Adverse selection and capital structure: evidence from venture capital.
This article considers the screening of venture capitalist (VC) investments. In particular, this article investigates the issue of whether different financial instruments (including debt, common equity, preferred equity, and convertible securities) attract different types of entrepreneurial firms in terms of the adverse selection risks associated with financing low-quality firms in relation to the financial security used. A central contribution of this article is the assessment of whether syndication mitigates such adverse selection risks.
It has been well established that firm-specific characteristics affect the choice of financial security (e.g., Jensen & Meckling, 1976; Noe & Rebello, 1996; Rajan & Zingales, 1995). It has also been well established that offers of different forms of finance attract different types of entrepreneurial firms. Seminal research on selection effects and capital structure focused on debt and common equity (DeMeza & Webb, 1987, 1992; Stiglitz & Weiss, 1981). Consistent with the seminal work of Akerlof (1970) (and the related work of Stiglitz (see, e.g., Stiglitz & Weiss, 1981) and Spence (1973) for the joint award for the 2001 Nobel Prize in Economics), the theoretical literature shows that equity attracts firms with low expected returns (consistent with the language adopted in the literature, we hereafter refer to firms with low expected mean returns as lemons), whereas debt finance attracts firms with high variability in returns (hereafter referred to as nuts). Related research on adverse selection and capital structure has considered convertible securities; e.g., Brennan and Kraus (1987) show that firms with low variability in expected returns are attracted to convertible securities. The intuition linking these different securities to different types of adverse selection risks is explained in the next section of this article.
The VC market offers an interesting forum for empirically examining the relation between adverse selection and forms of finance. Risk management is particularly important for VC funds. VCs have enhanced screening and monitoring abilities, and are active value-added investors. (1) Because they finance early stage firms for which there is a comparative dearth of information, VCs take significant steps to try to avoid financing low-quality firms. Research on VC is consistent with the proposition that VCs are among the most sophisticated financial intermediaries with abilities to mitigate agency problems. (2) Therefore, in considering the VC market, we study not only pronounced problems of adverse selection, but also specialized investors that take active steps to mitigate such problems.
This article makes three main contributions. First, we introduce a new data set that indicates, consistent with VCs in Europe and in all countries (where data are available) other than the United States, VCs in Canada use a variety of forms of finance. Second, we introduce evidence that certain types of investors have a preference for financing certain types of entrepreneurial firms and syndicate with different degrees of frequency. Third, and perhaps most important, we empirically assess the relation between capital structure and adverse selection across a wide array of entrepreneurial firms with different characteristics for the complete class of forms of finance (debt, convertible debt, preferred equity, convertible preferred equity, common equity, and combinations thereof). We further provide new and unique tests of the effect of syndication on adverse selection costs pertaining to capital structure.
With regards to the link between adverse selection and capital structure, the most notable results in this article are as follows. We show that firms in the startup stage (with little or no track record) tend to select away from common equity and tend to select toward offers of debt finance. This indicates that the risk of financing startup stage firms is better characterized as a risk of financing a nut, and not a lemon. Similarly, we show that high-tech firms are more likely to select toward debt and are less likely to select toward convertible securities that provide the investor with upside potential and downside protection. This is consistent with the widely accepted view that high-tech firms are riskier than their nontech counterparts. Importantly, we further show that syndication significantly mitigates problems of adverse selection, consistent with research on why VCs syndicate their investments (Lerner, 1994; Lockett & Wright, 1999, 2001; Manigart et al., 2002b; Wright & Lockett, 2003). Our results on syndication complement this prior literature by providing a unique and new test of the role of syndication mitigating problems of adverse selection as they pertain to the use of different types of securities in venture finance.
This article is organized as follows. In the next section, we develop hypotheses pertaining to adverse selection in entrepreneurial finance. The Canadian Venture Capital Association (CVCA) data are then described. Empirical evidence is then presented, and the last section concludes.
Adverse Selection, Capital Structure, Firm Characteristics, and Syndication
In this section, we develop hypotheses relating problems of adverse selection among alternative forms of finance to entrepreneurial firms at different stages of development and in different industries. We also consider the effect of syndication on adverse selection. Our theoretical discussion in this section and empirical tests in the subsequent sections are focused on first-round VC investments only, and not on follow-on staged financing rounds.
Theoretical Principles on Adverse Selection and Capital Structure
To understand the link between adverse selection and capital structure, it is helpful to review the properties of the different types of securities. Common equity is analogous to a call option. Common equity holders have a claim to all of a firm's cash flows, but only after all other claimants have been satisfied. If the firm goes bankrupt, then equity holders are the last to be paid (debt holders have priority over preferred equity holders, who in turn have priority over common equity holders); however, common equity holders cannot lose more than their investment in the firm by virtue of the firm's limited liability. Common equity is therefore analogous to a call option on the firm, whereby exercising the option requires that the firm be liquidated and the face value of the debt (the option price) be paid off. The payoff function to common equity is bounded below by the cost of the common equity investment and by unlimited potential on the upside. The payoff to straight (nonconvertible) debt, by contrast, is limited to the present discounted value of all interest payments and to the principal on debt (similarly, the payoff to nonconvertible preferred equity is limited to the present discounted value of all prespecified preferred dividends). In the event of an entrepreneurial firm's failure to pay interest on debt, an investor can force a firm into bankruptcy. In the event of an entrepreneurial firm's failure to pay a prespecified preferred dividend, an investor cannot force a firm into bankruptcy, but all preferred dividends must be paid before any common dividends can be paid.
The well-known lemons principle (Akerlof, 1970), in the context of capital structure theory, is based on the idea that the form of finance offered by an investor attracts the worst possible type of firm for that form of finance. Theoretical research has well established the propositions that common equity attracts firms with low expected returns (lemons) (adverse selection costs of common equity) (DeMeza & Webb, 1987, 1992), straight debt and straight preferred equity attract firms with high variability in returns (nuts) (adverse selection costs of straight debt and straight preferred equity) (Stiglitz & Weiss, 1981), and convertible securities attract firms with low variability of returns (adverse selection costs of convertible securities) (Brennan & Kraus, 1987). Investors anticipate these problems and structure contracts accordingly (see, e.g., Brennan & Kraus, 1987). We review the intuition underlying these results in this subsection.
The intuition underlying the theoretical work of DeMeza and Webb (1987, 1992) linking common equity and lemons is as follows. A VC that finances an entrepreneurial firm with common equity receives significant upside potential in terms of obtaining an ownership interest in the entrepreneurial firm, but does not receive downside protection in terms of priority in bankruptcy. Entrepreneurial firms with lower expected mean returns have lower opportunity costs associated with giving up larger ownership interests to VCs through common equity contracts relative to entrepreneurial firms with high expected returns. Entrepreneurial firms with lower expected returns also have higher opportunity costs associated with giving up downside protection (in bankruptcy) to their VC investors. As such, VCs that offer common equity contracts to their entrepreneurial investee firms face a more pronounced adverse selection problem in terms of attracting entrepreneurial firms with lower expected average returns.
The intuition underlying the theoretical work of Stiglitz and Weiss (1981) linking nonconvertible debt (and similarly nonconvertible preferred equity) and nuts is as follows. A VC that uses debt to finance a firm does not receive any ownership interest and therefore does not derive the benefit of capturing the upside potential associated with an improvement in the quality of the firm (i.e., above and beyond an increase in the probability in being paid back the interest and principal on the debt contract). Entrepreneurial firms with high expected variability (standard deviation) of returns have higher opportunity costs associated with giving up larger ownership interests to VCs through equity contracts. The reason is as follows. If a risky venture happens to be successful, then the entrepreneur financed with debt captures 100% of the upside potential success. In the literature, this is sometimes referred to as a situation in which the entrepreneur thinks "Heads I win" in that the firm happened to do well, and the entrepreneur captured all of the benefits associated with winning the lottery. By contrast, if a risky venture happens to fail, then the entrepreneur financed with debt has not lost to the same degree as the investor that has not been repaid. In the literature, this is sometimes referred to as a situation in which the entrepreneur thinks "Tails you lose" in that the firm happened to fail, but the loss in capital provided to finance the firm is more heavily borne by the investor that provided such capital. In short, investors that offer debt contracts to their entrepreneurial investee firms face a more pronounced adverse selection problem in terms of attracting entrepreneurial firms with higher expected variability in returns.
Adverse Selection Risks in Venture Capital Finance
Adverse selection risks exist because of information asymmetry in that entrepreneurs have information that VCs do not have. In order to understand the extent to which entrepreneurs may have information that VCs do not have, it is necessary to identify and distinguish the entrepreneurial firm characteristics that are in fact known to VCs, such as the stage of development of the entrepreneur and the industry in which the entrepreneur operates. But not all entrepreneurial characteristics can be completely known by the VCs (such as the entrepreneur's intelligence, managerial abilities, planned work effort, scientific skills for developing patents, and technical processes for the firm, etc.). Hence, adverse selection risks will exist, and will exist to different degrees depending on the entrepreneurial firm characteristics that the VCs can identify.
Based on prior research, we may conjecture that, for entrepreneurial startup stage firms without a track record (or a significant track record) and for firms in high-tech industries, VCs are better able to screen the risks associated with financing a lemon relative to the risks associated with financing a nut. In other words, a "bad" firm is easier for a VC to screen than a "risky" one. The intuition is clear for four interrelated reasons. First, theoretical work (e.g., Hopenhayn & Vereshchagina, 2004; Landier & Thesmar, 2003) and empirical evidence (e.g., Moskowitz & Vissing-Jorgensen, 2002) consistently show that entrepreneurs bear substantial risk but do not earn a return that compensates for such risks. Most work in entrepreneurial finance is consistent with the view that entrepreneurs have significant risk tolerance and over optimism. This means that entrepreneurs downplay the risk of variance of potential outcomes associated with their startup firms. Second, entrepreneurs are not inclined to start a firm that they believe is a lemon due to the stigma of failure and the opportunity costs of not working elsewhere. Third, VCs can screen lemons through due diligence reviews based on the entrepreneur's track record in terms of leisure preference, (3) as well as managerial and technical competence that had been established before starting the particular firm seeking VC finance. (4) Fourth, while a VC can use past performance to mitigate the risk of financing a lemon, it would be much more difficult for a VC to ascertain whether an entrepreneur's latest idea is best characterized as nuts on the basis of the entrepreneur's past performance (i.e., it is easier to forecast the mean than the variance of expected value). Taken together, these four reasons suggest the following conjecture, which is tested in the subsequent sections of the article.
Hypothesis 1: The adverse selection risks associated with capital structure for financing risky nuts are more pronounced and difficult to mitigate through screening, relative to the adverse selection risks of financing low-quality lemons, for first-round VC financings of startup stage and high-tech entrepreneurial firms.
We further hypothesize that syndication reduces informational asymmetries and adverse selection costs through information sharing and improved screening. The intuition is based on the idea that due diligence is enhanced by bringing together complementary skills and stage and/or industry expertise from VC managers employed in different VC funds. The prediction that syndication mitigates adverse selection risks is consistent with analyses of syndication in different contexts (in the context of VC finance, see Dimov & De Clercq, 2006; Jaaskelainen, Maula, & Seppa, 2006; Lerner, 1994; Lockett & Wright, 1999, 2001; Manigart et al., 2002b; Wright & Lockett, 2003; for earlier work in a general context of syndication, see Leland & Pyle, 1977). We complement and build on this prior work by analyzing for the first time the effect of syndication on problems of adverse selection associated with different securities used in venture finance. With this literature that syndication facilitates due diligence and screening, we conjecture the following hypothesis:
Hypothesis 2: VC syndication in the first round of investment mitigates the adverse selection risks associated with capital structure decisions for financing startup stage and high-tech companies.
Further to H2, note that in Canada there are a variety of different types of investors, including private independent limited partnerships, corporate VCs, government VC funds, labor-sponsored venture capital corporations (LSVCCs), institutional investors, as well as foreign (U.S.) funds. In the empirical analyses in the next sections, we consider differences in the quality of the syndication among these different fund types. Our priority is to expect that U.S. VCs face greater hurdles in screening investments due to geographic distance. LSVCCs may also face greater difficulties in screening investments as a fund with statutory constraints that impose a time limit (from 1-2 years, depending on the jurisdiction in which they are incorporated) lot reinvestment of capital that has been contributed by investors (see, e.g., Cumming & Macintosh, 2003; Halpern, 1997; Osborne & Sandler, 1998 for the statutory details).
The data used to test these hypotheses are described in the next section. Then, empirical tests of the hypotheses developed herein are presented. We control for the different types of investors and make use of artificial regressions to explore problems of adverse selection for the complete class of forms of finance among different types of entrepreneurial firms. Limitations, alternative explanations, future research, and conclusion comprise the final sections of the article.
To investigate problems of adverse selection in capital structure decisions, it is both necessary and interesting to consider a country in which VC investors finance a wide variety of different types of entrepreneurial firms and to use a heterogeneous mix of forms of finance. Research on the U.S. VC industry is consistent with the proposition that VCs rarely use forms of finance other than convertible preferred equity (Gompers, 1998; Kaplan & Stromberg, 2003; Sahlman, 1990); therefore, the U.S. VC market does not offer a useful forum to address problems of adverse selection in capital structure. Nevertheless, there are reasons why VCs use forms of finance other than convertible preferred equity, and data indicate a variety of forms of finance are employed by VCs in all countries around the world other than the United States. (see, e.g., Cumming, 2005a, 2005b; Hege, Palomino, & Schwienbacher, 2004; Lerner & Schoar, 2005).
The data used in this article comprise 4,114 first-round Canadian VC financing transactions from 1991 (Q1)-2003 (Q3) (source: Macdonald and Associates, Ltd., for the CVCA; see http://www.canadavc.com). All VC investors in Canada finance a wide variety of entrepreneurial firms and use a wide variety of financial instruments (see Table 1, panels A-G). The tables show that the use of a variety of forms of finance by VCs in Canada is not due to the definition of "venture capital" in Canada (e.g., early stage only, or early and expansion stage (5)). Given the heterogeneity in the types of firms and the use of different forms of finance, the Canadian VC market offers an excellent forum in which to study risks of adverse selection in capital structure. In fact, Canada is the only country in the world with industry-wide venture finance data to empirically study adverse selection and capital structure. For the purpose of analyzing adverse selection in venture finance, we control for changes in market conditions over the period considered, as subsequently described. Our focus is on first-round investments in the startup and expansion stage firms only in order to focus on the screening of new pure VC investments (and not later stage private equity investments). The data are summarized in Table 1.
There exist seven types of VC investors in Canada (see also Cumming, 2005a; Cumming & Macintosh, 2003): private independent limited partnerships, corporate, government, institutional, labor sponsored, foreign (U.S.), and "other." Private independent funds (Table 1, panel A) are similar to U.S. VC limited partnerships, but Canadian private independent VC funds generally have fewer and less restrictive covenants placed on the investment managers (Cumming & Macintosh, 2003). Canadian corporate VCs (Table 1, panel B) are analogous to U.S. corporate VCs (Gompers & Lerner, 1999), but tend to finance a somewhat more heterogeneous group of entrepreneurial firms (see Table 2). Government VC funds in Canada (Table 1, panel C) are managed by independent professional VC managers and finance a wide variety of different entrepreneurial firms. Institutional investors (Table l, panel D) are larger institutions that make direct VC investments. LSVCCs (Table 1, panel E) are significantly different than the other Canadian VC funds. These differences arise through statutory restrictive covenants placed upon LSVCC managers (Cumming & Macintosh, 2003; Osborne & Sandler, 1998). The tests below control for the fact that the constraints in which LSVCCs operate may influence not only their choice of types of firms in which they invest, but also the types of securities they employ, irrespective of the characteristics of their investees. Foreign investors (Table 1, panel F) are U.S. limited partnership investors that invest in Canadian entrepreneurial firms. The category of other (including "unknown") investors (Table 1, panel G) encompasses all other types of investors that less regularly appear in the CVCA data.
Table 1 panels A-G clearly show that convertible preferred equity has not been the most commonly used form of finance by Canadian VCs. This result is not due to the type of VC fund, or the type of firms being financed, as documented in Table 1 panels A-G. It is also not due to the definition of VC, as the results in Canada hold for the funds most similar to U.S. limited partnerships (the transactions for Canadian private independent funds are presented in Table 1, panel A). The 4,114 transactions in Table 1 panels A-G are used in the following section to test the hypotheses developed in section 2. Because the evidence in Table 1 panels A-G drastically differs from U.S. evidence on venture finance on the use of convertibles (e.g., Gompers, 1998; Kaplan & Stromberg, 2003; Sahlman, 1990), we exhaustively present the data by type of VC and entrepreneurial firm. The differences between the Canadian and U.S. evidence are consistent with a tax bias in favor of convertibles in the United States (Gilson & Schizer, 2003), and this tax bias is absent in Canada (Sandler, 2001 ; see also Cumming, 2005a for further details on tax aspects and security choice in Canada and in the United States).
The CVCA data consider syndication as more than one VC fund from different VC organizations that simultaneously fund the entrepreneurial firm in the first round of the investment. Consistent with Gompers and Lerner (1999), we do not consider "coinvestment" (different VC funds that are part of the same VC organization) as a syndicated investment.
The subsequent empirical analysis focuses on the standard financial instruments: debt, convertible debt, preferred equity, convertible preferred equity, common equity, and combinations of debt and common equity. A broader contract space does exist in practice. For example, convertible debt is a hybrid bond that allows its bearer to exchange it for a given number of shares of common stock anytime up to and including the maturity date of the bond. A broader contract space, however, is not considered herein for a number of reasons. Articles that do consider a broader contract space (e.g., Aghion & Bolton, 1992) yield optimal contracts that resemble the standard financial instruments. The fact that the standard instruments in corporate finance are used most frequently suggests that considerable gains arise from standardization, and greater transaction costs arise from designing contracts to mimic the standard forms of finance (Berglof, 1994; Macdonald, 1992). Regardless, the data used to test the hypotheses developed herein have been recorded such that if a contract was designed to mimic one of the standard forms of finance, then the standard form was recorded. (6)
The econometric analyses in the next section consider the entire sample of 4,114 first-round investments without segmenting the sample by type of VC fund. Different VC funds syndicate more than 50% of their investments (see Table l, panels A-G, and Table 2). Separating the data by sectors would fail to account for VC syndicates across different investor types (Table 2), and would therefore significantly bias the results. Table 2 in fact shows a massive degree of integration across different types of VC funds in Canada insofar as different types of investors syndicate with one another. The use of the full data set enables the richest array of results by accounting for interactions among different VC funds in Canada, while recognizing that the Canadian VC market is not segmented by sectors.
The empirical evidence in this section is presented as follows. First, econometrics methods for the determinants of forms of finance used in prior work are discussed. Second, we consider the first step of a Durbin-Wu-Hausman (DWH) test for the purpose of setting up the test of adverse selection. Third, the test of adverse selection is provided with the second step of the DWH test. Finally, robustness checks are provided.
The Determinants of Forms of Venture Capital Finance. In the Canadian context (with the absence of the U.S. tax bias; Gilson & Schizer, 2003; Sandler, 2001), Cumming (2005a) considered the determinants of the selected form of finance by estimating multinomial logit regressions of the following form:
(*) Probability (security selected) = [alpha] + [[beta].sub.1] Startup + [[beta].sub.2] High-tech + [[beta].sub.3] Deal size + [[beta].sub.4] LSVCC + [[beta].sub.5] Capital gains tax + [[beta].sub.6] Toronoto Stock Exchange (TSX) + [[beta].sub.7] Trend + [epsilon]
The right-hand-side variables in equation * include a dummy variable for startup firms (a dummy for expansion-stage firms is suppressed for reasons of perfect collinearity; other later stage firms are excluded from the data herein), a dummy for high-tech firms, the capital required by the entrepreneurial firm (in real 2000 Canadian dollars and in logs), a dummy variable for LSVCCs, a variable for the capital gains tax changes, (7) a dummy variable for the Toronto TSX index, (8) and a trend term to account for changes over time. These variables are also used herein in the subsequent sections, particularly in regard to assessing the potential effect of potential endogeneity of the first three variables in equation *.
In the following subsections, we consider in more detail the nature of this adverse selection problem associated with different forms of finance and the presence of certain types of entrepreneurial firms in the market for VC finance. Adverse selection would exist with respect to hypothesis 1, where the variables startup and high-tech are endogenous in equation *. To consider this possibility, we proceed in two steps. In the first step, we consider instrumental variables that impact the types of firms that may exist in the market for VC. In the second step, we use the residuals from the first-step regressions to test the effect of endogeneity on the estimated coefficients in equation *. This artificial regression in the second step, a version of the DWH test, (9) is not a direct test for exogeneity, but does indicate the effect of endogeneity on the coefficient estimates in equation *, and therefore provides insight into the nature of adverse selection associated with different forms of finance and different types of entrepreneurial firms.
Do Different Types of Investors Select Different Types of Entrepreneurial Firms? VCs typically receive around 1,000 requests for financing each year (Sahlman, 1990). This may create noise in the matching of entrepreneurial firms to VC funds. But VC funds tend to have a preference for financing certain types of entrepreneurial firms according to their stage of development, required capital, and/or type of technology, among other things (Gompers & Lerner, 1999; Mayer et al., 2005). (10) Herein, we relate the presence of different types of firms in the VC market to geographic location of investees (11) and to the type of VC fund. We estimate the binary logit regression equations 1 and 2, and ordinary least squares regression equation 3 of the following form:
(1) Probability (Startup) = [Year fixed effects] + [[beta].sub.1] Corporate + [[beta].sub.2] Government + [[beta].sub.3] Institutional + [[beta].sub.4] LSVCC + [[beta].sub.5] LSVCC + [[bet].sub.6] Foreign investor + [[beta].sub.7] Other type of VC + [[beta].sub.8] British Columbia + [[beta].sub.8] Alberta + [[beta].sub.9] Saskatchewan + [[beta].sub.10] Manitoba + [[beta].sub.11] Ontario + [[beta].sub.12] Quebec + [[epsilon].Startup]
(2) Probability (High-tech) = [Year fixed effects] + [[beta].sub.1] Corporate + [[beta].sub.2] Government + [[beta].sub.3] Institutional + [[beta].sub.4] LSVCC + [[beta].sub.5] LSVCC + [[beta].sub.6] Foreign investor + [[beta].sub.7] Other type of VC + [[beta].sub.8] British Columbia + [[beta].sub.8] Alberta + [[beta].sub.9] Saskatchewan + [[beta].sub.10] Manitoba + [[beta].sub.11] Ontario + [[beta].sub.12] Quebec + [[epsilon].sub.High-up]
(3) Probability (Syndication) = [Year fixed effects] + [[beta].sub.1] Startup + [[beta].sub.2] High-tech + [[beta].sub.3] Log (Deal size) + [[beta].sub.4] Corporate + [[beta].sub.5] Government + [[beta].sub.6] Institutional + [[beta].sub.7] LSVCC + [[beta].sub.8] LSVCC + [[beta].sub.9] Foreign investor + [[beta].sub.10] Other type of VC + [[beta].sub.11] British Columbia + [[beta].sub.12] Alberta + [[beta].sub.13] Saskatchewan + [[beta].sub.14] Manitoba + [[beta].sub.15] Ontario + [[beta].sub.16] Quebec + [[epsilon].sub.Syndication]
In equations 1-3, a dummy variable for private independent limited partnerships is suppressed to avoid perfect collinearity problems; as such, the other coefficients are relative to the private independent funds. A dummy variable for the Maritime Provinces is suppressed to avoid perfect collinearity as well. Year fixed effects are used to account for the possibility of different types of firms (by stage of development, industry type, and capital requirements) present in the market for VC in different years due to market conditions. [[epsilon].sub.Startup], [[epsilon].sub.High-tech], and [[epsilon].sub.Syndication] are vectors of residuals for equations 1-3, respectively, and are used in the subsequent endogeneity tests. Diagnostic tests suggested in the main that the right-hand-side variables for equations 1-3 were orthogonal to the different forms of finance (see Table 3) and are therefore suitable instruments for the DWH tests of endogeneity subsequently described. (12) The main results are not sensitive to the choice of instruments (alternative regressions are on file with the author, as well as an earlier version of the article yielding similar results).
The estimates for equations 1-3 are presented in Table 4. Also, analogous to regression equation 3, Table 4 also presents regression equations 4-6 for syndications with LSVCCs only (regression equation 4 as specified in Table 4), foreign VCs only (regression equation 5 in Table 4), and for syndications for all other types of investors excluding foreign VCs and LSVCCs (regression equation 6 in Table 4).
There are two main results in Table 4. First, investments in certain types of entrepreneurial firms are related to overall investment activity in different geographic regions in Canada. In particular, VC-backed startups are less prevalent in Alberta and Saskatchewan, and high-tech VC-backed companies are more prevalent in British Columbia and less prevalent in Alberta, Saskatchewan, Manitoba, and Quebec. Second, there is support for the proposition that different types of VC funds often finance different types of entrepreneurial firms in Canada and display differences in the propensity to syndicate investments. Subsequently, we elaborate on this second result, as it is more generalizable to contexts in other countries where comparative work could be considered.
Table 4 indicates that corporate funds are 29.8% less likely to invest in startup stage firms and 30.0% less likely to invest in high-tech firms relative to private limited partnerships. Institutional investors are 34.3% less likely to invest in startup stage and 32.8% less likely to invest in high-tech firms relative to limited partnerships. LSVCCs are 19.2% less likely to invest in startup stage and 34.6% less likely to invest in high-tech firms relative to limited partnerships. By contrast, government funds are 5.7% more likely to invest in startup stage firms and 77% less likely to invest in high-tech firms relative to limited partnerships. Foreign (U.S.) VCs are 20.3% less likely to invest in startups, but are 12% more likely to invest in high-tech companies relative to limited partnerships.
Table 4 regression equation 3 indicates differences in the propensity of different types of VC funds to syndicate their investments. Relative to private independent limited partnerships, corporate VCs are 12.% more likely to syndicate, government VCs are 12% more likely to syndicate, institutional VCs are 12.8% less likely to syndicate, LSVCCs are 90% less likely to syndicate, and foreign (U.S.) investors are 12.2% more likely to syndicate. For all types of VC funds considered together (regression equation 3), we observe syndication 16.9% more frequently for startups and 16.5% more frequently for high-tech companies.
Syndication is also more likely the greater the total value of the deal. In regard to the economic significance of that effect, an increase in deal size from C$1,000,000 to C$2,000,000 increases the probability of syndication by 6.7%, whereas an increase in deal size from C$11,000,000 to C$12,000,000 increases the probability of syndication by 0.8. (13)
Table 4 regressions equations 4-6 consider the determinants of the propensity of LSVCCs to syndicate (regression equation 4), U.S. VCs to syndicate with Canadian VCs (regression equation 5), and all other types of VCs taken together (that is, all excluding LSVCCs and U.S. VCs; regression equation 6). LSVCCs are considered separately for reasons indicated earlier (see also Osborne & Sandler, 1998). U.S. VCs are considered separately because they are geographically distinct, if not in other ways as well.
Regression equations 4 and 5 in Table 4 do in fact indicate significant differences exhibited by LSVCCs and U.S. VCs. The propensity of LSVCCs to syndicate is invariant for startups and for high-tech firms, unlike other VCs (compare regression equation 4 with regression equations 3, 5, and 6). U.S. VCs, by contrast, are less 2.5% likely to syndicate with Canadian VCs for startup investments and 1.5% more likely to syndicate with Canadian VCs for high-tech investments.
In the next subsection, we use the vectors of residuals in regression equations 1-6 in Table 4 to construct artificial regressions and test for the effect of endogeneity associated with estimating equation * (the DWH test). That is, while the choice of form(s) of finance depends on the type of firm and the particular agency problems associated with the investment, are different types of firms more attracted to different forms of finance, and does syndication mitigate such adverse selection problems?
Adverse Selection, Forms of Finance, and Firm Characteristics: Empirical Evidence. In this subsection, we test for the effect of adverse selection associated with the choice of form(s) of finance by estimating a system of artificial regressions of the following form:
(7) Probability (Security selected) = [alpha] + [[beta].sub.1] Startup + [[beta].sub.2] High-tech + [[beta].sub.3] Log (Deal size) + [[beta].sub.4] LSVCC + [[beta].sub.5] Capital gains tax + [[beta].sub.6] TSX + [[beta].sub.7] Trend + [[beta].sub.8] [[epsilon].sub.Startup] + [[beta].sub.9] [[epsilon].sub.High-tech] + [[beta].sub.10] [[epsilon].sub.Syndication] + [epsilon]
Recall from the first subsection that different forms of finance may be more appropriate for different entrepreneurial firms (equation *); but the causal relation between type of firm and form of finance is not unidirectional. The key empirical insight is that where the residual vectors ([epsilon]) are not orthogonal in equation 7, different types of firms are attracted to different forms of finance. Statistically significant estimates indicate the residuals [[epsilon].sub.Startup], [[epsilon].sub.High-tech], and [[epsilon].sub.Syndication] are not orthogonal to the selected security; that is, they are endogenous, such that startups, high-tech, and syndicated investments are observed in the data because of the presence of the particular security that is part of the VC financing transaction. A positive coefficient is evidence of selection toward that security and vice versa for a negative coefficient.
Instead of providing the extended version of all of the multinomial logit regression coefficients of equation 7, we present only the marginal effects of such estimates. Only the marginal effects and statistical significance for [[beta].sub.8]-[[beta].sub.10] are reported in Table 5, as these coefficients are pertinent to the issue of adverse selection by virtue of the effect of endogeneity on the coefficient estimates of equation *. This has the benefit of conciseness, and shows economic as well as statistical significance.
Table 5 provides evidence relating selection effects to forms of finance for the different types of firms financed by Canadian VCs. Panel A of Table 5 considers all of the different types of VC funds that syndicate together. Panel B of Table 5 distinguishes syndication among LSVCCs and foreign VCs only from the rest of the types of VCs.
Three primary results are apparent from panels A and B of Table 5. The first primary result is that startup stage firms manifest adverse selection problems vis-a-vis debt finance. That is, startup stage firms are more likely to be attracted by offers of debt finance (Table 5, panel A). This result is statistically significant at the 1% level of significance, and the estimated economic significance is such that adverse selection problems are 4.3% more pronounced for startup stage firms (relative to expansion stage) in debt-financed transactions. Similarly, the data indicate that adverse selection problems for startup stage firms are 5.3% less pronounced for common equity finance and 1.7% more pronounced for mixes of debt and common equity finance (panel A). This evidence is consistent with the view that startup stage entrepreneurs do not like to "give up" common equity to their investors and would prefer to be financed by debt. Consistent with the prediction in section 2 (hypothesis 1), the data therefore indicate that startup stage firms manifest greater uncertainty as nuts as opposed to lemons. The economic and statistical significance of each of these effects are similar in panel B in Table 5, illustrating the robustness of the results.
The second primary result in Table 5 is that adverse selection problems are much more pronounced for firms in high-tech industries, as evidenced by the greater number of statistically and economically significant coefficients for high-tech firms. The data indicate that high-tech firms are attracted toward debt. In particular, adverse selection effects for high-tech firms are 5.3% more pronounced (relative to nontech firms) for debt finance (and similarly, they are more likely to select toward mixes of debt and common, and preferred and common). The economic and statistical significance is similar for panel B in Table 5 for high-tech investments. There is further evidence in both panels A and B that high-tech entrepreneurs are more inclined to select away from securities in which they would be required to give up equity ownership to their investors; for example, in support of theoretical research on adverse selection and convertible securities (Brennan & Kraus, 1987), the evidence indicates that high-tech firms are 9.2% more likely to select away from convertible preferred equity, and 4.1% more likely to select away from convertible debt. Overall, as with the evidence for startup stage firms discussed immediately above, this evidence for high-tech entrepreneurial firms is consistent with the conjecture in hypothesis 1 that the uncertainty associated with investment in high-tech firms is better characterized as uncertainty in the variability of returns (nuts) as opposed to uncertainty in average expected returns (lemons).
The third primary result in Table 5 is that, in support of hypothesis 2, adverse selection is more problematic for high-tech investments than for syndicated investments. That is, there are fewer statistically significant coefficients for syndication than for high-tech investments in Table 5, panel A. (Note that the number of significant coefficients for early stage and syndicated investments is similar in Table 5, but the evidence in the next subsection explores differences from a companion perspective in more detail.) Further, note in panel B of Table 5 that there are two statistically significant coefficients for syndication among LSVCCs, two for foreign VCs, and one for syndication among the other VC types. Overall, therefore, adverse selection appears to be slightly more problematic for LSVCCs and U.S. investors investing in Canada. As discussed earlier (see text accompanying hypothesis 2), these latter results are expected, since U.S. VCs face distance hurdles in carrying out due diligence and the screening Canadian investees. Likewise, LSVCCs face statutory covenants that require they invest capital within 1-2 years of receipt from investor (or risk paying a fine to the government, or even revocation of the opportunity to operate as a LSVCC; Osborne and Sandler ); this statutory covenant on LSVCCs compromises the time that LSVCC managers have for carrying out their due diligence and hence exacerbates adverse selection problems for LSVCCs.
Robustness Cheeks on the Adverse Selection Tests. In this subsection, we carry out two robustness checks on the adverse selection tests. Robust check number 1 (Table 6) presents estimates similar to that in Table 5, but with syndicated and nonsyndicated investments treated separately as two different subsamples (we do not treat different types of investors in different subsamples for reasons previously indicated). The second robustness check (Table 7) again presents similar tests, but with the difference that the dollar value of the deal size is treated as a potentially endogenous variable as well. Overall, the tests in Table 5 yield quite similar results to these two alternative specifications, as explicitly shown in Tables 6 and 7, and as described subsequently.
In Table 6, panel A, the tests for the subsample of nonsyndicated investments show four statistically significant coefficients for startups; by contrast, Table 6, panel B, for the subsample of syndicated investments, shows only one statistically significant coefficient for startups. This reinforces the view that syndication facilitates screening such that the security does not "cause" the presence of the particular entrepreneurial characteristics; that is, syndication mitigates adverse selection problems associated with startups. Similarly, in Table 6, panel A, there are five significant coefficients for high-tech companies, but only four significant coefficients for high-tech firms in panel B. This latter difference in the number of significant coefficients for the syndication versus nonsyndication subsamples is less pronounced for high-tech companies, but again still illustrates a benefit to syndication with respect to mitigating adverse selection problems. As discussed, this evidence is consistent with prior work on the rationales underlying syndication (hypothesis 2; see also Lerner, 1994; Lockett & Wright, 1999, 2001; Manigart et al., 2002b; Wright & Lockett, 2003). Furthermore, note in Table 6 that the coefficients are significant in a way that is very similar to those in Table 5; as described in subsection 4.3, this supports the views that startups and high-tech companies are better characterized as nuts and not as lemons (hypothesis 1).
Table 7 carries out very similar tests as in Table 6. The difference in Table 7, relative to Table 6, is that for equation *, we also consider the possibility that Log(Deal size) is endogenous as well. In Tables 6 and 7, we did not consider that variable as endogenous since capital requirements to start a project are typically taken as exogenous in theoretical (e.g., Kanniainen & Keuschnigg, 2003, 2004; Keuschnigg, 2003) and empirical work (e.g., Manigart et al., 2002b). The estimates in Table 7 are presented separately for the first-round transactions that were not syndicated (panel A) and those that were syndicated (panel B), as in Table 6. The results in Table 7 are quite consistent with Table 6 in that for startup and high-tech firms, adverse selection problems are mitigated by syndication. Therefore, the results pertaining to hypotheses 1 and 2 previously described are not affected by the treatment of the deal size variable as either exogenous or endogenous.
As an aside, note that Table 7 indicates selection effects associated with a firm's capital requirements are not mitigated by virtue of syndication. Panel A (for nonsyndicated investments) indicates no statistically significant coefficients (even at the 10% level of significance) for capital requirements, while panel B (for syndicated investments) indicates two statistically significant coefficients (one at the 1% level and one at the 10% of significance). First-round syndicated investments naturally do tend to involve larger deal sizes (average deal size for first-round syndicated investments was $1,501,126, whereas average deal size for first-round nonsyndicated investments was $1,078,892). It is therefore not surprising that adverse selection problems vis-a-vis capital requirements are not smaller by virtue of syndication (unlike the role of adverse selection problems vis-a-vis startup stage and high-tech firms previously discussed) since syndication most often involves larger deals.
Finally, it is important to stress that we have consistently demonstrated in Tables 5-7 that syndication mitigates adverse selection problems for startups and high-tech companies. We have never claimed (and the data do not show) that syndication completely eliminates adverse selection. The view that agency problems can be mitigated but not eliminated has been well documented (see, e.g., Farmer & Winter, 1986).
Conclusion and Discussion
The Canadian VC market is a unique setting in which one may study nonsegmented financial intermediaries that are not restricted by the type of security they may employ. Within this setting, we build on previous research (e.g., Brennan & Kraus, 1987; DeMeza & Webb, 1987; Stiglitz & Weiss, 1981) by relating and tracking the presence of selection effects in capital structure to a variety of types of entrepreneurial firms and for the complete class of forms of finance. This article provided new tests of the empirical tractability in adverse selection in a nonsegmented market in which financial intermediaries use any number of different forms of finance and finance a variety of types of entrepreneurial firms. We controlled for changes in the economic environment over the 1991-2003 period both in regard to types of firms in the market for venture finance and in regard to selected securities.
The data introduced in this article indicated two primary results in terms of adverse selection and capital structure in the VC context. First, the evidence enabled a characterization of the type of uncertainty faced by investors in different types of entrepreneurial financing transactions in terms of the risk of financing a nut or a lemon. Canadian VC investors appear to face the greatest uncertainty among firms with high expected variability in returns (nuts) in financing startup stage and high-tech firms. Second, the evidence indicated strong support for the conjecture that syndication mitigates problems of adverse selection (consistent with earlier work on the rationales underlying the decision to syndicate; see Lerner, 1994; Lockett & Wright, 1999, 2001; Manigart et al., 2002b; Wright & Lockett, 2003). VC syndicates therefore facilitate an important role in screening potential investees and mitigating the risk of financing low-quality firms associated with different capital structure decisions.
As we are unaware of prior work that has carried out similar tests, we provide a few remarks about the robustness of our results and extensions to other contexts. First, we employed artificial regressions to test for the effect of endogeneity on the coefficient estimates in equation *. To the extent that equation * is not appropriate for other contexts, the implied adverse selection evidence may not be generally applicable.
Second, we do not consider transaction-specific relative price changes across different securities, among certain other factors; this may limit the extent to which the results are generalizable. Increasing the price of one security relative to another may lead some entrepreneurs to switch to an alternative source of finance (e.g., from the type of financial intermediary providing debt to the type of financial intermediary providing equity; see e.g., DeMeza & Webb, 1987, 1992; Stiglitz & Weiss, 1981). Given the limitation that our data do not indicate relative security prices for forms of finance in Canadian venture capital, we hold constant relative security prices and focus on adverse selection associated with the use of alternative forms of finance. (14) Similarly, our data do not indicate information pertaining to specific control rights (Gompers, 1998; Kaplan & Stromberg, 2003; Sahlman, 1990). To the extent that selection effects are associated with the allocation of specific control rights, there may be "noise" in our empirical evidence. Nevertheless, following DeMeza and Webb (1987, 1992), Myers and Mjluf (1984), Noe and Rebello (1992, 1996), Rebello (1995), and Stiglitz and Weiss (1981), among others, we suspect that selection effects are more broadly associated with form(s) of finance at the outset, and not the particular underlying terms of the financial contract determined pursuant to contractual negotiations and fine tuning.
Third, pronounced differences may exist among financial intermediaries other than VCs, as well as across countries, for institutional and legal reasons (see, e.g., Manigart et al., 1996; Manigart et al., 2002b, 2002c; Manigart et al., 2006; Mayer et al., 2005), for differences in governance, syndication, and returns across countries).
An important finding in this article is that problems of adverse selection are in fact context dependent and vary depending on firm characteristics and on the presence of investor syndicates. The context-dependent nature of our results suggests a number of avenues for future research. Further theoretical research could also incorporate different and/or broader contract spaces for alternative forms of finance, different types of entrepreneurial firms, and different financial intermediaries. Analyses of adverse selection in the context of work on team formation (e.g., in the spirit of Forbes, Borchert, Zellmer-Bruhn, & Sapienza, 2006; Ruef, Aldrich, & Carter, 2003) could be carried out with more detailed hand-collected data. Empirical work in different countries and/or among different financial intermediaries could also be fruitful in providing an improved understanding of adverse selection and capital structure among entrepreneurial firms.
I owe special thanks to Mary Macdonald for helpful comments and insights into venture capital in Canada. I owe thanks to seminar participants at the University of Alberta School of Business, the Schulich School of Business, York University, the University of British Columbia Sauder School of Business, the Center for Financial Studies, University of Frankfurt, the University of Cambridge Judge Institute of Management, the Universitita di Bologna Forli School of Business, and the Nottingham University Business School University of Nottingham Institute of Enterprise and Innovation Entrepreneurship Research Workshop. I have also greatly benefited from comments from an anonymous referee, as well as discussions and comments from Dirk De Clercq, Andy Lockett, Deniz Ucbasaran, and John Butler, and comments on an earlier draft from Varouj Aivazian, Paul Halpern, Mark Huson, Ted Liu, Frank Mathewson, Mary Macdonald, Randall Morck, Tom Ross, Barry Scholnick, and Ralph Winter. l am grateful for financial support from the University of Alberta Pearson Fellowship and the University of British Columbia Entrepreneurship Research Alliance Scholarship. Macdonald & Associates, Limited (Toronto) generously provided the data.
Aghion, P. & Bolton, P. (1992). An incomplete contracts approach to financial contracting. Review of Economic Studies, 59, 473-494.
Akerlof, G. (1970). The market for lemons: Qualitative uncertainty and the market mechanism. Quarterly Journal of Economics, 84, 488-500.
Angrist, J.D. (2001). Estimation of limited-dependent variable models with dummy endogenous regressors: Simple strategies for empirical practice. Journal of Business and Economic Statistics, 19, 2-28.
Berger, A. & Udell, G. (1998). The economics of small business finance: The roles of private equity and debt markets in the financial growth cycle. Journal of Banking and Finance, 22, 613-673.
Berglof, E. (1994). A control theory of venture capital finance. Journal of Law, Economics, and Organization, 10, 247-267.
Bergmann, D. & Hege, U. (1998). Venture capital financing, moral hazard, and learning. Journal of Banking and Finance, 22, 703-735.
Brennan, M. & Kraus, A. (1987). Efficiency financing under asymmetric information. Journal of Finance, 42, 1225-1243.
Casamatta, C. (2003). Financing and advising: Optimal financial contracts with venture capitalists. Journal of Finance, 58, 2059-2086.
Casamatta, C. & Haritchabalet, C. (2003). Learning and syndication in venture capital investments. Working paper 3876. CEPR.
Cumming, D.J. (2005a). Agency costs, institutions, learning and taxation in venture capital contracting. Journal of Business Venturing, 20, 573-622.
Cumming, D.J. (2005b). Capital structure in venture finance. Journal of Corporate Finance, 11, 550-585.
Cumming, D.J. & Macintosh, J.G. (2003). Canadian labor-sponsored venture capital corporations: Bane or boon? In A. Ginsberg & I. Hasan (Eds), New venture investment: Choices and consequences (pp. 169-200). Greenwich, CT: JAI Press.
Davidson, R. & MacKinnon, J.G. (1993). Estimation and inference in econometrics. New York: Oxford University Press.
De Clercq, D. & Sapienza, H.J. (2001). The creation of relational rents in venture capitalist-entrepreneur dyads. Venture Capital: An International Journal of Entrepreneurial Finance, 3(2), 107-127.
DeMeza, D. & Webb, D.C. (1987). Too much investment: A problem of asymmetric information. Quarterly Journal of Economics, 102, 281-292.
DeMeza, D. & Webb, D.C. (1992). Efficient credit rationing. European Economic Review, 36, 1277-1290.
Dimov, D. & De Clercq, D. (2006). Venture capital investment strategy and portfolio failure rate: A longitudinal study. Entrepreneurship Theory and Practice, 30, 207-223.
Farmer, R.E.A. & Winter, R.A. (1986). The role of options in the resolution of agency problems. Journal of Finance, 41, 1157-1170.
Forbes, D.P., Borchert, P.S., Zellmer-Bruhn, M.E., & Sapienza, H.J. (2006). Entrepreneurial team formation: An exploration of new member addition. Entrepreneurship Theory and Practice, 30, 225-248.
Gilson, R. & Schizer, D. (2003). Venture capital structure: A tax explanation for convertible preferred stock. Harvard Law Review, 116, 875-916.
Gompers, P.A. (1998). Ownership and control in entrepreneurial firms: An examination of convertible securities in venture capital investments. Working paper, Harvard University.
Gompers, P.A. & Lerner, J. (1999). The venture capital cycle. Cambridge, MA: MIT Press.
Halpern, P. (1997). Financing growth in Canada. Calgary, Alberta: University of Calgary Press.
Hege, U., Palomino, F., & Schwienbacher, A. (2004). Determinants of venture capital performance: Europe and the United States. Working paper, HEC School of Management, Tilburg University and the University of Amsterdam.
Hopenhayn, H.A. & Vereshchagina, G. (2004). Risk taking by entrepreneurs. Working paper, University of Rochester.
Jensen, M.C. & Meckling, W. (1976). Theory of the firm: Managerial behavior, agency costs, and capital structure. Journal of Financial Economics, 3, 305-360.
Jaaskelainen, M., Maula, M., & Seppa, T. (2006). Allocation of attention to portfolio companies and the performance of venture capital firms. Entrepreneurship Theory and Practice, 30, 185-206.
Kanniainen, V. & Keuschnigg, C. (2003). The optimal portfolio of startup firms in venture capital finance. Journal of Corporate Finance, 9, 521-534.
Kanniainen, V. & Keuschnigg, C. (2004). Startup investment with scarce venture capital support. Journal of Banking and Finance, 28(8), 1935-1959.
Kaplan, S.N. & Stromberg, P. (2003). Financial contracting theory meets the real world: An empirical analysis of venture capital contracts. Review of Economic Studies, 70, 281-315.
Keuschnigg, C. (2003). Taxation of a venture capitalist with a portfolio of firms. Oxford Economic Papers, 56, 285-306.
Landier, A. & Thesmar, D. (2003). Financial contracting with optimistic entrepreneurs. Working paper, NYU Stem School of Business.
Leland, H. & Pyle, D. (1977). Informational asymmetries, financial structure, and financial intermediation. Journal of Finance, 32, 371-387.
Lerner, J. (1994). The syndication of venture capital investments. Financial Management, 23, 16-27.
Lerner, J. & Schoar, A. (2005). Does legal enforcement affect financial transactions? The contractual channel in private equity. Quarterly Journal of Economics, 120, 223-246.
Lockett, A. & Wright, M. (1999). The syndication of private equity: Evidence from the U.K. Venture Capital: International Journal of Entrepreneurial Finance, 1,303-324.
Lockett, A. & Wright, M. (2001). The syndication of venture capital investments. Omega: The International Journal of Management Science, 29, 375-390.
Macdonald, M. (1992). Venture capital in Canada: A guide and sources. Toronto: Canadian Venture Capital Association.
Maddala, G.S. (2001). Introduction to econometrics. New York: Wiley.
Manigart, S., DeWaele, K., Wright, M., Robbie, K., Desbrieres, P., Sapienza, H., et al. (2000). Venture capital, investment appraisal, and accounting information: A comparative study of the US, UK, France, Belgium and Holland. European Financial Management, 6, 380-404.
Manigart, S., DeWaele, K., Wright, M., Robbie, K., Desbrieres, P., Sapienza, H.J., et al. (2002c). The determinants of the required returns in venture capital investments: A five-country study. Journal of Business Venturing, 17, 291-312.
Manigart, S., Korsgaard, M.A., Folger, R., Sapienza, H., & Baeyens, K. (2002a). The impact of trust on private equity, contracts. Working paper, Vlerick Leuven Ghent Management School.
Manigart, S., Lockett, A., Meuleman, M., Wright, M., Landstrom, H., Bruining, H., et al. (2002b). Why do European venture capital companies syndicate? Working paper, Vlerick Leuven Ghent Management School.
Manigart, S., Lockett, A., Meuleman, M., Wright, M., Landstrom, H., Bruining, H., et al. (2006). Venture capitalists' decision to syndicate. Entrepreneurship Theory and Practice, 30, 131-153.
Manigart, S., Sapienza, H., & Vermeir, W. (1996). Venture capital governance and value-added in four countries. Journal of Business Venturing, 11, 439-469.
Mayer, C., Schoors, K., & Yafeh, Y. (2005). Sources of funds and investment strategies of VC funds: Evidence from Germany, Israel, Japan and the UK. Journal of Corporate Finance, 11,586 608.
Moskowitz, T. & Vissing-Jorgensen, A. (2002). The returns to entrepreneurial investment: A private equity premium puzzle? American Economic Review, 92, 745-778.
Myers, S.C. & Mjluf, N. (1984). Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economies, 13, 187-222.
Noe, T.H. & Rebello, M.J. (1992). Adverse selection, contract design and investment distortion. Journal of Financial Intermediation, 2, 347-375.
Noe, T.H. & Rebello, M.J. (1996). Asymmetric information, managerial opportunism, financing and payout policies. Journal of Finance, 51,637-660.
Osborne, D. & Sandler, D. (1998). A tax expenditure analysis of labor-sponsored venture capital corporations. Canadian Tax Journal, 46, 499-574.
Rajan, R.G. & Zingales, L. (1995). What do we know about capital structure? Some evidence from international data. Journal of Finance, 50, 1421-1460.
Rebello, M.J. (1995). Adverse selection costs and the firm's financing and insurance decisions. Journal of Financial Intermediation, 4, 21-47.
Rivers, D. & Vuong, Q. (1988). Limited information estimators and exogeneity tests for simultaneous probit model. Journal of Econometrics, 39, 347-366.
Ruef, M., Aldrich, H.E., & Carter, N.M. (2003). The structure of founding teams: Homophily, strong ties and isolation among U.S. entrepreneurs. American Sociological Review, 68, 195-222.
Sahlman, W.A. (1990). The structure and governance of venture capital organizations. Journal of Financial Economics, 27, 473-521.
Sandler, D. (2001). The tax treatment of employee stock options: Generous to a fault. Canadian Tax Journal, 49, 259-302.
Sapienza, H. (1992). When do venture capitalists add value? Journal of Business Venturing, 7, 9-27.
Spence, M. (1973). Job market signaling. Quarterly Journal of Economics, 87, 355-374.
Stiglitz, J. & Weiss, A. (1981). Credit rationing in markets with imperfect information. American Economic Review, 73, 393-409.
Trester, J.J. (1998). Venture capital contracting under asymmetric information. Journal of Banking & Finance, 22, 675-699.
Wooldridge, J.M. (2002). Econometric analysis of cross section and panel clam. Cambridge, MA: MIT Press.
Wooldridge, J.M. (2000). Introduction to econometrics: A modern approach. Cincinnati, OH: South-Western College.
Wright, M. & Lockett, A. (2003). The structure and management of alliances: Syndication in the venture capital industry. Journal of Management Studies, 40, 2073-2102.
(1.) See, e.g., Berger and Udell (1998), Bergmann and Hege (1998), Casamatta (2003), Casamatta and Haritchabalet (2003), De Clercq and Sapienza (2001), Gompers and Lerner (1999), Kanniainen and Keuschnigg (2003, 2004), Keuschnigg (2003), Manigart, Sapienza, and Vermeir (1996), Manigart et al. (2000, 2002a, 2002b, 2002c), Mayer, Schoors, and Yafeh (2005), Sapienza (1992), Trester (1998), and Wright and Lockett (2003).
(2.) See, e.g., references listed in footnote 1.
(3.) For example, at a Master of Business Administration lecture at the University of Alberta in 2003, a practicing VC described the due diligence process as including, but not limited to, meetings and barbecues with the entrepreneurs' family and spouse.
(4.) Technical competence can be assessed, e.g., from academic transcripts and peer reviews from practitioners.
(5.) Following the CVCA definitions, a startup stage firm is based on a concept without a product or any marketing, or it may have a product being developed, but not yet sold commercially. An expansion stage firm requires significant capital for plant expansion and marketing to initiate full commercial production and sales.
(6.) The inclusion/exclusion of less commonly used forms of finance was immaterial to the analyses and results.
(7.) In 1990, the effective capital gains rate in Canada was 36%. Because marginal income tax rates rose through the 1990s, the effective capital gains rate rose as high as 39.8% (in 1995) before falling back to 36.6% in 1999. In March 2000, the effective capital gains rate was reduced to 23.2%. The different capital gains tax rates can influence security choice in Canada as it directly affects the attractiveness of equity (capital gains) to investors (see Cumming, 2005a).
(8.) Different market conditions can influence security design, as investors and entrepreneurs change with changing market conditions (see Cumming, 2005a).
(9.) Davidson and MacKinnon (1993) discuss DWH tests for continuous variables. These specification tests have also been referred to as simply "Hausman specification tests" in various texts. Angrist (2001), Rivers and Vuong (1988), Wooldridge (2000, 2002), and others consider various ways to deal with endogeneity in limited dependent variable models. The problem of causal inference is not fundamentally different (see, e.g., Angrist, 2001, section 4), but limited dependent variables do present additional challenges in certain contexts, particularly for censored regression models. The DWH artificial regressions with (potentially endogenous) standard binary dummy variables in the context of this article are quite robust to alternative methods and specifications. The tests are also quite robust to various possible adjustments for multiple potential endogenous variables as described in, e.g., Maddala (2001, chap. 12).
(10.) See also mission statements from venture capital firms; links are available from http://www.vfinance.com and http://www.cvca.ca.
(11.) The location variables are included to account for the fact that different types of entrepreneurial firms may be more prevalent in different provinces in Canada due to regional economic differences.
(12.) Different types of VC funds in Canada are not restricted from using different forms of finance. Different types of funds may prefer to finance certain types of entrepreneurial firms, and different forms of finance may be more appropriate for different types of entrepreneurial firms, but there is generally no direct incentive for different types of funds to use different forms of finance. A possible exception is the case of LSVCCs (see Cumming, 2005a, for details); therefore, we include the LSVCC variable in the endogeneity tests.
(13.) The calculations are [log(12,000,000)-log(11,000,000)] x 0.224 = 0.008, and [log(2,000,000)--log(1,000,000)] x 0.224 = 0.067. Logs were used to account for a diminishing effect of deal size on the probability of syndication. Linear specifications did not materially change the results reported. Alternative specifications are available upon request.
(14.) But this may not be a significant limitation for two reasons. Since all types of VCs in Canada employ all forms of finance (see Table 1), relative security prices for entrepreneurs seeking venture financing in Canada are likely quite stable, especially compared with situations in which different forms of finance are only available from different types of financial intermediaries. Moreover, startups are hard to value, and investors and entrepreneurs therefore delay valuation to the second round, when more information is available. I owe thanks to an anonymous referee for this latter helpful comment.
Please send correspondence to: Douglas Cumming, e-mail: firstname.lastname@example.org at Severino Center for Technological Entrepreneurship, Lally School of Management and Technology, Rensselaer Polytechnic Institute, Troy, New York.
Douglas Cumming is an associate professor and director at the Severino Center for Technological Entrepreneurship at the Lally School of Management and Technology, Rensselaer Polytechnic Institute, Troy, New York.
Table 1 Summary of Canadian VC Investment Data by Type of Security, VC Fund, and Entrepreneurial Firm Panel A. Private independent limited partnership venture capital funds Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 91 0.769 0.231 Common equity and/or warrants 367 0.796 0.204 Convertible preferred equity 109 0.826 0.174 Debt 91 0.703 0.297 Convertible debt 164 0.732 0.268 Debt and common equity 31 0.419 0.581 Preferred and common equity 18 0.722 0.278 Mixes of other securities 124 0.798 0.202 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.857 1,437.808 0.648 Common equity and/or warrants 0.766 870.575 0.477 Convertible preferred equity 0.826 888.457 0.578 Debt 0.659 647.483 0.308 Convertible debt 0.774 753.058 0.390 Debt and common equity 0.452 638.415 0.516 Preferred and common equity 0.667 591.940 0.500 Mixes of other securities 0.855 757.091 0.605 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.341 0.319 0.341 Common equity and/or warrants 0.529 0.302 0.169 Convertible preferred equity 0.404 0.330 0.266 Debt 0.692 0.176 0.132 Convertible debt 0.537 0.268 0.195 Debt and common equity 0.871 0.032 0.097 Preferred and common equity 0.667 0.056 0.278 Mixes of other securities 0.339 0.387 0.274 Panel B. Corporate venture capital funds Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 27 0.593 0.407 Common equity and/or warrants 82 0.622 0.378 Convertible preferred equity 47 0.553 0.447 Debt 80 0.313 0.688 Convertible debt 43 0.628 0.372 Debt and common equity 26 0.346 0.654 Preferred and common equity 4 0.250 0.750 Mixes of other securities 77 0.662 0.338 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.704 1,638.536 0.778 Common equity and/or warrants 0.671 1,411.269 0.646 Convertible preferred equity 0.809 1,715.965 0.766 Debt 0.438 1,194.438 0.450 Convertible debt 0.651 976.280 0.698 Debt and common equity 0.346 1,081.462 0.577 Preferred and common equity 0.750 739.335 0.750 Mixes of other securities 0.584 793.493 0.623 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.333 0.296 0.370 Common equity and/or warrants 0.390 0.280 0.329 Convertible preferred equity 0.234 0.447 0.319 Debt 0.325 0.538 0.138 Convertible debt 0.302 0.372 0.326 Debt and common equity 0.577 0.269 0.154 Preferred and common equity 0.500 0.500 0.000 Mixes of other securities 0.065 0.234 0.701 Panel C. Government venture capital funds Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 44 0.773 0.227 Common equity and/or warrants 284 0.824 0.176 Comrrtible preferred equity 49 0.796 0.204 Debt 65 0.692 0.308 Convertible debt 96 0.771 0.229 Debt and common equity 28 0.643 0.357 Preferred and common equity 8 0.750 0.250 Mixes of other securities 86 0.849 0.151 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.909 780.352 0.795 Common equity and/or warrants 0.574 575.024 0.342 Comrrtible preferred equity 0.837 998.616 0.816 Debt 0.523 537.813 0.492 Convertible debt 0.615 505.848 0.615 Debt and common equity 0.607 606.096 0.536 Preferred and common equity 0.625 770.450 0.875 Mixes of other securities 0.651 579.790 0.547 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.318 0.227 0.455 Common equity and/or warrants 0.718 0.120 0.162 Comrrtible preferred equity 0.265 0.163 0.571 Debt 0.477 0.169 0.354 Convertible debt 0.448 0.177 0.375 Debt and common equity 0.536 0.107 0.357 Preferred and common equity 0.750 0.000 0.250 Mixes of other securities 0.384 0.151 0.465 Panel D. Institutional investors Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 31 0.323 0.677 Common equity and/or warrants 113 0.442 0.558 Convertible preferred equity 18 0.667 0.333 Debt 144 0.438 0.563 Convertible debt 56 0.429 0.571 Debt and common equity 47 0.404 0.596 Preferred and common equity 6 0.167 0.833 Mixes of other securities 53 0.661 0.340 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.645 1,150.588 0.516 Common equity and/or warrants 0.575 1,813.993 0.372 Convertible preferred equity 0.778 1,471.400 0.722 Debt 0.340 878.360 0.347 Convertible debt 0.429 1,572.134 0.393 Debt and common equity 0.319 1,176.706 0.398 Preferred and common equity 0.667 1,805.777 0.500 Mixes of other securities 0.528 3,152.704 0.528 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.419 0.161 0.419 Common equity and/or warrants 0.496 0.345 0.159 Convertible preferred equity 0.389 0.222 0.389 Debt 0.306 0.292 0.403 Convertible debt 0.429 0.321 0.250 Debt and common equity 0.447 0.383 0.170 Preferred and common equity 0.833 0.167 0.000 Mixes of other securities 0.321 0.377 0.302 Panel E. Labor-sponsored venture capital funds Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 64 0.688 0.313 Common equity and/or warrants 289 0.578 0.422 Convertible preferred equity 85 0.741 0.259 Debt 150 0.520 0.480 Convertible debt 114 0.596 0.404 Debt and common equity 97 0.536 0.464 Preferred and common equity 30 0.600 0.400 Mixes of other securities 130 0.662 0.338 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.609 1,666.579 0.516 Common equity and/or warrants 0.429 1,407.547 0.391 Convertible preferred equity 0.776 1,179.064 0.682 Debt 0.360 1,658.972 0.273 Convertible debt 0.658 1,408.740 0.518 Debt and common equity 0.320 1,075.662 0.278 Preferred and common equity 0.267 780.316 0.367 Mixes of other securities 0.585 1,182.964 0.523 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.344 0.297 0.359 Common equity and/or warrants 0.561 0.232 0.208 Convertible preferred equity 0.282 0.271 0.447 Debt 0.513 0.267 0.220 Convertible debt 0.430 0.237 0.333 Debt and common equity 0.660 0.227 0.113 Preferred and common equity 0.400 0.467 0.133 Mixes of other securities 0.385 0.262 0.354 Panel F. Foreign (U.S.) venture capital funds Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 17 0.765 0.235 Common equity and/or warrants 48 0.688 0.313 Convertible preferred equity 15 0.800 0.200 Debt 8 0.625 0.375 Convertible debt 9 1.000 0.000 Debt and common equity 2 1.000 0.000 Preferred and common equity 3 0.000 1.000 Mixes of other securities 80 0.513 0.488 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 1.000 2,460.133 0.824 Common equity and/or warrants 0.896 4,473.606 0.750 Convertible preferred equity 0.867 3,238.532 0.933 Debt 0.750 2,048.213 0.875 Convertible debt 1.000 2,096.885 0.889 Debt and common equity 0.500 1,121.992 0.500 Preferred and common equity 0.667 833.526 1.000 Mixes of other securities 0.975 6,348.063 0.900 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.059 0.412 0.529 Common equity and/or warrants 0.333 0.375 0.292 Convertible preferred equity 0.200 0.467 0.333 Debt 0.625 0.375 0.000 Convertible debt 0.556 0.222 0.222 Debt and common equity 1.000 0.000 0.000 Preferred and common equity 0.667 0.333 0.000 Mixes of other securities 0.038 0.713 0.250 Panel G. Other types of investors Total number of Proportion first-round Proportion of Type of security investments of startup expansion Preferred equity 24 0.667 0.333 Common equity and/or warrant 157 0.783 0.217 Convertible preferred equity 53 0.755 0.245 Debt 50 0.720 0.280 Convertible debt 62 0.710 0.290 Debt and common equity 20 0.800 0.200 Preferred and common equity 12 0.833 0.167 Mixes of other securities 86 0.767 0.233 Average Proportion amount Proportion of invested of Type of security high-tech ('000) syndication Preferred equity 0.708 1,297.772 0.917 Common equity and/or warrant 0.662 1,346.061 0.873 Convertible preferred equity 0.736 1,138.705 0.925 Debt 0.460 1,125.385 0.900 Convertible debt 0.581 755.826 0.968 Debt and common equity 0.250 753.110 0.900 Preferred and common equity 0.667 2,701.023 0.750 Mixes of other securities 0.791 3,756.433 0.756 Proportion Proportion Proportion in years in in Type of security 1991-1998 1999-2000 2001-2003 Preferred equity 0.250 0.375 0.375 Common equity and/or warrant 0.363 0.299 0.338 Convertible preferred equity 0.377 0.321 0.303 Debt 0.640 0.180 0.180 Convertible debt 0.452 0.306 0.242 Debt and common equity 0.450 0.200 0.350 Preferred and common equity 0.333 0.500 0.167 Mixes of other securities 0.093 0.372 0.535 Note: This table presents the number of first-round investments by type of VC fund. entrepreneurial firm, and security. Different securities over different years and by (non-) syndication are presented. The proportions of startup and expansion sum to 1.0. The proportion of nonhigh-tech is equal to I minus the proportion of high-tech. The proportion of nonsyndication is equal to 1 minus the proportion of syndication. The average amounts invested are in real $2.000 Canadian dollars. The different types of VC funds are indicated in each panel (A-G) separately. VC, venture capitalist. Table 2 Syndication and Integration among Different Types of VC Funds in Canada (1) Total (2) Number of (3) Proportion number of first-round of first-round first-round syndicated syndicated fund fund fund investments investments investments has been invested (all rounds) Private independent 995 489 0.491 limited partnership VCs Corporate VCs 386 242 0.627 Government VCs 660 332 0.503 Institutional VCs 468 188 0.402 LSVCCs 959 410 0.428 Foreign VCs 182 155 0.852 Other types of VCs 464 405 0.873 Totals 4,114 2,221 0.540 (4) Total (5) Total (6) Proportion number of number of of different different different companies in companies in companies in which the which fund which fund partnership type has been type fund type VCs invested as a has invested (all rounds) syndicated as a investor syndicated (all rounds) investor (all rounds) Private independent 1,322 736 0.557 limited partnership VCs Corporate VCs 692 433 0.626 Government VCs 913 563 0.617 Institutional VCs 668 365 0.546 LSVCCs 1,516 710 0.468 Foreign VCs 279 257 0.921 Other types of VCs 840 770 0.917 Totals 3,696 * 1,440 * 0.390 (7) Number (8) Number (9) Number of different of different of different companies companies companies syndicated syndicated syndicated with private with corporate with limited VCs government VCs Private independent 39 limited partnership VCs Corporate VCs 197 25 Government VCs 263 151 12 Institutional VCs 165 121 163 LSVCCs 328 175 247 Foreign VCs 178 109 100 Other types of VCs 392 214 321 Totals 736 * 433 * 563 * (10) Number (11) Number of different of different companies companies syndicated syndicated with with LSVCCs institutional VCs Private independent limited partnership VCs Corporate VCs Government VCs Institutional VCs 16 LSVCCs 158 34 Foreign VCs 68 116 Other types of VCs 155 379 Totals 365 * 710 * (12) Number (13) Number of different of different companies companies syndicated syndicated with foreign with other VCs types of VCs Private independent limited partnership VCs Corporate VCs Government VCs Institutional VCs LSVCCs Foreign VCs 7 Other types of VCs 134 8 Totals 257 * 770 * * Note: This table presents the total number of first-round fund investments (column 1), the total number of syndicated first-round fund investments (column 2), and the proportion of first-round syndicated investments (column 3, which is column 2 divided by column 1). In addition, to illustrate the extent of integration among different fund types in Canada, this table presents, by the number of investee companies for all investment rounds, the following information: the total number of different companies in which the fund type has been involved as a syndicated investor (column 4), the number of different companies in which the fund type has been involved as a syndicated investor (column 5), the proportion of different companies in which the fund type has been involved as a syndicated investor (column 6, which is column 5 divided by column 4), and the number of companies in which the fund type has syndicated with different types of other VCs (columns 7-13). Note that the * next to some of the totals indicates the column does not add up to the total because multiple syndications (i.e., syndications with more than two investors) across different fund types give rise to double counting of companies. VC, venture capitalist; LSVCC, labor-sponsored venture capital company. Table 3 Correlations between Securities and Type of VC Fund and Location Common equity Convertible Preferred and/or preferred equity warrants equity Private limited partnership 0.04 0.05 0.04 Corporate VC 0.00 -0.08 0.03 Government VC -0.01 0.10 -0.03 Institutional VC -0.01 -0.06 -0.07 LSVCC -0.01 -0.03 -0.01 Foreign (U.S.) investor 0.02 -0.03 -0.01 Other type of VC -0.03 0.01 0.03 British Columbia 0.02 -0.03 0.12 Alberta 0.01 0.04 0.02 Saskatchewan -0.03 0.01 -0.03 Manitoba -0.01 0.04 0.00 Ontario 0.06 0.01 0.04 Quebec -0.05 -0.02 -0.11 Debt and Convertible common Debt debt equity Private limited partnership -0.08 0.05 -0.07 Corporate VC 0.06 -0.02 0.01 Government VC -0.06 0.02 -0.03 Institutional VC 0.17 -0.01 0.06 LSVCC 0.02 -0.02 0.09 Foreign (U.S.) investor -0.06 -0.05 -0.04 Other type of VC -0.04 0.00 -0.03 British Columbia -0.05 -0.01 -0.05 Alberta -0.02 -0.03 -0.03 Saskatchewan 0.07 -0.02 0.00 Manitoba 0.00 -0.02 0.00 Ontario -0.10 -0.01 -0.07 Quebec 0.11 0.04 0.10 Preferred Mixes of and common other equity securities Private limited partnership -0.01 -0.05 Corporate VC -0.02 0.04 Government VC -0.02 -0.03 Institutional VC -0.02 -0.04 LSVCC 0.05 -0.03 Foreign (U.S.) investor 0.00 0.17 Other type of VC 0.02 0.03 British Columbia -0.02 0.03 Alberta 0.00 -0.01 Saskatchewan -0.02 -0.01 Manitoba 0.05 -0.03 Ontario -0.04 0.07 Quebec 0.04 -0.08 Note: This table presents correlation coefficients across securities and type of VC fund and location of entrepreneurial firms. Correlations greater than 0.04 are statistically significant at the 5% level of significance. Correlations are presented for first-round investments only (4,114 observations). VC, venture capitalist. Table 4 Logit Regressions for Startup, High-Tech, and Syndication: First-Stage DWH Regressions (3) Syndication: (1) (2) All types Startup High-tech of VCs Year fixed effects? Yes Yes Yes Startup -- -- 0.169 *** High-tech -- -- 0.165 *** Log (deal size) -- -- 0.224 *** Corporate -0.298 *** -0.300 *** 0.120 *** Government 0.057 ** -0.772 *** 0.102 *** Institutional -0.343 *** -0.328 *** -0.128 *** LSVCC -0.192 *** -0.346 *** -0.869 *** Foreign investor -0.203 *** 0.120 ** 0.122 ** Other type of VC -0.037 -0.202 *** 0.390 *** British Columbia 0.002 0.165 *** 0.132 ** Alberta -0.130 * -0.167 ** -0.955 Saskatchewan -0.268 *** -0.499 *** 0.106 Manitoba -0.057 -0.333 *** 0.136 * Ontario -0.073 0.810 -0.548 Quebec -0.084 -0.128 ** 0.150 ** [chi square] 403.240 *** 897.314 *** 1,794.612 *** Log likelihood -2,114.683 -2.270.576 -1,941.212 Pseudo [R.sup.2] 0.077 0.165 0.316 (5) (6) Syndication: Syndication: (4) Foreign Non-LSVCCs Syndication: (U.S.) and non-U.S. LSVCCs only VCs only VCs only Year fixed effects? Yes Yes Yes Startup 0.003 -0.025 *** 0.155 *** High-tech 0.009 0.015 *** 0.116 *** Log (deal size) 0.026 *** 0.004 *** 0.106 *** Corporate -- -- -- Government -- -- -- Institutional -- -- -- LSVCC -- -- -- Foreign investor -- -- -- Other type of VC -- -- -- British Columbia 0.042 -0.008 ** 0.223 Alberta -0.003 -0.017 *** -0.259 Saskatchewan 0.028 -- 0.161 ** Manitoba 0.139 * -- -0.564 Ontario 0.026 -0.020 *** -0.119 ** Quebec -0.008 -0.046 *** 0.942 * [chi square] 206.219 *** 159.099 *** 649.963 *** Log likelihood -1,231.203 -580.696 -2,447.950 Pseudo [R.sup.2] 0.077 0.120 0.117 *, **, and *** denote significance at 10, 5, and 1% levels, respectively. Note: This table presents logit regression estimates of the determinants of startup stage (equation 1) and high-tech (equation 2) investments, and the determinants of syndicated investments (equation 3). Also, equations 4-6, respectively, present regressions for syndications with LSVCCs only, foreign VCs only, and other types of investors excluding foreign VCs and LSVCCs. Year fixed effects for 1991-2003 are used. Dummy variables for private limited partnerships and for the Maritime Provinces are suppressed to avoid perfect collinearity. The marginal effects are presented to explicitly show economic significance for the logit regressions. Only first-round investments are considered for a total of 4,114 observations. The variable "--" is excluded because of collinearity in the syndication regressions, and because it is not applicable in the startup and high-tech regressions. VC, venture capitalist; LSVCC, labor-sponsored venture capital company; DWH, Durbin-Wu-Hausman. Table 5 DWH Tests of the Effect of Endogeneity vis-a- vis Firm Characteristics and Forms of Finance Panel A. All types of syndicated investments considered together Convertible Preferred Common preferred Startup -0.005 -0.053 ** 0.022 High-tech -0.013 0.013 -0.092 *** Syndication (all types of VCs) 0.000 -0.058 ** -0.040 *** Convertible Debt and Debt debt common Startup 0.043 *** 0.016 0.017 * High-tech 0.053 *** -0.041 ** 0.030 *** Syndication (all types of VCs) 0.026 0.024 -0.007 Preferred Other and combinations common of securities Startup -0.011 -0.029 High-tech 0.017 *** 0.033 ** Syndication (all types of VCs) -0.003 0.057 *** [chi square] = 948.8370 ***; log likelihood = -7,162.593; pseudo [R.sup.2] = 0.062 Panel B. LSVCC, foreign and other syndications considered separately Convertible Preferred Common preferred Startup -0.006 -0.061 ** 0.016 High-tech -0.016 0.016 -0.088 *** Syndication among LSVCCs -0.040 * 0.027 0.026 Syndication among foreign VCs 0.014 0.012 -0.014 Syndication among other VCs -0.011 -0.021 0.020 * Convertible Debt and Debt debt common Startup 0.046 *** 0.014 0.017 ** High-tech 0.048 *** 0.040 ** 0.029 *** Syndication among LSVCCs -0.031 0.043 -0.018 Syndication among foreign VCs -0.058 -0.085 ** -0.037 Syndication among other VCs -0.012 0.009 -0.001 Preferred Other and combinations common of securities Startup -0.010 -0.015 High-tech 0.017 *** 0.033 ** Syndication among LSVCCs -0.005 0.052 * Syndication among foreign VCs 0.016 0.179 *** Syndication among other VCs 0.004 0.013 [chi square] = 989.935 ***; log likelihood = -7,142.044; pseudo [R.sup.2] = 0.065 *, **, and *** denote significance at 10, 5, and 1% levels, respectively. Note: This table presents DWH tests of the effect of endogeneity vis-a-vis forms of finance and firm characteristics (stage of development, industry). The regressions are multinomial logit models of equation 7 described in the text of the article. The multinomial logit regression coefficients are not presented; rather, the marginal effects are presented to explicitly show economic significance. The reported right-hand-side variables are the residuals from the regressions reported in Table 4. Other independent right-hand-side variables are not reported for conciseness. Only first-round investments are considered. Panel A considers all syndications among all types of VCs together. Panel B considers separately syndications that involved LSVCCs, foreign VCs, and all other different types of VCs. The full sample of 4,114 first-round investments is used. VC, venture capitalist; LSVCC, labor-sponsored venture capital company; DWH, Durbin-Wu- Hausman. Table 6 DWH Tests: Robustness Check Number 1 Panel A. Nonsyndicated first-round investments Convertible Preferred Common preferred Startup 0.014 -0.082 ** 0.016 High-tech -0.021 -0.007 0.048 *** Convertible Debt and Debt debt common Startup 0.055 ** -0.031 0.002 High-tech 0.089 *** -0.013 0.052 *** Other Preferred combinations and common of securities Startup -0.019 *** 0.045 *** High-tech 0.018 *** -0.069 *** [chi square] = 459.326 ***; log likelihood = -3.147.875; pseudo [R.sup.2] = 0.068 Panel B. Syndicated first-round investments Convertible Preferred Common preferred Startup 0.003 -0.080 0.050 High-tech -0.071 ** 0.061 -0.092 ** Convertible Debt and Debt debt common Startup 0.022 0.021 -0.059 ** High-tech 0.079 *** -0.001 0.021 * Other Preferred combinations and common of securities Startup 0.000 0.042 High-tech -0.015 0.018 [chi square] = 482.706 ***; log likelihood = -3,932.897; pseudo [R.sup.2] = 0.058 *, and *** denote significance at 10, 5, and 1% levels, respectively. Note: This table presents DWH tests of the effect of endogeneity vis-a-vis forms of finance and firm characteristics (stage of development, capital requirements, and industry). The regressions are multinomial logit models of equation 7 described in the text of the article. The multinomial logit regression coefficients are not presented; rather, the marginal effects are presented to explicitly show economic significance. The reported independent variables are the residuals from regressions analogous to those reported in Table 4 (for the respective subsamples for syndicated and nonsyndicated investments). Other independent variables are not reported for conciseness. Only first-round investments are considered. Panel A considers 1,893 investments that were not syndicated. Panel B considers 2,221 investments that were syndicated. DWH, Durbin-Wu-Hausman. Table 7 DWH Tests: Robustness Check Number 2 Panel A. Nonsyndicated First-Round Investments Convertible Preferred Common preferred Startup 0.007 -0.064 * 0.012 High-tech -0.017 -0.015 -0.046 *** Log ($ deal size ['000]) -0.017 0.045 -0.010 Convertible Debt and Debt debt common Startup 0.044 * -0.036 0.011 High-tech 0.093 *** -0.011 0.047 *** Log ($ deal size ['000]) -0.026 -0.012 0.023 Other Preferred combinations and common of securities Startup -0.019 ** 0.044 * High-tech 0.018 *** -0.069 *** Log ($ deal size ['000]) 0.000 -0.003 [chi square] = 466.027 ***; log likelihood = -3,144.524; pseudo [R.sup.2] = 0.069 Panel B. Syndicated first-round investments Convertible Preferred Common preferred Startup 0.001 -0.075 0.047 High-tech -0.069 ** 0.058 -0.091 ** Log ($ deal size ['000]) -0.004 0.017 -0.002 Convertible Debt and Debt debt common Startup 0.026 0.028 -0.056 ** High-tech 0.076 *** -0.006 0.018 Log ($ deal size ['000]) 0.012 0.025 0.017 * Other Preferred combinations and common of securities Startup 0.000 0.028 High-tech -0.015 0.029 Log ($ deal size ['000]) -0.001 -0.064 *** [chi square] = 504.376 ***; log likelihood = -3.922.062; pseudo [R.sup.2] = 0.060 *, **, and *** denote significance at 10, 5, and 1% levels, respectively. Note: This table presents DWH tests of the effect of endogeneity vis- a-vis forms of finance and firm characteristics (stage of development, capital requirements and industry). The regressions are multinomial logit models of equation 7 described in the text of the article. The multinomial logit regression coefficients are not presented; rather, the marginal effects presented to explicitly show economic significance. The reported independent variables are the residuals from regressions analogous to those reported in Table 4 (for the respective subsamples for syndicated and nonsyndicated investments), and the residuals from an analogous regression explaining the log of deal sizes. Other independent variables are not reported for conciseness. Only first-round investments are considered. Panel A considers 1,893 investments that were not syndicated. Panel B considers 2,221 investments that were syndicated. DWH, Durbin-Wu-Hausman.
|Printer friendly Cite/link Email Feedback|
|Publication:||Entrepreneurship: Theory and Practice|
|Date:||Mar 1, 2006|
|Previous Article:||Venture capitalists' decision to syndicate.|
|Next Article:||Allocation of attention to portfolio companies and the performance of venture capital firms.|