WPS6771 Policy Research Working Paper 6771 Surges and Stops in FDI Flows to Developing Countries Does the Mode of Entry Make a Difference? Martijn J. Burger Elena I. Ianchovichina The World Bank Middle East and North Africa Region Office of the Chief Economist February 2014 Policy Research Working Paper 6771 Abstract This paper investigates the factors associated with kind, while decline in global economic growth and a foreign direct investment “surges” and “stops,” defined surge in the preceding year are the only predictors of a as sharp increases and decreases, respectively, of gross stop. Greenfield-led surges and stops are more likely in foreign direct investment inflows to the developing low-income and resource-rich countries than elsewhere. world and differentiated based on whether these events Global growth, financial openness, and domestic are led by waves in greenfield investments or mergers economic and financial instability enable mergers and and acquisitions. Greenfield-led surges and stops occur acquisitions-led surges. These results differ from those more frequently than mergers and acquisitions-led ones in the literature on surges and stops and are particularly and different factors are associated with the onset of the relevant in countries where foreign direct investments two types of events. Global liquidity is the only factor dominate capital flows. significantly associated with a surge, regardless of its This paper is a product of the Office of the Chief Economist, Middle East and North Africa Region. 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 eianchovichina@worldbank.org or mburger@ese.eur.nl. 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 Surges and Stops in FDI Flows to Developing Countries: Does the Mode of Entry Make a Difference? ∗ Martijn J. Burger ♦ and Elena I. Ianchovichina ♣ Key words: foreign direct investment (FDI), mode of entry, greenfield (GF) investment, mergers and acquisitions (M&A), capital flows, surges, stops, developing countries JEL codes: F21, F23, F43, F44 Sector Board: EPOL ∗ We would like to thank Sergio Schmukler for providing extensive comments on several drafts of this paper and Bob Rijkers for his useful suggestions on an early version of this paper. The research was supported by funding through the World Bank’s Knowledge for Change Program. The findings, interpretations, and conclusions expressed in this paper are entirely ours and should not be attributed to 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. ♦ Martijn J. Burger is assistant professor at the Department of Applied Economics, Erasmus University Rotterdam, Tinbergen Institute and Erasmus Research Institute of Management (ERIM), P.O. Box 1738, 3000 DR Rotterdam, the Netherlands. Tel: +31 (0)10 4089579. Fax: +31 (0)10 4089141. E-mail: mburger@ese.eur.nl. URL: http://www.mjburger.net. ♣ Elena Ianchovichina is lead economist at the Chief Economist Office, Middle East and North Africa Region, World Bank, 1818 H Street NW, Washington, DC 20433, USA, Tel: +1 202 458 8910, E-mail: eianchovichina@worldbank.org . 1. Introduction Over the past three decades the world has witnessed a dramatic rise in foreign direct investment (FDI) flows. Prior to 1985 the growth rate of FDI flows was comparable to the growth rates of world trade and output, but after that FDI flows grew at a much faster pace than either world trade or world output. The growing importance of FDI flows has spurred burgeoning literatures on the causes and effects of FDI in international and financial economics, international business, and economic geography. 1 Still, the rise of FDI has not proceeded in a smooth fashion. Since the 1980s, there have been distinct waves of FDI with corresponding surges and stops, especially in developed countries (Andrade et al., 2001). Although developed countries have generally received more FDI flows than developing ones and host the majority of the inward FDI stock, developing countries have caught up (UNCTAD, 2011). In 2010, for the first time the developing world received more FDI flows than their developed counterparts, but some developing countries were more successful in attracting FDI than others. The distribution of inward FDI flows has been persistently disproportionate in that investments have concentrated within a limited number of developing countries (see also Noorbakhsh et al., 2001; UNCTAD, 2011). This paper investigates the nature and determinants of FDI “stops” and “surges”, defined as sharp increases and decreases, respectively, of gross FDI flows 2 to the developing world, initiated by foreign investors and differentiated based on whether these events are led by waves in greenfield investments (GF) or mergers and acquisitions 1 See Barba Navaretti and Venables (2004), Blonigen (2005), and Brakman et al. (2006) for overviews of these literatures. 2 These flows include equity capital, reinvestment of earnings, other long-term capital, and short-term capital less disinvestments. 2 (M&A). In particular, we explore whether the mode of entry affects the incidence and determinants of FDI surges and stops. Several studies compare different types of financial flow events (e.g., Sula, 2010; Cardarelli et al., 2010; Agosin and Huaita, 2012; Forbes and Warnock, 2012; Furceri et al., 2012; Ghosh et al., 2012), but to our knowledge, none looks separately at the incidence and determinants of GF-led and M&A-led stops and surges in FDI flows to the developing world. The distinction is important for a number of reasons. The two types of FDI flows occur for different reasons and have different effects, characteristics, and incidence. 3 GF investment inflows finance the construction of new facilities which augment the stock of physical capital and thus expand the production capacity in countries, increasing market competition and employment (Mattoo et al., 2004). M&As predominantly involve a change in ownership via the purchase of existing assets, although they might result in a more efficient allocation of resources (Kim, 2009; Wang and Wong, 2009; Harms and Méon, 2012). 4 Importantly, whereas most global FDI waves have been associated with an increase in mergers and acquisitions (M&A) (Brakman et al., 2006), the extent to which greenfield (GF) investments contribute to surges and stops in FDI in developing countries remains unclear. It is, however, important to explore this question because GF investments dominate FDI flows to the developing world (Markusen and Stähler, 2011; UNCTAD, 2012), especially in resource-rich countries where local companies often have privileged access to the resources and, hence, host country government policies encourage GF investments into joint ventures. It is also true in low-income countries where large price differentials between the home and host countries and the absence of attractive 3 However, both modes are associated with increases in aggregate productivity. 4 M&A sales create rents for the previous owners which are not necessarily channeled into new investments (Harms and Méon, 2012). Yet, M&As might rely more on local and regional supplier networks than multinationals entering through greenfield investments (Wes and Lankes, 2001). 3 corporate assets make GF FDI more likely as an entry mode. Our focus on the developing world is also motivated by the study of Blonigen and Wang (2005) who show that the determinants of FDI flows to developing countries differ from those to developed ones. Understanding the role played by the mode of entry in the incidence of FDI surges and stops is valuable in the context of rising FDI flows to the developing world. These types of flows have become an important and sometimes dominant source of finance in developing countries, so there is a concern that economic growth might be harmed in countries exposed to extreme fluctuations of either type of these flows (Lensink and Morrissey, 2006; Herzer, 2012). There is also the long-standing concern that sudden stops and surges in foreign capital flows might contribute to and arise as a result of macroeconomic volatility (Calvo et al., 2006) and crises (Reinhart and Reinhart, 2009; Furceri et al., 2012) as well as complicate macroeconomic management in developing economies. Abiad et al. (2011) and Cowan and Raddatz (2011), for instance, point to a connection between sudden stops and credit market imperfections. Gall et al. (2013) find that high past exposure to FDI flows may impede an economy’s ability to respond to sudden stops in FDI, especially in industries relying on external financing, and more so in countries with less developed financial markets. The paper is related to the broader literature on net capital flows, which are volatile, pro- cyclical, and, during crises, prone to large “sudden stops,” defined as sharp slowdowns in net capital inflows. The literature originated with Calvo (1998) and broadened to include different conditions as well as the opposite events such as “surges”, defined as sharp 4 increases in net capital flows (Reinhart and Reinhart, 2009). 5 However, this paper studies the behavior of gross FDI flows to developing countries as we are interested in surges and stops due to actions of foreigners. Cowan et al. (2008) and Rothenberg and Warnock (2011) make the point that measures of “sudden stops” constructed from data on net inflows are not able to differentiate between stops that are due to the actions of foreigners and those due to locals fleeing the domestic markets. In addition, Broner et al. (2013) show that gross capital flows are pro-cyclical and are larger and more volatile than net capital flows. While Levchenko and Mauro (2007) show that FDI is the least volatile form of financial flow, when the average size of net or gross flows is taken into account, this paper shows that FDI surges and stops in the developing world are not rare events 6 and therefore are worth an in-depth look. Specifically, the paper contributes to the literature in the following ways. First, we build a database of episodes when foreign investors substantially increase or decrease FDI inflows to a developing country and distinguish between these episodes based on the dominance of the mode of entry. Using this database, which covers the period from 1990 to 2010 and includes 95 developing economies, we then document the incidence of sudden stops and surges by mode of entry, region, and resource status of the receiving economy. Second, we identify the factors associated with FDI surges and stops by mode of entry (i.e. GF-led and M&A-led surges and stops). We show that GF-led and M&A-led extreme events such as surges and stops have different determinants and therefore must be studied separately. 5 Other papers that belong to this literature include, for example, Kaminsky et al. (1998), Levchenko and Mauro (2007), and Mendoza (2010). 6 All developing countries experienced at least one such event during the period of investigation and most countries experienced multiple FDI surges and stops. 5 Our approach yields different results from previous studies on surges and stops in FDI flows which do not differentiate between these events based on the mode of entry (e.g., Dell’Erba and Reinhart, 2012). We show that different factors are associated with the onset of GF-led and M&A-led FDI surges and stops. Global liquidity is the only common predictor of the two types of FDI surges, while a decline in global growth and a FDI surge in the preceding year are the only significant and consistent predictors of FDI stops. GF-led sudden stops and surges are more likely in lower income and resource-rich countries than elsewhere. Policies aimed at increasing financial openness are enablers of M&A-led surges, which are also more likely during periods of global growth and domestic economic and financial instability. The results are also policy relevant as we show that GF-led extreme events occur more frequently than M&A-led ones. 7 Thus, countries relying mostly on GF investments, the more stable type of FDI inflows, are not immune to sudden stops in capital flows and should prepare to withstand them. Knowledge of the factors behind different types of FDI surges and stops can help policy makers in developing countries craft policies to successfully weather such episodes. The remainder of this paper is organized as follows. The next section defines FDI surges and stops and presents information on the types and frequency of such surges. In section 3, we turn attention to the empirical models for predicting GF-led and M&A-led surges and stops, the discussion of the econometric results, and the robustness checks. Section 4 summarizes the findings and offers concluding remarks. 2. Identifying GF-led and M&A-led FDI Surges and Stops Using UNCTAD data on gross FDI inflows from the World Investment Report (UNCTAD, 2011) and building on the work by Calvo et al. (2004), Reinhart and Reinhart 7 However, M&A-led surges are more likely to be short-lived and followed by a stop. 6 (2009), and Forbes and Warnock (2012), we define a surge episode as an increase in inflows in a given year that is more than one standard deviation above the country- specific (five-year rolling) average. The surge episode begins when the FDI-to-GDP ratio increases more than one standard deviation above its rolling mean and ends when the FDI-to-GDP ratio falls below one standard deviation above its rolling mean. In addition, we pose a restriction to the definition of an FDI surge in that the increase in the FDI-to- GDP ratio should fall within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth. This not only ensures that the increase in FDI inflows is substantial, but also that only large surges by international standards are included in our definition of a surge (Ghosh et al., 2012). This approach combines the two main empirical strategies present in the literature on surges and stops. One involves looking at deviations from the mean while the other requires factoring in minimum threshold values. Stops are defined in a symmetric way, with a stop episode defined as a decline in inflows in a given year that is more than one standard deviation below the rolling average. The stop episode starts when the ratio of FDI to GDP declines more than one standard deviation below its rolling mean and ends when the ratio increases above one standard deviation below its mean. We impose similar restrictions on stops as on surges. To identify whether a FDI surge or a stop can be mainly attributed to an increase in M&A activity or GF investments, we use Thompson ONE Source data on M&A inflows, available from 1990 onwards. Following Calderon et al. (2004), Wang and Wong (2009), and Bogach and Noy (2013), GF FDI is defined as the difference between gross FDI and M&A inflows. Using this information, we assess whether a surge in a given year is dominated by an increase in M&A activity or by an increase in GF investments. A surge is M&A-led when more than half of the increase in FDI can be attributed to an increase 7 in M&A activity. Likewise, a surge is GF-led when more than half of the increase in FDI can be attributed to an increase in GF investments. As indicated by Calderon et al. (2004), there are several potential problems with combining aggregate FDI inflow data with M&A data. First, FDI flows are measured on an accruals basis, while M&As are recorded at the time of an announcement or closure of a deal. As such, an FDI surge in a given year, for example, may be attributed to a sharp increase in M&A in the year before. Second, calculating GF FDI as the residual of FDI and M&A may possibly pollute the data with some international transactions that are not GF FDI. We control for these problems by verifying whether a FDI surge can indeed be attributed to either an increase in M&A or GF FDI and correct for anomalies in the data by accounting for the share of M&A activity in the year before the surge. When values for M&A activity are missing, it is assumed that the surge could be attributed to a surge in GF FDI. Exactly the same procedure is applied to identifying GF-led and M&A-led FDI stops. Figure 1 on FDI surges and stops in Algeria illustrates how such events are identified. The left panel of Figure 1 assesses the first condition, namely that we speak of a surge if the FDI-to-GDP ratio increases more than one standard deviation above its five year rolling mean. We see that in Algeria this is the case in several years in the mid1990s and the early and late 2000s. Likewise, a stop is identified when the FDI-to-GDP ratio decreases more than one standard deviation below its five year rolling mean. This is the case in 1992, 2003 and 2010. However, in order to qualify for a surge, the increase should fall within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth, meaning an increase of at least 0.82 percentage points, marked by the top horizontal line in the right panel of Figure 1. Likewise, in order to qualify for a stop, the increase should fall within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth, 8 meaning a decrease of at least 0.55 percentage points, marked by the bottom horizontal line in the right panel of Figure 1. This means that Algeria experienced a FDI surge only in 2001, while in 1999, 2003, and 2010 it experienced a FDI stop. As can also be observed from right panel of Figure 1, Algeria received few M&A flows. Hence, all surges and stops in Algeria are classified as GF-led. Figure 1: Identification of Surge and Stop Episodes in Algeria In total, the 95 developing economies in the sample experienced 264 surge-episodes during the period 1991-2010, of which 207 were led by a surge in GF investments and 57 were dominated by a surge in M&A activity (see Appendix A1). 8 The unconditional probability of experiencing a surge in GF investments and M&A activity was 11.7% and 3.2%, respectively. Almost all countries in the sample (98%) experienced at least one surge. However, whereas the majority of countries (86%) experienced at least one GF-led surge in the period under research, the percentage of countries that experienced at least one M&A-led surge was much lower (42%). 8 Some countries had to be excluded from the empirical analysis because explanatory variables for these countries were not available. 9 Although M&A flows are much more volatile than GF flows, 9 GF-led FDI surges outnumber M&A-led FDI surges in developing countries: around 80% of the FDI surges in developing economies can be attributed to an increase in GF FDI. Regions with either a high share of resource-rich or low-income countries or both such as the Middle East and North Africa and Sub-Saharan Africa, where GF investments represent a large share of FDI flows, have had the highest occurrence of GF-led FDI surges (Table 1). In resource-rich countries, governments encourage GF investments as local firms typically have privileged access to the resources. In low-income economies, large labor cost differentials between the home and host economies make GF FDI more likely as an entry mode. Regions with relatively strong links to global financial markets have had the lowest incidence of GF-led FDI surges and the highest incidence of M&A-led FDI surges. In addition, our analysis suggests that some countries experience more FDI surges than others and FDI surges occur at different times in different developing countries and usually last only a year. They were most prevalent in Europe and Central Asia and the Middle East and North Africa in the mid-2000s, in East Asia and Pacific and Sub-Saharan Africa in the late 1990s and late 2000s, and in Latin America and Caribbean in the mid-1990s. We identify FDI stops in a symmetric way. The 95 developing economies in the sample experienced 282 stop-years during the period 1991-2010, of which 225 were GF-led stops and 57 were M&A-led stops(see also Appendix A1). The unconditional probability of experiencing a GF-led stop was 12.8%, while that for an M&A-led stop was 3.3%. All countries in the sample experienced at least one stop-year and most stops were GF-led. Yet, M&A-led surges are significantly more frequently followed by a stop in the next year (51%) than GF-led surges (28%) (p-value for the Fischer’s exact test < 0.01). This 9 Whereas the average coefficient of variation for M&A flows was 4.00, the coefficient of variation for GF flows was only 0.82. The average coefficient of variation is based on the mean value for the coefficients of variation for all countries in the sample. 10 suggests that M&A-led surges are more likely to be short-lived and followed by a stop than GF-led events. Table 1: Incidence and Types of FDI Surges and Stops in Developing Countries Incidence of % GF-led Incidence of % GF-led Surge Surge Stop Stop East Asia and Pacific 15.5% 72.3% 15.0% 67.6% Europe and Central Asia 16.5% 54.5% 21.9% 65.6% Latin America and Caribbean 15.8% 68.2% 14.5% 67.2% Middle East and North Africa 13.1% 83.3% 15.7% 85.7% South Asia 6.3% 80.0% 8.8% 77.8% Sub-Saharan Africa 15.5% 89.2% 14.1% 92.6% Resource-Rich Economies* 15.3% 87.9% 15.4% 86.7% All Economies in Sample 14.9% 78.4% 15.1% 80.3% * Hydrocarbon and Mineral Rich Countries as defined by IMF Although at the global level the unconditional probability of experiencing a surge is similar to that of a stop, the occurrence of surges varies by region and over time. Stops are more frequent in Europe and Central Asia than in the other world regions and least frequent in lower income developing countries, but differences between different country groups are never statistically significant. 10 As in the case of FDI surges, stops occur at different times in different developing countries and most last only a year. 3. Predicting GF-led and M&A-led FDI Surges and Stops Estimation Approach and Variables We inform the selection of variables that might be associated with a FDI surge or a stop by drawing on the literature on sudden stops and bonanzas. Following this literature, we conjecture that the probability of a GF-led and M&A-led surge or stop depends on three sets of factors – global, contagion (regional), and domestic (Calvo et al., 1996; Fernandez-Arias and Montiel, 1996; Dell’Erba and Reinhardt, 2012; and Forbes and 10 This inference is based on a Fischer’s exact test. 11 Warnock, 2012). Hence, to examine the role of these global, contagion, and domestic factors in the conditional probability of having a GF-led or M&A-led FDI surge or a stop, we estimate the model: ( = 1) = (−1 + ,−1 + −1 ), (1) The variable εit is an indicator of the occurrence of an event in country i and year t and is equal to one of six episode dummy variables defined as follows. The first one, sit, assumes the value 1 if there is a FDI surge, either GF-led or M&A-led one, in country i and year t. In all other cases, sit is 0. The dummy variable, hit, equals 1 if a country i is experiencing a GF-led surge in a given year t and 0 otherwise. Finally, the dummy variable mit is 1 in the case of an M&A-led surge in a given year t and country i, and 0 otherwise. In a similar fashion, three dummy variables for stops are defined: one for GF-led FDI stops, one for M&A-led FDI stops, and one for FDI stops, regardless of their kind (GF-led or M&A- led). G is a vector of variables capturing global factors, R is a vector of variables capturing regional factors, and D is a vector of variables capturing domestic factors. Since surges and stops occur irregularly, (.) is asymmetric and, therefore, we use the complementary logistic regression (see also Forbes and Warnock, 2012) which assumes that (.) is the cumulative distribution function (cdf) of the extreme value distribution. We estimate the model separately for the six types of events; for the regressions on the likelihood of GF-led and M&A-led surges and stops, covariance across surge-years and stop-years is accounted for using seemingly unrelated regression, estimated with clustering of standard errors at the country level. The variables representing domestic stop or surge, GDP per capita, and natural resources are lagged by one year, and the latter two are winsorized at the 1% level. 12 Since economic developments in developed markets, which are the primary source of this type of finance, trigger big fluctuations in FDI flows to developing countries (Aleksynska and Havrylchyk, 2013), we include several global factors, including global risk, global liquidity, and global growth. Global risk is a volatility measure given by the VXO index of the Chicago Board Options Exchange. Global liquidity measures the availability of finance in global markets and is given by the sum of the change in the following two ratios – the ratio of stock market capitalization to GDP and the ratio of domestic private sector credit to GDP (Beck et al., 2000). 11 The size of the financial market is expected to be positively related to the ability to mobilize capital. Global growth measures the real growth of the world economy and is obtained from the World Development Indicators. Regional contagion reflects the extent to which surges or stops occurred in the region of the country experiencing the surge. The indicator used to measure this factor is the share of countries in the same macro region which experienced a surge in the preceding year (see also Dell’Erba and Reinhardt, 2012). The domestic set of factors include experiencing a surge in the preceding year, experiencing a stop in the preceding year (see also Sula, 2010), 12 per capita GDP, natural resource rents as a share of GDP, and the change in the following set of variables – trade and financial openness, economic and financial stability, and political stability. We expect M&A-led surges and stops to be more prevalent in higher income developing economies (Nocke and Yeaple, 2008; Qiu and Wang, 2011) because of the presence of attractive corporate assets in terms of quality of inputs and technology 13 and a narrower gap in 11 See Forbes and Warnock (2012) for a similar operationalization of this variable. 12 It can be expected that some surges concur with the recovery from a stop in FDI and some stops happen after a sudden surge in FDI. 13 Foreign firms typically ‘cherry pick’ high quality targets (Bertrand et al., 2012). 13 production costs between the destination and the source country. 14 A large price differential between the home and host countries might make GF investments more likely as an entry mode in lower income developing economies. These differentials are needed to offset the relatively higher start-up costs associated with the construction of new facilities. Investments in resource-intensive industries also usually take the form of GF FDI. The reason for this is that local companies often have privileged access to these natural resources and, hence, host country governments prefer joint ventures in the form of GF FDI (Demirbag et al., 2008). M&A-led surges and stops are also more likely in riskier and uncertain macroeconomic environments because of the existence of discounts on the prices of existing assets (Buiter et al., 1998). Such events have been associated with fire-sale FDI during the Latin American and Asian financial crises of the 1990s (Krugman, 2000; Aguiar and Gopinath, 2005). M&A-led surges and stops have been encouraged by capital market imperfections that result in undervaluations of firm assets, sales of assets at unrealistic prices, and stripping of firms for purely financial gains. Changes in financial openness in particular might have an effect on the likelihood of an M&A-led surge or stop. Furthermore, the existence of capital controls in the form of restrictions on foreign ownership and short-selling may limit possibilities for the earlier- mentioned fire-sale FDI. 15 There seems to be a strong political preference for GF investments, which are perceived to be more beneficial than M&As (Heinemann, 2012). Domestic unrest can also deter some types of FDI or trigger the pullout of investors from some markets (Schneider and Frey, 1985). Still, results from econometric studies 14 Aleksynska and Havrylchyk (2013) estimated that for the period 1996-2007, 56% of all FDI into developing countries originated from developed countries. 15 Likewise, government interventions against the takeover of domestic companies through cross-border M&A have become more common recently (Heinemann, 2012). 14 are ambiguous about the relationship between the degree of political stability in a country and stops in FDI inflows (Salomon and Ruiz, 2012). The source for data on GDP, trade, and natural resource rents is the World Bank’s World Development Indicators. Financial openness is represented by the Chinn and Ito (2008) capital account openness index. The International Country Risk Guide is the source for the economic, financial, and political stability measures. Definitions and sources for the independent variables are presented in Appendix A2, while descriptive statistics can be found in Appendix A3. Econometric Results The results from the complementary logistic regressions presented in Table 2 suggest that in general FDI surges are difficult to predict. The baseline regressions have low pseudo R2 varying between 3% and 10% of the observed variance. 16 Stops appear easier to predict in that the pseudo R2 varies between 10% and 34%. This difference can be attributed mainly to the fact that a surge in the preceding year is a good predictor of a FDI stop, but not vice versa. Hence, it can be inferred that surges are followed by stops. At the same time, a FDI surge in the preceding year is not a good predictor of a surge and a FDI stop in the preceding year is not a good predictor of a stop. FDI Surges Global liquidity is associated positively with a FDI surge, regardless of its kind. Although global liquidity is only significantly correlated with the probability of a GF-led surge (Table 2), the effect of global liquidity on the likelihood of a GF-led surge does not significantly differ from the effect of global liquidity on the likelihood of an M&A-led surge (χ2=0.08, p=0.77). Furthermore, the effect of global liquidity on the likelihood of 16 This figure is based on McFadden’s R2. 15 an M&A-led surge becomes significant under alternative definitions of a surge (Appendix Table B3), 17 alternative definitions of a resource-rich economy, economic, financial and political stability (Appendix Table B7), and additional control variables (Appendix Table B11). Overall, these results suggest that loosening of global credit conditions in response to global financial crises tends to increase the frequency of FDI surges in developing countries. These results are in line with earlier findings of Di Giovanni (2005) and Baker et al. (2009) that the availability of cheap financial capital stimulates the expansion of multinational activity. Regional contagion increases the probability of a surge, but not of a stop (Table 2). The result is robust when we use random effects probit, which allows for country-specific unobserved heterogeneity (Table 3). The conclusions that can be drawn from this estimation are to a large extent identical to the ones from a seemingly unrelated estimated complementary logistic regression with cluster-robust standard errors presented in Table 2. Likewise, considering a more restricted sample of events (Table 4), which excludes surges followed by stops and focuses on the so-called “sustainable” surges, did not lead to substantially different results. However, when we only focus on “sustainable” M&A- led surges and “relentless” stops (stops followed by a stop and not a surge), the effect of regional contagion on the probability of a surge becomes insignificant. There are pronounced differences with regard to the factors that predict the onset of M&A-led and GF-led surges. Global growth is positively and significantly correlated with the incidence of M&A-led surges and FDI-surges in general, but not with GF-led surges in general. This difference in effect sizes between M&A-led and GF-led surges was 17 The increase in the FDI-to-GDP ratio is more than one and a half standard deviation above its rolling mean. 16 statistically significant (χ2=16.56, p<0.01). The result is robust across different estimation strategies, alternative definitions of surges, other variables’ definitions, and additional control variables (see results in Table 3, Appendix Tables B1 and B3, Appendix Table B5 and B7, and Appendix Table B9 and B11). In the case of M&A-led surges, the result is in line with the fact that the two most recent global FDI and M&A waves took place during periods of strong economic growth, and their ends coincided with global downturns. In the case of GF-led surges, it reflects the fact that firms are often driven to invest in operations located in developing countries as a cost cutting measure and not necessarily during periods of strong global growth. However, when we only focus on sustainable surges, we find that global growth is a significant predictor for sustainable GF-led surges. A country with lower per capita income level is significantly more likely to experience GF-led surges (Table 2). In contrast, M&A-led surges are significantly more likely in countries with higher per capita incomes. The difference in the effect of per capita income levels on the likelihood of having a GF-led versus M&A-led surge is statistically significant (χ2=13.10, p<0.01). These results are robust to alternative estimation (Table 3), specification methods (Appendix Tables 9 and 11), and different definitions of a surge (Appendix Tables 5 and 7). Resource-rich countries are more likely to experience a GF-led surge, but less likely to incur an M&A-led surge. This result is not significant across estimation techniques and variable definitions, and when additional control variables are added. However, the difference in the effect sizes is statistically significant (χ2=5.01, p=0.025) and this result holds when we replace the natural resource variable with a dummy variable that takes the value 1 if a country is hydrocarbon or mineral-rich as defined by the IMF (Appendix B5 and B7; χ2=7.26, p<0.01). 17 In line with the fire-sale FDI hypothesis of Krugman (2000) and Aguiar and Gopinath (2002) and as shown by Bogach and Noy (2013), M&A-led surges are significantly more likely in countries which experience deterioration in economic and financial stability. Although an improvement in economic and financial stability tends to be positively associated with the probability of a GF-led surge, the variable is not significant in the baseline model and alternative tests. These results are robust to changes in the model specification, estimation technique, and variables definitions. The difference between the effects of changes in economic and financial stability on the likelihood of the two types of FDI surges is statistically significant (χ2=7.83, p<0.01) and becomes more pronounced under stricter definitions of a surge. The findings are supported by the results when additional control variables related to changes in domestic macroeconomic economic conditions are added to the model (Appendix Tables B9 and B11). A decrease in the exchange rate and an increase in the inflation rate are positively associated with the probability of M&A-led surges, but not of GF-led surges. This difference in effect size of the exchange rate variable is statistically significant (χ2=9.35, p<0.01). A decrease in capital controls (i.e. increase in financial openness) is associated with a higher probability of an M&A-led surge, but not with a higher probability of a GF led surge. This finding is in line with the idea that there is a strong political preference for GF FDI and capital controls particularly affect FDI in the form of M&As. The result is robust across different definitions of a surge and independent variables, as well as additional controls, but not alternative estimation techniques. Furthermore, the difference between the effects of changes in financial openness on the probability of the two types of FDI surges is not statistically significant. Finally, global risk and changes in political stability do not affect the likelihood of an FDI surge regardless of its type in nearly all specifications. 18 FDI Stops An FDI surge in the preceding year is the only significant and robust predictor of an FDI stop, regardless of its kind (Table 2). This result is in line with the recent findings of Agosin and Huaita (2012) who show that the best predictor of a sudden stop is a preceding capital boom, where stops are downward overreactions to sharp preceding positive overreactions. Tightening of global credit conditions diminishes the frequency of FDI surges, but conditional on a surge in the previous year do not necessarily have a contemporaneous effect on the likelihood of FDI stops. However, since global liquidity is a significant predictor of FDI surges, and a surge is a significant predictor of a stop in the next period, global liquidity is an important predictor of both FDI surges and stops. A decline in global growth has a significant and positive effect on the likelihood of a FDI stop, regardless of its kind, where the difference in effect sizes between the two modes of entry is not statistically significant (χ2=0.79, p=0.38). This result, however, loses its significance under stricter definitions of stops, be they GF-led or M&A-led (Appendix Tables B2 and B4). GF-led and M&A-led FDI stops are, on average, more likely in poorer and richer countries, respectively. The effect itself is not significant, but the difference in effect sizes is statistically significant (χ2=4.57, p=0.033) and becomes pronounced under stricter definitions of FDI stops (Appendix Tables B2 and B4). This difference in effect sizes holds under alternative estimation methods and variable definitions and when additional variables are included. The frequency of FDI stops is not higher in resource-rich countries than elsewhere in the world. Resource-rich economies appear to be more likely to have GF-led episodes and less likely to have M&A-led episodes, but these average effects are not significant across 19 a range of alternative model and variables’ specifications. Moreover, the difference in effect sizes of natural resources on GF-led and M&A-led stops is not statistically significant (χ2=1.70, p=0.192) in the baseline model in Table 2 as well as in the model using the alternative definition of natural resources (Appendix Tables B6 and B8; χ2=1.29, p=0.257). Removal of capital controls increases the probability of an M&A-led stop, but not of a GF-led stop. The difference in effect sizes is statistically significant (χ2=7.05, p<0.01) and holds when we control for changes in trade openness and tariffs (Appendix Tables B10 and B12), but this result becomes less pronounced under stricter definitions of a stop (Appendix Tables B2 and B4). Global risk and changes in political, economic, and financial stability do not affect the likelihood of a FDI stop regardless of its type and model specification. 20 Table 2: Regression Results on Likelihood of a FDI Surge and Stop – Complementary Logistic Regression Estimates Surges Stops All GF-led M&A-led All GF-led M&A-led Global Factors Global Risk 0.015 (.013) 0.010 (.014) 0.045 (.028) 0.023 (.010)* 0.013 (.013) 0.046 (.031) Global Liquidity 0.014 (.003)** 0.013 (.004)** 0.015 (.007) 0.003 (.003) 0.002 (.003) 0.003 (.006) Global Growth 0.150 (.055)** 0.076 (.058) 0.542 (.097)** -0.127 (.042)** -0.112 (.053)* -0.235 (.112)* Regional Contagion 0.020 (.006)** 0.021 (.008)* 0.032 (.015)* 0.000 (.006) -0.007 (.010) 0.028 (.017) Domestic Factors Stop previous year 0.217 (.180) 0.263 (.231) 0.685 (.501) -0.391 (.218) -0.735 (.332)* 0.848 (.612) Surge previous year -0.182 (.183) -0.014 (.205) -0.195 (.540) 1.623 (.141)** 1.533 (.153)** 3.915 (317)** GDP per Capita (ln) -0.082 (.023)** -0.148 (.034)** 0.173 (.068)* -0.008 (.028) -0.067 (.036) 0.106 (.068) Natural Resources 0.001 (.003) 0.007 (.003)* -0.036 (.018)* 0.002 (.003) 0.006 (.003)* -0.007 (.009) Δ Financial Openness 0.181 (.132) 0.118 (.165) 0.433 (.212)* -0.069 (.156) -0.247 (.195) 0.708 (.247)** Δ Economic and Financial 0.002 (.013) 0.020 (.015) -0.068 (.027)* -0.004 (.017) -0.011 (.018) 0.026 (.031) Stability Δ Political Stability -0.030 (.018) -0.038 (.020) 0.011 (.042) -0.025 (.019) -0.031 (.021) 0.013 (.041) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Surges or Stops 264 207 57 282 225 57 Pseudo R2 0.034 0.033 0.102 0.125 0.104 0.340 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. 21 Table 3: Regression Results on Likelihood of a FDI Surge and Stop – Random Effects Probit Estimates Surges Stops All GF-led M&A-led All GF-led M&A-led Global Factors Global Risk 0.010 (.007) 0.006 (.007) 0.024 (.012) 0.013 (.008) 0.007 (.008) 0.021 (.016) Global Liquidity 0.008 (.002)** 0.007 (.002)** 0.006 (.004) 0.001 (.002) 0.001 (.002) 0.001 (.003) Global Growth 0.086 (.032)** 0.038 (.033) 0.250 (.068)** -0.094 (.030)** -0.076 (.031)* -0.110 (.056) Regional Contagion 0.012 (.004)** 0.013 (.005)* 0.019 (.010)* 0.033 (.372) -0.372 (.472) 0.014 (.011) Domestic Factors Stop previous year 0.140 (.103) 0.172 (.117) 0.327 (.265) -0.221 (.132) -0.368 (.163)* 0.342 (.326) Surge previous year -0.101 (.110) -0.004 (.125) -0.132 (.329) 1.108 (.131)** 1.003 (.102)** 2.261 (.199)** GDP per Capita (ln) -0.048 (.021)* -0.083 (.023)** 0.082 (.036)* -0.002 (.022) -0.036 (.023) 0.077 (.042) Natural Resources 0.001 (.002) 0.004 (.002) -0.013 (.005)* 0.002 (.003) 0.004 (.002) -0.002 (.005) Δ Financial Openness 0.181 (.132) 0.078 (.096) 0.166 (.155) -0.039 (.096) -0.131 (.102) 0.316 (.161)* Δ Economic and Financial 0.002 (.013) 0.013 (.009) -0.030 (.015)* -0.004 (.009) -0.007 (.009) 0.016 (.016) Stability Δ Political Stability -0.018 (.012) -0.022 (.013) 0.009 (.021) -0.018 (.013) -0.018 (.013) -0.002 (.025) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Surges or Stops 264 207 57 282 225 57 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. Dependent variable is a 0-1 variable indicating if there is a surge or stop episode (either GF-led surge or an M&A dominated surge). The variables Regional Contagion, GDP per capita and Natural Resources are lagged (one year). The variables GDP per capita and natural resources are windsorized at the 1% level. 22 Table 4: Sustainable Surges and Relentless Stops Sustainable Surges: Relentless Stops: Surges not followed by Stops Stops not followed by Surges All GF-led M&A-led All GF-led M&A-led Global Factors Global Risk -0.006 (.007) 0.002 (.017) 0.025 (.043) 0.024 (.013) 0.015 (.014) 0.051 (.034) Global Liquidity 0.015 (.005)** 0.011 (.006)* 0.017 (.011) 0.003 (.003) 0.002 (.003) 0.001 (.007) Global Growth 0.215 (.088)* 0.195 (.092)* 0.710 (.180)** -0.143 (.051)** -0.117 (.059)* -0.237 (.118)* Regional Contagion 0.023 (.007)** 0.020 (.009)* 0.027 (.024) 0.002 (.675) -0.010 (.011) 0.025 (.018) Domestic Factors Stop previous year 0.351 (.240) 0.444 (.281) 0.963 (.653) -0.352 (.268) -0.636 (.358) 1.062 (.619) Surge previous year -0.422 (.280) -0.293 (.320) 0.176 (.596) 1.663 (.142)** 1.559 (.154)** 4.098 (.34)** GDP per Capita (ln) -0.051 (.044) -0.109 (.052)* 0.226 (.103)* -0.027 (.038) -0.091 (.038)* 0.084 (.079) Natural Resources 0.003 (.005) -0.000 (.005) -0.021 (.018) 0.002 (.004) 0.006 (.003)* -0.004 (.010) Δ Financial Openness 0.166 (.190) 0.049 (.211) 0.768 (.261)** 0.141 (.166) -0.027 (.183) 0.801 (.241)** Δ Economic and Financial 0.013 (.018) 0.029 (.017) -0.087 (.037)* 0.002 (.015) -0.002 (.019) 0.029 (.033) Stability Δ Political Stability 0.002 (.023) -0.002 (.023) -0.031 (.057) -0.033 (.023) -0.038 (.023) 0.019 (.043) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Surges or Stops 152 132 29 228 188 51 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. Dependent variable is a 0-1 variable indicating if there is a surge or stop episode (either GF-led surge or an M&A-led surge). The variables Regional Contagion, GDP per capita and Natural Resources are lagged (one year). The variables GDP per capita and natural resources are windsorized at the 1% level. 23 Sensitivity Analysis This section provides an overview of the most important results from the sensitivity analysis presented in Appendix B, but not discussed in the previous paragraph. The estimations which use alternative variable definitions (Appendix Tables B5-B8) and additional control variables (Appendix Tables B9-B12) did not yield very different results. Differences with the baseline regression under different definitions of FDI surges and stops (Appendix Tables B1-B4) were most pronounced. Under the strictest definition the number of surges and stops falls down to 143 (8.0% of the all country-years) and 108 (6.1% of all the country-years), respectively. Compared to the baseline regressions, the effect of global growth on the likelihood of GF-led and M&A-led stops, the effect of regional contagion on the likelihood of GF-led surges, and the effect of a stop in the preceding year on GF-led stops lose their significance under stricter definitions of a surge or stop. In addition to the differences described above, a decrease in political stability has a positive and significant effect on the likelihood of a GF-led surge under some of the stricter definitions of a surge. This result can be explained by the fact that some firms try to take advantage of new opportunities that arise through the changes in political regimes and reap the benefits in conflict locations through first mover advantages or market power. In particular, firms in the primary sector (which mainly enter through GF FDI) would be less deterred by a decrease in political stability because activities in this sector are bound by physical geography and risk adjusted rents are higher (Burger et al. 2013). 4. Concluding Remarks This paper investigates the factors triggering FDI surges and stops episodes, differentiated based on whether these events are led by waves in GF investments or M&As. In particular, we analyze the effect of the mode of entry on the incidence and 24 determinants of FDI surges and stops in the developing world. The focus on this topic is warranted because during the past decade there has been a significant increase of FDI flows to developing countries but the rise has not proceeded in a smooth fashion, prompting concerns about sudden stops even in countries where FDI inflows dominate capital flows. Furthermore, whereas most global FDI waves have been associated with an increase in M&As, it is not clear to what extent GF investments have contributed to surges and stops in FDI in developing countries. It is important to answer this question because GF investments dominate FDI flows to the developing world, especially in resource-rich and low-income countries, and as shown in the paper, GF-led surge and stop episodes occur more frequently than M&A-led ones. This paper contributes to the literature by constructing a database of episodes when foreign investors substantially increase or decrease FDI inflows to developing countries and distinguish between these episodes based on the dominance of the entry mode. We use this database to document the incidence of FDI surges and stops by mode of entry, region, and resource status. Using this database, we analyze the factors associated with GF-led and M&A-led surge and stop events and show that the two types of surges and stops have different incidence and determinants, and therefore must be studied separately. Our analysis shows that global liquidity is the only common predictor of different types of FDI surges, while a decline in global growth and an FDI surge in the preceding year are the only significant and consistent predictors of FDI stops. GF-led stops and surges are more likely in low income and resource-rich countries than elsewhere. Policies aimed at increasing financial openness are enablers of M&A-led surges, which are also more likely during periods of global growth and domestic economic and financial instability. 25 The results are policy relevant as the paper shows that GF-led extreme events occur more frequently than M&A-led ones. Thus, countries relying mostly on GF FDI, the more stable type of FDI inflows, are not immune to sudden stops in capital flows and should prepare to withstand them. Knowledge of the factors behind different types of FDI surges and stops can help policy makers in developing countries, especially those relying heavily on FDI flows, prepare to weather such events. 26 References Abiad, A., Dell’Ariccia, G., and Li, B. (2011) “Creditless Recoveries” IMF Working Paper 11/58, International Monetary Fund. Agosin, M.R. and Huaita, F. (2012) “Overreaction in Capital Flows to Emerging Markets: Booms and Sudden Stops”, Journal of International Money and Finance, 31(5), pp. 1140-1155. Aguiar, M. and Gopinath, G. (2005) “Fire-Sale Foreign Direct Investment and Liquidity Crises,” Review of Economics and Statistics, 87(3), pp. 439-452. Aleksynska, M. and Havrylchyk, O. (2013) “FDI from the South: The Role of Institutional Distance and Natural Resources,” European Journal of Political Economy, 29 (March), pp. 38-53. Andrade, G., Mitchell, M. and Stafford, E. (2001) “New Evidence and Perspectives on Mergers,” Journal of Economic Perspectives, 15(2), pp. 103-120. Baker, M., Foley, C.F. and Wurgler, J, (2009) “Multinationals as Arbitrageurs: the Effects of Stock Market Valuations on Foreign Direct Investment,” Review of Financial Studies, 22(1), pp. 337-369. Barba Navaretti, G. and Venables, A.J. (2004) Multinational Firms in the World Economy. Princeton, NJ: Princeton University Press. Beck, T., Demirgüç-Kunt, A. and Levine, R. (2000) “A New Database on the Structure and Development of the Financial Sector,” World Bank Economic Review, 14(3), pp. 597- 605. Bertrand, O., Hakkala, K.N., Norbäck, P-J. and Persson, L. (2012) “Should Countries Block Foreign Takeovers of R&D Champions and Promote Greenfield Entry?” Canadian Journal of Economics, 45(3), 1083-1124. 27 Blonigen, B.A. (2005) “A Review of the Empirical Literature on FDI Determinants,” Atlantic Economic Journal, 33(4), pp. 383-403. Blonigen, B.A. and Wang, M. (2005) “Inappropriate Pooling of Wealthy and Poor countries in Empirical FDI Studies”, in Moran, T., Graham, E. and, Blomstrom M. (eds) Does Foreign Direct Investment Promote Development?, Washington, DC: Institute for International Economics, pp. 221-243. Bogach, O. and Noy, I. (2013) “Fire-Sale FDI? The Impact of Financial Crises on Foreign Direct Investment”. Review of Development Economics (forthcoming) Brakman, S., Garretsen, H., Van Marrewijk, C. and Van Witteloostuijn, A. (2006) Nations and Firms in the Global Economy: An Introduction to International Economics and Business. Cambridge, UK: Cambridge University Press. Broner, F., Didier, T., Erce, A., Schmukler, S. (2013) “Gross Capital Flows: Dynamics and Crisis,” Journal of Monetary Economics, 60(1), pp. 113-133. Buiter, W., Lago, R. and Rey, H. (1998) “Financing in Transition: Investing in Enterprises during Macroeconomic Transition”. EBRD Working Paper 35, London: European Bank for Reconstruction and Development). Burger, M., Ianchovichina, E. and Rijkers, B. (2013) “Risky Business: Political Instability and Greenfield Foreign Direct Investment in the Arab World,” Policy Research Working Paper 6716, World Bank. Calderón, C., Loayza, N. and Servén, L. (2004) “Greenfield Foreign Direct Investment and Mergers and Acquisitions: Feedback and Macroeconomic Effects,” World Bank Policy Research Working Paper No. 3192, Washington DC. 28 Cardarelli, R., Elekdag, C. and Kose, M.A. (2010) “Capital Inflows: Macroeconomic Implications and Policy Responses,” Economic Systems, 34(4), pp. 333-356. Calvo, G. (1998) “Capital Flows and Capital Market Crisis: The Simple Economics of Sudden Stops,” Journal of Applied Economics, 1(1), pp. 35-54. Calvo, G., Leiderman, L. and Reinhart, C. (1996) “Inflows of Capital to Developing Countries in the 1990s,” Journal of Economic Perspectives, 10(2), pp. 123‐139. Calvo, G., Izquierdo, A., and Meija, L-F. (2004) “On the Empirics of Sudden Stops: the Relevance of Balance-Sheet Effects”. NBER Working Paper 10520. Calvo, G., Izquierdo, A. and Loo-Kung, R. (2006) “Relative Price Volatility under Sudden Stops: The relevance of Balance Sheet Effect,” Journal of International Economics, 69(1), pp. 231-254. Chinn, M. and Ito, H. (2008) “A New Measure of Financial Openness,” Journal of Comparative Policy Analysis, 10(3), pp. 309-322. Cowan, K., De Gregorio, J., Micco, A., and Neilson, C. (2008) “Financial Diversification, Sudden Stops, and Sudden Starts,” In Cowan, K, Sebastian, E., and Valdes, R. (editors), Current Account and External Financing: An Introduction. Santiago, Central Bank of Chile, pp. 159-194. Cowan, K. and Raddatz, C. (2011) “Sudden Stops and Financial Frictions” World Bank Policy Research Working Paper No. 5605. Dell’Erba, S. and Reinhardt, D. (2012) “Surfing the Capital Waves: A Sector-level Examination of Surges in FDI Inflows,” Working Paper 11.07, Study Center Gerzensee. 29 Demirbag, M., Tatoglu, E., Glaister, K.W. (2008) “Factors Affecting Perceptions of the Choice between Acquisition and Greenfield entry: The Case of Western FDI in an Emerging Market,” Management International Review, 48, pp. 5-38. Di Giovanni, J. (2005), “What Drives Capital Flows? The Case of Cross-border M&A Activity and Financial Deepening”, Journal of International Economics, 65, pp.127-149. Fernandez-Arias, E., Montiel, P. (1996) “The Surge in Capital Inflows to Developing Countries: An Analytical Overview,” World Bank Economic Review, 10(1), pp. 51-77. Forbes, K.J and Warnock, F.E. (2012) “Capital Flow Waves: Surges, Stops, Flight and Retrenchment,” Journal of International Economics, 88(2), pp. 235-251. Furceri, D., Guichard, S. and Rusticelli, E. (2012) “Episodes of Large Capital Inflows, Banking and Currency Crises, and Sudden Stops”, International Finance, 15(1), pp. 1-35. Gall, T., Schiffbauer, M., and Kubny, J. (2013) “Dynamic Effects of Foreign Direct Investment When Credit Markets are Imperfect” Macroeconomic Dynamics (forthcoming). Ghosh, A.R., Kim, J., Qureshi, M.S., Zalduendo, J. (2012) “Surges,” IMF Working Paper 12/22. Harms, P. and Méon, P-G. (2012) “Good and Bad FDI: The Growth Effects of Greenfield Investment and Mergers and Acquisitions in Developing Countries,” Working Paper. Heinemann, A. (2012) “Government Control of Cross-border M&A: Legitimate Regulation or Protectionism?” Journal of International Economic Law, 15(3), pp. 843-870. Herzer, D. (2012) “How Does Foreign Direct Investment Affect Developing Countries’ Growth?” Review of International Economics, 20(2), pp. 396-414. 30 Kaminsky, G., Lizondo, S., Reinhart, C. (1998) “Leading Indicators of Currency Crises,” IMF Staff Papers, 45(1), pp. 1-48. Kim, Y-H. (2009) “Cross-border M&A vs. Greenfield FDI: Economic Integration and its Welfare Impact,” Journal of Policy Modeling, 31(1), pp. 87-101. Krugman, P. (2000) “Fire-sale FDI.” In S. Edwards (ed.) Capital Flows and the Emerging Economies: Theory, Evidence and Controversies. Chicago, IL: University of Chicago Press, pp. 43-59. Levchenko, A. and Mauro, P. (2007) “Do Some Forms of Financial Flows Protect From Sudden Stops?” World Bank Economic Review, 21(3), pp. 389-411. Lensink, R. and Morrissey, O. (2006) “Foreign Direct Investment: Flows, Volatility, and the Impact on Growth,” Review of International Economics, 14(3), pp. 478-493. Markusen, J.R. and Stähler, F. (2011) “Endogenous Market Structure and Foreign Market Entry,” Review of World Economics, 147(2), pp. 195-215. Mattoo, A., Olarreaga, M. and Saggi, K. (2004) “Mode of Foreign Entry, Technology Transfer, and FDI Policy,” Journal of Development Economics, 75(1), pp. 95–111. Mendoza, E. (2010) “Sudden Stops, Financial Crises, and Leverage,” American Economic Review 100(5), pp. 1941-1966. Nocke, V. and Yeaple, S. (2008) “An Assignment Theory of Foreign Direct Investment,” The Review of Economic Studies 75, pp. 529-557. Noorbakhsh, F., Paloni, A. and Youssef, A. (2001) Human Capital and FDI Inflows to Developing Countries: New Empirical Evidence,” World Development, 29(9), pp. 1593- 1610. 31 Qiu, L.D. and Wang S. (2011) “FDI Policy, Greenfield Investment and Cross-border Mergers,” Review of International Economics, 19(5), pp. 836-851. Reinhart, C.M. and Reinhart, V.R. (2009) “Capital Flow Bonanzas: An Encompassing View of the Past and the Present”, in NBER Macroeconomics Annual. Chicago, IL: Chicago University Press. Rothenberg, A.D. and Warnock, F.E. (2011) “Sudden Flight and True Sudden Stops,: Review of International Economics, 19(3), pp. 509-524. Salomon, B. and Ruiz, I. (2012) “Political Risk, Macroeconomic Uncertainty and the Patterns of Foreign Direct Investment”, International Trade Journal, 26, pp. 181-191. Schneider, F. and Frey, B. (1985) “Economic and Political Determinants of Foreign Direct Investment,” World Development, 13(12), pp. 161-175. Sula, O. (2010) “Surges and Sudden Stops of Capital Flows to Emerging Markets”, Open Economies Review 21(4), pp. 589-605. UNCTAD (2011) World Investment Report: 2011. New York/Geneva: United Nations Press. UNCTAD (2012) World Investment Report: 2012. New York/Geneva: United Nations Press. Wang, M. and Wong, M.C.S. (2009) “What Drives Economic Growth? The case of Cross-border M&A and Greenfield FDI Activities,” Kyklos, 62(2), pp. 316–330. Wes, M. and Lankes, H.P. (2001) “FDI in Economies in Transition: M&As versus Greenfield investment,” Transnational Corporations, 10(3), pp. 113-129. 32 Appendix A1: Surge-Years and Stop-Years by Country; Increases and Decreases of 2 Standard Deviations, Top or Bottom 20th Percentile, Denoted by an Asterisk (*) Country First Last Average Annual GF-led Surges M&A-led Surges GF-led Stops M&A-led Stops Year Year Growth (percentage points) Albania 1996 2010 0.42 2000*, 2007 1994, 2002 2005 Algeria 1991 2010 0.08 2001 1999*, 2003, 2010* Angola 1994 2010 -0.33 1995, 1998*, 1999* 1996, 2000* Argentina 1991 2010 -0.00 1999* 1993, 2009* 1998, 2000* Armenia 1999 2010 0.33 2006 2004 2010* 1999*, 2005 Azerbaijan 1999 2010 0.06 2002* 2003 1998, 1999, 2000 2005* Bahrain 1991 2010 -0.98 1991*, 1996, 2006* 1993*, 1997, 2007*, 2009 Bangladesh 1991 2010 0.05 1999, 2009 Belarus 1999 2010 0.14 1999 2007 1998*, 2000*, 2010 Bolivia 1991 2010 0.06 2006* 1995*, 1997* 2000*, 2003* Botswana 1991 2010 0.21 2002*, 2009* 1991, 1993*, 1999, 2010* Brazil 1991 2010 0.10 1999, 2004 1997, 1998 2003, 2009 2001*, 2002 Burkina Faso 1991 2010 0.02 1994 2007* 2001, 2008 Cameroon 1991 2010 0.07 1996, 1998*, 2000, 1999*, 2003, 2010 2002*, 2009* Chile 1991 2010 0.31 1994*, 1996, 2007 1999* 2000* China 1991 2010 -0.01 1992, 1993* 1995, 2006 Colombia 1991 2010 0.06 1997 1996*, 2005 1995, 1998* 2006 Congo-Brazzaville 1991 2010 1.13 1993*, 1999*, 2005 1992, 1994*, 2000*, 2008* 33 Congo-Kinshasa 1991 2010 1.23 1998, 2003*, 2007*, 1997*, 2005* 2009 2010 Costa Rica 1991 2010 0.08 1998, 2004, 2006* 2002 2000* 2009* Côte d'Ivoire 1991 2010 0.05 1993, 1998* 1997* 1999* Croatia 1999 2010 0.03 1999*, 2006 2002, 2009, 2010* 2000*, 2004 Dominican Republic 1991 2010 1.23 1995, 1997, 2007, 1999 1996*, 2000, 2009 2008* Ecuador 1991 2010 0.12 1993, 2001, 2008* 1995, 2000* Egypt 1991 2010 0.11 1993, 2004*, 2005*, 2008* 2001, 2009 2006 El Salvador 1991 2010 0.02 1998*, 2007* 1999*, 2006, 2008 Ethiopia 1991 2010 0.06 1997*, 2001, 2003 1999*, 2005*, 2007 Gabon 1991 2010 0.04 1992, 1998*, 2010* 1993, 1995, 1999, 2005 Gambia 1991 2010 0.11 1999*, 2002, 2004 1991, 2003, 2009 Ghana 1991 2010 0.42 1993, 1994*, 2006*, 1995*, 2001* 2009 Guatemala 1991 2010 0.05 1998* 1999* Guinea 1991 2010 0.10 1999*, 2002, 2003, 1992, 2000*, 2009* 2007* Guinea-Bissau 1991 2010 0.18 1997, 2004, 2006*, 1998, 2008* 2010 Haiti 1992 2010 0.14 2006* 1992, 2007* Honduras 1991 2010 0.22 2007, 2010 1999*, 2000 1998, 2009* 2001 Hong Kong SAR, 1991 2010 1.55 1993, 1998, 1999*, 2010 2001* 2009 China 2000* India 1991 2010 0.07 2006*, 2008 2010 2009* Indonesia 1991 2010 0.04 1995 2005 1998* 2006 Iran 1991 2010 0.05 2002* 2003, 2006 34 Iraq 2001 2010 0.27 2003* 2004* Jamaica 1991 2010 -0.15 1998*, 1999, 2003, 1993, 2000, 2004, 2002* 2008 2009* Jordan 1991 2010 0.29 1997*, 2005, 2006 2000* 1991*, 1993, 1999, 2001* 2007* Kazakhstan 1999 2010 0.39 1999*, 2001*, 2006 2000, 2002, 2003, 2005, 2010 Kenya 1991 2010 0.03 2007* 2001 2008* Kuwait 1991 2010 0.00 1996, 2009 1997*, 2010 Lebanon 1991 2010 0.59 1997*, 2003* 1998, 2004, 2010 Liberia 1991 2010 3.07 1994, 1997*, 2003* 2010* 1991*, 1996, 2004 Libya 1991 2010 0.13 2005, 2006, 2007* 2001, 2008* 2010* Madagascar 1991 2010 0.51 1999, 2006*, 2007* 2002, 2010 Malawi 1991 2010 0.08 1994, 1999* 1991*, 1995, 2002, 2005* Malaysia 1991 2010 -0.25 1991, 1999, 2002*, 1994*, 1998, 2001*, 2008 2010* 2009* Mali 1991 2010 0.29 1995*, 2000, 2002, 1992, 1996*, 1998, 2009* 2003*, 2010 Mexico 1991 2010 0.05 1994*, 2007 2001* 2002* Moldova 1999 2010 0.15 2000*, 2007 2009* 2001 Mongolia 1991 2010 1.51 1997, 1999, 2000*, 1992, 2004* 2003*, 2008*, 2010* Morocco 1991 2010 0.02 1999 1997*, 2001* 2000, 2005 1998*, 2002* Mozambique 1991 2010 0.51 1998, 1999, 2007, 2000* 2009 Namibia 1995 2010 0.13 2000*, 2001*, 2007 1993, 1999, 2002, 2009 Nicaragua 1991 2010 0.35 1991*, 1995, 1997*, 2000, 2001, 2009* 1999*, 2007, 2008* 35 Niger 1991 2010 0.82 1992, 2007*, 2008*, 1991, 1993*, 2002 2009* Nigeria 1991 2010 -0.02 1996, 2005* 1995, 1997, 2004, 2006, 2010 Oman 1991 2010 0.06 2003*, 2005*, 2007 2004, 2008 Pakistan 1991 2010 0.03 2005 2006* 1995, 2009* Panama 1991 2010 0.38 1996 1997*, 2003*, 2006* 1995*, 2009 1998, 1999, 2007 Papua New Guinea 1991 2010 -0.12 1995*, 2007, 2009* 1991, 1996, 2000, 2008 2010* Paraguay 1991 2010 -0.04 1997, 1998 1993*, 1999*, 2009 Peru 1991 2010 0.31 1993*, 2002 1994* 1997, 2003 1995 Philippines 1991 2010 -0.04 1998, 2000, 2002 1999, 2008 2001* Qatar 1991 2010 0.18 1996*, 2002, 2005, 1995, 1999*, 2008*, 2009* 2010 Russian Federation 1998 2010 0.15 2003, 2006 2000, 2009* 2005 Saudi Arabia 1991 2010 0.35 2005* 2010* Senegal 1991 2010 0.10 1994*, 2006* 1997* 1991*, 1995 1998 Sierra Leone 1991 2010 0.26 2000*, 2004 2001*, 2008 Singapore 1991 2010 0.98 1993, 1999*, 2003, 1991, 1992, 1998*, 2006 2002, 2008* Somalia 1991 2010 0.58 2005, 2006*, 2010 2008* South Africa 1991 2010 0.02 1997*, 2001, 2007* 1998*, 2002 South Korea 1991 2010 0.04 1998 2001 Sri Lanka 1991 2010 -0.01 1997* 1995, 2009 1998* Sudan 1991 2010 0.14 1994, 1997, 1998* 2003 1995 2004, 2007* Syrian Arab Republic 1991 2010 0.15 1999, 2007 2001 Tanzania 1991 2010 0.23 1995*, 1999*, 2005* 2000, 2006* Thailand 1991 2010 0.06 1997* 1998* 1991*, 2000, 2009 1999, 2008* 36 Togo 1991 2010 0.19 1994*, 2000, 2001, 1992, 1996, 2002*, 2010 2003, 2007 Trinidad and Tobago 1991 2010 -0.02 1993*, 1997* 2007* 1995*, 2005 1998, 2009* Tunisia 1991 2010 0.00 1992* 1998*, 2000*, 2006* 1995, 1996, 2001, 1999, 2007* 2003 Turkey 1991 2010 0.05 2001* 2005, 2006 2002* 2008, 2009 Uganda 1994 2010 0.17 2006* 2008*, 2010 Ukraine 1999 2010 0.25 2003 2005* 2006 United Arab Emirates 1991 2010 0.09 2001, 2003*, 2004 1994, 1999*, 2002, 2005, 2008, 2009 Uruguay 1991 2010 0.31 2003*, 2005, 2006 2004, 2007, 2009 Venezuela 1991 2010 -0.17 1992*, 1997, 2003* 1991*, 1993*, 1998, 1999 2006 Viet Nam 1991 2010 0.16 1991, 1994*, 2007* 1995*, 2009 Yemen 1996 2010 -0.85 2006* 1994*, 2005, 2009 Zambia 1991 2010 0.52 1993*, 2002, 2007* 2010 1991*, 1994*, 2005, 2008 Zimbabwe 1991 2010 0.11 1998*, 2005* 1999*, 2006 37 Appendix A2 – Definitions of Variables Included in Baseline Regression Name Definition Source Global Factors Global Risk Volatility index measuring Chicago Board Options Exchange Global Liquidity Sum of the change in the stock market capitalization to GDP World Bank Development Indicators ratio and domestic private sector credit to GDP ratio Global Growth Growth in Real GDP World Bank Development Indicators Regional Contagion (FDI surges; GF-led Percentage of countries located in the same world region (East Own estimation surges; M&A-led surges; FDI stops; GF-led Asia Europe and Central Asia, East Asia and Pacific, Latin stops; M&A-led stops) America and Caribbean, Middle East and North Africa, South Asia, or Sub-Saharan Africa) experiencing the same event in the preceding year. Domestic Factors Stop previous year (FDI stops; GF-led stops; M&A- Takes value 1 if the country experienced a stop in the preceding Own estimation led stops) year. Surge previous year (FDI surges; GF-led surges; Takes value 1 if the country experienced a surge in the preceding Own estimation M&A-led surges) year. GDP per Capita (ln) Natural logarithm of real GDP per capita World Bank Development Indicators Natural Resources Natural resources rents to GDP ratio World Bank Development Indicators Change Financial Openness Change in the degree of openness to cross-border capital Chinn and Ito (2008) transactions. Change Economic and Financial Stability Change in the sum of the economic stability and financial International Country Risk Guide stability index. Economic stability is a composite score based on the economic risk assessment of the ICRG, which takes in GDP per capita, Real GDP growth, annual inflation, budget balance as percentage of GDP, and current account as percentage of GDP. Financial stability is a composite score based on the financial risk assessment of the ICRG, which takes in foreign debt as a percentage of GDP, foreign debt service as a 38 percentage of exports of goods and services, current account as a percentage of exports of goods and services, net international liquidity as months of import cover, and exchange rate stability. A higher score indicates an improvement in stability. Change Political Stability Change in the sum of the political stability index. The political International Country Risk Guide risk index is a composite index consisting of 12 scores: (1) investment profile, (2) law and order, (3) bureaucracy quality, (4) corruption, (5) democracy, (6) government stability, (7) military in politics, (8) socio-economic conditions, (9) ethnic tensions, (10) religious tensions, (11) internal conflict, and (12) external conflict. A higher score indicates an improvement in stability. 39 Appendix A3 – Descriptive Statistics of Independent Variables in Baseline Regression (N=1768) Mean Standard Deviation Minimum Maximum Global Factors Global Risk 21.21 6.97 12.15 34.66 Global Liquidity 2.85 21.70 -66.28 34.83 Global Growth 2.71 1.49 -2.25 4.33 Regional Contagion (FDI surges) 14.90 10.36 0 50.00 Regional Contagion (GF-led surges) 11.78 8.41 0 36.36 Regional Contagion (M&A-led surges) 3.12 5.51 0 30.00 Regional Contagion (FDI stops) 14.92 11.57 0 100.0 Regional Contagion (GF-led stops) 11.73 9.41 0 75.0 Regional Contagion (M&A-led stops) 3.18 5.74 0 40.0 Domestic Factors Stop previous year (FDI stops) 0.15 0.36 0 1 Stop previous year (GF-led stops) 0.12 0.32 0 1 Stop previous year (M&A-led stops) 0.03 0.18 0 1 Surge previous year (FDI stops) 0.15 0.36 0 1 Surge previous year (GF-led stops) 0.12 0.32 0 1 Surge previous year (M&A-led stops) 0.03 0.17 0 1 GDP per Capita (ln) 24.62 1.80 20.76 28.44 Natural Resources 12.29 16.29 0 85 Change Financial Openness 0.04 0.42 -2.56 3.25 Change Economic and Financial Stability 0.90 4.72 -27.83 24.95 Change Political Stability 0.55 3.53 -19.92 29.75 40 Appendix B: Sensitivity Analysis Alternative definitions of surges and stops (B1-B4): Surges Alternative Definition 1: 1 standard deviation above the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 2: 1.5 standard deviation above the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 3: 1.5 standard deviations above the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 4: 2 standard deviations above the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 5: 2 standard deviations above the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Stops Alternative Definition 1: 1 standard deviation below the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 2: 1.5 standard deviation below the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 3: 1.5 standard deviations below the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth 41 Alternative Definition 4: 2 standard deviations below the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition 5: 2 standard deviations below the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative variable definitions (B5-B8): 1. Replace sum of the change in the stock market capitalization to GDP ratio and domestic private sector credit to GDP ratio by global interest rate measure: The global liquidity measure is now based on the global interest rate. The global interest rate is measured as the average of the long-term bond yields in Japan, Germany, the United Kingdom, and the United States, using data from the IMF’s International Financial Statistics. 2. Replacing Natural Resources Rents to GDP ratio by Resource-Rich dummy variable: Natural resources are now measured as a dummy variable that take the value 1 if a country is Hydrocarbon or Mineral Rich as defined by the IMF. 3-4. Replace Change in Economic and Financial Stability by Change in Economic Risk (3) or Change in Financial Stability (4; See Appendix A2 for definitions of these variables). 5-7. Replace Change in Political Stability by one of the indices related to (1) Absence of Conflict, (2) Institutional Quality or (3) Democracy. These indices are based on the ICRG data. Absence of Conflict is a composite score consisting of internal conflict and external conflict. Institutional Quality is a composite score consisting of investment 42 profile, law and order, bureaucracy quality, corruption and government stability. Democracy is a composite score based on democracy and military in politics. Variables Added to Baseline Specification (B9-B12) 1. Region Dummies: Dummy variables indicating world region (East Asia and Pacific, Europe and Central Asia, Latin America and Caribbean, Middle East and North Africa, South Asia or Sub-Saharan Africa) in which the country is situated. 2. Domestic GDP Growth: Real GDP Growth, World Development Indicators, World Bank. 3. Change in Exchange Rate: Change in natural logarithm of the local currency unit-to- US dollar ratio, World Development Indicators, World Bank. 4. Change in Inflation: Change in inflation at consumer prices, World Development Indicators, World Bank. 5. Change in Trade Intensity: Change in Trade-to-GDP ratio, where trade is measured as the sum of exports and imports, World Development Indicators. 6. Change in Tariffs: Change in average tariff applied (unweighted), World Development Indicators, World Bank and Nicita (2007). 7. Country size: Measured by population, World Development Indicators, World Bank. 43 Appendix Table B1: Sensitivity Analysis – GF-Led Surges: Alternative Definitions of Surge Baseline 1 SD, Top 20% 1.5 SD, Top 25% 1.5 SD, Top 20% 2 SD, Top 25% 2 SD, Top 20% Regression Percentile Percentile Percentile Percentile Percentile Global Factors Global Risk 0.010 (.014) 0.013 (.015) 0.020 (.017) 0.024 (.018) 0.021 (.019) 0.025 (.020) Global Liquidity 0.013 (.004)** 0.011 (.004)* 0.013 (.004)** 0.010 (.004)* 0.015 (.005)** 0.012 (.005)* Global Growth 0.076 (.058) 0.084 (.059) 0.076 (.071) 0.070 (.074) 0.049 (.073) 0.061 (.079) Regional Contagion 0.021 (.008)* 0.018 (.010) 0.009 (.012) 0.008 (.015) 0.019 (.015) 0.027 (.017) Domestic Factors Stop previous year 0.263 (.231) 0.390 (.238) 0.258 (.259) 0.379 (.271) 0.156 (.455) 0.204 (.492) Surge previous year -0.014 (.205) -0.021 (.288) -0.147 (.285) -0.287 (.396) -0.020 (.341) -0.147 (.457) GDP per Capita (ln) -0.148 (.034)** -0.171 (.038)** -0.137 (.041)** -0.168 (.045)** -0.130 (.044)** -0.177 (.047)** Natural Resources 0.007 (.003)* 0.010 (.004)* 0.009 (.003)** 0.011 (.004)** 0.010 (.003)** 0.011 (.004)** Δ Financial Openness 0.118 (.165) 0.129 (.166) 0.252 (.173) 0.285 (.186) 0.439 (.138)** 0.447 (.151)** Δ Economic and Financial Stability 0.020 (.015) 0.017 (.016) 0.018 (.018) 0.025 (.018) 0.018 (.017) 0.023 (.019) Δ Political Stability -0.038 (.020) -0.052 (.022)* -0.034 (.024) -0.066 (.025)** -0.029 (.025) -0.057 (.027)* Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Surge Years 207 175 160 136 125 107 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. Dependent variable is a 0-1 variable indicating whether there is a surge episode (either GF-led surge or an M&A-led surge). All models are estimated using complementary logistic regression. The variables GDP per capita and natural resources are windsorized at the 1% level. Covariance across surge episodes is accounted for using seemingly unrelated estimated with clustering of standard errors at the country level. Alternative Definition Column 1: 1 standard deviation from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 2: 1.5 standard deviation from the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 3: 1.5 standard deviations from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 4: 2 standard deviations from the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 5: 2 standard deviations from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth 44 Appendix Table B2: Sensitivity Analysis – GF-Led Stops: Alternative Definitions of Stops Baseline 1 SD, Bottom 1.5 SD, Bottom 1.5 SD, Bottom 2 SD, Bottom 2 SD, Bottom Regression 20% Percentile 25% Percentile 20% Percentile 25% Percentile 20% Percentile Global Factors Global Risk 0.013 (.013) 0.032 (.016)* 0.008 (.015) 0.024 (.016) 0.002 (.020) 0.021 (.024) Global Liquidity 0.002 (.003) 0.004 (.003) 0.000 (.004) 0.001 (.004) -0.002 (.004) -0.001 (.005) Global Growth -0.112 (.053)* -.065 (.059) -0.106 (.061) -0.063 (.067) -0.132 (.076) -0.083 (.085) Regional Contagion -0.007 (.010) -0.019 (.011) -0.025 (.016) -0.024 (.020) -0.048 (.030) -0.050 (.038) Domestic Factors Stop previous year -0.735 (.332)* -0.398 (.365) -0.498 (.420) -0.292 (.462) -1.44 (1.02) -1.00 (1.02) Surge previous year 1.533 (.153)** 1.823 (.169)** 1.588 (.190)** 1.882 (.206)** 1.581 (.211)** 1.858 (.243)** GDP per Capita (ln) -0.067 (.036) -0.115 (.042)** -0.059 (.038) -0.116 (.043)** -0.072 (.049) -0.128 (.049)** Natural Resources 0.006 (.003)* 0.006 (.003)* 0.007 (.003)* 0.008 (.003)** 0.001 (.006) 0.005 (.006) Δ Financial Openness -0.247 (.195) -0.398 (.199)* -0.284 (.228) -0.491 (.210)* -0.225 (.306) -0.499 (.269) Δ Economic and Financial Stability -0.011 (.018) -0.014 (.021) -0.029 (.022) -0.036 (.025) -0.046 (.025) -0.055 (.028) Δ Political Stability -0.031 (.021) -0.031 (.021) -0.020 (.027) -0.023 (.029) -0.018 (.035) -0.021 (.036) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Stop Years 225 182 159 130 108 85 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. Dependent variable is a 0-1 variable indicating whether there is a surge episode (either GF-led surge or an M&A-led surge). All models are estimated using complementary logistic regression. The variables GDP per capita and natural resources are windsorized at the 1% level. Covariance across surge episodes is accounted for using seemingly unrelated estimated with clustering of standard errors at the country level. Alternative Definition Column 1: 1 standard deviation from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 2: 1.5 standard deviation from the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 3: 1.5 standard deviations from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 4: 2 standard deviations from the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 5: 2 standard deviations from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth 45 Appendix Table B3: Sensitivity Analysis – M&A-led Surges: Alternative Definitions of Surge Baseline 1 SD, Top 20% 1.5 SD, Top 25% 1.5 SD, Top 20% 2 SD, Top 25% 2 SD, Top 20% Regression Percentile Percentile Percentile Percentile Percentile Global Factors Global Risk 0.045 (.028) 0.040 (.028) 0.067 (.031)* 0.057 (.032) 0.060 (.034) 0.055 (.035) Global Liquidity 0.015 (.007) 0.017 (.009) 0.023 (.009)** 0.022 (.009)* 0.019 (.010) 0.018 (.010) Global Growth 0.542 (.097)** 0.576 (.115)** 0.576 (.101)** 0.564 (.120)** 0.553 (.111)** 0.562 (.129)** Regional Contagion 0.032 (.015)* 0.071 (.029)* 0.057 (.029)* 0.097 (.041)* 0.030 (.046) 0.049 (.052) Domestic Factors Stop previous year 0.685 (.501) 0.613 (.562) 0.286 (.772) -0.212 (1.08) X X Surge previous year -0.195 (.540) -0.791 (.852) -1.289 (.895) X X X GDP per Capita (ln) 0.173 (.068)* 0.137 (.068)* 0.156 (.073)* 0.109 (.071) 0.099 (.076) 0.067 (.078) Natural Resources -0.036 (.018)* -0.036 (.016)* -0.034 (.020) -0.033 (.019) -0.054 (.030) -0.048 (.027) Δ Financial Openness 0.433 (.212)* 0.414 (.202)* 0.524 (.212)** 0.509 (.227)* 0.569 (.242)* 0.547 (.243)* Δ Economic and Financial Stability -0.068 (.027)* -0.041 (.027) -0.086 (.029)** -0.057 (.028)* -0.113 (.033)** -0.076 (.029)** Δ Political Stability 0.011 (.042) 0.000 (.050) 0.034 (.046) 0.011 (.056) 0.042 (.049) 0.022 (.056) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Surge Years 57 52 48 44 38 36 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. X – omitted due to perfect multicollineaerity Dependent variable is a 0-1 variable indicating whether there is a surge episode (either GF-led surge or an M&A-led surge). All models are estimated using complementary logistic regression. The variables GDP per capita and natural resources are windsorized at the 1% level. Covariance across surge episodes is accounted for using seemingly unrelated estimated with clustering of standard errors at the country level. Alternative Definition Column 1: 1 standard deviation from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 2: 1.5 standard deviation from the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 3: 1.5 standard deviations from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 4: 2 standard deviations from the 5 year average and falling within the top 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 5: 2 standard deviations from the 5 year average and falling within the top 20th percentile of the entire sample’s FDI-to-GDP ratio growth 46 Appendix Table B4: Sensitivity Analysis – M&A-led Stops: Alternative Definitions of Stops Baseline 1 SD, Bottom 1.5 SD, Bottom 1.5 SD, Bottom 2 SD, Bottom 2 SD, Bottom Regression 20% Percentile 25% Percentile 20% Percentile 25% Percentile 20% Percentile Global Factors Global Risk 0.046 (.031) 0.049 (.036) 0.019 (.035) 0.019 (.038) 0.080 (.044) 0.076 (.048) Global Liquidity 0.003 (.006) 0.009 (.007) 0.002 (.007) 0.005 (.007) 0.005 (.008) 0.007 (.009) Global Growth -0.235 (.112)* -0.179 (.121) -0.260 (.131)* -0.203 (.140) 0.021 (.134) 0.019 (.153) Regional Contagion 0.028 (.017) 0.038 (.032) 0.046 (.028) 0.048 (.034) 0.110 (.034)** 0.130 (.035)** Domestic Factors Stop previous year 0.848 (.612) 0.117 (1.14) X X X X Surge previous year 3.915 (317)** 4.367 (.328)** 4.004 (.380)** 4.638 (.397)** 4.542 (.411)** 4.834 (.435)** GDP per Capita (ln) 0.106 (.068) 0.073 (.070) 0.229 (.076)** 0.261 (.085)** 0.248 (.134) 0.197 (.145) Natural Resources -0.007 (.009) -0.001 (.009) -0.009 (.010) -0.004 (.011) 0.007 (.014) 0.013 (.014) Δ Financial Openness 0.708 (.247)** 0.466 (.395) 0.371 (.466) 0.648 (.469) 0.656 (.580) 0.447 (.715) Δ Economic and Financial Stability 0.026 (.031) 0.467 (.031) 0.016 (.031) 0.053 (.028) -0.054 (.038) -0.026 (.046) Δ Political Stability 0.013 (.041) -0.016 (.050) -0.027 (.048) -0.068 (.059) 0.036 (.070) 0.002 (.079) Number of Observations 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 Number of Stop Years 57 48 41 36 25 23 *p<0.05, **p<0.01; cluster-robust standard errors in parentheses. X – omitted due to perfect multicollineaerity Dependent variable is a 0-1 variable indicating whether there is a surge episode (either GF-led surge or an M&A-led surge). All models are estimated using complementary logistic regression. The variables GDP per capita and natural resources are windsorized at the 1% level. Covariance across surge episodes is accounted for using seemingly unrelated estimated with clustering of standard errors at the country level. Alternative Definition Column 1: 1 standard deviation from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 2: 1.5 standard deviation from the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 3: 1.5 standard deviations from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 4: 2 standard deviations from the 5 year average and falling within the bottom 25th percentile of the entire sample’s FDI-to-GDP ratio growth Alternative Definition Column 5: 2 standard deviations from the 5 year average and falling within the bottom 20th percentile of the entire sample’s FDI-to-GDP ratio growth 47 Appendix Table B5: Sensitivity Analysis – GF-Led Surges: Alternative Variable Definitions Baseline Global Interest Resource Rich Economic Financial Absence of Institutional Democracy Regression Rate Stability Stability Conflict Quality Global Factors Global Risk 0.010 (.014) -0.015 (.015) 0.007 (.014) 0.010 (.014) 0.010 (.014) 0.014 (.014) 0.015 (.014) 0.015 (.014) Global Liquidity 0.013 (.004)** -0.103 (.055) 0.012 (.004)** 0.013 (.004)** 0.013 (.004)** 0.013 (.004)** 0.013 (.004)** 0.013 (.004)** Global Growth 0.076 (.058) 0.014 (.061) 0.083 (.060) 0.078 (.059) 0.085 (.057) 0.092 (.057) 0.093 (.058) 0.092 (.056) Regional Contagion 0.021 (.008)* 0.018 (.009)* 0.022 (.008)** 0.023 (.008)* 0.022 (.008)* 0.020 (.008)* 0.020 (.008)* 0.020 (.008)* Domestic Factors Stop previous year 0.263 (.231) 0.258 (.233) 0.252 (.234) 0.269 (.234) 0.260 (.231) 0.260 (.239) 0.260 (.238) 0.269 (.239) Surge previous year -0.014 (.205) -0.021 (.203) -0.011 (.204) -0.021 (.203) -0.020 (.205) 0.001 (.207) -0.000 (.208) 0.004 (.207) GDP per Capita (ln) -0.148 (.034)** -0.158 (.034)** -0.144 (.038)** -0.148 (.034)** -0.149 (.034)** -0.144 (.035)** -0.144 (.035)** -0.143 (.034)** Natural Resources 0.007 (.003)* 0.005 (.003) 0.296 (.130)* 0.007 (.003)* 0.007 (.003) 0.007 (.003)* 0.007 (.003)* 0.007 (.003)* Δ Financial Openness 0.118 (.165) 0.133 (.168) 0.133 (.162) 0.128 (.164) 0.129 (.163) 0.123 (.163) 0.124 (.162) 0.118 (.163) Δ Economic and 0.020 (.015) 0.017 (.016) 0.004 (.021) 0.019 (.023) 0.032 (.023) 0.013 (.015) 0.013 (.015) 0.013 (.015) Financial Stability Δ Political Stability -0.038 (.020) -0.040 (.020) -0.032 (.020) -0.033 (.020) -0.036 (.018)* -0.012 (.068) -0.020 (.091) 0.070 (.072) Number of 1768 1768 1768 1768 1768 1768 1768 1768 Observations Number of Countries 95 95 95 95 95 95 95 95 Number of Surge 207 207 207 207 207 207 207 207 Years 48 Appendix Table B6: Sensitivity Analysis – GF-Led Stops: Alternative Variable Definitions Baseline Global Interest Resource Rich Economic Financial Absence of Institutional Democracy Regression Rate Stability Stability Conflict Quality Global Factors Global Risk 0.013 (.013) 0.010 (.013) 0.013 (.013) 0.012 (.013) 0.014 (.013) 0.014 (.014) 0.014 (.013) 0.014 (.013) Global Liquidity 0.002 (.003) 0.018 (.058) 0.001 (.003) 0.002 (.003) 0.003 (.003) 0.003 (.003) 0.002 (.003) 0.003 (.003) Global Growth -0.112 (.053)* -0.124 (.053)* -0.119 (.053)* -0.098 (.056) -0.119 (.052)* -0.108 (.052)* -0.113 (.053)* -0.108 (.052)* Regional Contagion -0.007 (.010) -0.006 (.009) -0.008 (.010) -0.007 (.010) -0.008 (.010) -0.006 (.010) -0.006 (.010) -0.007 (.010) Domestic Factors Stop previous year -0.735 (.332)* -0.735 (.331)* -0.737 (.331)* -0.735 (.331)* -0.745 (.333)* -0.953 (.318)** -0.951 (.317)** -0.952 (.317)** Surge previous year 1.533 (.153)** 1.531 (.153)** 1.507 (.153)** 1.529 (.154)** 1.541 (.154)** 1.515 (.156)** 1.519 (.157)** 1.520 (.156)** GDP per Capita (ln) -0.067 (.036) -0.066 (.035) -0.080 (.035)* -0.067 (.035) -0.066 (.035) -0.062 (.035) -0.062 (.035) -0.063 (.035) Natural Resources 0.006 (.003)* 0.006 (.003)* 0.282 (.126)* 0.006 (.003)* 0.006 (.003)* 0.006 (.003) 0.005 (.003) -.006 (.003)* Δ Financial Openness -0.247 (.195) -0.253 (.194) -0.236 (.192) -0.239 (.196) -0.265 (.200) -0.256 (.197) -0.265 (.199) -0.252 (.195) Δ Economic and -0.011 (.018) -0.014 (.019) -0.009 (.016) -0.036 (.023) 0.014 (.031) -0.019 (.018) -0.018 (.019) -0.020 (.018) Financial Stability Δ Political Stability -0.031 (.021) -0.032 (.020) -0.031 (.020) -0.028 (.020) -0.040 (.019)* -0.018 (.082) 0.100 (.126) -0.061 (.099) Number of Observations 1768 1768 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 95 95 Number of Stop Years 225 225 225 225 225 225 225 225 49 Appendix Table B7: Sensitivity Analysis – M&A-led Surges: Alternative Variable Definitions Baseline Global Interest Resource Rich Economic Financial Absence of Institutional Democracy Regression Rate Stability Stability Conflict Quality Global Factors Global Risk 0.045 (.028) 0.019 (.027) 0.058 (.029)* 0.047 (.029) 0.046 (.028) 0.046 (.028) 0.040 (.027) 0.043 (.027) Global Liquidity 0.015 (.007) -0.159 (.111) 0.018 (.007)* 0.016 (.007)* 0.017 (.007)* 0.014 (.007)* 0.012 (.007) 0.014 (.007)* Global Growth 0.542 (.097)** 0.475 (.103)** 0.550 (.094)** 0.537 (.100)** 0.512 (.096)** 0.548 (.100)** 0.537 (.099)** 0.544 (.097)** Regional Contagion 0.032 (.015)* 0.037 (.015)* 0.035 (.014)* 0.030 (.015)* 0.033 (.015)* 0.032 (.015)* 0.031 (.015)* 0.032 (.014)* Domestic Factors Stop previous year 0.685 (.501) 0.629 (.485) 0.669 (.505) 0.677 (.503) 0.685 (.501) 0.723 (.499) 0.667 (.512) 0.741 (.515) Surge previous year -0.195 (.540) -0.206 (.553) -0.221 (.556) -0.165 (.557) -0.227 (.547) -0.152 (.522) -0.150 (.527) -0.139 (.514) GDP per Capita (ln) 0.173 (.068)* 0.167 (.069)* 0.153 (.067)* 0.171 (.067)* 0.169 (.068)* 0.170 (.069)* 0.175 (.068)* 0.167 (.070)* Natural Resources -0.036 (.018)* -0.037 (.017)* -0.819 (.322)* -0.036 (.017)* -0.036 (.018)* -0.035 (.017)* -0.035 (.017)* -0.034 (.017)* Δ Financial Openness 0.433 (.212)* 0.428 (.178)** -0.377 (.194) 0.374 (.189)* 0.382 (183)* 0.405 (.179)* 0.395 (.181)* 0.410 (.179)* Δ Economic and -0.068 (.027)* -0.076 (.029)** -0.052 (.022)* -0.084 (.042)* -0.090 (.051) -0.069 (.029)* -0.066 (.029)* -0.069 (.029)* Financial Stability Δ Political Stability 0.011 (.042) 0.007 (.043) 0.006 (.046) 0.001 (.044) 0.000 (.044) 0.053 (.130) 0.286 (.241) -0.023 (.200) Number of Observations 1768 1768 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 95 95 Number of Surge Years 57 57 57 57 57 57 57 57 50 Appendix Table B8: Sensitivity Analysis – M&A-led Stops: Alternative Variable Definitions Baseline Global Interest Resource Rich Economic Financial Absence of Institutional Democracy Regression Rate Stability Stability Conflict Quality Global Factors Global Risk 0.046 (.031) 0.012 (.029) 0.049 (.031) 0.045 (.031) 0.044 (.031) 0.039 (.030) 0.036 (.030) 0.039 (.031) Global Liquidity 0.003 (.006) -0.552 (.136)** 0.003 (.006) 0.003 (.006) 0.003 (.006) 0.002 (.006) 0.001 (.006) 0.002 (.006) Global Growth -0.235 (.112)* -0.231 (.100)* -0.210 (.106)* -0.286 (.114)* -0.212 (.105)* -0.260 (.108)* -0.286 (.106)** -0.259 (.108)* Regional Contagion 0.028 (.017) 0.018 (.018) 0.028 (.016) 0.029 (.017) 0.025 (.017) 0.026 (.017) 0.024 (.017) 0.026 (.016) Domestic Factors Stop previous year 0.848 (.612) 0.765 (.609) 0.898 (.627) 0.863 (.618) 0.861 (.622) 0.873 (.619) 0.905 (.621) 0.872 (.623) Surge previous year 3.915 (317)** 3.940 (.314)** 3.927 (.329)** 3.960 (.318)** 3.905 (.321)** 3.863 (.324)** 3.874 (.327)** 3.861 (.321)** GDP per Capita (ln) 0.106 (.068) 0.080 (.071) 0.091 (.064) 0.111 (.068) 0.094 (.070) 0.123 (.068) 0.132 (.068) 0.123 (.070) Natural Resources -0.007 (.009) -0.011 (.009) -0.166 (.293) -0.006 (.009) -0.006 (.009) -0.006 (.009) -0.007 (.009) -0.006 (.009) Δ Financial Openness 0.708 (.247)** 0.822 (.272)** 0.623 (.225)** 0.680 (.249)** 0.700 (.249)** 0.702 (.240)** 0.704 (.236)** 0.701 (.241)** Δ Economic and 0.026 (.031) 0.046 (.032) 0.023 (.029) 0.101 (.059) -0.021 (.038) 0.025 (.030) 0.031 (.030) 0.025 (.030) Financial Stability Δ Political Stability 0.013 (.041) 0.034 (.045) 0.018 (.041) 0.008 (.040) 0.023 (.040) 0.008 (.108) 0.366 (.249) 0.012 (.265) Number of Observations 1768 1768 1768 1768 1768 1768 1768 1768 Number of Countries 95 95 95 95 95 95 95 95 Number of Stop Years 57 57 57 57 57 57 57 57 51 Appendix Table B9: Sensitivity Analysis – GF-Led Surges: Additional Control Variables Baseline Region Domestic Exchange Rate Inflation Trade Tariffs Country Size Regression Dummies Growth Openness Global Factors Global Risk 0.010 (.014) 0.011 (.015) 0.010 (.014) 0.009 (.014) 0.012 (.014) 0.002 (.015) 0.001 (.015) 0.010 (.015) Global Liquidity 0.013 (.004)** 0.013 (.004)** 0.013 (.004)** 0.014 (.004)** 0.013 (.004)** 0.013 (.004)** 0.012 (.004)** 0.013 (.004)** Global Growth 0.076 (.058) 0.008 (.059) 0.082 (.060) 0.080 (.060) 0.073 (.059) 0.009 (.060) 0.011 (.059) 0.074 (.050) Regional Contagion 0.021 (.008)* 0.021 (.008)* 0.021 (.008)* 0.021 (.008)* 0.021 (.008)* 0.017 (.009) 0.023 (.009)** 0.021 (.008)** Domestic Factors Stop previous year 0.263 (.231) 0.264 (.229) 0.261 (.232) 0.259 (.237) 0.272 (.232) 0.259 (.248) 0.106 (.262) 0.205 (.236) Surge previous year -0.014 (.205) -0.021 (.207) -0.008 (.206) -0.038 (.214) -0.013 (.205) -0.057 (.218) -0.016 (.216) -0.009 (.206) GDP per Capita (ln) -0.148 (.034)** -0.163 (.048)** -0.143 (.036)** -0.145 (.035)** -0.148 (.034)** -0.124 (.039)** -0.174 (.038)** -0.148 (.044)** Natural Resources 0.007 (.003)* 0.008 (.004)* 0.007 (.003)* 0.005 (.004) 0.006 (.003) 0.008 (.004)* 0.006 (.004) 0.007 (.004) Δ Financial Openness 0.118 (.165) 0.118 (.167) 0.124 (.165) 0.090 (.167) 0.118 (.165) 0.186 (.187) 0.246 (.167) 0.114 (.167) Δ Economic and 0.020 (.015) 0.020 (.014) 0.023 (.014) 0.026 (.015) 0.021 (.015) 0.033 (.143) 0.019 (.015) 0.020 (.016) Financial Stability Δ Political Stability -0.038 (.020) -0.038 (.020) -0.035 (.020) -0.042 (.019)* -0.038 (.020) -0.061 (.021)* -0.051 (.022)* -0.030 (.020) Domestic Growth -0.012 (.012) Δ Exchange Rate 0.423 (.111) Δ Inflation 0.081 (.091) Δ Trade Openness 0.025 (.006)** Δ Average Tariff -0.020 (.013) Population (ln) -0.211 (.174) Number of 1768 1768 1768 1750 1768 1601 1552 1717 Observations Number of Countries 95 95 95 94 95 93 93 93 Number of Surge 207 207 207 205 207 186 184 203 Years 52 Appendix Table B10: Sensitivity Analysis – GF-Led Stops: Additional Control Variables Baseline Region Domestic Exchange Rate Inflation Trade Tariffs Country Size Regression Dummies Growth Openness Global Factors Global Risk 0.013 (.013) 0.012 (.013) 0.013 (.013) 0.012 (.013) 0.013 (.013) 0.016 (.014) 0.023 (.015) 0.012 (.014) Global Liquidity 0.002 (.003) 0.002 (.003) 0.002 (.003) 0.003 (.003) 0.002 (.003) 0.003 (.003) 0.004 (.003) 0.002 (.003) Global Growth -0.112 (.053)* -0.121 (.053)* -0.108 (.052)* -0.126 (.052)* -0.113 (.053)* -0.102 (.062) -0.109 (.062) -0.124 (.053)* Regional Contagion -0.007 (.010) -0.009 (.010) -0.007 (.010) -0.007 (.010) -0.007 (.010) -0.006 (.010) -0.014 (.009) -0.008 (.010) Domestic Factors Stop previous year -0.735 (.332)* -0.727 (.333)* -0.733 (.332)* -0.728 (.333)* -0.734 (.332)* -0.825 (.324)* -0.490 (.349) -0.717 (.332)* Surge previous year 1.533 (.153)** 1.542 (.156)** 1.541 (.156)** 1.564 (.152)** 1.532 (.153)** 1.553 (.175)** 1.486 (.182)** 1.522 (.152)** GDP per Capita (ln) -0.067 (.036) -0.068 (.044) -0.062 (.037) -0.064 (.037) -0.067 (.035) -0.040 (.039) -0.048 (.044) -0.073 (.041) Natural Resources 0.006 (.003)* 0.004 (.003) 0.006 (.003)* 0.007 (.003)* 0.006 (.003)* 0.006 (.003) 0.004 (.004) 0.006 (.003)* Δ Financial Openness -0.247 (.195) -0.242 (.205) -0.238 (.193) -0.253 (.196) -0.246 (.196) -0.296 (.200) -0.339 (.216) -0.258 (.198) Δ Economic and -0.011 (.018) -0.012 (.019) -0.009 (.019) -0.012 (.019) -0.011 (.019) -0.002 (.020) -0.002 (.020) -0.008 (.019) Financial Stability Δ Political Stability -0.031 (.021) -0.032 (.021) -0.029 (.021) -0.033 (.021) -0.031 (.021) -0.030 (.171) -0.002 (.019) -0.032 (.021) Domestic Growth -0.009 (.015) Δ Exchange Rate -0.097 (.157) Δ Inflation 0.047 (.093) Δ Trade Openness -0.013 (.009) Δ Average Tariff 0.017 (.024) Population (ln) 0.105 (.217) Number of 1768 1768 1768 1750 1768 1601 1552 1717 Observations Number of Countries 95 95 95 94 95 93 93 93 Number of Surge 225 225 225 216 225 191 179 212 Years 53 Appendix Table B11: Sensitivity Analysis – M&A-led Surges: Additional Control Variables Baseline Region Domestic Exchange Rate Inflation Trade Tariffs Country Size Regression Dummies Growth Openness Global Factors Global Risk 0.045 (.028) 0.048 (.028) 0.043 (.028) 0.048 (.027) 0.045 (.028) 0.043 (.029) 0.039 (.028) 0.045 (028) Global Liquidity 0.015 (.007) 0.017 (.008)* 0.015 (.007)* 0.016 (.007)* 0.016 (.007)* 0.016 (.007)* 0.015 (.007)* 0.016 (.007)* Global Growth 0.542 (.097)** 0.539 (.091)** 0.554 (.098)** 0.424 (.095)** 0.529 (.097)** 0.483 (.098)** 0.511 (.099)** 0.542 (.097)** Regional Contagion 0.032 (.015)* 0.014 (.017) 0.032 (.015)* 0.028 (.013)* 0.036 (.014)* 0.032 (.014)* 0.030 (.015)* 0.029 (.014)* Domestic Factors Stop previous year 0.685 (.501) 0.523 (.498) 0.688 (.500) 0.592 (.403) 0.695 (.503) 0.743 (.401) 0.755 (.517) 0.632 (.497) Surge previous year -0.195 (.540) -0.231 (.533) -0.194 (.539) -0.225 (.549) -0.186 (.539) -0.142 (.540) -0.214 (.537) -0.278 (.533) GDP per Capita (ln) 0.173 (.068)* 0.174 (.079)* 0.179 (.070)* 0.186 (.070)** 0.160 (.068)* 0.167 (.070)* 0.197 (.067)** 0.242 (.072)** Natural Resources -0.036 (.018)* -0.036 (.020) -0.035 (.017)* -0.036 (.017)* -0.037 (.018)* -0.031 (.018) -0.039 (.021) -0.036 (.017)* Δ Financial Openness 0.433 (.212)* 0.339 (.173)* 0.406 (.184)* 0.414 (.190)* 0.428 (.183)* 0.410 (.183)* 0.416 (.188)* 0.387 (.181)* Δ Economic and -0.068 (.027)* -0.064 (.028)* -0.064 (.026)* -0.084 (.031)** -0.064 (.028)* -0.067 (.028)* -0.068 (.028)* -0.071 (.027)** Financial Stability Δ Political Stability 0.011 (.042) 0.006 (.043) 0.014 (.042) 0.003 (.048) 0.011 (.042) 0.015 (.043) 0.013 (.045) 0.012 (.042) Domestic Growth -0.019 (.023) Δ Exchange Rate (ln) -1.421 (.488)* Δ Inflation 0.451 (.188)* Δ Trade Openness 0.016 (.010) Δ Average Tariff 0.044 (.016)* Population (ln) -1.391 (.723) Number of 1768 1768 1768 1750 1768 1601 1552 1717 Observations Number of Countries 95 95 95 94 95 93 93 93 Number of Surge 57 57 57 57 57 55 55 57 Years 54 Appendix Table B12: Sensitivity Analysis – M&A-led Stops: Additional Control Variables Baseline Region Domestic Exchange Inflation Trade Tariffs Country Size Regression Dummies Growth Rate Openness Global Factors Global Risk 0.046 (.031) 0.052 (.032) 0.046 (.031) 0.045 (.032) 0.046 (.031) 0.019 (.032) 0.035 (.032) 0.045 (.031) Global Liquidity 0.003 (.006) 0.002 (.006) 0.003 (.006) 0.003 (.006) 0.003 (.006) -0.006 (.006) 0.002 (.006) 0.003 (.006) Global Growth -0.235 (.112)* -0.265 (.116)* -0.236 (.112)* -0.247 (.116)* -0.238 (.113)* -0.242 (.136) -0.264 (.117)* -0.236 (.113)* Regional Contagion 0.028 (.017) 0.012 (.019) 0.028 (.017) 0.027 (.016) 0.027 (.016) 0.021 (.016) 0.026 (.017) 0.026 (.017) Domestic Factors Stop previous year 0.848 (.612) 0.789 (.588) 0.849 (.622) 0.763 (.620) 0.841 (.619) 0.975 (.629) 0.834 (.616) 0.813 (.618) Surge previous year 3.915 (317)** 3.819 (.313)** 3.915 (.317)** 3.863 (.316)** 3.910 (.318)** 4.059 (.348)** 3.829 (.336)** 3.867 (.323)** GDP per Capita (ln) 0.106 (.068) 0.069 (.063) 0.106 (.069) 0.109 (.071) 0.107 (.068) 0.093 (.069) 0.099 (.072) 0.130 (.073) Natural Resources -0.007 (.009) -0.009 (.009) -0.007 (.009) -0.009 (.009) -0.007 (.009) -0.014 (.011) -0.012 (.012) -0.006 (.009) Δ Financial Openness 0.708 (.247)** 0.734 (.262)** 0.709 (.246)** 0.785 (.262)** 0.711 (.248)** 0.744 (.237)** 0.719 (.259)** 0.698 (.248)** Δ Economic and 0.026 (.031) 0.032 (.030) 0.025 (.0331) 0.006 (.036) 0.027 (.031) 0.016 (.035) 0.022 (.033) 0.027 (.032) Financial Stability Δ Political Stability 0.013 (.041) -0.007 (.042) 0.013 (.041) 0.012 (.044) 0.013 (.041) 0.004 (.032) 0.018 (.045) 0.016 (.043) Domestic Growth 0.002 (.036) Δ Exchange Rate -2.310 (1.02)* Δ Inflation 0.175 (.069) Δ Trade Openness -0.041 (.015)* Δ Average Tariff 0.020 (.030) Population (ln) 0.046 (.943) Number of 1768 1768 1768 1750 1768 1601 1552 1717 Observations Number of Countries 95 95 95 94 95 93 93 93 Number of Surge 57 57 57 56 57 52 54 56 Years 55