Journal of Empirical Finance

Transcription

1 Journal of Empirical Finance 19 (2012) Contents lists available at SciVerse ScienceDirect Journal of Empirical Finance journal homepage: Economic freedom and cross-border venture capital performance Lanfang Wang a,, Susheng Wang b a Institute of Accounting and Finance, Shanghai University of Finance and Economics,China b Hong Kong University of Science and Technology, Hong Kong article info abstract Article history: Received 15 September 2010 Received in revised form 16 October 2011 Accepted 21 October 2011 Available online 30 October 2011 JEL classification: G15 G24 G34 Keywords: Cross-border VC Economic freedom Legality Entrepreneurial activity We investigate the determinants of cross-border venture capital (VC) performance using a large sample of 10,205 cross-border VC investments by 1906 foreign VC firms (VCs) in 6535 domestic portfolio companies. We focus on the impact of a domestic country's economic freedom on the performance of both VC investments and portfolio companies using a probit model and the Cox hazard model. After controlling for other related factors of domestic countries, portfolio companies, VCs and the global VC market, as well as year and industry fixed effects, we find that a domestic country's economic freedom is crucial to cross-border VC performance. In particular, in a more economically free country, as measured by the raw values of, quartiles of or the ranking in the index of economic freedom (IEF), a foreign VC-backed portfolio company is more likely to pull off a successful exit through an IPO (initial public offering) or an M&A (merger and acquisition), and a foreign VC firm is likely to spend a shorter investment duration in the portfolio company. We also identify interesting evidence on the impact of many other level factors of domestic countries, portfolio companies, VCs and the global VC market on cross-border VC performance Elsevier B.V. All rights reserved. 1. Introduction Venture capital (VC) is widely regarded as an active driving force in fostering entrepreneurship, financing startups, creating public companies and encouraging innovations in the U.S. over the past several decades (Barry et al., 1990; Kaplan and Strömberg, 2003; Kortum and Lerner, 2000; Lerner, 1994; Sahlman, 1990). It is widely believed that VC is instrumental in bringing innovations to markets at a rapid pace, thereby enhancing economic growth. VC has been increasingly internationalized, as marked by an increasing number of cross-border entries of VCs starting from the mid-1990s. Cross-border VC is crucial to growth-oriented but high-risk new ventures, especially in countries where the domestic supply of private equity is limited. Our focus is to examine the impact of a domestic country's economic freedom on cross-border VC performance. The idea that economic freedom is essential for economic efficiency has been a cornerstone in economic theory since Smith (1776), who argues for the role of the invisible hand in well-functioning markets. In theory, a free economy is defined as the so-called Arrow Debreu world, where economic efficiency is guaranteed in general equilibrium (Arrow and Debreu, 1954; Hart, 1980; McKenzie, 1959). In empirical studies, economic freedom has been widely observed to be important to economic growth (Gwartney et al., 1999; Haan and Sturm, 2000; Heckelman, 2000; Wu and Davis, 1999), income equality (Berggren, 1999; Scully, 2002) and employment (Feldmann, 2007, 2008). For VC investments, the role of economic freedom lies in two aspects. On the one hand, free markets are widely believed to provide more opportunities for early stage and mostly not-yet-profitable companies, which is essential to VC investments (BjØrnskov and Foss, 2008; Kreft and Sobel, 2005; Nyström, 2008; Sobel et al., 2007; Wennekers et al., 2002). The more Corresponding author. addresses: wang.lanfang@mail.shufe.edu.cn (L. Wang), s.wang@ust.hk (S. Wang) /$ see front matter 2011 Elsevier B.V. All rights reserved. doi: /j.jempfin

2 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) constraints there are on the allocation of resources in production and consumption, the fewer growth opportunities and chances of success for VCs. On the other hand, the government is more likely to act as a helping hand instead of a grabbing hand in free markets, which is also essential for the survival and success of VCs. In an economically free market, the burden of bureaucracy and corruption would be smaller and the government would try to provide a steady and reliable monetary environment, a free and open investment environment and a transparent and open financial system. Contract enforcement and investor protection, especially regarding private properties, also tend to be better in free markets. For cross-border VC investments, economic freedom is particularly important since investment risk, transaction costs, asymmetric information and agency problems are much more severe, which makes these types of investments more sensitive to the domestic country's economic environment (Cumming et al., 2006; Gompers et al., 2003; Jeng and Wells, 2000). Using a large sample of 10,205 cross-border VC investments from 1906 foreign VCs in 6535 domestic portfolio companies covering 35 countries, we investigate the impact of a domestic country's economic freedom on cross-border VC performance. We provide analysis of both VC investments and portfolio companies using both a probit model and the Cox hazard model. After controlling for other related factors of domestic countries, portfolio companies, VCs and the global VC market, as well as industry and year fixed effects, we find that a domestic country's economic freedom is crucial to the performance of crossborder VC. In particular, in an economically free country, as measured by either the raw values of, quartiles of or the ranking in economic freedom over the investment period, a foreign VC-backed portfolio company is more likely to pull off a successful exit through an IPO (initial public offering) or an M&A (merger and acquisition), and a foreign VC firm is likely to spend a shorter investment duration in a portfolio company. Specifically, a one standard deviation increase in our measure of economic freedom Index of Economic Freedom, Economic Freedom Quartile or Ranking in Economic Freedom at the mean level increases the likelihood of a successful exit by 2.02%, 2.76% or 3.02%, and the hazard of a successful exit by 5.35%, 12.59% or 13.69%, respectively. Our main findings are robust to many tests, including tests on a change in economic freedom, tests on exit choices (IPO versus M&A), tests on decomposed effects of economic freedom, tests considering domestic country fixed effects and tests using subsamples and alternative measures. We further find several interesting results from other related factors of domestic countries, portfolio companies, VCs and the global VC market. For example, we find legal quality to be positively related to the likelihood of a successful exit through an IPO or an M&A, which is consistent with the empirical work by Cumming et al. (2006). GDP per capital is found to be negatively related to cross-border VC performance, which is somewhat consistent with the well-known convergence hypothesis in the exogenous growth literature (Barro, 1991, 1996). Also, a domestic country's entrepreneurial activity as measured by the number of VC deals and the number of patents granted over the investment period is positively associated with the likelihood of a successful exit through an IPO or an M&A. Further, our results provide evidence confirming the positive impact of portfolio company quality and local VCs' participation and the negative impact of early stage investments and VCs' portfolio size on cross-border VC performance, which has been examined in previous studies (Cumming, 2006; Cumming and Dai, 2010, 2011; Dai et al., 2010; Mäkelä and Maula, 2008; Tykvová and Schertler, 2011). In particular, our finding that market conditions have a strong impact on the exit choice between an IPO and an M&A sheds light on the exit timing of cross-border VCs. There are few studies on cross-border VC performance that are closely related to ours. 1 Wang and Wang (2011) investigate the determinants of cross-border VC performance in China using a small sample of 495 VC investments in 243 portfolio companies by 84 foreign VCs. They document that foreign VCs' human capital has little importance to the likelihood of successful exits. In contrast, the domestic entrepreneurs' experience is crucial to VC performance. Our study differs significantly from that of Wang and Wang (2011) in that we provide an international study on cross-border VC performance covering 35 domestic countries, while Wang and Wang (2011) focus on one domestic country, China. Also, we investigate the impact of a domestic country's economic freedom on cross-border VC performance after controlling for many other level factors of domestic countries, portfolio companies, VCs and the global VC market, while Wang and Wang (2011) focus on foreign VCs' human capital and domestic entrepreneurs' experience. Dai et al. (2010) partially address cross-border VC performance using a sample of 4254 rounds of VC financing by 468 VCs in six Asian countries. They suggest a positive role of local VCs; that is, cooperation with local VCs helps mitigate information asymmetry and contributes to the likelihood of a portfolio company's successful exit through an IPO. Different from Dai et al. (2010), we emphasize the role of a country-level factor, namely the domestic country's economic freedom, in cross-border VC performance while controlling for many other factors including local VCs' participation which is also examined in their study. Our results on the impact of local VCs' participation are quite consistent with those of Dai et al. (2010), and provide supporting evidence to their finding. Another closely related study is that of Watson and George (2010), who examine the influence of several country-level factors, such as the size of the economy, business freedom, trade protection, burden of taxation, government size, price stability, openness, corruption and cultural distance, on the rate of return in 72 foreign acquisitions from 144 transactions in 24 different countries. Even though the research question addressed by Watson and George (2010) is similar to ours, there are substantial differences between the two studies. First, they use a small sample consisting of only foreign acquisitions of private equity, which cannot represent the bulk of cross-border VC investments. We use a large sample of 10,205 cross-border VC investments by 1906 foreign VCs in 1 For a comparison of VC investment characteristics in different countries, see Black and Gilson (1998), Lockett et al. (2002), Cumming and MacIntosh (2003), Bruton et al. (2004), Lerner and Schoar (2005), Mayer et al. (2005), Ahlstrom and Bruton (2006), Cumming et al. (2006), Schwienbacher (2008), Cumming et al. (2010), etc. For other aspects of cross-border VC investments (such as economic factors in attracting funding from foreign VCs and the role of local investors), see Mäkelä and Maula (2008), Balcarcel et al. (2009), Schertler and Tykvová (2011), Chemmanur et al. (2011), Tykvová and Schertler (2011), etc. For determinants of domestic VC performance, see Hochberg et al. (2007), Zarutskie (2007), Bottazzi et al. (2008), Gompers et al. (2008), Nahata (2008), etc.

3 28 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) domestic portfolio companies during covering 35 domestic countries. Second, their model is quite simple. Many related factors of countries, investors, investees and the global market that have been examined in existing studies are not considered. Also, they put all nine country-level factors into a regression at the same time, which may result in multicollinearity. Lastly, most of the country-level factors they estimate are captured by the economic freedom variable in our paper. We have done a robustness study on the decomposed effects of economic freedom (Section 5.3) and our findings confirm some of their results. Our study is also related to studies on economic freedom. Economic freedom is widely investigated in macroeconomic analyses, especially those on long-term economic development across countries. The general equilibrium theory shows that economic efficiency can be achieved in the Arrow Debreu world, where the market is complete, perfect and competitive. Economic freedom measures the degree to which an economy is consistent with these underlying assumptions. Empirical studies have found that economic freedom indeed contributes significantly to economic growth. In a study of 98 countries for the period , Barro (1991) suggests that economic growth is inversely related to market distortions. Barro (1996) further finds that economic freedom has a favorable impact on economic growth. Using an augmented Solow model with crosssectional data, Vanssay and Spindler (1994) also find that economic freedom has a significant and substantial effect on economic growth. Over the last decade or so, a number of indicators on economic freedom have become available and they have subsequently been applied to empirical growth models. Wu and Davis (1999) identify the role of economic freedom in enhancing economic growth after controlling for political freedom. In contrast, Gwartney et al. (1999) and Haan and Sturm (2000) find that the level of economic freedom does not contribute significantly to economic growth; instead, a positive change in economic freedom does. Also, Heckelman (2000) suggests that the average level of freedom in a nation, as well as many of the specific underlying components of freedom, precedes economic growth, after addressing the possibility of reverse causality with a series of Granger-causality tests. The existing studies also explore the role of economic freedom in income equality, employment and welfare. Berggren (1999) finds that during the period , the higher the degree of economic freedom, the higher the degree of income equality in a country. Scully (2002) uses a structural model and a reduced-form model to show that economic freedom is beneficial to both economic growth and income equality. Feldmann (2007, 2008) confirms a positive role of economic freedom in employment. Esposto and Zaleski (1999) conclude that the quality of life is positively associated with economic freedom. However, the empirical literature on economic freedom has focused on its long-term macroeconomic consequences. This paper investigates the impact of economic freedom on cross-border VC performance at the VC investment level and the portfolio company level. The theoretical link between economic freedom and cross-border VC performance, as implied by the general equilibrium theory, is that an economically free economy ensures free markets for early stage companies and in turn improves the economic efficiency of resource allocation. More specifically, a more economically free country can provide better growth opportunities, greater chances of success, and more protection with less grabbing from the government for cross-border VCs. Indeed, we find a positive association between a domestic country's economic freedom and cross-border VC performance after controlling for other related factors of domestic countries, portfolio companies, VCs and the global VC market. We contribute to the literature in several ways. First, we investigate the role of a domestic country's economic freedom on cross-border VC performance. This is our main contribution. The literature on economic freedom has focused on long-term effects, typically economic growth, at the macro level. We provide an analysis of economic freedom at the micro level, specifically on individual cross-border VC investments and domestic portfolio companies. Such an analysis has policy implications for fostering cross-border VC investments. Second, our study deepens the understanding of cross-border VC investments around the world and enriches the literature on the internationalization of VC. Studies on cross-border VC investments are rare. With rapid globalization in recent years, cross-border VC investments have become a significant phenomenon. Such investments can have a huge impact on the countries involved in innovations, technology transfers and entrepreneurship. Third, we also document the effects of other country-level factors on cross-border VC performance, especially such factors as economic development, stock market capitalization, stock market performance, legal quality and entrepreneurial activity. Our results also address the influence of many other related factors of portfolio companies, VCs and the global VC market. The remainder of this paper is organized as follows. Section 2 describes the sample and the research methodology. Sections 3 and 4 provide analyses of cross-border VC performance at the VC investment level and at the portfolio company level, respectively. Section 5 conducts extensive robustness tests. Section 6 concludes the paper with a summary of our main results. 2. Sample and methodology 2.1. Sample For cross-border VC investments, we rely primarily on the SDC VentureXpert database from Thomson Financial. 2 We focus on the cross-border VC investments made by foreign VCs in domestic portfolio companies. Here, the VCs and portfolio companies are 2 An advantage of using the SDC VentureXpert database is its comprehensive coverage. This database claims to cover over 2300 VC funds worldwide. According to Kaplan et al. (2002), the SDC VentureXpert database captures around 85% of the financing rounds made by American VCs. Du and Vertinsky (2008) point out that this database also includes most of the venture-related VC deals in Europe and most of the big VC deals in Asia. Aizenman and Kendall (2008) consider this database to be a comprehensive data source for VC deals worldwide, with over 30 years of historical data.

4 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) located in different countries. 3 The primary sample comprises all cross-border VC investments made during in portfolio companies that received their initial VC funding in the same period. We have employed the following sample selection procedure to obtain the final sample of VC investments: (1) we exclude those investments made by angels and buyout funds and concentrate on the investments made by VC funds; (2) we exclude those investments made in the buyout/acquisition stage or an unknown stage of portfolio companies; (3) we focus on the first investment made by each VC firm in a portfolio company, because the performance in later rounds depends heavily on investments in earlier rounds; and (4) we exclude those investments for which we do not have matching data for our regression variables. We ended up with 10,205 cross-border VC investment observations involving 1906 foreign VCs in 6535 domestic portfolio companies. The sample covers 35 domestic countries Argentina, Australia, Austria, Belgium, Brazil, Canada, Chile, China, 4 Denmark, Finland, France, Germany, Hong Kong, India, Indonesia, Ireland, Israel, Italy, Japan, Malaysia, Mexico, the Netherlands, New Zealand, Norway, Philippines, Portugal, Singapore, South Africa, South Korea, Spain, Sweden, Switzerland, Thailand, the U.K. and the U.S. Table 1 presents the distribution of 10,205 VC investments by funding year and development stage of a portfolio company, the distribution of 6535 portfolio companies by country and industry, and the distribution of 1906 VCs by global region and organizational type. The number of observations and the corresponding percentages (in parentheses) are given. As shown, the internet bubble in the late 1990s has affected the timing of cross-border VC investments. The cross-border VC investments peaked in 2000 and then decreased during Around 38.1% of VC investments occurred at the early or seed stage of development of a portfolio company. The U.S. is the most active in VC imports, constituting roughly 40.9% of our sample. Those countries following the U.S., with each constituting just over 3% of our sample, are the UK, Germany, France, Canada, China, and Israel in that order. Our sample spans the following six industries based on the Venture Economics Industry Classification: biotechnology, communication and media, computer-related, medical/health/life science, semiconductors/other electronics, and non-high technology. Similar to the situation in the U.S., the majority of VC-backed portfolio companies belong to high-tech sectors, such as the communications and media sector and the computer-related sector, which constitute about 19.5% and 40.1% of our sample, respectively. As indicated in the table, VCs from Asia, Europe and North America constitute respectively 31.4%, 23.4% and 41.1% of the sample. VCs from other regions constitute only 4.1% of the sample. Following the literature on VC strategies (Bottazzi et al., 2008; Hellmann et al., 2008; Nahata, 2008), we identify four VC firm types: independent VCs, and VCs affiliated with financial institutions, corporations, or governments. For simplicity, we refer to them respectively as traditional VCs, institutional VCs, corporate VCs and government VCs. As presented in the table, around 58.5% of VCs are traditional VCs, whereas institutional VCs, corporate VCs and government VCs constitute respectively 18.3%, 17.6% and 4.7% of the sample Methodology We measure VC performance using the likelihood of a successful exit by the end of 2009, and call this Investment Success. A portfolio company is treated as successful if it went public or was acquired by the end of We conduct analyses of the likelihood and timing of successful exits at both the VC investment level and the portfolio company level. Specifically, we conduct a probit analysis with Investment Success as the dependent variable and a survival analysis using a Cox hazard model with Investment Duration as the dependent variable. The Cox hazard model allows for right-censored data and time-varying variables, and it is a semi-parametric model in which the hazard function is not dependent on a specific distribution of the survival time. A positive (or negative) coefficient indicates that the variable increases (or decreases) the hazard of a successful exit and shortens (or lengthens) the expected duration. We take three alternative measures of a domestic country's economic freedom during the investment period Index of Economic Freedom, Economic Freedom Quartile and Ranking in Economic Freedom as our primary explanatory variables. Following the literature and taking into account the characteristics of cross-border VC markets, we further control for other related factors of domestic countries, portfolio companies, VCs and the global VC market, as well as industry and year fixed effects. Specifically, the domestic country-related factors are GDP per Capital, Market Capitalization, Market Return, Legality, Foreign Country Legality, Same Legal Indicator, VC Deals and Patents. The portfolio company-related factors are Total VC Funding, Valuation Disclosure Indicator, Early Investment Indicator and Distance. The VC firm-related factors are Local VC Ratio, VC Experience, VC Size, VC Size Squared, VC Portfolio Size, Institutional VC Indicator, Corporate VC Indicator, Asia VC Indicator and Europe VC Indicator. The global VC market-related factors are VC Industry Competition, IPO Market Conditions and M&A Market Conditions. The Appendix provides a detailed discussion of our methods and the definitions of all regression variables. Appendix Tables 1 3 describe the components of economic freedom, the measures of variables and the correlation matrix, respectively. 3 Since the SDC VentureXpert database only provides updated information on the location of portfolio companies and VCs, we use such information to identify cross-border VC investment observations, an approach that is consistent with existing studies on cross-border VC investments using public data sources (Balcarcel et al., 2009; Dai et al., 2010; Schertler and Tykvová, 2011; Wang and Wang, 2011; Watson and George, 2010). We do not have historical information for each portfolio company at the time of VC funding and the time of exit. Therefore, we cannot define cross-border VC investments as investments by foreign VCs in domestic portfolio companies at the time of investment. There might be a bias here since Cumming et al. (2009) find that relocation to the U.S. implies much greater returns to Asia-Pacific VCs than investing in companies already based in the U.S. at the time of VC investment using a hand-collected sample of 468 portfolio companies. As suggested by Cumming et al. (2009) that most relocations (all but three in their sample) were moves to the U.S., we have tried to rule out the influence of relocation on our findings by excluding the U.S. portfolio companies in a robustness test. 4 La Porta et al. (1997, 1998) do not provide legality values for China. Considering China's importance in cross-border VC investments, and in line with Cumming et al. (2006), we employ the average German legal origin values for China. These values are somewhat consistent with the legal index estimates for China by Allen et al. (2005) and Durnev and Kim (2005). In a robustness test, after excluding China from our sample, all our results remain unchanged.

5 Table 1 The sample distribution. The sample consists of 10,205 identifiable cross-border VC investments by 1906 VCs in 6535 portfolio companies that received their initial VC funding during and for which relevant data are available. The table presents the distribution of VC investments by funding year and the development stage of portfolio companies, the distribution of VC-backed portfolio companies by country and industry, and the distribution of VCs by global region and type. The number of observations and the corresponding percentages (in parentheses) are reported. VC investments Portfolio companies VCs Year Obs. Stage Obs. Country Obs. Industry Obs. Global region Obs. Type Obs (1.31) Seed/Startup 1191 (11.67) U.S (40.86) Biotechnology 503 (7.70) Asia 599 (31.43) Traditional VC 1114 (58.45) (3.03) Early stage 2698 (26.44) U.K 595 (9.10) Communications and media 1273 (19.48) Europe 445 (23.35) Institutional VC 348 (18.26) (3.52) Expansion stage 5217 (51.12) Germany 323 (4.94) Computer related 2621 (40.11) North America 783 (41.08) Corporate VC 336 (17.63) (5.20) Later stage 1099 (10.77) France 316 (4.84) Medical/health/life science 521 (7.97) Other 79 (4.14) Government VC 89 (4.67) (11.63) Total 10,205 (100) Canada 312 (4.77) Semiconductors/other electronics 562 (8.60) Total 1906 (100) Total 1906 (100) (24.07) China 274 (4.19) Non-high technology 1055 (16.14) (15.69) Israel 219 (3.35) Total 6535 (100) (9.36) India 173 (2.65) (8.42) Sweden 169 (2.59) (10.03) Japan 153 (2.34) (7.74) Other 1331 (20.37) Total 10,205 (100) Total 6535 (100) 30 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) 26 50

6 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) We avoid an investigation using aggregate data at the country level, which is often subject to the simultaneous critique. It is much more sensible to examine data at the VC investment level and the portfolio company level where the environment is constant and much finer controls are available. For the analysis of VC investments, we focus only on each VC firm at the time of its first investment in a portfolio company, with a sample size of 10,205; and likewise, for the analysis of portfolio companies, we focus only on each portfolio company at the time of its receiving the initial VC funding, with a sample size of The reason for this treatment is that there is typically a substantial time gap between a VC firm's first investment and its exit, as well as between a portfolio company's first VC funding and its exit. This methodology can effectively address concerns regarding reverse causality where better cross-border VC performance can conversely lead to more economic freedom in the domestic country. Similar approaches have been adopted by Hochberg et al. (2007), Nahata (2008), etc. In the Cox hazard analysis, we further allow for time variation of the variables Index of Economic Freedom, Economic Freedom Quartile, Ranking in Economic Freedom, GDP per Capita, Market Capitalization, Market Return, VC Deals, Patents, IPO Market Conditions and M&A Market Conditions to alleviate the problem of measuring unsuccessful exits. The hazard framework has the added advantage of being able to capture the market conditions at the time of a domestic country's macroeconomic development and exit to determine their impact on the likelihood of a successful exit, for which a simple probit model cannot do. In summary, our basic regression models are: The probit model: Investment Success ¼ α 0 þ α 1 Economic Freedom þ α 2 Country related f actors þ α 3 Portf olio company related f actors þ α 4 VC firm related f actors þ α 5 Global VC market related f actors þ Industry dummies þ Year dummies: ð1þ The Cox hazard model: Investment Duration ¼ β 1 Economic Freedom þ β 2 Country related f actors þ β 3 Portf olio company related f actors þ β 4 VC firm related f actors þ β 5 Global VC market related f actors þ Industry dummies þ Year dummies: ð2þ Here, Economic Freedom is either Index of Economic Freedom, Economic Freedom Quartile or Ranking in Economic Freedom Summary statistics and univariate analysis Table 2 presents the summary statistics at the VC investment level. 5 The quartiles, means, standard deviations and the number of observations are presented. We have very similar data distributions at the portfolio company level (not reported). Only 15.8% of the cross-border VC investments exit successfully through IPOs or acquisitions, a level that is much lower than the 33.5% reported by Nahata (2008) and the 41.8% reported by Cochrane (2005) for U.S. domestic VC investments. The average Investment Duration is 2826 days with a median of 3072 days. 6 The means (medians) of Index of Economic Freedom, Economic Freedom Quartile and Ranking in Economic Freedom are respectively (78.57), 2.47 (2) and 0.65 (0.79). The means (medians) of GDP per Capita, Market Capitalization, Market Return, Legality, Foreign Country Legality, VC Deals and Patents are respectively 33,593 (36,909) US dollars, % (127.2%), 1.9% (1.8%), (16.45), (16.08), 8.43 (9.95) per million people and 0.42 (0.48) per thousand people. Around 48.2% of domestic countries' legal systems originate from the same tradition as that of the corresponding foreign countries. The total VC funding across all financing rounds in a portfolio company is million US dollars on average with a median of about million US dollars. About 29.1% of portfolio companies voluntarily disclose their market valuation after VCs' first investment in them, which is slightly lower than the 34.4% reported by Cumming and Dai (2011) for the U.S. VC investments made during Further, 38.1% of first investments by a VC firm in a portfolio company occur at the seed or early stage of development, which is much lower than the 63.7% reported by Nahata (2008) andthe62%reportedbygompers et al. (2010) for the U.S. VC industry. The average geographical distance between a portfolio company and a VC firm is 7205 km with a median of 7837 km. On average, 34.4% of VCs participating in the syndicate of a portfolio company are domestic VCs. The average VC age is about years old at the time of its first investment in a portfolio company with a median of about 8 years old. The aggregate amount of funds raised by a VC firm over the ten years prior to its first investment in a portfolio company is about 919 million US dollars with a median of only 143 million US dollars. The mean (median) number of companies in a VC firm's portfolio is (44). The reported portfolio size is much larger than that suggested by Cumming (2006) since our unit of analysis is VC firm instead of VC fund. Around 15.4% and 16.8% of VCs are respectively institutional and corporate VCs; these percentages are slightly larger than 5 We report the original values for all the variables in Table 2, even though we have taken logarithms of some variables in our regressions (as indicated by ** in Appendix Table 2). 6 The average Investment Duration of VC investments is only 1239 days for successful exits and is 2826 days for all VC investments (taking into account the fact that we do not know the exact time of unsuccessful exits and that Investment Duration of unsuccessful exits is right-censored at the end of 2009).

7 32 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) Table 2 Summary statistics and univariate analysis. This table presents the statistics of 10,205 cross-border VC investments during in those portfolio companies that received initial VC funding during The definitions, measures and data sources of the variables are described in Appendix Table 2. The quartiles, means, standard deviations and the number of observations are presented. The results of the univariate analysis are also reported, which are from the t-tests for equality of the means and Wilcoxon tests for equality of the medians between successful and unsuccessful exits. The significance levels at the 1%, 5% and 10% are identified by ***, ** and *, respectively. Variable Obs. Mean Std Successful exits Unsuccessful exits Mean Median Mean Median Investment success 10, Investment duration (days) 10, Index of economic freedom 10, *** 78.8*** Economic freedom quartile 10, *** 3*** Ranking in economic freedom 10, *** 0.829*** Domestic country-related factors GDP per capita (US$) 10,205 33,593 12,397 31,988 36,909 40,815 34,002 36,381*** 33,517 37,280 Market capitalization (%) 10, *** *** Market return 10, Legality 10, *** *** Foreign country legality 10, *** 15.95*** Same legal origin indicator 10, VC deals (per million people) 10, *** *** Patents (per thousand people) 10, ** 0.484*** Portfolio company-related factors Total VC funding (US$mil) 10, *** *** Valuation disclosure indicator 10, *** 1*** Early investment indicator 10, *** 0*** Distance (kilometers) 10, *** 8409*** VC firm-related factors Local VC ratio 10, *** 0.667*** VC experience (years) 10, *** 6*** VC size (US$mil) 10, *** 100*** VC portfolio size 10, *** 35*** Institutional VC indicator 10, ** 0** Corporate VC indicator 10, *** 0*** Asia VC indicator 10, *** 1*** Europe VC indicator 10, Global VC market-related factors VC industry competition (US$bil) 10, * * IPO market conditions 10, *** 16*** M&A market conditions 10, *** 103*** those reported by Bottazzi et al. (2008) for European VCs and by Nahata (2008) for the U.S. VCs. As shown, about 37.8% and 25.7% of the VCs are respectively located in Asia and Europe. 7 The aggregate inflow of VC funds in the world in the year prior to a VC firm's first investment is around billion US dollars with a median of about billion US dollars. Also, there are on average (with a median of 13.79) VC-backed IPOs and (with a median of ) M&A transactions with VC-backed portfolio companies as the targets in the quarter prior to the exit of a portfolio company. In Table 2, we also report the results of the univariate analysis the t-tests for the equality of means and Wilcoxon tests for the equality of medians between successful exits and unsuccessful exits. As shown in the table, successful exits are more likely in domestic countries with higher economic freedom as measured by Index of Economic Freedom, Economic Freedom Quartile or Ranking in Economic Freedom at the 1% significance level. Also, as both the t-tests and Wilcoxon tests indicate strong significant associations between most control variables and the likelihood of successful exits, there is the need to control for these factors. 3. Analysis at the VC investment level In this section, we investigate the impact of economic freedom on the performance of cross-border VC investments. The unit of analysis is VC investment. Our sample consists of 10,205 unique VC firm-portfolio company paired observations. Table 3 reports the results. All the z-statistics (in parentheses) are adjusted for heteroskedasticity (White, 1980) and clustered by portfolio company (Petersen, 2009) to account for the correlation among multiple VC investments in the same portfolio company. We first present the results of the probit analysis in models 1 3 of Table 3 with Investment Success as the dependent variable using Eq. (1). Our model fits the data well, with a high pseudo R-squared of around 0.5. We find that each of the three explanatory variables Index of Economic Freedom, Economic Freedom Quartile and Ranking in Economic Freedom has a significant impact, that is, a domestic country's economic freedom has a significant positive effect on Investment Success. The coefficients of Index of Economic 7 Since the summary statistics in Table 2 corresponds to 10,205 VC firm-portfolio company paired observations, the means of Institutional VC Indicator, Corporate VC Indicator, Asia VC Indicator and Europe VC Indicator are slightly different from the distribution of VCs reported in Table 1, which corresponds to 1906 individual VCs.

8 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) Table 3 Probit and Cox hazard analyses at the VC investment level. The sample consists of 10,205 cross-border VC investments during in those portfolio companies that received their initial VC funding during Models 1 3 and 4 6 present probit and Cox hazard regression results using Eqs. (1) and (2), respectively. The definitions, measures and data sources of the variables are described in Appendix Table 2. The robust z-values, cluster-adjusted by portfolio company, are in parentheses. The significance levels at the 1%, 5% and 10% are identified by ***, ** and *, respectively. Variable Probit Cox hazard Index of economic freedom 0.036** (2.15) 0.006* (1.94) Economic freedom quartile 0.569*** (5.76) 0.106*** (7.45) Ranking in economic freedom 1.760** (2.49) 0.477*** (3.61) Domestic country-related factors GDP per capita 0.280*** ( 16.46) 0.328*** ( 13.29) 0.280*** ( 16.62) 0.043*** ( 24.19) 0.049*** ( 26.10) 0.044*** ( 23.62) Market capitalization (0.02) ( 0.45) ( 0.15) (1.46) (1.38) (0.66) Market return 2.898* (1.92) 3.215** (2.15) 3.015** (2.01) (1.35) (1.60) (1.45) Legality 1.270*** (12.76) 1.377*** (12.65) 1.198*** (10.94) 1.398*** (9.60) 1.377*** (10.69) 1.198*** (7.55) Foreign country legality 0.048** ( 2.37) 0.034* ( 1.69) 0.046** ( 2.26) ( 1.54) ( 1.16) ( 1.48) Same legal origin indicator 0.117* ( 1.73) 0.121* ( 1.74) 0.137** ( 2.02) 0.164** ( 2.08) 0.162** ( 2.04) 0.215*** ( 2.66) VC deals 0.208*** (15.31) 0.263*** (11.61) 0.204*** (15.25) 0.033*** (22.14) 0.039*** (20.74) 0.033*** (21.94) Patents 1.909*** (9.33) 2.557*** (8.57) 1.881*** (9.53) 0.328*** (11.05) 0.421*** (12.46) 0.328*** (10.92) Portfolio company-related factors Total VC funding 0.111*** (4.30) 0.098*** (3.95) 0.110*** (4.29) 0.233*** (4.82) 0.220*** (4.57) 0.236*** (4.90) Valuation disclosure indicator 0.245*** (3.61) 0.204*** (3.06) 0.235*** (3.46) 0.230** (2.57) 0.195** (2.24) 0.227** (2.53) Early investment indicator 0.275*** ( 4.18) 0.237*** ( 3.52) 0.270*** ( 4.11) 0.541*** ( 7.35) 0.502*** ( 6.82) 0.538*** ( 7.32) Distance 0.083** (2.25) (0.38) 0.075* (1.96) 0.105** (2.24) (0.57) (1.54) VC firm-related factors Local VC ratio 0.861*** (6.69) 0.772*** (6.01) 0.847*** (6.60) 0.812*** (4.58) 0.672*** (3.71) 0.780*** (4.42) VC experience 0.050** (2.33) 0.044** (2.02) 0.049** (2.28) (1.21) (0.91) (1.13) VC size (0.90) (0.67) (0.89) 0.057* (1.95) (1.61) 0.058** (1.97) VC size squared ( 0.09) (0.15) ( 0.03) ( 1.41) ( 1.06) ( 1.32) VC portfolio size 0.086*** ( 4.46) 0.069*** ( 3.58) 0.084*** ( 4.34) 0.090*** ( 4.15) 0.076*** ( 3.45) 0.089*** ( 4.14) Institutional VC indicator 0.119** (2.09) 0.098* (1.69) 0.114** (2.00) (1.38) (0.94) (1.23) Corporate VC indicator 0.163*** (2.80) 0.151** (2.57) 0.163*** (2.79) 0.159** (2.23) 0.139** (2.00) 0.166** (2.34) Asia VC indicator (0.32) ( 0.67) (0.24) (0.37) ( 0.34) (0.39) Europe VC indicator (1.11) (0.11) (0.92) (0.65) ( 0.07) (0.48) Global VC market-related factors VC industry competition 0.330*** ( 6.78) 0.335*** ( 7.01) 0.323*** ( 6.65) 0.171* ( 1.83) 0.242*** ( 2.77) 0.163* ( 1.74) IPO market conditions 0.582*** ( 4.77) 0.611*** ( 4.95) 0.595*** ( 4.87) 0.027** ( 2.56) 0.030*** ( 3.24) 0.027*** ( 2.60) M&A market conditions 2.118*** (5.56) 1.847*** (4.52) 2.183*** (5.73) 0.203*** (4.44) 0.146*** (3.15) 0.204*** (4.52) Industry and year fixed effects Present Present Present Present Present Present Number of observations 10,205 10,205 10,205 10,199 10,199 10,199 Pseudo R

9 34 L. Wang, S. Wang / Journal of Empirical Finance 19 (2012) Freedom, Economic Freedom Quartile and Ranking in Economic Freedom are respectively 0.036, and 1.760, with the z-statistics respectively at the 5%, 1% and 5% significance levels. For the marginal effects, a one standard deviation increase in Index of Economic Freedom, Economic Freedom Quartile or Ranking in Economic Freedom at the mean level increases the likelihood of a successful exit by 2.02%, 2.76% or 3.02%, respectively. Some other variables also affect the performance of VC investments, mostly as expected. First, GDP per Capita is negatively associated with the likelihood of a successful exit, which is consistent with the well-known convergence hypothesis in the exogenous growth literature (Barro, 1991, 1996). Market Return is found to be positively associated with the likelihood of a successful exit, which is consistent with the studies on the impact of stock market performance on IPO volume (Loughran et al., 1994; Lowry, 2003). Legality is found to be significantly positively related to VC performance, which is consistent with Armour and Cumming (2006), Cumming et al. (2006) and Cumming and Walz (2010). The higher legality is, the more protected an investor's legal rights and the higher the likelihood of a successful exit. We also find that the coefficients of Foreign Country Legality and Same Legal Origin Indicator are typically negative, suggesting that the legal qualities of a domestic country and a foreign country complement each other. Also, we find a strong positive association between a domestic country's entrepreneurial activity and the likelihood of a successful exit. The greater the number of VC deals or patents granted in a domestic country, the higher the likelihood of a successful exit. However, the relationship between a domestic country's stock market capitalization and the likelihood of a successful exit is typically insignificant, which is inconsistent with the findings of Black and Gilson (1998) and Cumming (2008), but consistent with those of Cumming et al. (2006), who show that a country's stock market capitalization has no significant impact on exit choices in 12 Asia-Pacific countries. Second, our results on the coefficients of portfolio company-related factors are quite consistent with our expectations. Total VC Funding and Valuation Disclosure Indicator positively affect cross-border VC performance at the 1% significance level. The better the quality of a portfolio company, the larger the VC investment amount, the higher the likelihood of a voluntary disclosure of postinvestment valuation, and the higher the likelihood of a successful exit. Also, Early Stage Indicator is found to be strongly negatively associated with the likelihood of a successful exit at the 1% significance level. Unexpectedly, the variable Distance is found to be positively associated with the likelihood of a successful exit, but at a low significance level. In later estimations, we generally find insignificant results on Distance at the portfolio company level. Later in Section 5.2, after a robustness check of the exit choices between an IPO and an acquisition, we further find a different role of Distance in exits through IPOs than in exits through acquisitions. 8 Third, Local VC Ratio is positively related to the likelihood of a successful exit at the 1% significance level. This is consistent with the value-added view of domestic investors on cross-border VC investments (Dai et al., 2010; Mäkelä and Maula, 2008; Tykvová and Schertler, 2011). VC Experience is found to be positively related to the likelihood of a successful exit at the 5% significance level, which is consistent with the findings for the U.S. domestic market, such as those by Sorensen (2007), Gompers et al. (2008) and Nahata (2008). We find a significantly negative relationship between VC Portfolio Size and the likelihood of a successful exit, which is quite consistent with the findings of Kanniainen and Keuschnigg (2003, 2004) and Cumming (2006). Also, Institutional VC Indicator and Corporate VC Indicator have positive effects on Investment Success, due possibly to the value-added services offered by VCs to portfolio companies. As shown, there is no statistically significant association between VC Size or VC Size Squared and the likelihood of a successful exit. Also, neither Asia VC Indicator nor Europe VC Indicator has a significant effect on the likelihood of a successful exit. Fourth, the global VC market-related factor VC Industry Competition is negatively associated with Investment Success at the 1% significance level. The greater the worldwide inflow of VC funds, the more severe VC competition is in the international market and the less likely it is for a successful exit to occur. However, the global market conditions affect the likelihood of a successful exit in different ways. Specifically, IPO Market Conditions is negatively related to the probability of a successful exit, contrary to our expectations. M&A Market Conditions is positively related to the probability of a successful exit at the 1% significance level, consistent with our expectations. Later in Section 5.2, after a robustness check of the exit choices between an IPO and an M&A, we find that the negative coefficient of IPO Market Conditions arises from its other role in the choice between an IPO and an M&A. We next turn to the Cox hazard analysis in models 4 6 of Table 3 with Investment Duration as the dependent variable using Eq. (2). We find that most results mirror the evidence reported in models 1 3. Taking the right-censored nature of the data into account, we still find that the coefficients of Index of Economic Freedom, Economic Freedom Quartile and Ranking in Economic Freedom are significantly positive. Specifically, a one standard deviation increase in Index of Economic Freedom, Economic Freedom Quartile or Ranking in Economic Freedom is associated with an increase of 5.35%, 12.59% or 13.69% respectively in the hazard of a successful exit. The results on the control variables are very similar to our earlier results with some changes in the significance level. Specifically, GDP per Capita, Same Legal Origin Indicator, Early Investment Indicator, VC Portfolio Size, VC Industry Competition and IPO Market Conditions remain negatively associated with the hazard of a successful exit, while Legality, VC Deals, Patents, Total VC Funding, Valuation Disclosure Indicator, Local VC Ratio, Corporate VC Indicator and M&A Market conditions remain significantly positively associated with the hazard of a successful exit. Further, we have stronger evidence on the impact of the size of a VC firm, with a positive association between VC Size and the hazard of a successful exit. However, the effects of Market Return, Foreign Country Legality, VC 8 We have tried some alternative measures of Distance, including the geographical distance between the foreign country and the domestic country, the flight distance between the VC firm and the portfolio company, etc. The coefficients of Distance are generally positive but not very significant. However, if we define successful exits as IPO exits only, as Tian (2011) does, we find a generally negative effect of Distance on the likelihood and hazard of IPO exits. By comparing our results with the two related studies focusing on the influence of the geographical distance on VC investments (Chemmanur et al., 2011; Tian, 2011), we conclude that the coefficient of Distance is generally found to be positive in our primary regressions because Distance has a different effect on exits through IPOs than on exits through acquisitions. Also, the inclusion of geographical distance may suffer from an endogeneity problem due to omitted variables. If we exclude the variable Distance, our results remain very much unchanged. Since Distance is not the primary focus of our paper, we have not conducted a thorough investigation of it. It would be interesting and worth exploring in the future.