WPS5953 Policy Research Working Paper 5953 Bilateral M&A Activity from the Global South Mansoor Dailami Sergio Kurlat Jamus Jerome Lim The World Bank Development Economics Prospects Group January 2012 Policy Research Working Paper 5953 Abstract This paper studies the factors associated with outbound according to whether the acquisition targets are in other bilateral mergers and acquisitions (M&A) activity by emerging economies or advanced countries, and that firms located in emerging economies. The authors these differences can be attributed to differing theoretical document recent trends in emerging market M&A flows, motivations behind foreign direct investment. The which have risen dramatically over the past decade, authors also consider the implications of their model for and explore the factors that may have contributed to future M&A originating in the global South, in light of this rise. They find distinct patterns for M&A deals the global financial crisis of 2008. This paper is a product of the Prospects Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The authors may be contacted at mdailami@worldbank.org, skurlat@worldbank.org, and jlim@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Bilateral M&A Activity from the Global South Mansoor Dailami, Sergio Kurlat, and Jamus Jerome Lim∗ Keywords: Mergers and acquisitions, South-South capital flows JEL Classification: F21, G34, G11 Sector Board: EPOL ∗ The authors are with the Development Prospects Group at the World Bank. Their respective emails are: mdailami@worldbank.org, skurlat@worldbank.org, and jlim@worldbank.org. We thank Robert Hauswald for generously providing the data, and Jonathon Adams-Kane, Yueqing Jia, and Harris Kim for valuable comments. This paper served as a technical background paper for a subset of the policy-oriented discussions in Chapter 2 of the World Bank’s Global Development Horizons 2011 report, led by Dailami. The �ndings, interpretations, and conclusions expressed in this article are entirely those of the authors. They do not necessarily represent the views of the World Bank, its Executive Directors, or the countries they represent. 1 Introduction The shift of global �nancial investment activity away from the advanced world toward emerg- ing economies has been widely documented, and nowhere is this shift more evident than in recent patterns of cross-border mergers and acquisitions (M&A) deals. Between 1997 and 2003, companies based in emerging economies engaged in outbound cross-border M&A deals worth $189 billion, or 4 percent of the total value of all global cross-border M&A investment. In the equivalent period from 2004 to 2010, that amount had increased to $1.1 trillion, or 17 percent of the global total. This sharp rise in emerging economies’ share of cross-border M&A has been accompanied by a deepening reach of emerging-market �rms in international capital mar- kets overall, not just via foreign direct investment (FDI) but also through equity cross-listings, participation in international loan syndicates, and debt issues on international bond markets. Given the rising importance of emerging economies in international M&A, the question of what forces drive their investment activity becomes ever more important. In particular, do cross-border M&A choices by emerging market �rms differ systematically across destination markets, and if so, what are the dimensions in which these choices differ? This question is increasingly pertinent as government regulators grapple with the dramatic pickup in M&A investment by Southern �rms in both Northern markets—exempli�ed by Chi- nese carmaker Geely’s high-pro�le acquisition of Sweden’s Volvo from Ford in 2010—as well as (and perhaps more importantly) their purchasing activities in the other developing economies of the South. Indeed, the trend of rising South-North and South-South M&A activity suggests yet another wrinkle to the Lucas (1990) paradox of “uphill� South-North capital flows: It is not sufficient for theoretical explanations to merely explain why emerging market �rms may be investing in (ostensibly) less risky Northern markets, but also why they may choose to in- vest in other economies that are otherwise very similar to their own (insofar as their level of development is concerned), rather than investing at home. This paper seeks to examine the factors associated with the flow of M&A investment orig- inating in emerging market economies.1 The empirical analysis relies on bilateral outbound cross-border M&A data for �rms based in 61 emerging markets, collected for the period be- tween 1997 and 2010. This coverage includes economies from all major developing regions, which makes this, to our knowledge, one of the most comprehensive analyses of bilateral M&A activity by the Global South. The picture of Southern cross-border M&A that emerges from our paper is a fairly sophis- ticated one. Consistent with other forms of cross-border economic activity, M&A deals reflect standard gravity components, such as economic size and distance. But the strength of existing trading and investment relationships also matter, and for acquisitions in advanced economy targets, the informational advantages gleaned from such prior economic relationships appear to overcome frictions due to physical distance. Moreover, deals in advanced economies tend to 1 For the purposes of this paper, emerging economies are de�ned as 61 (mostly middle-income) economies traditionally classi�ed as emerging markets by the �nancial community. The full list is provided in the annex. 2 reflect the extent to which FDI can substitute for direct exporting activity, or offer possible di- versi�cation bene�ts. In contrast, acquisitions in other emerging countries tend to be associated more with considerations of factor price differentials. Finally, the ease of �nancial access—both in the home and host economies—appears to facilitate M&A transactions, a result consistent with the notion that limitations to trade flows may be overcome by substituting capital flows for goods exports. We also �t our empirical model of M&A deals to a set of growth assumptions for emerging and advanced economies to obtain a projection of outbound cross-border M&A by emerging market �rms for the period 2010–25. Under plausible scenarios of relative growth rates, we �nd that M&A activity is expected to recover from the crisis-induced decline, and grow at an average of 8.2 percent annually over the period. This respectable rate nevertheless represents a moderation in the rate of growth relative to the past, where—for the decade leading up to the crisis (1998–08)—average annual growth was signi�cantly higher, at 14.3 percent. Our �ndings corroborate, and extend, the existing literature in several ways. Existing empir- ical papers on FDI have tended to focus on testing one theoretical framework against another a (Braconier, Norb¨ck & Urban 2005; Brainard 1997; Head & Ries 2008; Helpman, Melitz & Yeaple 2004). In contrast, we adopt an agnostic view on the different competing theories and seek instead to test a fairly eclectic set of potential hypotheses. Like several papers in the liter- ature (Anand & Delios 2002; Blonigen & Piger 2011; Carr, Markusen & Maskus 2001; Makino, Lau & Yeh 2002), we indeed �nd that, depending on the circumstance, different theoretical motivations may drive FDI. Our innovation is to frame these distinct cases in terms of emerging versus advanced economy target acquisitions. While a small number of papers have relied on cross-border M&A data (di Giovanni 2005; e M´on & Delannay 2006; Rossi & Volpin 2004), these have tended to be relatively limited in terms of time period and/or country coverage, and none have explicitly focused on M&A by emerging economies. Finally, unlike several recent papers, we eschew an explicit focus on policy- or politically-related factors driving FDI—such as political risk (Busse & Hefeker 2007) u or investment agreements (B¨the & Milner 2008; Neumayer & Spess 2005)—and instead embed these factors in our overall gravity framework. The paper is organized as follows. In the following section, we describe some broad stylized patterns of M&A activity originating from emerging economies (Section 2). This is followed by an overview, in Section 3, of the different theoretical streams that have informed economists’ understanding of bilateral M&A (and FDI, more generally). Section 4 follows with a description of the dataset, econometric speci�cation, and estimation methodology. Benchmark results and robustness checks follow (Section 5), before a �nal section concludes with some brief thoughts on policy and future research directions. 3 $ billions deals 1,800 3,000 Value (Emerging) 1,600 Value (Advanced) 2,500 Deals (Emerging) 1,400 1,200 2,000 1,000 1,500 800 600 1,000 400 500 200 0 0 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Source: Authors' calculations, based on Thomson-Reuters SDC Platinum Figure 1: Total cross-border M&A deals by �rms from advanced economies and emerging- market economies, 1997–2010. The upward trend held by emerging market �rms is evident in both absolute deal number and relative deal values. 2 Recent Trends in Emerging Economy M&A Emerging market multinationals have become far more assertive in their M&A activities on the global stage over the past decade. Save a dip during the global crisis in 2008 and 2009, the overall trend in cross-border M&A has been upward, especially in the post-dot-com period since 2001. This is evident both in terms of the total number of deals (increasing from 661 deals in 2001 to 2,447 in 2010), as well as—perhaps more dramatically—in the value of M&A deals concluded by emerging economy �rms (the rise from $30 to $254 billion over this period represented an increase in shares from 6 to 29 percent) (see Figure 1). Although not the focus of this paper, it is illuminative to consider, by way of comparison, whether patterns in emerging market M&A are also replicated in the other component of FDI, green�eld activity.2 The share of emerging market green�eld investment did indeed rise between 2001 and 2009 (the latest year data are available), from 12 to 15 percent of total global green�eld activity. This rise, while clearly more modest, was nevertheless a signi�cant absolute increase: the value of cross-border green�eld investment rose from an estimated $98 billion to $250 billion over the same period.3 2 Green�eld investment, as opposed to M&A, typically represents internal, organic corporate growth, while M&A activity, in addition to satisfying growth objectives, may capture other more complex corporate goals, such as strategic market penetration or the acquisition of new technology. 3 Green�eld data were sourced from UNCTAD and fDi Markets but were, unfortunately, only available for 2003 through 2009. The value for 2001 given here is an exponential projection from the available time series, and is not meant to be an authoritative �gure, but rather to give a sense of the magnitude involved. 4 South-North South-South $ billions 700 $ billions China 400 China 600 350 500 UAE Singapore 300 Singapore Malaysia 400 250 Brazil India 200 Russia 300 India Russia UAE 150 Mexico 200 Mexico 100 Other MNA Other MNA Other LAC 100 50 Other EAP Other EAP 0 AFR 0 Other Source: Authors' calculations, based on Thomson-Reuters SDC Platinum Figure 2: Top source countries of emerging-market cross-border M&A in emerging economies (left) and advanced economies (right), by value, 1997–2010 total. AFR, EAP, LAC, and MNA correspond to the World Bank’s regional classi�cations for Africa, East Asia and the Paci�c, Latin America and the Caribbean, and Middle East and North Africa. As may be expected, large and fast-growing emerging economies are responsible for the bulk of cross-border M&A activity. China is the single largest emerging market source country for M&A deals, and accounts for $80 and $132 billion of the total $426 and $698 billion invested in emerging and advanced economies, respectively, over the 1997–2010 time period. Other emerging economies with signi�cant presence among source countries include Singapore, the United Arab Emirates,4 India, and Russia (see Figure 2). Much of these flows are destined for developed markets—primarily the United States, United Kingdom, Canada, and Australia—but China, Singapore, and Brazil are major destination markets as well. In terms of sectoral composition, the major emerging market M&A transactions appear in high-value, nontradable service sectors: �nancial services (the top sector for cross-border M&A activity among emerging-market �rms, amounting to $227 million of the total of $1.12 billion for 1997–2010), telecommunications (a distant second, amounting to $103 million), resource extraction, and utilities. There is little difference in the sectoral composition of M&A deals in emerging versus advanced economies, which suggests that, to the extent that there are dis- tinct patterns between Southern and Northern investments, they lie more in the nature of the respective economies, rather than in the type of businesses involved.5 4 The relatively high standing of Singapore and the UAE, in spite of their relatively small size, is attributable to the large number of sovereign wealth fund acquisitions in these economies. Singapore’s Temasek Holdings and GIC Real Estate, for instance, accounted for 32 and 31 deals, respectively, in the sample period. 5 The sectoral distribution differs somewhat when considering the number, rather than value, of deals, with 5 A casual examination of the actual transactions data suggests that the overall pattern of cross-border M&A investment by emerging-market �rms is consistent with the typical interna- tional growth strategy of individual corporations. When companies venture abroad, they often �rst establish a small foothold in new markets through branch or representative offices, small distribution networks, or maintenance centers. Such small green�eld investments can be the �rst step toward execution of a �rm’s globalization strategy, allowing companies with limited international exposure to gain experience and local knowledge before making a major com- mitment to a particular market through an outright acquisition or large-scale investment via mergers.6 In carrying out M&A transactions, companies often appear to seek more immediate access to local markets. Firms may potentially capitalize on technological and informational advantages that may be gained by their foreign acquisitions, or when they can apply their unique expertise to the same industries abroad. In particular, emerging-market �rms with expertise overcoming the difficult institutional environment in their home countries may be eager to apply this informational ad- vantage to similar environments in other emerging markets. Some M&A may also be motivated by the desire to exploit factor cost differentials in target markets (relative to their own). Finally, international M&A activity may demonstrate some persistence, when initial investments lead to additional cross-border investments through the necessity of the restructuring or upgrading of acquired assets, or as part of acquiring other �rms’ vertical- or horizontal-integration growth strategies. 3 Potential Factors Related to M&A Activity Economic models of bilateral trade flows have most successfully been modeled on the basis of an empirical gravity equation, which has more recently been contextualized in the form of a broad variety of theoretical models (Anderson & van Wincoop 2004). We therefore rely, as a point of departure, on a gravity equation, where cross-border M&A flows are positively related to the pair’s respective output and negatively related to the bilateral distance between them. Bilateral country distances capture not only explicit trade costs associated with shipping and transportation, but could also embed implicit transactions costs related to the deteriorating quality of an investor’s (or acquirer’s) knowledge of, and ability to obtain information about, a potential acquisition target as physical distances between the two countries increase (in line with the argument made by Loungani, Mody & Razin (2002)). Naturally, existing bilateral trade flows are also likely to be associated with bilateral M&A flows. sectors such as professional and technical services, and electronics manufacturing featuring more prominently in emerging market M&A deals. However, the pattern of overlap in South-North and South-South remains the same. 6 Such staged investment strategies emphasize the real-option aspects. Consequently, the initial green�eld investment serves is a stepping-stone to understanding a local economy. As uncertainties about demand and supply become resolved over time, follow-up investments then create a permanent presence in the foreign market by extending the scope and reach of the initial unit. Gilroy & Lukas (2006) provide the theoretical justi�cation for this phenomenon, while Brouthers & Dikova (2010) establish empirical evidence. 6 In addition to this baseline, however, we supplement the model with a range of theories that have been put forward to explain cross-border investment activity. The �rst class of theories posit that companies seek growth opportunities abroad as they outgrow their home markets; a problem especially acute in developing countries. The decision of multinationals to either horizontally expand to access foreign markets or vertically integrate production across borders, in turn, depends on both market size and the ability to exploit factor price differentials between the two production locations (Helpman & Krugman 1985; Markusen 2002; Markusen & Venables 1998). This result, which relies on relative factor proportions, suggests that, in addition to absolute GDP, per capita incomes—as a proxy for factor costs— could be important for M&A choices. Of course, the implications of the factor proportions hypothesis is not limited to contempo- raneous differences in factor prices, but also possible future differentials. Consequently, relative growth in both home and destination countries could affect deal flows. This hypothesis can thus be further tested by including variables that measure GDP and sectoral growth rates. Following this rationale, faster growth in the home (host) country will exert greater pressure on domestic (foreign) factor prices and hence increase (reduce) incentives to engage in cross-border M&A.7 A second class of theories revolves around structural economic characteristics of the home and host countries, especially those related to the extent of trade openness (Brainard 1997; Helpman et al. 2004; Horstmann & Markusen 1992), but also with regard to differential access to �nance (either domestic or international), or differences in the speed of diffusion of tech- nological advances. This tradeoff—between proximity to the customer versus concentration of production—tends to privilege the former especially when transport costs and trade barriers are substantial or, conversely, when economies of scale favoring home production and subsequently exporting are relatively low. Indeed, for economies heavily invested in high-�xed, low-marginal cost activities such as research and development (R&D), the proximity-concentration hypothesis would argue against FDI (or, at the least, geographically-diffused FDI, since pockets of research excellence may exist in more than one location). Given that emerging economies have now become important contributors to the advancement of science and technology in their own right, one can further test this group of hypotheses by including variables directly related to the home country’s capabilities in science and technology, such as the number of patents granted, or through other indirect measures of innovative capacity, such as the percentage of the population attaining a tertiary education or the number of engineering graduates in the population. Financial access can also be captured via measures of international �nancial openness (by, for example, private capital flows as a share of GDP) or the level of domestic �nancial development (by, for instance, the ratio of stock-market capitalization to GDP). Innovation in the host country may also serve as a justi�cation for M&A. This class of theories, which focuses on the potential for FDI to facilitate technological and other types of 7 However, host growth could also increase its market size in the future, in which case growth could increase the desire for M&A, leaving the sign of the coefficient ambiguous. 7 a s a ıguez-Clare spillovers (Ethier 1986; Fosfuri, Motta & Rønde 2001; Havr´nek & I˘ov´ 2012; Rodr´ 1996), suggests that the desire for technological and knowledge transfer could motivate emerging market �rms to acquire �rms in an advanced economy. At the same time, emerging-market �rms may have specialized managerial and operational expertise which the �rms could spillover to markets very similar in nature to their home markets. This technology transfer hypothesis argues that it is not only the home country’s innovative capacity that may influence M&A choices, but also that of the host country. Finally, political and policy factors may play a role in international M&A as well. Possibly the most likely channel where public policy could affect M&A deal flows is the residual accumu- lation of reserves as a consequence of existing trade patterns. In addition, policy factors that may affect M&A could take other forms, such as the presence of bilateral investment treaties u (BITs) (B¨the & Milner 2008; Neumayer & Spess 2005), or risks associated with economic pol- icy or political conditions (Busse & Hefeker 2007). Accordingly, we cluster these variables into the class of political economy explanations, which we consider in our empirical work. 4 Data Description and Econometric Methodology 4.1 Data sources and description The cross-border M&A investment database used for this paper was compiled from a variety of sources. The primary data for M&A deals were drawn from a larger dataset compiled by Thompson-Reuters SDC Platinum, which covers all publicly disclosed cross-border transactions for which the ultimate acquiring company was based in an emerging-market country, and the immediate target company was located in a country other than that of the ultimate acquirer. Transactions that were included involved either two or more companies pooling their assets to form a new entity (a merger ), or a foreign company gaining a portion of a domestic company (an acquisition). All completed and partially completed deals were included, as well as intended and pending deals announced after September 1, 2009. The de�nition of a cross-border M&A transaction used in this paper includes any deal where any equity stake is obtained by the acquirer �rm.8 When no deals were recorded for any country and year, the dependent variable was coded as zero.9 The compilation resulted in a working database that covers some 10,000 companies from 61 emerging-market economies, over the period between 1997 and 2010. These were merged with the main independent variables of interest and additional con- trols, which were drawn from a variety of additional sources. These include macroeconomic conditions from the World Bank’s World Development Indicators (WDI) and the IMF’s In- ternational Financial Statistics (IFS); �nancial factors from Dealogic DCM Analytics, MSCI, 8 This grouping includes investments where share purchases resulted in acquisitions of less than 10 per cent of a �rm’s voting shares, a narrower but commonly-used de�nition of FDI. 9 The database also provides historical information on acquirer and target countries (both immediate and ultimate), status, sector, and consideration offered. These were used in the section describing stylized facts in the data, but were excluded from the econometric analysis. 8 and J.P. Morgan; commodity prices from Goldman Sachs and the World Bank’s Development Prospects Group; bilateral investment treaties from UNCTAD; country risk and institutional indicators from the PRS Group’s International Country Risk Guide (ICRG); and technology and innovation indicators from the World Intellectual Property Organization. Depending on the speci�cation, the dataset is an unbalanced panel that includes between 29,995 and 55,497 observations.10 4.2 Econometric speci�cation and estimation The econometric model we use is an augmented gravity model that speci�es that the number of cross-border M&A deals originating in country i (“home�) and destined for country j (“host�) at time t, Mijt , is a function of each country’s output in that period, Yit and Yjt , the (time- invariant) bilateral distance between them, Dij , and additional factors: Mijt = β1,k Yit + β2,k Yjt + β3,k Dij + Γk Xit + Λk Zjt + Φk Bijt + Ψk Gt + ijt , (1) where X and Z are vectors of home- and host-country characteristics, respectively, B is a vector of other variables capturing the bilateral economic relationship between the home and host countries, and G is a vector of additional controls representing global macroeconomic and �nancial conditions. To maintain parsimony, we nest the two possible host targets within (1), so that the various coefficients—β, γ, λ, φ, and ψ—are allowed to vary by host-country class (advanced, AD, or emerging, EM ), so that k = {AD, EM }. Variables considered within X and Z are informed by the different theoretical approaches outlined in Section 3. These include, inter alia, GDP growth (corresponding to the factor pro- portions hypothesis), trade openness (corresponding to the proximity-concentration hypothesis), patents granted (corresponding to technology transfer arguments), and international reserve holdings (corresponding to political economy explanations). Additional variables included to account for the bilateral relationship between country pairs include factors such as the existing size of bilateral trade and the existence of a BIT between the two economies. In our benchmark regressions, (1) was estimated using ordinary least squares (OLS) with two-dimensional clustering for standard errors (by country-pair and time), designed to correct for both heteroskedasticity across countries and serial correlation within countries. Because distance is time-invariant (and we are interested in the signs and magnitudes of the coefficient β3 ), we do not introduce country-pair �xed effects. Moreover, given the inclusion of global variables, we exclude time �xed effects from the benchmark in order to minimize the incidence of multicollinearity, although we explore this possibility in our robustness checks. 10 The technical appendix provides detailed descriptions of each variable and its source in Table A.2, along with summary statistics for major variables of interest in Table A.3. 9 5 Empirical Results and Robustness Checks 5.1 Benchmark results Our benchmark results are reported in Table 1, for three main speci�cations: (B1 ) A bare-bones speci�cation that comprises the standard components of a gravity model (GDP and distance); (B2 ) A parsimonious speci�cation that includes only one representative variable from each family of hypotheses elaborated on in Section 3,11 along with the main control variables such as bilateral trade flows and global macroeconomic conditions; and (B3 ) A fully speci�ed model that includes all the variables of interest associated with the various theoretical hypotheses.12 Although we consider the �nal speci�cation, (B3 ), to be the most complete representation of (1), the sample size is much smaller (about half of the �rst speci�cation), and the goodness-of-�t improves only marginally from speci�cation (B2 ). Nevertheless, some additional insight can be gleaned from the more stripped-down speci�cations. In particular, variables that are signi�cant in a less elaborate speci�cation typically survive the more comprehensive one (and always retain their signs). This suggests that such variables carry considerable explanatory power. In general, the basic gravity model variables—economic size and distance—enter with highly signi�cant coefficients (with the exception of distance in the �nal speci�cation). The coefficients on both home and host GDP are positive and signi�cant, which is consistent with theoretical pri- ors: large economies tend to engage in a greater amount of cross-border economic activity, M&A included. In terms of magnitude, the effect is several times larger for acquisitions in developed versus emerging markets; this suggests that only �rms from relatively large emerging economies have the means to pursue expansion in advanced economies through M&A. Interestingly, these relative magnitudes are reversed when considering host GDP, which means that these �rms are also far more likely to pursue opportunities in larger emerging economies, compared to advanced ones. While distance from other emerging economies—a proxy for transactions costs, which can include informational costs13 —decreases M&A when considering emerging market targets, it actually increases when considering advanced economy �rms. While seemingly paradoxical, a this can be understood by recognizing that transactions costs vis-`-vis developed countries are likely to be fairly low, and so other factors are more likely to predominate in any M&A decision. Indeed, the positive coefficient is only statistically signi�cant in one speci�cation, suggesting 11 These are (theoretical family in parentheses): per capita GDP (factor proportions), trade openness (proximity-concentration), host patents granted (technology transfer), and both reserves and economic risk, since these capture distinctly different types of political-economy effects. 12 In the third speci�cation, the standard error for emerging host GDP is not reported. This arises due to collinearity. However, given the importance of GDP in the model, we have chosen to retain the variable in the speci�cation, keeping in mind that the coefficients of variables that are highly correlated with GDP should be interpreted with caution (coefficients for non-collinear variables are unaffected). 13 The quality of an investor’s information about a potential acquisition target decreases as the distance between the two counties increases, whereas the costs of communication, coordination, and monitoring all increase with distance. At the same time, �rms tend to be more knowledgeable about the �nancial, legal, and political environments of economies in close geographical proximity to their own. Proximity would thus reduce the cost of acquiring and operating subsidiaries. 10 Table 1: Benchmark regressions for factors associated with number of cross-border outbound M&A investments from emerging economies, 1997–2009 B1 B2 B3 to EM to AD to EM to AD to EM to AD Home country characteristics GDP 4.781 17.060 2.728 8.391 3.714 8.158 (1.10)∗∗∗ (4.00)∗∗∗ (1.01)∗∗∗ (0.92)∗∗∗ (1.38)∗∗∗ (2.20)∗∗∗ GDP per capita -0.672 -0.616 -0.321 -1.770 (0.64) (0.64) (1.00) (0.83)∗∗ GDP growth -0.350 1.136 (0.48) (0.76) Trade 5.096 3.751 5.232 2.458 openness (1.81)∗∗∗ (1.32)∗∗∗ (2.11)∗∗ (1.69) Financial 0.227 -0.737 openness (0.33) (0.45) Stock market 1.816 5.298 capitalization (0.50)∗∗∗ (1.83)∗∗∗ Patents -5.496 -1.993 granted (2.80)∗∗ (2.56) Reserves -1.276 1.179 -2.218 1.356 (0.44)∗∗∗ (0.31)∗∗∗ (0.72)∗∗∗ (0.78)∗ Economic 0.684 0.481 0.986 -0.296 risk (0.29)∗∗ (0.34) (0.44)∗∗ (0.66) Political -0.733 -0.055 risk (0.47) (0.34) Host country characteristics GDP 6.899 1.652 3.706 0.752 1.139 0.844 (3.28)∗∗ (0.45)∗∗∗ (1.99)∗ (0.36)∗∗ (0.41)∗∗ GDP per capita -1.227 0.921 -0.646 0.701 (0.49)∗∗ (0.30)∗∗∗ (0.53) (0.30)∗∗ GDP growth 0.312 0.355 (0.56) (1.37) Trade 1.994 -1.865 1.797 -3.257 openness (0.72)∗∗∗ (0.81)∗∗ (0.78)∗∗ (1.36)∗∗ Financial 0.339 0.122 openness (0.15)∗∗ (0.08) Stock market -0.080 2.926 capitalization (0.53) (0.75)∗∗∗ Patents -6.219 -2.622 -8.283 -5.886 granted (2.69)∗∗ (0.83)∗∗∗ (3.54)∗∗ (1.28)∗∗∗ Reserves -0.694 -0.552 0.616 -0.099 (0.28)∗∗ (0.31)∗ (1.15) (0.32) Economic 0.420 -1.224 0.163 -2.656 risk (0.35) (0.76) (0.49) (1.21)∗∗ Political -0.332 2.512 risk (0.25) (0.60)∗∗∗ Country-pair characteristics Distance -3.810 0.229 -1.236 2.025 -0.838 0.325 (0.76)∗∗∗ (0.56) (0.55)∗∗ (0.73)∗∗∗ (0.68) (0.80) Bilateral 2.583 0.551 3.013 0.501 trade flows (0.90)∗∗∗ (0.17)∗∗∗ (0.94)∗∗∗ (0.16)∗∗∗ BITs 1.226 -0.639 (0.70)∗ (0.77) Global controls 10-yr Treasury -1.575 4.472 -1.646 2.156 rate (1.34) (1.23)∗∗∗ (0.67)∗∗ (0.87)∗∗ Energy price -0.958 -0.288 -1.311 -0.536 index (0.06)∗∗∗ (0.18) (0.53)∗∗ (0.53) Agricultural 0.164 -0.302 0.921 1.098 price index (0.01)∗∗∗ (0.27) (0.59) (0.60)∗ Adjusted R2 0.078 0.078 0.275 0.275 0.289 0.289 F 128.162∗∗∗ 128.162∗∗∗ 32.686∗∗∗ 32.686∗∗∗ 20.998∗∗∗ 20.998∗∗∗ N 55,497 55,497 37,909 37,909 29,995 29,995 † Heteroskedasticity and autocorrelation-robust standard errors, clustered by time and country-pair, are reported in parentheses. A constant term were included in the regressions, but not reported. ∗ indicates signi�cance at 10 percent level, ∗∗ indicates signi�cance at 5 percent level, and ∗∗∗ indicates signi�cance at 1 percent level. 11 that other factors are in fact more critical. In addition, the positive and signi�cant coefficient on bilateral trade flows corroborates this notion, since preexisting trading relationships implies lower transactions costs, which serves to facilitate greater M&A activity. The signs and statistical signi�cance of coefficients on the other theoretically-motivated variables suggest that many of the hypotheses put forth in Section 3 have some merit. The level of host-country development, as measured by per capita GDP, negatively affects acquisitions in emerging destination countries, but is positive for advanced-country targets. This would suggest that �rms acquire assets in other emerging economies that have not yet attained a certain level of development—as measured by per capita GDP—so as to exploit factor price differentials (recall, the emerging economy acquirer �rms in the sample typically hail from middle-income countries). Acquisitions in advanced economies, in contrast, do not offer positive wage differentials, and so the positive coefficient on per capita GDP in that case would simply mean that more developed economies tend to attract more M&As, since there are likely to be more acquisition targets available. This could especially be the case for acquisitions in services—as discussed in Section 2, deals in service-related sectors account for a signi�cant share of all M&A activity—since countries with higher levels of per capita GDP also typically possess larger service sectors. As expected, the greater a home country’s participation in the global economy, as measured by its trade flows, the greater its M&A flows. To a lesser extent, this is also reflected by its for- eign currency reserves. A country whose �rms trade more frequently with advanced (emerging) economies tends to build up foreign reserves faster (slower), which makes the country’s com- panies more (less) likely to engage in acquisitions in their target markets (hence the positive coefficient on home reserve holdings for advanced economy M&A, and conversely for emerging economy M&A). The coefficients on trade openness in the host country are far more interesting. For ac- quisitions in advanced countries, less restrictive economies—as proxied by a greater degree of trade openness—will tend to attract less M&A, since emerging markets can simply choose the exporting rather than FDI route to penetrate those markets. In this sense, the two are sub- stitutes, an interpretation entirely consistent with the proximity-concentration hypothesis. In contrast, acquisitions in emerging markets appear to be complements. Since barriers to the flow of goods and services tend to be signi�cantly higher in most developing countries, a given marginal reduction in trade restrictiveness will have less of an effect. Thus, instead of choos- ing the route of exporting goods, �rms export capital instead, by establishing an operational presence in such countries. The positive and signi�cant coefficient on the distance variable for flows to advanced economies can, in fact, be viewed as further corroboration of this proximity- concentration tradeoff. The additional measures that capture �nancial access also indicate that, overall, a greater level of access to �nance is associated with more M&A activity. For instance, the ability of �rms in the home economy to raise capital (through its domestic stock market, for example) 12 can promote M&A, as can �nancial depth in the host country. By relaxing constraints to �nancing, barriers to horizontal �rm expansion are lowered, and �rms are encouraged to pursue the M&A route. The negative and signi�cant coefficients on the innovation variable in the host country lend little support to the technology transfer argument. Indeed, across all speci�cations, emerging economy �rms appear to invest less in countries with more granted patents.14 This could be, in part, because emerging economies now already account for a signi�cant share of global innovation (Aizenman & Noy 2007; World Bank 2011), and it is the emerging economies that engage in the technology transfer to less innovative host nations. Whatever the motivation, the evidence does suggest that, if �rms choose to pursue cross-border M&A, it seems unlikely that they do so for reasons of acquiring technology. Finally, there is some evidence that political and policy variables make a difference to M&A. BITs are positively related to acquisitions in emerging economy country-pairs, and the magni- tude of this effect is substantial (although the coefficient is only marginally statistically signif- icant). In particular, the positive and signi�cant effect of political stability on acquisitions in advanced countries seems to suggest that �rms actively seek to lower their political-risk expo- sure through their M&A activities in advanced economies (since higher values of the measure indicate less risk).15 Similarly, �rms from more economically stable emerging economies are more likely to seek to diversify their exposure by acquisitions in other emerging markets. 5.2 Robustness checks To examine the strength of the results reported in Table 1, we experiment with two sets of additional robustness checks. The �rst set of checks allows for additional factors that may be associated with cross-border M&A, or alternative measures of existing variables. In the interest of space, and given the relatively good performance of the parsimonious model (B2 ), we rely on this speci�cation as the basis for the robustness tests to follow.16 The results of this �rst set of checks are reported in Table 2. In the benchmark models, per capita GDP was used as a proxy for factor prices. While doing so considerably expands the data coverage, there may be concern that this measure may be capturing other relevant factors beyond factor prices per se. For example, GDP per capita— 14 One argument against this is that patents may reflect the extent to which the legal structure of the country supports patent �lings, rather than innovation. While this may be true, the legal environment is controlled, in part, by our inclusion of the political risk variable. Furthermore, we regard the patent data as the best proxy that we have available for measuring innovative capacity. As an additional robustness check, however, we substitute the total patents measure with data that includes only cross-border patents granted (which arguably better controls for differences in domestic patent law). While the results are somewhat weaker, the qualitative message remains unchanged; these additional results are available on request. 15 M&A activity does not, however, appear to respond to political risk measures in South-South acquisitions. This suggests that the hypothesis that emerging market �rms may exploit their comparative advantage in more challenging institutional environments is not supported by the evidence presented here, a result that has also been corroborated by others (Arita forthcoming). 16 Results obtained with the fully speci�ed model (B3 ) were qualitatively similar, and are available from the authors on request. 13 Table 2: Robustness regressions for additional and alternative factors associated with number of cross-border outbound M&A investments from emerging economies, 1997–2009 R1 R2 R3 R4 R5 R6 to EM to AD to EM to AD to EM to AD to EM to AD to EM to AD to EM to AD Home country characteristics GDP 10.975 10.667 3.497 8.451 4.897 7.308 6.098 7.851 2.721 8.486 3.730 9.057 (3.67)∗∗∗ (7.27) (1.07)∗∗∗ (0.53)∗∗∗ (1.12)∗∗∗ (3.21)∗∗ (1.55)∗∗∗ (1.90)∗∗∗ (1.00)∗∗∗ (0.91)∗∗∗ (1.27)∗∗∗ (1.70)∗∗∗ GDP per capita -0.481 -0.369 -0.837 -0.935 -0.898 -0.847 -0.350 -1.349 -0.676 -0.842 -0.456 -0.926 (2.05) (0.80) (0.88) (0.79) (0.68) (1.15) (0.55) (0.61)∗∗ (0.65) (0.65) (0.76) (0.72) Wages -0.000 -0.000 (0.00)∗∗ (0.00) Trade 6.735 1.911 6.051 3.959 5.348 4.353 5.805 3.889 5.096 3.813 5.327 3.336 openness (2.75)∗∗ (0.90)∗∗ (2.13)∗∗∗ (1.39)∗∗∗ (2.10)∗∗ (2.82) (2.15)∗∗∗ (1.35)∗∗∗ (1.81)∗∗∗ (1.31)∗∗∗ (2.04)∗∗∗ (1.47)∗∗ Stock market -0.328 0.161 turnover (0.27) (0.36) Domestic 0.853 7.588 credit/GDP (0.90) (3.85)∗∗ Reserves -1.828 1.286 -1.494 1.172 -2.611 6.509 -2.506 1.063 -1.279 1.024 -1.405 1.028 (1.12) (0.91) (0.52)∗∗∗ (0.28)∗∗∗ (0.68)∗∗∗ (2.37)∗∗∗ (0.61)∗∗∗ (0.79) (0.43)∗∗∗ (0.26)∗∗∗ (0.47)∗∗∗ . Economic 0.397 1.001 0.671 0.549 0.727 -0.665 1.040 1.074 0.588 -1.100 0.780 -0.012 risk (0.81) (0.39)∗∗∗ (0.39)∗ (0.39) (0.27)∗∗∗ (0.66) (0.36)∗∗∗ (0.26)∗∗∗ (0.33)∗ (0.72) (0.32)∗∗ (0.40) Financial 0.133 2.508 risk (0.40) (0.89)∗∗∗ Corporate -3.831 -5.453 bond issuance (4.16) (6.45) Sovereign bond 6.697 -22.848 rating (10.26) (12.98)∗ Host country characteristics GDP 1.456 1.879 4.894 0.874 2.163 0.696 0.712 1.113 3.774 0.621 4.212 0.829 (1.99) (0.74)∗∗ (2.51)∗ (0.40)∗∗ (0.42)∗ (0.52)∗∗ (1.96)∗ (0.36)∗ (2.25)∗ (0.39)∗∗ 14 GDP per capita -0.600 0.419 -1.462 1.054 -1.124 0.584 -2.653 0.209 -1.209 1.447 -1.363 1.002 (0.73) (0.29) (0.59)∗∗ (0.34)∗∗∗ (0.43)∗∗∗ (0.54) (0.99)∗∗∗ (0.27) (0.50)∗∗ (0.47)∗∗∗ (0.53)∗∗ (0.33)∗∗∗ Wages -0.000 -0.000 (0.00) (0.00)∗ Trade 1.474 -0.473 2.431 -2.218 1.587 -4.241 2.282 -1.194 2.010 -2.304 2.212 -2.062 openness (1.21) (0.58) (0.88)∗∗∗ (1.03)∗∗ (0.75)∗∗ (3.00) (0.95)∗∗ (0.90) (0.73)∗∗∗ (0.95)∗∗ (0.79)∗∗∗ (0.89)∗∗ Stock market -0.321 -0.114 turnover (0.23) (0.45) Domestic 1.240 -0.014 credit/GDP (0.60)∗∗ (0.01) Patents -8.983 -2.200 -6.603 -2.999 -6.948 -3.939 -6.275 -2.443 -6.716 -2.853 granted (4.58)∗∗ (0.89)∗∗ (2.93)∗∗ (0.99)∗∗∗ (3.17)∗∗ (1.23)∗∗∗ (2.68)∗∗ (0.74)∗∗∗ (2.89)∗∗ (0.90)∗∗∗ R&D/GDP -0.002 -0.020 (0.03) (0.02) Reserves 2.480 -0.965 -0.851 -0.620 0.198 -0.811 0.938 -1.194 -0.690 -0.154 -0.764 -0.590 (2.57) (0.33)∗∗∗ (0.32)∗∗∗ (0.35)∗ (0.91) (0.32)∗∗ (1.20) (0.42)∗∗∗ (0.28)∗∗ (0.35) (0.25)∗∗∗ (0.33)∗ Economic -0.383 -0.228 0.407 -1.441 0.052 -2.744 0.367 0.649 0.618 -0.578 0.469 -1.269 risk (0.85) (1.25) (0.43) (0.81)∗ (0.38) (2.50) (0.29) (1.09) (0.37)∗ (0.80) (0.38) (0.83) Financial -0.294 -2.971 risk (0.26) (1.19)∗∗ Country-pair characteristics Distance -1.324 2.868 -1.667 2.028 -1.388 0.637 -1.497 0.950 -1.242 1.609 -1.343 2.418 (0.88) (0.97)∗∗∗ (0.68)∗∗ (0.81)∗∗ (0.62)∗∗ (1.41) (0.59)∗∗ (0.75) (0.54)∗∗ (0.78)∗∗ (0.61)∗∗ (0.87)∗∗∗ Bilateral 3.073 0.662 2.599 0.546 2.582 0.453 2.777 0.534 2.583 0.549 2.596 0.547 trade flows (1.04)∗∗∗ (0.16)∗∗∗ (0.92)∗∗∗ (0.17)∗∗∗ (0.98)∗∗∗ (0.13)∗∗∗ (0.95)∗∗∗ (0.18)∗∗∗ (0.90)∗∗∗ (0.17)∗∗∗ (0.92)∗∗∗ (0.17)∗∗∗ Adjusted R2 0.303 0.303 0.280 0.280 0.287 0.287 0.284 0.284 0.276 0.276 0.277 0.277 F 27.401∗∗∗ 27.401∗∗∗ 24.865∗∗∗ 24.865∗∗∗ 26.158∗∗∗ 26.158∗∗∗ 29.885∗∗∗ 29.885∗∗∗ 29.361∗∗∗ 29.361∗∗∗ N 14,326 14,326 30,197 30,197 29,479 29,479 23,617 23,617 37,909 37,909 34,211 34,211 † Heteroskedasticity and autocorrelation-robust standard errors, clustered by time and country-pair, are reported in parentheses. Global controls and a constant term were included in the regressions, but not reported. ∗ indicates signi�cance at 10 percent level, ∗∗ indicates signi�cance at 5 percent level, and ∗∗∗ indicates signi�cance at 1 percent level. as a measure of the overall level of development of the host country—could reflect the overall quality of acquisition targets in a country, or perhaps the extent to which property rights of foreign entities are respected. While we have sought to account for these additional intervening effects via our controls, it is worthwhile including factor prices directly into the benchmark. This is done in column (R1 ), where we have introduced total wages paid to employees in the manufacturing sector as an additional measure of factor price differentials. The negative (and signi�cant) coefficient further corroborates the factor proportions hypothesis, although the magnitude of the effect is quantitatively small (a reduction of one advanced economy M&A deal requires an increase in host wages of twenty percent). To further explore the robustness of the proximity-concentration �nding, we consider two perturbations to the benchmark: in column (R2 ), we substitute stock market capitalization with the stock market turnover ratio; and in column (R3 ), we add domestic credit to the private sector (as a share of GDP) to further approximate the importance of domestic �nancial depth.17 As was the case in the benchmark, �nancial depth (whether in the home or host country), when signi�cant, is positively associated with greater M&A (stock market turnover was statistically insigni�cant). An alternative measure of host-country innovative capacity is expenditure on R&D, as a share of GDP. When we substitute the patents measure of the benchmark—shown in column (R4 )—the sample size decreases signi�cantly, but the negative coefficient fails to reverse (al- though in this case it is insigni�cant).18 Finally, we consider supplementing the political economy variables with two alternatives. First, we include �nancial instead of political risk, as reported in column (R5 ). Home economies that experience lower levels of �nancial risk tend to increase their acquisitions in advanced markets—perhaps, as before, to meet diversi�cation objectives—and the converse holds true for host economies: less risky Northern markets may attract less Southern M&A, perhaps because they offer a less attractive risk-return reward. In the �nal column, (R6 ), we introduce corporate bond issuance and sovereign risk of the home economy as additional factors that may influence cross-border M&A. These appear to have little effect, although the coefficient for sovereign risk is marginally (and negatively) signi�cant. In all of these speci�cations, the coefficients for the other main variables of interest remain largely unchanged. The second set of robustness checks that we consider are different estimation strategies for the benchmark. As before, we utilize the parsimonious model (B2 ) for our analyses. Table 3 reports three alternative estimation methods that we consider. In column (E1 ), we substitute 17 Another alternative could be to substitute the de facto measure of �nancial openness with a de jure one, such as the Chinn & Ito (2008) index. Doing so changes the signi�cance of the coefficient on �nancial openness for advanced economy acquisitions (it becomes negative and signi�cant), results analogous to the �nding for trade openness. The other coefficients are qualitatively unchanged, and we do not report this speci�cation, although these are available on request. 18 We also considered, but do not report, the share of researchers in the population as yet another measure of innovative capacity. Again, the qualitative results remain, and details are available on request. 15 two-dimensional standard errors clustering with two-way �xed effects (by country-pair and time), along with Huber-White robust standard errors. In column (E2 ), we introduce three- way �xed effects (by each respective country and time), and in the third column, (E3 ), we use a random effects (RE) model with robust standard errors and errors clustered by country-pair. The �nal column (E4 ) apples seemingly unrelated regressions (SUR) with correlated country- pairs and a common AR(1) error.19 The �rst two FE approaches allow for ever-greater levels of unobserved heterogeneity, while the latter two error components models accept somewhat less clustering in the error structure, in exchange for greater efficiency in estimation (while still correcting standard errors for the panel nature of the data); the latter approach has also been applies by others in the literature (such as, for example, Head & Ries (2008)). It is evident that the main results that obtained from our benchmark speci�cations remain unaffected by these alternative estimation approaches. It is useful to recognize that this is in spite of the far greater heterogeneity that is afforded by the �xed effects estimators (E1 ) and (E2 ). Indeed, the qualitatively consistent results reported in Table 3 suggest that the simpler (and more efficient) OLS estimation employed in our benchmark was a reasonable choice. 5.3 Will M&A flows change substantially in the aftermath of the 2008 crisis? As discussed in Section 2, cross-border M&A flows by emerging market �rms fell substantially during the later crisis period, and in the year following. The natural question that arises is whether such flows are likely to change substantially as a result of the crisis, especially with anticipated slower growth in advanced economies. Using the model developed in Section 4,20 we project M&A deal trends for 2010 through 2025. Obviously, forecasts of most independent variables were not available. We retained 2009 values for all but GDP and net international investment position (IIP)-related variables (speci�cally, GDP, GDP per capita, GDP growth, and net IIP, used to infer reserve holdings), and simulated the model using these assumptions for emerging and advanced aggregates. The main assumptions are summarized in Table 4 for two scenarios: a baseline where emerging economies are assumed to grow, on average, twice as fast as advanced economies; and a high- growth scenario for emerging markets where the largest emerging countries (China, India, and Russia) grow faster than the rest of the large emerging economies (such as Brazil, Indonesia, and Korea). The results of the two scenarios are summarized in Figure 3. Future cross-border deals are likely to grow at a sustained, albeit slower, pace. Projections along the baseline scenario suggest that the pace of cross-border deal growth is likely to slow from the 14.3 percent annual growth rate recorded between 1998 and 2008, to an average of 8.9 percent annual growth over 2010– 19 Variants of this feasible generalized least squares approach, for example by allowing panel-speci�c errors or independent country-pairs, did not qualitatively change the results. 20 In contrast to the regression analyses, we utilize a slightly different speci�cation (B3 ) here in order to minimize data demands while maximizing the historical �t of the model. Details of the model used for the projections are detailed in the annex. 16 Table 3: Robustness regressions for alternative estimation methods for number of cross-border outbound M&A investments from emerging economies, 1997–2009 E1 E2 E3 E4 to EM to AD to EM to AD to EM to AD to EM to AD Home country characteristics GDP -5.448 21.707 -3.838 2.067 1.661 11.670 5.082 6.794 (4.48) (8.49)∗∗ (2.31)∗ (2.37) (1.85) (3.47)∗∗∗ (0.57)∗∗∗ (0.26)∗∗∗ GDP per 4.723 2.315 3.144 2.990 1.340 0.614 -0.472 -0.056 capita (2.35)∗∗ (1.95) (1.03)∗∗∗ (1.05)∗∗∗ (0.94) (0.87) (0.10)∗∗∗ (0.05) Trade 2.858 0.722 2.353 0.977 3.966 2.763 4.562 3.282 openness (2.43) (1.07) (0.82)∗∗∗ (0.85) (1.80)∗∗ (1.14)∗∗ (0.60)∗∗∗ (0.06)∗∗∗ Reserves 0.144 -0.231 -0.787 1.608 -0.648 0.652 -0.749 1.663 (0.27) (1.19) (0.25)∗∗∗ (0.28)∗∗∗ (0.30)∗∗ (0.93) (0.12)∗∗∗ (0.20)∗∗∗ Economic 0.048 -0.330 -0.081 0.046 0.182 -0.146 0.481 0.529 risk (0.17) (0.28) (0.33) (0.42) (0.20) (0.28) (0.16)∗∗∗ (0.11)∗∗∗ Host country characteristics GDP 19.805 2.425 13.623 1.847 9.287 0.896 3.143 0.811 (10.97)∗ (2.22) (2.64)∗∗∗ (0.68)∗∗∗ (4.93)∗ (0.50)∗ (0.77)∗∗∗ (0.77)∗∗∗ GDP per -0.370 1.036 -1.439 1.041 -1.311 0.691 -1.518 0.704 capita (1.64) (0.61)∗ (1.39) (0.88) (0.77)∗ (0.29)∗∗ (0.40)∗∗∗ (0.31)∗∗ Trade 2.453 -3.639 2.058 -4.144 2.945 -1.672 1.712 -2.194 openness (1.35)∗ (1.62)∗∗ (0.94)∗∗ (2.27)∗ (1.10)∗∗∗ (0.96)∗ (0.45)∗∗∗ (0.93)∗∗ Patents -3.511 0.842 -5.142 0.253 -4.028 -2.223 -3.111 -2.892 granted (1.60)∗∗ (1.38) (1.42)∗∗∗ (2.06) (1.90)∗∗ (0.85)∗∗∗ (0.89)∗∗∗ (0.23)∗∗∗ Reserves -1.934 0.065 -2.005 0.062 -1.059 -0.012 -0.909 -0.722 (0.96)∗∗ (0.20) (0.27)∗∗∗ (0.24) (0.46)∗∗ (0.19) (0.14)∗∗∗ (0.11)∗∗∗ Economic -0.350 -0.038 -0.267 0.533 -0.220 0.011 0.476 0.040 risk (0.36) (0.75) (0.33) (0.91) (0.34) (0.70) (0.12)∗∗∗ (0.46) Country-pair characteristics Distance -2.942 -2.131 -2.150 1.747 -2.649 2.283 (0.24)∗∗∗ (0.44)∗∗∗ (0.51)∗∗∗ (0.59)∗∗∗ (0.42)∗∗∗ (0.42)∗∗∗ Bilateral 1.310 0.493 2.510 0.541 1.540 0.527 1.642 0.587 trade flows (0.62)∗∗ (0.09)∗∗∗ (0.03)∗∗∗ (0.01)∗∗∗ (0.69)∗∗ (0.09)∗∗∗ (0.11)∗∗∗ (0.04)∗∗∗ Adjusted R2 0.096 0.096 0.207 0.207 0.263 0.263 R2 (within) 0.097 0.097 0.093 0.093 R2 (between) 0.310 0.310 F 6.767∗∗∗ 6.767∗∗∗ 251.5∗∗∗ 251.5∗∗∗ Wald χ2 279.1∗∗∗ 279.1∗∗∗ 10,182.5∗∗∗ 10,182.5∗∗∗ Estimator 2-way FE 2-way FE 3-way FE 3-way FE RE RE SUR-PCSE SUR-PCSE N 37,909 37,909 37,909 37,909 37,909 37,909 37,909 37,909 † Heteroskedasticity and autocorrelation-robust standard errors (all speci�cations), clustered by country-pair (third spec- i�cation), are reported in parentheses. Global controls and a constant term were included in the regressions, but not reported. ∗ indicates signi�cance at 10 percent level, ∗∗ indicates signi�cance at 5 percent level, and ∗∗∗ indicates signi�cance at 1 percent level. 17 Table 4: Growth and net IIP assumptions for emerging and advanced economies, by scenario, 2010–25† Baseline High growth 2010 2025 Average 2010 2025 Average GDP growth (%) Emerging 6.2 3.9 4.7 6.2 4.4 4.9 Advanced 2.2 1.5 2.3 2.2 1.5 2.3 GDP (USD trillion) Emerging 21.6 43.7 21.6 44.8 Advanced 39.5 56.2 39.5 56.2 GDP per capita growth (%) Emerging 5.6 3.6 4.2 5.6 4.0 4.4 Advanced 1.7 1.3 1.9 1.7 1.3 1.9 Net IIP (USD trillion) Emerging 2.7 15.2 2.7 15.2 Advanced -2.4 -9.8 -2.4 -9.8 * GDP levels are measured in constant 2009 U.S. dollars. 20 (9.0 percent in the high-growth scenario), and to an average of 6.7 percent annual growth between 2020 and 2025 (7.0 percent in the high-growth scenario). Consistent with the past decade, the expansion of �nancial globalization, as measured by the rate of growth of cross-border deals, is expected to exceed that of real economic growth: Growth in cross-border deals is likely to outpace expected emerging-market GDP annual growth rates of 4.9 percent over 2010–20, and 4.1 percent over 2020–25. This expected growth in cross-border deals echoes a global trend of �nancial growth generally exceeding growth in real economic variables. Preliminary evidence from M&A activity for 2010 and 2011 is supportive of this reasonably rapid recovery from the crisis. Indeed, as evident in Figure 1, deal activity by emerging market �rms in 2010 has already exceeded pre-crisis peaks, and M&A activity in 2011 appears well on track for the full recovery, in 2013, implied by our projections. This is also supported by country-level evidence. Outbound cross-border M&A deals concluded by Chinese �rms, for example, reached 744 deals in 2010 and 909 deals in 2011, levels comparable to (and in excess of) the pre-crisis peak of 760 deals, attained in 2007. 6 Conclusion In this paper, we have explored the factors associated with bilateral M&A activity by emerging market �rms. Our main �nding is that these �rms’ decisions to pursue acquisition targets depend critically on whether they are investing in advanced or emerging market targets. Southern acquisitions tend to be located in countries with lower levels of per capita income, which likely reflects a desire to take advantage of lower wage costs in those countries. Acquisitions of 18 Number of deals Projection Number of deals Projection 5,000 5,000 4,500 4,500 4,000 4,000 3,500 3,500 3,000 3,000 2,500 2,500 2,000 2,000 1,500 1,500 1,000 1,000 500 500 0 0 1998 2001 2004 2007 2010 2013 2016 2019 2022 2025 1998 2001 2004 2007 2010 2013 2016 2019 2022 2025 Source: Authors' calculations Source: Authors' calculations (a) Baseline scenario (b) High-growth scenario Figure 3: Model projection (hollow and dotted lines) and actual historical data (solid line) for outbound cross-border M&A deals by �rms in emerging-market economies, 1998–2025, for baseline (left panel) and high-growth (right panel) scenarios. Shaded areas indicate the projec- tion period. Forward projections based on growth and net IIP assumptions for emerging and advanced economies. Northern targets, in contrast, occur in greater frequency when these host countries are more closed to trade, which suggests that such deals may be due to an implicit tradeoff that favors siting production directly in the country, rather than exporting to it. Another important insight that emerges from the analysis is that �rms appear to seek to diversify their equity holdings through acquisitions. More speci�cally, while emerging economy acquirers are likely to be located in economically less volatile economies, they tend to seek out targets in advanced economies that exhibit both lower levels of political risk (to insure the protection of their investments), but higher levels of economic and �nancial risk (possibly because such economies can offer better returns). The extent to which such diversi�cation occurs is also materially affected by economic policies, such as those governing bilateral investment or �nancial access. In light of this, emerging market �rms are likely to press for future economic policies that will strengthen investment climates both at home and abroad. In doing so, emerging market �rms can act as catalysts that spur increased integration of developing countries into the global economy, since enhanced integration offers additional support for open trading and investment regimes. But these �rms will also serve as a growing source of global competition, especially when they invest in other emerging economies. Emerging market acquisitions are increasingly driven by resource- and efficiency-seeking motives—motives traditionally considered the pre- serve of �rms based in advanced countries—and in making such cross-border investments, these �rms will also challenge advanced-country �rms’ preeminence in industrial production. Such competition will, in the longer run, drive global factor price equalization in general, and wage convergence in particular. Countries can support such positive competition by enhancing �nan- cial access within their countries, which is also positively associated with cross-border M&A. 19 The slow post-crisis recovery in the developed world, coupled with the relatively rapid recovery in the developing one, has underscored the future economic potential of emerging markets. Projections of post-crisis M&A volume by emerging market �rms suggest that future M&A activity, while moderating somewhat, is likely to remain fairly robust. Emerging market �rms are thus likely to be at the forefront of this process of global economic convergence, and are fast becoming a potent force for globalization in their own right. Future research in this area will do well to study how the economic behavior of such globalized emerging market multinationals may differ from those of advanced country corporations, beyond their choices in cross-border M&A. References Aizenman, Joshua & Ilan Noy (2007). “Prizes for Basic Research: Human Capital, Economic Might and the Shadow of History�. Journal of Economic Growth 12(3) (September): 261–282 Anand, Jaideep & Andrew K. Delios (2002). “Absolute and Relative Resources as Determinants of International Acquisitions�. Strategic Management Journal 23(2) (February): 119–134 Anderson, James E. & Eric van Wincoop (2004). “Trade Costs�. Journal of Economic Literature 42(3) (Septem- ber): 691–751 Arita, Shawn (forthcoming). “Do Emerging Multinational Enterprises Possess South-South FDI Advantages?� International Journal of Emerging Markets Blonigen, Bruce A. & Jeremy Piger (2011). “Determinants of Foreign Direct Investment�. Working Paper 16704, National Bureau of Economic Research a Braconier, Henrik, Pehr-Johan Norb¨ck & Dieter M. Urban (2005). “Multinational enterprises and wage costs: vertical FDI revisited�. Journal of International Economics 67(2) (December): 446–470 Brainard, S Lael (1997). “An Empirical Assessment of the Proximity-Concentration Trade-off between Multina- tional Sales and Trade�. American Economic Review 87(4) (September): 520–544 Brouthers, Keith D. & Desislava Dikova (2010). “Acquisitions and Real Options: The Green�eld Alternative�. Journal of Management Studies 47(6): 1048–1071. URL http://econpapers.repec.org/RePEc:bla:jomstd: v:47:y:2010:i:6:p:1048-1071 Busse, Matthias & Carsten Hefeker (2007). “Political Risk, Institutions and Foreign Direct Investment�. European Journal of Political Economy 23(2) (June): 397–415 u B¨the, Tim & Helen V. Milner (2008). “The Politics of Foreign Direct Investment into Developing Countries: Increasing FDI through International Trade Agreements?� American Journal of Political Science 52(4) (October): 741–762 Carr, David L., James R. Markusen & Keith E. Maskus (2001). “Estimating the Knowledge-Capital Model of the Multinational Enterprise�. American Economic Review 91(3) (June): 693–708 Chinn, Menzie D. & Hiro Ito (2008). “A New Measure of Financial Openness�. Journal of Comparative Policy Analysis 10(3) (September): 309–322 20 di Giovanni, Julian (2005). “What Drives Capital Flows? The Case of Cross-Border M&A Activity and Financial Deepening�. Journal of International Economics 65(1) (January): 127–149 Ethier, Wilfred J. (1986). “The Multinational Firm�. Quarterly Journal of Economics 101(4) (November): 805–833 Fosfuri, Andrea, Massimo Motta & Thomas Rønde (2001). “Foreign Direct Investment and Spillovers Through Workers’ Mobility�. Journal of International Economics 53(1) (February): 205–222 Gilroy, Bernard Michael & Elmar Lukas (2006). “The Choice Between Green�eld Investment and Cross-Border Acquisition: A Real Option Approach�. Quarterly Review of Economics and Finance 46(3) (July): 447–465 a as s a Havr´nek, Tom´˘ & Zuzana I˘ov´ (2012). “Estimating Vertical Spillovers from FDI: Why Results Vary and What the True Effect Is�. Journal of International Economics 85(2) (November): 234–244 Head, C. Keith & John C. Ries (2008). “FDI as an Outcome of the Market for Corporate Control: Theory and Evidence�. Journal of International Economics 74(1) (January): 2–20 Helpman, Elhanan & Paul R. Krugman (1985). Market Structure and Foreign Trade: Increasing Returns, Im- perfect Competition,and the International Economy. Cambridge, MA: MIT Press Helpman, Elhanan, Marc J. Melitz & Stephen R. Yeaple (2004). “Export Versus FDI with Heterogeneous Firms�. American Economic Review 94(1) (March): 300–316 Horstmann, Ignatius J. & James R. Markusen (1992). “Endogenous Market Structures in International Rrade (Natura Facit Saltum)�. Journal of International Economics 32(1–2) (February): 109–129 Loungani, Prakash, Ashoka Mody & Assaf Razin (2002). “The Global Disconnect: The Role of Transactional Distance and Scale Economies in Gravity Equations�. Scottish Journal of Political Economy 49(5) (December): 526–543 Lucas, Robert E., Jr. (1990). “Why Doesn’t Capital Flow from Rich to Poor Countries?� American Economic Review 80(2) (May): 92–96 Makino, Shige, Chung-Ming Lau & Rhy-Song Yeh (2002). “Asset-Exploitation Versus Asset-Seeking: Implica- tions for Location Choice of Foreign Direct Investment from Newly Industrialized Economies�. Journal of International Business Studies 33(3) (September): 403–421 Markusen, James R. (2002). Multinational Firms and the Theory of International Trade. Cambridge, MA: MIT Press Markusen, James R. & Anthony J. Venables (1998). “Multinational Firms and the New Trade Theory�. Journal of International Economics 46(2) (December): 183–203 e M´on, Pierre-Guillaume & Anne-France Delannay (2006). “The Impact of European Integration on the Nineties’ Wave of Mergers and Acquisitions�. DULBEA Working Papers 06-12.RS, Brussels, Belgium: Universite Libre de Bruxelles Neumayer, Eric & Laura A. Spess (2005). “Do Bilateral Investment Treaties Increase Foreign Direct Investment to Developing Countries?� World Development 33(10) (October): 1567–1585 ıguez-Clare, Andr´s (1996). “Multinationals, Linkages, and Economic Development�. American Economic Rodr´ e Review 86(4) (September): 852–873 21 Rossi, Stefano & Paolo F. Volpin (2004). “Cross-Country Determinants of Mergers and Acquisitions�. Journal of Financial Economics 74(2) (November): 277–304 World Bank (2011). Global Development Horizons 2011: Multipolarity: A New Global Economy. Washington, DC: The World Bank 22 Technical Appendix Projection Model Details Our M&A growth projections are based on a modi�ed version of the model (1), and is designed to minimize data requirements for forward variables while maximizing the �t of the model to historical data. We use speci�cation (B3 ) in Table 1 as a starting point, and introduce a lagged dependent variable to the regressors. The model is then estimated with the full set of explanatory variables, and insigni�cant variables are then sequentially dropped. The process is repeated until the most parsimonious speci�cation is reached. Due to data limitations, a smaller panel is used, which includes 53 countries for which full data are available. The country-speci�c variables are further simpli�ed by classifying host countries two groups, either advanced or emerging. The projections are based on the two sets of assumptions regarding GDP, GDP per capita, and net IIP detailed in Table 4. These assumptions correspond to a growth scenario where all countries grow at a rate consistent with their potential output, allowing for short-run deviations in the short term (from 2011–12, where we use the World Bank’s forecasts published in the Global Economic Prospects report). Per capita calculations are obtained by supplementing the growth- implied GDP levels with UN population projections. Reserve holdings are assumed to grow as a constant fraction of the net IIP position. A number of variables, such as corporate bonds issues, are assumed to grow at historical rates (for the 1997–2008 period). Other variables, such as economic risk, political risk, and participation in global trade are assumed to remain constant throughout the projection period. The �tted model is generally robust to small deviations in most of the independent variables, but is sensitive to assumptions about reserve holdings growth. 23 Additional Tables Table A.1: Countries included in the database† Emerging economies Algeria Ghana Peru Argentina Guatemala Philippines Azerbaijan Hungary Poland Bahamas, The India Qatar Bahrain Indonesia Romania Barbados Jamaica Russian Federation Belarus Jordan Saudi Arabia Brazil Kazakhstan Singapore Bulgaria Kenya South Africa Chile Kuwait Korea, Rep. of China Latvia Sri Lanka Colombia Lebanon Thailand Costa Rica Lithuania Trinidad and Tobago Croatia Malaysia Turkey Czech Republic Mexico Ukraine Dominican Republic Mongolia UAE Ecuador Morocco Uruguay Egypt, Arab Rep. Nigeria Venezuela, RB El Salvador Oman Vietnam Estonia Pakistan Georgia Panama † The de�nition of emerging economies used in the paper were chosen on the basis of markets traditionally classi�ed as emerging by invest- ment banks, and to illustrate the economies that were distinct from the historically advanced economies of North America, Western Eu- rope, Japan, and Oceania. China data aggregate the mainland and the special administrative regions of Hong Kong and Macau. 24 Table A.2: Data description and sources Variable Description Source 10-year Treasury rate 10-year constant maturity Treasury bond rate J.P. Morgan Agricultural price index Agricultural commodity prices, weighted index World Bank DECPG BITs Existence of bilateral investment treaty between two countries UNCTAD Capital openness Total private capital inflows and outflows (% GDP) World Bank WDI Corporate bond issuance Total number of corporate bonds issued Dealogic DCM Analytics Cross-border M&A deal Acquisition of an equity stake of ¿ 1 percent in target �rm located in country other than country of domicile of acquirer Thompson-Reuters SDC Platinum Distance Bilateral geodesic distance between national capitals CEPII Domestic credit/GDP Domestic credit to private sector (% GDP) World Bank WDI Economic risk ICRG economic risk rating, annual average (higher values indicate lower risk) Political Risk Services Energy price index Energy commodity prices, weighted index Goldman Sachs 25 Financial risk ICRG �nancial risk rating, annual average (higher values indicate lower risk) Political Risk Services GDP Gross domestic product, in 2000 constant U.S. dollars World Bank WDI GDP growth Growth of real gross domestic product World Bank WDI GDP per capita Per capita gross domestic product, in 2000 constant U.S. dollars World Bank WDI Patents granted Total patents granted per million people in population WIPO Political risk ICRG political risk rating, annual average (higher values indicate lower risk) Political Risk Services Population Total population World Bank WDI R&D/GDP R&D expenditure (% GDP) World Bank WDI Reserves Total international reserves, end of year IMF IFS Sovereign bond rating Long-term ratings on sovereign bonds J.P. Morgan Stock market capitalization Total market capitalization of listed companies MSCI Stock market turnover Stocks traded, turnover ratio (%) MSCI Wages Home total manufacturing wages/salaries to employees (USD) World Bank WDI Table A.3: Summary statistics for major variables of interest† To advanced To emerging N Mean Std. Dev. Min Max N Mean Std. Dev. Min Max M&A deals 19,825 0.286 1.785 0 65 43,030 0.158 1.616 0 98 GDP, home 18,200 0.018 0.035 0.000 0.347 39,481 0.018 0.035 0.000 0.347 GDP, host 19,032 0.156 0.242 0.003 1.323 41,470 0.019 0.036 0.000 0.347 GDP per capita, home 18,200 0.056 0.058 0.004 0.336 39,481 0.056 0.058 0.004 0.336 GDP per capita, host 19,032 0.381 0.123 0.160 0.916 41,470 0.054 0.056 0.002 0.336 Trade openness, home 18,200 0.089 0.055 0.016 0.438 39,481 0.089 0.055 0.016 0.438 26 Trade openness, host 19,032 0.082 0.053 0.019 0.324 41,470 0.091 0.058 0.016 0.438 Patents granted, host 16,653 0.046 0.038 0.000 0.199 39,897 0.005 0.018 0.000 0.221 Reserves, home 19,175 0.038 0.145 0.000 2.418 41,623 0.038 0.145 0.000 2.418 Reserves, host 19,703 0.052 0.134 0.000 1.024 41,223 0.041 0.153 0.000 2.418 Economic risk, home 18,975 0.359 0.051 0.200 0.493 41,180 0.359 0.051 0.200 0.493 Economic risk, host 19,825 0.410 0.031 0.303 0.484 40,122 0.351 0.057 0.098 0.493 Distance 19,825 0.072 0.042 0.001 0.189 43,030 0.078 0.048 0.001 0.198 Bilateral trade flows 16,735 0.180 0.995 0.000 35.500 35,495 0.051 0.292 0.000 15.664 BITs 19,825 0.041 0.049 0.000 0.100 43,030 0.028 0.045 0.000 0.100 † Variables were rescaled for regression purposes and reported values may not reflect actual magnitudes.