Cross-border Buyout Performance

Transcription

1 Cross-border Buyout Performance Siyang Tian 30 April 2017 Abstract This study examines the determinants of the cross-border buyout performance, with a focus on country-specific institutional, cultural and learning factors. I find that the probability of a successful exit increases when the country-specific institutional environment of the private equity (PE) s portfolio company is of higher quality. In addition, PE firms do not suffer from the adverse effect of cultural differences in buyout investments. Further, PE firms international experience, industrial sophistication, and reputation help them overcome institutional barriers and improve buyout success. Additional analyses examining exit strategies of PE s portfolio companies suggest that PE group syndication increases the probability of choosing initial public offering as an exit route while institutional factors and PE experience are associated with the choice of mergers and acquisitions. Keywords: Cross-border buyout, Institutions, Learning Siyang Tian is at the Cass Business School, City, University of London, 106 Bunhill Row, London EC1Y 8TZ, United Kingdom. Telephone: +44 (0) siyang.tian.2@cass.city.ac.uk (Tian). I gratefully acknowledge the helpful comments from Cyril Auger, Erwan Morellec, Daniela Fabbri, Sonia Falconieri, Meziane Lasfer, Paolo Volpin, Scott Moeller, Anh Tran, and seminar participants in Cass Mergers and Acquisitions Research Center Advisory Board Meeting and Cass PhD Research Day Event.

2 1. Introduction A leveraged buyout (LBO), an important part of private equity (PE) investments, is an acquisition of a company financed with a substantial portion of borrowed funds. During a typical PE investment, PE firms (1) improve the portfolio company s value by conducting financial, governance, and operational engineering (Kaplan and Stromberg, 2008), (2) divest portfolio companies as PE funds have limited contractual lifetime, and (3) return capitals to their investors. Cross-border PE investments have become a phenomenon since the late 1990s (Bottazzi, Da Rin, and Hellmann, 2004; Wright, Pruthi, and Lockett, 2005; Maula, 2010; Cao, Cumming, Qian and Wang, 2015). The previous literature on global PE investments mainly either focuses on early stage venture capital (VC) s cross-border performance (Nahata, Hazarika and Tandon, 2014) or compares PE performance in domestic markets within different countries (Cumming, Fleming and Schwienbacher, 2006; Cumming, 2008; Stromberg, 2008) while remaining limited on the factors predicting a cross-border buyout s eventual successful exit and their impact on specific exit strategies. In this study, I attempt to fill this gap in the PE literature by examining the determinants of cross-border buyout performance. Lerner, Sorensen and Stromberg (2009) show that LBO outcomes and exit ratios (both domestic and cross-border deals) vary across different countries with the highest exit ratio in North American market and the lowest in developing Asian markets. Their evidence suggests that there are important country-specific factors facilitating successful buyout exits. La Porta, Lopez-de-Silanes, Shleifer and Vishny (1998) study how investor protection, legal rules, and contract enforcement affect capital market development and financing patterns. Their research offers a law and finance framework to explain international differences in financial transactions. Rossi and Volpin (2004) show that legal factors serve as the key determinants in cross-border merger and acquisition transactions. Further, when PE firms invest abroad, they face unfamiliar environment and lack of awareness of local cultural and social norms and 2

3 therefore the information asymmetry problem could be severe. The cultural distance can thus adversely affect PE s investment performance. However, PE firms could accumulate business, institutional and internationalization knowledge in their ongoing activities and overcome such institutional barriers. In this paper, I analyse the effects of institutional, cultural and organisational learning factors on the likelihood of PE firms bringing their portfolio companies to successful exits. Following Hochberg, Ljungqvist and Lu (2007) and Nahata et al. (2014), I define the PE s portfolio company exit to be successful if it is later brought to the market through an initial public offering (IPO) or acquired by another company. I then examine the probability of a successful exit and the time-to-successful exit. Finally I study the impact of the above factors on the choice between IPO and mergers and acquisitions (M&A) as exit routes. Using a sample 2,665 cross-border buyout investments in 40 countries between 1998 and 2007, I find that the likelihood of a successful exit increases when the quality of the portfolio company country s institutional environment is higher. I measure this country-specific institutional environment quality using a composite index Country Index which captures the political risk (including creditor rights, legal system and investor protection), economic stability and financial stability. In addition, cultural distance between PE firms and their portfolio companies has no effect on cross-border buyout performance. Further, I find that the PE firm s experience is positively related to the likelihood of the exit success. In terms of exit strategies, I find evidence that the probability of going IPO increases when PE firms form a club in the cross-border buyout and the probability of choosing M&A increases when the quality of the institutional environment is higher and when PE firms are more experienced. The results are robust to the self-selection issue in which the performance of PE firms may be attributable to the quality of their portfolio companies rather than PE firms experience and reputation (Nahata, 2008). I follow Nahata (2008) and adopt a variation of Heckman s (1979) 3

4 endogeneity correction procedure. In the first stage, I estimate the reputable PE firms likelihood of investing in a portfolio company. In the second stage, I include the inverse Mills ratio received from the first-step probit model as an additional control variable to estimate the buyout performance. This study contributes to the PE literature in several ways. First, this paper is one of the few studies that examines LBO investments across countries. Most PE studies focus on the U.S. and European markets while this paper covers not only developed markets but also emerging markets such as China and India. Axelson, Jenkinson, Stromberg and Weisbach (2013) and Cao et al. (2015) study LBO activities across countries but they examine issues at the stage of entry including leverage, pricing, deal frequency, and formation of club deals. This paper investigates the performance at the stage of exit which completes the investment process. Second, this paper studies the cross-border buyout and previous papers mainly compare domestic PE behaviour within different countries. Moreover, this paper shows how the country-specific institutional environment (political risk, legal system, economic stability and financial stability) influences PE s cross-border buyout performance. Such cross-country analyses of law and institution on the financial intermediation have garnered popularity in many empirical studies (La Porta et al., 1998; Rossi and Volpin, 2004; Lerner and Schoar, 2005; Cumming, Fleming and Schwienbacher, 2006; Bris and Cabolis, 2008; Nahata et al., 2014; Cao et al., 2015). Lastly, to alleviate common data limitations at the deal-level in prior PE studies, I construct the sample by extracting portfolio companies details of 1,008 PE firms around the world and obtain a sample of 2,665 cross-border buyouts. The rest of the paper is structured as follows. Section 2 provides a review of the literature, the institutional settings, and the hypotheses. Section 3 presents the data collection procedure and the construction of all variables. Section 4 discusses the empirical analyses. Section 5 4

5 concludes. 2. Literature Review and Hypothesis Development 2.1 Literature review The literature on PE performance can be categorised into two groups: fund-level studies and deal-level studies. Metrick and Yasuda (2011) outline both advantages and limitations for these studies. The net of fund fees and carry could be calculated at the fund level, however, there is missing information about timing and exits of individual projects. Also, investment write-offs which incur losses are not observed at the fund level. In contrast, deal-level data could alleviate the selection bias problem as the outcome of unsuccessful investments could be tracked. Nevertheless, deal-level studies suffer from data incompleteness (Kaplan, Sensoy and Stromberg, 2002) and a novel data set or a model which could overcome the data problem is thus required. Kaplan and Schoar (2005) report that the average of net of fees fund returns is closed to the S&P 500 index but the fund heterogeneity is strong. They also find that returns are persistent across subsequent funds of a partnership and these partnerships tend to raise more follow-on funds. Phalippou and Gottschalg (2009) make several adjustments to the measure of returns including fund maturity adjustment, performance weight adjustment and risk adjustment. In contrast to Kaplan and Schoar (2005), they claim that average net of fees fund performance underperforms the S&P 500 Index by 3% per year and gross net of fees fund performance over-performs by 3% per year. Driessen, Lin and Phalippou (2012) develop and apply a generalised methods of moments (GMM) model to estimate the market beta for funds. They report a low beta and mixed evidence on the buyout fund performance. In terms of deallevel performance studies, Groh and Gottschalg (2008) compare buyout returns to a mimicking portfolio of equally risk levered investments in the S&P 500 Index and report a positive and 5

6 statistically significant alpha of buyout funds. Lopez-de-Silanes, Phalippou and Gottschalg (2015) construct a dataset of 7,500 investments based on the private placement memorandums. They report that 10% of the deals end up with bankruptcy and 25% of the deals achieve an internal rate of return (IRR) of more than 50%. Further, they conclude the diseconomies of scale in PE activities. Several deal-level studies shed lights on real effects (productivity and efficiency). Lichtenberg and Siegel (1990) report that management buyout plants achieve higher total factors productivity than the comparable group in the same industry before the buyout and improve the productivity after the buyout. Alperovych, Amess and Wright (2013) claim that buyout vendor source (divisional buyout) and PE firms experience exert a positive influence on the post-buyout efficiency during the first three years after the buyout. There are also studies investigating cross-border VC and PE investments. Maula (2010) summarises the existing literature and outlines the determinants of cross-border VC and PE volume, including human capital, patents, creditor rights and legal framework, supportive exit markets including M&A market and IPO markets. The frictions include distance, different currency and language, lack of information and trust and lack of regulatory environment. However, few studies shed lights on the exit performance of cross-border investments. Jaaskelainen and Maula (2008) report that direct and indirect network ties of VC investors in foreign markets influence the likelihood of divesting the portfolio company via trade sale or IPO. Dai, Jo and Kassicieh (2012) explore cross-border VC exit performance in six Asian countries and claim that the partnership with a local VC firm helps to alleviate the information asymmetry and monitoring problems and has a positive influence on the likelihood of a successful exit. Nahata et al. (2014) find that superior legal rights and better-developed stock market significantly enhance cross-border VC performance. Remarkably, they find that cultural differences increase the propensity of a successful exit and they argue that this is because VC 6

7 managers recognise cultural differences and conduct the due diligence and screening more carefully. Chemmanur, Hull and Krishnan (2014) adopt the quasi-natural experiment methodology to study cross-border LBO investment performance of U.S. PE firms. They exploit the exogenous shock to the effective proximity of U.S. PE investors to other countries due to the ease of travel afforded by the open sky agreement between the U.S. and countries of PE s portfolio companies. They find that the presence of the open sky agreement is positively related to the likelihood of successful exits. This paper differs from previous literature in several perspectives. Most studies look into determinants of VC investment performance in a single geographic market such as the U.S. market. I exam how cross-country differences affect the exit performance of cross-border buyouts in both developed and developing markets. This paper uses the deal-level data and each observation is a cross-border buyout transaction. To resolve the data incompleteness problem, I construct a sample by tracking the exit outcomes of 1,008 PE firms portfolio companies. Finally, this paper not only considers the cross-border investments of U.S. PE investors but also the non-u.s. investors as non-u.s. investors become increasingly important overtime (Axelson and Matinovic, 2013). 2.2 Hypothesis Development Gompers and Lerner (2004) state that there are little theoretical attention being paid to the divestment aspect of PE and therefore limited established theories could explain the sale of portfolio companies. In cross-country studies, the effect of law and institution on investments is well documented (Rossi and Volpin, 2004; Cumming et al., 2006; Bris and Cabolis, 2008; Nahata et al., 2014; Cao et al., 2015). North (1990) defines the institution as the rule of the game in a society and humanly devised constraints that shape human interaction. The institution emphasizes the role of information asymmetry and transaction cost in economic 7

8 activities and the key function of the institution is to reduce the uncertainty by establishing a stable structure to facilitate interaction among people. Williamsons (2000) establishes four hierarchies of the institution and the first two levels are country-specific and vary across different countries. The first level is the informal institution which comes from socially transmitted information and is part of the culture (North, 1990). The second level is the formal constraints, including constitutions, contract laws and enforcement of property rights. Institutional differences result in different level of information asymmetry, transaction cost and complexity. This framework could thus help to explain why cross-border buyout performance varies across different countries Formal Institutions Formal rules contain political and legal rules, economics rules and contracts. The purpose of these rules is to facilitate political or economic exchange (North, 1990). There are two contrasting views on effects of law and institution on financial transactions. Under the law matters view, La Porta et al. (1997, 1998) show that legal system exerts a positive influence on investor protection and capital market development. Common law countries have the strongest investor protection and well-developed capital markets while French civil law countries have the weakest investor protection and less-developed capital markets, with German and Scandinavian civil laws countries in the middle. Glaeser, Johnson and Shleifer (2001) state that appropriate laws and regulations and high enforcement of shareholder and creditor rights are instrumental to build up the market confidence and attract investments. The law and institution could affect the PE exit in several ways (Cumming et al., 2006). Firstly, all else being equal, in the country with stronger legal protection and well-developed institutional system, the capital market will be more active and cross-border PE firms have more opportunities to divest their portfolios. Further, the exit decision will be influenced by the information asymmetry between the seller and buyer (Neus and Walz, 2004) and the transaction 8

9 complexity. In the country with high institutional environment quality, information asymmetry and transaction cost could be reduced as there are stronger legal protection and contract enforcement, facilitating the exit process. Under the Coasian view, legal and institutional differences do not matter as sophisticated investors could privately negotiate and optimize the contract to mitigate the legal impediments (Bergman and Nicolaievsky, 2007). For example, Allen, Qian and Qian (2005) state that the emerging economy China has underdeveloped legal and financial system. However, the legal impediments do not prohibit China s fast growth. The impact of law and institution on PE investor behavior has been documented by several scholars. In terms of contract design, Lerner and Schoar (2005) study how the legal system affects the financial contract. They report that PE investors in high enforcement and common law nations are more likely to use the convertible preferred stock. By contrast, PE firms incline to use common stock and debt and rely more on equity and board control in low enforcement and civil law nations. Similar results could also be found in Bottazzi et al. (2009) and Cumming, Schmidt and Walz (2010). However, Kaplan, Martel and Stromberg (2007) show that legal regime does not matter and more experienced VC investors adopt U.S. style sophisticated contract. Further, Cao et al. (2015) report that the level of creditor rights in a certain country is positively related to the prosperity of LBO activities. Regarding VC exits, Cumming et al. (2006) and Nahata et al. (2014) find that the quality of legal system is positively related to the exit success. Compared to previous studies which focus the legal system and origin, contracts enforcement, and creditor rights separately, I use the time-varying country composite index from International Country Risk Guide database to capture the impact of institutional environment quality on the exit success. This index takes the political risk, economic stability 9

10 and financial stability into consideration and has been used by several studies to proxy the institutional environment quality (Gelos and Wei, 2005; Humphery-Jenner, Sautner and Suchard, 2016). The higher the index score, the lower the risk of the country and the better the institutional environment. As an additional test, I follow Berkowitz, Pistor and Richard (2003), construct the legality index and measure the impact of legal system quality on the cross-border buyout success. In brief, in the country with the high quality of institutional environment, there will be stronger investor protection and contract enforcement, less political, economic and financial uncertainty, buyout specialists could thus facilitate the exit process. On the other hand, PE investors are sophisticated and could avoid the institutional impediments so that the impact of law and institution is minimal. Consequently, the first hypothesis is: Hypothesis 1: Higher country index is associated with a higher likelihood of a successful exit Informal Institutions In Williamson s framework, the informal institution is referred to the culture. Culture could exert influences on economic activities as it shapes economic individuals choices and perceptions (Hofstede and Bond, 1988). The influence of cultural differences on cross-border investments has been examined in recent studies. Guiso, Sapienza and Zingales (2009) investigate the role of trust origins in determining of capital flows between countries. Ahern, Daminelli and Fracassi (2015) report that the cultural distance adversely affects the crossborder mergers volume and combined announced returns, after controlling country and year fixed effects, and economic, legal, financial and deal-level factors. They argue that different cultural values could lead to impediments such as mistrust, misunderstanding or mismatched goals in cooperation. Nahata et al. (2014) examine the impact of cultural differences on the outcomes of VC investments and they find that cultural distance actually increases the VC 10

11 success. They argue that the cultural disparity between VC firms and their portfolio companies leads to more rigorous due diligence and deal selection, and hence improves the performance. The influence of cultural differences on the outcome of buyout investments has not been seriously examined. Cultural differences between PE firms and their portfolio companies could adversely affect the cross-border buyout performance. Deal negotiation, contract negotiation, corporate policy design, and working relationship development could be affected by several cultural factors such as individualism, uncertainty avoidance and gender equality. The differences arising from cultural diversity could lead to conflicts and investment failures. On the other hand, in the buyout transaction, PE firms usually fully acquire the portfolio company and PE firms targets are mature companies in the late development stage which could generate stable cash flow to meet the debt repayment requirement (Jensen, 1989). Also, buyout investors are sophisticated. To add firm value, they restructure the capital structure, replace the management team with industrial experts and guide the operational change. In comparison, VC firms invest in early-stage and start-up companies. Understanding of business ideas and entrepreneurs is essential to VC investments (Kaplan and Stromberg, 2004). Consequently, buyout investors might not suffer from the adverse influence of cultural differences because they rely more on the hard information while the VC investors could be subject to cultural differences as they rely more on the soft information. The second hypothesis is: Hypothesis 2: Cultural differences adversely influence the cross-border buyout performance Learning Under the experience matters view, Meuleman and Wright (2011) claim that PE firms may be able to reduce institutional barriers through learning. Experiential learning is the process in which firms accumulate knowledge about a certain market through their ongoing activities. De Clercq and Dimov (2007) argue that PE firms obtain knowledge about local 11

12 business and institution through prior investments and acquire skill sets in the process of evaluation, selection and management. Also, PE firms could establish their local networks through prior investments in the target country (Humphery-Jenner et al., 2016). As cross-border investments can be considered as the internationalization process, multinational experience about a rich array of environments with a broad range of institutional characteristics also plays a vital role in the cross-border investment process (Barkma and Vermeulen, 1998). The experiential learning could mitigate the information asymmetry created by intuitional barriers, lower the transaction complexity and hence facilitate the exit process. Further, as PE firms approach the buyout market repeatedly, building reputation is necessary because the reputation can serve as certification and help to mitigate the information asymmetry between PE firms and potential buyers (Gompers, 1996). Various empirical studies provide insights into PE learning and experience and confirm the positive role of experience and reputation in investment activities. From the view of corporate governance, firms with better track records have higher present values of rents from a favorable credit market and thus they could obtain less risky projects (Diamond, 1989). From the view of debt financing and capital structure, Cotter and Peck (2001) demonstrate that companies sponsored by buyout specialists tend to use less senior debt and they are less likely to experience financial distress. Axelson, Stromberg and Weisbach (2009) develop a theoretical model and claim that general partners have incentives to use higher leverage and risk shifting because their compensations rely on the fund return. However, general partners in reputable PE firms with marked records will conduct less risk shifting and invest more conservatively. Demiroglu and James (2010) argue that reputable PE firms have persistent performance and this confirms PE firms skills in selecting, monitoring, and restructuring. Reputation will help PE firms to deliver the impression of being less risky to investors and banks, resulting in better lending terms. Ivashina and Kovner (2011) report that PE firms which develop good 12

13 relationship with banks through repeated interactions could reduce inefficiencies from the information asymmetry and hence receive favorable loan terms from banks. Humphery-Jenner et al. (2016) find that PE-backed acquirers obtain higher announcement returns in cross-border M&A if targets are in poor information environment. They claim this is because PE firms experience and networks in the target country could help to build up PE firms reputation and PE backing hence can serve as a positive market signal. Further, Stromberg (2008) shows that, in U.S. market, experienced PE firms divest their portfolios companies more quickly. In brief, previous studies suggest that the experience not only helps PE firms to obtain more favorable terms but also facilitates the exit process. The third hypothesis is: Hypothesis 3: More experienced PE firms are more likely to perform better in the cross-border buyout. 3. Data and Variables Construction 3.1 Data and sample My sample of global LBOs comes from Mergermarket, a data provider for M&A transactions. Mergermarket tracks investment records for 1,008 worldwide PE firms (as of 31 st December 2015). Unlike other databases such as Capital IQ M&A and SDC Platinum M&A, which track investments at the transaction level, Mergermarket categorises investments into exit portfolios and holding portfolios at the PE firm level. It provides information on holding periods, buyout/exit types, transaction value, deal description and financial characteristics. I obtain the sample as follows. Firstly, I select transactions of which the deal type buyout is specified. I only keep the transaction with the leading PE 1 in club deals where more than 1 The leading PE is defined as the one invests largest stake or the oldest firm in the club deals if stake information is missing (Nahata et al. 2014). To confirm the correctness of leading PE firms, I also go through the deal description and check if the PE firm is leading the consortium/group. As Mergermarket keeps records at the PE firm level, same transaction will be recorded several times for club deals. I check the deal ID and target name and 13

14 one PE firm participate in the transaction. Since my aim is to examine the cross-country determinants of the cross-border buyout performance, I select deals if the dominant country of the portfolio company is different from the country in which the PE firm is headquartered. In addition, I keep deals of which transaction dates and holding periods are non-missing. Although Mergermarket tracks the deal history back to 1997, it provides more reliable information since Consequently, following Nahata (2008) and Nahata et al. (2014), I include buyout transactions from 1 st January 1998 and exclude all countries with less than ten observations to avoid the adverse effects of outliers. I stop the sample at the end of 2007 to be able to track the outcome of all buyout transactions during an eight-year window until the end of The final sample has 2,665 deals from 40 countries from 1998 to To supplement other deal characteristics such as deal value and management participation, I match the sample with two other buyout databases: SDC Platinum M&A and Zephyr. Zephyr has better coverage for European deals and smaller deals. I match these databases using the PE name, the portfolio company s name, and the transaction date. 2 Since some PE firms have changed their name (for example, HSBC PE is renamed as Montagu PE), I therefore extensively check the company website, confirm the change and carefully match different databases. In addition, Mergermarket s industry classification is different from the SIC classification. 3 Following Cao et al. (2015), I carefully check the industry based on the subindustry description in Mergermarket and reclassify them into 11 SIC industries according to the U.S. Securities and Exchange Commission (SEC) SIC Code List. delete duplicates. Only transactions with leading PE firms are kept. 2 Deal is labelled as Leveraged Buyout in SDC and Institutional Buyout in Zephyr; deal status is completed deals ; the time span is from 1 st January 1998 to 31 st December The geographically area is worldwide and the dominant country is defined as the place where the portfolio company is located in. In Zephyr, the PE name is not always specified so that I go through each deal description to figure out the PE firms behind each deal. 3 For example, the Computer Hardware industry in Mergermarket includes manufacturers, distributors, and retailers while the corresponding SIC codes for these industries are 3570, 5045 and

15 Finally, I obtain country index and legality index data from the International Country Risk Guide database following La Porta et al. (1998) and cultural distance data from Taras, Steel and Kirkman (2012). The market development data come from SDC Platinum M&A database. Other country-pair controls are from the Central Intelligence Agency (CIA) World Factbook, the Foreign Law Guide database and CEPII database. 3.2 Variable construction Dependent variables There are five exit routes for PE firms: IPO, secondary buyout, trade sale, buyback, and write-off. Secondary buyout and trade sale are considered as the M&A route (Povaly, 2007). Mergermarket identifies the following four exit types: IPO, secondary buyout, trade sale, and other exits. Other exits are exit routes except for IPO, secondary buyout and trade sale and Mergermarket could not track details for these routes. As a result, these deals drop out from the study before the ending date. For the analysis on the probability of exit success, I follow the previous PE literature (Hochberg et al., 2007; Nahata et al., 2014) and code exits to be successful if PE firms could exit portfolio companies either via IPO or M&A. The data allow me to track the PE s portfolio companies during an eight-year window from the initial buyout. I assign a value of one to a portfolio company that was divested within eight years 4 from its buyout date and zero to unsuccessful exits, which including other exit routes and non-exit ones. For the survival analysis of time-to-successful exit, I define the variable Holding Time which is the number of months between the buyout date of and the successful exit date. For portfolio companies with unsuccessful exits, the holding time is the number of months between the buyout date and 31 st December 2015 or the last available tracking date. 4 For robustness check, I also test the seven-year (the median duration of this sample) and nine-year window length and results are qualitatively similar. 15

16 3.2.2 Explanatory variables The explanatory variables can be categorized into four sets of variables. The first set pertains to the formal and informal institutions. The second set consists of learning factors from different perspectives. The third category contains the deal characteristics. Finally, the fourth category represents the set of country-pair control variables. Appendix A presents the definitions of these variables and their data sources Factors related to institutions Firstly, I obtain the country index from International Country Risk Guide database to proxy the quality of institutional environment. La Porta et al. (1998) extract four key variables from this database to measure the legal enforcement: the rule of law, corruption, the risk of expropriation, and repudiation of contracts by the government. The variables in this database has been intensively used in the following law and intuition studies (Lerner and Schoar, 2005; Gelos and Wei, 2005; Cumming et al., 2006; Cumming, 2008; Nahata et al., 2014; Humphery- Jenner et al., 2016). The country index is a composite index. In the index calculation, political risk components account for 50% and the rest consists of economic and financial risk components. Low risk countries are defined as those with rating scores higher than 80 points while high risk countries are those with rating scores less than 50 points. The manner of composite index calculation is developed and used in several law and institution studies such as Berkowitz et al. (2003), Cumming et al. (2006), Cumming (2008) and Nahata et al. (2014). As argued by Nahata et al. (2014), the positive aspect of using a single composite index is to alleviate the influence of multicollinearity when all individual variables are included. As a further test, I follow Berkowitz et al. (2003) and Cumming et al. (2006) and construct a legality index: 16

17 Legality Index = 0.381*(Efficiency of Judiciary) *(Rule of Law) *(Corruption) *(Risk of Expropriation) * (Risk of Contract Repudiation) The legality index captures the impact of the quality of legal system on the buyout performance. Further, Meuleman and Wright (2011) state that the development of local LBO market is one of key factors in the institutional context. A mature LBO market could facilitate the exit process because there are more buyout-related investment banks, law firms, and financial advisors and the transaction complexity could therefore be reduced. Also, PE firms could find more financial buyers and sell their portfolios via secondary buyout. In line with Nahata et al. (2014), I construct the LBO market development as the aggregate number of LBOs in the country of the portfolio company from 1990 to the year prior to the initial buyout and normalize it by the world total number of LBOs in the same period. To proxy the cultural difference, I adopt the Hofstede s cultural distance. In his 1980 book Culture s Consequences: International Differences in Work Related Values, he emphasises influences of culture on society and economic development. There are four dimensions in his cultural evaluation: power distance, individualism, masculinity, and uncertainty avoidance. Hofstede (2001) states that power distance describes how less powerful groups within a country expect and accept that power is distributed unequally. Individualism describes the society in which connections between individuals are loose. In addition, masculine society is the one in which men are supposed to be assertive, tough and focused on material success while the role of the female is supposed to be more modest, tender and life quality focused. Finally, uncertainty avoidance is the degree to which people of a culture feel threatened by uncertainty and ambiguity. I compute the multidimensional cultural distance between the country of the PE firms and the country of the portfolio company as follows: 17

18 Cultural Distance = ( 4 i=1 (C TC,i C PE,i ) 2 ) 1/2 where C TC,i is the portfolio company s national culture measured on element i and C PE,i is the leading PE firm s national culture measured on element i. The cultural distance is measured in the buyout year. To capture changes in the cultural distance, I use culture scores from Teras et al. (2011). These data are country-specific and time-varying over three periods: 1980s, 1990s, and 2000s. If the buyout year is between 1998 and 1999, the 1990s data will be used; if the buyout year is between 2000 and 2007, the 2000s data will be used. Following Nahata et al. (2014), if the data are missing for the 1990s, I will use 2000s score as the proxy. If the data are missing for both 1990s and 2000s, the 1980s score will be used as the proxy. For a few countries where there are missing information, I use the average in the same buyout decade as the proxy Factors related to learning I construct four variables to measure different aspects of learning. Stromberg (2008) shows that the experience of PE firms consistently explains the global exit behaviour and the variation in holding periods. Experience (measured by the age) is considered as a PE firm s reputation in many studies (Gompers, 1996; Gompers and Lerner, 2000; Alperovych et al., 2013). However, the age may not distinguish between funds in the cross-border studies. In this study, the first learning variable is the Country-specific Experience. Following Kogut and Singh (1998) and Humphery-Jenner et al. (2016), I construct this variable as the number of buyouts which the PE firm completed in the country of the portfolio company from 1990 to the year prior to the initial buyout. The second variable Multinational Experience is constructed as the number of foreign countries in which the PE firm invested from1990 to the year prior to the initial buyout (Barkema and Vermeulen, 1998). The next variable Industry Experience aims to capture the industrial specialization as each PE firm has its industrial focus. For example, 18

19 the UK PE group 3i focuses more on Service and Industrial Products industries. Industry experience is calculated as the number of buyouts which the PE firm completed in the industry of the portfolio company from 1990 to the year prior to the initial buyout. Further, as argued by Demiroglu and James (2010), the accumulating experience will increase over time and could not distinguish between funds. Following Demiroglu and James (2010), I construct another variable Reputation measuring recent experience: the total number of buyout transaction completed by the PE firm three years prior to the initial buyout. As there is a time gap between the initial buyout date and the final exit date, all measures link PE firms past experience to their future performance and thus avoid the reverse causality Factors related to deal characteristics With the management team participating in the buyout transaction, the information asymmetry between PE firms and the portfolio company could be reduced and hence a better performance is anticipated. To account for the corporate governance characteristic, I adopt the dummy variable Management which equals to one if the deal is defined as management buyout in Mergermarket, acquirer including management in SDC Platinum M&A database, or management buyout in Zephyr. Further, to account for the syndication among PE firms, the variable Club Size is included. Officer, Ozbas and Sensoy (2010) demonstrate that the PE club pays less for the buyout transaction and such lower pricing might be an inadvertent by-product of an unobserved motivation for club deals. Meuleman and Wright (2011) find that institutional differences induce U.K. PE firms to cooperate with a local PE firm when they invest in continental Europe. The variable Club Size 5 is calculated as the number of PE firms in the deal. Moreover, Nahata (2008) includes the total funding amount across all rounds to capture the quality of the portfolio 5 To calculate the club size in Mergermarket, I read through the details in buyer description, seller description, equity provider and deal description and figure out the number of PE firms. 19

20 company. The higher the total funding across all rounds, the better the company quality. He admits that because of the data limitation, the proxy is imperfect and involves the look-ahead bias. Mergermarket, SDC and Zephyr provide little information on portfolio companies other than transaction details. In a similar manner of Nahata (2008), I include the deal value to measure the size effect and the quality of the portfolio company. The larger the deal value, the larger the investment the PE firm makes. The deal value thus could capture the quality of the portfolio company to a certain degree. The deal value information is from the buy value in Mergermarket, transaction value in SDC Platinum M&A or deal value in Zephyr Factors related to country-pair controls I also consider other types variables measuring the link between the country of the PE firm and the country of the portfolio company. Guiso et al. (2003) and Aizenman and Kendall (2008) show that religion and language have an impact on the economic development. The religion and language information are extracted from Central Intelligent Agency (CIA) World Factbook. Variable Common Religion is a dummy variable that equals to one if the country of the PE firm and the country of the portfolio company have the same primary religion. Variable Common Language is the dummy variable that equals to one if they have the same first official language. Also, I track the law origin and commercial code of both portfolio companies nations and PE firms nations in Foreign Law Guide database. Following previous law and finance studies such as La Porta et al. (1998) and Lerner and Schoar (2005), the world legal systems are divided into six groups: English common law, French civil law, German civil law, Scandinavian civil law, Islamic law and Socialism background law. Variable Common Law- Origin is a dummy variable that equals to one if country of the PE firm and the country of the portfolio company have the same legal origin. Finally, geographic proximity could favour the participation of PE firms in portfolio companies and improve the performance (Chen, Gompers, 20

21 Kovner and Lerner, 2010). I measure the geographic proximity by using the geographic distance between the most populated city of the country of the portfolio company and the country of the PE firm. The data are from CEPII database. 3.3 Descriptive statistics Table 1 provides descriptive statistics for the sample of cross-border buyouts. Panel A reports the incidence of buyout based on the buyout year. I classify the exit outcome types based on the Mergermarket records as of 31 st December 2015: successful exits and unsuccessful exits. Successful exits represent 64% of the sample and are divested via IPO (4%) or M&A (60%). The sample suggests that PE firms prefer M&A as the way to divest their portfolios. Unsuccessful exits include other exits (7%) and non-exit ones (30%). Other exits are portfolio companies for which Mergermarket loses tracking information and non-exit portfolios are still privately held by PE firms. The sample is similar to Stromberg (2008) who reports one-third of portfolio companies are still private until 2008 and M&A earns its popularity as the divestment alternative to IPO. In terms of the distribution of the sample, the majority of buyout portfolios took place in early years are exited successfully. For the buyout portfolios took place in 2007, 47% of them are still private until The holding time is thus right censored. [Insert Table 1 about here] In panel B of Table 1, I show the incidence of buyout portfolios based on industry using two-digit SIC codes. Industry categories are defined with respect to the SEC industry classification. The majority of portfolio companies operates in the Manufacturing industry, amounting for 45% of the deals. Service industry and Retails industry account for 24% and 7% of the buyouts, respectively. The distribution in terms of industry is similar to Cao et al. (2015). The successful exit ratios within each industry are highest in Manufacturing (69%) and Service 21

22 (66%) industries. I also find that in the Finance industry, the successful exit ratio (45%) is lower than other groups. To control for the industry heterogeneity, I include the industry fixed effects in all tests. Panel C of Table 1 shows the clustering of buyouts across countries of portfolio companies. Cross-border buyouts in Germany constitutes 12% of the deals and U.K. accounts for 10%. Cao et al. (2015) use a sample of which 60% of buyout transactions are from the U.S. and U.K. markets. My sample is more representative as I include deals in both developed and developing markets. Further, emerging Asian countries attract foreign investors attention in recent years, especially China and India as there are rapid macroeconomic growth, demographic change, and legal and financial reforms (Naqi and Hettihewa, 2007; Dai et al., 2012). India consists 5% and China accounts for 4% of the deals, respectively. I find that in these countries, the successful exit ratio is lower and less than 50%. This result is consistent with Lerner et al. (2009) who find a low exit ratio in emerging Asian countries. I list countries of PE firms in panel D of Table 1. U.S. and U.K. markets are generally believed to be the most developed LBO markets, and PE specialists from these two countries contribute to 67% of the buyouts. Table 2 reports the descriptive statistics for explanatory variables including the mean, median, standard deviation, minimum, and maximum. I also test the difference between successful exit and unsuccessful exit groups. The successful group has the higher country index and legal index values. The successful group also has a higher cultural distance value. This might provide marginal evidence to Nahata et al. (2014) and PE firms could recognise cultural differences and perform a more rigorous due diligence research. In terms of experience across all types of measurements, more experienced PE firms are more likely to exit successfully. In the empirical test, I transform these variables in the logarithm format and details are presented 22

23 in Appendix A. In addition, the successful group usually has management participation and a smaller club size. Finally, the average deal value is around million USD 6. [Insert Table 2 about here] 4. Empirical Results and Discussions 4.1 Likelihood of a successful exit Probability of a successful exit Table 3 reports the results of the logit analysis. Following Stromberg (2008), the dependent variable Exit within Eight Years takes a value of one if the portfolio company is divested via IPO or M&A within eight years of its initial buyout and zero if the portfolio company is still privately held by the PE firm or Mergermark loses tracking records. The specification is: P i = eα i+x i βi 1 + e α i+x i β i, logit (P i ) = log ( P i 1 P i ) = α i + x i β i [Insert Table 3 about here] Across models 1-4, I relate the all institutional variables, deal characteristics and countrypair control variables to the likelihood of a successful exit. To avoid the multicollinearity, I include different measurements of learning variables separately. I find that in all specifications, coefficients of the variable Country Index are positive and statistically significant. The results are consistent with the hypothesis that the quality of institutional environment is positively associated with the exit success and support the law matters view (La Porta et al., 1998). The results are in line with previous studies of the impact of law and institution on exit performance of cross-border VC investments such as Nahata et al. (2014), suggesting that the quality of institutional environment is vital not only for VC 6 The average deal value in Cao et al. (2015) is million USD. As one of their sample selection requirements is that the deal value must be larger than 5 million USD, the average deal value in their sample is reasonably larger. 23

24 investments but also for buyout investments. Coefficients of the variable Culture Distance are insignificant across all models. The results imply that PE firms which are sophisticated investors (Cao et al., 2015) suffer from minimal adverse influences of cultural differences. This is different from Nahata et al. (2014) who report a positive influence of cultural differences on VC exit performance. Compared to VC firms, buyout firms conduct the LBO to acquire the late-stage and mature firms which could generate enough operating cash flow to repay the debt (Jensen, 1989). Consequently, one plausible explanation for this finding could be that, the sophisticated buyout specialist reply on the hard information and thus overcome the barrier of cultural differences. In line with Nahata (2008), Demiroglu and James (2010) and Meuleman and Wright (2011), PE firms experience and reputation have positive impacts on investment performance. Specifically, multinational experience which brings in the knowledge of different institutions, industrial experience which offers deep industrial insights and reputation which serves as the certification to resolve asymmetric information problems help PE firms achieve higher likelihoods of a successful exit in the cross-border buyout. I also find that management participation which reduces the information asymmetry between insiders and PE managers helps to improve the buyout performance. The coefficient of the dummy variable Management is around In terms of economic significance, ceteris paribus, management participation increases the odds of that the portfolio company to get divested successfully by a factor of Regarding the probability interpretation, management participation increases the probability by In terms of club size, I find the diseconomies of scale of the PE club as the larger PE club takes a longer time to successful divest the portfolio company. In the later section 4.3, I find that the club size is positively associated with the 7 The increase in odds is equal to e = 1.44 and the probability is equal to 24 odds 1+odds.

25 likelihood of going IPO and negatively associated with the choice of M&A. PE firms frequently adopt M&A as the route to divest portfolio companies but IPO is considered as the most successful way (Gompers, 1996). Combining these results, one could argue that although it takes more time for the larger PE club to divest the portfolio company, the large PE club improves the cross-border buyout performance in a modest way as it increases the probability of bringing the portfolio company to the market through an IPO. When the geographic distance is taken into consideration, I provide consistent evidence to the geographic proximity studies (Chen et al., 2010). PE firms are less likely to exit successfully if they are far away from their portfolio companies. In model 5, instead of using the composite country index, I follow Berkowitz et al. (2003) and Cumming et al. (2006) and include the legality index. In addition, to control the size effect of the deal, I add the variable Deal Value. I find that the legality index is positively related to the investment performance. The result is consistent with previous law and finance and crosscountry studies (Rossi and Volpin, 2004; Bris and Cabolis, 2008; Nahata et al., 2014; Cao et al., 2015). The result indicates that stronger contract enforcement, better investor protection and well-developed legal system which generally reduce the information asymmetry and transaction complexity are beneficial to improve the buyout performance. Overall, the results of the logit analysis indicate the probability of a successful exit increases when the quality of the institutional environment and legal system is higher and when PE firms have more international experience, more industrial experience, and stronger reputation certification. In addition, sophisticated LBO specialists do not suffer from the adverse influence of cultural differences. These findings support the law matters view and experience matters view. 25