Syndication in International Venture Investing. June 26, 2006 rev. January 23, 2007

Transcription

1 Syndication in International Venture Investing Isin Guler* Anita M McGahan Assistant Professor Professor Boston University Boston U. and visiting London Business School 595 Commonwealth Avenue amcgahan@bu.edu Boston MA (617) guler@bu.edu (617) June 26, 2006 rev. January 23, 2007 Venture investors typically syndicate their interests in the financing of new ventures to mitigate risks. In this paper, we examine differences in the composition of investor syndicates across US and non- US ventures. After controlling for exogenous characteristics such as the industry affiliations of the ventures and endogenous deal characteristics such as the amount invested and the number of rounds, we find evidence that non-us ventures are not managed through syndication as if they are more risky than ventures in other parts of the world. Investors do not syndicate more broadly internationally than in the US. One explanation is that investors select higher quality ventures outside the US than in the US. The conclusions emphasize that the venture community has not developed fully in its understanding of international investing. * Corresponding author. Thanks to Mercedes Delgado, Joanne Oxley, Joseph LiPuma and seminar participants at the Atlanta Conference on Competitive Advantage, the NBER Productivity Lunch, the Rotman School of the University of Toronto, and the Royal Complutense Colegio at Harvard for comments. McGahan thanks the Everett Lord Fund at the Boston University School of Management for financial support. Copyright 2006 Isin Guler and Anita McGahan. All rights reserved.

2 2 A fundamental question facing investors, entrepreneurs and policy-makers is whether new ventures in regions outside the United States (US) are more or less risky, overall, than new ventures in the US. Entrepreneurs are concerned about the basic business risks of operating internationally. Policy makers seek insights about how structural interventions can enhance economic development internationally. Venture capital investors assess country conditions as they make decisions about whether to sponsor particular ventures because international investments pose unique opportunities and challenges that may vary by region. In this paper, we ask how venture capitalists manage risk in international ventures, and whether risk management practices differ by the location of the venture. In particular, we develop empirical evidence on the syndication practices employed in ventures from other regions of the world as compared with management of US ventures so as to examine whether investors manage US ventures for risk less intensively than other ventures through syndication. Benefits and risks of non-us investing More venture-capital investors (called VCs or investors throughout this paper) are located in the US than in any other country. From the perspectives of these investors, both advantages and disadvantages are tied to investments outside the US. The advantages reflect that many of the world s industries continue to globalize and offer opportunities to entrepreneurs who are willing to apply both new and established business models in remote geographic settings. Entrepreneurial opportunities in other regions of the world have expanded significantly with movements toward freer trade, outsourcing, off-shoring, and foreign-investment incentives. In addition, the relative attractiveness of international venturing over US venturing may have improved with increasing competition over domestic ventures. Several disadvantages of investing outside the US have also been suggested in prior research. Investors may be exposed to higher levels of risk due to variation across countries in institutional infrastructure, which may in turn increase political and market uncertainty (Black and Gilson, 1998; Guillén, 2002; Henisz & Delios, 2001; Jeng and Wells, 2000; Martin, Swaminathan, & Mitchell, 1998). Investors and entrepreneurs may suffer a liability of foreignness that arises from lack of knowledge

3 3 about a new context (Zaheer, 1995) and from the difficulties in monitoring investments due to geographical distance (Sorenson & Stuart, 2001). Thus, investing in non-us ventures carries unique benefits and risks. Previous studies have documented that, in specific instances, investors and entrepreneurs manage the risk of internationalism by choosing locations strategically. For example, ventures sometimes follow their peers into countries after the initial investment (Guillén, 2002; Henisz & Delios, 2001; Martin et al., 1998) or co-invest with partners who are in close proximity to the investment (Sorenson & Stuart, 2001). Once a VC and entrepreneur are committed, they also may design their interactions to manage various aspects of the risks that accrue to each party (Cumming, Schmidt, & Walz, 2005; Kaplan, Martel, & Stromberg, 2005; Lerner & Schoar, 2006). To date, relatively little empirical research compares how VCs and entrepreneurs design syndicates to mitigate the specific risks of investing across country borders. Overview of the approach The purpose of this paper is to assess whether risk-mitigating syndication is more prevalent for ventures from different regions of the world than for US ventures. The results indicate that, on the whole, non-us ventures are not managed as if they are inherently riskier. The controls in the analysis include the industry of the venture, the venture s age cohort, the venture s age at the inception of financing, the amount of disbursed financing, the number of rounds of financing, and the frequency of financing rounds. The analysis proceeds along the following lines. We begin by discussing the theoretical bases for expecting differences in the underlying risks of ventures and of the effectiveness of risk-mitigation practices in the regions of the world. Second, we develop a dataset on US and non-us ventures. For each deal, we identify the venture s characteristics, the total amount invested, the number of rounds of financing, the elapsed time between rounds, and the number and identities of syndicate investors. We then employ a model that identifies relationships between risk-management practices and the regional affiliations of ventures after controlling for observable, systematic characteristics that investors typically consider in making an investment. The analysis is designed to minimize the imposition of structure on the

4 4 relationships between the regional identities of the ventures and the prevalence and extent of syndication. The precise specifications are explained below. THEORY AND HYPOTHESES The risks in venture investing are extensive Venture investments take place under unusually high levels of uncertainty. In the United States, a small number of entrepreneurial ventures account for a large proportion of the returns to venture-capital investors (Scherer, Harhoff, & Kukies, 2000). Only 10 to 30% of investments result in an initial public offering of stock to the public (Fenn, Liang, & Prowse, 1997). Sixty-two percent of the returns to investments between 1969 and 1988 were associated with just 10% of funded ventures (Scherer et al., 2000). During the period, over 30% of venture investments resulted in losses (Sahlman, 1990). The literature suggests several sources of risk in these investments. First, uncertainty results from the inherent nature of venture opportunities. These are the risks that are of primary interest in this study. Companies that receive venture financing are at early stages of development and may pursue unprecedented business models, unusual technologies, and unfamiliar product niches. Uncertainty arises about whether proposed technologies will result in a successful product or service offering as well as about the existence and size of market demand for the product or service. Even when the risks are evident, they may be difficult to manage as both the investing VC and the entrepreneurial venture may have little recourse for reducing them (Gompers & Lerner, 2000; Guler, 2003; Kaplan & Stromberg, 2002). Thus, some risk is irreducibly identified with the investments themselves. In the analysis, we control for those facets of venture identity that are observable to investors when syndicates are formed: namely, the venture s industry sector, founding year, and age at first investment. Our main hypotheses deal with how the residual risks vary by region of the world. The second main source of risk for venture investors originates in an agency problem between the investing capitalist and the entrepreneurial venture. The entrepreneur inevitably possesses private information about capabilities and effort levels that cannot be signaled credibly to the investor, and that

5 5 may vary depending on the contractual restrictions placed on the venture (Amit, Glosten, & Muller, 1990; Gompers, 1995; Kaplan & Stromberg, 2004). These agency problems may be compounded by physical distance. US venture capital firms typically operate in localized fashion, since proximity between the investor and the venture facilitates monitoring and control (Gupta & Sapienza, 1992; Lerner, 1995; Norton & Tenenbaum, 1993; Sorenson & Stuart, 2001). Especially when an entrepreneurial venture is at an early stage, investors keep close track of the venture s operations, and provide guidance and expertise through frequent interaction and office visits (Gorman & Sahlman, 1989; Lerner, 1995; Sorenson & Stuart, 2001). Since distance makes it harder to perform these monitoring and guidance functions, it leaves investors more vulnerable to the agency problems described above. International ventures, especially those in countries which are farther away from the investor s home base, therefore expose investors to higher risk. The risk tolerances of investors also may vary. Some may elect to support ventures with higher levels of risk, or may manage their portfolios in ways that allow for greater risks. For example, some investors may pursue high-technology industries despite the availability of lower-risk opportunities in other sectors. Others may carry greater risks because they sponsor entrepreneurs with very low collateral as long as all available collateral is available in the case of default. Thus, differences in the characteristics of the investors due to structural and/or policy variation may arise. As a result of these deal-specific risks, we control in the analysis for several facets of each deal: namely, participation in the syndicate by US investors and non-us investors, each of which may exhibit idiosyncratic investment approaches, Please note that the controls for regional identity of the investors are only indicators of whether at least one syndicate member is from the US or a non-us region. These indicator variables represent contextual variation in agency risks due to proximity for monitoring and to investing tradition. These indicator variables do not provide any information on the right-hand side about the number of investors in the syndicate. In the main empirical analysis, we also control for alternative risk-mitigation mechanisms that investors may employ as alternatives to syndication: the total amount invested, the staging of the

6 6 investment into rounds, and the duration between rounds. Each of these control variables is explained in greater detail below in the section called data and methods. Risks arise from regional affiliations The variables of primary interest for testing our hypotheses are the regional affiliations of the ventures in our dataset. Recent research emphasizes that the semi-globalization of economic activities creates important regional effects (Ghemawat, 2003, McGahan and Victer, 2006). Aspects of a country s institutional infrastructure can influence the returns to entrepreneurial ventures (Guler & Guillen, 2005) for the following reasons: (1) The legal system: Contractual mechanisms that US investors use to limit potential agency problems in US investments may not be transferable or enforceable under different legal systems (Lerner & Schoar, 2006; Guler & Guillen, 2005). (2) Financial markets: VCs primarily exit their investments through IPOs or acquisitions, each of which is accessible in countries where developed capital markets exist (Black & Gilson, 1998; Guler & Guillen, 2005; Jeng & Wells, 2000). The ability of VCs to successfully exit their investments and achieve high returns is substantially reduced in the absence of well-developed capital markets. (3) Political environment: In countries where rules and regulations for business change frequently, investor returns are subject to higher risk of expropriation (Henisz, 2000; Henisz & Williamson, 1999). The unpredictability created by larger political hazards may also add to the riskiness of international VC investments (Guler & Guillen, 2005). (4) Agreements that create regional ties: A slate of regional trade agreements such as ASEAN, NAFTA, and the European Union link colocated countries economically. These agreements typically remove barriers to competition between member countries but sometimes isolate regions by enacting barriers that block importing and exporting across regional boundaries. Investors rely on syndication to manage risks. A principal benefit of syndication for investors is to reduce the exposure of each investor to the risk of failure by amortizing the consequences of failure across a larger group of investors. Thus, a common mechanism for managing risk is to recruit additional investors into a deal syndicate (Admati &

7 7 Pfleiderer, 1994; Bygrave, 1987; Lerner, 1994; Sorenson & Stuart, 2001). Syndication also may reduce the risk of deal failure directly by providing the entrepreneurial venture with access to specialized investor skills, talent, and insight. Additional investors often prefer ventures that are referred by other capitalists (Fried & Hisrich, 1994). The venture and its investors benefit because the presence of reputable investors in a syndicate signals the underlying quality of the venture (Stuart, Hoang, & Hybels, 1999). Furthermore, syndication may reduce the information asymmetry problem between entrepreneurs and investors by enhancing the mechanisms for providing investors with verified, credible data on venture performance (Admati & Pfleiderer, 1994; Lerner, 1994; Sorenson & Stuart, 2001). Thus, deal risk for each investor in a syndicate is reduced as the number of investors in the syndicate increases. We hypothesize that the potential for greater risk in investing in ventures from regions other than the US may lead to greater syndication in their financing than for US ventures: Hypothesis 1. After controlling for observable risk characteristics of ventures (i.e., industry affiliation, age cohort of the venture, and age of venture at first financing) and the deal characteristics (i.e., indicators of the regional affiliations of participants in the investor syndicate, the total amount invested, the number of rounds, and the frequency of rounds), the number of investors in the syndicates for ventures from regions other than the US is greater than for US ventures. A deterrent to syndication is the reduction in returns to each investor as returns are shared among participating investors (Brander, Amit, & Antweiler, 2002). When returns are forecasted as especially high, the lead investors in a venture may elect not to pursue syndicate partners because the risk-reducing benefits of syndication are insufficient to compensate for the dilution of prospective returns. Increasing membership in a syndicate also may complicate agency and contractual relationships between investors and thereby increase costs (see, for example, Admati & Pfeiderer, 1994). Thus, there are two major classes of reasons why syndicates may be small: first, syndication may not be necessary because venture risk is low relative to the return; and second, syndication may not be attractive because the risk-reducing benefits of syndication are insufficient to offset the dilution effect of sharing prospective returns. This last situation may arise because the underlying quality of the venture is high. In both instances, small syndicates are associated with an attractive ex ante risk-return profile for

8 8 the venture. Large syndicates arise when the benefits of sharing risks and of reducing them directly by involving more investors are strong enough to outweigh the costs to each investor of dilution and of additional coordination. Therefore an alternative to hypothesis 1 is that investors do not manage non-us venture ventures more intensively by building larger syndicates than for US ventures: Hypothesis 2 (alternative). After controlling for observable risk characteristics of ventures (i.e., industry affiliation, age cohort of venture, and age of venture at first financing) and the deal characteristics (i.e., indicators of the regional affiliations of participants in the investor syndicate, the total amount invested, the number of rounds, and the frequency of rounds), the number of investors in the syndicates for ventures from regions other than the US is not greater than for US ventures. The unit of analysis for testing these hypotheses is the venture, and the dependent variable is the number of members of an investment syndicate. Empirical support for hypothesis 1 indicates that investors diversify the risks of non-us venture investment through a larger syndicate than for US deals. Support for hypothesis 2 indicates that the syndicates of non-us ventures are smaller, all else equal, than those for US ventures. DATA The data for this study were drawn from the Venture Economics surveys, conducted since the late 1970 s and compiled by Thompson Financial. Venture Economics is the official database of the National Venture Capital Association, which cooperates with the European Venture Capital Association in conducting annual surveys of over 1000 US venture-capital investors and approximately 300 European investors. The database has been used extensively in research on venture capital (Barry et al., 1990; Gompers & Lerner, 2000; Megginson & Weiss, 1991; Sahlman, 1990; Shane & Stuart, 2001). Although this study analyzes many non-us ventures, the results must be interpreted as preliminary and suggestive because the quality of Venture Economics coverage on non-us ventures is unknown. When a US venture capitalist makes international investments, it is reasonable to assume that the accuracy of reporting is comparable for domestic and international deals. Nonetheless, to the best of

9 9 our knowledge, no direct verification of the quality of non-us reporting has occurred, and hence, we do not know if inaccuracies are present. The sample includes all venture-backed companies that were founded between 1989 and The screened dataset covers 13,415 ventures, of which 8,672 were located in the US and 4,743 were located outside the US. The investors that funded the 13,415 companies included 783 US and 1,316 non- US investors. 664 of the 783 US investors (i.e., 84.8%) and 925 of the 1,316 non-us investors (i.e., 70.3%) participated in more than one deal. Table 1 presents descriptive statistics for non-us ventures, US ventures, and the pooled sample. The average syndicate involved about 3 investors. The US ventures attract a larger average number of investors (5.42 in US ventures versus 2.77 in non-us ventures). This feature of the dataset is also difficult to interpret as an aggregate statistic because of the smaller sizes of the non-us deals. US ventures received a larger average amount of investment ($32 million on average for US ventures versus $14 million for non-us ventures), in a larger number of rounds (3.5 rounds for US ventures versus 2.1 rounds for non-us ventures). The elapsed time between rounds is longer for US investments (208 days for US ventures versus 143 days for non-us ventures). Table 2 compares the US and non-us ventures in detail on several dimensions: founding year, number of rounds received, industry affiliation, and region. On average, the non-us ventures in the sample were founded more recently than the US ventures. Consistent with this observation, the non-us ventures more frequently received only one round of financing than the US ventures. The last panel of Table 2 shows that non-us and US ventures were similarly distributed across industries, with somewhat more representation of US ventures in the computer software & services and Internet-specific sectors. The last panel in Table 2 shows the locations of entrepreneurial ventures by region. Europe is most highly represented among the non-us ventures at 18.04% of the pooled sample, followed by Asia at 12.99%. Nearly two-thirds of the observations in the dataset are associated with US ventures. Potential biases and a cautious interpretation of results

10 10 We have no way of knowing whether our dataset is skewed towards representing US investors. Yet we proceed, interpreting our results carefully, because any potential bias that arises from the omission of non-us investors likely makes the results that we report too weak rather than too strong. In other words, if more non-us investors were included in the dataset, then we would likely find stronger evidence that non-us ventures attract smaller syndicates than US investments. As additional data become available, we hope to analyze absolute differences in the use of risk-mitigation mechanisms in various countries. The assessment of the native risks of US and ventures from other regions of the world is constrained by the absence of reliable information about structural differences across countries in venture bankruptcies, valuations, and performance. Without these data, we evaluate the risk mitigation practices imposed by investors and accepted by ventures as a signal of the underlying risks of deals. Following precedents set by other researchers, we conjecture that a greater observed prevalence of risk-management practices is an important sign of ex-ante risk-return profile that the venture presents to the investors. Specifically, we focus on the relative risk-mitigating actions among investors that are exposed to both domestic and/or international ventures. In the conclusion, we highlight the limitations of this approach, and explore alternative mechanisms that may account for our results. MODEL, METHOD AND MEASURES The model used to test hypothesis 1 against the alternative of hypothesis 2 is analyzed using a negative binomial regression because the dependent variable is expressed as a count (i.e., the number of investors in the syndicate for the venture): SYNj = EXP(( α r) + χjcontrolj) (1) r where SYNj is the number of firms in the syndicate funding venture j, r α r are a series of dummy variables representing the various geographic regions of the world, and CONTROLj represents the vector of controls: (a) the industry-sector affiliation of the venture

11 11 (b) the founding year of the venture; (c) the age of the venture at first investment; (d) the total amount of financing received (e) the number of rounds (i.e., the prevalence of staging) (f) the frequency of rounds in number of days between rounds (g) a US VC investor dummy; (h) a non-us VC investor dummy; and The right-hand-side variables are represented by the following measures. Regional Dummy Variables. The principal independent variables used to test our hypotheses are sets of dummy variables that represent venture locations by region. The represented regions are the United States, Australia-New Zealand, Asia, Europe, North America (excluding US), South America & Caribbean and Africa. The United States is the omitted category except in the supplementary analyses of only non-us ventures, in which Africa is the omitted category.) (a) Industry-sector affiliation of the ventures. Entrepreneurial ventures are exposed to observable risks that are tied to the industry sectors in which they participate (Gompers & Lerner, 2000). As a result, we include controls for the industry-sector affiliations of the entrepreneurial ventures by including industry-sector dummy variables. The nine variables represent whether the venture is affiliated with biotechnology, communications & media, computer hardware, computer software and services, consumer products, industrial & energy, internet-specific, medical & health, and semiconductors & other electronics. The omitted category is other products, which includes agriculture, business services, construction, financial services, manufacturing, transportation, and utilities. (b) Founding year. A venture s founding year may influence investment practices. For instance, older ventures may have received more investment rounds, or more investors simply because more time has passed since founding. Differences in the characteristics of entrepreneurial ventures may also vary across founding years because of variation in laws, customs and governance requirements. Underlying technological progress may cause an interaction between the year of founding and the industry affiliation of the ventures.

12 12 (c) Age at first investment. The age of the venture may affect its risk profile. Older ventures may have more information available for VCs, which likely reduces agency risks in the relationships between capitalists and ventures (Gompers, 1995). (d) Total amount of financing received. This variable is unambiguously among the most significant indicators of the aggregate risk-return relationship associate with an investment. As a result, we include it as a control to provide for the fact that ventures which receive greater absolute levels of financing also may attract larger numbers of investors into a syndicate. (e) Number of rounds.. An option for mitigating risk is to stage the disbursement of funds from investors into rounds. At each new round, investors reevaluate the venture s prospects, update assessments, and calibrate the offer of funding to reflect better information about the underlying opportunity. Staging allows the investor to monitor the resolution of critical technological and demand uncertainties and to implement new procedures for managing a venture. At each round, the investor typically retains the option of terminating the investment based on the venture s environment and performance (Gompers, 1995). Entrepreneurs may find that they can reveal their own commitments and the venture s capabilities through successive stages, and thereby reduce agency problems. Thus, the benefit of staging an investment into rounds is increased control over several major sources of uncertainty: the venture s specific risks and the risks tied to the partnership between the venture and the investors. The value of the investment also may increase with the resolution of agency problems. The costs of staging an investment are also substantial, however. VC investors and the entrepreneurial venture alike must expend resources in the monitoring process. For the venture, the effort required to sustain investor enthusiasm may even detract from the effectiveness of the enterprise in its product and service markets. The costs of monitoring are likely to increase with geographic distance and with the complexity of the local environment. Thus, the benefits and costs of staging an investment into rounds are both likely to vary internationally.

13 13 Thus, the net impact of staging into rounds on risk mitigation in international investing is ambiguous because of both these benefits and countervailing costs. We control nonetheless for the staging of financing into rounds as an alternative mechanism to syndication for risk mitigation. (f) Duration between rounds. Another mechanism that investors use to manage risk is to hold each round of financing to a short average duration of days. While venture capitalists typically continually keep abreast of an entrepreneurial venture s progress through the regular course of operations, a unique opportunity arises at each round of investment to formally assess the developments in the venture s operations. By increasing the frequency of rounds and by simultaneously reducing the duration between rounds, the venture capitalists quickly obtain information about the entrepreneurial venture s business prospects and underlying operations. Gompers (1995) shows that ventures that are subject to higher uncertainty tend to receive their funding through more frequent rounds with smaller amounts of financing at each round. Increasing the frequency of rounds may also be costly, however. Both investors and entrepreneurs expend resources for each financing event, and the costs of monitoring increase with larger distance and differences in investing environment. When little time elapses between rounds, these costs may be increased as the investor and entrepreneur seek to develop additional information above that available at the prior round. For international ventures, these costs may be especially high because of greater volatility in the environment and greater physical distances between the venture and its investors. Thus, we also control for the frequency of rounds as an alternative but costly mechanism to syndication for managing risk. The inclusion of this control leads to a significant loss of observations because this variable can only be assessed for deals that attract more than one round of financing. In a secondary specification, we exclude this variable to recover the lost observations and affirm the robustness of the results.

14 14 (g) A US VC investor dummy. We control for the involvement of US VC investors to allow for the possibility that US investors may follow protocols, processes and approaches to investing that are idiosyncratic to the US. (h) A non-us VC investor dummy. The involvement of international investors in a venture those outside the US may coincide with different structural features in the syndicate. The inclusion of this control variable allows for this possibility. RESULTS Table 3 shows our principal results. The rows represent the independent variables in the regressions and the columns represent pairs of models on Non-US ventures, US ventures, and both Non- US and US ventures in a pooled sample. Each pair of columns shows results first on the full model as specified above, and then on a reduced model that excludes the dummy variables on participation by US and non-us investors. The estimated coefficients in columns (5) and (6) on the regional dummy variables are negative and significant (with the exception of the coefficient in (6) for North America, excluding the US). Thus, hypothesis 1 is rejected in favor of hypothesis 2: syndicate size is *not* greater for ventures located outside the US than for ventures located in the US. This finding is consistent with the idea that non-us ventures are not managed for higher risk than US ventures. The size of the estimated coefficients on the regional variables in columns (1) and (2) on non-us ventures indicate that the largest syndicates occur in North America, excluding the US (i.e., Canada, Mexico). Syndicate size in Europe and Asia is roughly equivalent but smaller than in North America (excluding the US) on average. Note that the characteristics of venture deals in Australia-New Zealand and South America & the Caribbean are substantially similar, which creates co-linearity in the variables and prevents their assessment in a columns (1) and (2). The results on the control variables in the models reported in Table 3 yield additional insights. First, the significance of the coefficients on each of the industry-sector variables is consistent with precedent. More VC investors are involved in syndicates on ventures in biotechnology, communications

15 15 & media, computer hardware, computer software & services, industrial/energy, Internet, Medical/health and Semiconductors & Other Electronics than in the other products category. Syndication is less prevalent with a higher founding year (i.e., the newer the venture) and a younger firm at first investment, perhaps because of limits on the time required to attract investors to the venture. These results are significant and consistent for both non-us ventures and US ventures with just a few minor exceptions such as on the insignificance of age at first investment for non-us ventures. The results on the controls for the total amount of financing are positive and significant in all six models represented in Table 3, which indicates that, as expected, larger amounts of financing are associated with greater overall syndicate size. The coefficients on the number of rounds of financing and the number of days between rounds are also both significantly and positively associated with syndication (although the sizes of the coefficients on the number of days between rounds are small). Note that inclusion of the number of days variable requires that we restrict our sample to include only deals with more than one round of financing. As a result of this restriction, we lose significant number of observations. In Table 4, we recover these observations but exclude the number of days variable. The results are discussed further below. The final sets of control variables in Table 3 reflect the participation by US and Non-US investors in the syndicate. Recall that the US investor dummy is an indicator variable coded as 1 for those ventures that attract at least one US investor to the syndicate, and the Non-US investor dummy is coded as 1 for those that attract at least one non-us investor. The purpose of this control is to indicate the effect on syndication of participation by investors who may have processes, approaches, and other idiosyncrasies that affect syndicate size. As expected, the results suggest that participation by US investors is associated with larger syndicate size, perhaps because of the large size of the US investor community and/or the customary practice of syndication among US investors. Notably, participation by non-us investors is not significant. Note that the main results are not sensitive to the exclusion of these dummy variables (as reported in columns (1), (3) and (5)).

16 16 Table 4 reports the results of sensitivity analyses in which the main results are replicated for a larger sample in which we recover information on deals that involved only one round of financing. These deals were excluded from the sample used in the principal analysis (reported in Table 3) in order to control for the number of days between rounds as an alternative risk-mitigation mechanism to syndication on the theory that investors can mitigate their risks by increasing the frequency of rounds instead of inviting more investors into the syndicate. Observations are lost to include this variable because the number of days between rounds can only be assessed for ventures that receive more than one round of financing. The models reported in Table 4 thus exclude the number of days variable but include many more observations. It is worthwhile to note that there are a number of reasons why ventures may receive more than one round of financing: in some cases, they may be so unusually successful after the first round that no additional financing is needed; in other cases, they may be immediate failures. Thus, we expect that ventures that receive just one round of financing to have higher variation in overall riskiness than the ventures in the main sample represented in Table 3. The results in Table 4 affirm that hypothesis 2 over hypothesis 1: Non-US ventures are *not* tied to greater levels of syndication. This result is evident in the negative signs on all estimated coefficients on the models in columns (5) and (6) on the regional dummy variables. As in Table 3, many of these signs are significantly negative. The main difference on the regional variables is in the significance of the signs in the model reported in columns (1) and (2) on non-us ventures. While the coefficients on the regional dummies in these models here are negative (thus rejecting hypothesis 1 in favor of hypothesis 2), the significance levels on the regional dummies are not as great as in Table 3. This outcome is consistent with the idea that variation in syndication size for international ventures located in varying regions is greater for ventures that receive multiple rounds of financing. For international ventures that attract just one round of financing, syndicate size does not differ by region as much as for ventures that attract more than one round of financing. Thus, regional variation in syndicate size may be related to the decision among investors to offer secondary and subsequent rounds of financing.

17 17 The results on the control variables in Table 4 are also consistent with those in Table 3, although, in several instances, the levels of significance are somewhat lower than in the main results. This outcome is also consistent with higher variation in the riskiness of ventures that receive just one round of financing regardless of geography. One interesting difference in the signs on the controls is the significance of the non-us investor dummy variable in Table 4. We included this control in the analysis on the theory that investors may impose idiosyncratic processes on ventures that reflect local differences in risk mitigation, but found that this variable did not have significance in the main results presented in Table 3. The positive and weakly significant coefficients on this variable in Table 4 suggest that international investors may seek to manage ventures that receive just one round of financing differently than those that ultimately receive multiple rounds. Thus, the estimated coefficients on the control variables in Tables 3 and 4 are largely consistent with generally lower levels of significance on the controls reported in Table 4 (except for the non-us investor dummy ). Close inspection of the coefficients on the controls suggests interesting avenues for future research particularly on the ways in which investors adjust their idiosyncratic processes for particularly risky international ventures. The main results on regional variation are affirmed, though: The international address of a venture is not associated with higher levels of syndication relative to US ventures regardless of region after controlling for such characteristics as the total amount of financing received, the number of rounds, the age of the venture and the industry sector of the venture. DISCUSSION AND CONCLUSION This study is motivated by questions about whether whether risk management practices differ for ventures in other regions of the world than US ventures. The findings from our study affirm the alternative hypotheses 2, which stipulates no greater evidence of risk-mitigation practices for ventures from outside the US than from the US. Three major explanations arise: first, the underlying risks of non- US ventures may not be greater than for US ventures; second, syndication may not be employed to manage the non-us risks, if they exist; and third, the data may not accurately represent the issues at hand.

18 18 The last of these three possibilities on the limitations of data may indeed be significant in this analysis. US investors may sponsor ventures of only the highest quality, and therefore the ventures included in this analysis may not be representative of the broad range of entrepreneurship outside the US. If non-us ventures of only the highest quality are employed in the analysis, then the levels of syndication normally reserved for high-risk ventures may not be employed in the represented sample. The second of the possibilities that US investors may not be employed to manage the non-us risks is suggested by various anomalies in the signs of coefficients on the control variables as discussed earlier. These results suggest that US investors tend to employ syndication differentially across sectors, founding years, and ages at first investment regardless of the venture s regional affiliation. Yet syndication is not employed more intensively in non-us ventures, which by conventional logic on the difficulties of monitoring due to distance, etc., would seem to warrant more intensive management. This raises questions for subsequent study: Are US investors less able to employ syndication to manage the risks of international investing for some structural reason? Is syndication less effective for managing international risks?.the evidence on venture capitalists utilization of other risk mitigation mechanisms in international investments than syndication suggests that this may be the case (Cumming Schmidt and Walz 2005, Kaplan and Stromberg 2002). More research is needed to understand this issue comprehensively. The evidence reported here presents a puzzle because we find a positive and significant sign on the dummy variable that signifies the involvement of US investors in an international syndicate. This result indicates that US investors manage international ventures more aggressively for risk than non-us investors. Yet the overall size of the syndicates for international ventures is not greater. This apparent inconsistency raises the possibility that the risk in international deals may be related to the interaction of investor characteristics and the venture firms. US investors may be less adept at monitoring or managing international ventures than US ventures due to distance, for example, and thus may push for relatively greater levels of syndication than would otherwise occur. More research is needed to reconcile this

19 19 possibility with the finding that the overall level of syndication is significantly lower for non-us ventures. Finally, our results may be obtained because non-us ventures are not riskier than US ventures after controlling for their industry-sector affiliations, year of founding, age at first investment, total financing, number of rounds of financing, days between rounds. Perhaps international ventures that receive venture-capital attention do not pose higher inherent risk, when we control for regional differences, sectoral affiliations, and investment practices. Additional research which relies on information on structural variables such as rates of bankruptcy, time to IPO and value at IPO is needed to investigate this possibility. Contribution to the academic literature on VC behavior This study contributes to a small but growing body of literature on international venture- capital investment patterns (for a review, see Zalan, 2004). Prior work has examined factors that affect VCs willingness to invest abroad (Bottazzi et al., 2005), importance of country-level institutions in selection of investments (Black & Gilson, 1998; Guler & Guillen, 2005; Jeng & Wells, 2000), and the differences in contractual mechanisms for oversight (Kaplan et al., 2005; Lerner & Schoar, 2006). We contribute to this stream by developing evidence that suggests that VCs do not syndicate more widely for international ventures, but rather by observable venture and deal characteristics. Implications for practice The results reported here, while preliminary, have implications for venture-capital investors, entrepreneurs, and policy makers. A primary implication of our findings for the venture-capital investing community is that significant, untapped opportunity may exist in international investing, particularly by developing more sophisticated valuation and risk-management techniques that account for the idiosyncratic contexts in which non-us entrepreneurs operate. Ventures that seek capital from US and non-us investors should expect to be managed for risk using conventional tools such as staging into frequent rounds. The evidence here is consistent with the idea that geographic distance between ventures

20 20 and investors as well as idiosyncratic conditions in the country and regional environment may influence the costs of imposing these risk-mitigation practices on the non-us ventures. The findings here also have implications for policy-makers for policy-makers that seek to cultivate a local venture-capital investing industry. The evidence here is consistent with conservatism in the established venture community: conservatism in international deal selection and conservatism in the imposition of risk-mitigation practices that are adapted from the US environment. Policy-makers may induce cross-border venture capital investment by reducing country-specific risk factors, through measures such as supporting the enforcement of contracts and facilitating a more stable investment environment (see Black & Gilson, 1998; Guler & Guillen, 2005; Jeng & Wells, 2000). More importantly, they may trip off a virtuous cycle of growth by encouraging the establishment of local VCs with a deeper knowledge of local conditions and of localized opportunities.

21 21 REFERENCES Admati, A. R. & Pfleiderer, P Robust financial contracting and the role of venture capitalists. Journal of Finance, 43: Amit, R., Glosten, L., & Muller, E Entrepreneurial ability, venture investments and risk sharing. Management Science, 36(10): Barry, C. B., Muscarella, J. W., Peavy, J. W. I., & Vetsuypens, M. R The Role of Venture Capital in the Creation of Public Companies: Evidence from the Going Public Process. Journal of Financial Economics, 27: Black, B. S. & Gilson, R. J Venture capital and the structure of capital markets: Banks versus stock markets. Journal of Financial Economics, 47: Bottazzi, L., Da Rin, M., & Hellmann, T What Role of Legal Systems in Financial Intermediation? Theory and Evidence. Working paper. Brander, J. A., Amit, R., & Antweiler, W Venture-Capital Syndication: Improved Venture Selection vs. the Value-Added Hypothesis. Journal of Economics & Management Strategy, 11(3): Bygrave, W. D Syndicated Investments by Venture Capital Firms: A Networking Perspective. Journal of Business Venturing, 2: Cumming, D. J., Schmidt, D., & Walz, U Legality and Venture Capital Governance Around the World. Working paper. Fenn, G. W., Liang, N., & Prowse, S The Private Equity Market: An Overview. Financial Markets and Instruments, 6(4): Fried, V. H. & Hisrich, R. D Toward a model of venture capital investment decision making. Financial Management, 23(3): Ghemawat, P Semiglobalization and International Business Strategy. Journal of International Business Studies, 34(2):

22 22 Gompers, P Optimal Investment, Monitoring, and the Staging of Venture Capital. Journal of Finance, 50(5): Gompers, P. & Lerner, J The Venture Capital Cycle. Cambridge, MA: MIT Press. Gorman, M. & Sahlman, W What do venture capitalists do? Journal of Business Venturing, 4: 231. Guillén, M. F Structural Inertia, Imitation, and Foreign Expansion: South Korean Firms and Business Groups in China, Academy of Management Journal, 45(3): Guler, I. & Guillen, M. F Institutions, Networks, and Organizational Growth. Working paper. Gupta, A. K. & Sapienza, H. J Determinants of Venture Capital Firms' Preferences Regarding the Industry Diversity and Geographic Scope of Their Investments. Journal of Business Venturing, 7(5): Henisz, W. J. & Williamson, O. E Comparative Economic Organization - within and between Countries. Business and Politics, 1: Henisz, W. J Institutional Environment for Economic Growth. Economics and Politics, 12: Henisz, W. J. & Delios, A Uncertainty, Imitation, and Plant Location: Japanese Multinational Corporations, Administrative Science Quarterly, 46: Jeng, L. A. & Wells, P. C The Determinants of Venture Capital Funding: Evidence Across Countries. Journal of Corporate Finance, 6: Kaplan, S. N., Sensoy, B., & Stromberg, P How well do venture capital databases reflect actual investments? Working paper. Kaplan, S. N. & Stromberg, P Financial Contracting Theory Meets the Real World: An Empirical Analysis of Venture Capital Contracts. Review of Economic Studies: Kaplan, S. N. & Stromberg, P Characteristics, Contracts and Actions: Evidence from Venture Capitalist Analyses. Journal of Finance, 54(5): Kaplan, S. N., Martel, F., & Stromberg, P How Do Legal Differences and Learning Affect Financial Contracts? Working Paper. Lerner, J The Syndication of Venture Capital Investments. Financial Management, 23(3):

23 23 Lerner, J Venture capitalists and the oversight of private firms. Journal of Finance, 50(1): Lerner, J. & Schoar, A Does Legal Reinforcement Affect Financial Transactions? The Contractual Channel in Private Equity. Quarterly Journal of Economics, forthcoming. Martin, X., Swaminathan, A., & Mitchell, W Organizational Evolution in the Interorganizational Environment: Incentives and Constraints on International Expansion Strategy. Administrative Science Quarterly, 43: McGahan, A.M. and R. Victer The Effects of Industry and Headquarters Country on Firm Profitability. Boston U. Working Paper. Megginson, W. L. & Weiss, K. A Venture Capitalist Certification in Initial Public Offerings. Journal of Finance, 46(3): Norton, E. & Tenenbaum, B. H The effects of venture capitalists' characteristics on the structure of the venture capital deal. Journal of Small Business Management, 31(4): Sahlman, W. A The Structure and Governance of Venture Capital Organizations. Journal of Financial Economics, 27: Scherer, F. M., Harhoff, D., & Kukies, J Uncertainty and the size of distribution of rewards from innovation. Journal of Evolutionary Economics, 10: Shane, S. & Stuart, T Organizational Endowments and Performance of University Start-Ups. Management Science, 48(1): Sorenson, O. & Stuart, T. E Syndication networks and the spatial distribution of venture capital investments. American Journal of Sociology, 106(6): Stuart, T. E., Hoang, H., & Hybels, R. C Interorganizational Endorsements and the Performance of Entrepreneurial Ventures. Administrative Science Quarterly, 44: Zaheer, S Overcoming the liability of foreignness. Academy of Management Journal, 38: Zalan, T The Secret Multinationals of the New Millennium: Internationalization of Private Equity Firms, JIBS Literature Review.