THE OPTIMAL PORTFOLIO OF START-UP FIRMS IN VENTURE CAPITAL FINANCE: THE MODERATING EFFECT OF SYNDICATION AND AN EMPIRICAL TEST
Mikko Jääskeläinen, Helsinki University of Technology
Markku Maula, Helsinki University of Technology
Tuukka Seppä, Helsinki University of Technology
Recent theoretical research on the optimal size of venture capitalists’ portfolio has posited that the number of portfolio companies a venture capitalist manages and the total returns of the venture capital partnership will exhibit an inverted U-shaped relationship. We test this theory and further expand it proposing that syndication moderates the relationship so that the higher the level of syndication, the higher the optimal number of portfolio companies per venture capitalist. Our empirical analysis of longitudinal data of the largest 97 U.S. venture capital investors provides strong evidence for the hypotheses.
Previous research has identified that venture capitalists participate actively in the monitoring and management of their portfolio companies in addition to the money invested. Several empirical studies have reported important value added benefits provided by venture capitalists. These benefits include help in obtaining additional financing, help in strategic planning, and help in recruiting key executives (Gorman and Sahlman, 1989; MacMillan et al., 1988, Rosenstein, 1993; Sapienza, 1992; Sapienza et al. 1996; Hellman and Puri, 2001). Although venture capitalists provide ventures with value-added services, their time is a scarce resource. Venture capitalists need to both nurture their existing portfolio companies and evaluate new ones on a constant basis. Thus, venture capitalists need to allocate their time optimally between the current portfolio in order to ensure its performance, and new investments to guarantee continuity.
While previous research suggests that venture capitalists attempt to add value to their investments beyond money, the limitations venture capitalists face in this task have received less attention. Although the value-adding activities of venture capitalists have received fair attention, the focus in previous studies has been on the portfolio company (e.g. Barney et al., 1996; Gorman and Sahlman, 1989; Hellman and Puri, 2001; MacMillan et al., 1988; Rosenstein, 1993; Sapienza, 1992, 1996). Furthermore, most of these studies measure value-added as it is perceived by various stakeholders rather than using measures for the outcomes of these ventures. This is mostly due to the limited availability of performance data from venture capitalists and the cross-sectional nature of the studies. To our knowledge, there are no studies linking the value-added involvement of venture capitalists in their ventures to the performance of the venture capitalists.
Venture capital firms are typically organised as limited partnerships, where the managers of funds act as general partners of the firms (Sahlman, 1990). The number of partners is typically small; Gorman and Sahlman (1989) report the average number of active partners per firm to be 4.7. While the small number of partners implies limited resources, it also means that there are fewer to divide the profits. Thus, an additional partner would increase the total amount of resources, but unless it also increases the overall efficiency of the firm, the outcome is negative for the existing partners. Venture capitalists can, however, co-operate with other venture capitalists by syndicating their investments, and thus potentially increase their capacity to manage portfolio companies. Earlier research on the syndication activities of venture capitalists has considered both the rationales of an individual venture capitalist to syndicate (Lerner, 1994; Brander et al., 1999), the network structures that syndication relationships create among venture capitalists (Bygrave, 1988), and their implications for information sharing (Sorenson and Stuart, 2001). Syndication has been linked to performance on the level of individual investments (Brander et al., 1999), but the link between the venture capitalists’ value-adding involvement and syndication has not been studied.
This study sets out to examine how the size of the venture capital firm’s portfolio and its ability to co-operate with other firms is related to performance of the firm. Firstly, we establish an empirical relationship between the number of portfolio firms per partner and the performance of the venture capitalist. Secondly, we examine how this relationship is moderated by the amount of syndication activity with other venture capital firms.
Our findings have both theoretical and practical implications. The results support the recent theoretical research on the optimal portfolio size and expand the theory by introducing the moderating role of syndication on the optimal size of the portfolio. Furthermore, our findings provide guidance for venture capitalists on successful strategies and resource allocation.
Venture capitalists are investors that typically attempt to add value to their investments beyond money. Previous research has identified several mechanisms that venture capitalists use to add value to their portfolio companies. For instance, MacMillan et al. (1988) reported that activities with the highest degree of venture capitalists involvement were (1) serving as a sounding board to the entrepreneur team, (2) helping the firm obtain alternative further sources of equity financing, (3) interfacing with the investor group, (4) monitoring financial performance, (5) monitoring operating performance, and (6) helping their portfolio firms attract alternative sources of debt financing. Similarly, Gorman and Sahlman (1989) documented a ranked order of the forms of assistance as follows: (1) help with obtaining additional financing, (2) strategic planning, (3) management recruitment, (4) operational planning, (5) introductions to potential customers and suppliers, and (6) resolving compensation issues. Sapienza et al. have conducted several studies on sources of value-added provided by both U.S. and European venture capitalists (Sapienza et al., 1994; 1996). In their studies, strategic roles have ranked highest followed by interpersonal roles. Strategic roles include acting as a sounding board as well as being a finance and business advisor. Rosenstein et al. (1993) have also examined this issue from the CEO’s perspective, concluding that the most important roles for professional equity providers are principally monitoring financial performance, serving as a sounding board to the entrepreneur team and recruiting/replacing the CEO. Synthesizing the results of the earlier studies, the most valuable contributions of independent venture capitalists other than the initial provision of capital are in arranging additional financing, supporting strategy making, and recruiting key executives.
While most venture capitalists would probably like to add as much value to their portfolio companies as possible, there are natural limitations to their ability to do so. The most evident limitation is the number of companies each managing partner needs to take care of. Managing partners are typically the key contact between the venture capital partnership and the portfolio company, and the time managing partners have in their use is a critical resource for the partnership (Sahlman, 1990). Intuitively, there should exist an optimal number of portfolio companies per managing partner—a number large enough to take full advantage of the focal managing partner’s abilities, but small enough to be manageable for the partner.
In their theoretical framework proposing a unique optimal number of portfolio companies, Kanniainen and Keuschnigg (2000) proposed that the effort cost be convex in relation to the number of portfolio companies. This proposition is based on an assumption that diluting the value-adding capacity of a venture capitalist for additional start-up companies decreases the expected returns from every company in the portfolio because the chances of survival decrease. An individual venture capitalist has only a limited amount of working hours available for advising portfolio companies. Stretching beyond the limits will automatically imply lower attention per portfolio company. To balance the entrepreneurs’ incentives, they must be compensated with a higher share of returns, which causes the diminishing total returns for venture capitalists after certain number of portfolio companies.
Besides the double-moral hazard framework of Kanniainen and Keuschnigg (2000), there are also other explanations suggesting a similar curvilinear relationship between the number of portfolio companies and the performance of the venture capital firm. In the strategic alliance literature, it has been demonstrated that there exist a curvilinear relationship between the number of alliances and the performance of focal firms (Rothaermel and Deeds, 2001). Rothaermel and Deeds also identified several reasons for this curvilinear relationship, of which some are applicable to our context. One prominent argument is based on transaction cost theory. Based on the transaction cost logic, an increasing number of portfolio companies may cause a venture capitalist’s transaction costs to rise up to, and possibly beyond, a point where the gains from additional portfolio companies are outweighed by their costs, resulting in negative marginal returns for high numbers of portfolio companies per partner (Jones and Hill, 1988). We assume a curvilinear (inverted U-shaped) relationship with respect to the number of portfolio companies per VC partner and the performance of the VC firm. In particular, we propose that the relationship between the number of portfolio companies per venture capitalist and the performance of the venture capital firm will exhibit diminishing marginal returns, and may even pass the point of diminishing total returns and thus eventually exhibit negative marginal returns for each additional portfolio company.
Hypothesis 1: There is a curvilinear (inverted U-shaped) relationship between the number of portfolio companies per VC partner and the performance of the VC partnership.
This simple framework on the optimal number of portfolio companies does not yet take into account the potential importance of VC firm specific factors in explaining the relationship between the number of portfolio companies per VC partner and the performance of the VC firm. We argue that the level of syndication is one firm specific factor that influences a VC firm’s capability to simultaneously manage multiple portfolio companies.
In their insightful paper, Brander, Amit and Antweiler (1999) formally model and empirically compare two possible reasons for syndication. First, syndication might improve project selection, as an additional venture capitalist provides a valuable “second opinion.” Alternatively, syndication might arise because additional venture capitalists add value to a venture through complementary skills. These alternatives imply contrasting predictions about comparative returns to syndicated and stand-alone investments. The empirical analysis, using Canadian data, favours the value-added interpretation. Their results suggest that syndication is done in order to combine heterogeneous competencies of venture capitalists. This saves the efforts of the venture capitalists because each syndicating venture capitalist can concentrate on what he/she is best at. Furthermore, the lead investor typically takes a time-consuming board seat but often the other syndicating investors do not. Other syndicating investors trust the monitoring of the lead investor and thus are able to reduce their workload.
Based on these arguments, we hypothesize that VC firms with higher level of syndication should be able to simultaneously manage a larger number of portfolio companies per VC partner. Accordingly, the point of diminishing total returns to the number of portfolio companies per VC partner should be reached at a higher number of portfolio companies for more syndicating VC firms than for less syndicating VC firms.
Hypothesis 2: The higher the share of syndicated deals, the higher the optimal number of portfolio companies per VC partner.
Data
Our sample consists of the 97 largest private U.S. venture capital organizations and their investments. We identify them based on the number of portfolio companies the firms had invested in by the end of year 2000. The sample is restricted 1) to contain only U.S. venture capitalists and investments, 2) to those U.S. venture capital partnerships that Venture Economics classifies as “Independent private partnerships” thus excluding investment bank affiliates, corporate investors, endowments, individuals, or other private equity investors, 3) to those U.S. venture capital firms that invest mainly in U.S. companies in order to standardise both the investments and their performance as well as the nature of syndication relationships excluding cross-border activities, and 4) to standard venture capital investments removing records that Venture Economics classifies as “Leveraged buyout,” “Secondary purchase,” “Open market purchase,” “Private investment in public company” or “Turnaround.”
Altogether, the data includes 27,848 investments in 4,755 portfolio companies during the years 1986–1996. The follow-up of portfolio companies is extended until the end of May 2001 in order to track the outcomes of these investments. Furthermore, the syndication relationships are tracked from an extended sample of the 160 largest private venture capital organisations to enhance the identification of the actual syndication relationships.
The venture capital investment data of this paper is obtained from Securities Data Corporation’s Venture Economics database. This extensive database contains information on over 150,000 private equity investments and it is widely recognised as a leading source of U.S. venture capital investment data. Database has been used also in previous venture capital research (e.g. Bygrave, 1987; Gompers, 1995; Lerner, 1994; Sorenson and Stuart, 2001).
Data on the personnel resources of the sample firms is gathered from the back issues of the publications called Pratt’s Guide to Venture Capital Sources. This widely used publication lists the managing partners, the key personnel, and a variety of other parameters of most of the U.S. venture capital firms each year. The most relevant records for our study are the managing partners of the firms. This data is reported consistently each year with names and positions. Using the Pratt’s Guide data, we tracked the total number of partners and their names in each venture capital firm each year. To further ensure the validity of the managing partner data, we collected the current resumes of 28% of the partners of our sample firms and backtracked their career years in each venture partnership. We then compared the Pratt’s Guide listings of partners to the sub-sample of resumes without observing significant inconsistencies between these two sources.
Operationalization of Variables
Performance: We analyse the performance of the VC firm using the number of portfolio company initial public offerings as the key measure. IPO track record is a quantifiable performance measure of a venture capitalist and offers two clear advantages. Firstly, the largest valuations and returns to venture capitalists are most often realized in IPOs (Bygrave and Timmons, 1992). Although only a fraction of venture capital investments reach the IPO, most of the total value to the investors is created in these exits (Bygrave and Timmons, 1992, Gompers and Lerner, 1999). Thus, we can consider the IPO as the preferred exit vehicle of the most venture capital firms. Secondly, an IPO is a publicized event in the market and conveys information that should have a direct reputational effect for the focal venture capital firm, as an IPO indicates that the firm has been able to make a very successful investment.
Portfolio size: We measure the size of portfolio of a venture capitalist as the number of the companies the VC is involved in as an investor. We calculate the size of the portfolio relative to the number of partners in each venture capital firm. Thus, we are able to count for the differences in the sizes of the firms and to make the measure comparable across firms. Furthermore, the number of portfolio companies per partner readily measures the capacity of an individual venture capitalist to manage the portfolio companies.
We identify the portfolio companies of a venture capitalist using the investment records of Venture Economics. These records hold information of the dates of the investment as well as whether a company has made an IPO and when. We view that a company enters the portfolio when the venture capitalist invests in it for the first time, and it exist from the portfolio either through an IPO or after last recorded investment. The number of partners each year is taken from our combination of the records from Pratt’s Guide to Venture Capital Sources and the resumes of the partner sub-sample. We classified personnel as partners if their position title included one of the terms ‘partner,’ ‘vice president’ or ‘managing director.’
Syndication: We measure the syndication activity of a venture capital firm using the frequency of syndication. We define the frequency as the ratio of syndicated investments of all investments in a given year. We identify the syndicated investments by tracking the investments in companies each year and then identifying those with multiple investors. We extended the original sample of 97 companies to include the 160 largest venture capital investors in order to achieve exhaustive coverage for the syndication relationship between the largest firms and the remaining. The investments by the 160 largest firms capture approximately 90% of all the portfolio companies in the Venture Economics database between 1986—1996, and thus forms a reliable measure for syndication.
Control variables: Resources of venture capitalists were measured as the number of partners in the firm and as the amount of capital under management. Age of VC firm was measured in years since the founding of the venture capital firm. Investment stage mix was calculated as the percentage of investments made in early-stage, expansion stage and later-stage companies for each firm each year. Industry mix of investments was measured as the percentage of investments to each industry segment. Location was measured using dummy variables to indicate whether the VC firm was located in California, Massachusetts or another state. Finally, we controlled for potential Time-dependency using dummy variables indicating years.
Statistical Methods
The number of IPOs per year constitutes a variable that is discrete, positive and consists of zeros and ones to a significant degree. Thus, an ordinary linear regression is not applicable in this context, as it relies on the normality of the dependent variable. A standard model with count data such as ours is Poisson regression (Greene, 2000). However, Poisson regression assumes that the mean of the sample and the variance are equal. This assumption is not met in our data, and we thus utilize a negative binomial model with random effects (Hausman et al., 1984).
As our sample consists of panel data, we are able to account for the unobserved heterogeneity of units in the sample. These individual effects can be modelled either random or fixed. A random-effects model should be chosen when inferences are made for all the population, whereas a fixed-effects model is appropriate when inferences are made conditional to the sample (Hsiao, 1986). Furthermore, with a fixed-effects model we cannot comfortably use time-independent variables, such as dummy variables indicating the location of the venture capital firm. This is a significant control in our analysis. Thus, we use a negative binomial random-effects model. However, we checked the robustness of the analysis also with a fixed-effects model finding no qualitative differences.
Table 1 reports the descriptive statistics for the variables used in the study. The data set consist of 793 observations on the 97 venture capital firms. The statistics are from the whole sample and as such, they pool observations across firms and years. Table 2 presents the correlations between dependent and independent variables of the sample. The dummy variables for years are excluded, since there were no significant correlations with other variables.
Inverted U-shaped Relationship
Table 3 presents the results of the negative binomial random-effects regression analysis on the number of portfolio companies that ultimately reached the initial public offering. The non-standardised regression coefficients are presented with the corresponding standard deviations. Although not presented in the table, all regression analyses include year dummy variables to control for potential differences between years.
Model 1 in Table 3 shows the base model including the control variables. The sized related controls—the amount of capital under management, number of managing partners, and the number of portfolio companies—have an expected impact on the number IPOs. The market share of the investments has a positive effect on the dependent variable. The age of the venture capital organisation is negatively related to the number of IPOs, indicating that the younger the firm, the faster its portfolio companies tend to do an IPO. This finding supports Gompers’ (1996) grandstanding hypothesis. The location of a venture capital firm affects the number of IPOs it is involved in. Firms located in California or Massachusetts have more IPOs than those outside these regions. Firms that invest in later stage companies appear to be involved in more IPOs that those firms that concentrate on early stage investments.
In hypothesis 1, we suggested that there exist a curvilinear relationship between the number of portfolio companies per VC partner and the performance of the VC partnership. We find support for the hypothesis in models 2 and 3 in Table 3. The coefficient of the linear term is positive, whereas the coefficient of the quadratic term is negative. These results indicate that there exist negative marginal returns on the number of portfolio companies per partner.
To calculate the optimal size of the portfolio, we took a partial derivative with respect to the number of companies per partner. The data suggest that venture capitalists reach their respective point of diminishing total returns, and thus the optimum, in slightly over 12 portfolio companies per venture capitalist, given the mean level of syndication. We would like to point out that this number is contingent upon the sample and period under investigation; thus, the result is more suggestive than prescriptive. Furthermore, the figure is per partner, which makes it larger than is the actual number of investments per investment manager (including also others than partners). The important message, however, is that at some point there are diminishing and subsequently negative returns to adding more portfolio companies per venture capitalist.
Moderating Effect
Hypothesis 2 suggested that the share of syndicated deals of all deals, or the frequency of syndication, be positively related to a VC firm’s capability to manage portfolio companies. Model 4 in Table 3 adds the frequency of syndication as an additional variable. Syndication frequency has a positive and significant effect on the firm’s performance. However, as we enter the interaction term into the analysis in the model 5, the significance of the linear term vanishes. The effect of the interaction between the frequency of syndication and the number of companies per partner squared on performance is positive and significant.
The effect of the interaction term on the optimal number of portfolio companies per partner is two-fold. First, as syndication frequency increases, the optimum moves to right on the axis of the number of companies per managing partner, thus increasing the optimal number of companies per partner. Second, the number of IPOs from investments in the optimum increases as syndication frequency increases. Thus, an increase in the frequency of syndication increases both the number of portfolio companies a partner is able manage optimally and the number IPOs these companies ultimately produce. We thus find support for Hypothesis 2.
Figure 1 illustrates the moderation effect. The picture plots the equation in model 5 Table 3 on three different levels of syndication frequency; the sample mean and within one standard deviation from the mean. The optimal number of companies per partner moves right as the frequency of syndication increases.
In this paper, we set to analyse the optimal number of portfolio companies per venture capital firm managing partner, and the moderating effect of syndication on the manageable number of portfolio companies. We utilized an extensive data set composed of the investments and ultimate IPOs of all the investments by the 97 largest U.S. venture capital partnerships in 1986–1996.
The results of this study show that there exist an inverted U-shaped relationship between the number of portfolio companies a venture capital partner manages and the performance of these portfolio companies. At first, as the number of companies per partner increases, performance also improves, although with diminishing marginal returns. When the number of portfolio companies exceeds the optimum, performance starts to deteriorate. Given the shape of the relationship, there exists an optimal number of portfolio companies for a venture capitalist to manage. This optimum is moderated by syndication activity. The more a venture capitalist syndicates its investments, the higher is the size of the portfolio it can manage optimally.
Implications for Research
Our findings have important theoretical contributions because they (a) empirically test and validate recent theoretical research on the optimal portfolio size for a venture capitalist, (b) expand the extant theory by introducing the moderating role of syndication on the optimal portfolio size, and (c) empirically test and validate the usefulness of this proposed expansion. Aside these three specific contributions, this paper contributes to the wider body of literature on strategies and success factors of venture capitalists.
Practical Implications
Our findings have also important practical implications. The results provide valuable guidance for venture capitalists on successful strategies and resource allocation. Venture capitalists should carefully consider how many companies they can manage and still add value. Too few companies imply that the venture capitalist’s value-adding potential is not fully utilized. However, exceeding the optimal number of companies results in deteriorated performance as the venture capitalist’s value-adding activities become too fragmented. Syndication can be used to overcome resource constraints.
CONTACT: Mikko Jääskeläinen, Helsinki University of Technology, Institute of Strategy and International Business, P.O. Box 9500, FIN-02015 HUT, Finland; (T) +358 50 341 5267; (F) +358 9 451 3095; mikko.jaaskelainen@hut.fi
Barney, J.B., L. Busenitz, J.O. Fiet, & D. Moesel. (1996) “New Venture Teams’ Assessment of Learning Assistance from Venture Capital Firms.” Journal of Business Venturing 11(4): 257–272.
Brander, J., R. Amit, & W. Antweiler. (1999) “Venture Capital Syndication: Improved Venture Selection versus the Value-Added Hypothesis.” University of British Columbia.
Bygrave, W.D., & J.A. Timmons. (1992) Venture capital at crossroads. Boston: Harvard Business School Press.
Bygrave, W.D. (1987) “Syndicated Investments of Venture Capital Firms: A Networking Perspective.” Journal of Business Venturing 2:139–154.
Gompers, P.A. (1995) “Optimal Investment, Monitoring, and the Staging of Venture Capital.” Journal of Finance 50(5):1461–1489.
Gompers, P.A., & J. Lerner. (1999) The Venture Capital Cycle. Cambridge, MA: MIT Press.
Gorman M, & W.A. Sahlman. (1989) “What Do Venture Capitalists Do?” Journal of Business Venturing 4: 231–248.
Greene, W.H. (2000) Econometric analysis. 4th edition. Upper Saddle River, NJ: Prentice Hall.
Hausman, J., B.H. Hall, & Z. Griliches (1984). “Econometric Models for Count Data with an Application to the Patents-R&D Relationship.” Econometrica 52(4):909–938.
Hellman, T. & M. Puri. (2002) “Venture Capital and the Professionalization of Start-up Firms: Empirical Evidence.” Journal of Finance 57(1):169–197.
Hsiao, C. (1986) Analysis of panel data. Cambridge: Cambridge University Press.
Jones, G.R., & C.W.L. Hill. (1988) “Transaction cost analysis of strategy-structure choice.” Strategic Management Journal 9: 159–172.
Kanniainen V, & C. Keuschnigg. (2000) “The Optimal Portfolio of Start-Up Firms in Venture Capital Finance.” CESifo Working Paper Series No. 381.
Lerner, J. (1994) “The Syndication of Venture Capital Investments.” Financial Management 23(3):16–27.
MacMillan I.C., D.M. Kulow, & R, Khoylian. (1988) “Venture Capitalists’ Involvement in Their Investments: Extent and Performance.” Journal of Business Venturing 4: 27–47.
Rosenstein J., A.V. Bruno, W.D. Bygrave, & N.T. Taylor. (1993) “The CEO, Venture Capitalists, and the Board.” Journal of Business Venturing 8: 99–113.
Rothaermel F.T., & D.L. Deeds. (2001) “More Good Things Are Not Necessarily Better: An Empirical Study of Strategic Alliances, Experience Effects, and New Product Development in High-Technology Start-Ups.” Forthcoming in: in: Creating Value: Winners in the New Business Environment, ed. M.A. Hitt, R. Amit, C. Lucier, & R.D. Nixon. Oxford: Blackwell Publishers.
Sahlman, W.A. (1990) “The structure of governance of Venture Capital Organizations.” Journal of Financial Economics 27:473–521.
Sapienza, H.J., A.C. Amason, & S. Manigart. (1994) “The Level and Nature of Venture Capitalist Involvement in Their Portfolio Companies: A Study of Three European Countries.” Managerial Finance 20(1): 3–17.
Sapienza, H.J., S. Manigart, & W. Vermeir. (1996) “Venture Capitalist Governance and Value Added in Four Countries.” Journal of Business Venturing 11:439–469.
Sapienza, H.J. (1992) ”When do venture capitalists add value?” Journal of Business Venturing 7:9–27.
Sorenson, O. & T.E. Stuart. (2001) “Syndication Networks and the Spatial Distribution of Venture Capital Investments.” American Journal of Sociology 106(6):1546–1588.
2002 Babson College. All Rights Reserved. Last Updated March 2003.