk in the period thereafter and vice versa. Cumming et al (2005) used the early-‐stage and expansion-‐stage investments as a proxy for high and low technological risk respectively. As explained in the theory technological risk includes all other types of risk when investing in a project of uncertain quality. It is interesting to examine whether the trade-‐off theory holds for the industry of the project as well. I expect that VCs specialized in earlier stages adjust their investment decisions according to the liquidity risk they face and reduce investments with high technological risk after a period of high liquidity risk. Thereby choosing early-‐stage projects as the less risky option of the two in a low-‐risk industry. Conversely I expect the specialized VCs to opt for seed-‐stage projects in high-‐risk industries after a period of low liquidity risk.
Hypothesis 6: Low liquidity risk increases investments in seed-‐stage projects in high-‐risk industries in the period thereafter (increase technological risk).
Hypothesis 7: High liquidity risk increases investments in early-‐stage investments in low-‐risk industries in the period thereafter (reduce technological risk).
I expect the same relation for the VCs specialized in later stages, which is tested with the following hypothesis:
Hypothesis 8: Low liquidity risk increases investments in expansion-‐stage investments in high-‐risk industries in the period thereafter (increase technological risk).
Hypothesis 9: High liquidity risk increases investments in later-‐stage investments in low-‐risk industries in the period thereafter (decrease technological risk).
