SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

Managing the different companies in which they invest while at the same time performing portfolio optimization for themselves, venture capitalists position themselves as a pure-play or diversified conglomerate through their cumulative portfolios. I examine the effects of two investment strategies of venture capitalists: 1) a specialist “pure-play” strategy that maximizes venture capital involvement and 2) a more generalist strategy of diversification at the “firm” level that minimizes portfolio risk. I find that neither strategy optimizes both venture capital growth and time to entrepreneurial exit, which highlights a need for institutional investors to clarify fund objectives at the time a fund is established.

Venture capitalists (VCs) spend a lot of time deciding in which (and how many) entrepreneurial ventures to invest, since this is the basis for their business. As part of this function, VCs must determine which criteria they will use to choose the investments they will make. For individual investments, the analysis is straightforward but laborious. The VC determines whether an investment is worthy based on the likelihood of its profitability and survivorship. VCs bet that these portfolio companies (PCs) will not only last but will also be profitable. Historically, 50% of VC profits come from only 7% of investments. Up to 33% of venture capital investments result in losses, with as many as 15% going broke, that is, a −100% return (National Venture Capital Association, 2004).

With statistics like this, it is easy to understand why VCs would be inclined to diversify. Indeed, in its 2004 report, the National Venture Capital Association notes that a typical VC will raise funds that

… may be similar to other funds in the firm. However, the firm may have one fund with a specific focus and another with a different focus and yet another with a broadly diversified portfolio. This depends on the strategy and focus of the venture firm itself.

Since VCs make several investments within a portfolio, they must also make decisions across investments, effectively choosing the composition of their portfolio. Although similar to the decisions made by portfolio fund managers, investment composition decisions are not as clear-cut for VCs.1 Entrepreneurs often prefer venture capital to angel capital based on the active management role VCs take (Hellmann, 2000; Hellmann and Puri, 2000, 2002).2 This expertise component in the financing arrangement is advantageous for entrepreneurs who excel in the service offered by their company (e.g., technology expertise for a high-tech company entrepreneur) but lack the business savvy to succeed in the critical early years (Kaplan and Strömberg, 2001).

The resulting relationship between the VC and the entrepreneur creates an organization that is not unlike a multidivisional firm. Thus, the VC's investment decisions must be made both intrafund and interfund, and must deal with the scope of the conglomerate those decisions create. If the newly formed group consists of PCs and/or funds that are similar in their characteristics, the VC forms a pure-play strategy. This type of VC is called a “specialist investor.” Alternatively, if PCs and/or funds are heterogeneous in nature, the VC is termed a “generalist investor” and forms a diversification strategy. VCs can diversify in several ways: 1) geographically (both domestically and internationally), 2) development stage, and 3) industry. The potential benefit of diversification is similar across all forms, one of risk reduction (Markowitz, 1952; Sharpe, 1963, 1964). That said, the optimal diversification strategy of an individual VC will likely vary across VC specifics such as expertise. The potential costs of diversification are based on the limited nature of resources: 1) time and 2) expertise. Depending upon the extent of the costs and benefits, the VC should adopt a specialist or generalist strategy.

In this paper, I examine the optimal strategy of VCs in their dual roles: 1) general manager to the portfolio companies in which they invest and 2) portfolio fund manager to both themselves and the limited partners.3 I examine how the composition of a VC's portfolio impacts the probability of success for the PCs in which they invest.

The findings suggest that although VCs benefit with regard to their growth using a diversification strategy (a 3.5% to 7.5% increase in growth for a one standard deviation increase in diversification), the time to exit of the PCs in which they invest is delayed by such a strategy (a 0.5% to 2% decrease in the probability of PC exit with a one standard deviation increase in diversification). Examinations control for potential grandstanding by young VCs in order to guarantee funding for future funds (Gompers, 1996). These results may be of interest to several different parties. Institutional investors may use these results when creating covenants that can limit the diversification of venture capital fund managers (see Gompers and Lerner, 1996 for a discussion of these covenants). To the extent that VCs are concerned with an expeditious exit of PCs, they might wish to consider a relatively pure-play strategy (e.g., investing in one industry, one stage, one international region, etc.). PCs that utilize VCs for their active management role or as a swift vehicle to the public markets need to consider the level of diversification potential VCs practice (this discrimination is not always possible, depending upon the available supply of venture capital).

This paper contributes to three main areas of literature. The first is VC portfolio size. Like many of these papers, I examine VC portfolio optimization. Unlike Kanniainen and Keuschnigg (2003, 2004), Cumming (2006), Bernile, Cumming, and Lyandres (2007), and Cumming and Walz (2009) who examine the optimal size of the VC portfolios, I examine the composition of the portfolios and how they relate to PC success.

This paper also contributes to the building of literature regarding VC strategy and performance. To this author's knowledge, the only examination of VC diversification versus specialization is Norton and Tenenbaum (1993). Their paper examines how VCs reduce risk. Specifically, they use survey data to examine whether VC firms reduce risk through stage diversification, information sharing, networking, and/or specialization. My approach differs from theirs in that I analyze four diversification dimensions and use information obtained from both Galante's Venture Capital and Private Equity Directory and Thomson Financial's SDC Platinum, providing a larger sample from which to derive conclusions.

Finally, this paper contributes to the literature on corporate diversification in that it expands the breadth of corporations that are analyzed in the conglomerate literature. The literature that most closely relates to my paper in the conglomerate literature is Stein (1997). This theoretical paper examines the role of corporate headquarters in strategically dispersing funds to profitable projects. My work differs from Stein's in that it is empirical in nature and it examines the venture capital level of financing rather than the public arena.

The rest of the paper is structured as follows. Section I motivates the paper. Section II develops the empirical methodology of the paper, while Section III describes the data. Section IV lists the results of the study, Section V offers robustness tests, and Section VI concludes.

I. The Risky Business of Venture Capital

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

VCs, in their role of mutual fund managers, choose investments based on their expected payoff. These managers can be thought to be investing in a set of Arrow-Debreu state-contingent securities; the potential investments have a payoff of one in only one state of nature and zero in all other states of nature. The VC, acting on behalf of himself and his limited partners in the capacity of a mutual fund manager, faces the following expected payoff

  • image(1)

where π is the probability assessment by VCi of state of nature ω occurring and x is the random payoff for security j in state of nature ω.

In the event that the VC invests in more than one security, one can compare the expected payoff of a single investment portfolio to that of a multiple investment portfolio. The relationship between the expected payoffs of an equally weighted investment portfolio and any single investment is as follows:

  • image(2)

This is because there may be more states of nature that will provide a payoff of one with securities that are not correlated. Note that with optimal weighting, the portfolio can be further enhanced. However, I only intend to demonstrate the impact of a simple diversification here.

In fact, if I can assume that the securities are not perfectly correlated (e.g., the payoffs are weakly correlated), it is possible to claim that

  • image(3)

since there will be more states of nature that provide a payoff of one. In such a case, we can say that the VC is never hurt by diversification and, in many cases (perhaps the majority), he benefits from it.

Although diversification does not by itself guarantee a greater payoff, holding risk constant, standard portfolio theory would suggest that higher payoffs are possible. Given the aforementioned high level of risk associated with venture capital investments, it would seem likely, even probable, that VCs would diversify on at least one dimension (e.g., geography, stage, or industry). Further, assuming that investments are weakly correlated allows for a greater payoff.

I use growth in capital under management instead of return, the typical proxy for payoff, due to a lack of good performance data in this area of research.4 I express the notion in Hypothesis 1a:

  • H1a: The growth of the venture capital firm is significantly and positively related to a diversification strategy that is determined by the number of geographical locations, stages, and/or industries managed by a VC.

Although the actual impact of diversification on venture capital firm growth is important, it is also worth examining this impact across diversification dimensions (i.e., geography, stage, and industry). Given that diversification across these different dimensions may not have the same impact, I examine whether these expected payoffs differ from one another. More concretely stated, I examine, for example, whether the payoff of diversifying across industries is the same as diversifying geographically. Will these payoffs be equal to diversifying across stages?

Equation (1) refers to the expected payoff of “security j.” For the purposes of this paper, I extend this to different forms of diversification: 1) geography, 2) stage, and 3) industry. Equations (4)–(7) show these extensions

  • image(4)
  • image(5)
  • image(6)
  • image(7)

where E(Pγ) refers to the expected payoff of international region γ, E(Pϕ) refers to the expected payoff of US region ϕ, E(Pθ) refers to the expected payoff of stage θ, and E(Pσ) refers to the expected payoff of industry σ. To address the aforementioned question, I examine whether these expected payoffs are equal, as suggested in Equation (8).

  • image(8)

Beyond simple correlation between the diversification dimensions, there is no reason to believe that these expected payoffs are, in fact, equivalent to one another. As such, I propose that these expected payoffs are not equal to one another. More formally stated:

  • H1b: The impact of diversification on venture capital firm growth will not be equal across diversification dimensions.

The potential costs of diversification for the VC (and the PC) are primarily based on the VC's limited resources. Depending upon the monitoring/advising (hereafter “involvement”) style of the VC and/or the stage of development (Gompers and Lerner, 1999), diversification could entail considerable time and expense. For the majority of the diversification dimensions, additional time for involvement is required (for theoretical motivation of this see Fulghieri and Sevilir, 2009). Time absorbed in involvement is time taken away from remaining clients (i.e., PCs), as well as new investment due diligence, holding staff constant. The expense of involvement across geographically dispersed portfolios (Lerner, 1995; Kanniainen and Keuschnigg, 2003; Cumming, 2006) exceeds that of more focused investment portfolios. Other diversification dimensions (e.g., stage and industry) could impose a cost based on finite expertise.5 Such resource limitations impose a “hand holding” or management cost to diversification that the market generally enforces in the pricing of public firms (Servaes, 1996). If these costs become substantive, the growth and development of these fledgling firms could be negatively impacted, ultimately delaying the exit of these firms. PCs funded by diversified VCs might be unable to attain the growth necessary to enter the public market through initial public offering (IPO) or merger/acquisition (M&A) in a timely manner. Tested empirically:

  • H2: The entrepreneurial time to exit, as measured by the probability of the PC's current status being exit (stagnation/failure), is significantly negatively (positively) related to the VC diversification strategy (e.g., the number of geographical locations, industries, and/or stages managed by a VC).6

II. The Impact of Diversification

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

A. VCs

To test whether VCs do indeed benefit from diversification, annual rankings and diversification preferences of VCs are hand collected from Galante's Venture Capital and Private Equity Directory for years 1998-2006. I examine the impact a diversification strategy has on the growth of a venture capital firm. To avoid biases based on ranking methods for firms with equivalent capital under management, I use growth instead of the rank of the venture capital firm. The model specification is as follows:

  • image(9)

where Growthi,t is the log difference in capital under management by VCi from time t to t + 1 scaled by time t + 1 minus time t (to account for any missing years in the data). Xi,t is a vector of VC firm characteristics including the VC's age, preferred role in a syndication (Prefer to Originate), a proxy for the previous successes of the VC (Previous IPOs), a proxy for the skill set of the VC (Expertise), a dummy indicating whether the VC is corporate or not (Corporate VC), and Risk, an indicator of the level of risk the VC undertakes in his investments. Risk is an index from zero to two, which is the sum of two dummy variables, IT Dummy and Early-Stage Dummy. Y is a vector of macroeconomic variables to control for such things as the VC fundraising (Number of Deals) (see Gompers, Lerner, Blair, and Hellmann, 1998, for an analysis of what drives VC fundraising), factors that may affect the probability of exit such as the IT bubble time period (Bubble) and general market conditions (S&P 500 Return) (see Cumming, Fleming, and Schwienbacher, 2005, for the rationale of VC investment changes in different IPO markets). Finally, Diversificationi,t−1, the variable of interest, which is lagged to avoid any causality ambiguity, is the diversification strategy employed by VCi at time t − 1 scaled by capital under management at time t − 1. These strategies include international and domestic geographical locations, stages, and industries. All strategies are scaled to account for the size of VCi. This scaling is essential since diversifying into 10 international regions, for example, is very different for a small VC than for a large VC.

To examine these diversification measures controlling for interrelation, I create diversification indexes. These indexes differentiate between a homogeneous case, where a VC has four industries all within the same two-digit SIC code, and a heterogeneous case, where a VC has four industries, all of which have different one-digit SIC codes.

I create a modified Herfindahl index of industry diversification to gauge the extent of diversification on the industry dimension. The industry diversification index (IDI) is calculated as follows:

  • image(10)

where ln is the natural log of the expression in parentheses, ηi,t−1 is the number of preferred industries within a single two-digit SIC code for VCi at time t − 1, and τi,t−1 is the total number of industries preferred by VCi at time t − 1. This index will be high when the VC portfolio is diversified and low when it is concentrated.

To take into account the ratio of geographical locations to branch offices of the VC, I create a measure of geographic diversification. The measure (GDI) is calculated as follows:

  • image(11)

where πi,t−1 is the number of geographic locations by VCi at time t − 1, and λi,t−1 is the number of branch locations VCi has at time t − 1. This measure accounts mainly for time involvement within a fund, controlling for the fact that branch offices of the VC might exist in international (US) regions for which the VC prefers to invest.7 That is, how much time will be consumed by involvement in these firms or education in culture and international business practices given existing branch locations (i.e., how knowledgeable can the VC be in venture capital investments in Africa?). The measure implicitly assumes that the firm has at least one branch office in the international region for which it prefers to invest and the branches are of equal size. It will be higher when the VC is more diversified and lower when it is more concentrated.

The stage diversification index (SDI) is the percentage of the entrepreneurial firm life cycle in which the VC invests. This figure is calculated as follows:

  • image(12)

where δi,t−1 is the number of portfolio company stages VCi prefers at time t − 1 and γ is the number of stages in the venture capital cycle. This measure describes the percentage of the life cycle in which the VC invests. It is higher (lower) when the VC is diversified (concentrated) on this dimension.

After I calculate these measures on all three diversification dimensions, I estimate, once again, Equation (9) using the diversification indexes. The resulting equation is

  • image(13)

where variables are as they are defined in Equation (9) with the exception of the diversification index, which is a vector of the indexes listed in Equations (10)–(12). Note that these indexes are also lagged to avoid any issues with causality.

B. Entrepreneurial Firms

To ascertain the impact of diversification strategies on PCs, I collect annual rankings and diversification preferences of VCs from Galante's Venture Capital and Private Equity Directory for years 1998-2006 and information on the PCs of the firms listed from SDC Platinum. I use VC characteristics as in Equations (9) and (13), the current status of the PC, and investment specifics such as the investment term, the number of years since the last VC investment, the portfolio size per manager, and the market-to-book ratio of the industry for which the PC belongs. I purge the database of any VC/PC relationships whose last investment preceded my sample period (e.g., prior to 1998) and create a relationship data set from which I determine diversification impact.

To ascertain how PCs are impacted by VC diversification, I use a multinomial logit model to regress the following:

  • image(14)

where CurrentStatus is the current standing of the entrepreneurial (PC) company j (i.e., defunct, private, public, or subsidiary).8Ψ is the cumulative logistic probability distribution function. Invi is a vector of VC investment characteristics that control for the timing of the investment, such as the term of the investment (Investment Term) and the number of years since the last VC investment (Years Since Last Inv). Xi is a vector of VC characteristics including the VC's preferred role in a syndication (Prefer to Originate), whether the VC is corporate or not (Corporate VC), the number of previous IPOs, the number of successful funds the VC has raised (Expertise), the level of risk the VC takes on (Risk). I is the market-to-book ratio for the industry to which the PC belongs (Industry M/B). Y is a vector of macroeconomic variables including a proxy for the level of VC fundraising (Number of Deals), general market conditions (S&P 500 Return), and a dummy variable indicating whether the last year of VC investment was during the IT bubble (Bubble). The figures in Xi are averaged over the investment term since the observations in this section are VC/PC relationship specific. Figures in Y are consistent with the last year of VC investment. Diversification dimensions are once again based on geography, stage, and industry. Both the diversification measure and index are tested. Robust errors are clustered around PC to control for firm effects.9

III. Data Resources

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

The lack of empirical research in venture capital is a consequence of the private status of the companies involved. Private firms do not have to adhere to the disclosure standards or the quarterly filings imposed by the Securities and Exchange Commission. Moreover, due to the illiquidity of private equity funds and the longer-term investment duration, credible intertemporal valuation is almost impossible. Accordingly, performance data are not available. The industry, therefore, relies on methods such as the internal rate of return (IRR) to discern levels of performance, which can be problematic in that it results in multiple measurements in instances of irregular cash flows (Ross, Westerfield, and Jordan, 2006). I circumvent this potential problem by using the investment preferences of VCs from Galante's Private Equity and Venture Capital Directory, which lists the largest 500 US venture capital and private equity firms. This annual ranking, “The 500 Largest US Venture Capital and Private Equity Firms” from 1998 to 2006, is based on capital under management. Information on these firms is either voluntarily provided by each firm or estimates are provided by Galante's based on news, data in their possession, or inquiries of the firms themselves. The data included in this ranking are firm name, city, state, capital under management, and the rank (based on capital under management).

The data collected for this empirical analysis include a number of characteristics of the VC beyond that which is included in the Galante's ranking. These include Age, Prefer to Originate, Corporate VC, Expertise, Number of Previous IPOs, IT Dummy, and Early-Stage Dummy. Age is included to proxy such things as VC affiliation and skill, which has been determined to affect PC success in exits (Megginson and Weiss, 1991). These data were collected from Thomson Financial's SDC Platinum.

Some VCs are just more knowledgeable than others due to experience and their gained skill set, leading to implications for both VC growth and PC current status. To control for this, I include a proxy for VC skill, Expertise. The number of funds a VC has successfully raised derives this proxy. This proxy implicitly assumes retention of VC management. This assumption should not be problematic as long as venture capital firms are able to hire similarly talented executives to lead their firms. Further, both Age and Expertise serve to control for VC grandstanding, which was brought to light by Gompers (1996).

Prefer to Originate is included to control for the VC's preferred role in a syndication and its influence on PC exit (Cumming, Fleming, and Schwienbacher, 2006). According to Lerner (1994), “Syndicating first-round venture investments may lead to better decisions about whether to invest in firms.” This implies that VCs that lead (or even participate in) a syndication will invest in higher quality PCs and the resulting probability of exit should be higher. Corporate VC is a dummy variable that indicates whether a VC is corporate or not. It is included to control for VC fund characteristics and follows the work of Cumming, Fleming, and Schwienbacher (2006).

Similar to expertise, but judging more specifically a VC's track record, is Number of Previous IPOs. This is the number of IPOs for which a VC is responsible. The implication of a VC's track record on PC exit is obvious. The more successful a VC has been in the past, the more successful (i.e., PC exit via IPO or M&A) a VC will be in the future.

Growth is the log difference in VC capital under management (or rank in the robustness section) scaled by time t + 1 minus time t. This is averaged over the investment term in the PC regressions since these observations are VC/PC relationship based and time invariant. As the VC becomes larger and attains more clout in the industry, it will be able to offer its PCs more expertise, financial assistance, and certification in the ultimate exit strategy (Megginson and Weiss, 1991; Hsu, 2004).

Dummy variables for the riskiest sectors of the industry and stage diversification dimensions are added together to create the variable Risk. Gompers and Lerner (1999) and Norton and Tenenbaum (1993) explain that investment at certain stages entails more risk, and accordingly, more opportunity than others. Similarly, there are some industries that are riskier than others. Due to the different opportunity sets available in these categories, I include an index that sums the two dummy variables for the stage and industry perceived as riskier than the rest: 1) information technology (IT Dummy) and 2) early-stage investments (Early-Stage Dummy). Since each dummy variable can be at least zero and at most one, Risk is an index from zero to two. I include this index to neutralize any additional motivation to diversify and to account for any fund effects.

Thomson Financial's SDC Platinum provides data on the PCs in which VCs invest. There are 121,106 PC/VC investment observations. Specifics about the VC/PC investment relationship are also obtained. They are: 1) Total Investment, 2) Investment Term, 3) Years Since Last Inv, 4) Portfolio Size per Manager, and 5) Industry Market-to-Book. Total Investment is the total amount of capital the VC fund has invested. This is included to control for fund size in the fundraising regressions. This is important in cases where there are funds that limit the size of the fund for the sample term. Investment Term and Years Since Last Inv are included to control for the average term of investment. It is more likely that a firm would have exited the venture capital cycle if the term is longer or if the last investment occurred less recently. Portfolio Size per Manager accounts for the number of companies that each manager must oversee and has a direct implication on the costs of diversification. In fact, the inclusion of this variable, in particular, allows us to run a horse race of sorts to see which form of diversification is more influential in the current status of the PC. Industry Market-to-Book is included to control for any cyclical impact regarding the industry. This is included based on several papers in the area including Cumming and MacIntosh (2003b) and Cumming, Fleming, and Schwienbacher (2006). Brau, Francis, and Kohers (2003) find that this variable increases the propensity of a given PC to exit via M&A.

Macroeconomic variables such as Number of Deals, S&P 500 Return, and Bubble are included to control for the general state of the VC industry and the market. Number of Deals provides a proxy for the general fundraising levels. S&P 500 Return is included to control for public market conditions. This variable will likely pick up the countercyclical nature of the venture capital industry (Groshen and Potter, 2003). Following studies such as Cumming, Fleming, and Schwienbacher (2006), I include Bubble to account for the increased probability of exit during the IT bubble period (1998-2000).

Number of Geographic Locations is included to measure the extent to which the VC is diversified.10 Since many VCs invest in PCs within their general area, this variable could disclose information as to how the VC handles its monitoring of the PC, and perhaps how “hands on” it is in its approach to monitoring (MacMillan, Kulow, and Khoylian, 1988).

Another dimension of diversification, the Number of Stages, is used to calculate the third dimension of possible venture capital diversification, stage diversification, which provides information on the implied management expertise in the life cycle of the entrepreneurial venture.11Gompers (1995) finds that VCs concentrate in stages where monitoring is valuable, alluding to the fact that all stages do not offer this advantage and that dispersion among stages would not offer significant benefits. The VCs desired level of involvement (Macmillan Kulow, and Khoylian, 1988) coupled with the amount of monitoring that is typical of particular stages of investment (Elango, Fried, Hisrich, and Polonchek, 1995), most likely influence this decision.

Number of Industries, a dimension on which to diversify, is included to examine the more typical means of diversification among conglomerates.12 As the typical gauge for level of diversification, it discloses the VC's approach to investment. Equally important, however, it implies how diverse the management expertise of the VC is, since advising is one of the most important functions of the VC. Leshchinskii (1999) finds that interindustry externalities, if negative, can necessitate the termination of some positive net present value (NPV) projects within the portfolio of the VC (which could hurt individual PCs) for the benefit of the VC value overall.

A list of summary statistics is found in Table I.

Table I.  Data Characteristics VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. Age is the natural log of the number of years VCi has been in business. Prefer to Originate describes the VCs preferred role in a syndication. Corporate VC is a dummy variable that takes on a value of one if the VC is corporate and zero otherwise. Previous IPOs is the number of IPOs for which VCi has been responsible. Expertise is the number of funds the VC has raised before time t. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. Total Investment is the natural log of the total capital ($ 000s) that the VC fund has invested in all portfolio companies. Growth is the log difference in VC capital under management from time t to time t + 1 scaled by the difference between time t + 1 and time t. Intl Geography is the number of international regions in which VCi invests at time t − 1. IGDI is the number of international regions per branch in which VCi invests at time t − 1. Dom Geography is the number of US regions in which VCi invests at time t − 1. DGDI is the number of US regions per branch in which VCi invests at time t − 1. Stages is the number of stages in which VCi invests at time t − 1. SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle. Industries is the number of industries in which VCi invests at time t − 1. IDI is an altered Herfindahl index of industries in which VCi invests. All diversification variables are scaled by capital under management to control for VC size. Investment Term is the natural log of the difference between the year of last investment minus the year of first investment. Years Since Last Inv is 2006 minus the year that the VC made its last investment in the PC. Portfolio Size/Mgr is the number of PCs in which VC invests divided by the number of managerial staff in VC. Industry M/B is the market-to-book ratio of the industry to which PCj belongs. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 is the annual return on the S&P 500 index. Bubble is an indicator variable that is equal to one if time t is the year 1998, 1999, or 2000.
 Obs.MeanStd. Dev.MinMax
VC Characteristics
 Age 1,893 2.82 0.51 1.61 4.54
 Prefer to Originate 1,893 0.76 0.4301
 Corporate VC 1,893 0.45 0.5001
 Previous IPOs 1,893 1.69 4.41032
 Expertise 1,893 3.75 2.83028
 Risk 1,893 0.90 0.8202
 Total Investment 1,89213.04 1.50 7.0217.26
 Growth (capital) 1,893 0.13 0.34−2.39 2.48
 Intl Geography 1,893 0.01 0.010 0.12
 IGDI 1,892 0.00 0.010 0.12
 Dom Geography 1,886 0.01 0.020 0.08
 DGDI 1,892 0.01 0.010 0.08
 Stages 1,893 0.01 0.02 0.00 0.15
 SDI 1,893 0.07 0.11 0.00 0.88
 Industries 1,893 0.03 0.05 0.00 0.30
 IDI 1,893 1.18 2.05 0.0211.41
Investment Characteristics
 Investment Term43,677 2.59 0.940 3.95
 Yrs Since Last Inv43,677 2.61 5.39−135
 Portfolio Size/Mgr43,677 4.34 3.05−8.57 9.90
 Industry M/B    54 3.16 5.17−7.4620.43
Market Conditions
 Number of Deals    9 8.28 0.36 7.99 9.00
 S&P 500 Return    9 2.2517.48−21.9828.41
 Bubble    9 0.23 0.4201

A vast majority of variables used within specifications do not exhibit any real conformance. There are only two pairwise correlation values of concern: 1) Number of Deals and 2) Bubble in the VC regressions (Table II, Panel A) at 0.88 and Expertise and Previous IPOs (Table II, Panel B) at 0.47. Neither correlation is surprising. Specifications excluding these variables (e.g., using time dummies instead of a Bubble dummy) exhibit qualitatively identical results; the inclusion of these variables does not seem to confound the results. As such, the variables in question are included based on their relevance to the analysis. Correlation values of all variables are displayed in Table II, Panels A and B.

Table II.  Correlations VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. This table shows pairwise correlation of the variables for VCs (Panel A) and entrepreneurs (Panel B). Age is the natural log of the number of years VCi has been in business. Prefer to Originate describes the VCs preferred role in a syndication. Corporate VC is a dummy variable that takes on a value of one if the VC is corporate and zero otherwise. Previous IPOs is the number of IPOs for which VCi has been responsible. Expertise is the number of funds the VC has raised before time t. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. Total Investment is the natural log of the total capital ($ 000s) that the VC fund has invested in all portfolio companies. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 is the annual return on the S&P 500 index. Bubble is an indicator variable that is equal to one if time t is the year 1998, 1999, or 2000. Growth is the log difference in VC capital under management from time t to time t + 1 scaled by the difference between time t + 1 and time t. Intl Geography is the number of international regions in which VCi invests at time t − 1. IGDI is the number of international regions per branch in which VCi invests at time t − 1. Dom Geography is the number of US regions in which VCi invests at time t − 1. DGDI is the number of US regions per branch in which VCi invests at time t − 1. Stages is the number of stages in which VCi invests at time t − 1. SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle. Industries is the number of industries in which VCi invests at time t − 1. IDI is an altered Herfindahl index of industries in which VCi invests. All diversification variables are scaled by capital under management to control for VC size. Investment Term is the natural log of the difference between year of last investment minus the year of first investment. Years Since Last Inv is 2006 minus the year that the VC made its last investment in the PC. Portfolio Size/Mgr is the number of PCs in which VC invests divided by the number of managerial staff in VC. Industry M/B is the market-to-book ratio of the industry to which PCj belongs. Bold font indicates a significance level of 5% or higher.
Panel A. Venture Capitalists
 123456789101112131415161718
Age (1) 1.00                 
Prefer to Originate (2) 0.03 1.00                
Corporate VC (3) 0.04−0.01 1.00               
Previous IPOs (4) 0.34−0.02 0.10 1.00              
Expertise (5) 0.27 0.04 0.06 0.34 1.00             
Risk (6)−0.02−0.02 0.05 0.16 0.26 1.00            
Total Investment (7) 0.24 0.06 0.09 0.34 0.34 0.201.0           
Number of Deals (8) 0.11−0.06 0.01 0.03−0.10−0.12−0.02 1.00          
S&P 500 Return (9)−0.01 0.03−0.04−0.02 0.02−0.04−0.00−0.28 1.00         
Bubble (10) 0.11−0.05 0.00 0.01−0.10−0.16 0.03 0.88 0.00 1.00        
Growth (capital) (11) 0.03 0.00−0.02 0.06 0.03 0.03 0.09 0.28−0.20 0.16 1.00       
Intl Geography (12)−0.09−0.03 0.02−0.11−0.19−0.05−0.22 0.24−0.01 0.26 0.11 1.00      
IGDI (13)−0.07−0.05 0.00−0.11−0.21−0.10−0.22 0.23 0.01 0.26 0.10 0.92 1.00     
Dom Geography (14)−0.16−0.03 0.00−0.13−0.31−0.09−0.35 0.37 0.02 0.42 0.20 0.46 0.46 1.00    
DGDI (15)−0.13−0.06−0.05−0.12−0.31−0.16−0.33 0.35 0.03 0.40 0.17 0.39 0.49 0.92 1.00   
Stages (16)−0.13 0.00 0.00−0.13−0.24−0.09−0.32 0.36−0.04 0.38 0.21 0.47 0.45 0.69 0.61 1.00  
SDI (17)−0.13 0.00 0.00−0.13−0.24−0.09−0.32 0.36−0.04 0.38 0.21 0.47 0.45 0.69 0.61 1.00 1.00 
Industries (18)−0.07−0.03−0.07−0.13−0.23−0.23−0.39 0.31 0.01 0.33 0.14 0.21 0.24 0.54 0.54 0.50 0.50 1.00
IDI (19)−0.07−0.03−0.07−0.14−0.29−0.30−0.40 0.40−0.01 0.43 0.15 0.27 0.32 0.66 0.67 0.57 0.57 0.87
Panel B. Entrepreneurs
 1234567891011121314151617181920
Investment Term (1) 1.00                   
Yrs Since Last Inv (2)−0.31 1.00                  
Portfolio Size/Mgr (3) 0.17 0.05 1.00                 
Industry M/B (4)−0.05 0.09−0.05 1.00                
Prefer to Originate (5)−0.01 0.02−0.06 0.04 1.00               
Growth (6) 0.02−0.04−0.06 0.02 0.07 1.00              
Corporate VC (7)−0.07 0.11 0.02 0.00−0.02−0.02 1.00             
Previous IPOs (8) 0.15 0.15 0.20 0.01 0.05 0.14−0.04 1.00            
Expertise (9) 0.15 0.06 0.03 0.00 0.10 0.23−0.08 0.47 1.00           
Risk (10) 0.09−0.11 0.30−0.10 0.01 0.05−0.03 0.14 0.22 1.00          
Number of Deals (11)−0.08 0.28 0.02 0.29 0.01−0.01 0.05 0.05 0.02−0.05 1.00         
S&P 500 Return (12) 0.11 0.19 0.07−0.09 0.01 0.02 0.00 0.08 0.07 0.04−0.43 1.00        
Bubble (13)−0.08 0.22 0.00 0.31 0.01−0.01 0.03 0.03 0.01−0.04 0.90−0.32 1.00       
Intl Geography (14)−0.02−0.01 0.05−0.01 0.01−0.10 0.00−0.14−0.21−0.04−0.01 0.01−0.01 1.00      
IGDI (15)−0.02 0.01 0.07 0.00−0.01−0.12 0.00−0.13−0.21−0.05 0.00 0.01 0.00 0.94 1.00     
Dom Geography (16)−0.04−0.01 0.12 0.01−0.03−0.16 0.02−0.21−0.41−0.06−0.01−0.01−0.01 0.60 0.60 1.00    
DGDI (17)−0.04 0.02 0.13 0.02−0.05−0.17 0.03−0.19−0.39−0.09 0.01 0.00 0.00 0.52 0.60 0.92 1.00   
Stages (18)−0.03−0.01 0.01 0.02 0.01−0.09−0.03−0.19−0.34−0.11−0.01−0.01−0.01 0.52 0.48 0.70 0.59 1.00  
SDI (19)−0.03−0.01 0.01 0.02 0.01−0.09−0.03−0.19−0.34−0.11−0.01−0.01−0.01 0.52 0.48 0.70 0.59 1.00 1.00 
Industries (20) 0.00 0.03 0.05 0.03−0.04−0.11−0.01−0.17−0.34−0.26 0.00 0.00 0.00 0.32 0.34 0.52 0.52 0.60 0.60 1.00
IDI (21) 0.00 0.03 0.06 0.03−0.04−0.12−0.01−0.17−0.37−0.29 0.00 0.00 0.00 0.39 0.41 0.60 0.59 0.63 0.63 0.94

IV. Consequences of Diversification as a Strategy

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

A. VC Growth

Results of the VC analysis in Table III show that, after controlling for fund effects, all forms of diversification offer the VC a significant benefit with regard to firm growth. Domestic geographic diversification offers one of the most extensive marginal effects. The statistically significant marginal effects (4.903 for the dimension measure in Specification (3) and 4.182 for the index measure in Specification (4)) suggest that for a VC increasing his domestic geographic diversification dimension (number of US regions scaled by capital under management) by one standard deviation, his growth would increase by 7.74%.13 Once the number of branches is taken into consideration for the index measure, this impact drops to 6.03%.14

Table III.  Venture Capital Growth The regression used for venture capitalists is Growthi,t=α+β0 Xi,t1Yt +β2 Diversificationi,t−1i,t. Venture capitalist growth is the log difference in VC capital under management from time t to time t + 1 scaled by time t + 1—time t. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. X is a vector of VC firm characteristics including Age, Prefer to Originate, Corporate VC, Previous IPOs, Expertise, and Risk. Age is the natural log of the number of years VCi has been in business. Prefer to Originate describes the VC's preferred role in a syndication. Corporate VC is a dummy variable indicating whether VCi is a corporate VC or not. Previous IPOs describes the number of IPOs for which VCi is responsible. Expertise is the number of funds VCi has successfully raised. Risk is an index from zero to two which sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return, and Bubble. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 Return is the return on the S&P 500 index. Bubble is an indicator variable describing whether time t is during the bubble (i.e., 1998, 1999, or 2000) or not. Measures of diversification in specifications are as follows: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, and 8) IDI is an altered Herfindahl index of industries in which VCi invests. All diversification variables are scaled by capital under management to control for VC size. Robust standard errors (clustered around VC) appear in parentheses.
 Dependent Variable: Venture Capitalist Growth
 Intl Geog. (1)IGDI (2) Dom Geog. (3)DGDI (4)Stages (5)SDI (6) Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Age−0.014−0.0150.001−0.006−0.005−0.005−0.014−0.013
(0.013)(0.013)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Prefer to Originate0.0150.0170.0140.0190.0080.0080.0130.012
(0.015)(0.015)(0.015)(0.015)(0.015)(0.015)(0.014)(0.014)
Corporate VC−0.024*−0.023−0.025*−0.019−0.022−0.022−0.017−0.016
(0.014)(0.014)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Previous IPOs0.0030.0030.0030.0030.004**0.004**0.003*0.003*
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Expertise0.006***0.006***0.011***0.009***0.007***0.007***0.006***0.007***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Risk0.013*0.014*0.0110.016**0.0110.0110.018**0.021***
(0.007)(0.007)(0.008)(0.008)(0.007)(0.007)(0.007)(0.007)
Number of Deals0.551***0.554***0.549***0.553***0.532***0.532***0.537***0.533***
(0.066)(0.066)(0.065)(0.065)(0.063)(0.063)(0.063)(0.063)
S&P 500 Return−0.076*−0.075*−0.088**−0.084**−0.070*−0.070*−0.081**−0.080**
(0.040)(0.040)(0.040)(0.040)(0.040)(0.040)(0.041)(0.041)
Bubble−0.290***−0.290***−0.340***−0.325***−0.304***−0.304***−0.285***−0.293***
(0.060)(0.060)(0.061)(0.061)(0.055)(0.055)(0.054)(0.055)
Div. Dimension2.820*** 4.903*** 3.453*** 0.793*** 
(0.892) (0.711) (0.633) (0.178) 
Div. Index 3.154*** 4.182*** 0.587*** 0.023***
 (0.919) (0.686) (0.108) (0.005)
N (# of firms) 1,893 (634) 1,892 (634) 1,886 (631) 1,885 (631) 2,013 (655) 2,013 (655) 2,013 (655) 2,013 (655)
Wald stat252.36***254.61***275.94***271.14***259.24***259.24***257.90**267.73***
= Intl Geography
= Dom Geography0.015**0.071*
= Stage0.079*0.3950.5130.052*
= Industry0.8210.7360.000***0.008***0.025**0.001***

A diversification strategy through number of stages offers the second largest impact on growth. A statistical significance of 1% and economically significant marginal effects (3.453 for the diversification measure in Specification (5) and 0.587 for the diversification index in Specification (6)) demonstrate this nicely. Indeed, for a VC increasing his stage diversification dimension (number of stages scaled by capital under management) by one standard deviation, this translates into a 6.37% increase in growth. Expressing the diversification as a percent of the life cycle for PCs does not change this impact.

Diversification across industries offers a benefit to the VC firm as well. The marginal effect of the diversification measure and index (0.793% and 0.023% in Specifications (7) and (8), respectively) are statistically significant at a 1% level and are also economically significant. These coefficients translate into a 4.19% increase in VC growth for a one standard deviation increase in the industrial diversification (number of industries scaled by capital under management) and a 4.72% increase in growth for a one standard deviation change in the modified Herfindahl index. Intuitively appealing is the fact that the increased impact on growth of the industry diversification index, which is a measure that takes into consideration the diversification into unlike industries, is more beneficial than diversifying into like industries.

Geographic diversification outside the US offers the smallest impact on growth, although it is still a significant one as demonstrated by the significantly positive relationship of the marginal effects (2.820 in Specification (1) and 3.154 in Specification (2)). For a VC increasing the international geographic diversification dimension (number of international regions scaled by capital under management) by one standard deviation, this translates into a 3.59% increase in growth. Once the number of branches is taken into consideration for the index measure, the impact falls slightly to 3.46%.

The marginal effects of control variables such as Number of Deals, as well as VC specifics such as Expertise, Number of Previous IPOs, and Risk, all display expected signs. The coefficients on these variables are all positively related to VC growth, with the vast majority being statistically significant. Other variables, such as S&P 500 Return and Bubble exhibit a negative relationship. These signs are not surprising given the countercyclical nature of the VC industry. Variables such as Age, Prefer to Originate, and Corporate VC do not seem to have a significant impact on VC firm growth, as demonstrated by mostly insignificant coefficients. Results for the regression analysis are shown in Table III.

Figure 1 demonstrates the positive relationship between VC Growth and the diversification dimensions after including the numerous control variables in the analysis. Confirming the results found in Table III, all panels within Figure 1 show that VC growth increases as diversification (i.e., international in Figure 1(a), domestic in Figure 1(b), stages in Figure 1(c), industries in Figure 1(d)) increases. The slopes of these graphs fall in line with the coefficients found in Table III. That said, the relative impact of diversification dimensions may only be found after taking into consideration the standard deviation of each dimension.

image

Figure 1. Partial Scatter Plot of VC Growth against Diversification Measures

These figures are based on the odd specifications in Table III (i.e., 1, 3, 5, and 7, respectively) and represent the partial scatter plots of the regression plane in VC Growth/Diversification Dimension space. Residuals are collected from Equation (9) and labeled e(Growth VC | X). Each of the diversification dimension measures (i.e., Intl Geography, Dom Geography, Stages, and Industries, respectively) is individually regressed on all of the control variables listed in Equation (9). Residuals are collected and labeled e(Diversification Measure | X). The graph represents e(VC Growth | X) plotted against e(Diversification Measure | X). VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms).

Download figure to PowerPoint

Estimating the coefficients on the diversification measures (indexes) in the same specification allows for testing of equality. In doing so, I find that approximately half are statistically equivalent to another measure (index) when estimating the coefficients in this manner. Since there is significant correlation among the diversification dimensions, these results should be taken lightly. The results suggest that the effects of diversification on VC growth can be similar, offering the VC different options if individual diversification dimensions are harmful to the PCs in which they invest.

The cumulative results display ample evidence to indicate that diversification is a dominant strategy for these financiers when not considering the impact of this strategy on the PCs in which they invest. Although the economic significance of the coefficients varies with the form and measurement of diversification, all coefficients suggest that there are benefits to diversification with regard to growth for the VC. Deviations from the specification offered in Equations (9) and (13), which I will refer to as the VC base specifications, are offered in the robustness section of this paper.15

B. Entrepreneurial Outcome

1. Exit

The results for entrepreneurial exit via IPO (see Table IV, Panel A) demonstrate that all forms of diversification by VCs, save international diversification, are significantly damaging to the probability of PC exit through IPO (or equivalently, time to IPO since PCs that are private at the end of the sample term can still go public going forward). Industrial diversification is the most damaging form of diversification, as evidenced by the statistically significant (1%) and economically large marginal effects of a one standard deviation increase in the diversification dimension and index measures (−2.09% for the dimension measure in Specification (7) and −1.87% for the index in Specification (8)).16 Recall that VC growth is positively influenced (over 4%) by a one standard deviation increase in these ratios. Although the deleterious impact on PC time to IPO is only roughly half of the benefit to the VC, it is still an amount that is noteworthy. Perhaps even more important is that the negative relationship of VC diversification with the probability of PC exit through IPO provides evidence that the incentives of the agent (VC) in this investment relationship are not aligned with its principal, the entrepreneur (PC), inciting further research to reconcile these discrepancies.

Table IV.  Entrepreneurial Company Outcome The multinomial logit model used for entrepreneurs is inline image, where Ψ is the cumulative logistic probability distribution function. Current Status is the current status of the PC: Public, Subsidiary, Private, or Defunct. The base specification is Public Status = Private. Inv is a vector of investment characteristics such as Investment Term, Years Since Last Inv, and Portfolio Size/Mgr. Investment Term is the natural log of the difference of date of last investment and date of first investment. Years Since Last Inv is 2006 minus the year that VCi made its last investment in PCj. Portfolio Size/Mgr is the number of invested PCs in VC fund scaled by the number of managers. X is a vector of venture capitalist characteristics including Prefer to Originate, Growth, Corporate VC, Previous IPOs, Expertise, and Risk. Prefer to Originate describes the VCs preferred role in a syndication. Growth is the log difference in VC capital under management from time t to time t + 1 scaled by the difference between time t + 1 and time t. Corporate VC is a dummy variable that takes on a value of one if the VC is corporate and zero otherwise. Previous IPOs is the number of previous IPOs for which VCj is responsible. Expertise is the number of funds the venture capitalist has raised before time t. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. I is the industry market-to-book ratio for the industry to which the PC belongs. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return, and Bubble. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 Return is the annual percentage return on the S&P 500 for the year that the VC last invested in the PC. Bubble is an indicator variable that takes on a value of one if the year that the VC last invested in the PC is 1998, 1999, or 2000. The specifications use the following definitions for diversification: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, and 8) IDI is an altered Herfindahl index of industries in which VCi invests. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. Marginal effects are reported and robust standard errors (clustered around PC) are given in parentheses.
Panel A. Public
 Dependent Variable: Portfolio Current Status—Public (N = 8,078)
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
Investment term0.034***0.034***0.034***0.034***0.035***0.035***0.036***0.036***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Yrs Since Last Inv0.009***0.009***0.009***0.009***0.009***0.009***0.009***0.009***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Portfolio Size/Mgr−0.003**−0.003*−0.002−0.001−0.002*−0.002*−0.001−0.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Industry M/B−0.001**−0.001**−0.001**−0.001**−0.001**−0.001**−0.001**−0.001**
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Prefer to Originate0.0030.0030.0040.0030.0040.0040.0030.003
(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)
Growth0.0000.0000.0000.0000.0000.0000.0000.000
(0.011)(0.011)(0.012)(0.012)(0.012)(0.012)(0.012)(0.012)
Corporate VC0.011**0.011**0.010**0.011**0.009*0.009*0.008*0.009*
(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)
Previous IPOs0.001***0.001***0.001**0.001**0.001**0.001**0.001**0.001**
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Expertise0.002***0.002***0.0010.001*0.001**0.001**0.0010.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Risk−0.015***−0.015***−0.015***−0.016***−0.016***−0.016***−0.021***−0.022***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.003)(0.003)
Number of Deals−0.018−0.018−0.019−0.019−0.021−0.021−0.025*−0.023
(0.014)(0.014)(0.014)(0.014)(0.014)(0.014)(0.014)(0.014)
S&P 500 Return−0.165***−0.166***−0.169***−0.168***−0.171***−0.171***−0.176***−0.173***
(0.019)(0.019)(0.020)(0.020)(0.020)(0.020)(0.020)(0.020)
Bubble0.044***0.045***0.047***0.047***0.048***0.048***0.052***0.050***
(0.013)(0.013)(0.013)(0.013)(0.014)(0.014)(0.014)(0.014)
Div. Dimension−0.15 −0.942*** −0.836*** −0.523*** 
(0.148) (0.150) (0.129) (0.051) 
Div. Index −0.399** −1.089*** −0.142*** −0.012***
 (0.191) (0.167) (0.022) (0.001)
Panel B. Subsidiary
 Dependent Variable: Portfolio Current Status—Subsidiary (N = 13,508)
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
Investment term0.066***0.065***0.065***0.065***0.065***0.065***0.066***0.066***
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Yrs Since Last Inv0.016***0.016***0.016***0.016***0.016***0.016***0.016***0.016***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Portfolio Size/Mgr0.0020.0020.0020.0020.0020.0020.0020.002
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Industry M/B−0.002***−0.002***−0.002***−0.002***−0.002***−0.002***−0.002***−0.002***
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Prefer to Originate−0.005−0.005−0.005−0.006−0.005−0.005−0.006−0.006
(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)(0.005)
Growth0.0000.0000.0000.0000.0000.0000.0000.000
(0.015)(0.015)(0.015)(0.015)(0.015)(0.015)(0.015)(0.015)
Corporate VC0.016***0.015**0.015**0.015**0.015**0.015**0.014**0.015**
(0.006)(0.006)(0.006)(0.006)(0.006)(0.006)(0.006)(0.006)
Previous IPOs0.002***0.002***0.002***0.002***0.002***0.002***0.002***0.002***
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Expertise−0.001−0.001−0.001−0.001−0.001−0.001−0.001−0.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Risk0.008**0.008**0.008**0.008**0.008**0.008**0.0050.005
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Number of Deals−0.034*−0.035*−0.034*−0.034*−0.034*−0.034*−0.037**−0.036*
(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)
S&P 500 Return−0.223***−0.223***−0.223***−0.222***−0.221***−0.221***−0.224***−0.223***
(0.026)(0.026)(0.026)(0.026)(0.026)(0.026)(0.026)(0.026)
Bubble0.043***0.044***0.042***0.042***0.042***0.042***0.043***0.043***
(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)
Div. Dimension−0.274 −0.424** −0.077 −0.171*** 
(0.190) (0.176) (0.144) (0.055) 
Div. Index −0.437* −0.428** −0.013 −0.004***
 (0.239) (0.194) (0.024) (0.001)
Panel C. Private
 Dependent Variable: Portfolio Current Status—Private (N = 16,730)
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
Investment term−0.130***−0.130***−0.130***−0.130***−0.131***−0.131***−0.132***−0.132***
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Yrs Since Last Inv−0.031***−0.031***−0.031***−0.031***−0.031***−0.031***−0.031***−0.031***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Portfolio Size/Mgr−0.002−0.003−0.004−0.004*−0.002−0.002−0.004*−0.004*
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Industry M/B0.006***0.006***0.006***0.006***0.006***0.006***0.006***0.006***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Prefer to Originate0.012*0.012*0.0110.012*0.010.010.012*0.012*
(0.007)(0.007)(0.007)(0.007)(0.007)(0.007)(0.007)(0.007)
Growth0.0000.0000.0000.0000.0000.0000.0000.000
(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)(0.019)
Corporate VC−0.034***−0.034***−0.033***−0.034***−0.032***−0.032***−0.030***−0.031***
(0.008)(0.008)(0.008)(0.008)(0.008)(0.008)(0.008)(0.008)
Previous IPOs−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Expertise−0.001−0.0010.0010.001000.0010.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Risk−0.006−0.006−0.006−0.005−0.005−0.0050.0020.002
(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)
Number of Deals0.067**0.068**0.068**0.068**0.070***0.070***0.075***0.072***
(0.027)(0.027)(0.027)(0.027)(0.027)(0.027)(0.027)(0.027)
S&P 500 Return0.484***0.485***0.486***0.484***0.487***0.487***0.494***0.490***
(0.037)(0.037)(0.037)(0.038)(0.037)(0.037)(0.038)(0.038)
Bubble−0.108***−0.109***−0.109***−0.108***−0.109***−0.109***−0.114***−0.112***
(0.022)(0.022)(0.022)(0.022)(0.022)(0.022)(0.022)(0.022)
Div. Dimension0.392 1.224*** 0.892*** 0.647*** 
(0.246) (0.230) (0.190) (0.073) 
Div. Index 0.811*** 1.376*** 0.152*** 0.015***
 (0.310) (0.255) (0.032) (0.002)
Panel D. Defunct
 Dependent Variable: Portfolio Current Status—Defunct (N = 5,361)
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Investment Term0.031***0.031***0.031***0.031***0.031***0.031***0.030***0.030***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Yrs Since Last Inv0.006***0.006***0.006***0.006***0.006***0.006***0.006***0.006***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Portfolio Size/Mgr0.003***0.003***0.003***0.003***0.003***0.003***0.003***0.003**
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Industry M/B−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Prefer to Originate−0.010***−0.010***−0.010***−0.010***−0.010***−0.010***−0.010***−0.010***
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Growth0.0000.0000.0000.0000.0000.0000.0000.000
(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)
Corporate VC0.008*0.008**0.008*0.008*0.008*0.008*0.008**0.008**
(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)(0.004)
Previous IPOs0.0000.0000.0000.0000.0000.0000.0000.000
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Expertise−0.001−0.001−0.001−0.001−0.001−0.001−0.001−0.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Risk0.013***0.013***0.013***0.013***0.013***0.013***0.014***0.014***
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Number of Deals−0.015−0.015−0.015−0.015−0.015−0.015−0.014−0.014
(0.011)(0.011)(0.011)(0.011)(0.011)(0.011)(0.011)(0.011)
S&P 500 Return−0.096***−0.096***−0.094***−0.094***−0.095***−0.095***−0.094***−0.093***
(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)(0.016)
Bubble0.020*0.020*0.020*0.020*0.020*0.020*0.019*0.019*
(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)(0.010)
Div. Dimension0.032 0.142 0.021 0.048 
(0.116) (0.107) (0.092) (0.033) 
Div. Index 0.025 0.141 0.004 0.002*
 (0.143) (0.117) (0.016) (0.001)
N (total)43,67743,67743,67743,67743,67743,67743,67743,677
Chi-squared2,503***2,503***2,546***2,542***2,535***2,535***2,603***2,582***
Log likelihood−54,039−54,039−54,014−54,013−54,016−54,016−53,967−53,983
Pseudo R20.0480.0480.0490.0490.0490.0490.0500.049

Stage diversification is the second most damaging form of diversification, as evidenced by a −1.67% marginal effect for a one standard deviation increase in the dimension measure (Specification (5)) and −1.28% marginal effect for a one standard deviation increase in the index (Specification (6)), as well as a statistical significance of 1%. This dimension, in particular, demonstrates a misalignment of interests since stage diversification provides the second largest positive impact on VC growth (6.37%). Evidence of the cost imposed on PC time to IPO (i.e., the negative marginal effect) supports the literature that posits that the value of the role of VCs is the “hand holding” that they provide. This hand holding necessitates knowledge of PC specifics such as stage and industry (Gompers and Lerner, 1999).

Results for geographic diversification are mixed. Domestic geographic diversification is detrimental to PC time to IPO as demonstrated by the negative marginal effect and its statistical significance at 1%. This negative marginal effect (Specification (3)) translates into a −0.94% for a one standard deviation increase in the dimension measure. Once branches are taken into consideration, this impact increases slightly to a −1.09% (Specification (4)). The results for international diversification are less than impressive. The impact of the dimension measure is effectively zero, since the coefficient is not statistically significant (Specification (1)). The diversification index, which examines the per branch number of international regions, is marginally statistically significant (at the 5% level) and marginally economically significant (−0.4%; Specification (2)). The lack of significance for the dimension measure combined with the significance of the index measure suggests that investment in a different international region is executed through a branch location in that international region. It is only once diversification is executed beyond that foreign branch that a cost is imposed.

Although the economic significance of geographic diversification is less than that of the other diversification dimensions, it is still significant. The relevance of this form of diversification in PC time to IPO once again supports the hand holding role of the VC, and suggests that the more geographically spread the VC, the less time he is able to spend with individual PCs. This supports the findings of Lerner (1995), who indicates that the oversight of distant businesses is more costly than that of local firms. When speaking of international diversification, this could also suggest that VCs are less knowledgeable about business practices and cultural items for foreign PCs, at least at first. Practically speaking, the fact that geographic diversification has a smaller deleterious impact of PC time to IPO is likely due to the available technology (i.e., conference calls) and the ease/frequency of business travel.

Control variables such as Investment Term, Years Since Last Inv, Corporate VC, Previous IPO, VC Expertise, and Bubble are positively related to the probability of PC time to IPO. S&P 500 Return is negatively related to the probability of PC time to IPO. The negative relationship of Risk with PC exit via IPO supports the findings of Cumming, Fleming, and Schwienbacher (2005), who find that VCs that invest in PCs that are in the early stage or the IT industry are not in a rush to exit their investments (in cases where the market is illiquid). All of these relationships are as expected. Relationships not as obvious are those for Industry Market-to-Book, Number of Deals, and Growth. The negative coefficient on Industry Market-to-Book most likely reflects the countercyclical nature of venture capital. The impact of Number of Deals may reflect the negative impact of another form of diversification, volume. As the number of investments increases, the VC is spread further and further, imposing a limit to the hand holding that he is able to offer individual PCs. Perhaps most important is the impact of growth on time to IPO. The negative relationship between this control variable and time to IPO implies that larger VCs (those with smaller growth) are more likely to get their PC to the public market in a timely fashion. This relationship supports the literature that puts forth the idea that VC affiliation and/or reputation is important in the success of its PCs (Megginson and Weiss, 1991; Hsu, 2004). Results are found in Table IV, Panel A.

Evidence in Panel B of Table IV points to the fact that only certain diversification dimensions hinder time to PC exit through M&A. Although all four diversification techniques impose negative influences on the probability of a PC's current status being an exit through M&A, not all coefficients exhibit statistical significance. The most harmful form of diversification is industrial. Whether measured by the dimension measure or the index, both are significantly harmful to PC time to M&A. This is demonstrated by the statistically significant marginal effects (−0.171 in Specification (7) and −0.004 in Specification (8)). These translate into a 0.68% (0.62%) decrease in the probability of M&A with a corresponding one standard deviation increase in the industrial diversification dimension (index).

Also damaging is domestic diversification. The marginal effects of both the dimension measure and the index are statistically significant (−0.424 in Specification (3) and −0.428 in Specification (4)), suggesting that a one standard deviation increase in the domestic diversification dimension (index) causes a 0.42% (0.43%) decrease in the probability of M&A.

International geographic diversification marginally influences PC time to M&A. This is demonstrated by the lack of significance in the diversification dimension (Specification (1)) and the marginal significance (10%) of the diversification index (Specification (2)). The economic significance of the index, however, is in line with that of domestic diversification (−0.44%).

Interestingly, the economic significance of those forms of diversification that do exhibit statistical significance is less than what is seen for exit through IPO. Also different is the fact that stage diversification does not significantly influence the time to exit via M&A as it did for exit via IPOs. This implies that there are differences in the PCs that exit through M&A, in the VCs that exit their PCs through M&A, or both. It is also possible that the specification belies the diversity of the impact of diversification on PC time to exit via M&A (i.e., this impact could depend on another control variable). This relationship will be examined in the robustness section of this paper. Looking to the control variables for some answers, distinctions in the signs of marginal effects for both Risk and Expertise in Panels A and B can be found. The former implies that VCs that take on more risk, are more likely to exit through M&A.17 This finding is loosely supported by Cumming (2008) who finds that VCs who take on more syndication partners (and thus, less risk) are more likely to exit via IPO. The latter implies that VC expertise or equivalently, track record, has little effect on the probability of PC exit through M&A. These two distinctions are somewhat supported by the findings of Schwienbacher (2002), Fleming (2004), and Cumming and MacIntosh (2003a, 2003b) who consider M&A an inferior exit to IPO (i.e., firms exiting via M&A are of lower quality than those exiting via IPO). Other control variables exhibit the same relationships as in Panel A.

The results found in Panels A and B of Table IV collectively imply that diversification based on PC industry is detrimental to the time to PC exit. Stage diversification is harmful with regard to PC exit through IPO. Relatively speaking, geographical diversification, both domestic and international, seems to be fairly benign with regard to the effects on PC exit. As such, it is possible that VCs can use this form of diversification to reduce risk and potentially grow their firm. It is important to note that these results do not mean to imply that the PCs of diversified VCs never exit the venture capital cycle. Rather, the results imply that VC diversification can delay PC exit.

2. Stagnation and Failure

Panel C of Table IV demonstrates the positive and significant impact of VC diversification on the probability that the PC's current status will be private (i.e., stagnation). These results mesh nicely with the results seen in the previous section for PC exit. All forms of diversification have a statistically significant positive influence on the probability that the PC will remain private. Marginal effects seen in Panel C are greater than those seen in either Panels A or B. This should not be too surprising considering the fact that remaining private means that the PC did not exit through IPO or M&A. The relative economic significance of these marginal effects is in line with those in Panel A, for PC exit through IPO. That is, industrial diversification seems to have the most deleterious effect, a 2.59% increase in the probability of the PC remaining private for a one standard deviation increase in the dimension measure. Stage diversification is next, at 1.78%. Domestic diversification follows at 1.28%, and international diversification is only significant once the number of branches is considered (0.81%). Index measures of the first three forms of diversification have similar economic impacts of the marginal effects to the dimension measures.

Intuitively appealing and asserting that the incentives of VCs and PCs are not completely misaligned is the fact that there exists a basically neutral (i.e., zero) impact of diversification on PC failure. This implies that although exit strategies might be delayed, diversification at the VC level does not threaten the PCs' viability. Ultimately, both the PC and the VC benefit from an entrepreneurial exit. Results here indicate that the VC can potentially employ diversification dimensions that do not materially delay PC exit without fear of risking PC failure. Full results are seen in Panel D of Table IV.

Deviations from the specification found in Equation (14), which I will refer to as the base PC specification, are found in the Robustness section of the paper.

V. Robustness

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

A. Sample Selectivity

To control for sample selectivity issues in the choice of IPO (Cochrane, 2005; Cumming, Fleming, and Schwienbacher, 2006), I use a two-step Heckman model (Heckman, 1979). Specifically, I perform the following regression:

  • image(15)

Following the work of Cumming, Fleming, and Schwienbacher (2006), I include only Bubble year dummies in the first step of the regression (i.e., the choice of exiting versus not exiting) to avoid any multicollinearity in the specifications. Once the possible effects of sample selection are taken into consideration, the results are enhanced considerably. Results of this methodology, seen in Table V, assert that the harmful affects of VC diversification on PC exit via IPO are much closer to the impacts on VC growth. Domestic diversification hinders the probability of IPO given PC exit by 4.23% and 4.67% (based on marginal effects in Specifications (3) and (4)) for the diversification dimension and index, respectively. This is up from −0.94% to −1.09% in the base PC regression in Table IV and compared to a 7.74% and 6.03% impact on VC growth from Table III, the base VC regression. The probability of IPO given PC exit is similarly negatively impacted by stage diversification. The economic impact of this marginal effect (seen in Specifications (5) and (6)) is similar to that of the positive impact for VCs (−5.42% for PCs vs. 6.37% for VCs). What is most noteworthy, however, is the economic impact of the marginal effect of industrial diversification. A one standard deviation change in industrial diversification has a −8.25% impact on the probability of PC exit via IPO (given that the PC has exited) as compared to a 4.19% impact on VC growth. This is an important point given the literature's opinion that M&A is inferior to IPO as a form of exit (Schwienbacher, 2002; Cumming and MacIntosh, 2003a, 2003b; Fleming, 2004).

Table V.  Nonrandomness of the Decision to Exit The Heckman Selection Model is used for entrepreneurs is inline image, where the first-stage regression is inline image. Exit is defined as PC current public status of subsidiary or public. T is a vector of last investment year Bubble time period dummies (i.e., last investment year = 1998, 1999, or 2000). Inv is a vector of investment characteristics such as Investment Term, Years Since Last Inv, and Portfolio Size/Mgr. Investment Term is the natural log of the difference of date of last investment and date of first investment. Years Since Last Inv is 2006 minus the year that VCi made its last investment in PCj. Portfolio Size/Mgr is the number of invested PCs in VC fund scaled by the number of managers. X is a vector of venture capitalist characteristics including Prefer to Originate, Growth, Corporate VC, Previous IPOs, Expertise, and Risk. Prefer to Originate describes the VC's preferred role in a syndication. Growth is the log difference in VC capital under management from time t to time t + 1 scaled by the difference between time t + 1 and time t. Corporate VC is a dummy variable that takes on a value of one if the VC is corporate and zero otherwise. Previous IPOs is the number of previous IPOs for which VCj is responsible. Expertise is the number of funds the venture capitalist has raised before time t. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. I is the industry market-to-book ratio for the industry to which the PC belongs. Y is a vector of macroeconomic variables including Number of Deals and S&P 500 Return. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 Return is the annual percentage return on the S&P 500 for the year that the VC last invested in the PC. The specifications use the following definitions for diversification: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, (7) Industries is the number of industries in which VCi invests at time t − 1, and (8) IDI is an altered Herfindahl index of industries in which VCi invests. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. Marginal effects are reported and robust standard errors (clustered around PC) are given in parentheses.
Dependent Variable: Pr(Exit) in (1); Pr(IPO|Exit) in (2)
 
 Intl Geog.IGDIDom Geog.DGDIStagesSDIIndustriesIDI
(1)(2)(1)(2)(1)(2)(1)(2)(1)(2)(1)(2)(1)(2)(1)(2)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Investment term −0.02 −0.02 −0.02 −0.02 −0.02 −0.02 −0.01 −0.01
 (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.01) (0.02)
Yrs Since Last Inv −0.01 −0.01 −0.01 −0.01 −0.01 −0.01 −0.01 −0.01
 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Portfolio Size/Mgr −0.02* −0.02* −0.01 −0.01 −0.02* −0.02* −0.01 −0.01
 (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01) (0.01)
Prefer to Originate 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.03
 (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
Growth −0.18** −0.18** −0.20** −0.20** −0.19** −0.19** −0.19** −0.19**
 (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06) (0.06)
Corporate VC 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00
 (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03)
Previous IPOs 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Expertise 0.01** 0.01** 0.01 0.01 0.01 0.01 0.00 0.00
 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Risk −0.08*** −0.08*** −0.08*** −0.08*** −0.08*** −0.08*** −0.10*** −0.10***
 (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02)
Industry M/B 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
 (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00) (0.00)
Number of Deals 0.07 0.07 0.07 0.07 0.06 0.06 0.05 0.06
 (0.10) (0.10) (0.10) (0.10) (0.10) (0.10) (0.10) (0.10)
S&P 500 Return −0.20 −0.20 −0.20 −0.20 −0.21 −0.21 −0.22 −0.21
 (0.12) (0.12) (0.12) (0.12) (0.12) (0.12) (0.12) (0.12)
Div. Dimension 0.03   −2.68***   −2.94***   −1.56***  
 (0.82)   (0.80)   (0.70)   (0.26)  
Div. Index   −0.55   −3.24***   −0.50***   −0.04***
   (1.05)   (0.88)   (0.12)   (0.01)
Invest in 19980.59*** 0.59*** 0.59*** 0.59*** 0.59*** 0.59*** 0.59*** 0.59*** 
(0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) 
Invest in 19990.56*** 0.56*** 0.56*** 0.56*** 0.56*** 0.56*** 0.56*** 0.56*** 
(0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) (0.04) 
Invest in 20000.53*** 0.53*** 0.53*** 0.53*** 0.53*** 0.53*** 0.52*** 0.52*** 
(0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) (0.03) 
N 43,677  43,677  43,677  43,677  43,677  43,677  43,677  43,677 
N(IPOs) 8,078  8,078  8,078  8,078  8,078  8,078  8,078  8,078 
Censored obs. 22,091  22,091  22,091  22,091  22,091  22,091  22,091  22,091 
Uncensored obs. 21,586  21,586  21,586  21,586  21,586  21,586  21,586  21,586 
Log likelihood −43,685  −43,685  −43,679  −43,685  −43,674  −43,674  −43,664  −43,668 
Chi-squared70.67*** 71.00*** 82.43*** 84.84*** 90.81*** 90.81*** 110.96*** 102.71*** 

B. Lagged Growth

To ensure that VC results are not spurious based on the VC specification (e.g., absence of previous VC growth), I include a lagged growth rate and instead use an AR(1) methodology.18 The empirical test becomes:

  • image(16)

Inclusion of the lagged growth rate changes the results only slightly. The magnitude of the diversification measures/indexes are changed slightly, but not enough to change the implications or the economic significance of the results. Statistical significance, however, is changed for the international geographic diversification index suggesting that once lagged growth is considered, the benefits of per branch international diversification are slightly less significant (significance drops from 1% to 5%). All other variables remain statistically significant at the 1% level, leaving the vast majority of the VC fundraising results unchanged. Results are seen in Table VI, Panel A.

Table VI.  Alternate Specifications The AR(1) used for venture capitalists is Growthi,t=α+β0 Xi,t1Yt +β2  Growthi,t−13  Diversificationi,t−1i,t. Venture capitalist growth is the log difference in VC capital under management from time t to time t + 1. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. X is a vector of VC firm characteristics including Age, Prefer to Originate, Corporate VC, Previous IPOs, Expertise, and Risk. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return, and Bubble. Measures of diversification in specifications are as follows: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, and 8) IDI is an altered Herfindahl index of industries in which VCi invests. All diversification variables are scaled by capital under management to control for VC size. Control variables are left out for brevity. Panel A includes lagged growth to control for the growth rate of the previous period. Panel B uses a two-step regression wherein the first stage is Diversificationi,t−101Growthi,t−1+μ and the second step follows the regression found in Panel A except the figures used for diversification are predicted residuals from the first stage. Robust standard errors (clustered around VC) are given in parentheses. R2 are reported from the first stage for Panel B.
Panel A. Lagged Growth
 
 Dependent Variable: Venture Capitalist Growth
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
Growthi,t−1−0.174***−0.176***−0.162***−0.170***−0.151***−0.151***−0.145***−0.152***
(0.021)(0.021)(0.021)(0.021)(0.020)(0.020)(0.020)(0.020)
Div. Dimension2.704*** 5.567*** 3.883*** 0.663*** 
(1.014) (0.695) (0.550) (0.191) 
Div. Index 2.747** 4.624*** 0.660*** 0.022***
 (1.285) (0.741) (0.094) (0.005)
N (# of firms) 1,749 (605) 1,748 (605) 1,743 (602) 1,742 (602) 1,845 (615) 1,845 (615) 1,845 (615) 1,845 (615)
Wald statistic 306*** 304*** 375*** 346*** 354*** 354*** 307*** 315***
Panel B. Lagged Growth and Diversification Residuals
 Dependent Variable: Venture Capitalist Growth
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

Growthi,t−1−0.179***−0.180***−0.187***−0.187***−0.164***−0.164***−0.155***−0.162***
(0.021)(0.021)(0.021)(0.021)(0.020)(0.020)(0.020)(0.020)
Div. Dimension2.704*** 5.567*** 3.883*** 0.663*** 
(1.014) (0.695) (0.550) (0.191) 
Div. Index 2.747** 4.634*** 0.660*** 0.022***
 (1.285) (0.739) (0.094) (0.005)
N (# of firms) 1,749 (605) 1,748 (605) 1,743 (602) 1,742 (602) 1,845 (615) 1,845 (615) 1,845 (615) 1,845 (615)
Wald statistic 306*** 304*** 375*** 347*** 354*** 354*** 307*** 315***
F-test (1st stage) 24*** 30*** 84*** 59*** 33*** 33*** 71*** 44***

Although the above results confirm those produced by using Equations (9) and (13), it could be argued that there exists multicollinearity between the diversification and growth rates at time t − 1 in Equation (16). To allay any fears that the results are being influenced by existing multicollinearity between the lagged growth variable and the diversification variable, I estimate, once again, the specification in Equation (16) using a two-step process. The first step regresses diversification on the lagged growth rate as follows:

  • image(17)

where Diversification, Div.Index (regressed separately), and Growth are as defined in Equation (9) and (13). Step 2 involves estimating the regressions found in Equation (16), substituting the predicted residuals from Equation (17) for existing values of Diversification/Div. Index.

The only change in the results involves the international diversification index, which once again shows a drop in significance from 1% to –5%. Statistical significance remains exactly the same otherwise, suggesting that multicollinearity is not a problem in the specification found in Equation (16). Coefficients for the lagged growth change only very slightly and all other coefficients remain the same. This, once again, confirms the results found in the base VC specifications. Results are found in Table VI, Panel B.

C. Alternate Definition of VC Growth

To ensure that results are not dependent on the measurement of venture capital growth, I use ordinal ranking of capital management (instead of the capital under management level) to calculate VC firm growth. The results are nearly identical. The significance remains across the board, as does the relative magnitude. This should not be surprising given the difference between the range in capital under management and rank, as well as the resulting percentage change. Results can be seen in Table VII.

Table VII.  Growth Definition The regression used for venture capitalists is Growthi,t=α+β0 Xi,t1Yt +β2  Diversificationi,t−1i,t. Venture capitalist growth is the log difference in VC rank (Galante's 500 Largest VCs) from time t to time t + 1. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. X is a vector of VC firm characteristics including Age, Prefer to Originate, Corporate VC, Previous IPOs, Expertise, and Risk. Age is the natural log of the number of years VCi has been in business. Prefer to Originate describes the VC's preferred role in a syndication. Corporate VC is a dummy variable indicating whether VCi is a corporate VC or not. Previous IPOs describes the number of IPOs for which VCi is responsible. Expertise is the number of funds VCi has successfully raised. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return, and Bubble. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 Return is the return on the S&P 500 index. Bubble is an indicator variable describing whether time t is during the bubble (i.e., 1998, 1999, or 2000) or not. Measures of diversification in specifications are as follows: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, (6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle. 7) Industries is the number of industries in which VCi invests at time t − 1, and (8) IDI is an altered Herfindahl index of industries in which VCi invests. All diversification variables are scaled by capital under management to control for VC size. Robust standard errors (clustered around VC) appear in parentheses.
 Dependent Variable: Venture Capitalist Growth
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Age0.0030.005−0.010−0.004−0.004−0.0040.0050.003
(0.014)(0.014)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Prefer to Originate−0.015−0.016−0.014−0.018−0.007−0.007−0.012−0.012
(0.016)(0.016)(0.016)(0.016)(0.015)(0.015)(0.015)(0.015)
Corporate VC0.0160.0140.0170.0110.0120.0120.0080.007
(0.014)(0.014)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Previous IPOs−0.004*−0.004*−0.004*−0.004*−0.004**−0.004**−0.004*−0.004*
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Expertise0.0010.001−0.003−0.002−0.001−0.0010.000−0.001
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Risk−0.011−0.012−0.009−0.014*−0.009−0.009−0.015**−0.019**
(0.007)(0.007)(0.008)(0.008)(0.007)(0.007)(0.007)(0.008)
Number of Deals−0.176***−0.179***−0.175***−0.178***−0.146**−0.146**−0.151**−0.146**
(0.066)(0.066)(0.065)(0.065)(0.062)(0.062)(0.063)(0.063)
S&P 500 Return−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***−0.003***
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Bubble0.225***0.223***0.269***0.255***0.217***0.217***0.199***0.209***
(0.060)(0.060)(0.061)(0.061)(0.055)(0.055)(0.055)(0.055)
Div. Dimension−2.850*** −4.542*** −3.169*** −0.713*** 
(0.702) (0.567) (0.500) (0.144) 
Div. Index −2.935*** −3.891*** −0.539*** −0.023***
 (0.713) (0.560) (0.085) (0.004)
N (# of firms) 1,893 (634) 1,892 (634) 1,886 (631) 1,885 (631) 2,013 (655) 2,013 (655) 2,013 (655) 2,013 (655)
Wald stat113.05***111.47***167.90***145.82***136.54***136.54***116.40***122.53***

D. Interaction of VC Characteristics and Diversification

To ascertain whether the impact of diversification on PC success or failure varies across groups within the sample and to ensure that the impact of a missing interaction of VC characteristics and diversification is not confounding the results, I add two possible interactions to the base PC regressions. I first add an interaction term between VC Expertise and Diversification. I then add an interaction term between VC's preferred role in a syndication, Prefer to Originate, and Diversification.

The specifications that include interaction terms for VC Expertise and Diversification yield results that are qualitatively similar to those in Table IV. Overall, diversification measures/indexes do not significantly influence the probability of PC failure and positively influence the probability of PC stagnation. The cumulative impact of stage and industry diversification remains negative and, in most cases, significant for PC exit via M&A and IPO. Specification (7) in Panel A indicates that when a VC diversifies with regard to industries, there is a different impact of this diversification across VC Expertise with regard to PC time to exit via M&A. The results suggest that for a VC with average expertise (Expertise = 3.31), a one standard deviation increase in the industrial diversification measure (Specification (7)) causes a decrease in the probability of PC time to M&A of 1.826%. By increasing the value of expertise by one standard deviation (Expertise = 4.985), the impact of industrial diversification becomes more substantial. VCs with this level of expertise will see a 2.727% drop in the probability of PC time to M&A. Moving in the other direction, decreasing the value of expertise by one standard deviation (Expertise = 1.635), the impact of industrial diversification becomes less substantial. VCs with this level of expertise will see a 0.937% drop in the probability of PC time to M&A. What is noteworthy, however, is the fact that once Expertise is interacted with Diversification, the impact is significant for stage diversification (stage diversification is insignificant in the base specification for PCs in Table IV). This suggests that the impact of stage diversification on time to exit via M&A is significant at some levels of VC expertise.

Results for PC time to exit via IPO remain qualitatively similar to those previous. Results imply that the impact of diversification on PC exit via IPO is dependent on the level of VC expertise, although all levels see a significant negative impact. Moving from a VC that has average expertise to one that has expertise that is one standard deviation higher increases the impact of stage diversification (e.g., in Specification (5), from a 2.34% decrease in the probability in PC time to IPO to a 3.306% decrease). As a VC becomes more of an expert, the cost of stage diversification increases. All forms of diversification display similar relationships for PC exit via IPO. The comprehensive results are found in Table VIII, Panel A.

Table VIII.  Interactions The multinomial logit model used for entrepreneurs is inline image, where CurrentStatus is the current status of the PC: Defunct, Private, Subsidiary, or Public. Ψ is the cumulative logistic probability distribution function. The base specification is Public Status = Private. Inv is a vector of investment characteristics such as Investment Term, Years Since Last Inv, and Portfolio Size/Mgr. X is a vector of venture capitalist characteristics including Growth, Corporate VC, Previous IPOs, and Risk. I is the industry market-to-book ratio for the industry to which the PC belongs. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return, and Bubble. (Exp/PTO) * Div is the interaction term between Expertise and Diversification in Panel A and Prefer to Originate and Diversification in Panel B. Expertise is the number of funds the venture capitalist has raised before time t. Prefer to Originate describes the VCs preferred role in a syndication. The specifications use the following definitions for diversification: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, and 8) IDI is an altered Herfindahl index of industries in which VCi invests. Control variables have been left out for brevity. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. Marginal effects are reported and robust standard errors (clustered around PC) are given in parentheses.
Panel A. Expertise/Diversification Interaction
 Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Dependent Variable: Current Public Status = Defunct (N = 5,361)
Expertise−0.001−0.0010.0000.0000.0000.0000.0000.000
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Expertise * Diversification0.0110.025−0.067−0.046−0.030−0.005−0.0030.000
(0.061)(0.069)(0.048)(0.053)(0.043)(0.007)(0.017)(0.000)
Diversification0.000−0.0440.265*0.2300.0750.0130.0540.002*
(0.149)(0.192)(0.141)(0.158)(0.122)(0.021)(0.042)(0.001)
Dependent Variable: Current Public Status = Private (N = 16,730)
Expertise−0.002*−0.002−0.003**−0.002−0.004***−0.004***−0.004***−0.003***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Expertise * Diversification0.326**0.406***0.520***0.439***0.550***0.094***0.274***0.008***
(0.133)(0.151)(0.104)(0.116)(0.087)(0.015)(0.037)(0.001)
Diversification−0.0080.2200.2420.524−0.136−0.0230.200**0.004
(0.297)(0.382)(0.303)(0.342)(0.249)(0.042)(0.094)(0.002)
Dependent Variable: Current Public Status = Subsidiary (N = 13,508)
Expertise0.0000.0000.0000.0000.0010.0010.0010.001
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Expertise * Diversification−0.115−0.136−0.181**−0.142−0.209***−0.035***−0.101***−0.003***
(0.101)(0.115)(0.079)(0.088)(0.067)(0.011)(0.028)(0.001)
Diversification−0.139−0.252−0.089−0.1530.3040.052−0.0120.000
(0.235)(0.304)(0.232)(0.260)(0.188)(0.032)(0.069)(0.002)
Dependent Variable: Current Public Status = Public (N = 8,078)
Expertise0.003***0.003***0.003***0.002***0.003***0.003***0.004***0.003***
(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)(0.001)
Expertise * Diversification−0.222***−0.296***−0.272***−0.251***−0.311***−0.053***−0.170***−0.005***
(0.079)(0.092)(0.066)(0.074)(0.059)(0.010)(0.024)(0.001)
Diversification0.1470.076−0.418**−0.600***−0.243−0.041−0.242***−0.006***
(0.176)(0.231)(0.191)(0.216)(0.163)(0.028)(0.060)(0.002)
N (total) 43,677 43,677 43,677 43,677 43,677 43,677 43,677 43,677
Chi-squared 2,514*** 2,517*** 2,564*** 2,552*** 2,582*** 2,582*** 2,662*** 2,632***
Log likelihood −54,034 −54,030 −54,001 −54,005 −53,993 −53,993 −53,932 −53,952
Pseudo R20.0490.0490.0490.0490.0490.0490.0400.050
 
Panel B. Syndication/Diversification Interaction
 
Dependent Variable: Current Public Status = Defunct (N = 5,361)
Prefer to Originate−0.009**−0.009**−0.004−0.006−0.010**−0.010**−0.011***−0.009**
(0.004)(0.004)(0.005)(0.004)(0.004)(0.004)(0.004)(0.004)
Prefer to Originate * Diversification−0.203−0.233−0.459−0.3650.0010.0000.0440.000
(0.365)(0.420)(0.298)(0.313)(0.267)(0.045)(0.080)(0.002)
Diversification0.1920.1930.429**0.3590.0240.0040.0260.002
(0.299)(0.328)(0.213)(0.219)(0.199)(0.034)(0.054)(0.001)
Dependent Variable: Current Public Status = Private (N = 16,730)
Prefer to Originate−0.0010.0010.0080.0110.0010.001−0.0020.003
(0.008)(0.008)(0.010)(0.009)(0.009)(0.009)(0.009)(0.008)
Prefer to Originate * Diversification2.746***3.111***0.2350.0210.7730.1310.481***0.009**
(0.804)(0.957)(0.632)(0.678)(0.557)(0.095)(0.179)(0.005)
Diversification−1.703**−1.390*1.078**1.371***0.3650.0620.392***0.010***
(0.664)(0.749)(0.469)(0.493)(0.423)(0.072)(0.120)(0.003)
Dependent Variable: Current Public Status = Subsidiary (N = 13,508)
Prefer to Originate0.0030.0010.002−0.0020.0060.0060.0040.000
(0.006)(0.006)(0.008)(0.007)(0.007)(0.007)(0.007)(0.006)
Prefer to Originate * Diversification−1.712***−1.896***−0.649−0.386−0.971**−0.165**−0.308**−0.006*
(0.598)(0.703)(0.487)(0.517)(0.424)(0.072)(0.136)(0.003)
Diversification1.025**0.899*−0.012−0.1900.568*0.097*−0.007−0.001
(0.487)(0.545)(0.356)(0.371)(0.316)(0.054)(0.089)(0.002)
Dependent Variable: Current Public Status = Public (N = 8,078)
Prefer to Originate0.0070.007−0.005−0.0030.0020.0020.010*0.007
(0.005)(0.005)(0.006)(0.005)(0.006)(0.006)(0.006)(0.005)
Prefer to Originate * Diversification−0.830*−0.982*−0.873**−0.731*−0.197−0.034−0.217*−0.003
(0.464)(0.549)(0.398)(0.433)(0.352)(0.060)(0.126)(0.003)
Diversification0.4850.299−1.494***−1.540***−0.957***−0.163***−0.411***−0.010***
(0.378)(0.428)(0.297)(0.319)(0.263)(0.045)(0.081)(0.002)
N (total) 43,677 43,677 43,677 43,677 43,677 43,677 43,677 43,677
Chi-squared 2,509*** 2,507*** 2,554*** 2,546*** 2,542*** 2,542*** 2,613*** 2,587***
Log likelihood −54,032 −54,030 −54,010 −54,011 −54,013 −54,013 −53,963 −53,981
Pseudo R20.0490.0490.0490.0490.0490.0490.0500.049

Results for specifications that include the interaction between the VC's preferred role in syndications, Prefer to Originate, and Diversification also achieve qualitatively similar results to those previous (i.e., diversification positively impacts PC stagnation and negatively impacts PC time to exit). The significance of the interactive effect implies that the impact of diversification differs for the two groups of VCs, those that prefer to originate in a syndication and those that do not. Specifically, the negative impact of diversification is greater for those VCs that originate in a syndication. In fact, in the case of stage diversification and time to exit via M&A, VCs that do not originate in a syndication see a weakly positive impact of this form of diversification. This diversity in the affect of diversification across syndication role helps to explain the lack of significance for stage diversification in Table IV, Panel C.

Looking to the last section of Panel B, time to PC exit via IPO, the relative lack of significance in the interactive terms suggests that there is not much of a difference in the negative impact of diversification on PC time to IPO across syndication role groups. The economic significance of the cumulative impact of diversification suggests that this is indeed the case. A systematic relationship does not exist between the magnitude of the impact of diversification between VCs that prefer to originate and those that do not.

E. Fund Size

To control for the fact that not all of the venture capitalist in Galante's are limited partnerships and that fund size may be fixed across the sample period, I include a variable to control for the size of the fund.19 This results in enhanced marginal effect for the diversification measures/indexes when compared to those in Table III. The fund size variable is statistically significant at the 1% level and positively influences VC growth as we would expect it to. Results are found in Table IX.

Table IX.  Investment Amount The regression used for venture capitalists is Growthi,t=α+β0 Xi,t1Yt +β2  TotalInv +β3  Diversificationi,t−1i,t. Venture capitalist growth is the log difference in VC capital under management from time t to time t + 1. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. X is a vector of VC firm characteristics including Total Investment, Age, Prefer to Originate, Corporate VC, Previous IPOs, Expertise, and Risk. Total Investment is the natural log of the total capital (in 000s) that the VC fund has invested in all portfolio companies. Age is the natural log of the number of years VCi has been in business. Prefer to Originate describes the VC's preferred role in a syndication. Corporate VC is a dummy variable indicating whether VCi is a corporate VC or not. Previous IPOs describes the number of IPOs for which VCi is responsible. Expertise is the number of funds VCi has successfully raised. Risk is an index from zero to two that sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early- Stage PCs, respectively. Y is a vector of macroeconomic variables such as Number of Deals, S&P 500 Return, and Bubble. Number of Deals is the natural log of the number of deals (investments) in the VC industry at time t. S&P 500 is the return on the S&P 500 index. Bubble is an indicator variable describing whether time t is during the bubble (i.e., 1998, 1999, or 2000) or not. Measures of diversification in specifications are as follows: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, 8) IDI is an altered Herfindahl index of industries in which VCi invests. Robust standard errors (clustered around VC) appear in parentheses.
 Dependent Variable: Venture Capitalist Growth
Intl Geog. (1)IGDI (2)Dom Geog. (3)DGDI (4)Stages (5)SDI (6)Industries (7)IDI (8)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Total Investment0.022***0.022***0.035***0.031***0.032***0.032***0.032***0.033***
(0.006)(0.006)(0.007)(0.007)(0.006)(0.006)(0.006)(0.006)
Age−0.020−0.022−0.005−0.012−0.012−0.012−0.022*−0.021
(0.013)(0.013)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Prefer to Originate0.0100.0120.0060.0130.0000.0000.0070.006
(0.015)(0.015)(0.015)(0.015)(0.015)(0.015)(0.014)(0.014)
Corporate VC−0.027**−0.026*−0.030**−0.022−0.027**−0.027**−0.021−0.02
(0.014)(0.014)(0.014)(0.014)(0.013)(0.013)(0.013)(0.013)
Previous IPOs0.0020.0010.0010.0010.0010.0010.0010.001
(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)(0.002)
Expertise0.004*0.004*0.008***0.007**0.005*0.005*0.0040.005*
(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)(0.003)
Risk0.0080.0090.0030.0110.0040.0040.014**0.019***
(0.007)(0.007)(0.008)(0.007)(0.007)(0.007)(0.007)(0.007)
Number of Deals0.550***0.554***0.548***0.553***0.529***0.529***0.534***0.528***
(0.066)(0.066)(0.064)(0.065)(0.062)(0.062)(0.063)(0.063)
S&P 500 Return−0.001*−0.001*−0.001**−0.001**−0.001*−0.001*−0.001**−0.001**
(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)(0.000)
Bubble−0.293***−0.293***−0.355***−0.335***−0.314***−0.314***−0.293***−0.305***
(0.060)(0.060)(0.060)(0.061)(0.055)(0.055)(0.054)(0.054)
Div. Dimension3.375*** 6.116*** 4.270*** 1.152*** 
(0.904) (0.735) (0.668) (0.182) 
Div. Index 3.794*** 5.140*** 0.726*** 0.034***
 (0.938) (0.699) (0.114) (0.005)
N (# of firms) 1,892 (633) 1,891 (633) 1,885 (630) 1,884 (630) 2,012 (654) 2,012 (654) 2,012 (654) 2,012 (654)
Wald stat268.46***271.34***303.90***295.21***285.21***285.21***289.56***305.79***

F. Differences in VC Expertise

In order to take a closer look at the impact of diversification on the probability of PC success at different levels of VC Expertise, I divide the sample into terciles: 1) Novice VCs (observations with expertise less than or equal to the 25th percentile), 2) Average VCs (observations with expertise greater than 25th percentile, but less than 75th percentile), and 3) Expert VCs (observations with expertise greater than or equal to the 75th percentile). Overall, the results suggest that the impact of diversification on PC current status depends on the expertise of the VC. Results in the base PC specification belie the differences across VC expertise. Thus, important distinctions can and should be noted. Probably the most interesting is the distinction in the results for probability of PC failure (i.e., PC Public Status = Defunct). Novice VCs actually increase the probability of PC failure with all forms of diversification except international. This suggests that there is a learning curve in the role of VCs and that the hand holding function needs to be mastered before diversifying on any dimension. It could also imply that the affiliation that VCs gain over time is more important than the hand holding at some level. Average VCs actually decrease the probability of PC failure by diversification and Expert VCs see no effect on the probability of PC failure with an increase in diversification. This variance in the impact of diversification across VC expertise in the probability of PC failure is telling of the importance of VC expertise in its role of manager to its PCs.

The other main difference that can be found is that stage diversification, which was found to have a statistically insignificant effect on PC exit via M&A (see Table IV) is actually statistically significant for Average and Expert VCs in this analysis. Novice VCs see no affect on probability of PC exit via M&A for an increase in diversification. If M&A is an inferior form of PC exit (Schwienbacher, 2002; Cumming and MacIntosh, 2003a, 2003b; Fleming, 2004), these results could reflect the fact that VCs do not become skilled enough to choose PCs that have the quality necessary to exit via IPO until they are more experienced. Only the impact of diversification on PC exit via IPO is relatively consistent across levels of VC expertise. Results are found in Table X.

Table X.  Differences in VC Expertise The multinomial logit model used for entrepreneurs is inline image, where Ψ is the cumulative logistic probability distribution function. Current Status is the current status of the PC: Public, Acquired, Private, or Defunct. The base specification is Public Status = Private. The regression is run on subsamples of the original sample that include Novice VCs (≤ 25th percentile of sample), Average VCs (> 25th percentile and < 75th percentile of sample), and Expert VCs (≥ 75th percentile of sample). Inv is a vector of investment characteristics such as Investment Term, Years Since Last Inv, and Portfolio Size/Mgr. Xi is a vector of venture capitalist characteristics including Prefer to Originate, Corporate VC, Previous IPOs, Expertise, and Risk. I is the industry market-to-book ratio for the industry to which the PC belongs. Y is a vector of macroeconomic variables including Number of Deals, S&P 500 Return and Bubble. The specifications use the following definitions for diversification: 1) Intl Geography is the number of international regions in which VCi invests at time t − 1, 2) IGDI is the number of international regions per branch in which VCi invests at time t − 1, 3) Dom Geography is the number of US regions in which VCi invests at time t − 1, 4) DGDI is the number of US regions per branch in which VCi invests at time t − 1, 5) Stages is the number of stages in which VCi invests at time t − 1, 6) SDI is the number of stages in which VCi invests at time t − 1 divided by the number of stages in the portfolio firm life cycle, 7) Industries is the number of industries in which VCi invests at time t − 1, and 8) IDI is an altered Herfindahl index of industries in which VCi invests. Control variables are left out for brevity. VC rankings are from Galante's Private Equity and Venture Capital Directory, 1998-2006 (only US firms). Investment (PC) data specifics are from VentureXpert. Marginal effects are reported and robust standard errors (clustered around PC) are given in parentheses.
 Novice VCsAverage VCsExpert VCs
Intl Geog. (1)Dom Geog. (2)Stages (3)Industries (4)Intl Geog. (1)Dom Geog. (2)Stages (3)Industries (4)Intl Geog. (1)Dom Geog. (2)Stages (3)Industries (4)
  1. ***Significant at the 0.01 level.

  2.  **Significant at the 0.05 level.

  3.   *Significant at the 0.10 level.

Dependent Variable: Public Status = Defunct (N: 1,747 (25th percentile); 1,460 (50th percentile); 2,154 (75th percentile)
Div. Dimension0.1540.462***0.180*0.076**−2.220***−1.014***−0.629**0.0610.267−0.550−0.3210.093
(0.112)(0.109)(0.093)(0.036)(0.766)(0.344)(0.314)(0.086)(0.431)(0.358)(0.330)(0.159)
Dependent Variable: Public Status = Private (N: 6,364 (25th percentile); 4,194 (50th percentile);6,172 (75th percentile)
Div. Dimension−0.0760.3470.0750.396***5.461***2.959***3.726***0.871***1.5622.987***2.749***1.855***
(0.261)(0.266)(0.214)(0.084)(1.351)(0.669)(0.631)(0.192)(0.950)(0.686)(0.613)(0.308)
Dependent Variable: Public Status = Subsidiary (N: 4,380 (25th percentile); 3,418 (50th percentile); 5,710 (75th percentile)
Div. Dimension−0.065−0.0960.233−0.102−2.121**−1.227**−1.108**−0.274*−0.970−0.940*−1.304***−0.737***
(0.197)(0.200)(0.156)(0.062)(1.071)(0.512)(0.482)(0.145)(0.750)(0.549)(0.498)(0.256)
Dependent Variable: Public Status = Public (N: 2,534 (25th percentile); 2,071 (50th percentile); 3,473 (75th percentile)
Div. Dimension−0.013−0.714***−0.488***−0.370***−1.119−0.718*−1.989***−0.658***−0.858−1.497***−1.124***−1.211***
(0.150)(0.171)(0.140)(0.055)(0.796)(0.379)(0.399)(0.125)(0.631)(0.482)(0.409)(0.229)
N (total) 15,025 15,025 15,025 15,025 11,143 11,143 11,143 11,143 17,509 17,509 17,509 17,509
Chi-squared 1,124*** 1,156*** 1,136*** 1,168*** 952*** 942*** 960*** 962*** 1,322*** 1,338*** 1,352*** 1,374***
Log likelihood −18,303 −18,287 −18,295 −18,275 −13,833 −13,833 −13,822 −13,824 −21,799 −21,791 −21,790 −21,776
Pseudo R20.0440.0440.0440.0450.0520.0520.0530.0530.0510.0510.0510.052

VI. Conclusions

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

This paper examines a potential misalignment of interests between venture capitalists and the firms in which they invest, portfolio companies. Diversification dimensions for the VC that seem to have a significantly positive affect on their growth, stage, and industrial diversification seem to have the biggest delaying impact on PC exit (i.e., IPO or in some cases M&A). The evidence presented in this paper suggests that the diversification dimensions that could undermine the expertise role in the VC/PC relationship are the most harmful to these young companies. This supports extant literature that finds that venture capital investment has value in the hand holding that it provides its investment companies. If the PC is using venture capital as a vehicle to the public markets, the lack of hand holding and the resulting delay in exit that is found with more diversified VCs is not desirous.

Institutional investors may find these results important in establishing covenants that constrain VCs in their investment scope. To the extent that VCs make a significant portion of their profit via expeditious PC exit, diversification may not ultimately be desirable for the VC either. If risk reduction must be undertaken through diversification, dimensions that exhibit a less negative impact on the probability of timely entrepreneurial exit should be undertaken. Examples of these are domestic and international geographic diversification. These diversification dimensions might be undertaken as a relatively beneficial strategy since these dimensions are positively related to VC growth, have the smallest deleterious affect on the probability of PC exit, and, in some cases, are even insignificantly related to PC exit (in the case of international diversification). Hence, geographic diversification can offer comfort in providing an option for diversification that will not risk a PC's viability. Hot and cold IPO markets might be considered when evaluating these forms of diversification since the viability of exit options will be dependent on the temperature of the market.

To the extent that PCs are able to shop around for venture capital funding, and that their ultimate goal is to obtain access to public markets quickly, they may want to seek out a VC who is either a pure-play or is diversified across geographic locations only. VCs that diversify across industry or stage should be avoided if delay of exit is of prime importance. ▪

Appendix

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References

Appendix. Variable Definitions and Sources

VariableDefinitionSource
VC Characteristics
Prefer to OriginateA dummy variable describing the preferred role VCi takes in syndications equal to one if the VC prefers to originate and zero otherwise.Galante's
Corporate VCA dummy variable that takes on a value of one where VCi is a corporate venture capitalist and zero otherwise.SDC Platinum
Previous IPOsThe number of IPOs for which VCi is responsible.SDC Platinum
AgeThe natural log of the number of years VCi has been in existence (in 2006—Firm Founding Date). 
ExpertiseThe number of successful funds VCi has closed.SDC Platinum
GrowthThe logarithmic difference in capital under management (Table III) or rank (Table VII) scaled by time t + 1 minus time t.Galante's; own calculation
Total InvestmentThe natural log of the total amount (in 000s) VCi has invested in all PCs.SDC Platinum
VC Investment Preferences
Intl GeographyThe number of international regions in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
IGDIThe number of international regions per branch in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
Dom GeographyThe number of US regions in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
DGDIThe number of US regions per branch in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
StagesThe number of stages in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
SDIThe number of stages in which VCi invests at time t − 1, divided by the number of stages in the portfolio firm life cycle, scaled by capital under management at time t − 1.Galante's
IndustriesThe number of industries in which VCi invests at time t − 1, scaled by capital under management at time t − 1.Galante's
IDIAn altered Herfindahl index of industries in which VCi invests, scaled by capital under management at time t − 1.Galante's
RiskAn index from zero (low) to two (high) which sums IT Dummy and Early-Stage Dummy, indicators of whether VCi invests in the IT and/or Early-Stage PCs, respectively.Galante's
Investment Specifics
Investment TermThe year VCi last invested in PCj minus year VCi first invested in PCj.SDC Platinum
Year Last InvThe year in which VCi last invested in PCj.SDC Platinum
Industry M/BThe market-to-book ratio for the industry to which PCj belongs (Data Item 24 * Data Item 25)/Data Item 60).Compustat
Portfolio Size/MgrThe number of PCs in which VC fund invests divided by the number of managerial staff in the VC.SDC Platinum
Market Conditions
Number DealsThe natural log of the number of VC deals (investments) at time t.VentureExpert
S&P 500 ReturnThe return on the S&P 500 index.Standard & Poor's
BubbleDummy variable if time t is during the market bubble (t = 1998, 1999, 2000). 
Footnotes
  • 1

    Venture capital funds are legally distinct from mutual funds.

  • 2

    An “angel” is an alternative source of capital for portfolio companies. These wealthy individuals provide their own capital (versus funds raised from others) and do so in smaller amounts. According to http://www.angelcapitaleducation.org, this amount is generally $5,000-$100,000.

  • 3

    Limited partners are investors in venture capital funds that enjoy a portion of the partnership's investment cash flows but cannot be held liable for any legal issues within the contractual partnership between the VC and its investment, the portfolio company.

  • 4

    Growth in ranking is used in the robustness section of the paper.

  • 5

    See Fried and Hisrich, 1994 for information-gathering benefits of the VC specialist relative to their generalist counterparts.

  • 6

    Since both VCs and PCs benefit from entrepreneurial exit, the probability of PC exit is meant to represent time to exit.

  • 7

    Geographic indexes might be biased due to the fact that number of branches is not broken down into domestic versus otherwise.

  • 8

    PC outcome = private is the base outcome.

  • 9

    Since observations are based on the term of investment in the portfolio companies and the current status, clustering around time cannot be done.

  • 10

    In cases where Galante's lists “global” (“anywhere in the United States”) for geographic preference of a given VC, the number of international (US) regions is coded at the maximum, which is the number of international (US) regions possible for the directory, or 12 (eight).

  • 11

    In cases where Galante's lists “all stages” for stage preference of a given VC, the number of stages is coded at the maximum, which is the number of stages possible for the directory, or 17.

  • 12

    In cases where Galante's lists “diversified” for industry preference of a given VC, the number of industries is coded at the maximum, which is the number of industries possible for the directory, or 30.

  • 13

    This figure is obtained by multiplying the standard deviation of the domestic geography diversification variable, approximated to the nearest hundredth in Table I, by the coefficient, 4.903 (Table III, Specification (3)).

  • 14

    One standard deviation for a VC falling in the 10th percentile is an increase of one US region. For the median VC, this increase amounts to an increase of approximately eight US regions. These should be taken lightly, however, since approximating the economic impact of a marginal effect of a one region increase in domestic diversification assumes that capital is held constant. In fact, both the numerator and the denominator can (and do) vary in the diversification measures. This causes estimated values of the numerator to be overestimated as the denominator's value increases. Discussions of economic significance will, therefore, be given in terms of standard deviation and are meant to reflect increases in the ratio, not in the numerator only.

  • 15

    Not reported are the results obtained from altering the base specification for the VC by substituting S&P 500 and the Number of Deals with time dummies to account for differences in VC investing across time. Results remain and, thus, are left out for brevity, but are available on request.

  • 16

    These two figures are obtained by multiplying the coefficients −0.523 and −0.012 (Specifications (7) and (8) of Table IV, Panel A) with their corresponding standard deviations rounded to the nearest hundredth in Table I.

  • 17

    This result cannot be compared directly with that of Brau, Francis, and Kohers (2003) who find that firms that are in the high-tech industry are more likely to exit via IPO. Risk here tells us whether the VC invests in either the IT industry or the early stage of the PC life cycle. A given PC may or may not be in the IT industry if a VC has a risk score above zero.

  • 18

    Specifications run with a modified Sharpe ratio equal to VC Growth divided by the risk variable as the dependent variable (instead of growth in capital under management) produce results in confirmation with those of previous specifications.

  • 19

    Cumming (2006), for example, demonstrates that “there is decreasing returns to scale in the number of entrepreneurial firms financed by a venture capital fund,” suggesting that there could be a point at which VCs would optimally want to limit growth.

References

  1. Top of page
  2. Abstract
  3. I. The Risky Business of Venture Capital
  4. II. The Impact of Diversification
  5. III. Data Resources
  6. IV. Consequences of Diversification as a Strategy
  7. V. Robustness
  8. VI. Conclusions
  9. Appendix
  10. References
  • Alternative Investor, 1998-2006, Galante's Venture Capital & Private Equity Directory, Alternative Investor, Wellesley .
  • Bernile, G., D. Cumming, and E. Lyandres, 2007, “The Size of Private Equity Fund Portfolios: Theory and International Evidence,” Journal of Corporate Finance 13, 564-590.
  • Brau, J., B. Francis, and N. Kohers, 2003, “The Choice of IPO versus Takeover: Empirical Evidence,” Journal of Business 76, 583-612.
  • Cochrane, J., 2005, “The Risk and Return of Venture Capital,” Journal of Financial Economics 75, 3-52.
  • Cumming, D., 2006, “The Determinants of Venture Capital Portfolio Size: Empirical Evidence,” Journal of Business 79, 1083-1126.
  • Cumming, D., 2008, “Contracts and Exits in Venture Capital Finance,” Review of Financial Studies 21, 1947-1982.
  • Cumming, D., G. Fleming, and A. Schwienbacher, 2005, “Liquidity Risk and Venture Capital Finance,” Financial Management 34, 77-105.
  • Cumming, D., G. Fleming, and A. Schwienbacher, 2006, “Legality and Venture Capital Exits,” Journal of Corporate Finance 12, 214-245.
  • Cumming, D. and J. MacIntosh, 2003a, “Venture Capital Exits in Canada and the United States,” University of Toronto Law Journal 55, 101-200.
  • Cumming, D. and J. MacIntosh, 2003b, “A Cross-Country Comparison of Full and Partial Venture Capital Exits,” Journal of Banking and Finance 27, 511-548.
  • Cumming, D. and U. Walz, 2009, “Private Equity Returns and Disclosure around the World,” Journal of International Business Studies, forthcoming.
  • Elango, B., V. Fried, R. Hisrich, and A. Polonchek, 1995, “How Venture Capital Firms Differ,” Journal of Business Venturing 10, 157-179.
  • Fleming, G., 2004, “Venture Capital Returns in Australia,” Venture Capital: An International Journal of Entrepreneurial Finance 6, 23-45.
  • Fried, V. and R. Hisrich, 1994, “Toward a Model of Venture Capital Investment Decision Making—Venture Capital Special Issue,” Financial Management 23, 28-37.
  • Fulghieri, P. and M. Sevilir, 2009, “Size and Focus of a Venture Capitalist's Portfolio,” Review of Financial Studies, forthcoming.
  • Gompers, P., 1995, “Optimal Investment, Monitoring, and the Staging of Venture Capital,” Journal of Finance 50, 1461-1489.
  • Gompers, P., 1996, “Grandstanding in the Venture Capital Industry,” Journal of Financial Economics 42, 133-156.
  • Gompers, P. and J. Lerner, 1996, “The Use of Covenants: An Empirical Analysis of Venture Capital Partnership Agreements,” Journal of Law and Economics 39, 463-498.
  • Gompers, P. and J. Lerner, 1999, The Venture Capital Cycle, 1st Ed., Cambridge , MA , MIT Press.
  • Gompers, P., J. Lerner, M. Blair, and T. Hellmann, 1998, “What Drives Venture Capital Fundraising? Brookings Papers on Economic Activity, Microeconomics 149-204.
  • Groshen, E. and S. Potter, 2003, “Current Issues in Economics and Finance,” Federal Reserve Bank of New York 9, 1-7.
  • Heckman, J., 1979. “Sample Selection Bias as a Specification Error,” Econometrica 47, 153-161.
  • Hellmann, T., 2000, “Venture Capitalists: The Coaches of Silicon Valley” in W.Miller, C.M.Lee, M.Hanock, and H.Rowen, Eds., Silicon Valley Edge: A Habitat for Innovation and Entrepreneurship, Stanford , CA , Stanford University Press.
  • Hellmann, T. and M. Puri, 2000, “The Interaction between Product Market and Financing Strategy: The Role of Venture Capital,” Review of Financial Studies 13, 995-984.
  • Hellmann, T. and M. Puri, 2002, “Venture Capital and the Professionalization of Start-Up Firms: Empirical Evidence,” Journal of Finance 57, 169-198.
  • Hsu, D., 2004, “Do Entrepreneurs Pay for Affiliation? Journal of Finance 59, 1805-1844.
  • Kanniainen, V. and C. Keuschnigg, 2003, “Optimal Portfolio of Start-Up Firms in Venture Capital Finance,” Journal of Corporate Finance 9, 521-534.
  • Kanniainen, V. and C. Keuschnigg, 2004, “Start-Up Investment with Scarce Venture Capital Support,” Journal of Banking & Finance 9, 521-534.
  • Kaplan, S. and Strömberg, P., 2001, “Venture Capitalists as Principals: Contracting, Screening, and Monitoring,” American Economic Review 91, 426-430.
  • Lerner, J., 1994, “The Syndication of Venture Capital Investments,” Financial Management 23, 16-27.
  • Lerner, J., 1995, “Venture Capital and the Oversight of Private Firms,” Journal of Finance 50, 301-318.
  • Leshchinskii, D., 1999, “Venture Capitalists as Benevolent Vultures: The Role of Network Externalities in Financing Choice,” HEC Working Paper.
  • MacMillan, I., D. Kulow, and R. Khoylian, 1988, “Venture Capitalists, Involvement in Their Investments, Extent and Performance,” Journal of Business Venturing 4, 27-47.
  • Markowitz, H., 1952, “Portfolio Selection,” Journal of Finance 7, 77-91.
  • Megginson, W. and K. Weiss, 1991, “Venture Capital Certification in Initial Public Offerings,” Journal of Finance 46, 879-903.
  • National Venture Capital Association, 2004, “The Venture Capital Industry—An Overview.” Available at: http://www.nvca.org/def.html.
  • Norton, E. and B. Tenenbaum, 1993, “Specialization versus Diversification as a Venture Capital Investment Strategy,” Journal of Business Venturing 8, 431-442.
  • Ross, S., R. Westerfield, and B. Jordan, 2006, Fundamentals of Corporate Finance, New York , NY , McGraw-Hill Irwin, 278-282.
  • Schwienbacher, A., 2002, “An Empirical Analysis of Venture Capital Exits in Europe and the United States,” University of Amsterdam Working Paper.
  • Servaes, H., 1996, “The Value of Diversification during the Conglomerate Merger Wave,” Journal of Finance 51, 1201-1226.
  • Sharpe, W., 1963, “A Simplified Model of Portfolio Analysis,” Management Science 9, 277-293.
  • Sharpe, W., 1964, “Capital Asset Prices: A Theory of Market Equilibrium Under Conditions of Risk,” Journal of Finance 19, 425-442.
  • Stein, J., 1997, “Internal Capital Markets and the Competition for Corporate Resources,” Journal of Finance 52, 111-133.
  • Thomson Financial, 2006, SDC Platinum, New York , NY , Thomson Financial.