Page | 1

Investors as Capabilities: Intra-Investor

Complementarities and Startup Performance

Shai Harel

ABSTRACT

The contribution of investors to the performance of their portfolio companies within the

VC industry has been researched thoroughly. A novel aspect of this paper is that it

examines the impact of a syndicated investment using two different measures: (a) the

total number of investors; and (b) the total number of different investor types. Using a

novel dataset and while controlling for the endogeneity, this paper finds that having more

investors of the same type has a positive curvy-linear impact on the portfolio company's

performance. However, a larger number of investor types has only a positive linear

impact on the company's performance. It is suggested that a tradeoff between having

more investor complementarities and increased coordination costs is the driving

mechanism of these results.

Key Words: Venture Capital Industry, Investor Types, Heterogeneity vs. Homogeneity,

Syndication, Startup Performance

Page | 2

Investors as Capabilities: Intra-Investor

Complementarities and Startup Performance

INTRODUCTION

The venture capital industry comprises many types of investors including: venture

capital funds (VC), corporate venture capital (CVC), angel investors, incubators,

industrial companies, financial institutions and more. Each of these investor types

has different attributes, investment preferences, constraints or hands-on approach

toward their portfolio companies. On one hand, it is possible that having more

investors means greater support and contribution to the portfolio company. On the

other, having "too many" investors may mean more conflicts, conflicting interests

and additional coordination costs. The aim of this paper is to examine the impact of

having more investors on the performance of portfolio companies. It does so by

deconstructing the concept of having “more investors” into two sub-questions: (a)

what is the impact of having a larger number of investors on the performance of the

portfolio company? and (b) what is the impact of having a larger number of investor

types on the performance of the portfolio company?

The main finding of this paper is that having more investors of the same type

has a curvilinear impact on the portfolio company's performance. However, a larger

number of investor types has only a linear impact on the company's performance.

That is, the paper shows that there is an inverted U-shaped relationship between

the number of investors and the exit prospects of a startup (either by an M&A or an

IPO). Similarly, the paper shows a U-shaped relationship between the number of

Page | 3

investors and the write-off prospects of the startup. However, when the

contribution of more investor types to the performance of the firm is examined, the

paper shows a linear relationship to exit probability and a negative linear

relationship to write-off. When examining the impact on the time to exit, a U-shaped

relationship is found between the number of investors and the time it takes a

company to exit, and a negative linear relation between the 'number of investor

types' to the time it takes a company to exit.

The explanation for these findings relies on the understanding that a startup

in its early stages relies on external complementarities to succeed (Prahalad &

Hamel, 1990; Teece, 1986). The investors of the startup in most cases supply these

complementarities in the form of hands-on support or active involvement (for

example: Ber & Yafeh, 2007; Davila, Foster, & Gupta, 2003; Engel & Keilbach, 2007;

Hellmann & Puri, 2000, 2002; Hochberg, Ljungqvist, & Lu, 2007; Kortum & Lerner,

2000; Manigart & Van Hyfte, 1999). However, having more investors increases the

coordination costs and conflicts of interest. Thus, beyond a certain number of

investors, the costs exceed the benefits. However, when the 'number of investor

types' increases, the variety of complementarities supplied by more investor types

compensates for the coordination costs and the conflicts of interest.

Investments in the venture capital industry are far from random. Startups

select their investors from among several offers, and similarly, the various investor

types carefully examine potential investments in startups. Several methods were

used to address these selection concerns: First, the empirical analysis control for

endogeneity using an instrumental variable approach. Second, in parts in which

Page | 4

survival analysis was used, an attempt was made to address endogeneity. Third, a

set of robustness tests were performed, both for the instrumental variable and for

the confounding effect that might occur between the two main dependent variables.

To examine these questions, a unique database was assembled. This database

comprises 1003 angels, 60 CVCs, 601 VCs, 35 incubators, 982 industry-related

companies, 879 finance-related companies, and 55 other investors, participating in

9675 financing rounds investing in 2409 companies during 15 years of activity

(1990-2005). The database contains information on investments made in the Israeli

market, which is one of the world’s most vivid venture capital investment markets.

The rest of the paper is organized as follows: Section 2 sets the theoretical

background and presents the hypotheses. Section 3 describes the database,

empirical approach and presents descriptive statistics. Section 4 contains the

empirical results and Section 5 concludes and sets directions for future research.

LITERATURE REVIEW

The aim of this paper is to assess the impact of the number of investors on

performance of their portfolio companies. However, clarification of what it means to

have "more investors" is required. The first option is that it means the company has

a larger number of investors (i.e. 10 investors rather than two). The second option is

that the company has more investor types (i.e. one angel, one VC and one CVC vs.

just VCs). The analysis in this paper addresses both options. Thus, based on the

importance of the investors' hands-on approach to the performance of the startup,

the literature review begins with an elaboration of the concept of complementarities

Page | 5

between firms; next is a review of the literature concerning the impact of the

number of syndicate members on the performance of a portfolio company. This is

followed by a review of various investor types, their hands-on approach and an

examination of the impact of number of investor types on the portfolio company's

performance.

INVESTORS AND COMPLEMENTARITIES

Starting any new business is, in most cases, a complex and risky action, all the more

so in the case of startups. Startups usually face great technological challenges,

require longer periods of research & development (that sometimes can even take

years) and need considerable resources. Even when a potentially successful

product/service is developed, the difficulties and complexity associated with

launching it or with other post-sales activities are usually greater than for a

"standard" good or service. Moreover, in many cases of technology-based startups,

the entrepreneurs who initiate them are not experienced business people. For

example, the entrepreneurs can be scientists or engineers with a great and

promising innovative idea; these entrepreneurs might even have the basic skills to

launch their business, but it can be safely assumed that their "personal competitive

advantage" rests in the technological sphere and that their inexperience or limited

managerial skills endanger the process of transforming innovation into a

marketable product/service (regardless of its quality). The company's investors are

aware of these risks and thus take a hands-on approach as they often possess the

complementarities the startup needs.

Page | 6

But what are these complementarities exactly? The concept of

complementarities evolved from Penrose's (1959) work on the growth of the firm.

According to Penrose, the company’s competitive advantage lay in its ability to

create and sustain a set of unique capabilities, and with these aptitudes, the

company may gain an advantage over its competitors and create a sustainable

competitive advantage (Barney, 1991; Prahalad & Hamel, 1990). Later work by

Teece (1986) and Prahalad and Hamel (1990) suggest that a company should focus

on its unique knowledge and capabilities while obtaining complementary assets and

capabilities from outside the firm (Prahalad & Hamel, 1990; Teece, 1986).

As stated above, within knowledge-intensive companies, the issues

associated with such complementarities are more complex. Due to limited resources

and time constraints, the startup is unable to develop them on its own (Aghion &

Tirole, 1994); thus, the startup must rely on outside support for its development

and commercialization process (Gans, Hsu, & Stern, 2002; Teece, 1986; Tripsas,

1997). The support is needed in almost every aspect of the firm: R&D, production,

marketing, sales, pre- and post-sale service, human resources and financing. Indeed,

most investors in the venture capital industry view themselves as providers of

"Smart Money,” that is, beyond the money they invest in the company, they also

provide additional added value that, they hope, improves the success prospects of

the portfolio company. The second part of this paper reviews this literature, while

the first part reviews the contribution of the syndicate to its portfolio companies.

Page | 7

THE BENEFITS OF HIGHER NUMBER OF SYNDICATE INVESTORS

The paper by Brander, Amit and Antweiler (2002) aims to understand the

motivations for VCs to syndicate. More specifically, they test two alternate reasons,

(a) an additional VC brings about a useful second opinion; and (b) additional VCs

contribute additional capabilities and added value. Their empirical analysis

supports the added-value explanation and shows that syndicated investments

perform better than standalone investments. Additionally, they find that a larger

number of VCs investing in the same company is associated linearly with increased

returns; they also test for non-linear association (between the number of investors

and the returns) but do not find such an association.

Similarly, Tian (2011) compares the performance of portfolio companies

receiving an investment from a syndication of VCs vs. companies financed by a

single VC investor. His main finding is that a syndication of VCs invest more in young

firms in early financing rounds; this also leads to a better product market value for

portfolio firms. Additionally, VC syndicate-backed firms are more likely to have a

successful exit, receive a higher IPO market valuation, incur less IPO underpricing

and perform better after their IPO.

Das, Jo and Kim (2011) examine an interesting viewpoint concerning the

performance of companies financed by VC syndicates vs. those who receive

financing from non-syndicated VCs. More specifically, they try to understand

whether performance is driven by value added or solely by better selection.

Controlling for endogeneity, they find that companies financed by VC syndicates

have higher exit probabilities and faster time-to-exit. Additionally, their results

Page | 8

show that the magnitude of exit multiples is a result of the selection of portfolio

companies, whereas the probability and speed to exit are the result of the value-

added they gain.

This paper differs from the above three in several respects. First, it

incorporates a new measure for the syndicate size and deals with a broader range of

investor types. Second, it looks at the impact of varying (continuous) syndicate size

as opposed to the former two (Tian's and Das, Jo and Kim), which examine a single

VC vs. a VC syndicate. Third, while their papers only examine the impact of

syndication on positive results (i.e. exit) this paper also examines the impact on

negative results (i.e. if the company ceases to operate) and on the time to exit.

Fourth, the findings in this paper differ from theirs as the analysis shows a non-

linear relationship between the number of investors to: (a) the exit prospect of the

startup; (b) the failure prospect of the startup; and (c) the startup's time to exit.

THE DETRIMENTS OF HIGHER NUMBER OF SYNDICATE INVESTORS

The literature presented above showed the benefits of having more investors. A

company with more investors has more complementarities and capabilities at its

service. However, not all that glitters is gold; having more investors may bring about

difficulties as well. In a closely related working paper, Du (2009) examines the

benefits and costs VCs should consider when selecting syndication partners. More

specifically, Du looks into the impact of syndicating with homogeneous vs.

heterogeneous partners on the VC itself and on the performance of the portfolio

companies. Du claims that potential syndicate members should be viewed through

Page | 9

the lens of two transaction-cost types: coordination costs and agency costs. For

example, VCs with heterogeneous venture experience may have more

disagreements regarding their advice to portfolio companies, and thus face more

difficulties coordinating actions. Moreover, following research on teamwork, Du

claims that VCs' actions alignment will take longer and require more effort in

heterogeneous syndications. Concerning agency costs, Du claims that when less

experienced VCs syndicate with more experienced VCs, the VC with more

experience may face agency costs as the VC with less experience will exert less effort

in monitoring and advising portfolio companies. On the other hand, Du also

discusses the benefits of heterogeneity and claims that it may also be helpful for the

VCs to syndicate, as one of the properties of the VC industry is that the various

investors face many complex and non-routine problems. The decisions the VC

industry deals with (such as: opportunity assessment, recruiting policies and

preferences, exit or write-off decisions) are subjective and thus, co-investing with

diverse partners can lead to better decision-making (this is similar to the

motivations suggested by Brander, Amit and Antweiler (2002)). The major

difference between this paper and Du's is that she examines whether VCs prefer

similar or different syndicate partners and what the prospects are for the formation

of various syndicate mixtures. Additionally, she examines what the impact of this

choice is on the portfolio company and on the VC itself. This paper, on the other

hand, does not deal with prospects for formation of various syndicate mixtures, nor

does it deal with their impact on the VC itself. Similarly to her work this paper

focuses on the syndicate's implications on its portfolio companies but differs

Page | 10

empirically in several aspects: (a) in her empirical analysis, Du considers only the

first round of investment while this paper considers all rounds; (b) this paper uses

different measures to examine the impact of the size of the syndicate ('number of

investors') and of diversity in investor types (i.e. VCs, Angels) on the performance of

the portfolio company; (c) this paper has more outcome measures such as write-off

dummy and time to exit; and (d) this paper finds a non-linear relationship between

number of investors to the three dependent variables mentioned above.

Indeed, a few papers find evidence for a negative impact of the number of

investors on the startup's performances. For example, in a working paper, Guler &

McGahan (2007) claim that, in countries with weak legal and institutional systems, a

member of a VC syndicate may behave opportunistically and harm the performance

of portfolio companies. Other evidence for a negative impact was found in a paper

closely related to this one by Agarwal (2011). Similarly to this paper, Agarwal claims

that as the number of investors increases, a negative impact may occur as problems

such as coordination or conflicts of interest arise. Moreover, Agarwal suggests that

adding more investors in later rounds may cause the startup to receive less

monitoring and less added value. Using both a game theoretical model and an

empirical analysis, Agarwal finds that the marginal contribution of each additional

investor diminishes and creates an inverted U-shaped relation between having

more investors and the performance of the portfolio company. Several differences

distinguish his paper and this one. First, this paper tests two facets of additional

investors (such as: greater number of investors; a larger number of investor types)

while his tests only for one. Second, this paper’s analysis examines a broader scope

Page | 11

of performance indicators (exit dummy, closure/write-off dummy and time to exit).

Third, Agarwal examines this question through the lens of the marginal effort

invested by an additional investor relative to its share in the startup; thus, Agarwal

tests the impact of an additional investor between rounds. This paper examines the

cumulative impact of an additional investor; that is, it tests the impact of all

investors in all rounds together. The reasoning behind this approach is that, when

assessing the impact of an additional investor between rounds, the measures for

failure or success (the dependent variable) should be of the same nature and not the

final outcome of the startup (exit or write-off). For example, a round-based analysis

may address questions such as: by how much has the value of the firm increased or

decreased in that round? How much money was raised in the following round/s?

Though these are intriguing and important questions, the database does not contain

enough information to address them.

NUMBER OF INVESTOR TYPES

Most (if not all) the literature that deals with the impact of a syndicate on the

performance of its portfolio company examined it by counting the number of

investors. However, past research showed that not all investor types support the

same set of activities. This is only logical, as not every investor has the same

capabilities (and even if they did have the same capabilities they probably vary in

quality); thus, different investors may contribute to different aspects and in varying

intensity. To clarify this, the next part briefly elaborates the main finding of the

Page | 12

hands-on approach of three major types of investors: business angels, VC funds and

CVCs.

The literature on VC funds has shown that VC funds are closely involved in

the operations of portfolio firms. This includes recruiting policies (Davila, Foster, &

Gupta, 2003; Hellmann & Puri, 2000, 2002), innovation effectiveness (Hellmann &

Puri, 2000; Kortum & Lerner, 2000), CEO turnover, and the probability of hiring a

marketing VP or of introducing an employee stock-option scheme (Hellmann & Puri,

2002). VC funding has also been associated with faster growth rates (Davila, Foster,

& Gupta, 2003; Engel & Keilbach, 2007), increased survival probability (Ber & Yafeh,

2007; Hochberg, Ljungqvist, & Lu, 2007; Manigart & Van Hyfte, 1999) and higher

profit volatility (Manigart & Van Hyfte, 1999).

Business angels are also actively involved in their portfolio companies. They

are shown to be involved in the startups in day-to-day operations (Amatucci & Sohl,

2004; Benjamin & Margulis, 1999; Brettel, 2003; Freear, Sohl, & Wetzel, 1995;

Madill, Haines, & Riding, 2005; Stevenson & Coveney, 1996), human resource

operations (Ardichvili, Richard, Tune, & Reinach, 2002; Brettel, 2003; Harrison &

Mason, 1992b), mentoring and business advice (Ardichvili, Richard, Tune, &

Reinach, 2002; Brettel, 2003; Ehrlich, De Noble, Moore, & Weaver, 1994; Lumme,

Mason, & Suomi, 1998; Mason & Harrison, 1996; Paul, Whittam, & Johnston, 2003;

SæTre, 2003; Stevenson & Coveney, 1996; Tashiro, 1999), networking activities

(Amatucci & Sohl, 2004; Ardichvili, Richard, Tune, & Reinach, 2002; Brettel, 2003;

Lumme, Mason, & Suomi, 1998; Mason & Harrison, 1996; Paul, Whittam, & Johnston,

2003; SæTre, 2003; Sørheim, 2005), strategic planning (Amatucci & Sohl, 2004;

Page | 13

Ardichvili, Richard, Tune, & Reinach, 2002; Brettel, 2003; Ehrlich, De Noble, Moore,

& Weaver, 1994; Harrison & Mason, 1992a; Lumme, Mason, & Suomi, 1998; Mason

& Harrison, 1996) and supervision (Ehrlich, De Noble, Moore, & Weaver, 1994;

SæTre, 2003).

In the case of CVCs, the issue of active involvement and added value is more

complex. On one hand, CVC have been found to support various types of added value

activities and provide complementarities such as supplying infrastructure for

product development, supportive manufacturing resources and marketing, sales

and post-sales activities (Dushnitsky & Lenox, 2005; Katila, Rosenberger, &

Eisenhardt, 2008). On the other, the motivations behind CVC investments are

somewhat different from those of other investors; while other investors mainly seek

financial gains, CVCs view the strategic nature of their investment as more

important than financial considerations; thus, conflicts may arise between the CVC

and the invested company. For example, it was found that the investing CVC may

produce competing products or expropriate the intellectual property of the invested

company (Hellmann, 2002, Dushnitsky and Shaver, 2009). Moreover, the CVC's

oversight teams are less experienced in supporting such investments (compared to

independent VCs) and less financially incentivized, and therefore the quality of their

added value is generally lower than that of independent VCs (Dushnitsky & Shapira,

2010; Ivanov & Xie, 2010). Another conflict potentially arises when other

corporations are reluctant to cooperate with the invested company, as it is at least

partly owned by a competitor (Park & Steensma, 2012). Having said all that, and

despite the drawbacks that accompany CVC, past research has shown that CVC-

Page | 14

backed companies perform somewhat better than companies backed by

independent VCs (Gompers & Lerner, 2000a; Maula & Murray, 2001).

As this short review shows, only some of the complementarities supplied by

the various investor types overlap. Thus, it may well be that it is not the number of

investors that influences the success prospects of a company, but the number of

investor types a company has that matters (i.e. it is better for a company to have

several investor types such as an angel, a VC and a CVC, rather than having only

VCs). The underlying assumption here is that when several types of investors are

jointly involved in the same company, then the scope of potential complementarities

standing at the service of a startup is greater than when only a single type of

investor is involved.

However, heterogeneity also comes with a price. When more investor types

are involved, the scope of considerations and constraints each type has is much

larger. For example, VCs and angels do not have the same investment strategy,

available capital or investment time horizon. A VC may find it easier to coordinate or

share goals with other VCs but it is not obvious that it can do this as easily with

angels or CVCs. These differences may increase the coordination costs and may even

harm the portfolio company.

As a case study for heterogeneity between investors, the next section briefly

reviews the literature that examines the relationship between two types of

investors: angels and VCs. Chemmanur and Chen (2006) formulate a model to

explain the startup's shift from angels to VC financing as the company reaches later

stages. They claim that VCs are able to add value to portfolio companies while angels

Page | 15

cannot. By contrast, Schwienbacher (2012) claims that the difference between these

types lies elsewhere. While both can add value to the company, VCs possess greater

financial resources, crucial to firm development as it matures. Angels, on the other

hand, are aware of their disadvantage and compensate for it by investing more

effort, believing it will help attract additional investors in later stages. For example,

87% of angels in the US were found to possess experience in operations (Freear,

Sohl, & Wetzel, 1991) while VCs have very little or no experience in operations at all

(Van Osnabrugge & Robinson, 2000). Moreover, Benjamin and Margulis (1999)

argue that some angels work on a regular basis in the startup, making their

investment much more "personal.” A great example of this is the difference in time

perspective. Angels are said to do less intensive and less time-consuming due

diligence (Van Osnabrugge, 1998), but have a longer investment time horizon

(Freear, Sohl, & Wetzel, 1994; Wetzel, 1983). VCs, on the other hand, may not

always take decisions in the best interest of their portfolio companies. VCs were

found to perform “grandstanding,” meaning selling their shares at an IPO at a lower

price to raise further funds (Gompers, 1996; Johnson & Sohl, 2011; Lee & Wahal,

2004). Angels, on the other hand, usually do not face pressure to quickly signal their

high performance to the market, and therefore have a longer time horizon. This also

means their decisions will be in the portfolio company’s best interest (Johnson &

Sohl, 2011). However, this also has some negative effects; Goldfarb, Hoberg, Kirsch

and Triantis (2009) find that companies having only angel investors are more likely

to become “living dead” than when a VC is also involved. They suggest that either

Page | 16

these firms need more patience or that angels are less prone to force unsuccessful

startups to liquidate.

HYPOTHESES

Summing up the possible impact of the syndication's size and structure on the

performance of the portfolio company, it seems there are two contradictory forces

working simultaneously. On one hand, having more investors/investor types

increases the availability of complementarities and capabilities, but on the other,

having more investors/ investor types increases the likelihood of conflict and raises

coordination costs. Thus, the first set of hypotheses that concern the 'number of

investors,’ the 'number of investor types' and the exit probabilities are:

Hypothesis 1a: There is an inverted U-shaped relationship between the

'number of investors' a company has and its exit probabilities

Hypothesis 2a: There is an inverted U-shaped relationship between the

'number of investor types' a company has and its exit probabilities

However, when the measure for performance is the probability of a write-off,

the expectation is to get the opposite relations. On one hand, as the number of

investors/investor types increases, the bundle of resources the startup has at its

disposal increases; thus, prospects of the company’s failure diminish. On the other

hand, as the number of investors/investor types grows further, the relative share of

Page | 17

each investor decreases and the company may face a moral hazard problem as each

investor counts on the others to support the startup. Thus, the second set of

hypotheses concerning the 'number of investors,’ the 'number of investor types' and

the write-off probabilities are:

Hypothesis 1b: There is a U-shaped relationship between the 'number of

investors' a company has and its write-off probabilities.

Hypothesis 2b: There is a U-shaped relationship between the 'number of

investor types' a company has and its write-off probabilities.

Last, drawing upon the same reasoning mentioned above, the third set of

hypotheses that concern the 'number of investors,’ the 'number of investor types'

and the 'time to exit' are:

Hypothesis 1c: There is a U-shaped relationship between the 'number of

investors' a company has and the time to exit.

Hypothesis 2c: There is a U-shaped relationship between the 'number of

investor types' a company has and the time to exit.

The next section defines the dependent variables by which these hypotheses

will be tested.

Page | 18

VARIABLES AND METHOD

DEPENDENT VARIABLES

This paper’s main question concerns the contribution of investors to the

performance of their portfolio companies. In the case of the VC industry, financial

measures such as IRR are not available as the information required to compute it is

usually not disclosed. Thus, following former research (for example: Brander, Amit,

& Antweiler, 2002; Gompers & Lerner, 2000b; Sørensen, 2007) the paper uses exit

as a measure of successful investments. Thus, the first dependent variable is 'Exit

Dummy,' which equals one if the company performed an exit (either through an IPO

or an M&A) and zero otherwise. As the database also contains information on cases

when companies were written off, the second performance variable is 'Write-off

Dummy' (which again equals one if the company ceased operations and zero

otherwise). Last, as this paper aims to assess the complexities involved with

syndications, following Das, Jo and Kim (2011) the third dependent variable is 'Time

to Exit' (measured in years). This is defined as the time passed since the company

was established until it reported an M&A or an IPO. Notably, using and 'Exit' or

'Write-off' may be a noisy measure of performance as not all exits are necessarily

good for the company (for example, consider a case where the company was sold for

a smaller amount than was invested in it).

MAIN INDEPENDENT VARIABLES

The first main independent variable is the total 'number of investors' that invested

in the company in all rounds regardless of their types (if the same investor

Page | 19

performed a follow-up investment in the same company, it was counted only once).

The logic behind this measure is that more investors result in more

complementarities. The second main independent variable assumes that it is not the

total 'number of investors' that matters but rather the 'number of investor types'

that brings a larger scope of complementarities. Thus, this variable counts the total

number of types that invested in the company in all rounds. Investors are classified

into seven types. To test the non-linear relationships, the quadratic terms of both

variables (the number of investors/ investor types squared) are used.

Another important independent variable used in all regressions is the total

number of rounds each company has. It is reasonable to expect that a company that

undergoes more rounds will have more investors. Similarly, longer years of activity

may increase the chance that a company will have more investors or that it is more

likely to exit; thus, time variables were used by using a dummy variable of the year

the company was established. Additional control variables used are: industry

classification; the geographic location of the company's headquarters; and whether

this company was part of a governmental program.

SELECTION AND ENDOGENEITY TREATMENT

The matching between investors and portfolio companies is not exogenous. On one

hand, this matching may be correlated with firm characteristics that are not

observable to us (for example, a promising company may attract more investors).

On the other hand, an investor may have unobservable characteristics that lead the

portfolio company to prefer it over another (for example, former research showed

Page | 20

that investees may prefer to have a reputable VC as an investor and are willing to

incur a valuation discount of 10 to 14 percent for that (Hsu, 2004)). Thus, the

performance of the portfolio company may be the result of a selection bias and not a

result of added value activities that investors contribute to the company.

To control for unobserved characteristics and the matching process between

investor-investee, a two-stage model based on an instrumental variable is

employed. Following Giannetti and Yafeh's (2012) instrumental variable (in their

case for a bank-firm match) and implementing Du's (2009) rule of thumb for

potential investors, the paper postulates that investor-investee probability of

matching (independent of the investor's or the investee's characteristics) depends

on the distribution of the 'number of investors' of the same investor type who were

active (performed an investment) in the same industry at the same quarter. Thus,

the instrumental variable is equal to the summation of the number of all potential

investors who were active in the same quarter in the same industry. For example,

suppose that on February 15th 2004, company A, which is active in the 'Internet

industry,’ received an investment from one VC and two angels. The instrumental

variable examines how many VCs and how many angels invested in the 'Internet

industry' during the first quarter of 2004. Suppose that 10 angels and seven VCs

invested in the first quarter of 2004 in the 'Internet industry,' then the instrumental

variable is 17. This instrumental variable was used in the regressions as an

instrument for the linear independent variable. As an instrumental variable to test

the non-linear relationship, the quadratic term of this instrument (the potential

number of investors squared) was used.

Page | 21

It is important to understand that the instrument has two functions.

Following Bottazzi, Da Rin and Hellmann (2008), Giannetti and Yafeh (2012), and

Sørensen (2007), the identifying assumption is that the investors' characteristics do

not affect the investee's performance directly but affect the performance through

their added-value activity. The decision to invest in a specific company is correlated

with the investor's characteristics and thus any change in the performance of the

firm is a result of the investors' added value.

The analysis involves two main statistical methods: probit regression and

hazard models. To control for endogeneity, a two-stage probit regression was

performed. In the first-stage regression, the instrumental variable was used as a

predictor for the number of investors/investor types and then the fitted results of

this probit regression were entered into the main regression that measured the

impact of the number of investors/investor types on firm performance. To further

validate these regressions, the standard 2SLS model for linear regression was used,

with the same variables used in White's correction (1980) for heteroscedasticity.

Correcting for heteroscedasticity enables the use of a linear regression even though

the dependent variable is a dummy variable (for an elaboration on the validity of

such a treatment see Angrist, 2001).

In the case of hazard models, there is no "formal" treatment for endogeneity.

Thus, to further validate the results and control for endogeneity, the paper mimics

the process of the 2SLS. As a first stage regression, as in the probit models, a simple

regression was performed with the number of potential investors as a predictor for

the actual number of investors/investor types. Then, the fitted results were used in

Page | 22

the second stage as the hazard model regressions. This process is similar to the

process performed by Das, Jo and Kim (2011) when they used a hazard model to

predict the same dependent variable except that, in their case, the first-order

regression was a probit as their dependent variable was a dummy variable.

Last, due to the high correlation between the two main independent

variables, the 'number of investors' and the 'number of investor types,’ the

regressions are performed separately for each of these variables. However,

regressing them separately may result in a confounding effect, meaning it might be

that the results in the regressions when controlling for the 'number of investors'

variable is actually driven by the variable 'number of investor types,’ and vice versa.

To rule out this bias, several robustness tests for the results were performed and

will be discussed in the analysis section.

DATA AND DESCRIPTIVE STATISTICS

Insert Table I about here

The data in this study was taken from the Israel Venture Capital (IVC) database1, the

most comprehensive database available on the Israeli VC industry2. The entire

database was documented during October 2008. The original data included

information on 192 angels, seven CVCs, 481 foreign investors, 165 investment

companies, 215 VCs and 3281 unclassified investor types. Since such a large number

1 http://www.ivc-online.com/ 2 As such, formal publications of the Israeli Central Bureau of Statistics concerning the VC industry in Israel are also based on data from this website.

Page | 23

of unclassified investors may bias the results substantially, the unclassified

investors were recoded into seven types: (1) VC; (2) angel investors; (3) CVC; (4)

industry-related companies; (5) finance-related companies; (6) incubators; and (7)

others. The recoding yields the following distribution: 1003 angels (this includes

any individual reported to invest in a company3); 60 CVCs; 601 VCs (405 foreign VCs

and 196 local VCs); 35 incubators; 982 industry- related companies (this includes all

investors who invested in startups and who do not have a CVC arm; this can be, for

example, a consulting company, a communications company, etc.); 879 finance-

related companies (this includes any investor whose primary occupation is in the

finance sector: investment banks, holding companies, hedge funds, etc.); and 55

other investor types (this includes investors who do not follow any of the above

definitions; this could be, for example, a governmental fund, a charity fund, a

hospital, university technology transfer company, etc.). Table I presents descriptive

statistics by investor types. Interestingly, all types of investors have a fairly similar

percent of write-offs, varying between 10 and 20 percent. Similarly, their success

rates are somewhat alike and vary between 30 and 40 percent except for

incubators, probably because incubators are, by definition, biased toward early-

stage investments. The average amount invested is fairly different and varies

between USD10M and USD37M. Another interesting observation is that the

involvement of foreign investors is relatively high and that all investor types include

3 Following Harrison and Mason (2000) and Johnson and Sohl (2011), which exclude family finance as angel's investment, any individual who had the same last name as the founder/s was not considered as an angel investor.

Page | 24

foreign investors. It can also be seen that CVCs and VCs invest on average in more

companies and have more follow-up investments.

Insert Table II about here

The data also include information on 7131 startup companies. For 2910 of

these, the data include detailed information on 16,245 investment rounds (of these,

10,958 rounds are first investments in a company and the rest are follow-up

investments). However, years preceding 1990 were not used in this research as the

Israeli VC industry in those years was fairly small and the data may be partial or

inaccurate. Additionally, the analysis involves performance measures based on the

exit or the write-off of companies. Exit or write-off of companies does not take place

instantly after the company's inception; thus, companies established during the

years 2006-2008 were also not used in this study, and hence the final sample

includes 2409 companies participating in 9675 investment rounds.

Table II presents descriptive statistics of the companies in the sample. Most

of them (57%) are located in the center of Israel, either in Tel Aviv or the central

area, and most belong to the Information and Communications Technology (ICT)

industry in its various forms (Internet, IT & enterprise software, communications)

or the life-science industry, which is also fairly developed, accounting for 24% of

startups. Regarding their performance, 22% of the companies performed an exit,

which on average, took 5.75 years since inception with an average exit amount of

USD152M (at 2008 values). Thirty percent of the companies were closed and, on

Page | 25

average, 4.15 years passed until they were closed. Within the context of this

research it is interesting to acknowledge that the average total 'number of investors'

in a single company is 4.02 with 2.11 investor types participating in an average of

2.24 rounds.

Insert Table III about here

ANALYSIS AND RESULTS

Table III presents the correlation matrix between the main independent variables.

As this table shows, all variables are significantly and highly correlated. To avoid

multicollinearity, each of the two main independent variables was used separately

and not together in the same regression. Though running these variables separately

does not allow “horse racing” between the two variables and understanding which

has the greatest impact, running them together would result in highly multicollinear

results that would not be valid anyway. However, this issue is addressed extensively

in the robustness tests that follow the main statistical analysis.

EXIT AND WRITE-OFF PROSPECTS

Table IV presents the analysis for exit and write-off prospects using a probit model

where the dependent variables are 'Exit Dummy' or 'Write-off Dummy.’ In Models 1

– 4 the main independent variable is the 'number of investors,' while in Models 5 – 8,

the main independent variable is the 'number of investor types.’ Model 1 indicates a

significant and positive relation between the 'number of investors' and 'Exit

Dummy.’ Model 2 adds the 'number of investors squared'; this variable is also

Page | 26

significant, suggesting a non-linear relation between the number of investors and

the exit prospects of the company. Similarly, Model 3 shows a negative significant

relationship between the 'number of investors' and write-off prospects. However,

once the 'number of investors squared' is added in Model 4, a positive and

significant relation is visible. This means there is a U-shaped relationship between

the number of investors and the write-off prospects of the startup.

The right side of Table IV analyses the relationship between the 'number of

investor types' and the performance of the startup. Model 5 shows a significant and

positive relation between the 'number of investor types' and 'Exit Dummy.’ Yet,

Model 6 shows that once the 'number of investor types squared' is added, there is a

negative relation, though not statistically significant. This means the relationship

between the exit prospects and the number of investor types is linear and positive.

Models 7 and 8 show a similar result—a linear and negative significant relationship

between the number of investor types and the 'Write-Off Dummy,' but not a U-

shaped relationship.

Insert Table IV about here

TIME TO EXIT

Another aspect of performance that was suggested concerns the impact of the

'number of investors' and of the 'number of investor types' on the 'Time to Exit.’

These relations between the variables were tested using a Cox proportional hazard

model. Table V reports the coefficients of the hazard model. Model 1 examines the

relationship between the 'number of investors' and the time to exit and shows that

Page | 27

there is a significant and negative relation. Model 2 shows that when the 'number of

investors squared' is added, its coefficient is positive and significant; hence, there is

a U-shaped relationship between the 'number of investors' and the time to exit.

Model 3 presents the impact of the 'number of investor types' on the 'Time to Exit'

and shows a linear and negative relationship between the 'number of investor types'

and the 'time to exit.’ However, Model 4 shows that when the 'number of investor

types squared' is added, its coefficient is positive but not significant; thus, contrary

to expectation, this relationship is not U-shaped. This suggests that the relation

between the 'number of investor types' and 'time to exit' is negative and linear; thus,

more investor types lead to a shorter time to exit.

Insert Table V about here

To conclude this section, the main finding is that of an inverted U-shaped

relationship between the 'number of investors' and the exit prospects of the firm.

Additionally, findings indicate that the 'number of investors' has a U-shaped

relationship with the 'write-off prospect's of the firm and with the 'time to exit.’ The

'number of investor types' is linearly and positively related with the startup's exit

prospects and linearly and negatively related with the 'write-off prospect's of the

firm and with the 'time to exit.' The next section validates these findings using

several robustness tests.

Page | 28

Robustness Tests

To further validate the results, several robustness tests were performed. The

robustness tests aim to address two central issues that may bias the findings. The

first bias may occur as a result of endogeneity; hence, the first part of the robustness

tests discusses the measures taken to address this issue. The second issue that may

bias the results is that, as the correlation between the 'number of investors' and the

'number of investor types' is high, the regressions performed regressed these two

variables separately. However, it may be that interpreting the results as derived

from the 'number of investors' is flawed, and the results are actually driven by

having a larger 'number of investor types' (and vice versa). Thus, the second part of

the robustness tests addresses this issue and validates the results both for the

'number of investors' and for the 'number of investor types.’

ENDOGENEITY

Insert Table VI about here

Models 1 – 4 in Table VI describe several robustness tests for endogeneity. Models 1

and 2 use the instrumental variable approach and a probit model to address the

issue of endogeneity. As can be seen, the results are the same as before. However,

one of the requirements of the probit model when using a instrumental variable

approach is that the exogenous variables will be continuous variables. The

instrument used is discrete; to verify the robustness of the results in Models 3 and 4,

the standard 2SLS model was used. As mentioned above, the standard 2SLS process

Page | 29

may be used with a binary dependent variable while performing White's (1980)

correction for heteroscedasticity. Both models indicate the results to be the same;

thus, there is an inverted U-shaped relation between the number of investors and

the success prospects of a startup. Similarly, there is a U-shaped relation between

the write-off prospects of the startup and the number of investors.

Models 1 – 4 in Table VII control for endogeneity for the results of the

'number of investor types' variable. Models 1 and 2 show the results when an

instrumental variable with a probit model was used. Model 1 indeed validates

former findings but in Model 2, though the coefficient is negative, it is not

statistically significant. Again, to further validate the results, Models 3 and 4 use

2SLS regressions while controlling for heteroscedasticity; this time the results are

the same as in Models 5 and 7 in Table IV. Hence, this validates the results and the

conclusion is that there is a linear and positive connection between the 'number of

investor types' and the prospects for exits of the firm. Additionally, there is a linear

and negative connection between the 'number of investor types' and the prospect

for write-off.

Insert Table VII about here

The next section addresses the issue of endogeneity in the hazard models. As

mentioned above, there is no direct way to control for endogeneity in hazard

models. Hence, following Das, Jo and Kim (2011) a process was performed that

mimics the logic of instrumental variables in regressions. Models 1 and 2 in Table

Page | 30

VIII show the results using this technique, which uses the residual of the main

independent variables4 in the hazard model. Both models validate former findings;

Model 1 shows a U-shaped relation between the 'number of investors' and the time

it takes the company to exit. Model 2 shows a linear and negative relation between

the 'number of investor types' and the 'time to exit.’

Lastly, a robustness test was performed for the instrumental variable used. A

constructed variable was created as an instrument to control for endogeneity. This

variable was based on the number of potential investors of the same type investing

at the same quarter in a company in the same industry as the controlled company.

To further validate this inclusion criterion, another instrument was constructed, this

time with the criterion that the same investor type invested in a company in the

same industry within a six-month time frame. The results stayed the same (these

regressions are not reported but the results are available upon request).

Validating the Results of the Number of Investors

As explained above, there may be a bias in the analysis as a result of a confounding

effect between the 'number of investors' and the 'number of investor types.’ To

further validate the finding regarding the 'number of investors,' several robustness

tests were performed. Models 5 and 6 in Table VI examine a subset of the database in

which the 'number of investor types' is held constant and is equal to one. The

reasoning behind this is that in this case, if the results are the same as before, then it

is certainly driven solely by the 'number of investors' as the 'number of investor

types' is fixed. Indeed, Model 5 shows that while holding the 'number of investor

4 'Number of Investors,’ 'Number of Investors2' and 'Number of Investor Types'

Page | 31

types' fixed, there is still an inverted U-shaped relation between the 'number of

investors' and the exit prospects. However, in Model 6, which examines the relation

between the write-off prospects and the number of investors, there is only a

significant negative relation but the positive relation of the squared term is not

significant. As this can be the result of the smaller set of observations, in Model 7 the

constraint of having robust standard errors was relaxed while the same probit

regression was performed as in Model 6. This time, the former results are validated

and the model shows a U-shaped relation between the write-off prospects and the

'number of investors.’

Insert Table VIII about here

Next, using the same method (holding the 'number of investor types' equal to

one) a test was performed for a confounding effect within the hazard models. Model

3 in Table VIII tests for this regarding the 'time to exit' and finds a U-shaped relation

between the two, thus validating the former findings.

Validating the Results of the Number of Investor Types

The next section validates the finding regarding the 'number of investor types' using

two validation techniques. First, a sub-sample of firms was used where the 'number

of investor' equals six. Second, an alternative measure is introduced and tested for

the variable 'number of investor types.’

Page | 32

Limiting the 'number of investors' to no more than six (nearly the maximum

number of investor types a company may have), Models 5 and 6 in Table VII confirm

the results from Models 5 and 7 in Table IV. An attempt to perform the same process

for the hazard model presented in Table V did not yield significant results (the

results are not reported). Hence, to confirm the finding, several regressions were

performed using an alternative measure of investor types.

The alternative measure is based on a Herfindahl index value of the

proportion of each group of investor type. The Herfindahl index is a measure usually

used to evaluate the level of concentration in ownership of an industry. The index

examines the share of each company (in percents) in the industry out of the total

number of companies. When computed to measure industry concentration, the

Herfindahl index is computed by the formula5:

The product of this computation is that the higher the value is, the more

concentrated the industry is. Hence, it is expected that this industry will be less

competitive. Within the context of this paper, the reason this measure is used as a

robustness test for the 'number of investor types' is that it is driven by both the

'number of investors' and the 'number of investor types.’ The underlying

assumption is that, for each type to actively contribute and be involved in the

5 Where si is the market share (in percents) of firms in the industry.

Page | 33

decision-making, the relative size of its group (as measured by the seven types

above) should be considerable. To better understand this point, let us consider, for

example, two hypothetical companies: (a) Company A has nine angels and one VC

investor; and (b) Company B has five angels and five VC investors. In this example,

Company A's Herfindahl index would be: 0.92+0.12=0.82 while Company B's would

be: 0.52+0.52=0.5. According to this measure, the composition of Company B’s

investors is less concentrated than Company A’s. Assuming that each type has

similar complementarities, Company A may be biased towards the capabilities of

angels, while Company B enjoys a more balanced set of complementarities.

Moreover, the investors are in most cases part of the company's board of directors

(hereby after BoD) and are intensely involved in its decisions. Thus, it may be

speculated that if the company has one dominant type of investor, it will be similar

to the case of having fewer investor types and this may reduce the contribution of

having several types6. It is important to note that, due to data availability, the entire

set of measures used in this research does not take into consideration the relative

share of each investor in the startup's ownership. Incorporating this information

into the measures would improve their accuracy.

6 The logic behind this claim derives from a different strand of literature: the literature on women's representation in BoDs and their impact on the performance of the firms. Kanter (1977), who introduced the critical mass theory of women’s representation in BoDs, suggests that the impact of women on the BoD's decisions and on the performance of companies requires a critical mass of about 30%-35% (Kanter, 1977; Rosener, 1995; Schwartz-Ziv, 2012; Shrader, Blackburn, & Iles, 1997). Thus, following this logic, what matters is the relative size of the group of each type of investor. This means that comparing a company with a diverse group of investor types but with unbalanced proportions of the various types to a company with a more balanced proportion of investor types (and the same diversity), will not benefit from the bundle of capabilities that stands at its disposal and thus will be less successful.

Page | 34

Model 7 in Table VII shows that a more concentrated group is negatively and

significantly related to the 'Exit Dummy.’ This means that a diverse group of

investor types with balanced proportions of the various types performs better than

an equally diverse group with less balanced proportions of investor types. This

validates the findings in Model 5 in Table IV. Similarly, Model 6 shows that a more

concentrated group of investors in positively and significantly correlated with the

write-off prospects, thus validating the findings in Model 7 in Table IV. Models 9-12

are the same regressions controlling for endogeneity7. Models 9 and 10 present IV

probit regression and Models 11 and 12 are regressions using 2SLS. As can be seen,

these regressions yield the same results as before. The first stage's results are also

reported at the bottom of the table, where all the relevant variables can be seen to

be statistically significant. Using the Herfindahl measure, the same logic is used in

Table VIII to further validate the hazard models. Model 4 is a hazard model using the

Herfindahl measure itself, while Model 5 is the residual value of the Herfindahl

measure, thus controlling for endogeneity. Both models yield similar results and

validate the above findings.

DISCUSSION AND CONCLUSIONS

This paper has several contributions. First, it contributes to the literature on

venture capital as it elucidates and sharpens the concept of what it means when a

company has more investors. It does so by distinguishing between two possible

7 As an instrumental variable for the Herfindahl index, the Herfindahl index for potential investors was computed. In the example used above for computing potential investors (in which ten angels and seven VCs invested in the first quarter of 2004 in the “Internet industry”) the instrumental variable

for the Herfindahl index would be:

Page | 35

options: (a) having a larger number of investors regardless of their type; and (b)

having a larger number of investor types. Second, this paper is one of the first

attempts to address the impact of a large variety of investor types on firm

performance. Moreover, in the VC industry, many investments are in the form of a

syndicate and involve several types of investors. Thus, past research that focused

mostly on one or two types of investors at most might not have captured the entire

picture. As this paper shows, the number of investor types influences the startup

performance; thus, analyzing one type of investor in companies with more than one

type may bias the results.

The most prominent finding of this paper is that having a larger number of

investors does not necessarily mean that the company's performance is better.

However, having more types of investors has a linear and positive impact. These

findings are consistent for all three performance measures: exit prospects, failure

prospects and the time it takes the company to exit. These findings are very

interesting within the context of the benefits and detriments of heterogeneity

among investors. It seems that in accordance with the concept of complementarities,

the benefits to the portfolio company from a greater scope of complementarities are

greater than the difficulties it may face as a result of more coordination efforts.

However, this is not the case when more investors of the same type are involved. In

this case, having more investors may harm the company's performance. This finding

emphasizes the tradeoff the entrepreneur faces and suggests that they should

carefully consider numbers and types of investors in seeking further financing. This

Page | 36

also bears a considerable importance for the financial intermediaries themselves,

stressing the importance of carefully choosing whether and with whom to syndicate.

LIMITATIONS AND FUTURE RESEARCH

This research has several limitations: First, though extensive effort was made to

address the issue of endogeneity, it may be that not all unknown variations were

addressed. For example, the instrument is based on the assumption that good

investments are available for everyone; it may be that once an investor invests in

what they conceive as a very promising company, they would prefer to avoid

additional investors. A second limitation may be that the analysis performed

referred to all rounds together and not per round. The main obstacle that should be

addressed is the formalization of a success/failure measure on a per round basis.

The most reasonable measure is company valuation; however, the database used in

this research contains only partial information on firm valuation between rounds.

This stream of research contains many interesting questions that may

provide a basis for future research. For example, it would be interesting to compare

different syndicates and see whether certain syndicate profiles outperform others

(i.e. maybe a syndicate of VCs and angels is better than a syndicate of CVCs,

incubators and VCs, or vice versa). Another option is to examine what types of

added value different syndicate profiles provide. It may also be interesting to

examine the pattern of syndicate formation along the life cycle of the startup. For

example, angels and incubators are usually early-stage investors, while VCs and

CVCs invest in later rounds; it might be interesting to examine whether syndicate

Page | 37

formation patterns exist, and which investor type enters at which stage and what

future investors it draws. Moreover, within the scope of "smart money" and

investor-added value it may be very interesting to see if one investor type is a

substitute for another. For example, it might be that angel investors are a substitute

for incubators or that VCs are a substitute for CVCs (i.e. each investor type can

provide the same added value at the same stages as the other type). Another avenue

for future research may be to examine the impact of foreign investors (of any type)

and compare their investment patterns within various local syndicates.

Page | 38

BIBLIOGRAPHY

Agarwal C. 2011. How Do VC Syndicates Affect Venture Performance: New Insights Challenging Old Studies. Journal of Management Research.

Aghion P, Tirole J. 1994. The Management of Innovation. Quarterly Journal of Economics 109(4): 1185-1209.

Amatucci FM, Sohl JE. 2004. Women Entrepreneurs Securing Business Angel Financing: Tales from the Field. Venture Capital: An International Journal of Entrepreneurial Finance 6(2-3): 181-196.

Angrist JD. 2001. Estimation of Limited Dependent Variable Models With Dummy Endogenous Regressors. Journal of Business and Economic Statistics 19(1): 2-28.

Ardichvili A, Richard NC, Tune K, Reinach J. 2002. The Role Angel Investors in the Assembly of Non-Financial Resources of New Ventures: Conceptual Framework and Empirical Evidence. Journal of Enterprising Culture 10(01): 39-65.

Barney J. 1991. Firm Resources and Sustained Competitive Advantage. Journal of Management 17(1): 99-120.

Benjamin GA, Margulis JB. 1999. Angel Financing: How to Find and Invest in Private Equity. John Wiley & Sons: New York.

Ber H, Yafeh Y. 2007. Can Venture Capital Funds Pick Winners? Evidence from Pre-IPO Survival Rates and Post-IPO Performance. Israel Economic Review 5(1): 23-46.

Bottazzi L, Da Rin M, Hellmann T. 2008. Who Are the Active Investors?: Evidence from Venture Capital. Journal of Financial Economics 89(3): 488-512.

Brander JA, Amit R, Antweiler W. 2002. Venture-Capital Syndication: Improved Venture Selection vs. The Value-Added Hypothesis. Journal of Economics and Management Strategy 11(3): 423-452.

Brettel M. 2003. Business Angels in Germany: a Research Note. Venture Capital: An International Journal of Entrepreneurial Finance 5(3): 251-268.

Chemmanur TJ, Chen Z. 2006. Venture Capitalists Versus Angels: the Dynamics of Private Firm Financing Contracts.

Das SR, Jo H, Kim Y. 2011. Polishing Diamonds in the Rough: the Sources of Syndicated Venture Performance. Journal of Financial Intermediation 20(2): 199-230.

Davila A, Foster G, Gupta M. 2003. Venture Capital Financing and the Growth of Startup Firms. Journal of Business Venturing 18(6): 689-708.

Page | 39

Du Q. 2009. Birds of a Feather or Celebrating Differences? the Formation and Impact of Venture Capital Syndication. Unpublished Working Paper.

Dushnitsky G, Lenox MJ. 2005. When Do Firms Undertake R&D by Investing in New Ventures? Strategic Management Journal 26(10): 947-965.

Dushnitsky G, Shapira Z. 2010. Entrepreneurial Finance Meets Organizational Reality: Comparing Investment Practices and Performance of Corporate and Independent Venture Capitalists. Strategic Management Journal 31(9): 990-1017.

Ehrlich SB, De Noble AF, Moore T, Weaver RR. 1994. After the Cash Arrives: a Comparative Study of Venture Capital and Private Investor Involvement in Entrepreneurial Firms. Journal of Business Venturing 9(1): 67-82.

Engel D, Keilbach M. 2007. Firm-Level Implications of Early Stage Venture Capital Investment - an Empirical Investigation. Journal of Empirical Finance 14(2): 150-167.

Freear J, Sohl J, Wetzel W. 1991. Raising Venture Capital to Finance Growth. Frontiers of Entrepreneurship Research.

Freear J, Sohl JE, Wetzel WE. 1994. Angels and Non-Angels: Are There Differences? Journal of Business Venturing 9(2): 109-123.

Freear J, Sohl JE, Wetzel WE. 1995. Angels: Personal Investors in the Venture Capital Market. Entrepreneurship and Regional Development 7(1): 85-94.

Gans JS, Hsu DH, Stern S. 2002. When Does Start-Up Innovation Spur the Gale of Creative Destruction? Rand Journal of Economics 33(4): 571+.

Giannetti M, Yafeh Y. 2012. Do Cultural Differences Between Contracting Parties Matter? Evidence from Syndicated Bank Loans. Management Science 58(2): 365-383.

Goldfarb B, Hoberg G, Kirsch D, Triantis A. 2009. Does Angel Participation Matter? an Analysis of Early Venture Financing. Unpublished Working Paper.

Gompers P, Lerner J. 2000a. The Determinants of Corporate Venture Capital Success: Organizational Structure, Incentives, and Complementarities. In Concentrated corporate ownership. University of Chicago Press.

Gompers P, Lerner J. 2000b. Money Chasing Deals? the Impact of Fund Inflows on Private Equity Valuation. Journal of Financial Economics 55(2): 281-325.

Gompers PA. 1996. Grandstanding in the Venture Capital Industry. Journal of Financial Economics 43: 133-156.

Page | 40

Guler I, McGahan A. 2007. The More the Merrier? Institutions and Syndication Size in International Venture Capital Investments, Working paper, University of Toronto, Rotman School.

Harrison R, Mason C. 2000. Venture Capital Market Complementarities: the Links Between Business Angels and Venture Capital Funds in the UK. Venture Capital: An International Journal of Entrepreneurial Finance 2(3): 223-242.

Harrison RT, Mason CM. 1992a. International Perspectives on the Supply of Informal Venture Capital. Journal of Business Venturing 7(6): 459-475.

Harrison RT, Mason CM. 1992b. The Role of Investors in Entrepreneurial Companies: a Comparison of Informal Investors and Venture Capitalists. University of Southampton, Department of Economics.

Hellmann T, Puri M. 2000. The Interaction Between Product Market and Financing Strategy: The Role of Venture Capital. Review of Financial Studies 13(4): 959-984.

Hellmann T, Puri M. 2002. Venture Capital and the Professionalization of Start-Up Firms: Empirical Evidence. The Journal of Finance 57(1): 169-197.

Hochberg YV, Ljungqvist A, Lu Y. 2007. Whom You Know Matters: Venture Capital Networks and Investment Performance. The Journal of Finance 62(1): 251-301.

Hsu DH. 2004. What Do Entrepreneurs Pay for Venture Capital Affiliation? The Journal of Finance 59(4): 1805-1844.

Ivanov VI, Xie F. 2010. Do Corporate Venture Capitalists Add Value to Start-Up Firms? Evidence from IPOs and Acquisitions of VC Backed Companies. Financial Management 39(1): 129-152.

Johnson W, Sohl J. 2011. Initial Public Offerings and Pre-IPO Shareholders: Angels Versus Venture Capitalists. In Unpublished Working Paper.

Kanter RM. 1977. Men and Women of the Corporation, Basic Books: New York.

Katila R, Rosenberger JD, Eisenhardt KM. 2008. Swimming with Sharks: Technology Ventures, Defense Mechanisms and Corporate Relationships. Administrative Science Quarterly 53(2): 295-332.

Kortum S, Lerner J. 2000. Assessing the Contribution of Venture Capital to Innovation. RAND Journal of Economics 31(4): 674-692.

Lee PM, Wahal S. 2004. Grandstanding, Certification and the Underpricing of Venture Capital Backed IPOs. Journal of Financial Economics 73(2): 375-407.

Lumme A, Mason C, Suomi M. 1998. Informal Venture Capital: Investors, Investments and Policy Issues in Finland. Kluwer Academic Publishers Boston.

Page | 41

Madill JJ, Haines JGH, Riding AL. 2005. The Role of Angels in Technology SMEs: A Link to Venture Capital. Venture Capital: An International Journal of Entrepreneurial Finance 7(2): 107-129.

Manigart S, Van Hyfte W. 1999. Post-investment Evolution of Belgian Venture Capital Backed Companies: an Empirical Study. Paper presented at the Babson Entrepreneurship Conference.

Mason CM, Harrison RT. 1996. Informal Venture Capital: a Study of the Investment Process, the Post-Investment Experience and Investment Performance. Entrepreneurship and Regional Development 8(2): 105-126.

Maula VJM, Murray G. 2001. Corporate Venture Capital and the Creation of US Public Companies: the Impact of Sources of Venture Capital on the Performance of Portfolio Companies. In Creating Value: Winners in the New Business Environment. Hitt M, Amit R, Lucier C, Nixon R (eds.), Blackwell: Oxford.

Park HD, Steensma HK. 2012. When Does Corporate Venture Capital Add Value for New Ventures? Strategic Management Journal 33(1): 1-22.

Paul S, Whittam G, Johnston JB. 2003. The Operation of the Informal Venture Capital Market in Scotland. Venture Capital: An International Journal of Entrepreneurial Finance 5(4): 313-335.

Penrose ET. 1959. The Theory of the Growth of the Firm. Basil Blackwell: Oxford.

Prahalad CK, Hamel G. 1990. The Core Competence of the Corporation. Harvard Business Review 68(3): 79-91.

Rosener JB. 1995. America’s Competitive Secret: Utilizing Women as a Management Strategy. Oxford University Press: New York.

SæTre A. 2003. Entrepreneurial Perspectives On Informal Venture Capital. Venture Capital: An International Journal of Entrepreneurial Finance 5(1): 71-94.

Schwartz-Ziv M. 2012. Does the Gender of Directors Matter? In Unpublished Working Paper.

Schwienbacher A. 2012. The Entrepreneur's Investor Choice: The Impact on Later-Stage Firm Development. In Unpublished Working Paper.

Shrader CB, Blackburn VB, Iles P. 1997. Women in Management and Firm Financial Performance: an Exploratory Study. Journal of Managerial Issues: 355-372.

Sørensen M. 2007. How Smart Is Smart Money? a Two Sided Matching Model of Venture Capital. The Journal of Finance 62(6): 2725-2762.

Page | 42

Sørheim R. 2005. Business Angels as Facilitators for Further Finance: an Exploratory Study. Journal of Small Business and Enterprise Development 12(2): 178-191.

Stevenson H, Coveney P. Year. A Survey of Business Angels: Fallacies Corrected and Six Distinct Types of Angel Identified. In Proceedings of the National Small Firms Policy and Research Conference.

Tashiro Y. 1999. Business Angels in Japan. Venture Capital: An International Journal of Entrepreneurial Finance 1(3): 259-273.

Teece DJ. 1986. Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing and Public Policy.

Tian X. 2011. The Role of Venture Capital Syndication in Value Creation for Entrepreneurial Firms. Review of Finance 16(1): 245-283.

Tripsas M. 1997. Unraveling the Process of Creative Destruction: Complementary Assets and Incumbent Survival in the Typesetter Industry. Strategic Management Journal 18(1): 119-142.

Van Osnabrugge M, Robinson RJ. 2000. Angel Investing: Matching Startup Funds with Startup Companies-The Guide for Entrepreneurs and Individual Investors. Jossey-Bass.

Van Osnabrugge MS. 1998. The Financing of Entrepreneurial Firms in the UK: a Comparison of Business Angel and Venture Capitalist Investment Procedures, University of Oxford.

Wetzel W. 1983. Angels and Informal Risk Capital. Sloan Management Review 24(4): 23-34.

White H. 1980. A Heteroskedasticity-Consistent Covariance Matrix Estimator and a Direct Test for Heteroskedasticity. Econometrica: Journal of the Econometric Society: 817-838.

Page | 43

Table I

Descriptive Statistics By Investor Type

Variables Angels CVCs

Finance Related

Companies

Incubators

Industry Related

Companies Other VCs

Average Amount Invested (M. USD)

mean 10.4 26.3 14.7 10.0 9.34 10.3 37.5 N 347 36 352 21 331 27 360

Foreign Dummy mean 0.22 0.85 0.69 0.029 0.44 0.60 0.67 N 967 60 859 35 775 43 601

Percent of Companies Written-off

mean 0.15 0.19 0.17 0.11 0.22 0.19 0.15 N 1,003 60 879 35 982 55 601

Percent of Companies Exits

mean 0.30 0.44 0.41 0.074 0.30 0.34 0.35 N 1,003 60 879 35 982 55 601

Total Activity Times in years

mean 1.53 4.24 1.80 3.16 1.36 2.93 3.95 N 1,003 60 879 35 982 55 601

Total Number Of Companies Invested

mean 1.80 5.98 2.69 5.17 1.76 4.67 5.83 N 1,003 60 879 35 982 55 601

Total Number Of Follow up Investment

mean 0.57 3.67 0.99 0.60 0.54 0.33 4.69 N 1,003 60 879 35 982 55 601

Total Number Of Rounds Participated

mean 2.37 9.65 3.67 5.77 2.30 5 10.5 N 1,003 60 879 35 982 55 601

The data in this table is based on the sample of 1003 angels ;60 CVCs; 601 VCs (405 Foreign VCs and 196 Local VCs); 35 incubators; 982 industry-related companies ; 879 finance-related companies; and 55 other investor types. 'Average Amount Invested' is the average amount in USD in 2008 values. 'Foreign Dummy' is a dummy variable that equals 1 if the investor is of foreign origin. 'Percent of Companies Written-off' is the number of companies closed out of total number of companies invested. 'Percent of Companies Exits' off' is the number of companies exited (IPO or M&A) out of total number of companies invested. 'Total Activity Times in years' is the time passed from first to last investment. 'Total Number Of Companies Invested' is the number of companies invested. 'Total Number Of Follow up Investment' is the total number of companies receiving additional investment from the same investor. 'Total Number Of Rounds Participated' is the total number of rounds a specific investor participated in.

Page | 44

Table II

Descriptive Statistics of Companies Variables N Mean SD Min Max

Company Stage

Seed 1,950 0.064 0.24 0 1

R&D 1,950 0.34 0.47 0 1

Initial Revenues 1,950 0.45 0.5 0 1

Revenue Growth 1,950 0.15 0.35 0 1

Industry

Miscellaneous Technologies 2,377 0.093 0.29 0 1

IT & Enterprise Software 2,377 0.24 0.43 0 1

Communications 2,377 0.2 0.4 0 1

Life Sciences 2,377 0.24 0.43 0 1

Semiconductors 2,377 0.059 0.24 0 1

Clean-tech 2,377 0.05 0.22 0 1

Internet 2,377 0.12 0.32 0 1

Geographic Location Of Main Offices

Tel Aviv 2,305 0.12 0.33 0 1

Center 2,305 0.45 0.5 0 1

Jerusalem 2,305 0.069 0.25 0 1

North 2,305 0.081 0.27 0 1

Haifa 2,305 0.054 0.23 0 1

South 2,305 0.032 0.18 0 1

West Bank 2,305 0.0043 0.066 0 1

Abroad 2,305 0.18 0.39 0 1

Investments

Originated From Incubators Dummy 2,409 0.18 0.38 0 1 Number of Rounds 2,409 2.24 1.69 1 14 Number of Investors 2,409 4.02 3.84 1 33 Number of Investor Types 2,409 2.11 1.17 1 6 Total Number of Potential Investors 2,409 6.12 7.74 1 66

Performance

Exit Dummy 2,409 0.22 0.42 0 1

Years To Exit 535 5.75 2.98 0.18 16

Closure Dummy 2,409 0.3 0.46 0 1

Years To Closure 717 4.15 2.41 0.13 14.4

Amount Of Exit 405 152 592 0.28 11,275

The data in this table is based on the sample of 2409 startups and the unit of analysis is a single startup. Industry is the dummy variable indicating the industry of the startup. 'Geographic Location Of Main Offices ' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. 'Number of Rounds' is the total number of rounds the company had. 'Number of Investors' is the accumulated 'number of investors' that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investor Types' is the total number of different types of investors a company got investment from. 'Entropy of The Total 'number of investors' is a measure calculated based on the entropy calculation presented in the paper. 'Exit Dummy' is a dummy variable indicating if the company exited (IPO or M&A). 'Years To Exit' is the total time in years from the company inception until its exit. 'Write-off Dummy' is a dummy variable indicating if the company was closed. 'Years To Write-off' is the total time in years from the company inception until it was closed. 'Amount Of Exit' is the total amount of exit either trough a IPO or M&A that a company got.

Page | 45

Table III

Correlation Matrix

Number of Investors

Number of Investors2

Number of Investor

Types

Number of Investor Types2

Number of Investors 1 Number of Investors2 0.903*** 1

Number of Investor Types 0.769*** 0.556*** 1

Number of Investor Types2 0.771*** 0.594*** 0.974*** 1

* p<0.05, ** p<0.01, *** p<0.001

The data in this table is based on the sample of 2409 startups and the unit of analysis is a single startup. 'Number of investors' is the accumulated number of investors that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investors2' is the same number squared. 'Number of Investor Types' is the total number of different types of investors a company got investment from. 'Number of Investor Types2' is the same number squared.

Page | 46

Table IV

Exit and Write-off Prospects by the 'Number of Investors' and by the 'Number Of Investor Types' Number of Investors Types of Investors

Treatment Probit Probit

Dependent Variable Exit Dummy Exit Dummy Failure Dummy

Failure Dummy

Exit Dummy

Exit Dummy Failure Dummy

Failure Dummy

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

Endogeneity Control X X X X X X X X

Number of Investors 0.082*** 0.15*** -0.050*** -0.10***

0.013 0.023 0.017 0.027

Number of Investors2

-0.0033***

0.0033**

0.00095 0.0013

Number of Investor Types

0.16*** 0.24* -0.092** -0.25**

0.037 0.12 0.037 0.13

Number of Investor Types2

-0.015

0.033

0.023 0.025

Industry

Communications 0.51** 0.50** 0.36** 0.36** 0.56*** 0.56*** 0.34** 0.34**

0.2 0.2 0.16 0.16 0.2 0.2 0.16 0.16

IT & Enterprise Software 0.61*** 0.60*** 0.24 0.24 0.64*** 0.64*** 0.23 0.23

0.2 0.2 0.15 0.15 0.2 0.2 0.16 0.15

Internet 0.25 0.25 0.52*** 0.52*** 0.34 0.34 0.49*** 0.50***

0.21 0.22 0.17 0.17 0.22 0.22 0.17 0.17

Life Sciences 0.36* 0.35* 0.044 0.045 0.40** 0.40** 0.033 0.035

0.2 0.2 0.15 0.15 0.2 0.2 0.15 0.15

Miscellaneous Technologies 0.23 0.22 0.057 0.06 0.23 0.23 0.062 0.071

0.21 0.22 0.17 0.17 0.22 0.22 0.17 0.17

Semiconductors 0.46** 0.44* 0.38* 0.38* 0.52** 0.52** 0.36* 0.36*

0.22 0.22 0.2 0.2 0.23 0.23 0.2 0.2

Other Controls

Number of Rounds -0.037 -0.057* -0.24*** -0.22*** 0.03 0.031 -0.28*** -0.28***

0.029 0.029 0.037 0.037 0.025 0.025 0.032 0.032

Originated From Incubators Dummy -0.47*** -0.47*** 0.031 0.031 -0.50*** -0.51*** 0.047 0.055

0.11 0.11 0.091 0.091 0.11 0.11 0.091 0.091

Headquarter Location Dummies √ √ √ √ √ √ √ √

Year Dummies √ √ √ √ √ √ √ √

Constant 0.86** 0.75* -5.29*** -5.18*** 0.65 0.57 -4.94*** -4.79***

0.44 0.44 0.18 0.2 0.44 0.46 0.19 0.22

Observations 2,278 2,278 2,278 2,278 2,278 2,278 2,278 2,278

Page | 47

Table IV

Exit and Write-off Prospects by the 'Number of Investors' and by the 'Number Of Investor Types' Number of Investors Types of Investors

Treatment Probit Probit

Dependent Variable Exit Dummy Exit Dummy Failure Dummy

Failure Dummy

Exit Dummy

Exit Dummy Failure Dummy

Failure Dummy

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

Endogeneity Control X X X X X X X X

Prob > chi2 0 0 0 0 0 0 0 0 Pseudo R2 0.18 0.19 0.18 0.18 0.17 0.17 0.18 0.18

The regressions in this table are based on the sample of 2409 startups and the unit of analysis is a single startup. 'Number of Investors' is the accumulative 'number of investors' that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investors2' is the same variable squared. 'Number of Investor Types' is the accumulative number of investor types that invested in that specific company (if the company has two investments from the same type of investor, it was counted only once). 'Number of Investor Types2' is the same variable squared. The dependent variable is 'Exit Dummy' or 'Write-off Dummy,' which equal 1 if Exited/Written-off and 0 if it was not. Industry is the dummy variable indicating the industry of the startup. 'Headquarter Location Dummies' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. In industries dummies, the missing dummy is 'Miscellaneous Technologies Dummy.’ The 'office dummies' indicate where the main office of the company is located; the company may have additional offices elsewhere. 'Number of Rounds' is the total number of rounds the company had. The symbols ***, **, and * denotes significance at the 1%, 5%, and 10% level (two-sided), respectively.

Page | 48

Table V 'Time To Exit' by 'Number of Investors' and by the 'Number Of Investor Types'

Number of Investors Types of Investors

Treatment Hazard Models Hazard Models

Dependent Variable Time to

Exit Time to

Exit Time to

Exit Time to

Exit

Labels Model 1 Model 2 Model 3 Model 4

Endogeneity Control X X X X

Number of Investors -0.028** -0.076**

0.014 0.031

Number of Investors2 0.0021* 0.0012

Number of Investor Types

-0.100* -0.36**

0.052 0.17

Number of Investor Types2

0.048

0.029

Industry

Communications 1.07*** 1.09*** 1.07*** 1.06***

0.39 0.39 0.39 0.39

IT & Enterprise Software 0.90** 0.92** 0.90** 0.89**

0.38 0.38 0.38 0.38

Internet 0.96** 0.96** 0.89** 0.85**

0.42 0.41 0.41 0.41

Life Sciences 0.73* 0.74* 0.73* 0.69*

0.38 0.38 0.38 0.38

Miscellaneous Technologies 0.81* 0.82** 0.81* 0.80*

0.41 0.41 0.41 0.41

Semiconductors 1.08** 1.11*** 1.07** 1.06**

0.42 0.42 0.42 0.42

Other Controls

Number of Rounds -0.16*** -0.14*** -0.17*** -0.17***

0.039 0.04 0.036 0.036

Originated From Incubators Dummy 0.058 0.061 0.069 0.072

0.2 0.2 0.2 0.2

Headquarter Location Dummies √ √ √ √

Observations 457 457 457 457 Prob > chi2 0 0 0 0 LR chi2 82.7 85.7 82.4 84.9

The Hazard models in this table are based on the sample of 2409 startups s and the unit of analysis is a single startup. 'Number of Investors' is the accumulative 'number of investors' that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investors2' is the same variable squared. 'Number of Investor Types' is the accumulative number of investor types that invested in that specific company (if the company has two investments from the same type of investor, it was counted only once). 'Number of Investor Types2' is the same variable squared. The dependent variable is 'Exit Dummy' or 'Write-off Dummy,' which equal 1 if Exited/Written-off and 0 if it was not. Industry is the dummy variable indicating the industry of the startup. 'Headquarter Location Dummies' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. In industries dummies, the missing dummy is 'Miscellaneous Technologies Dummy.’ The 'office dummies' indicate where the main office of the company is located; the company may have additional offices elsewhere. 'Number of Rounds' is the total number of rounds the company had. The symbols ***, **, and * denotes significance at the 1%, 5%, and 10% level (two-sided), respectively.

Page | 49

Table VI

Robustness Tests for Exit and Write-off Prospects by the 'Number of Investors'

Robustness for Number of Investors

Treatment Probit with IV 2SLS Types of Investors = 1

Dependent Variable Exit

Dummy Failure Dummy

Exit Dummy Failure Dummy

Exit Dummy

Failure Dummy

Failure Dummy

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Endogeneity Control √ √ √ √ X X X

Number of Investors 0.17*** -0.055 0.054*** -0.035*** 0.63*** -0.41** -0.41**

0.03 0.039 0.012 0.007 0.19 0.19 0.18

Number of Investors2 -0.0050*** 0.0018 -0.0016*** 0.0013*** -0.073** 0.05 0.050*

0.0013 0.0021 0.00059 0.00032 0.032 0.032 0.029

Industry

Communications 0.46** 0.35** 0.12*** 0.10** 0.42 0.41* 0.41*

0.22 0.16 0.045 0.044 0.37 0.22 0.23

IT & Enterprise Software 0.50** 0.24 0.13*** 0.068 0.31 0.37* 0.37*

0.21 0.16 0.043 0.043 0.37 0.21 0.22

Internet 0.19 0.51*** 0.044 0.15*** 0.19 0.56** 0.56**

0.24 0.17 0.048 0.048 0.39 0.23 0.24

Life Sciences 0.27 0.035 0.059 0.014 0.31 0.15 0.15

0.21 0.15 0.04 0.042 0.37 0.2 0.21

Miscellaneous Technologies 0.15 0.056 0.019 0.015 0.35 0.0054 0.0054

0.23 0.17 0.046 0.048 0.38 0.24 0.24

Semiconductors 0.41* 0.36* 0.10* 0.098* 0.39 0.34 0.34

0.24 0.19 0.057 0.052 0.44 0.29 0.3

Other Controls

Number of Rounds -0.049 -0.27*** -0.019* -0.049*** 0.11 -0.14* -0.14

0.034 0.043 0.011 0.0075 0.096 0.087 0.095

Originated From Incubators Dummy -0.48*** 0.033 -0.10*** 0.0076 -0.61*** 0.038 0.038

0.13 0.092 0.026 0.025 0.23 0.13 0.14

Headquarter Location Dummies √ √ √ √ √ √ √

Year Dummies √ √ √ √ √ √ √

Constant 0.76* -5.3 0.73*** 0.16** -0.34 -5.19*** -5.19

0.45 144 0.085 0.063 0.73 0.34 142

Observations 1,666 2,278 1,666 2,278 887 926 926 Prob > chi2 0 0 0 0 0 0 0 Pseudo R2

0.17 0.188 0.18 0.16 0.16

First Stage Probit

Dependent Variable Number of Investors

Number of Investors

Number of Investors

Number of Investors

Number of Potential Investors 0.643 0.657 0.643 0.657

0 0 0 0

Page | 50

Table VI

Robustness Tests for Exit and Write-off Prospects by the 'Number of Investors'

Robustness for Number of Investors

Treatment Probit with IV 2SLS Types of Investors = 1

Dependent Variable Exit

Dummy Failure Dummy

Exit Dummy Failure Dummy

Exit Dummy

Failure Dummy

Failure Dummy

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Endogeneity Control √ √ √ √ X X X

Number of Potential Investors2 -0.004 -0.005 -0.004 -0.005

0 0 0 0

Industry Dummies √ √ √ √

Number of Rounds √ √ √ √

Originated From Incubators Dummy √ √ √ √

Headquarter Location Dummies √ √ √ √

Year Dummies √ √ √ √

Observations 1666 2,278 1666 2278

Prob > F 0 0 0 0

Adj R-squared 0.861 0.863 0.864 0.865

Dependent Variable Number of Investors2

Number of Investors2

Number of Investors2

Number of Investors2

Total Number of Potential Investors 5.135 5.005 5.135 5.006

0 0 0 0

Total Number of Potential Investors2 0.093 0.096 0.093 0.096

0 0 0.001 0

Industry Dummies √ √ √ √

Number of Rounds √ √ √ √

Originated From Incubators Dummy √ √ √ √

Headquarter Location Dummies √ √ √ √

Year Dummies √ √ √ √

Observations 1666 2,278 1666 2278

Prob > F 0 0 0 0

Adj R-squared 0.736 0.742 0.741 0.745

The regressions in this table are based on the sample of 2409 startups s and the unit of analysis is a single startup. 'Number of Investors' is the accumulative 'number of investors' that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investors2' is the same variable squared. The dependent variable is 'Exit Dummy' or 'Write-off Dummy,' which equal 1 if Exited/Written-off and 0 if it was not. Industry is the dummy variable indicating the industry of the startup. 'Headquarter Location Dummies' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. In industries dummies, the missing dummy is 'Miscellaneous Technologies Dummy.’ The 'office dummies' indicate where the main office of the company is located; the company may have additional offices elsewhere. 'Number of Rounds' is the total number of rounds the company had. The symbols ***, **, and * denotes significance at the 1%, 5%, and 10% level (two-sided), respectively.

Page | 51

Table VII

Robustness Tests for Exit and Write-off Prospects by the 'Number Of Investor Types'

Dependent Variable Exit

Dummy Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Treatment Probit with IV 2SLS Number of investors

<= 6 Probit Probit with IV 2SLS

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12

Endogeneity Control √ √ √ √ X X X X √ √ √ √

Number of Investor Types 0.40*** -0.16 0.15*** -0.050* 0.084* -0.085*

0.12 0.15 0.047 0.028 0.049 0.044

Number of Investor Types2

Herfindahl of Number of Investors

-0.45*** 0.34*** -0.45*** 0.38*** -0.12*** 0.15***

0.14 0.13 0.15 0.13 0.042 0.037

Industry

Communications 0.50** 0.35** 0.13*** 0.097** 0.52** 0.36** 0.57*** 0.34** 0.51** 0.34** 0.13*** 0.095**

0.22 0.16 0.047 0.045 0.22 0.16 0.2 0.16 0.22 0.16 0.045 0.045

IT & Enterprise Software 0.55** 0.23 0.14*** 0.062 0.58*** 0.27* 0.64*** 0.23 0.53** 0.23 0.13*** 0.063

0.21 0.16 0.046 0.043 0.22 0.16 0.2 0.16 0.21 0.16 0.043 0.043

Internet 0.27 0.49*** 0.064 0.15*** 0.14 0.57*** 0.34 0.50*** 0.27 0.49*** 0.059 0.15***

0.24 0.17 0.051 0.048 0.24 0.17 0.22 0.17 0.24 0.17 0.049 0.048

Life Sciences 0.32 0.037 0.068 0.011 0.29 0.055 0.41** 0.032 0.34 0.029 0.070* 0.01

0.21 0.15 0.043 0.042 0.22 0.15 0.2 0.15 0.21 0.15 0.04 0.042

Miscellaneous Technologies 0.18 0.059 0.024 0.012 0.24 0.11 0.22 0.069 0.15 0.064 0.013 0.015

0.23 0.17 0.05 0.048 0.23 0.18 0.22 0.17 0.23 0.17 0.047 0.048

Semiconductors 0.47* 0.36* 0.11* 0.087* 0.34 0.53** 0.53** 0.36* 0.51** 0.36* 0.13** 0.086

0.24 0.19 0.06 0.053 0.26 0.21 0.23 0.2 0.24 0.19 0.058 0.053

Other Controls

Number of Rounds -0.082 -0.25*** -0.035* -0.050*** 0.061 -0.27*** 0.060*** -0.28*** 0.055** -0.28*** 0.018** -0.059***

0.057 0.069 0.021 0.013 0.038 0.043 0.022 0.03 0.023 0.03 0.0073 0.0055

Originated From Incubators Dummy -0.51*** 0.057 -0.11*** 0.017 -0.54*** 0.061 -0.50*** 0.057 -0.53*** 0.059 -0.12*** 0.019

0.13 0.094 0.027 0.025 0.13 0.095 0.11 0.091 0.13 0.092 0.026 0.025

Headquarter Location Dummies √ √ √ √ √ √ √ √ √ √ √ √

Year Dummies √ √ √ √ √ √ √ √ √ √ √ √

Constant 0.42 -4.84 0.61*** 0.17** 0.86* -4.86*** 1.22*** -5.35*** 1.32*** -5.39 0.89*** -0.021

0.47 83.2 0.1 0.074 0.48 0.21 0.45 0.24 0.47 82.9 0.087 0.069

Observations 1,666 2,278 1,666 2,278 1,876 1,876 2,278 2,278 1,666 2,278 1,666 2,278

Page | 52

Table VII

Robustness Tests for Exit and Write-off Prospects by the 'Number Of Investor Types'

Dependent Variable Exit

Dummy Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Exit Dummy

Failure Dummy

Treatment Probit with IV 2SLS Number of investors

<= 6 Probit Probit with IV 2SLS

Labels Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9 Model 10 Model 11 Model 12

Endogeneity Control √ √ √ √ X X X X √ √ √ √

Prob > chi2 0 0 0 0 0 0 0 0 0 0 0 0 Pseudo R2

0.11 0.18 0.17 0.16 0.17 0.18

0.15 0.18

First Stage Probit

Dependent Variable Number

of Investors

Number of Investors

Number of Investors

Number of Investors

Number of Investors

Number of Investors

Number of Investors

Number of

Investors

Total Number of Potential Investors 0.056 0.06 0.056 0.06

0 0 0 0

Herfindahl of Total Potential Investors 0.978 0.982 0.978 0.982

0 0 0 0

Industry Dummies √ √ √ √

√ √ √ √

Number of Rounds √ √ √ √

√ √ √ √

Originated From Incubators Dummy √ √ √ √

√ √ √ √

Headquarter Location Dummies √ √ √ √

√ √ √ √

Year Dummies √ √ √ √

√ √ √ √

Observations 1666 2278 1666 2278

1666 2278 1666 1666

Prob > F 0 0 0 0

0 0 0 0

Adj R-squared 0.519 0.509 0.527 0.516

0.978 0.981 0.978 0.981

The regressions in this table are based on the sample of 2409 startups s and the unit of analysis is a single startup. 'Number of Investor Types' is the accumulative number of investor types that invested in that specific company (if the company has two investments from the same type of investor, it was counted only once). 'Number of Investor Types2' is the same variable squared. The dependent variable is 'Exit Dummy' or 'Write-off Dummy,' which equal 1 if Exited/Written-off and 0 if it was not. Industry is the dummy variable indicating the industry of the startup. 'Headquarter Location Dummies' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. In industries dummies, the missing dummy is 'Miscellaneous Technologies Dummy.’ The 'office dummies' indicate where the main office of the company is located; the company may have additional offices elsewhere. 'Number of Rounds' is the total number of rounds the company had. The symbols ***, **, and * denotes significance at the 1%, 5%, and 10% level (two-sided), respectively.

Page | 53

Table VIII

Robustness Tests for 'Time To Exit' by 'Number of Investors' and by the

'Number Of Investor Types'

Number

of Investors

Types of Investor

Types

Number of

Investors

Types of Investor

Types

Types of Investor

Types

Treatment IV Proxy IV Proxy Types of

Investors = 1

Hazard Models

IV Proxy

Dependent Variable Time to

Exit Time to

Exit Time to

Exit Time to

Exit Time to

Exit

Labels Model 1 Model 2 Model 3 Model 4 Model 5

Endogeneity Control √ √ X X √

Number of Investors -0.54* 0.32

Number of Investors2

0.082*

0.05

Residual Of The Number of Investors -0.11***

0.039

Residual Of The Number of Investors2

0.0032*

0.0018

Residual Of The Number of Investor Types -0.36**

0.14

Herfindahl Of The Number of Investors

0.39*

0.21

Residual Of The Herfindahl Of The Number of Investors

0.39*

0.21

Industry

Communications 1.10*** 1.07*** 1.06 1.05*** 1.06***

0.39 0.39 0.81 0.39 0.39

IT & Enterprise Software 0.93** 0.89** 0.67 0.88** 0.90**

0.38 0.38 0.8 0.38 0.38

Internet 1.00** 0.95** 0.59 0.86** 0.87**

0.41 0.41 0.83 0.41 0.41

Life Sciences 0.75** 0.73* 0.54 0.70* 0.71*

0.38 0.38 0.76 0.38 0.38

Miscellaneous Technologies 0.81* 0.80* 1.1 0.80* 0.82**

0.42 0.42 0.91 0.41 0.41

Semiconductors 1.11*** 1.04** 0.82 1.05** 1.06**

0.42 0.42 0.89 0.42 0.42

Other Controls

Number of Rounds -0.10** -0.031 -0.1 -0.19*** -0.18***

0.047 0.075 0.17 0.032 0.032

Originated From Incubators Dummy 0.068 0.096 0.19 0.061 0.076

0.2 0.2 0.56 0.2 0.2

Headquarter Location Dummies √ √ √ √ √

Observations 457 457 109 457 457 Prob > chi2 0 0 0.6656 0 0 LR chi2 88.3 85.7 13.1 82 82

The Hazard models in this table are based on the sample of 2409 startups s and the unit of analysis is a single startup. 'Number of Investors' is the accumulative 'number of investors' that invested in that specific company (if the company has two investments from the same investor, it was counted only once). 'Number of Investors2' is the same variable squared. 'Number of Investor Types' is the accumulative number of investor types that invested in that specific company (if the company has two investments from the same type of investor, it was counted only once). 'Number of Investor Types2' is the same variable squared. The dependent variable is 'Exit Dummy' or 'Write-off Dummy,' which equal 1 if Exited/Written-off and 0 if it was not. Industry is the dummy variable indicating the industry of the startup. 'Headquarter Location Dummies' is the location of the main office of the startup. 'Originated From Incubators Dummy' is a dummy variable indicating whether the company originates in a governmental program. In industries dummies, the missing dummy is 'Miscellaneous Technologies Dummy.’ The 'office dummies' indicate where the main office of the company is located; the company may have additional offices elsewhere. 'Number of Rounds' is the total number of rounds the company had. The symbols ***, **, and * denotes significance at the 1%, 5%, and 10% level (two-sided), respectively.

Page | 54