Economic Globalization and Transnational Terrorism - Semantic Scholar

JOURNAL ARTICLE 10.1177/0022002703262869 Li, Schaub /OF ECONOMIC CONFLICTGLOBALIZATION RESOLUTION AND TERRORISM

Economic Globalization and Transnational Terrorism A POOLED TIME-SERIES ANALYSIS QUAN LI DREW SCHAUB

Department of Political Science Pennsylvania State University

The effect of economic globalization on the number of transnational terrorist incidents within countries is analyzed statistically, using a sample of 112 countries from 1975 to 1997. Results show that trade, foreign direct investment (FDI), and portfolio investment have no direct positive effect on transnational terrorist incidents within countries and that economic developments of a country and its top trading partners reduce the number of terrorist incidents inside the country. To the extent that trade and FDI promote economic development, they have an indirect negative effect on transnational terrorism. Keywords: economic globalization; transnational terrorism; economic integration; foreign direct investment

Do countries that are more integrated into the global economy also experience more

transnational terrorist incidents within their borders? In the months following September 11, 2001, many people questioned the future viability of an open global economy, believing that economic globalization had contributed to the transnational terrorist attacks on the World Trade Center and the Pentagon. Indeed, more than $1.4 billion of goods cross the borders of the North American Free Trade Agreement (NAFTA) countries every day (McDonald 2002). In the last year alone, cargo vessels off-loaded roughly 18 million, 40-foot-long cargo containers at American ports, often in single batches as large as 8,000. Ports and border crossings around the world have similarly experienced an increasing volume of daily shipping and trucking activity. The increasing numbers of trucks and container vessels that facilitate international commerce

AUTHORS’ NOTE: We thank Scott Bennett, Sarah Knight, Michael Koch, Manus Midlarsky, Todd Sandler, George Shambaugh, and participants of the Pennsylvania State University International Relations Reading Group for comments and suggestions. We also thank Todd Sandler and Edward Mickolus for help with data and Adrienne Lauzon and Andreea S. Mihalache for research assistance. An earlier version of this study was presented at the annual meeting of the International Studies Association, Portland, OR, February 2003. The replication data set is available at http://www.yale.edu/unsy/jcr/jcrdata.html. JOURNAL OF CONFLICT RESOLUTION, Vol. 48 No. 2, April 2004 230-258 DOI: 10.1177/0022002703262869 © 2004 Sage Publications

230

Li, Schaub / ECONOMIC GLOBALIZATION AND TERRORISM

231

increase the likelihood of a terrorist successfully smuggling himself or a weapon undetected across a vulnerable border. Financial markets have also experienced a drastic increase in the volume of cross-national transactions. The daily turnover in the foreign exchange market is nearly $2 trillion, exceeding the value of all traded goods and services. The growing number of international financial transactions threatens to overwhelm the enforcement officers who attempt to intercept money meant to sponsor acts of terrorism. Therefore, the risks facing international criminals—including terrorists—who use global networks to facilitate their operations decrease substantially. The global war on terrorism after September 11 has led to tightened security almost around the world, increasing the costs associated with conducting international business. In contrast, a small number of scholars and policy makers argue that economic openness will result in a reduction in transnational terrorism. Proponents of this view believe that economic globalization promotes economic development, which in turn alters the decision calculus of terrorist groups toward a reduction in terrorist activities. Progress in economic development due to trade and capital flows removes one of the main incentives for people to engage in terrorist activities out of desperation and poverty. Although this argument is relatively new and less well developed, many policy makers have turned to it for a solution to global terrorism. Little systematic empirical analysis has been conducted to estimate the effect of economic globalization on transnational terrorism. In this study, we assess empirically whether economic globalization increases or decreases transnational terrorist incidents inside countries. We consider economic globalization as implying the growing integration of a country’s economy into the world’s goods and financial and production capital markets, following the convention in the globalization literature (see, e.g., Frankel 2000; Bhalla 2002; Garrett 1998; Li and Reuveny 2003). We use economic globalization and economic integration interchangeably in this study. Although globalization is often interpreted to imply global diffusion of technology and culture, we focus on the impact of economic globalization. Following Enders and Sandler (1993, 1999, 2000), we define terrorism as the premeditated or threatened use of extra-normal violence or force to obtain a political, religious, or ideological objective through the intimidation of a large audience. A transnational terrorist incident in a country involves victims, perpetrators, targets, or institutions of another country. We choose to focus on transnational terrorist incidents, such as terrorist attacks that are initiated by foreign terrorists against some domestic target within a country or committed by domestic terrorists against some foreign target within a country.1 The suicide attacks on the World Trade Center and the Pentagon by foreign nationals on September 11 represent one type of transnational terrorist event. The targeting of U.S. banks and bank personnel by Greek terrorists in Greece offers an example of the other types of transnational terrorism described above. Although transnational terrorist incidents can affect a country’s integration into the global economy, we focus on how economic globalization affects transnational terrorist incidents inside a country. To date, there has been no systematic empirical analysis of whether economic globalization induces or reduces transnational terrorist incidents within 1. The bombing of the Alfred P. Murrah Federal Building in Oklahoma City by Timothy McVeigh in 1995 is an example of a domestic terrorist event.

232

JOURNAL OF CONFLICT RESOLUTION

countries. Our analysis fills this gap. The findings obviously bear significant policy implications for the war on terrorism and the future of the global economy. Our empirical analysis covers 112 countries from 1975 to 1997. The dependent variable is the number of transnational terrorist incidents initiated within a country in a year. The independent variables include a country’s openness to trade, foreign direct investment (FDI), portfolio investment, a country’s economic development, and the development of its top economic partner countries. We control for country size, level of democracy, government capability, interstate conflict, temporal persistence, regional variations, and temporal unit effects. The findings can be summarized as follows. Trade, FDI, and financial capital flows of a country have no direct positive effect on the number of transnational terrorist incidents initiated within the country. However, the economic development of a country and its top economic partners reduces the number of transnational terrorist incidents within the country. To the extent that trade and FDI promote economic development, economic globalization has an indirect negative effect on transnational terrorism. These results are robust to alternative model specifications and statistical estimators. The rest of the study proceeds as follows. In the first section, we discuss competing arguments on how economic globalization increases or decreases the number of transnational terrorist incidents within a country. In the second section, we present the research design for our empirical analysis. In the third section, we discuss our findings and conduct a sensitivity analysis, and in the last section, we conclude the article. HOW DOES ECONOMIC INTEGRATION AFFECT TRANSNATIONAL TERRORISM IN A COUNTRY? TERRORISTS AS RATIONAL ACTORS

Causal analysis of how economic globalization affects transnational terrorist incidents presupposes that terrorists behave rationally. The notion that terrorists are rational actors merely implies that terrorists are goal oriented, can rank order their preferences, and act to maximize their preferences within budgetary constraints (see, e.g., Crenshaw 1981; Sandler, Enders, and Lapan 1991). Several implications emerge from this interpretation of terrorist behavior. First, terrorist groups often are involved in nationalistic, separatist, racist, nihilistic, religious, Marxist, or other economic movements. These groups hold particular political, religious, or ideological goals and strive to attain them through long periods of struggle (Crenshaw 1981; Hoffman 1997, 1998; Midlarsky, Crenshaw, and Yoshida 1980; Stohl 1979; Wilkinson 1986). Many terrorist groups are also found to negotiate with authorities to maximize their benefits and refrain from typically “irrational” behaviors (behaviors not designed to fulfill the stated goal or those that work against the fulfillment of that goal), such as the indiscriminate killing of hostages. Second, terrorist groups allocate resources between legal and illegal activities to maximize utility subject to budgetary constraints (Sandler, Tschirhart, and Cauley 1983). The respective levels of legal and illegal activity carried out by a terrorist group depend on three factors: the relative costs between


233

legal and illegal activities, the relative gains between the two types of activities, and the total resources available. A rise in the total resources of a group should correspond with a rise in its overall activities. A decline in the cost per unit of illegal activity should result in a substitution from the legal activity if the relative gains remain the same. Third, rational terrorist groups not only maximize their preferences but also react to constraints. It is useful to distinguish between specific and structural constraints. Specific constraints, such as the implementation of metal detectors at airports, affect only the costs associated with an individual legal or illegal mode of operation, merely causing substitution among the various modes of operation (Enders and Sandler 1993). In contrast, structural constraints such as the end of the cold war (Enders and Sandler 1999) affect the relative costs of all modes of legal and illegal activities, causing overall increases or decreases in terrorist activity within a state. These conceptual expectations of rational terrorist behaviors appear to be substantiated empirically. Gurr (1979) and Mickolus (1979) show that terrorists employ more expensive and riskier modes of attack less often than less expensive and less risky modes of attack. Jenkins, Johnson, and Rondfeldt (1977) suggest that the high success rates of hostage missions indicate that terrorists can be rationally expected to favor modes of attack that are likely to lead to the attainment of a particular goal. The high success rates associated with this type of logistically complex activity also signify that terrorist groups effectively plan and execute operations with success as a goal, a further indication of rationality. The experience of the Greek government offers yet another example. As the Greek government took action to limit terrorism within Greece, the frequency of terrorist events in Greece declined over time while terrorist organizations conducted an increasing number of operations elsewhere in Europe. The increased deterrence by the Greek government changed the costs associated with terrorism in one state and affected the relative cost of terrorism in other states (Sandler and Lapan 1988; Sandler 1997). Enders and Sandler (1993) show that an increase in the cost of one mode of terrorist operation across the entire international system leads to long- and short-run changes in the global frequency of other modes. For example, the worldwide implementation of metal detectors in airports led to a decrease in the number of worldwide skyjackings but resulted in an increase in the use of other similar modes of attack. The higher relative price of skyjackings forced terrorists to substitute other logistically similar modes of attack, most notably kidnappings and barricade and hostage-taking missions. In addition to specific antiterrorist efforts, terrorist activities are also affected by structural constraints. Eyerman (1998) tests competing arguments on how regime type represents a structural constraint that affects the overall level of illegal terrorist activity experienced by a state. The breakup of the former Soviet Union provides another example of a structural constraint. Following the breakup of the Soviet Union, the total level of terrorist resources fell worldwide, as the funding previously provided by the Soviet Union to terrorist organizations slowly dried up. This overall decline in the level of resources controlled by terrorist organizations resulted in a change in the relative costs associated with both illegal and legal operations. The cost of illegal activity rose as the resources to fund such activity disappeared, causing the relative cost of legal

234


activity to fall. This resulted in a decrease in both the absolute and relative levels of terrorism during the last three-quarters of 1994 (Enders and Sandler 1999). In addressing these issues related to the rationality of terrorist behaviors, the previous literature on transnational terrorism has set the stage for analyzing how economic globalization affects transnational terrorist incidents. For economic globalization to affect such incidents within a country, it has to affect the incentives of terrorists, the structural constraints on terrorist behaviors, or both. Below, we discuss how economic globalization can increase or reduce transnational terrorist incidents within a country. POSITIVE EFFECT OF ECONOMIC GLOBALIZATION ON TRANSNATIONAL TERRORISM

Many policy makers, journalists, and scholars believe, especially after the terrorist attacks of September 11, that globalization causes an increase in the frequency of transnational terrorism. For example, “the World Trade Center (attack) exposed the dark side of global interconnections: the ease with which the West’s enemies and their resources can move around the world” (Is it at risk? 2002). Economic globalization can act as a structural constraint that alters the relative costs between legal and illegal activities and affects the decision calculus of transnational terrorists. More specifically, as globalization increases, the cost of illegal activity declines relative to the cost of legal activity, and the overall level of terrorism increases. This decreased risk results from the expansion of the trade, financial, and production investment networks in the global economy. Computers, chips, and satellites change significantly the structure of international finance, reducing the risks associated with illegal transnational financial transactions (Strange 1998, 25). The “digitization” of money allows the wide use of credit cards and smart cards, facilitates the instantaneous transfer of funds across borders, and decreases the probability of being caught in transporting and using illegally obtained funds. The integration of the global financial market also leads to an increase in the sheer volume of financial transactions. As the cost of financial transactions decreases, their volume increases significantly. Monetary authorities are less and less able to regulate transactions effectively because those transactions have multiplied so rapidly in number (Strange 1998, 25). The huge volume of financial transactions that occur on a daily basis—more than $1 trillion worth—makes tracing terrorist funds an “extremely difficult task” (Weintraub 2002). Financial globalization therefore limits the area of control that governments have over financial matters, as foreign exchange transactions become practically untraceable (Kobrin 1997). The decreasing effectiveness of governments to monitor financial transactions decreases the risks in the illegal transactions used to sponsor terrorist operations. Similarly, as the volume of international trade increases, the risk associated with illegal trading also decreases. Trade between the United States, Canada, and Mexico, for example, has more than doubled over the past decade to at least $1.4 billion worth a day. Over the same period, the number of customs agents responsible for discovering contraband or illegally traded goods has remained the same. Statistics concerning maritime shipments demonstrate the phenomenon even more clearly. Of the 18 mil-


235

lion cargo containers that arrive by sea each year in the United States, only 2% to 10% of them are searched (McDonald 2002). Other sources claim that the lower estimate is much more accurate (The Trojan box 2002). These figures allow a numerical value to be placed on the level of risk associated with illegal international trade. Obviously, a 2% chance of discovery is much too low to seriously prevent determined terrorist organizations from smuggling weapons of mass destruction across international borders or from conducting profitable illegal trade to finance terrorist operations. Transnational terrorists also often take advantage of the international trade network to market goods or services in an effort to marshal resources with which to carry out their criminal activities. Terrorist organizations often rely on the international trade network to trade contraband to fund their various operations. The increasingly fluid nature of the global investment and distribution networks makes such trade increasingly less risky and therefore much more likely (Matthew and Shambaugh 1998). For instance, Osama bin Laden’s Al Qaeda terrorist network relied on illegal as well as legal international trade to fund its operations (Shahar 2001). Governments, however, are not well equipped to deal with these changes that facilitate illegal international transactions. For one thing, the number of enforcement officers required to perform proper regulatory and monitory functions is inadequate. “The global financial system provides many more opportunities than it can ever hope to forestall or block . . . law enforcement is playing a game of catch up which it is almost certainly destined to lose” (Strange 1998, 128). Bank examiners and government investigators simply cannot supply the necessary man-hours to effectively monitor the enormous volume of transactions that take place daily (Strange 1998). The failure of governments to provide appropriate levels of enforcement stems not only from the enormous volume of transactions but also from the enormous salary gap for the highly skilled information technology specialists required to fill such positions. As technology continues to drive economic expansion, the government is less competitive financially than the private sector in recruiting the high-tech talent (Naím 2000). The government is limited in its ability to monitor and sanction tech-savvy money-launderers, transnational terrorists, drug traffickers, and other international criminals who rely on the international financial market. The international banking system also tends toward secrecy, further facilitating the sustained symbiotic relationship between the international banking industry and international criminals (Strange 1998). As banks compete internationally for clients, they must stress their historic pledge to protect the privacy of their customers. Government demands on banks to disclose the illegal banking activities of their customers represent “sticks, without corresponding carrots” (Strange 1998, 131). Banks, threatened with punitive action for failing to disclose any illegal actions by their customers, face a catch-22. If they decide to play above board and deny privacy to the suspicious clientele, they risk losing profit and alienating their entire customer base. Therefore, banks are likely to continue to protect the privacy of their customers and hence implicitly facilitate the continued use of the international financial system by transnational terrorists. An article in The Wall Street Journal (Phillips 2002), which discusses the current difficulties of the Bush administration to gain compliance from European banks in aiding the U.S. war on terrorism, suggests that this is indeed the case; international

236


banks are unlikely to disclose the private information of their clientele. Therefore, the relationship between states and banking interests demonstrates yet another way that globalization reduces the costs of operations for transnational terrorist organizations. Matthew and Shambaugh (1998) emphasize the link between the international banking community and transnational terrorist organizations, arguing that only through cooperation with the international banking community can governments hope to “disrupt the transnational web of incentives, opportunities, and capabilities enough to discourage terrorism.” Economic globalization may also positively affect the level of transnational terrorism in a state by increasing the density of desirable targets present within that state. According to Midlarsky, Crenshaw, and Yoshida (1980), most victims of transnational terrorism are Western. The increased presence of Western targets that necessarily accompanies expanding globalization, therefore, is likely to increase the probability that a terrorist group will choose a particular state as the location of an attack. In other words, holding other effects of globalization constant, economic globalization raises the probability of transnational terrorism simply by increasing the density of suitable targets in more integrated states. Examples of suitable targets, placed in states around the globe, abound. Multinational banks and corporations operated by individuals of many nationalities conduct business, send representatives, and maintain offices in multiple countries. Most terrorist activities in South America during the late 1960s and early 1970s were directed against American or European business personnel employed by American corporations (Midlarsky, Crenshaw, and Yoshida 1980). The extent to which many businesspeople travel internationally also increases the density of specific national targets in various locations. Targets representing national governments also abound and have frequently been the target of terrorist attack (Enders and Sandler 1999). NEGATIVE EFFECT OF ECONOMIC GLOBALIZATION ON TRANSNATIONAL TERRORISM

Advocates of the negative relationship between economic globalization and terrorism claim that economic globalization removes an important cause of transnational terrorism. The emergence of this argument is quite recent, mostly as an immediate reaction to the terrorist events of September 11. It is not surprising that the argument is not as well developed as those on the positive effect of globalization, nor has it received as much attention from students of transnational terrorism. Economic globalization is argued to reduce transnational terrorist incidents because it facilitates economic development, which in turn removes an incentive to engage in terrorism. A primary cause of transnational terrorism is underdevelopment and poverty, an argument that recently became popular among but was rarely formalized by policy makers and scholars (see, e.g., Biden 2001; Bush 2002; Johnston 2001; Merritt 2001; Rice 2001; Tyson 2001). Poor economic conditions create “terrorist breeding grounds,” where disaffected populations turn to transnational terrorist activities as a solution to their problems. In Marwan Bishara’s words, “When people feel so inferior militarily and economically, they adopt asymmetric means—not the usual means—to


237

get what they want” (Johnston 2001). In addition, poverty, underdevelopment, and instability are often associated with those states either willing to provide safe haven for terrorists or unable to successfully expel terrorists from their borders. Poverty and its accompanying instability in Afghanistan created the conditions that allowed the Taliban to gain power, a situation that in turn led to the provision of sanctuary for Al Qaeda and Osama bin Laden. Consistent with this argument, Bush (2002) claims in a widely cited speech, “We fight against poverty because hope is an answer to terror.” Recently, Krueger and Maleckova (2002) assessed empirically the link between poverty or low education and participation in politically motivated violence and terrorist activities. They show that the occurrences of hate crimes, which resemble terrorism in spirit, are largely unrelated to economic conditions. Then, based on a public opinion poll conducted in the West Bank and Gaza Strip, they show that Palestinians who have higher education or living standards are just as likely to support violence against Israeli targets. Next, they conducted a statistical analysis of participation in Hezbollah in Lebanon and found that education and poverty do not explain whether individuals choose to become martyrs for Hezbollah on suicide missions. Those who have a living standard above the poverty line or a secondary school or higher education are actually more likely to participate in Hezbollah. They also found that Israeli Jews involved in terrorist activities in the early 1980s were well educated and with well-paying jobs. Finally, they show that there is mixed evidence on the effect of real gross domestic product (GDP) growth on the number of terrorist acts each year from 1969 to 1996 in Israel. Krueger and Maleckova conclude that economic conditions and education are largely unrelated to individual participation in and support for terrorism. Although interesting and innovative, the study by Krueger and Maleckova (2002) needs to be put into perspective. First, most of their cases come from a region that is characterized by historical tension, hatred, and military violence. Their conclusion may not generalize across all countries. Second, the positive relationship between a Hezbollah suicide bomber’s education and suicide missions may merely reflect that terrorist leaders use education as a screening device to pick the most competent candidate possible. Without controlling for the job opportunities of the suicide bombers, Krueger and Maleckova are not directly testing their hypothesis.2 Third, the lack of correlation between poor economic conditions and terrorist activities at the individual level may not be inferred to hold at the country level. Better educated people, living under good conditions in the poor countries, are also better informed about conditions in other rich countries than their poor countrymen and hence are more conscious of the comparison between the rich and their own countries. The sense of relative deprivation can provide a strong incentive for them to engage in terrorist activities as the last resort to change the conditions of their own countries. Their individual behaviors can lead one to observe at the aggregate level more terrorist incidents in the poor countries. Hence, we concur with Krueger and Maleckova that whether economic development is related to transnational terrorism at the country level should be assessed in crosscountry analyses. 2. We thank a referee for suggesting this point.

238


For economic globalization to reduce transnational terrorism, globalization has to be able to promote economic development and reduce poverty. Many policy makers have endorsed the positive effect of globalization on development. Canadian Finance Minister Paul Martin (2001) argues that participation in the global economic system greatly enhances a state’s economic development. President Bush (2002) also says, “The vast majority of financing for development comes not from aid, but from trade and domestic capital and foreign investment. . . . So, to be serious about fighting poverty, we must be serious about expanding trade.” U.S. Federal Reserve Board Chairman Alan Greenspan (1997) also claims that “the extraordinary changes in global finance on balance have been beneficial in facilitating significant improvements in economic structures and living standards throughout the world.” Leaders of the seven major industrial democracies assert in the joint communiqué for the 1996 G7 Summit that “economic growth and progress in today’s interdependent world is bound up with the process of globalization” (Lyon G7 Summit 1996). Scholars, however, have relatively more diverse opinions about the effect of economic globalization on development. On one hand, trade was considered to hinder growth in the developing countries in the 1950s. More recently, Rodrik (e.g., 1992, 2001) has raised doubt about the positive effect of trade openness on economic growth for the developing countries. Trade is not sufficient to generate higher growth in these countries, and domestic factors are more important than economic openness. On the other hand, many studies (e.g., Edwards 1993; Frankel and Romer 1999; Stiglitz and Squire 1998) have shown that trade openness does promote economic development. The effects of portfolio and foreign direct investments are also debated in the literature. Although the earlier dependency arguments and the later contagious financial crises in the 1990s led to varying degrees of opposition or reservation concerning capital market integration, there have been equally strong, if not stronger, theoretical arguments and empirical evidence showing that financial and production capital market integration benefits the economic development of countries in general (see, e.g., Graham 2000; Obstfeld 1994; Quinn 1997). Although it is certainly not feasible to expect to settle definitely the controversy over the effect of economic openness on development, it suffices to note that the possible negative effect of economic globalization on transnational terrorism may be achieved by promoting economic development for the purpose of this analysis. The negative effect may be realized through two channels. As a country expands its trade, FDI, and financial capital, its growing integration into the global economy arguably improves not only its own economic conditions but also those of its economic partner countries. Economic development in its own national economy removes an incentive for its citizens to engage in transnational terrorist incidents against foreign targets within the country. In addition, economic progress in its major partner countries reduces the likelihood that their citizens will cross borders to this country to engage in terrorist activities. If economic globalization promotes development, a country’s economic integration affects simultaneously the development of itself and its major economic partners. Although a country’s own development and that of its major partners affect transnational terrorist incidents inside the country, following the same


239

theoretical logic, their effects are distinct empirically and should be separated in analysis. HYPOTHESES

The purpose of this study is to separate empirically the possible positive effect of economic globalization on transnational terrorism from its possible negative effect. Based on the above discussions, the following testable hypotheses are formulated for the empirical analysis below. Hypothesis 1: A country’s economic integration in terms of trade, FDI, and financial capital has a positive effect on the number of transnational terrorist incidents within its borders. Hypothesis 2: Economic development of a country decreases the number of transnational terrorist incidents within the country. Hypothesis 3: Economic development of a country’s economic partner countries reduces the number of transnational terrorist incidents within the country.

RESEARCH DESIGN: DATA, VARIABLES, AND METHOD In this section, we design an empirical analysis that separates and tests the potential positive and negative effects of economic globalization on transnational terrorist incidents. The unit of analysis is the country-year. The estimation sample covers 112 countries from 1975 to 1997. A list of all the countries is in Appendix A. Because the arguments on the effects of economic globalization are expected to apply to comparisons both cross-nationally and over time for individual countries, we employ a pooled time-series, cross-sectional (TSCS) design. DEPENDENT VARIABLE

The dependent variable is the number of transnational terrorist events that occur in a country in a year. Data are collected from the International Terrorism: Attributes of Terrorist Events (ITERATE) data sets from 1968 to 2000 (Mickolus 1982; Mickolus et al. 1989, 1993, 2002). The estimation sample for the statistical model is smaller than the size of the data on the dependent variable because of data limitations on several independent variables. The ITERATE data involve several research design issues. First, the data are compiled from publicly available media sources. Therefore, states with better developed or less government-controlled media may be overreported, whereas those with less developed or more government-controlled media may be underreported. The ITERATE data set, however, partially resolves this bias by drawing from both national and international sources and capturing events that government-controlled national media sources may have failed to report (Sandler 1995). However, it is worth noting that the bias may still lead one to find more incidents associated with more developed or more democratic countries. This implies that one may have reasonably strong confi-

240


dence in results that find that more developed or less government-controlled states experience fewer transnational terrorist incidents. A second design issue involves the method used to differentiate between the starting and ending locations of a terrorist event. Whereas in most cases, the starting and ending locations of a particular event are identical, in roughly 5% of all incidents, the starting and ending locations differ. Differences in starting and ending locations typically result from hijackings, in which a terrorist group hijacks a plane or other vehicle in state A and uses said vehicle to travel to state B. In this analysis, we focus only on those countries in which transnational terrorist incidents start. Because such a small portion of the cases have different starting and ending locations, their effect on the estimation results is likely to be small. Furthermore, the theoretical arguments are more applicable to countries where these incidents first occur because, although a terrorist may rationally choose a starting location based on the above discussed theory, random factors outside of the terrorist’s control may influence where an incident ends. SEPARATING POSITIVE AND NEGATIVE EFFECTS OF ECONOMIC GLOBALIZATION

We use five indicators to separate empirically the positive and negative effects of economic globalization on transnational terrorism. Three indicators—trade, foreign direct investment, and financial capital—capture the positive effect of a country’s economic integration into the global economy, as hypothesis 1 suggests. The remaining two indicators measure a state’s own and its top partners’ economic development. They represent the indirect, negative effect of integration on transnational terrorism via economic development, as hypotheses 2 and 3 suggest. The rationale is that with the negative effect of economic globalization controlled for by economic development variables, the globalization variables themselves reflect the leftover positive effect. Hence, we expect the globalization variables to be positive and significant according to hypothesis 1, whereas we expect the economic development variables to be negative and significant according to hypotheses 2 and 3. The three direct indicators of economic globalization are trade openness, foreign direct investment, and financial capital flows. In the globalization literature, both economists and political scientists have widely used these three indicators as standard measures of economic globalization and national integration into the global economy (see, e.g., Frankel 2000; Bhalla 2002; Garrett 1998; Li and Reuveny 2003). They capture, respectively, three distinct and significant dimensions of economic globalization: the cross-border flows of goods and services, production capital, and financial capital. They also reflect the degree of a country’s capital account and exchange rate restrictions that constrain terrorists’ ability to move money around to finance their operations.3 More specifically, TRADE denotes the annual sum of the value of imports and exports of the goods and services of a country, as a percentage of the country’s GDP. 3. Although these indicators do not measure directly the use of informal economic transfers (such as Islamist charities and labor remittances) that terrorists also use to finance their operations, they should correlate with and hence indirectly capture the level of ease of such informal economic transfers.


241

FDI is the sum of the absolute values of inflows and outflows of the foreign direct investments of a country, as a ratio to the country’s GDP. PORTFOLIO is the sum of a country’s financial flows in equity securities and debt securities, as a percentage of the country’s GDP. Each variable is divided by GDP to reflect its relative importance in the national economy. ECONOMIC DEVELOPMENT denotes the logged yearly GDP per capita in purchasing power parity adjusted to international prices. Data for all four variables are taken from the World Development Indicators (World Bank 2002). PARTNERS’ DEVELOPMENT denotes the logged yearly GDP per capita (PPP) average of the largest eight destination countries of a country’s exports.4 This variable captures the economic conditions of a country’s major economic partners. Because capital flows often follow the flows of commodities and services, and because data on bilateral capital flows are hard to obtain or limited, we use goods flows to proxy the effect of a country’s trade and capital flows on its partner countries. The variable PARTNERS’ DEVELOPMENT indicates how the economic development of a country’s economic partners, which is affected by trade and investments with this country, can remove an incentive for citizens from those partner countries to engage in terrorist activities in this country. The implicit assumption is that a country’s trade and investments are likely to affect its major economic partners the most. Data on PARTNERS’ DEVELOPMENT were computed using data on GDP per capita from the World Development Indicators (World Bank 2002) data set and the bilateral trade data from the International Monetary Fund (IMF) Direction of Trade Database. Because terrorist incidents can affect, in reverse order, trade, FDI, financial capital, and economic development, we control for this possibility by lagging these five variables 1 year behind the dependent variable. Because the hypotheses are directional, we apply one-tailed tests to the coefficients of these five variables. CONTROL VARIABLES

We include the following control variables in the statistical model. SIZE denotes the size of a country, measured as its total population. It is logged to control for skewed distribution and lagged 1 year behind the dependent variable. The size of a country is often positively associated with terrorist incidents. Larger countries tend to be more heterogeneous, where alienated segments of the population may resort to terrorism to influence their governments. In addition, according to Eyerman (1998), a larger country is likely to experience more attacks due to the increased difficulty of successfully policing a larger population. DEMOCRACY denotes the level of democracy in a country from the POLITY IV data set (Marshall and Jaggers 2000). This widely used data set provides a composite indicator of democracy on an annual basis from 1800 to 1999, based on the difference between a democracy index (DEMOC) and an autocracy index (AUTOC). The 10point democracy index (DEMOC) measures its democratic characteristics. The 10point autocracy index (AUTOC) measures the autocratic characteristics of the regime. The composite measure of democracy ranges between –10 (the most autocratic 4. Statistical results are robust whether we look at top 8, 10, or 15 export destination countries.

242


regime) and 10 (the most democratic regime), which is frequently used in many studies on democracy (see, e.g., Londregan and Poole 1996). Because terrorist incidents may affect the level of democracy in a country, the variable is lagged 1 year behind the dependent variable. The literature suggests contradictory theoretical expectations of the effect of democracy (see, e.g., Eubank and Weinberg 1994, 2001; Sandler 1995; Eyerman 1998). On one hand, democracy provides access for citizens to express and resolve their grievances, removing the need to resort to terrorist activities. On the other hand, by guaranteeing citizens’ political rights and civil liberties, democracy allows terrorist groups much greater room to maneuver, lowering the costs and risks for committing terrorism. GOVCAPABILITY denotes the capability or resources a government controls that may be applied to control terrorist activities relative to the other countries of the world. Because terrorist groups seek to achieve their goals with minimum costs, they often avoid countries that are more capable of preventing and controlling terrorist incidents and choose to act in those countries that are relatively less capable (Sandler and Lapan 1988; Sandler 1997). GOVCAPABILITY is measured as an annual composite percentage index of a state’s share of the world’s total population, GDP per capita, GDP per unit of energy, military manpower, and military expenditures. It is computed using data from the World Development Indicators (World Bank 2002) data set and lagged 1 year behind the dependent variable. It is worth noting that this variable captures state military and economic strength in a traditional sense and hence represents merely a proxy for potential resources the government can use to control terrorism. PAST INCIDENTS denotes the number of transnational terrorist events that occurred in a country in the previous year. This variable helps to capture the persistence of terrorist incidents within a country or the persistence of the absence of terrorist incidents in a country. Terrorist organizations tend to operate continuously within a chosen country rather than stop their activities after one event due to cost considerations. A country that has experienced some terrorist incident is likely to experience more of such attacks. According to Midlarsky, Crenshaw, and Yoshida (1980), the activity of one group may encourage the future activities of other groups within that state, thereby producing a greater probability of an increased number of events in the future. In addition, countries that experience long periods without terrorist activities may continue to enjoy an absence of such activity because it can be prohibitively expensive for terrorist groups to organized terrorist activities in new countries. The inclusion of the PAST INCIDENTS variable also helps to control for the effect of potentially relevant but omitted structural variables and to deal with possible serial correlation in the error term. Many studies in political science recommend and adopt this modeling strategy (see, e.g., Beck and Katz 1995; Burkhart and Lewis-Beck 1994; Li and Reuveny 2003). CONFLICT is coded 1 if a state is engaged in interstate military conflict or war and 0 otherwise. This measure captures the possibility that a state’s external conflict engenders enmity and causes foreign terrorists to attack its domestic targets or domestic terrorists to attack foreign targets within the country. Data are from Gleditsch et al. (2002).


243

Transnational terrorist incidents are unevenly distributed geographically. Some regions experience a concentration of terrorist incidents, whereas other regions are relatively tranquil. To control for these systematic regional variations, we construct five regional dummy variables: Europe, Middle East, Africa, Asia, and North and South America. We exclude Middle East from the statistical model as the reference category. The estimate for each of the other four regions represents a comparison between that region and the Middle East. Transnational terrorist incidents exhibit particular unit effects over time. Some years are relatively more tranquil, whereas others are characterized by some extremely large number of terrorist incidents. Such yearly unit effects often need to be controlled for in the pooled design. In addition, Enders and Sandler (1999) also show that global transnational terrorist incidents have declined due to the end of the cold war. Meanwhile, economic globalization and the per capita income have been rising and expanding since 1950 in the global economy (e.g., Frankel 2000; Bhalla 2002). Transnational terrorist incidents and economic globalization may trend in opposite directions in the sample period. We therefore include year dummy variables to control for year-specific effects. Although these variables make it harder to find statistically significant results, they ensure that the statistically significant results are nonspurious. STATISTICAL METHOD

Because the dependent variable is an event count variable, ordinary least squares (OLS) estimates can be inefficient, inconsistent, and biased (Long 1997). The Poisson regression model is often applied to model event counts, in which the mean of the Poisson distribution is conditional on the independent variables. But the Poisson regression model assumes that the conditional mean of the dependent variable equals the conditional variance. This assumption, which is often violated, causes underestimated standard errors and spurious statistical significance. Therefore, we apply the negative binomial regression, which allows the conditional variance of the dependent variable to exceed the conditional mean by adding a dispersion parameter to model the unobserved heterogeneity among observations. Statistical models for the pooled time-series, cross-sectional data often suffer heteroskedasticity and serial correlation in the error term. To deal with these potential problems, we estimate robust standard errors clustered over countries. These estimated standard errors are robust to both heteroskedasticity and to a general type of serial correlation within any cross-sectional unit (Rogers 1993; Williams 2000). The inclusion of the lagged incidents and the year dummies also helps to absorb temporal dependence in the data. FINDINGS AND IMPLICATIONS In this section, we discuss the statistical findings. Table 1 presents results for the benchmark model 1 and its variants for the robustness check. Table 2 includes addi(text continues on p. 248)

244

Europe

Conflict

Past incidents

Government capability

Democracy

Size

Partners’ development

Economic development

Portfolio

FDI

Trade

–0.001 (0.48) 0.019 (0.80) –0.018 (0.97) –0.215 (2.16)** –0.644 (1.93)** 0.341 (6.12)** 0.043 (4.09)** 0.631 (5.10)** 0.020 (4.84)** 0.107 (0.84) –0.240 (1.02)

Model 1

–0.019 (1.09) –0.204 (2.09)** –0.623 (1.93)** 0.340 (6.30)** 0.043 (4.18)** 0.615 (5.06)** 0.020 (4.88)** 0.107 (0.85) –0.240 (1.04)

–0.0002 (0.13)

Model 2

0.016 (0.70) –0.017 (0.94) –0.215 (2.15)** –0.628 (1.90)** 0.350 (6.76)** 0.043 (4.15)** 0.627 (5.13)** 0.020 (4.85)** 0.112 (0.87) –0.243 (1.04)

Model 3

TABLE 1

0.304 (5.35)** 0.035 (3.35)** 0.416 (3.64)** 0.020 (4.83)** 0.114 (0.85) –0.291 (1.26)

–0.001 (0.36) –0.007 (0.25) –0.021 (1.13)

Model 4 –0.001 (0.44) 0.019 (0.82) –0.018 (0.98) –0.236 (2.16)** –0.644 (1.93)** 0.341 (6.12)** 0.043 (4.09)** 0.631 (5.10)** 0.020 (4.84)** 0.107 (0.84) –0.240 (1.02)

Model 5

–0.001 (0.60) 0.024 (1.04) –0.022 (1.27) –0.187 (1.28)* –0.725 (1.90)** 0.347 (5.85)** 0.035 (2.94)** 0.610 (4.35)** 0.020 (4.40)** 0.108 (0.83) –0.359 (1.58)

Model 6

–0.001 (0.36) 0.020 (0.87) –0.018 (1.00) –0.130 (1.06) –0.640 (1.89)** 0.351 (5.87)** 0.044 (4.22)** 0.571 (4.17)** 0.020 (4.85)** 0.111 (0.94) –0.090 (0.33)

Model 7

Effects of Economic Globalization on Transnational Terrorist Incidents within Countries, 1975-1997

–0.002 (1.28)* 0.035 (0.99) –0.050 (3.27)** –0.293 (2.37)** –0.799 (1.94)** 0.352 (4.42)** 0.045 (4.03)** 0.658 (4.33)** 0.020 (4.76)** 0.064 (0.56) –0.062 (0.21)

Model 8

245

2.563 (0.88) 1,996 5.12** 636.5**

–1.051 (3.53)** –0.405 (1.71)* –1.176 (4.37)**

2.324 (0.82) 2,014 5.08** 603.6**

–1.036 (3.59)** –0.390 (1.68)* –1.144 (4.37)**

2.248 (0.81) 2,011 5.10** 637.6**

–1.064 (3.60)** –0.391 (1.68)* –1.169 (4.39)**

–1.051 (3.53)** –0.405 (1.71)* –1.176 (4.37)**

–3.855 2.722 (4.14)** (0.93) 2,011 1,996 5.25** 5.12** 566.5** 636.5**

–0.929 (3.15)** –0.327 (1.43) –0.949 (4.30)**

2.747 (0.79) 1,996 4.98** 912.8**

–1.012 (3.23)** –0.295 (1.15) –1.117 (3.77)** –0.830 (2.00)** 0.411 (0.93) 4.134 (1.99)** 0.020 (1.28)* 0.881 (0.26) 1,992 5.11** 610.1**

–0.889 (2.75)** –0.414 (1.76)* –1.214 (4.66)**

0.022 (1.70)** 3.698 (1.01) 1,232 5.93** 848.2**

–1.034 (3.84)** –0.595 (2.68)** –1.367 (5.49)**

NOTE: Robust z statistics, adjusted over countries, in parentheses. Coefficients for year dummy variables not shown. FDI = foreign direct investment; OECD = Organization of Economic Cooperation and Development. *Significant at the 10% level. **Significant at the 5% level.

Observations Dispersion = 1 2 Wald test (χ )

Constant

Income inequality

OECD

OECD × Partners’ Development

OECD × Economic Development

Africa

Americas

Asia

246

Asia

Europe

Conflict

Government capability

Democracy

Size

Partners’ development

Economic development

Portfolio

FDI

Trade

–0.001 (0.38) –0.007 (0.27) –0.006 (0.35) –0.128 (1.30)* –0.391 (2.29)** 0.357 (5.84)** 0.041 (3.89)** 0.575 (4.19)** 0.203 (1.52) –0.223 (0.88) –1.083 (3.54)**

Model 1

–0.005 (0.32) –0.137 (1.40)* –0.392 (2.32)** 0.356 (5.95)** 0.041 (3.96)** 0.579 (4.27)** 0.198 (1.50) –0.234 (0.94) –1.098 (3.66)**

–0.001 (0.46)

Model 2

–0.009 (0.37) –0.006 (0.31) –0.128 (1.30)* –0.395 (2.32)** 0.364 (6.46)** 0.041 (3.93)** 0.576 (4.20)** 0.209 (1.56) –0.225 (0.90) –1.093 (3.61)**

Model 3

TABLE 2

0.303 (5.36)** 0.028 (2.65)** 0.630 (5.91)** 0.208 (1.53) –0.171 (0.67) –0.849 (2.85)**

–0.001 (0.38) –0.039 (1.33) –0.011 (0.66)

Model 4 –0.001 (0.36) –0.007 (0.27) –0.006 (0.34) –0.138 (1.29)* –0.390 (2.28)** 0.356 (5.84)** 0.041 (3.88)** 0.573 (4.19)** 0.203 (1.52) –0.223 (0.89) –1.083 (3.54)**

Model 5 –0.001 (0.51) –0.004 (0.14) –0.011 (0.59) –0.060 (0.45) –0.313 (1.54)* 0.358 (5.63)** 0.036 (3.38)** 0.529 (3.81)** 0.220 (1.66)* –0.292 (1.16) –1.042 (3.28)**

Model 6

–0.001 (0.32) –0.005 (0.19) –0.006 (0.35) –0.069 (0.59) –0.408 (2.35)** 0.362 (5.63)** 0.041 (3.96)** 0.544 (3.74)** 0.202 (1.62) –0.110 (0.36) –0.961 (2.74)**

Model 7

–0.002 (0.91) –0.005 (0.15) –0.036 (2.32)** –0.181 (1.42)* –0.394 (1.83)** 0.355 (4.01)** 0.040 (3.38)** 0.619 (3.43)** 0.147 (1.25) –0.117 (0.35) –1.073 (3.56)**

Model 8

Effects of Economic Globalization on Transnational Terrorist Incidents within Countries, 1975-1997: Alternative Specifications

247

–0.103 (0.07) 1,996 6.52** 168.9**

–0.365 (1.51) –1.039 (3.91)**

–0.006 (0.00) 2,014 6.46** 166.9**

–0.379 (1.60) –1.051 (3.99)**

–0.229 (0.17) 2,011 6.50** 171.9**

–0.354 (1.48) –1.034 (3.91)**

–0.365 (1.51) –1.038 (3.92)**

–4.067 –0.022 (4.13)** (0.02) 2,011 1,996 6.85** 6.52** 176.0** 169.3**

–0.193 (0.80) –0.764 (3.21)**

–1.426 (0.79) 1,996 6.41** 202.0**

–0.290 (1.16) –0.945 (3.44)** –0.700 (1.30)* 0.332 (0.51) 3.499 (1.47)* 0.014 (0.88) –1.174 (0.57) 1,992 6.54** 166.2**

–0.374 (1.56) –1.066 (4.07)**

0.018 (1.27) –0.148 (0.07) 1,232 8.07** 219.0**

–0.552 (2.35)** –1.239 (4.85)**

NOTE: Robust z statistics, adjusted over countries, in parentheses. FDI = foreign direct investment; OECD = Organization of Economic Cooperation and Development. *Significant at the 10% level. **Significant at the 5% level.

Observations Dispersion = 1 2 Wald test (χ )

Constant

Income inequality

OECD

OECD × Partners’ Development

OECD × Economic Development

Africa

Americas

248


tional robustness tests based on alternative model specifications. Below, we first focus on the results of the benchmark model and then discuss the sensitivity analysis. In model 1, the test of the dispersion parameter indicates that the conditional variance of the dependent variable exceeds its conditional mean at a statistically significant level. It justifies the use of the negative binomial model over the Poisson regression. The Wald test of the model fit is statistically significant at the 1% level as well. Now we turn to the key independent variables in model 1. The globalization variables do not have the positive effect on transnational terrorist incidents expected by hypothesis 1. Trade openness and portfolio investments are both negative and FDI is positive, but none of the three coefficients is statistically different from 0. One may argue that the lack of statistical significance may have resulted from collinearity among the globalization variables. To assuage this concern, we test the joint hypothesis that all three variables have no effect simultaneously. The chi-square statistic of 1.6 is far from rejecting the hypothesis. In addition, the correlation matrix in Appendix B does not show unreasonably high correlation among these variables (highest correlation = 0.48 between FDI and trade). Therefore, the lack of significance does not result from high collinearity. Rising integration of a country into the world’s goods and capital markets is not associated with an increasing number of transnational terrorist incidents within the country. The finding offers no support for the argument that greater economic integration of a country causes more transnational terrorist incidents within its borders. The indirect, negative effects of the globalization variables, as hypothesized in hypotheses 2 and 3, receive support in model 1. A country’s GDP per capita, measuring its own level of development, is statistically significant and negative, consistent with hypothesis 2. As a country becomes more developed, the number of transnational terrorist incidents decreases within its borders. In addition, the average level of development of the country’s top export destination countries, measured by their average GDP per capita, is also statistically significant and negative. Consistent with hypothesis 3, as the level of development in a country’s economic partner countries improves, the number of transnational terrorist incidents decreases within the country. This implies that growing economic integration between the country and its economic partners can help to remove some incentive for foreign terrorists from those partner countries to attack targets within this country. Based on the coefficients of model 1, a 1% increase in the GDP per capita of a country decreases the expected number of transnational terrorist incidents within the country by 19.3%, holding all other variables constant. A 1% increase in the average GDP per capita of the country’s top eight export destination countries decreases the expected number of transnational terrorist incidents within this country by 47.5%, holding other variables constant. Although the globalization variables have no direct positive effect, their indirect negative effect through influencing economic development appears large and important. We also find interesting results for the control variables. The size of a country is positively associated with the number of transnational terrorist incidents inside the country. As expected, larger countries are exposed to greater risks of transnational terrorist incidents than smaller countries. A 1% increase in a country’s population size is


249

associated with a 40.6% increase in the expected number of transnational terrorist incidents within the country. The level of democracy variable is positive and statistically significant. As the democracy score of a country increases by 1 point, its expected number of transnational terrorist incidents increases by 4.4%, which is not a very substantial increase. The evidence is consistent with the argument that democracy, by ensuring the political rights of its citizens, gives terrorist groups more freedom in association, reducing the costs of their engaging in terrorist activities. Eubank and Weinberg (1994, 2001) also find that democracy is positively associated with terrorism. There exists, however, the possibility that the positive result of democracy may have been amplified by the ITERATE data’s reliance on open media sources for data collection on the dependent variable (see, e.g., Sandler 1995). Maybe incidents in democratic countries are more often reported in the media as a result of the fewer restrictions that democratic countries place on the media. Because of these limitations, we should interpret the positive effect of democracy with caution. More future research is in order. The government capability variable is statistically significant but positive. The result implies that although more capable governments may have more resources to control terrorist incidents, they also constitute more effective and valuable targets for terrorist groups. The ability to engage in terrorist activities against more capable governments gives a terrorist group more media coverage and wider influence. Although terrorist groups may have to pay higher costs to act against more capable governments, the returns can be expected to be much higher as well. The past incidents variable is statistically significant and positive, consistent with the expectation that both the presence and the absence of terrorist activities tend to be persistent. Terrorist groups, once operational organizationally, tend to engage in activities continuously. Similarly, terrorist groups may experience greater difficulties in setting up operations in countries that rarely experience terrorist incidents. The interstate military conflict variable is positive as expected, but it is not statistically significant. Results for the regional variables suggest certain patterns in the geographical distribution of terrorist incidents. With the Middle East serving as the reference group, Europe, Asia, America, and Africa all have a negative sign, and they are all statistically significant except for Europe. The Middle East appears to have the highest concentration of transnational terrorist incidents, with Europe ranking in second place. The other regions experience fewer transnational terrorist incidents than the Middle East and Europe. Asia, America, and Africa experience 65%, 33.3%, and 69.1% fewer incidents, respectively, than the Middle East.5 Are these results of model 1 robust to alternative model specifications or other variations? Models 2, 3, 4, 5, 6, and 7 present results from variants of model 1. Models 2, 3, and 4 investigate further the possible influence of collinear variables by excluding 5. The effect of the end of the cold war is subsumed by the year dummy variables. The dummy variables for the years 1991 to 1997 are not statistically significant. In addition, a dummy variable representing the end of the Cold War is statistically significant in a model excluding the year dummy variables. These results are consistent with Enders and Sandler’s (1999) finding that the end of the cold war has led to a decline in transnational terrorist incidents worldwide.

250


FDI, trade openness, and economic development variables sequentially. Statistical inferences for the globalization and development variables, whenever present in the models, remain consistent with those for model 1. Trade, FDI, or portfolio investment remains insignificant even when we take out the correlated trade or FDI. The development variables remain significant and negative. One might argue that economic development of a country is endogenous to the globalization variables and that the results in model 1 may be sensitive to this endogeneity bias. Model 5 assesses this possibility, with the country’s own economic development variable endogenized. We first estimate a two-way, fixed-effects model of economic development by regressing the logged GDP per capita on trade, FDI, portfolio investment, democracy, lagged dependent variable, and country and year dummies, with robust standard errors estimated. Then we obtain the predicted value of the GDP per capita and substitute it into model 1. Results for the first-stage development equation are presented in Appendix C. Trade openness and FDI have a positive and statistically significant effect on GDP per capita, whereas portfolio investments are not statistically significant. Results of the second-stage estimation, included in model 5, show that statistical inferences remain unchanged for the globalization and economic development variables. Results of model 1 are not sensitive to the endogeneity of development. One may also argue that the effect of economic development in the developed countries differs from that in the developing countries. Maybe the negative effect only exists in the developed countries, but not in the developing world. In model 1, economic development variables may have just captured such a developed country-less developed country (DC-LDC) distinction. In addition, factions that are against modernity and prone to attack foreign targets often operate in the less developed countries, which also implies that the negative effect of economic development may not apply to the developing countries. To assess the sensitivity of model 1 to this possibility, we construct an interactive model in which we include into model 1 three additional variables: (1) an Organization of Economic Cooperation and Development (OECD) dummy, which equals 1 if a country belongs to the OECD and 0 otherwise; (2) an interaction term between the OECD dummy and economic development; and (3) an interaction term between the OECD dummy and partners’ development. Economic development and partners’ development as main effect variables capture the effect of economic development for the non-OECD countries only. The interaction terms measure the additional effect of being an OECD country. Results in model 6 show that after controlling for development in the OECD countries, a non-OECD country’s economic development and its top trading partners’ development still reduce the number of transnational terrorist incidents inside the country. The coefficients do not differ substantially in size from those in model 1. In addition, the interaction term between the OECD dummy and economic development is statistically significant and negative. This suggests that an OECD country’s economic development has a greater negative effect on the number of terrorist incidents than another non-OECD country’s development. The interaction term between the OECD dummy and partners’ development is statistically insignificant. This indicates that the effect of partners’ development of an OECD country does not differ from that


251

of partners’ development of a non-OECD country. Hence, we conclude that the negative effect of economic development on the number of transnational terrorist incidents exists in both developed and developing countries. Finally, economic globalization is sometimes argued to raise income inequality in countries, causing an increase in transnational terrorism. Whereas economic globalization encourages development, the benefits and costs of globalization are often asymmetrically distributed, generating winners and losers and widening the gap between the rich and the poor in societies. Groups marginalized by economic globalization may resort to more terrorist activities. To assess the sensitivity of model 1 to this possibility, we include the Gini coefficient as a measure of national income inequality. The Gini coefficient ranges from 0 to 100, with larger values indicating greater inequality. We use the high-quality data on Gini from Deininger and Squire (1996), one of the most comprehensive data sources on income inequality. Because Gini is computed based on national income surveys, which are not conducted every year for every country, the Gini data contain many missing values. We fill up the missing values using two alternative strategies. For model 7 in Table 1, the missing values are filled with predictions from estimating Gini as a function of GDP per capita, GDP per capita squared, and regional dummies, following Feng and Zak (1999) and Li and Reuveny (2003). For model 8, the missing value of Gini is filled with the previous available value in a country, assuming income inequality remains stable for a period of time. In model 7, results for trade, FDI, and portfolio investment remain insignificant as before, even when we control for income inequality. Partners’ development still retains the same negative, significant effect. Although economic development still has the negative sign, it is no longer significant. This result is to be expected because missing values of Gini are filled based on GDP per capita, and these two variables are highly correlated (–0.74) in the estimation sample. High collinearity causes economic development to become statistically insignificant. Not surprisingly, Gini is positive as expected and significant for the one-tailed test. The size of the effect is not very substantial. As income inequality rises by 1 point along the 100-point scale, the expected number of incidents in a country increases by about 2%. In model 8, based on an alternative Gini measure, economic development and partners’ development remain negative and significant as in model 1. FDI is still positive and insignificant. Although trade and portfolio investment remain negative, trade becomes marginally significant and portfolio investment is highly significant. Gini is statistically significant and positive, encouraging transnational terrorism. The size of the effect is about the same as in model 7. Based on this model, economic globalization appears to increase terrorism mainly by raising income inequality. In addition, trade and portfolio investment reduce transnational terrorism even further, once we control for income inequality and development. Although the finding is interesting, we should interpret it with caution, given the crude nature of the Gini data. The issue warrants future investigation. Overall, the results of model 1 are, broadly speaking, robust even when we control for income inequality. In Table 1, we have followed the general-to-specific approach to model specification and used rather extensive statistical controls, particularly the lagged dependent

252


variable and the year dummies. This approach helps to prevent spurious findings from an underspecified model (see, e.g., Dougherty 2002, 362). In addition, as noted, the inclusion of the lagged dependent variable and the year dummies is motivated theoretically and empirically. One may argue plausibly, however, that the lagged dependent variable and the year dummies generate high collinearity and inefficiency, competing for explanatory power particularly against the globalization variables. The time frame from 1975 to 1997 is not long enough to justify including the lagged dependent variable. The results on the globalization and development variables may be an artifact of including these “irrelevant” variables. To examine this possibility, we estimate all models in Table 1 again, excluding the lagged dependent variable and the year dummies. The results are presented in Table 2. Statistical inferences from this sensitivity analysis are almost identical to those in Table 1. Trade, FDI, and portfolio investment remain statistically insignificant, except for portfolio investment in model 8. In general, as in Table 1, trade, FDI, and portfolio investment are not found to increase transnational terrorism. A country’s economic development and its major economic partners’ development are still found to reduce the number of transnational terrorist incidents within the country. As an exception, the non-OECD country’s economic development in model 6 of Table 2 is still negative but becomes statistically insignificant now. One may also note that the coefficients for the key independent variables—trade, FDI, portfolio investment, economic development, and partners’ development—have decreased in size almost by half. Whereas the lagged dependent variable and the year dummies may have generated collinearity and inefficiency for the models in Table 1, the results in Table 2 suggest that their inclusion is important for capturing relevant confounding forces. More important, the effects of trade, FDI, portfolio investment, economic development, and partners’ development are generally quite robust to alternative model specifications. LOCATION OF INCIDENT AND TARGET CHOICE OF ATTACK

In this analysis, we have focused on the effect of economic globalization on the venue of the terrorist attack. We do not analyze its effect on the target choice by terrorists. The ITERATE database, on which we rely for terrorist incident data, is the most comprehensive database on transnational terrorist incidents. The database does not have data on the targets of terrorist attacks because the motives of the terrorists are often not known. The database contains a variable that identifies the victim’s nationality. Two issues thwart our effort of using the victim variable to capture the target of an attack. First, the victim may be only collateral damage, not really the target. Second, the victim nationality variable in ITERATE does not list exhaustively all nationalities involved in each incident but rather three nationalities at most. Hence, targets of terrorist attacks are not observed in the ITERATE database. The key issue for our analysis is to what extent studying the venue of a terrorist attack does and does not inform us about the attack target choice, if such choice is solely what one is concerned about. The relationship between the location and the target of a terrorist attack involves four substantively meaningful scenarios: (1) country


253

A’s citizens are targeted and attacked at country A by foreign terrorists, (2) country A’s citizens are not targeted but attacked at country A, (3) country A’s citizens are targeted but attacked at country B, and (4) country A’s citizens are not targeted but attacked at country B. The conceptualization suggests that the location and the target provide different information regarding transnational terrorism. First, focusing on the attack location sheds light on the causes of scenarios (1) and (2), in which a country’s citizens are attacked domestically either as the target or merely as collateral damage. It does not explain scenarios (3) and (4), that is, why a country’s citizens are attacked abroad. Second, focusing on targeting, assuming data on targeting were available, explains scenarios (1) and (3), in which a country’s citizens are targeted and attacked at home or abroad. But it does not account for scenarios (2) and (4), in which a country’s citizens become collateral damage in terrorist incidents at home or abroad. Within the context of our analysis, such a conceptualization has two important implications. First, the effect of economic globalization is relevant to both scenarios (1) and (2). Although rising economic integration may render a country’s citizens targets of terrorist attacks based on scenario (1), it also reduces the costs of conducting terrorist attacks within the country, making it a desirable attack venue and causing its citizens to become collateral damage, based on scenario (2) (even though citizens of the country are not really the target of attacks). Apparently, focusing on the attack venue captures both scenarios (1) and (2), whereas focusing on targeting alone misses scenario (2), an important implication of economic globalization for terrorist activities. Second, economic globalization also affects scenario (3), in which rising integration of country A into the global economy may generate animosity, causing country A’s citizens to be targeted and attacked abroad rather than at home. Focusing on the venue of the terrorist attack does not explain to what extent economic globalization has this effect. Hence, findings of this analysis apply specifically to the question of how economic globalization affects transnational terrorist incidents within countries. The findings do not explain whether economic globalization causes a country’s citizens to be targeted and attacked abroad. This limitation of the analysis is caused by the lack of data on targeting choice and should be dealt with in future research. CONCLUSION After the September 11 terrorist attacks, the association of economic globalization and transnational terrorist activities has attracted wide attention from policy makers, journalists, and academics. Many believe that economic globalization contributes to transnational terrorist attacks, whereas a few others posit that economic globalization helps to prevent terrorist incidents. Despite heated public discourse, there has been no systematic, rigorous analysis of the relationship between economic globalization and transnational terrorist incidents. In this study, we analyze statistically the effect of economic globalization on the number of transnational terrorist incidents within countries.

254


Based on a sample of 112 countries from 1975 to 1997, we find interesting patterns of statistical association between economic globalization and transnational terrorist incidents. In general, trade, FDI, and portfolio investment of a country do not directly increase the number of transnational terrorist incidents inside its borders. Economic development of the country and its top trading partners reduces the number of terrorist incidents inside the country. To the extent that economic globalization promotes development, globalization can have an indirect negative effect on transnational terrorism. We should note that our analysis has focused on the linkages between economic globalization and terrorism as a first cut. We raise caution about generalizing the findings to other aspects of the globalization phenomenon. Future research should also look at the effects of immigration, cultural globalization, and the role of transnational nongovernmental organizations if we want to understand fully the effect of globalization on transnational terrorism. Our analysis is informative in terms of how a country’s economic integration affects the number of terrorist incidents within the country. Our research findings should, therefore, be interpreted accordingly. Future research could also investigate how economic integration affects the target choice of terrorists. Despite the caveats, our analysis suggests important policy implications for the war against terrorism. National governments should realize that economic globalization is not the cause of, but a possible partial solution to, transnational terrorism. Although opening up one’s border facilitates the movement of terrorists and their activities, our results show that the effect of such facilitation appears weak. It does not precipitate a significant rise in transnational terrorist attacks within countries. This is an important lesson for policy makers who are designing antiterrorism policies. More important, economic openness, to the extent that it promotes economic development, may actually help to reduce indirectly the number of transnational terrorist incidents inside a country. Closing borders to foreign goods and capital may produce undesirable effects. Economic closure and autarky can generate more incentives to engage in transnational terrorist activities by hindering economic development. Antiterrorism policy measures should be designed with caution. They should not be designed to slow down economic globalization. Promoting economic development and reducing poverty should be important components of the global war against terrorism. Such effects are structural and systemwide. It is in the best interest of the United States not only to develop by itself but also to help other countries to grow quickly. The effect of economic development on the number of transnational terrorist incidents is large. The role of economic development deserves much more attention from policy makers than it currently enjoys.


APPENDIX A List of Countries in the Estimation Sample Albania Algeria Angola Argentina Armenia Australia Austria Bahrain Belarus Bolivia Botswana Brazil Bulgaria Burkina Faso Burundi Cameroon Canada Central African Republic Chad Chile China Colombia Congo, Republic of Costa Rica Côte d’Ivoire Cyprus Denmark Dominican Republic Ecuador Egypt, Arab Republic of El Salvador Estonia Ethiopia Fiji France Gabon Germany Ghana

Greece Guatemala Guyana Haiti Honduras Hungary Iceland India Indonesia Iran, Islamic Republic of Ireland Israel Italy Jamaica Japan Jordan Kenya Korea, Republic of Kuwait Lao Peoples’ Democratic Republic Latvia Lesotho Lithuania Malawi Malaysia Mali Mauritania Mexico Morocco Mozambique Namibia Nepal The Netherlands New Zealand Nicaragua Niger Nigeria

Norway Pakistan Panama Papua New Guinea Paraguay Peru Philippines Poland Portugal Romania Russian Federation Rwanda Saudi Arabia Senegal Sierra Leone Singapore Slovenia South Africa Spain Sri Lanka Swaziland Sweden Switzerland Syrian Arab Republic Tanzania Thailand Togo Trinidad and Tobago Tunisia Turkey Ukraine United Kingdom United States Uruguay Venezuela, RB Zambia Zimbabwe

255

256


APPENDIX B In-Sample Correlation Matrix of Key Independent Variables

Trade Trade Foreign direct investment Portfolio Economic development Partners’ development

1.00 0.48 –0.11 0.11 –0.03

Foreign Direct Investment

Portfolio

Economic Development

1.00 0.03 0.06

1.00 0.33

1.00 –0.10 0.23 0.15

Partners’ Development

1.00

APPENDIX C Effects of Economic Globalization on Development Two-Way Fixed-Effects Model Trade Foreign direct investment Portfolio Democracy Economic developmentt – 1 Constant Observations 2 R

0.0003** (2.13) 0.0022*** (4.19) –0.0003 (0.50) –0.0002 (0.61) 0.9118*** (95.86) 0.7715*** (9.60) 2,430 0.998

NOTE: Robust z statistics, adjusted over countries, in parentheses. Coefficients for country and year dummies not shown. **Significant at the 5% level. ***Significant at the 1% level.

REFERENCES Beck, Nathaniel, and Jonathan N. Katz. 1995. What to do (and not to do) with time-series cross-section data. American Political Science Review 89 (3): 634-47. Bhalla, Surjit S. 2002. Imagine there’s no country: Poverty, inequality, and growth in the era of globalization. Washington, DC: Institute for International Economics. Biden, Joseph R. 2001. Toward economic growth in Africa: The fifth action in the war on terrorism. Retrieved February 5, 2003, from http://biden.senate.gov/press/speeches/economicgrowthinafrica.htm Burkhart, Ross, and Michael Lewis-Beck. 1994. Comparative democracy: The economic development thesis. American Political Science Review 88 (4): 903-10. Bush, George W. 2002. President outlines plan to help world’s poor. Retrieved February 2, 2003, from http:// www.whitehouse.gov/news/releases/2002/03/print/ 20020322-1.html Crenshaw, Martha. 1981. The causes of terrorism. Comparative Politics 13:379-99. Deininger, Klaus, and Lyn Squire. 1996. A new dataset measuring income inequality. World Bank Economic Review 10:565-91. Dougherty, Christopher. 2002. Introduction to econometrics. New York: Oxford University Press. Edwards, Sebastian. 1993. Openness, trade liberalization, and growth in developing countries. Journal of Economic Literature 31:1358-93.


257

Enders, Walter, and Todd Sandler. 1993. The effectiveness of antiterrorism policies: A vectorautoregression-intervention analysis. American Political Science Review 87:829-44. . 1999. Transnational terrorism in the post–cold war era. International Studies Quarterly 43:145-67. . 2000. Is transnational terrorism becoming more threatening? A time-series investigation. Journal of Conflict Resolution 44:307-32. Eubank, William, and Leonard Weinberg. 1994. Does democracy encourage terrorism? Terrorism and Political Violence 6:417-43. . 2001. Terrorism and democracy: Perpetrators and victims. Terrorism and Political Violence 13:155-64. Eyerman, Joe. 1998. Terrorism and democratic states: Soft targets or accessible systems. International Interactions 24:151-70. Feng, Yi, and Paul J. Zak. 1999. The determinants of democratic transitions. Journal of Conflict Resolution 43:162-77. Frankel, Jeffrey. 2000. Globalization of the economy. In Governance in a globalizing world, edited by Joseph Nye and John Donahue. Washington, DC: Brookings Institution. Frankel, Jeffrey, and David Romer. 1999. Does trade cause growth? American Economic Review 89:379-99. Garrett, Geoffrey. 1998. Global markets and national politics: Collision course or virtuous circle? International Organization 52:149-76. Gleditsch, Nils Petter, Peter Wallensteen, Mikael Eriksson, Margareta Sollenberg, and Håvard Strand. 2002. Armed conflict 1946-2001: A new dataset. Journal of Peace Research 39:615-37. Graham, Edward. 2000. Fighting the wrong enemy: Antiglobal activists and multinational enterprises. Washington, DC: Institute for International Economics. Greenspan, Alan. 1997. The globalization of finance. Retrieved from cato.org/pubs/ journal/cj17n3-1.html Gurr, Ted Robert. 1979. Some characteristics of the politics of terrorism. In The politics of terrorism, edited by Michael Stohl. New York: Marcel Dekker. Hoffman, Bruce. 1997. The confluence of domestic and international trends in terrorism. Terrorism and Political Violence 9:1-15. . 1998. Inside terrorism. New York: Columbia University Press. Is it at risk?—Globalization. 2002. The Economist, February 2. Retrieved February 5, 2003, from http:// web.lexis-nexis.com/universe/ Jenkins, Brian, Janera Johnson, and David Ronfeldt. 1977. Numbered lives: Some statistical observations from 77 international hostage episodes. Santa Monica, CA: RAND. Johnston, A. 2001. Disparities in wealth are seen as fuel for terrorism. International Herald Tribune, December 19. Kobrin, Steven. 1997. Electronic cash and the end of national markets. Foreign Policy 107:65-77. Krueger, Alan B., and Jitka Maleckova. 2002. Education, poverty, political violence and terrorism: Is there a causal connection? NBER Working Paper No. 9074, National Bureau of Economic Research, Cambridge, MA. Li, Quan, and Rafael Reuveny. 2003. Economic globalization and democracy: An empirical analysis. British Journal of Political Science 33:29-54. Londregan, John, and Keith Poole. 1996. Does high income promote democracy? World Politics 49:1-30. Long, J. Scott. 1997. Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage. Lyon G7 Summit. 1996. Economic communiqué. Retrieved from library.utoronto.ca/www/g77/ 96ecopre.html Marshall, Monty G., and Keith Jaggers. 2000. POLITY IV project: Political regime characteristics and transitions, 1800-2000 dataset users manual. Retrieved February 5, 2003, from http://www.bsos.umd.edu/ cidcm/inscr/polity/index.htm#data Martin, Paul. 2001. Globalization, terrorism and the world economy: Speech by the Honourable Paul Martin, Finance Minister of Canada, at a luncheon organized by the Reinventing Bretton Woods Committee and the Conference Board of Canada. Retrieved February 3, 2003, from http://www.fin.gc.ca news01/ 01-105e.html Matthew, Richard, and George Shambaugh. 1998. Sex, drugs, and heavy metal: Transnational threats and national vulnerabilities. Security Dialogue 29:163-75.

258


McDonald, Michael. 2002. Checkpoint terror: Border searches snarl the free flow of goods. US News and World Report, February 11. Merritt, Giles. 2001. Wake up to the perilous cost of the wealth gap. International Herald Tribune, October 30. Mickolus, Edward F. 1979. Transnational terrorism. In The politics of terrorism, edited by Michael Stohl. New York: Marcel Dekker. . 1982. International terrorism: Attributes of terrorist events, 1968-1977 (ITERATE 2). Ann Arbor, MI: Inter-University Consortium for Political and Social Research. Mickolus, Edward F., Todd Sandler, Jean M. Murdock, and Peter Fleming. 1989. International terrorism: Attributes of terrorist events, 1978-1987 (ITERATE 3). Dunn Loring, VA: Vinyard Software. . 1993. International terrorism: Attributes of terrorist events, 1988-1991 (ITERATE 4). Dunn Loring, VA: Vinyard Software. . 2002. International terrorism: Attributes of terrorist events, 1991-2000 (ITERATE 5). Dunn Loring, VA: Vinyard Software. Midlarsky, Manus I., Martha Crenshaw, and Fumihiko Yoshida. 1980. Why violence spreads: The contagion of international terrorism. International Studies Quarterly 24:262-98. Naím, Moisés. 2000. Missing links: The digital drain. Foreign Policy 120:119-20. Obstfeld, Maurice. 1994. Risk-taking, diversification, and growth. American Economic Review 84:1310-29. Phillips, Michael M. 2002. US-European divisions hinder drive to block terrorists’ assets. The Wall Street Journal, April 11. Quinn, Dennis. 1997. The correlates of change in international financial regulation. American Political Science Review 91:531-51. Rice, Susan E. 2001. The Africa battle. The Washington Post, December 11. Rodrik, Dani. 1992. The limits of trade policy reform in developing countries. Journal of Economic Perspective 6:87-105. . 2001. Trading in illusions. Foreign Policy 123:55-62. Rogers, William H. 1993. Regression standard errors in clustered samples. Stata Technical Bulletin 13:19-23. Sandler, Todd. 1995. On the relationship between democracy and terrorism. Terrorism and Political Violence 12 (2): 97-122. . 1997. The future challenges of NATO: An economic viewpoint. Defense and Peace Economics 8:319-53. Sandler, Todd, Walter Enders, and Harvey E. Lapan. 1991. Economic analysis can help fight international terrorism. Challenge 34:10-17. Sandler, Todd, and Harvey E. Lapan. 1988. The calculus of dissent: An analysis of terrorist’choice of target. Synthese 76:245-61. Sandler, Todd, John T. Tschirhart, and Jon Cauley. 1983. A theoretical analysis of transnational terrorism. American Political Science Review 77:36-54. Shahar, Yuval. 2001. Tracing bin Laden’s money: Easier said than done. International Policy Institute for Counter-Terrorism. Retrieved February 3, 2003, from http://www.ict.org.il/ Stiglitz, Joseph, and Lyn Squire. 1998. International development: Is it possible? Foreign Policy 110:138-51. Stohl, Michael. 1979. Introduction: Myths and realities of political terrorism. In The politics of terrorism, edited by Michael Stohl. New York: Marcel Dekker. Strange, Susan. 1998. Mad money: When markets outgrow governments. Ann Arbor: University of Michigan Press. The Trojan box. 2002. The Economist, February 9. Retrieved February 5, 2003, from http://web.lexisnexis.com/universe/ Tyson, Laura. 2001. It’s time to step up the global war on poverty. Business Week, December 3. Weintraub, Sidney. 2002. Disrupting the financing of terrorism. The Washington Quarterly 25:53-60. Wilkinson, Paul. 1986. Terrorism and the liberal state. Washington Square: New York University Press. Williams, Rick L. 2000. A note on robust variance estimation for cluster-correlated data. Biometrics 56:645-46. World Bank. 2002. The 2002 world development indicators CD-ROM. Washington, DC: World Bank.