Are Venture Capital Funds Changing Appetite? : Worldwide Evidence. Master Thesis MSc Business Administration

(1)

Master Thesis

MSc Business Administration

Are Venture Capital Funds Changing

Appetite? : Worldwide Evidence.

By:

Bagus Angga Wirajaya

Phone: (+31) 618557174

E-mail: angga81.ba@gmail.com/b.a.wirajaya@student.rug.nl

Student Number: S2934442

Supervisor: Dr. Samuele Murtinu.

Co Assessor: dr. Florian Noseleit.

(2)

1

Abstract

Venture Capital (VC) has been one of the most important choices for startups in terms of external financing, mainly for its financial and knowledge support. There are 2 types of VC based on their preference of industry to invest, which are generalist and specialist VC. Acknowledging past researches, the majority of them are focusing on how specialist VCs have better performance compared to generalist VC. Meanwhile, there is little focus on research regarding the changes in the preference, namely from being a specialist to generalist or vice versa. This is important to look into because as we have seen from several decades, different industry tops the most invested industry rank and it is interesting to see whether VC stays specialist once the industry it specialized into is no longer in the top of the rank or even considered as less favorable to invest in anymore just for the sake of better performance. Through this research, we investigate four factors that we deem to affect the VC decision to switch its preference. Overall, the result shows that as time passes, VC becomes more generalized in their investment portfolio, and certain preference of stage also impacts the decision to change their approach in investing. Last but not least, we also find a different method of measuring the degree of specialization, which shows how sensitive the data is compared to previous research.

(3)

2

Introduction

Venture capital (VC) is one of the options for equity financing, and it has become increasingly popular as the choice of external financing instrument (Bergemann & Hege, 1998; Gupta & Sapienza, 1992; Gompers et al., 2009). Some particular reasons for this are because companies backed by venture capital tend to have better innovation, in terms of research and development (R&D) as well as the number of patents, and shorter time of introducing a product to the market (Hellman & Puri, 2000). Apart from that, VC-backed companies also receive monitoring and supervision from the VC, which during this process VC provides the companies with the necessary knowledge based on the expertise of the VC (Gupta & Sapienza, 1992; Davila et al., 2003). There is also an impact on the economic growth as well, mainly in terms of increase in the creation of new companies, expansion number of jobs and increase in income (Samila & Sorenson, 2009). Based on this evidence, it is apparent that VCs, to a certain extent, have a positive impact on the development of startup companies. And in addition to that, the economic impact is also felt to the country’s economic development, since VC also helps to create companies that eventually increases employment opportunities.

VC raise capital by organizing what is called a fund. The fund consists of capital acquired from two parties which are General Partners (GP) and Limited Partners (LP). The LP usually consist of external investors such as banks and business angels, and the partnership with LP has an average term of ten years (Lerner & Schoar, 2004). Meanwhile, the GP itself is the VC that organizes the raising of the funds (Andrieu & Groh, 2017). GP will then invest the jointly raised capital from LP to designated companies (Andrieu & Groh, 2017). The ten years partnership with LP is divided into 2 stages, which are investment and harvesting stages. At the investment stage, the LP provide the agreed amount of capital, and the GP will invest the capital towards the companies (Andrieu & Groh, 2017). Around 7 to 10 years after the investment was made (initial closing), GP will divest the investment in which GP will liquidate the investment it made using the capital and distributed the return to the LP based on the agreed basis (Andrieu & Groh, 2017).

(6)

5

software industry remains the top industry to invest in (KPMG, 2018). Commercial goods, consumer services, and healthcare devices and systems sectors are also interesting to invest in as well; each raised from 10% to 15% from the total of VC investment in all industry constantly in the last 5 years (KPMG, 2018). With this in mind, there is always a debatable topic of whether to invest in the specific industry only or try to diversify the portfolio into the different industry as well, since the performance of each industry might differ from one another and might also differ every year.

Previous academic research attempted to address aforementioned issue, and distinguished two types of VC based on the preference of industry; generalists or specialists VC (Norton & Tenenbaum, 1993; Gompers et al., 2005; Gompers et al., 2009; Andrieu & Groh, 2018). In terms of definition, generalist VCs are those firms that diversify their portfolios in different types of industries, meanwhile specialists VC are firms that take an interest only in particular industry (Gompers et al., 2009). Several reasons why some VC prefers to be specialized are the organizational form of the VC funds and also the expertise that the VC has over certain criteria of investment (Andrieu & Groh, 2018).

Previous empirical research has focused on differentiating generalist and specialist VC not only on which industries they focus on but also regarding the performance of the investment measured by the return of the investment that VC makes between investing in different industries (generalist) or investing in the particular industry only (specialist). Generalist VC which invest in different types of industries was predicted to be better at allocating capital across industries (Stein, 1977). This is quite straightforward since generalist VC would have the insight into different industries that could be used as a consideration in where to invest. However, recent finding stated that specialist VC tends to outperform the generalist VC due to their greater average success of investment (Gompers et al., 2009). In the same research as well, it is found that as VC progresses over time, there is a tendency of shifting in approach as well, indicating that changing in structure of approach is quite possible (Gompers et al., 2009).

(7)

6

investment takes place (Sorenson & Stuart, 2001), and the performance of the investment at its liquidation stage (Hull, 2018). These are endogenous factors, or factors from inside the VC industry that can also explain the industry preference of VC (Gompers et al., 2009). These factors have been overlooked in the past research regarding VC’s industry preference. It is imperative to note that VC decision to invest in certain industry still majorly depends from within the VC itself and not solely from the factor such as the tempting return that VC may gain when specializing or diversifying its investment (Gompers et al., 2009). Aside of that, another interesting aspect to see is that with the ever-changing trend of investment in the industries every year, it might be possible to witness several VC making an investment out of its initial preference, or it would still invest in the same industry. Therefore, the purpose of this research is to answer which factors that influence the switching preferences of VC from generalist to specialist investment approach, or vice versa and to analyze the impact of some endogenous factors that might influence VC industry preference; hence the research question

“What factors determine the switching approach of generalist to specialist Venture Capital (VC)?”, “To what extent do these factors contribute for VC in choosing which approach to pursue?”.

The aim of the research is to provide further insight regarding the impact of these factors on how VC determine their industry preference. One aspect that needs to be taken into account by VC is that the trend of investment does not move in tandem, which means that we could expect that there would be a different amount of investment in different sectors over time (Cooney, 2005). Therefore, we could expect that there would be some changes that VC make in their industry preference. Moreover, this study also provides information on what factors VC can assess into when deciding to invest in a particular industry, apart from assessing the performance of the investment only.

Theoretical Framework

Generalist vs. specialist

(8)

7

(Gompers et al., 2009). Although this preferences prone to create higher risk due to investing in different industries, using generalist approach means that the firm may also hedge from the risk that it bears from losing profit in one industry, by making a profit in another (Stein, 1997; Gompers et al., 2009). In terms of performance, there are numerous findings that explained how VC that focus on specialization approach would excel, in terms of performance, compared to a generalist approach. By becoming a specialist in a certain industry, it also benefits from the effect of the knowledge that it has regarding a certain industry, therefore adding more value to its investment (i.e. making the precise decision on where to invest) (Gompers et al., 2009). Gompers et al. (2005) findings confirmed this statement, in which VC investment firms that were specialized in certain industries tended to have a greater profitable exit (i.e., IPO). These findings suggest that up to present time being, a specialist VC has some advantages on its own, but the question is would those become enough findings to justify for upcoming VC in the future to directly follow specialization approach?

To answer this question, there are some factors that could affect the VC in choosing the approach. Based on previous findings, the age of the firm impacts the preference of the approach, in which as a VC firm mature, its investment will eventually become more generalist (Gompers et al., 2009). Preference towards certain stage is also an important factor, as scholars found that preference towards an earlier stage of company’s development (e.g., seed and early stage) is correlated with a preference towards narrow types of industry (Gupta & Sapienza, 1992). In addition to stage preference, geographic proximity is also important as it allows ease of monitoring for VC which prefer to invest in a much riskier early stage company (Gupta & Sapienza, 1992). Therefore we will test these factors and analyze the impact of these factors toward the industry preference of VC. A more extensive explanation regarding these factors and the relation with a preference towards a certain approach will be discussed in the next section. Throughout this research, we will refer to this industry preference of VC as a degree of specialization, which is the extent of how specialized or generalized a VC when choosing an industry to invest in.

VC Firms’ Age and Degree of Specialization

(9)

8

industry, both in firm-level or individual level (Gompers et al., 2009). Based on their result, it showed that firms, which have operated for 10 years or more, would have lesser Herfindahl index compared to those that just perform at 3 years at least (Gompers et al., 2009).

Although Gompers (2009) did not specifically focus on the VC firms’ expertise regarding the choice of the approach and also did not explain further regarding this specific result, another scholar apparently provided the reason behind it. As VC firms gain experience through the investment they made, their preference towards investing in specific early stage development companies decrease (Dimov & Murray, 2008). As VC firms mature through the expertise they gained from investing in various industries, they started to develop and behave in a similar context like a private equity firm; which is a similar equity investment firm like VC but has more diversified industry portfolios due to its preference of investing in a more developed companies compared to VC. (Dimov & Murray, 2008).

VC Stage Preference and Degree of Specialization

VC firms also specialize in terms of stage preference (Gupta & Sapienza, 1992). There are numerous types of funding stage, but the most common are the seed funding stage, early stage, later stage, mezzanine stage, and buyout stage (Gompers et al., 2009; Gupta & Sapienza, 1992; Slbernagel & Vaitkunas, 2012; Ljungqvist et al., 2007). Seed stage and early stage invest at the state where the investee is still in its earlier development stage (developing product prototype, making blueprints, etc.), which might bear high risk in itself (Ruhnka & Young, 1991). Later stage, mezzanine and buyout stage focus on the already developed companies, however, VC funds investing in a buyout stage is intended to acquire an already developed private company just before it goes to become public (i.e. through IPO) whereas the others are more about investing with the expectation of substantial return for the investment (Ljungqvist et al., 2007, Jeng & Wells, 2000). Preference of seed and early-stage funds is typically related to narrow focus of industry choice, meanwhile, the later, buyout and mezzanine stage prefer to invest in the diversified industry (Norton & Tenenbaum, 1993; Silbernagel & Vaitkunas, 2012; Ljungqvist et al., 2007).

(10)

9

the “coaching” term a quite common role is for certain VC (Baum & Silvermann, 2004; Berger, 2018). And since the characteristic of such information is industry specific, therefore Gupta and Sapienza (1992) argued that VCs that preferred to invest in an early stage of development tend to prefer industries that are in accordance to the industry-specific information that VCs have. The findings of this specific test were positive, indicating that VCs which prefer to invest in an early stage of development tend to prefer investing in the less diversified industry, in other words, it has a specialized approach in choosing the industry (Gupta & Sapienza, 1992; Lin, 2016).

Based on this premise, we could, at certain point, argue that with the current condition in the global market, there would be a point where a change of preference towards industry could happen. Currently, the late-stage development market seems to have been in a dramatic inflation. This is marked by the amount of investment raised at this particular stage, which is almost double the amount of the investment raised at early stage ($11.5 million compared to $5.5 million) (KPMG, 2018). There are two assumptions behind this dramatic increase, first due to its security reason for investing in companies that are ready to commercialize, and the second is because this is the period where early stage investment made in the previous period (2015 and before) has reached its reversion state, meaning that some of the companies that survive up to this year are able to exit the investment by going to public or acquired via buyouts (KPMG, 2018). Therefore, based on the previous proposition from Gupta and Sapienza (1992) and with the current condition in the global market, we could expect that there will be some change in industry preference, specifically preference to a more diversified industry in VC due to the late development stage force from the market.

VC Location Preference and Degree of Specialization

(11)

10

VC firms will only invest in companies and area that they have experience in (Gupta & Sapienza, 1992, Amit et al., 1990, Sorensen & Stuart, 2001). However, there are not many researchers studying the impact of preference towards later stage development and the choice of the geographic location investment. Another plausible reason to explain this is due to the monitoring and supervising problem by the VC firms. Since VC firms do not only invest in capital aid but also provide assistance and “mentoring” for the development of the investee, supervising and monitoring the activity of the investee are necessity for VC firms, therefore it would be convenient to invest within the proximity of the VC firms especially for those who invest in the earlier stage development (Gupta & Sapienza, 1992; Sorenson & Stuart, 2001). Assistance from VC firms must be offered directly to the investee management in order to gain effective and efficient result for the development of the company itself (Sorensen & Stuart, 2001).

Past researches have been focusing on the investment in specific countries and region such as the United States and Europe. Little is known regarding the activity of investment in areas such as Asia, Africa, and Pacific region, meanwhile, in the past five years, the investment activity in these regions have been promising (KPMG, 2018). Take an example in India, in 2018 as much as $7 billion is raised through VC investment (KPMG, 2018). 26 new unicorns are founded in Asia, which is two times the amount of unicorns founded in Europe from 2017 to 2018 (KPMG, 2018). With this statistic, it is very tempting for new and incumbent VC to explore the other region in order to experience this fast-paced development that has been going on in the other part of the world.

IPO Performance of VC-backed Companies and Degree of Specialization

(12)

11

of options to exit an investment, IPO is still favorable for companies because it encompasses some benefits toward the traits that are very useful for them.

VC firms who invest in industries that are outside of its specialization are typically expected to yield less successful exit compared to the industry in where they specialized (Hull, 2018). In terms of numbers, as much as a 20% likelihood of failure in investment is expected when a firm is investing outside its industry preference (Hull, 2018). The underlying motive behind this result is that due to the extensive cost and the higher risk embedded in diversifying the investment portfolio, VC firms are suggested to keep track on specific industry only to maintain higher likelihood of making a successful investment (Hull, 2018). In regard to specialization and exit performance, VC firms who specialize in choosing which industry to invest tend to have better IPO performance (Hull, 2018). However there are also numerous VC firms diversifying their industry preference not only to hedge the uncertainty but also to gain experience in other industries outside the ones that they have invested in, because in the end, accumulative experience of investing in other industries would allow them to understand the industry even better and potentially could yield better performance of IPO as well (Matsusaka, 2001; Hull, 2018).

(13)

12

Based on the 4 factors that have been described, below is the conceptual model used in this research:

(14)

13

Research Method

Data Collection

The method that is going to be used in this research is a quantitative method. The type of the data is secondary data, which is sourced from Eikon, and IPO database provided by Jay Ritter (2018) on his website. Eikon provides the data regarding the fund’s activity, such as where the funds are being invested and at which stage the funds are invested. There is also information regarding the number of companies, deals, and equity raised for the funds to be invested in the designated companies. Jay Ritter (2018) IPO database provides the IPO data from 1980 to 2018 with several measurement items, such as proceed in the investment which entails of how much the dollar raised during the IPO, price-to-sales ratio of opening price and also market price, median sales of the share along with the adjustment of rates from 2014, age of the company when going for IPO, and the profit percentage that company raised for the return.

The object of the research is specifically ventures capital deals and the observations are regarding the stage at which the funds invested in, the location of the first investment that the fund made as a base identifier of the geographic location, fund size to represent a proxy showing the size of VC firms, the year of the first investment made by the funds, along with information regarding the industry focus of the funds which are determined by Eikon with the threshold of 60% of the total funds. Apart from using the focus of the funds provided by the Eikon, we will also determine the degree of funds specialization using Herfindahl Index (HHI) which are used previously by Gompers (2009) in his papers regarding the successful performance of specialized VC firms. The VC deals that we are looking into are from the period between 1968 and 2018, as the earliest data regarding VC investment that Eikon has recorded is 1968. The measurement for each variable will be discussed in detail in the following section.

Variable Measurement

VC Funds Expertise and Degree of Specialization

(15)

14

investment in the industry. As an example of the result from this index, if a VC fund yields 1 as its HHI index, the fund is investing in 1 particular industry only (100%2=1), and the lower (closer to 0) the result is, the more likely that a VC fund is pursuing a generalist approach. After determining the HHI of each VC fund, we will define the pattern of the HHI through time and see whether the HHI is decreasing towards zero (indicating preference to generalist) or increasing towards one (indicating a preference to a specialist).

VC Funds Stage Preference and Degree of Specialization

To measure the stage preference of a fund, we will use the measurement provided by Eikon, which is the stage at which the funds invested in. There are numerous types of stages defined by Eikon, but there are five specific stages that we are focusing on to measure this variable. The 6 stages are seed stage, early stage, later stage, mezzanine stage, buyout stage, and generalist stage. The generalist stage is defined as the investment made in different stages of development, which is included as a representative of a particular fund that has no preference for certain types of stage. These stages will then be analyzed with the HHI index to see the correlation between the two. Since stage preference is regarded as string data in STATA which will eventually unable to be included in the regression and probit regression, we changed the type of the variable into dummy, by assigning 1 for specific stage and 0 for the rest, therefore we have 6 dummy variables for each of the stage preference that we would like to focus on.

VC Investment Location and Degree of Specialization

The data regarding the investment location of the fund is collected from Eikon. Eikon defines the fund location as the location where the fund makes its first investment. Instead of using the country as the base of the analysis, we use the region as the base. Eikon divides the world location of the investment into 6 types which are Africa, America, Asia, Europe, Pacific and Not Available (NA). We focus on the first five of the regions in order to give a clear interpretation of the result. Reports such as KPMG (2018) also use this region division to emphasize the location of the investment activity. The location is, similar to stage preference, treated as a dummy variable, assigning 1 for a specific location and 0 for the rest.

IPO Performance and Degree of Specialization

(16)

15

value will also be used to portray the value of the company in the market. All the data will be presented in median due to the format provided by the database, therefore we also adjust the HHI by computing the median of HHI. The year of the IPO data is from 1980 to 2018, therefore we adjust the HHI by measuring the median of HHI only from 1980 to 2018. In the regression test, we will regress the IPO performance to lag behind by one year.

(17)

16

multicollinearity test will also be conducted and we will limit the mean VIF to below 10 to minimize the effect of multicollinearity.

Control variables

In addition to the existing factors, we also include two control variables to the test. The first control variable is VC fund size or also known as assets under management. The data for this control variable is available from Eikon, defined as Fund Size. VC fund size, according to Kaplan and Schoar (2005) is correlated to the record of the VC fund’s performance, in which it gives a signal to companies that big VC funds with measured by its fund size are associated to highly reputed and could possibly be associated with the ability to raise more rounds. Therefore, VC fund size is vis a vis with VC funds experience in terms of the impact towards the specialization because both entail the ‘records’ of VC funds and quite possibly both have some impacts on the degree of specialization.

(18)

17

Results and Discussion

The total population of investment data from 1968 until 2018 that we obtain from Eikon is 29.527 VC funds. However, due to our restriction of stage and location preference, we drop some funds which have stage preference outside of the six stages that we are focusing on and also the funds that do not indicate the location of the investment. After dropping the aforementioned funds, we obtained a total sample of 18.402 VC funds.

Fund Industry Focus

Frequency %

Biotechnology 659 3.6

Communications and Media 97 .5

Computer Hardware 166 .9

Computer Software and

Services 2408 13.1 Consumer Related 278 1.5 Industrial/Energy 332 1.8 Internet Specific 1862 10.1 Medical/Health 643 3.5 NA 11363 61.7 Other Products 466 2.5 Semiconductors/Other Elect. 128 .7 Total 18402 100.0

TABLE 1:FREQUENCY AND PERCENTAGE OF FUND INDUSTRY FOCUS

(19)

18

not that peculiar considering that post dot com era has brought a substantial impact in doing business up to present time.

Fund World Location

Frequency % Africa 129 .7 Americas 10393 56.5 Asia 3004 16.3 Europe 4623 25.1 Pacific 253 1.4 Total ₁₈₄₀₂ _100.0

TABLE 2:FREQUENCY AND PERCENTAGE OF FUND WORLD LOCATION

The table above describes the dispersion of VC funds investment location. It is determined as the location of the investment by the fund takes place. Therefore, this is regarded as the base location of the fund’s preference. Based on the table, America is listed as the most preferred location to invest, which covers about 50% of the total sample. This is also expected, as America is the biggest market for the VC industry in the last 5 years and could possibly in the last decades as well (KPMG, 2018). It is also reported that from 2013 until 2018 the number of closed deals in America contributes up to 60% of the total global deals (KPMG, 2018). The second most preferred location is Europe, which manages to raise up to $24 billion worth investment in 2018, which is an increase from 2017 (KPMG, 2018). However, it is still trailing behind Asia, which has raised up to 30% of the total VC invested in 2018, meanwhile, Europe only contributes 10% of the total global investment (KPMG, 2018). This is due to the numerous unicorns coming from Asia which valued at above $1 billion dollar for a single company, and in 2018 itself there are 4 companies valued at that rate dispersed across Asia in countries such as Indonesia (Tokopedia) and Singapore (Grab).

Fund Stage

(20)

19 Generalist 2614 14.2 Later Stage 2340 12.7 Mezzanine Stage 175 1.0 Seed Stage 2528 13.7 Total 18402 100.0

TABLE 3:FREQUENCY AND PERCENTAGE OF FUND'S STAGE PREFERENCE

This is the distribution of stage preference from the sample. The early stage is the most preferred stage in the sample which covers up to 50% of the total sample. It is also expected as the deals that are specified in this research is VC deals, in which the majority of venture capital deals are related to companies which are still in the early stage of development (Gupta & Sapienza, 1992). Generalist stage funds come second which describe that there are funds that do not have specific stage preference and instead invest in different stages.

Linear Regression Analysis

For the next part we will conduct the linear regression analysis test of VC funds experience towards degree of specialization. We will use the fund year as the independent variable and HHI as the dependent variable. The result shows that as VC fund progresses through year, the industry preference would be leaning to a more specialist approach (B=.003, p=.000). However, we are also using the similar test conducted by Gompers (2009) by dividing the timeline into 4 periods which are the mean HHI of funds which has 0 to 2 years, 3 to 4 years, 5 to 9 years and more than 10 years of experience in investment, to compare if the result would still be the same like the regression. Below is the table of the result from this test:

TABLE 4:FUND HHIMEAN COMPARISON

The table above describes that the degree of specialization of VC funds is decreasing as the funds gain experience in investing. Although, in comparison with Gompers (2009) result, the dispersion of the data is not as widely spread as the result in that specific research. Gompers

(21)

20

(2009) result encompassed a more specialist approach in the early year (0-2 years), which described as the mean of HHI above 0.6, and later on, at later year (>1 0 years) the approach becoming generalist due to the decreasing mean of HHI. The difference might be due to the different subject of the research and also the different amount of sample that is used to test the observation. However, one similarity that this result has with the result of Gompers (2009) is that despite the different subject and sample, the approach is still leaning towards generalist approach since the mean of HHI is closer to 0. To describe the dispersion of the data in a more detailed manner, a normal distribution graph is included in Appendix A.

Multiple Linear Regression Analysis

For the multiple linear regression analysis of model 1, the variables that are included are VC funds stage preference, VC funds location, VC funds size, VC funds year and HHI as the measurement for the dependent variable which is the degree of VC fund specialization. We drop several measurements of the variable due to high correlation result from the multicollinearity test (Generalist Stage and Africa), and the result of the multicollinearity test can be seen in Appendix B. Early (B=.147, t=17.73, p=.000) and Seed stage (B=.188, t=17.46, p=.000) preference are positively correlated with the specialization approach due to the positive sign. Later stage, however, showing a positive correlation towards specialization approach (B=.200, t=17.85, p=.000). Apart from America (B=-.057, p=.078) and Europe (B=-.079, p=.016) which, to a certain extent, show a significant impact of the degree of specialization leaning towards a more generalized preference of industry, all the other location for VC funds World Location are not significantly impacting the degree of specialization. Fund size is also not significantly impacting the degree of specialization (B=.000, t=1.21, p=.226). The year is positively correlated with the degree of specialization (B=.002, t=9.29, p=.000), however, we are focusing on a different method to explain the relationship between year and degree of specialization. Overall the model is significant but other predictors can be included to better explain the changes in the degree of specialization (R2=.055, F(11,18.390=134.24, p=.000). The result is shown in Table 5.

(22)

21

(1) (2)

VARIABLES Model1 Model2

(23)

22 (0.00218) Constant 0.232*** 0.366*** (0.0356) (0.0872) Observations 18,402 38 R-squared 0.055 0.230

Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

TABLE 5:RESULT OF MULTIPLE REGRESSION ANALYSIS OF MODEL 1 AND MODEL 2

The model is significant (R2=.230, F(3,34)=4.75, p=.007). Sales of IPO shares are positively correlated with specialized approach (B=.004, t=3.03, p=.005). Proceed of IPO is not significantly impacting the degree of specialization (B=-.003, t=-1.73, p=.092). Market value of the shares is not significantly correlated to the degree of specialization (B=.001, t=.61, p=.544). Therefore, we can refer that the more dollar an IPO raised would not impact the preference of the VC funds industry. Predictors are set in a one-year lag time against the dependent variables, so it could capture whether there is a change in degree of specialization in the VC funds industry when the IPO performance of the companies change. The result of the multiple linear regression analysis can be seen in Table 5.

Probit Regression Analysis

Referring back to the result of the multiple regression for the 1st model, we will conduct a probit regression analysis as a comparison towards the regression analysis. The dependent variable is still the degree of specialization, but we modified the measurement by adding the fund industry focus data from Eikon and change the data into a binary variable. The new dependent variable will be measured as 1 if the fund does not have an industry preference and the HHI is below 0.6. The result is shown in table 6.

(24)

23 Earlystage 0.0634** (0.0296) Generalist 0.461*** (0.0380) Laterstage 0.0168 (0.0389) Mezzanine 0.619*** (0.109) o.Seedstage - Africa -0.478*** (0.142) America -0.156* (0.0836) Asia -0.0921 (0.0860) Europe -0.143* (0.0849) o.Pacific - Year(included) - Constant -1.101*** (0.110) Observations 18,402 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

TABLE 6:RESULT OF PROBIT REGRESSION OF MODEL 1

(25)

24

collinearity issue, which is the year 2018, seed stage and Pacific. The result of probit regression is different from linear regression, where in probit regression the result is presenting the probability of the predictors impacting the change in the dependent variable. In general, the variable year is positively related to the probability of choosing the value 1 of the binary dependent variable. Based on the result, all of the fund stage preference showing positive probability towards choosing the generalist approach, however only three items are significant. When a fund invests in buyout stage, there is a 20% probability that the fund is a generalized fund (dy/dx=.191, p=.000). The same condition also applies to generalist stage and mezzanine stage, where funds preferring to invest in these stages have 16% (dy/dx=.160, p>z=.000) and 21% (dy/dx=.215, p>z=.000) probability of becoming generalized funds. For the location, funds located in Africa have 16% (dy/dx=-.166, p>z=.001) decrease in the probability of becoming generalized funds. In terms of the year, from 1968 until 2016 the funds are showing positive probability of becoming generalized funds, only 2017 shows not significant result. However, there is a period, which is from 1990 to 2006, showing a higher probability of choosing the generalized approach. This can be seen from the margins (dy/dx) shown in the result which it indicates above 50% probability of the funds in this period are generalized funds. The highest probability is in 1998, where the probability of the funds are generalized funds is 73,2% (dy/dx=.732, p=.000). Overall the result of the probit regression in comparison towards the multiple linear regression showed similarities especially in the fund stage variable, where funds preferring to invest in buyout stages, generalist stages and mezzanine stages are more related towards being generalized in terms of industry preference. Fund size, on both methods, does not have a significant impact on the degree of specialization of funds.

Logit Regression Analysis

(26)

25

The result, as seen in the Appendix D, shows that for fund stage preference, as VC switch from investing in buyout stage to early development stage such as seed stage and early stage, the probability of becoming generalist decreases to 28% and 24% (dy/dx=-0.281; dy/dx=-0.244). When switching to the later stage, it also shows that the probability of becoming generalist decreases by 21% (dy/dx=-0.212). As expected, there is no significant increase in the probability of becoming generalist when VC switch to Mezzanine stage, which means it stays as a generalist. As VC switch from preferring buyout stage to generalist, the probability of becoming generalist also decreases to 5% (dy/dx=-0.050). For the preference of world location to invest, if a VC switch from Africa to another location, the probability of becoming generalist will increase regardless of the location, which can be seen in the result where all the margin (dy/dx) of the location showed a positive sign, meaning that there is an increase in probability of having generalist preference if the VC fund changes the location from Africa. Therefore, we can conclude that for the logit test, stage preference and location preference are important factors that could determine the switching of the VC approach.

(27)

26

For the IPO performance of the investee, only sales and proceed of IPO have a significant impact on the degree of specialization. Sales indicate that the more shares are sold during the IPO on a specific year, the more likely that the fund is going to become specialist indicated from the increase of HHI as well. The more proceed that the IPO raises in a specific year, VC will be inclined to become more generalist. Overall, IPO performance has some impacts on the degree of specialization of VC, and with more added items to this factor, we can explain the impact of this particular factor in more detail.

(28)

27

Conclusion

Based on this, we can conclude that aside from the performance factors that are usually used to differentiate VC industry preference approach, the factors that we use as predictors in this research can be used in the same manner as well. Certain stage preferences have certain impacts towards the degree of specialization of VC funds, in a sense that preference towards early stage is related to specialist approach, while, to some extent, preference toward later stage is related to generalist approach, particularly in mezzanine stage and buyout stage. VC funds experience, measured in terms of the year, also has an impact towards VC funds industry preference as well, and according to the result of our test, the preference is leaning to a more generalized industry. Fund size does not have a significant impact on the tests. For location preference, VC funds investing in America and Europe, to a certain degree, are correlated to having a diversified preference of industry. Meanwhile, for Africa, it has a slight probability that VC fund investment made there is specialized in a particular industry only (AVCA, 2018).

Aside from finding the contribution of each factor toward the industry preference of VC funds, we also find there are certain factors that could impact the industry preference of a VC fund in such a way that it can change the preference of the VC fund itself (i.e., from generalist to specialist or vice versa). First is the fund stage preference, in we find that if a VC fund changes its preference from investing in companies at a later stage of its life cycle to investing in a companies at its earlier stage (seed stage or early stage), the probability of the VC fund to become specialist increases, where initially it has a generalist approach in industry preference due to investing in buyout stage. We also find that if a VC fund initially invests in a location with not many VC investment activities, it has a high probability of investing in a diversified industry once it changes the preference of investing in an area with high VC investment activities such as in America or Asia. Therefore, we can conclude based on these two results that the factors that we use in this study to a certain extent explain how VC funds derive their preference towards the certain industry.

Limitation

(29)

28

percentage of the company a VC fund invests in a particular period, we cannot find the exact percentage for it. This makes the computing of HHI a bit difficult since we cannot measure perfectly the percentage of the industry itself. Due to time constraint, manual analysis of industry percentage from each fund is not possible.

The next limitation is the measurement of the Herfindahl Index (HHI) itself. Although it is widely used by researchers to investigate the specialization degree of VC investment, it is still lacking in terms of defining specialization based on the number of the index itself. For example, using the formula of HHI, you only achieve HHI of 0.52 by investing 70% of your fund to a specific industry and 10% to other 3 industries. This is an ambiguous result as we cannot specify clearly whether it is a specialized or generalized VC because when we look at the composition of the investment portfolio it clearly put more efforts to invest in one industry but because of the HHI result it is regarded as indecisive upon its preference.

Future Research

For further research in this topic, using additional databases such as Crunchbase could become an additional source to obtain specific information. Since this research focuses more on the endogenous factors, which are factors that impact the changes of the industry coming within the industry itself, future research can focus on analyzing the external factors coming from the companies that the fund invested in. Some of the examples are investment return (ROI) and post-investment performance. These factors can explain the pattern of VC funds investment after the companies that it invested in are growing and whether VC would invest in the same industry as the companies or not.

Implications

The implications of this research are twofold. First, for academic purpose, this research can be used as a reference to specify in measuring the degree of specialization even further. The measurement that we provide in this research could serve as an additional proxy to measure the degree of specialization. When measuring the degree of specialization, using different factors and methods could yield different results, therefore researchers should pay attention in detail regarding the factors that they are going to use. A different approach could also mean different result, as the ones that we analyzed are endogenous factors, exogenous factors might present an entirely different result.

(30)

29

(31)

30

References

Amit, R., Brander, J., & Zott, C. (1998). Why do venture capital firms exist? Theory and Canadian evidence. Journal Business Venturing, 13, 441–466.

Andrieu, G., & Groh, A., P. (2018). Specialist vs Generalist Investor: Trading off support quality, Investment Horizon and Control rights. European Economic Review, 101, 459-478.

AVCA. (2018). Annual African Private Equity Data Tracker. Retrieved from

https://www.avca-africa.org/research-publications/data-reports/2018-annual-african-private-equity-data-tracker/.

Baum, J. A. C, & Silvermann, B. S. (2004). Picking winners or building them? Alliance, intellectual, and human capital as selection criteria in venture financing and performance of biotechnology startups. Journal of Business Venturing, 19, 411-436.

Bergemann, D., & Hege, U. (1998). Venture capital financing, moral hazard, and learning. Journal of Banking & Finance, 22(6-8), 703-735.

Berger, R. (2018). Venture Capital Fueling Innovation and Economic Growth. Retrieved from https://www.rolandberger.com/en/Publications/Venture-Capital-fuelling-innovation-and-economic-growth.html.

Cooney, A. M. (2005). Generalists vs. Industry Specialists –What You See and What You Don’t! Or The Perfect Investment Solution in the Perfect World. Institute for Fiduciary Education.

Davila, A., Foster, G., & Gupta, M. (2003). Venture capital financing and the growth of startup firms. Journal of business venturing, 18(6), 689-708.

(32)

31

Dimov, D., & Murray, G. (2008). Determinants of Incidence and Scale of Seed Capital Investment by Venture Capital Firms. Small Business Economics, 30, 127-152.

Dimov, D., Shepher, D. A., & Sutcliffe, K. M. (2007). Requisite expertise, firm reputation, and status in venture capital investment allocation decisions, Journal of Business Venturing, 22, 481-502.

Eisenhardt, K. M. (1989). Agency theory: An assessment and review. Academy of management

review, 14(1), 57-74.

EVCA, (2005). Pan-European Survey on Performance, EVCA.

Gompers, P., Kovner, A., Lerner, J., & Scharfstein, D. (2005). Venture capital investment cycle: the impact of public markets. NBER Working Paper, 11485.

Gompers, P., Kovner, A., & Lerner, J. (2009). Specialization and Success: Evidence from Venture Capital. Journal of Economics & Management Strategy, 18(3), 817-844.

Gompers, P., & Lerner, J. (2000). Money chasing deals? The impact of fund inflows On private equity valuation. Journal of Financial Economics, 55, 281–325.

Gupta, A. K., & Sapienza, H. J. (1992). Determinants of Venture Capital Firms’ Preferences Regarding the Industry Diversity and Geographic Scope of their Investment. Journal of

Business Venturing, 7, 347-362.

Harper, N., Schneider, A. (2004). Private equity new challenge. The Mckinsey Quarterly, 1–6.

Hellmann, T., & Puri, M. (2000). The interaction between product market and financing strategy: The role of venture capital. The review of financial studies, 13(4), 959-984.

(33)

32

Invest Europe. (2013). 2013 European Private Equity Activity. Retrieved from https://www.investeurope.eu/media/711867/invest-europe-2017-european-private-equity-activity.pdf.

Invest Europe. (2017). 2017 European Private Equity Activity. Retrieved from https://www.investeurope.eu/media/711867/invest-europe-2017-european-private-equity-activity.pdf.

Jeng, L. A., & Wells, P. C. (2000). The determinants of venture capital funding: evidence across countries. Journal of corporate Finance, 6(3), 241-289.

Kaplan, S. N., & Schoar, A. (2005). Private equity performance: Returns, persistence, and capital flows. The journal of finance, 60(4), 1791-1823.

KPMG. (2018). Venture Pulse Q3 2018. Retrieved from https://assets.kpmg/content/dam/kpmg/cn/pdf/en/2018/10/venture-pulse-q3-2018.pdf.

Lerner, J., & Schoar, A. (2004). The Illiquidity Puzzle: Theory and Evidence from Private Equity. Journal of Financial Economics, 72, 3-40.

Lin, T. (2016). Factors influencing Venture Capital Firms preferences regarding industry and

geographic diversification of their investments with the focus on US at the year of 2015.

Retrieved from http://www.scriptiesonline.uba.uva.nl/document/647741.

Ljungqvist, A., Richardson, M., & Wolfenzon, D. (2007). The investment behavior of buyout funds: Theory and evidence. Financial Management.

Maksimovic, V., & Pichler, P. (2001). Technological innovation and initial public offerings. The Review of Financial Studies, 14(2), 459-494.

(34)

33

Nanda, R., Samila, S., & Sorenson, O. (2018). The persistent effect of initial success: Evidence

from venture capital (No. w24887). National Bureau of Economic Research.

Norton, E., & Tenenbaum, B. H. (1993). Specialization versus diversification as a venture capital investment strategy. Journal of Business Venturing, 8(5), 431-442.

Pagano, M., Panetta, F., & Zingales, L. (1998). Why do companies go public? An empirical analysis. The journal of finance, 53(1), 27-64.

Rajan, R.G., Servaes, H., & Zingales L. (2000). The Cost of Diversity: Diversification Discount and Inefficient Investment. Journal of Finance, 55, 35–80.

Ritter, J. (2019, April 9). IPO 2018 Statistics. Retrieved from

https://site.warrington.ufl.edu/ritter/ipo-data/.

Ritter, J. R., & Welch, I. (2002). A review of IPO activity, pricing, and allocations. The journal

of Finance, 57(4), 1795-1828.

Ruhnka, J., & Young, J., M. (1991). Some Hypotheses about Risk in Venture Capital Investing.

Journal of Business Venturing, 6(2), 115-133.

Samila, S., & Sorenson, O. (2011). Venture capital, entrepreneurship, and economic growth. The Review of Economics and Statistics, 93(1), 338-349.

Scharfstein, D. S. (1998). The dark side of internal capital markets II: Evidence from

diversified conglomerates (No. w6352). National Bureau of Economic Research.

Scharfstein, D. S., & Stein, J. C. (2000). The dark side of internal capital markets: Divisional rent‐seeking and inefficient investment. The Journal of Finance, 55(6), 2537-2564.

(35)

34

Stein, J.C. (1997). Internal Capital Markets and the Competition for Corporate Resources.

Journal of Finance, 52, 111–133.

(36)

35

Appendix A

FIGURE 2:NORMAL DISTRIBUTION HISTOGRAM OF FUND HHI

This is the histogram of the regression result of variable Fund Year and HHI as the measurement for degree of specialization, which is conducted as part of supporting information for the first test. The histogram shows the residual is distributed normally throughout each level of HHI, with no outliers detected (more than +2 and less than -2).

(37)

36

TABLE 7:LINEAR REGRESSION ANALYSIS RESULT

Appendix B

Multicollinearity Test of Model 1

Variable VIF 1/VIF America 35.65 0.028047 Europe 27.62 0.036203 Asia 20.37 0.049093 Pacific 2.93 0.341658 Earlystage 2.38 0.85988 Laterstage 1.75 0.881447 Year 1.1 0.907387 Seedstage 1.1 0.907594 Buyouts 1.41 0.936908 FundSize 1.05 0.953596 Mezzaninestage 1.07 0.989218 Mean VIF 8.56

TABLE 8:MULTICOLLINEARITY TEST OF MODEL 1

The table shows the result of multicollinearity test which was conducted for Model 1. We exclude several items because otherwise it would increase the Mean VIF to more than 10. At this point, Model 1 is still within the boundary of allowed Mean VIF which is below 10.00.

Multicollinearity of Model 2

Variable VIF 1/VIF

ProceedMedian 6.13 0.163114 MedianMarketValue 5.17 0.193538 Mediansalenominal 2.36 0.424395 Mean VIF 4.55

(38)

37

(39)

(40)

(41)

(42)

(43)

(44)

43 Generaliststage 0.160838 0.000 0.135135 0.186541 Laterstage 0.00586 0.666 -0.02075 0.032474 Mezzaninestage 0.215969 0.000 0.1418 0.290137 Seedstage 0 Africa -0.16689 0.001 -0.26375 -0.07003 America -0.0543 0.063 -0.11144 0.002842 Asia -0.03212 0.284 -0.09092 0.02667 Europe -0.04993 0.092 -0.10796 0.008103 Pacific 0

TABLE 10:EXTENDED PROBIT REGRESSION RESULT OF MODEL 1

Based on table 10, we extend the detail of the probit regression, mainly by disclosing the result of the impact from the variable Fund Year. Throughout the year 1968 to 2018, we can infer that there is a period of time where it clearly showed higher probability of becoming a generalist, but generally all the funds are more generalist through this entire time. This period starts from 1994 up to 2004 (as shown in the blue box). As you can see, the probability of becoming generalist is high in this period, which is on average account as more than 60%. The lowest probability is in 1981 which the probability was only 23%.

(45)

44 Asia 0.101** (0.0439) Europe 0.152*** (0.0435) Pacific 0.163*** (0.0534) Observations 18,402

Are Venture Capital Funds Changing Appetite? : Worldwide Evidence. Master Thesis MSc Business Administration

Master Thesis

MSc Business Administration