Critical Criminology 10: 57–72, 2001. © 2001 Kluwer Academic Publishers. Printed in the Netherlands. 57 CRIME AND CRIME CONTROL IN AN AGE OF GLOBALIZATION: A THEORETICAL DISSECTION GREGG BARAK Eastern Michigan University Accepted 10 July 2001 Abstract. This article examines the impact of globalization on both crime and crime control at the national and global levels. To make conceptual sense out of the transforming nature of these activities at the turn of the 21st century, a threefold analysis is presented: (1) an overview of the three traditional developmental models of crime and crime control – modernization, world system, and opportunity; (2) a characterization of crime and crime control in relationship to the more recently emerging models of globalization; and (3) a discussion of the implications of the dialectical relations between the models of development and the models of globalization. Assessments of the models and other provisional conclusions are drawn based on a survey of both crime and crime control in 15 developed, developing, and post-traditional nation-states. Introduction Globalization refers to the process of growing interdependency among events, people, and governments around the world that are increasingly connected through a worldwide political economy and an expanding communications, transportation, and computer network. In today’s world of laissez-faire globalization, free trade and unregulated markets rule supreme. Largely under the aegis of the United States, these forces and relations of production have come to represent a hegemonic regime over the prevailing economic and political spheres of nation-states around the globe. As characterized by Weiss (2000: 1): The Soviet menace has been vanquished. China has turned away from Communism as an economic doctrine. Third World resistance groups are in disarray and retreat, western European social democracies (with the bold exception of Denmark and The Netherlands) are shrinking their welfare states, and governments in Latin America, the former Soviet bloc countries, and much of Asia are busy deregulating, downsizing, privatizing, contracting out, reducing taxes, and cutting social spending. On the cultural front, people throughout the world are emulating the voracious consumerism of Americans. The ascendance of capitalist 58 GREGG BARAK values and social standards among former peasants, apparatchiks, and state industrial workers is helping to finish off the old social orders in China, Eastern Europe, and Latin America. With globalization has come the polarization of wealth and income, as well as the intensification of disease, poverty, and hunger – both within and between rich and poor countries. Developmental studies of the United Nations reveal that the upper fifth of those living in high-income countries account for 86% of all the world’s private expenditures on consumption. At the same time, tens of millions of people succumb annually to famine and preventable diseases. For hundreds of millions of others, mostly in developing but also in developed countries, “life is a daily preoccupation with obtaining safe water, rudimentary health care, basic education, and sufficient nutrition” (Weiss 2000: 2). From the perspectives of both crime and crime control, critical criminologists at the turn of the century are trying to understand the comparative effects that development, globalization, and increasing inequalities are having on the phenomena that we study. In an effort to shed some light on this understanding, this discussion provides a conceptual overview of the theories of modernization, opportunity, and world system. The article assesses their strengths, weaknesses, and possibilities for integration. It develops an understanding of globalization, which has important implications for both transnational crime and as context for local factors that significantly shape crime and crime control policies. This article draws on a recent global survey of 15 nation-states (Barak 2000) to frame and contextualize important issues, as well as draw some provisional conclusions about crime and crime control in an age of globalization. For each of these nation-states, the researchers agreed to present (1) quantitative data on the trends and rates of victimization; (2) qualitative data on the historical evolution of crime and social control, incorporating a profile of the changing cultural styles of living and consuming; (3) a discussion on the general causes and responses to crime, including the circumstances surrounding legal and policy developments; and (4) a speculative discussion on future crime and crime control projections (Barak 2000). Countries were placed into one of three nation-state classifications based on (1) the social, political, and economic development/integration into the 20th century’s multinational corporatism and (2) the emerging 21st century’s lifestyle of consumerism (Waters 1995). These three classifications are: • Developed Nation-States: United States, Germany, United Kingdom, New Zealand, Taiwan, The Netherlands • Post-traditional Nation-States: Ghana, Nigeria, Navajo Nation • Developing Nation-States: Brazil, Poland, Russia, Iran, China, India GLOBALIZATION OF CRIME CONTROL 59 Theoretical Orientations to Crime and Crime Control In the field of comparative criminology, there are essentially three macro perspectives on the development of crime and crime control: modernization, world system, and opportunity. The models of modernization are based on the ideas of Emile Durkheim, the models of world system are based on the ideas of Karl Marx, and the models of opportunity are derivative of 18th century classical rationalism and 20th century evolutionary ecology. Each of these models incorporates various assumptions, concepts, and relationships about law, crime, and social change; and each model proffers a different explanation for the variation in crime and crime control. The modernization explanations argue that social changes such as urbanization and industrialization are associated with changes in crime and victimization patterns. According to these models, the changes in crime and crime control are primarily the products of the internal influences of development, regardless of time and place. World system explanations argue that contemporary developing nations are dependent on, and to varying degrees exploited by, already developed nations. Hence, the changes in crime and crime control in any country, regardless of its developmental status, are primarily the products of external influences in relationship to a changing political economy. Opportunity explanations argue that both crime and crime control reflect a mixture of developing material resources and environments. Thus, changes in the patterns of crime and victimization, over time, are primarily the products of interacting internal and external factors. Typically, these three models are used to explain cross-national differences in crime rates such as homicide and theft, and are viewed as alternative or as competing models. It makes more sense, however, to view them as overlapping and complementary. The modernization and opportunity approaches are both “evolutionary, focus on processes of adaptation, and emphasize industrialization, urbanization, cultural diversity and population growth,” while the opportunity and world system approaches are more “holistic, materialistic, relational, and emphasize hierarchical or dominance relations” (Neuman and Berger 1988: 288). It might also be that the ultimate value of these theories, especially in the context of globalization, may turn out to reside in some kind of integration of the various models. Modernization Theory Based on the Durkheimian (1964) notions of “the division of labor in society,” “mechanical/organic solidarity,” “anomie,” and “cultural lag,” modernization theorists have contended that it is the speed rather than the level of development which is important for understanding the patterns or variations in crime 60 GREGG BARAK and crime control. Rapid change intensifies conflicts and throws society into a temporary state of disequilibrium where deviance and crime tend to expand as values clash regarding appropriate norms. Advocates of this theoretical perspective explain cross-national variation “in terms of industrialization, urbanization, and the resultant social disorganization and anomie” brought on by modernization (Neapolitan 1997: 68). Shelley (1981) posited that all nations experience the same developmental stages and corresponding changes in their crime patterns. These models maintain that the developing nations of the world should be viewed as undergoing essentially the same kind of evolution in crime and crime control that contemporary developed nations experienced in the 19th century or earlier. Hence, modernists argue that the up and down trends in crime and crime control should be more or less the same for all nations as they adapt to similar levels of rapid development. Modernization theory regards urbanization and industrialization as explaining more about the patterns of property and violent crime from one nation to the next than do the cultural characteristics, politics, or local histories, for example. The processes of modernization are viewed as disrupting traditional family and community patterns of organization as peasants become workers and as societies transform from agricultural to industrial and service-oriented economies. During these transitional stages of development, a breakdown of norms occurs where new beliefs and values are substituted for older ones. In the process, both society’s formal (legalistic) and informal (customary) social controls are weakened. Crime increases because of these changes in community relations. In discussions of crime (and deviance) control, modernists employ rather narrow conceptions. For example, crimes are those acts subject to “repressive” sanctions (criminal law) and not to “restitutive” sanctions (civil law). These models of modernization incorporate a value-neutral perspective, as criminal law is regarded as embodying societal consensus and criminality is equated with official crime rates. These approaches do not examine how political processes affect crime rates. Although relatively apolitical, these approaches adopt pluralist theories of the state. Finally, these models view both individual and collective acts of crime and violence as increasing when institutions (especially political ones like the state) lag behind the pace of modernization or when relative deprivation is experienced by enough people who see large gaps between what they and what others possess and consume. Nonetheless, modernists consider the relations of inequality as of secondary importance since they argue that equality is increasing as nations develop economically, technologically, and otherwise. GLOBALIZATION OF CRIME CONTROL 61 World System Theory In contrast to modernization theory, world system theory views inequality as a primary cause in the production of crime and crime control. World system or dependency perspectives maintain that inequality increases as the capitalist mode of production expands and as developed nations penetrate developing nations. Utilizing Marxian perspectives on conflict and world order, which argue that crime and crime control are functions of expanding or contracting inequality, world system theory emphasizes the importance of the economic over non-economic factors. For example, world system theory contends that the dependence of developing nations on developed nations is fundamental to the trends in crime and crime control. World system theories further claim that nations do not move along a linear track toward development, with dependence and underdevelopment being temporary states. Instead, exploitation results in uneven development both within and across nations, creating inequalities that were not present when the developed nations modernized. World system theory argues further that what is important are the linkages between the internal processes of dependency and the external relations of the global economic order. As such, the main causal variables of world system theory are “the global economy and uneven expansion of the capitalist mode of production, the international system of states, class structure and conflict, economic inequality, the class nature of the state, and the spread of new ideologies” (Neuman and Berger 1988: 284). What is at the heart of these analyses of crime, crime control, and the world system are the interrelations or inequalities among the various social formations that are used as sites for examining structures and processes within and between nations. According to Wallerstein (1974), world system theory begins with an account of colonialism and its consequences for the uneven development of nations, and then examines the advancement of global capitalism in relation to the worldwide political-economic system of core, semi-periphery, and periphery nations. Core nations refer to the developed nations of the world, characterized by highly industrialized and prosperous formations where capitalist social relations are most elevated. Nations of the periphery or of the Third World, are the least developed and the most economically dependent formations, often providing a source of inexpensive natural resources for core nations. Recently emancipated from the colonial domination of core states, these dependent formations are the least industrialized, have the lowest standards of living, and are often the locales of civil war, military dictatorships, protest, and rebellion. Semi-periphery nations are partially industrialized and developed. These nations provide investment outlets for the core states and 62 GREGG BARAK often act as political and economic “buffers” between nations of the core and periphery. The consequences of the relations and social formations of capitalist development are that core nations typically exploit the periphery and semiperiphery nations to differing degrees. One significant result of the relationship between the nations of a global economy has been that the surplus or profit created by development in the periphery (and to a lesser extent in the semi-periphery) is extracted by domestic and international elite groups, acting legally and extra-legally. The experiences of today’s developing countries appears to be unlike the experiences of the already developed nations, where the money made had remained internal, circulating and propelling the local economies and even reducing inequality for a period of time. In discussions of crime (and deviance) control, world system theorists are not exclusively guided by criminal and legal considerations. They prefer to draw broader definitions of crime based on socio-political concepts grounded in relations of production and power as well as social injuries and harms. Thus, crimes may also include acts committed by corporations and states that may or may not fall within the narrow confines of the criminal law, such as those acts that include human rights violations, global treaties, and universal covenants (Schwendinger and Schwendinger 1970; Barak 1990). In this context, world system theory analyzes the state, law, and crime in terms of both political and class struggle and in terms of the developing modes of capitalist production. Using this approach, crime and crime control are viewed as products of inequality, poverty, and political oppression that accompany development based on exploitation and dependency (Humphries and Greenberg 1981; Platt and Takagi 1981). In the context of capitalist development and class struggle, the various forms and expressions of crime itself are regarded as structural adaptations, where there are essentially two kinds of crime happening (Quinney 1977). First, there are the “crimes of domination or repression” methodically committed by capitalist/governing classes and their agents. Second, there are the “crimes of accommodation,” “crimes of interpersonal violence,” and “crimes of resistance/rebellion” committed by working and subordinate classes, and referred to in their totality as “crimes of survival.” Examples of the crimes of domination or repression encompass those acts which systematically injure or harm consumers, workers, and the general public, including offenses like antitrust violations, worker-related illnesses and deaths, environmental pollution, organized corruption, abuses of law enforcement, and so forth. Examples of the crimes of survival include property crimes of theft and illegal entrepreneurial activities involving drug sales, gambling, and prostitution; violent crimes like murder, assault, rape, and GLOBALIZATION OF CRIME CONTROL 63 domestic violence; and offenses like political organizing, protests, riots, and strikes. When comparing Quinney’s classification of crimes in the context of developing political economies, the former acts can be viewed generally as the product of group and organizational activities engaged in on behalf of the direct or indirect accumulation of capital, while the latter acts can be viewed generally as the product of isolated and individual incidents of predation (although there are the less common social and collective acts of resistance and rebellion). Of course, organized criminal activities, especially those involving the distribution of contraband across nation-states, cut across all forms of criminality. Opportunity Theory A third perspective on cross-national crime and crime control blends elements of the ecological theories of crime and social change – a legacy of the Chicago School of the 1920s and 1930s. This approach emphasizes the interrelations between the physical environment, population groupings, and opportunity theory which maintains that “crime occurs in spatially and temporally organized social contexts that provide ‘favorable’ environmental conditions for the execution of criminal acts” (Neuman and Berger 1988: 287–288). Opportunity theory borrows from differential opportunities theory and routine activities theory (Cohen and Felson 1979) which argue that crime is a function of rational behavioral options. It also incorporates ideas from civilization theory (Elias 1982), which argues that there are fewer positive supports for violence as populations grow in density and wealth. Thus, opportunity models predict increasing levels of property crime and declining levels of personal crime. Increased criminality is associated with all groups and classes of people, including women and juveniles, especially in the latter stages of capitalist development when suitable targets and a surplus of goods are increasingly vulnerable and subject to socio-economic activities that are widely dispersed and less protected from capable guardians. The opportunity model argues that the variation in cross-national crime and crime control is the product of a mixture of growing resources and environments, which provide increased opportunities for unsanctioned criminal and deviant behavior. These acts will increase “when evolutionary processes create a societal surplus which expands the quantity of material goods available to be stolen” (Neuman and Berger 1988: 288). At the same time, when “population mobility and cultural diversity increase, the protection and social control mechanisms provided by large, stationary kinship groups” lose their effectiveness in suppressing crime (Neuman and Berger 1988: 288). 64 GREGG BARAK In discussions of crime (and deviance) control, opportunity theorists take a managerial or administrative approach. Their emphasis is on the inter-organizational environments, technology, rationalization, communication systems, and centralization. Like the modernists, the opportunists’ conceptions of crime and crime control do not question the criminal law produced by the state’s technical-administrative apparatus. Instead, their approaches – dominated by ecological systems of opportunity – argue that crime and crime control proceeds instrumentally in relation to rationally calculated interests within a context of scarce resources and changing openings for mobilization. Accordingly, crimes from above or below, individual and collective, occur in contexts of cost-and-benefit analyses. Assessments and Integrations As previously noted, there are similarities and differences as well as overlapping assumptions, definitions, and explanations shared between the modernization and opportunity models on the one hand, and between the opportunity and world system models, on the other hand. Moreover, concepts such as urbanization, industrialization, population mobility, and cultural diversity are shared by the three models. Based on a 15 nation-state survey (Barak 2000), it can be argued that each of these models underscores important dimensions and significant factors from which we can further our comparative understanding of crime and crime control. Hence, processes like bureaucratization, accumulation, and marginalization contribute to the co-production and distribution of crime and crime control. It remains a fundamental task of comparative criminology, however, to unravel the interactions and interrelations of these variables as they work themselves out in the socio-historical realities of crime and crime control. Future scholarship may very well contain characteristics of both modernist and postmodernist theory. In the former, there are models of multiple interactive causality or of dialectical causality involving reciprocities between the various elements of modernization, opportunity, and world system. In the latter, there are the models of recursive productions and routinized activities that have become part of the historically and culturally specific formations of modernization, opportunity, and world system (Barak 1998). For example, it is not only the case of industrialization, urbanization, inequality, and their relationships to crime and crime control over time that is crucial. Also important are local conditions in the context of time and place and the related political culture. The development of crime and crime control at the macro levels is about: (1) how different modes of production and levels of development interact with each other; (2) what constitutes the political arenas or venues of action; (3) when urbanization occurs and technology develops; GLOBALIZATION OF CRIME CONTROL 65 (4) why inequality expands or contracts the availability of licit and illicit opportunities; and (5) how each of these is influenced by the interaction of localization and globalization. Having established that cross-national specificity is still required to make sense out of the development of crime and crime control, we were able to reach some provisional conclusions. First, both crime and crime control are growing and expanding enterprises worldwide. This was generally true for developed, developing, and post-traditional countries. As inequality grows within and between nation-states, crimes against property generally expand more rapidly than crimes against the person. Although there are certainly differences within and between the expressions of criminality for the three groupings of nation-states, these are nevertheless more uniform and similar when compared to the responses to crime or to the reactions of crime control. In other words, across nation-states, regardless of the stage of development, the trends in crime are more alike than are the trends in crime control and criminal justice. Comparatively, the opportunity and world system models were more accurate than the modernization model. While urbanization and industrialization are certainly material to changes in crime and crime control as modernization theory forecasts, time and place also matter despite the arguments by Shelley (1981) and others to the contrary. Similarly, rapid change and relative deprivation are both related to changes in crime and crime control, but only in context of an expanding worldwide inequality in economic consumption. Differential opportunities, growing inequalities, privileges, and other relations of domination and exploitation characterize the interactions between internal and external influences on both crime and crime control, as suggested by opportunity and world system theories. Globalization, Crime, and Crime Control Globalization refers to the interdependency of events, people, and governments around the world, such as the Seattle and Washington meetings of the World Trade Organization in December 1999 and the protests it caused. Globalization appreciates that: the effects of the events on one side of the world are likely to ripple all the way around the globe. Calculations of national sovereignty are routinely affected by the interest and needs of 160 other nations. There is no longer a country on the face of this shrunken planet that can go it alone (Hufstedler 1980: 8). In particular, globalization refers to a worldwide political economy and an expanding communications, transportation, and computer network that has 66 GREGG BARAK the effect of reducing the planet to a global village, which also creates an excess of opportunities for capital(ist) and criminal(ist) expansion, or for the exchange of both licit and illicit goods or services. The latter appears as both innovative and spontaneous forms of old and new crimes against property and person alike. For example, embodied in the illicit customs of globalization are the fraudulent and unfair trade practices in commerce, the laundering of unauthorized drug and arms trade profits, the smuggling of illegal immigrants into and out of nations, the dumping of toxic waste and other forms of ecological destruction, the acts of terrorism committed by and against various states, and the behavior of multinationals to move capital and technology to exploit cheap labor. Other crimes of control and domination engaged in by corrupt police, militia, and other governmental agents, inside and outside of the criminal justice system, also reflect the illicit commodities of international productivity and service delivery. These crimes appear geographical and involve organized networks functioning in local, regional, national, and international markets. At the same time, the developing contours of some criminality are currently undergoing fundamental changes that “with decreases in barriers of language, communication, information and technology transfer and mobility, and the ever increasing globalization of the economy, there has been a growing trans-national character of organized, financial, sex-related, immigration, and computer crime” (Travis 1998: 1). In addition to the models of modernization, world system, and opportunity, there are the more recently articulated theories of global capitalism which have yet to fully address the issues of crime and crime control. Global capitalism refers to a domination by multinational firms over those corporate firms that tend to operate and compete as local or national enterprises. Fundamental to this view is the belief that the primary engine driving capitalism today is the system of global production or the internationalization of labor and capital relations. As Ross and Trachte (1990: 25) put it: The theory of global capitalism differs from that of monopoly capitalism because it sees the dominance of global production organizations as changing the national structures and processes characteristic of the monopoly era. The theory also differs from that of world system analysis whose starting point defines capitalism as a system of production for exchange, and which explicates the current world system as a hierarchical order of exchange rather than production relations. Despite the internationalization of the world economy, it does not appear from our global study that national cultures, economies, and borders are dissolving. Nor does it appear that “national economies and, therefore, GLOBALIZATION OF CRIME CONTROL 67 domestic strategies of national economic management are increasingly irrelevant” (Hirst and Thompson 1996: 1). At the same time, however, the emergence of the Euro on January 1, 1999 was a means of Western Europe “cooperatively” trying to compete with the US Dollar and the Japanese Yen. These new economic relations have many regional and global implications, not only for the monetary and fiscal policies around the world, but also in terms of the various kinds of marginality and crime and crime control. As the London Sun warned its readers, for example, “Britain may be forced to adopt Eurowide criminal laws laid down by Brussels” (quoted in Guttenplan 1999: 18), and, “while Prime Minister Tony Blair may like the Euro, many Britons are more circumspect, even though the average bloke probably feels closer to Europe today than at any time in recent history” (Guttenplan 1999: 18). Hence, endogenous struggles are still very important in shaping public and private policies. It is not the argument here some globalists make that the world economy is dominated by uncontrollable market forces and by transnational corporations that no longer owe any allegiance to the nation-state (Friday 1996; Jamieson, South, and Taylor 1997; Sheptycki 1998). Rather, like Hirst and Thompson, it is argued that these economic developments are more complex and more equivocal than many global analysts believe. The point being that from a developmental, comparative, and global perspective, our 15 nationstate study of crime and crime control revealed the importance of paying close attention to local conditions, domestic strategies, and national policies in relationship to the internationalization of the world economy. As Hazlehurst and Hazlehurst (1998: 18–19) have similarly concluded regarding their comparative project on gangs and youth subcultures: “What is now imperative is an international research strategy that encompasses both a myriad of local studies and . . . global processes, global units of analysis, and global institutions.” Ultimately, in what may be regarded as a contemporary period marked by both globalizing and particularizing trends in crime and punishment, it is up to criminologists to ferret out the interdependent relationships between capital, labor, crime, and crime control, on the one hand, and between the intensification of cultural, commercial, transportational, communicational, and homogenizing elements on the other hand. More generally, as Waters (1995: 9) has concluded: “the globalization of human society is contingent on the extent to which cultural arrangements are effective relative to economic and political arrangements.” Waters (1995: 10) further suggests “we can expect the economy and the polity to be globalized to the extent that they are culturalized, that is to the extent that the exchanges that take place within them are accomplished symbolically. We would also expect that the degree 68 GREGG BARAK of globalization is greater in the cultural arena than either of the other two” (Walters 1995: 10). Meanwhile, this global homogenization is in tension with the political or nationalistic culture of particular sites. From a transnational or global point of view, a negative outcome of the thawing of East/West relations has been the influx of conflict and crime worldwide. During the past decade, one consequence of the end of the Cold War has been an international increase in “border crimes,” especially those involving the smuggling of goods and services out from and into various nation-states. For example, as late as the late 1980s, the majority of stolen cars in the United States were hot-wired for joy riding or for domestic chop shops to sell the disassembled parts locally. With the breakup of the Soviet Union, the loosening of border controls across Eastern Europe, and the opening up of the free market in the early 1990s, the international demand for stolen cars (as operating vehicles) had increased and expanded worldwide. As a result, and in less than one decade, both the business of stealing cars and of protecting cars in the U.S. and elsewhere was completely transformed (Bradsher 1999). Crime control had similarly transformed its relations, especially in terms of the development of transnational policing and law enforcement plans, stratagems, and cooperation (Moore and Fields 1996; Sheptycki 1998). Cross-National Capital, Labor, Crime, and Crime Control: Implications for the Future In the context of capitalist development worldwide, the 21st century is witness to a highly mobile and international capital. At least since the end of the Cold War, capital has been actively engaged “in a worldwide political offensive in favor of free trade, deregulation, privatization and cuts in social spending” (Early 1998: 33). By contrast, labor has been fairly quiet, provincial, and regional in its orientation to the global economy. Hence, labor primarily operates “within the framework of a single national-state or, worse yet, one domestic industry, firm, or craft” (Early 1998: 34). Unlike capital, labor remains fragmented and disorganized. It finds itself still groping for ways to go beyond traditional forms of nationalism. Workers, whether organized or not, have been in retreat, on the defensive, and predominantly absorbed in struggles against the further erosion of their position in the capital-labor schemata of a worldwide swing to laissez-faire capitalism. At best, instead of contributing to the implementation of labor-based alternatives to the competitive agendas of capital, labor has been capitulating to labormanagement partnerships of free enterprise. The results have been a reduction in the average living standards of workers worldwide. GLOBALIZATION OF CRIME CONTROL 69 When it comes to criminality, both criminals and criminal organizations seem to resemble business more than they do labor and criminal justice/crime control. By analogy, organized criminals, like their counterparts in organized business, may resort to knowledge, wealth, and violence to accomplish their tasks. Tactically speaking, however, both prefer using knowledge and capitalizing on the “information revolution” rather than using force or wealth. For example, Moore and Fields (1996: 3) have observed that criminal organizations “have taken advantage of the information and technological revolution to as great or a greater degree than has government and business, and to a far greater degree than has the criminal justice system.” In fact, some criminal confederations are using intelligence systems, developing technology, and fine-tuning their criminal techniques on an international scale. On balance, the information and intelligence networks of some crime groups have been superior to those of law enforcement, but these networks are still no match for the most powerful multinational corporations. These changing developments involving capital, labor, crime, and crime control are motivated by three contemporary economic forces that are driving the capital mega-mergers of the new world order of global companies, such as the $39.5 billion merger announced in May, 1998, involving Germany’s Daimler-Benz and the USA’s Chrysler. These forces are: (1) the demand by money managers for quarterly performance and earnings growth that will propel stock prices upward; (2) the marketplace itself, whose logic demands companies integrate their global strategies with global capital markets; and (3) the rush for global market dominance. Some international bankers and globalists go so far as to argue that the world has already moved from the reality of globalization to the reality of “globality.” A world of globality, they argue, is not so much a process but rather a condition – a world economy in which traditional and familiar boundaries are either being surmounted or made inconsequential. While it is true that some of the old ways and rules of doing business and/or crime/crime control have changed, these relations have not necessarily wiped out the traditional and familiar ways. Accordingly, based on the findings from our 15 nation-state study of crime and crime control, globalism was regarded as a process rather than a condition. In effect, the globalization of capital, labor, crime and crime control over the last decade revealed a developing worldwide political economy where capitalist reformers in Mexico or in the former Eastern bloc nations, for example, may have promised prosperity for their nation’s workers; but so far, these laborers have been the primary benefactors of austerity, job insecurity, and greater underemployment and unemployment. Moreover, these workers have also experienced increasing levels of criminal victimization. In other 70 GREGG BARAK parts of the world – developed, developing, and underdeveloped – the same trends appear: an unparalleled and untrammeled degree of competition in which free trade, cheap labor, and crime are all abundantly available and in great demand. At the same time, the profits from capitalist expansion and global consumerism were at an all time high. In addition, with respect to crime control and the integration of criminal justice services, the ideology and practice of the privatization of crime and punishment fit neatly into the political economies of contracting welfare states and deregulating governments. Hence, the still nascent but emerging sectors of the world’s criminal justice systems are tinkering with something that has become a publicprivate apparatus of crime control. Often referred to as “police-industrial” and “penal-service” complexes, these public-private forms of crime control may contribute to multinational profits as they simultaneously create new forms of surveillance and “corrections” (Nuzum 1998). In sum, as part of the process of globalization, both the legitimate and illegitimate fields of criminal enterprise have been freed-up for the greater exploitation of all of humankind: The borders that constrained commerce – but also protected companies from the full brunt of competition – are eroding. Governments are retreating from control of the commanding heights of their economies: they are privatizing and deregulating. Barriers to trade and investment are coming down rapidly. Ever-cheaper communications and ever-faster computers, along with the Internet, are facilitating the flow of goods and services, as well as knowledge and information (Yergin 1998: 27). And so it goes: with global competition and the international flow of goods and services, knowledge and information, comes the worldwide growth in crime and crime control. Yet these global developments are not unencumbered by the local experiences of politics, nationalism, and domestic strategies of social change. Hence, the internationalization of markets in all kinds of criminal contraband – including weapons, drugs, sex, alcohol, tobacco, coffee, and computers – and the international efforts to combat this activity, are both illustrations of the globalization of crime, surveillance, and social control. Nevertheless, despite the development of transnational crime, the bulk of crime, violent and property, individual, organized, corporate, or governmental (state), is still largely confined within the geographic boundaries of the existing nation-states. Today, however, the roots of both crime and crime control may be found in the interplay between local, national, and global forces. GLOBALIZATION OF CRIME CONTROL 71 References Barak, G. (1990). Crime, criminology, and human rights: Toward an understanding of state criminality. Journal of Human Justice 2(1), 3–14. Barak, G. (1998). Integrating Criminologies. Boston: Allyn and Bacon. Barak, G. (2000). Crime and Crime Control: A Global View. Westport, CN: Greenwood Press. Bradsher, K. (1999). For car thieves, a technological arms race. New York Times (March 21), F1. Cohen, L. and Felson, M. (1979). Social change and crime rate trends: A routine activities approach. American Sociological Review 44, 588–608. Durkheim, E. (1964). The Division of Labor in Society. New York: Free Press. Early, S. (1998). Slicing the globaloney: A review of workers in a lean world: unions in the international economy. The Nation (February 16), 33–35. Elias, N. (1982). The Civilizing Process, II: State Formation and Civilization. Oxford: Oxford University Press. Friday, P. (1996). The need to integrate comparative and international criminal justice into a traditional curriculum. Journal of Criminal Justice Education 7, 227–239. Guttenplan, D.D. (1999). Letter from London: Tony Blair quite likes the euro, but many britons think it’s a foreign plot.” The Nation (January 11/19), 18–24. Hazlehurst, K. and Hazlehurst, C. (1998). Gangs and Youth Subcultures: International Explorations. New Brunswick, USA: Transaction Publishers. Hirst, P. and Thompson, G. (1996). Globalization in Question: The International Economy and the Possibilities of Governance. Cambridge: Polity Press. Hufstedler, S.M. (1980). World in transition. Change (May/June), 8–9. Humphries, D. and Greenberg, D. (1981). The dialectics of crime control. In D. Greenberg (ed.), Crime and Capitalism. Palo Alto: Mayfield, pp. 209–254. Jamieson, R., South, N., and Taylor, I. (1997). Economic Liberalisation and Cross-Border Crime: The North American Free Trade Area and Canada’s Border with the US. Salford Papers in Sociology, No. 22: The University of Salford, UK. Moore, R.H. and Fields, C.B. (1996). Comparative criminal justice: Why study? In C. Fields and R. Moore (eds.), Comparative Criminal Justice: Traditional and Nontraditional Systems of Law and Control. Prospects Heights, IL: Waveland Press, pp. 1–14. Neapolitan, J.L. (1997). Cross-National Crime: A Research Review and Sourcebook. Westport, CN: Greenwood Press. Neuman, W.L. and Berger, R.J. (1988). Competing perspectives on cross-national crime: An evaluation of theory and evidence. The Sociological Quarterly 29(2), 281–313. Nuzum, M. (1998). The commercialization of justice: Public good or private greed? The Critical Criminologist 8(3), 4–7. Platt, T. and Takagi, P. (1981). Crime and Social Justice. Totowa, NJ: Barnes and Noble. Ross, R.J.S. and Trachte, K.C. (1990). Global Capitalism: The New Leviathan. Albany, NY: State University of New York Press. Quinney, R. (1977). Class, State, and Crime. New York: Mckay. Schwendinger, H. and Schwendinger, J. (1970). Defenders of order or guardians of human rights? Issues in Criminology 5, 123–157. Shelley, L. (1981). Crime and Modernization: The Impact of Industrialization and Modernization on Crime. Carbondale: Southern Illinois University Press. Sheptycki, J.W.E. (1998). Policing, postmodernism and transnationalization. British Journal of Criminology 38(3), 485–503. 72 GREGG BARAK Travis, J. (1998). NIJ request for proposals for comparative, cross-national crime research challenge grants. U.S. Department of Justice: National Institute of Justice (April). Wallerstein, I.M. (1974). The Modern World-System. New York: Academic Press. Waters, M. (1995). Globalization. London: Routledge. Weiss, R.P. (2000). Introduction to “criminal justice and globalization at the new millennium.” Social Justice 27(2), 1–15. Yergin, D. (1998). The age of “globality”. Newsweek (May 18), 24–27. Oxford Economic Papers, 2015, 133–156 doi: 10.1093/oep/gpu040 Advance Access Publication Date: 10 November 2014 Does globalization mitigate the adverse effects of terrorism on growth? By Javed Younas Department of Economics, School of Business Administration, American University of Sharjah, PO Box 26666, Sharjah, UAE; e-mail: jyounas@aus.edu Abstract This study identifies the damaging influence wielded by terrorism on the economy. It investigates whether international openness limits the negative consequences of terrorism on economic growth. The analysis is focused on 120 developing countries over the period 1976–2008. The positive interaction effect of terrorism and globalization suggests that the latter ameliorates the adverse impact of the former on growth. I also identify the critical values of the globalization index where the negative effects of both domestic and transnational terrorism are offset by the positive impact of openness; this, however, happens at a significantly high level of openness. The findings are robust to using the disaggregated measure of globalization and some individual indicators of economic openness. The result helps explain why the growth consequences of terrorism vary across nations and hold important policy implications. JEL classifications: D74; F0 1. Introduction The detrimental effects of terrorism on the economy are well documented. It destroys human and physical capital, reduces foreign capital inflows, depresses the tourism industry, and increases the costs of doing business through higher wages and insurance premiums, among others.1 Terrorism is also known to have a negative influence on economic growth (e.g., Abadie and Gardeazabal, 2003; Blomberg et al., 2004, 2011; Gaibulloev and Sandler, 2008, 2009). Meierrieks and Gries (2013) find that terrorism is particularly harmful to growth in developing countries but not in the developed world. At the same time, globalization can be a powerful engine for growth in developing countries (e.g., Mishkin, 2009; Stiglitz, 2004). There is also some evidence of link between terrorism and globalization (Li and Schaub, 2004; Zimmermann, 2011). However, whether globalization mediates or exacerbates the negative effects of terrorism on growth has not yet been examined. I attempt to fill this gap in the literature. 1 See Sandler and Enders (2008) for the different mechanisms through which terrorism may affect an economy. C Oxford University Press 2014. V All rights reserved. 134 GLOBALIZATION, TERRORISM, AND GROWTH Globalization may not only reduce the effect of terrorism on growth by reducing terrorism itself, it may also reduce the effect of terrorism on growth even when it does not affect the level of terrorism. For instance, openness may allow for greater diversification of risk and more efficient international division of labor. It also promotes unification of goods and capital markets across the world by reducing restrictions on international trade and foreign investment. Many studies have demonstrated that reduction in trade protectionism and higher foreign investment increase economic growth (e.g., Alfaro et al., 2010; Vamvakidis, 1998). All of these factors may reduce a country’s economic vulnerability to terrorism. However, globalization may magnify the negative macroeconomic consequences of terrorism. It is not clear whether increased international competition due to rapid globalization can significantly benefit an economy (Gurgul and Lach, 2014). Openness may also coincide with more vulnerable chains of supply and production, which may make an open economy more exposed to terrorism. Since globalization is known to be a multifaceted phenomenon (e.g., Stiglitz, 2004; Pica and Mora, 2011), its interplay with terrorism and the resulting impact on growth is not clear a priori. On one hand, globalization may reduce terrorism, and thus, its negative influence on the economy in several ways. First, according to Li and Schaub (2004), progress in economic development due to elimination of restrictions on trade reduces incentives for people to engage in terrorism out of despair and poverty. Second, globalization increases the international division of labor, which in turn increases the opportunity cost of engaging in a terrorist act (Zimmermann, 2011). Third, the neoclassical model with decreasing returns and frictionless market predicts that economic openness will encourage capital to flow from capital-abundant rich to capital-scarce poor countries, expanding their reemployment base. Fourth, globalization can be a key factor in stimulating institutional reforms in developing countries (Mishkin, 2009). Finally, it enhances the free flow of ideas and knowledge across borders, which may attenuate extremist tendencies over time. On the other hand, indirect impacts of globalization could also lead to more acts of terrorism and result in lower growth. First, globalization may increase poverty by creating a bias in favor of capital-induced expansion through higher demand for skilled workers, but reduced demand for unskilled workers. Second, terrorism induces skilled migration (Dreher et al., 2011), and may also lead capital to flow overseas due to higher domestic risks. Given that globalization makes movements of labor and capital less costly, it may intensify the negative consequences of terrorism on growth. Third, globalization can widen the income inequality gap between rich and poor within and across nations (e.g., Bergh and Nilsson, 2010). Fourth, it may aggravate the ideological division between liberal and traditional forces in a society. Finally, it also facilitates the formation of networks, transfer of funds, and availability of sophisticated technology and weapons. All of these aid the participation in and implementation of a terrorist plot. As the foregoing discussion suggests, the theoretical literature cannot unambiguously establish whether openness helps mitigate the deleterious effects of terrorism on the economy. Also, it is unclear whether the relationship between growth and globalization is monotonic. Arguably, the positive impact of globalization may kick in only when it exceeds a certain threshold.2 I thus turn to an empirical investigation to understand these phenomena. 2 Stiglitz (2004) argues that adoption of institutions without proper adaptation can have opposing effects on economic growth. Likewise, Helpman et al. (2010) show that the distributional effects of trade on labor markets with heterogeneous firms are akin to those derivable in Heckscher-Ohlin models. J. YOUNAS 135 I estimate a dynamic panel data model for 120 developing countries over 1976–2008.3 To focus on the issues identified above, the model will include a measure of globalization, an indicator of terrorism, and an interaction term for the two. The reasons for choosing developing nations for the analysis are multifold. First, rich nations are more likely to escape the growth-repressing effects of terrorism because of their diversified economies (Gaibulloev and Sandler, 2009). Second, they are better equipped to employ monetary and fiscal policies to recover from a terrorism shock (Enders and Sandler, 2012). Third, because developing countries are economically smaller, a terrorist shock can have more adverse consequences for their economies. Fourth, since the 1980s there is greater proclivity for hosting terrorist groups in developing countries (Gaibulloev and Sandler, 2013). Finally, variations in the degree of openness are relatively large across developing nations, allowing for a more meaningful quantitative analysis of the effects of globalization. The unique feature of the terrorism data set is that it includes information on domestic and transnational incidents of terrorism. This allows me to explore whether different types of terrorism and their respective interactions with globalization yield different economic outcomes. The empirical procedure employs alternative econometric techniques and attempts to address a host of estimation issues such as possible measurement problem, omitted variables bias, and reverse causation. I also examine the presence of sample heterogeneity, as terrorism may cause different impacts for different countries (e.g., Blomberg et al., 2004; Meierrieks and Gries, 2013). The remainder of the article is structured as follows: Section 2 describes the data and variables. Section 3 outlines the empirical design. Section 4 presents estimation results, including some subsections of robustness analysis. Section 5 offers concluding remarks. 2. Description of variables and data 2.1 Main variables The terrorism data used in this article come from the Global Terrorism Database (GTD). This data set is maintained by the National Consortium for the Study of Terrorism and Responses to Terrorism (START, 2009). To distinguish between domestic and transnational terrorist incidents, I follow Enders et al. (2011) segregation of GTD incidents into three categories of terrorism: domestic, transnational, and ambiguous (whose category is unclear). Adding them provides the number of total terrorist incidents. This segregation would thus allow me to analyze the impacts of domestic and transnational terrorist incidents and their interplay with globalization on the economy, separately. As defined by Enders et al. (2011), ‘Terrorism is the premeditated use or threat to use violence by individuals or subnational groups against noncombatants in order to obtain a political or social objective through the intimidation of a large audience beyond that of the immediate victims’. I briefly describe the difference between domestic and transnational terrorist incidents. If victims, targets, perpetrators, or audience belong to the venue country, then the incident is deemed a domestic terrorist incident. If an incident in one country also involves victims, targets, perpetrators, or audience from another country, then the incident 3 Table A1 of Appendix A provides the list of countries in the study. I exclude Afghanistan, Iraq, Palestine, and western Gaza from the data set because they are outliers in terms of the number of terrorist incidents and warlike conditions there. The limitation on data availability for other variables is another reason for their exclusion. 136 GLOBALIZATION, TERRORISM, AND GROWTH Table 1. The number of terrorist incidents, fatalities and injuries in my sample Years 1976–79 1980–83 1984–87 1988–91 1992–95 1996–99 2000–3 2004–7 Total Total T. Total T. Total T. Dom. T. Dom. T. Dom. T. Trans. T. Trans. T. Trans. T. incidents fatalities injuries incidents fatalities injuries incidents fatalities injuries 2,224 5,713 7,686 10,734 7,405 5,219 2,366 2,478 43,825 2,354 9,174 14,454 18,369 14,356 19,520 7,480 8,382 94,089 1,959 6,564 12,937 15,911 17,536 24,982 12,418 17,152 109,459 1,306 4,394 6,031 8,695 5,332 3,560 2,048 2,138 33,504 1,580 6,401 10,586 14,286 10,018 13,472 6,170 6,870 69,383 1,406 5,558 9,588 14,351 13,785 16,984 10,711 13,690 86,073 550 752 658 1,256 1,081 826 237 265 5,625 367 444 559 1,167 1,140 1,845 943 1,318 7,783 206 392 656 723 2,267 5,125 1,537 3,145 14,051 Notes: T ¼ terrorist; Dom ¼ domestic; Tran ¼ transnational. Terrorism incidents, fatalities, and injuries are summed over the four-year period. Total terrorism incidents, fatalities, injuries are sum of domestic and transnational terrorist incidents, fatalities, injuries as well as the terrorist incidents and casualties that cannot be unambiguously assigned to either of these two types of terrorism. is considered to be a transnational terrorist incident. In transnational terrorism, the main targets are foreign personnel and assets, an example of which is an act that destroys the offices of a foreign company. All other damages are called auxiliary losses. Both domestic and transnational terrorism disrupt or destroy the physical and human capital of a country and heighten domestic risks and uncertainties. Although the relative impact of the two kinds of terrorism on the economy is an empirical question, an act of transnational terrorism creates more panic because it draws wider media attention. Bandyopadhyay et al. (2011) argue that poor countries are limited in their ability to cope with transnational terrorism, as their perpetrators are typically based in foreign lands. Table 1 displays the number of total, domestic, and transnational terrorist incidents, as well as the fatalities and injuries resulting from each type of terrorism during the sample period. Over the entire sample period, there were a total of 43,825 terrorist incidents with 94,089 fatalities and 109,486 injuries, of which 33,504 were domestic terrorist incidents with 69,383 fatalities and 86,073 injuries; only 5,625 were transnational terrorist incidents with 7,783 fatalities and 14,078 injuries. Transnational terrorism therefore accounted for only 16.8% of total terrorist incidents. However, as already mentioned, it may have more adverse effects on the economy. Table A2 in Appendix A shows how the terrorist incidents vary across the countries. I employ the 2011 version of the KOF Index of Globalization, compiled by Dreher (2006) and updated by Dreher et al. (2008). This index has been used by several recent studies on globalization (e.g., Bergh and Nilsson, 2010; Hessami, 2011). A unique feature of this index is that it covers three dimensions of openness of a country: economic, political, and social globalization, with their respective weights of 36%, 26%, and 38%. Its index values range between 0 and 100, with a higher value indicating more openness.4 There are some obvious advantages of using this composite measure over any individual indicator of openness in our empirical analysis. First, it minimizes the omitted variables bias (Dreher, 2006). Second, it is the overall effect of openness that matters for its role in containing 4 See http://globalization.kof.ethz.ch/ for detailed information on KOF index and the method of its calculation. J. YOUNAS 137 (or stimulating) the harmful influence of terrorism. Finally, including individual variables of openness in a regression can induce collinearity problem. Economic globalization contains information on actual flows (trade, foreign direct investment, portfolio investment, and income payments to foreign nationals) and restrictions (hidden import barriers, mean tariff rates, taxes on international trade, and capital account restrictions). Political globalization takes into account the number of embassies in a country, the number of international organizations it belongs to, the number of UN peacekeeping missions it has taken part in, and the number of international treaties it has signed. Social globalization is dependent on personal contact (telephone traffic, transfers, international tourism, foreign population in a country, international letters), information flows (Internet users, televisions, and trade in newspapers) and cultural proximity (number of McDonald’s restaurants, number of IKEA stores, and trade in books). 2.2 Other variables In addition to accounting for time and country-specific fixed effects in each regression, I include commonly used time-variant variables in the growth regressions, such as (i) the investment rate as proxied by gross fixed capital formation/GDP to account for the effect of capital accumulation, (ii) government consumption expenditure/GDP to capture the influence of government burden, (iii) log initial real GDP per capita to account for conditional convergence, (iv) log inflation rate for macroeconomic policies and stability, (v) log secondary school enrolments to capture the effect of human capital formation, (vi) an index of institutional quality for the effectiveness of formal institutions, and (vii) the lagged dependent variable to account for the effect of past economic shocks on current growth rate. Data for these variables are taken from the World Development Indicators (2011) of the World Bank, except for growth rate of real GDP per capita (constant year 2005) and institutional quality. Though the data for the former come from the Penn World Tables of Heston et al. (2011), the description of the latter is mentioned below. Institutional quality is measured by employing the data on ‘political rights and civil liberties’ from Freedom House (2010). The advantage of using institutional data from this source is that they have the widest coverage for developing countries, as needed for the application. ‘Political rights’ refer to the freedom to participate in the political process through voting, organizing political parties, forming opposition, and electing public representatives. Civil liberties entail freedoms of expression, religious belief, movement, exercising the rule of law, and the right to form unions. Each of these two indices is measured on a scale of 1 to 7, with a higher value reflecting lower institutional quality. Following others in the literature (e.g., Trumbull and Wall, 1994; Bandyopadhyay et al., 2012), I construct a combined freedom index by first adding and then reversing these two indices, such that the resulting index ranges from 2 to 14, with its higher value indicating a better institutional environment. Because terrorism in most countries is a low-probability event, its annual data show little variation over time. The same is the case for some other variables in the model such as globalization and institutional quality. Therefore, I average the data over a nonoverlapping four-year period, giving a total of eight time periods for each country. Besides introducing variation in the data over time, this also smooths out cyclical fluctuations. The descriptive statistics are reported in Table 2.5 Over the sample period, the average growth rate of real GDP per capita is 1.75%, with the maximum value of 23.1% (Azerbaijan) 5 I also tested the results by including a variable of civil conflicts in the model. Although its coefficient is insignificant in most of the regressions, the main results are robust with its inclusion. 138 GLOBALIZATION, TERRORISM, AND GROWTH Table 2. Descriptive statistics Variables Obs. Growth rate real GDP per capita Terrorism variables Total terrorist incidents Domestic terrorist incidents Transnational terrorist incidents Total terrorist incidents p.m.p. Domestic terrorist incidents p.m.p. Transnational terrorist incidents p.m.p. Globalization variables Globalization Economic globalization Political globalization Social globalization Other variables Investment (% of GDP) Government consumption (% of GDP) Ln (Initial real GDP per capita) Political rights and civil liberties Ln (Inflation) Ln (Secondary school enrolments) Foreign direct investment (% of GDP) Trade openness (% of GDP) 895 Mean 1.747 SD Min. Max. 4.36 23.03 23.10 0 0 0 0 0 0 481 430.25 52.75 83.89 67.69 7.44 960 960 960 959 959 959 11.41 8.73 1.46 0.95 0.69 0.17 42.47 34.38 4.58 4.11 3.32 0.55 942 798 942 942 41.01 43.64 49.32 33.81 12.66 16.23 20.10 16.24 15.29 8.29 6.59 6.14 78.61 92.12 93.14 82.38 861 870 896 909 842 829 819 871 21.12 15.54 7.91 7.46 2.34 3.70 2.35 76.31 7.36 6.51 1.13 3.42 1.39 0.84 3.53 40.43 2.08 1.38 5.38 2 1.78 0.64 20.63 11.39 68.83 59.18 11.51 14 9.02 4.74 26.94 347.20 Notes: p.m.p. ¼ per one million persons. Data for dependent variable are averaged over a nonoverlapping four-year period from 1977 to 2008, while the similar data averages for all independent variables are from from 1976 to 2007, giving me maximum of eight observations per country. and the minimum of 23.03% (Democratic Republic of Congo). The average number of total terrorist incidents per country stand at 11.41, with the maximum number of 481 (Peru). The average number of domestic terrorist incidents lie at 8.73, with the maximum number of 430.25 (Peru), whereas this average for transnational terrorist incidents stand at 1.46, with the maximum number of 52.75 (Colombia). The standard deviation of domestic terrorist incidents is 34.38 and is 4.58 for transnational terrorist incidents, suggesting that the frequency of the former is much more than that of the latter. The average value of the globalization index stands at 41.01, with the minimum of 15.29 (Bangladesh) and the maximum of 78.61 (Cyprus). 3. Empirical model and methodology 3.1 The empirical model and hypotheses I estimate a dynamic panel data model of growth rate of real GDP per capita of the following form: ðGrowth GDP p:c:Þit ¼ b0 þ b1 Tit þ b2 Git þ b3 ðGÞ2it þ b4 ðT  GÞit þ hZit þ st þ ci þ lit (1) where i refers to countries and t to time, st indicates time-effects, ci reflects country-specific effects, and lit is the usual error term. Although the time effects control for any time J. YOUNAS 139 varying common shocks, the country-specific fixed effects account for unobservable time-invariant heterogeneous factors that may influence the economy. T specifies the number of terrorist incidents, and G is the index of globalization. The quadratic term of G aims to capture any nonlinearity of globalization, as discussed in Section 1. Z is a vector of time-variant control variables. In eq. (1), the point estimate of the effect of a change in T is b1 þ b4 G. Therefore, the main focus is on the coefficients of b1 and b4. Given that higher value of T corresponds with more terrorism, I expect b1 to be negative. The partial effect of T will be evaluated at different levels of G in the sample. As discussed, the signs for b4 cannot be unambiguously assigned a priori. Our testable hypotheses, therefore, are as under: Hypothesis 1: Terrorism reduces growth: b1 < 0. Hypothesis 2: If openness mitigates the negative effect of terrorism on growth, then b1 þ b4 G > b1, where b1 < 0, but b4 > 0. But if openness exacerbates this effect of terrorism, then b1 þ b4 G < b1, where b1 < 0, but b4 < 0. Hypothesis 3: I expect b2 < 0 and b3 > 0, implying a U-shape relationship between globalization and growth. Initially, a local economy may lack the capacity to complement high level of foreign investment and capital flows. Moreover, foreign firms and entrepreneurs may undergo a gestation period to understand the local environment and cross the start-up hurdles. Therefore, positive impact of globalization may appear only after it has reached a certain threshold. 3.2 Econometric methodology I present estimation results using alternative econometric methodologies and by sequentially adding control variables in the baseline model to ensure that the findings are not spurious. I initially employ the feasible generalized least squares (FGLS) estimator because it can explicitly allow for the presence of heteroskedasticity across panels and serial correlation within a panel, which gives panel-corrected robust standard errors.6 A few recent panel data studies on terrorism have also employed the FGLS for the same reason (e.g., Gaibulloev and Sandler, 2008; Dreher et al., 2011). Theoretically, it is quite likely that both terrorism and globalization may be affected by the economic performance in a country. This raises concern about potential endogeniety in eq. (1). In the present model, this issue is more complex because of the interactive term. In fact, simultaneous causation is a pertinent issue for all the right-side variables in a growth model. The conventional solution to this problem calls for using the instrumental variables approach in the two-stage least squares. Since I use fixed-effects model specifications, any chosen instruments must display variation over time. Moreover, the exclusion restriction of instruments requires that they have high correlation with the instrumented variables, but be uncorrelated with the error term. Thus, the insuperable difficulty of finding such instruments for multiple endogenous variables (and their data for developing countries) renders this approach infeasible. The use of invalid instruments would rather contaminate the estimation results. Therefore, I first take nonoverlapping four-year data averaging for my dependent variable from the period 1977–2008, whereas similar data averaging for all time-variant independent variables is taken from the period 1976–2007. This strategy 6 I also derive estimation results using the ordinary least squares (OLS) with robust standard errors clustered by countries, and the main findings are qualitatively the same. 140 GLOBALIZATION, TERRORISM, AND GROWTH reduces contemporaneous correlation and also alleviates concern about endogeniety in the model.7 Thus, eq. (1) takes the following form: ðGrowth GDP p:c:Þit ¼ b0 þ b1 Ti;t1 þ b2 Gi;t1 þ b3 ðGÞ2i;t1 þ b4 ðT  GÞi;t1 þhZi;t1 þ st þ ci þ li;t1 (2) I acknowledge that the lagging of the independent variable does not properly resolve the concern of reverse causation. Therefore, I also report results by employing generalized method of moments (GMM) estimation technique, which has also been used in the recent literature of terrorism as well as of growth (e.g., Dreher et al., 2011; Loyaza et al., 2012). In a panel framework, two types of GMM estimators are commonly used in literature: the difference-GMM (DGMM) as proposed by Arellano and Bond (1991) and the system-GMM (SGMM) as proposed by Blundell and Bond (1998). DGMM takes the first difference of the data and then uses lagged values of the endogenous variables as their instruments. Arellano and Bover (1995), however, argue that the lagged levels are often poor instruments for the first differences. SGMM alleviates this problem by using additional moment conditions and combining regression equations in first differences and in levels. Thus, it uses both lagged differences and lagged levels as instruments for the endogenous variables. Furthermore, this estimator is also more suitable for large crosssections and small time periods. In implementing the SGMM estimator, I follow the guidelines in past studies. First, I only use internal instruments generated in the model. Second, for each regression, I report the Hansen J test of overidentifying restrictions and the autocorrelation test for confirming the validity of instruments and the absence of serial correlation in the residuals, respectively. In both these tests, failure to reject null hypothesis strengthens support for the model, which is the case in all of the regressions. However, these tests may lose power if the number of instruments, i, exceed the number of countries, n (e.g., Roodman, 2009). In all of the main regressions, the countries to instruments ratio, r ¼ n/i, is less than 1. I nevertheless check whether our results are robust to reducing instrument counts by limiting the number of lagged levels to be used as instruments. I employ the two-step GMM estimation technique because it is asymptotically more efficient and robust to all kinds of heteroskedasticity (e.g., Asiedu and Lien, 2011). In addition, I report estimation results that I derive using the finite-sample correction of standard errors in all GMM regressions, as proposed by Windmeijer (2005). 4. Estimation results 4.1 Total terrorist incidents and globalization In Table 3, columns (1) and (2) present results for FGLS regressions. Column (1) includes the main variables of interest along with the lagged dependent variable, time, and country-specific fixed effects. Column (2) also adds all other time-variant control variables. In both columns, the sign of the coefficient of total terrorist incidents is negative and statistically significant at 1% level. This is in line with past findings that terrorism causes damaging 7 The estimation results when I used contemporaneous values gave qualitatively the same results (see Table B1 of Appendix B). J. YOUNAS 141 Table 3. Total terrorist incidents and globalization Estimation technique! Independent variables ; FGLS (1) FGLS (2) SGMM (3) SGMM (4) SGMM (5) Total terrorist incidents, t–1 0.0196*** (0.001) 0.052 (0.320) 0.0004 (0.449) 0.0004** (0.013) 0.008 (0.715) 0.0264*** (0.005) 0.001 (0.978) 0.0001 (0.774) 0.0006*** (0.000) 0.064*** (0.005) 0.108*** (0.000) 0.085*** (0.000) 1.017 (0.104) 0.068 (0.118) 0.507*** (0.000) 0.365 (0.376) 8.705** (0.036) Yes Yes 0.0327*** (0.000) 0.223*** (0.000) 0.002*** (0.000) 0.0007*** (0.000) 0.113*** (0.000) 0.185*** (0.000) 0.150*** (0.000) 1.359*** (0.000) 0.137*** (0.000) 0.732*** (0.000) 2.412*** (0.000) 9.648*** (0.000) Yes 0.0336*** (0.000) 0.040 (0.407) 0.001* (0.077) 0.0008*** (0.000) 0.073*** (0.000) 0.143*** (0.000) 0.097*** (0.000) 1.248*** (0.000) 0.056 (0.102) 0.407*** (0.000) 2.246*** (0.000) 5.031*** (0.000) Yes 0.0327** (0.021) 0.223 (0.174) 0.002 (0.153) 0.0007*** (0.039) 0.113 (0.174) 0.185*** (0.003) 0.150*** (0.008) 1.359** (0.026) 0.137 (0.325) 0.732*** (0.002) 2.412*** (0.011) 9.648*** (0.003) Yes No No Yes No No Yes 632 120 118 1.02 632 120 109 1.10 632 120 118 1.02 0.331 0.417 0.275 0.317 0.331 0.436 0.004 55.75 0.008 0.004 Globalization, t–1 (Globalization, t–1)2 Total terrorist incidents, t–1  Globalization, t–1 Growth rate real GDP p.c., lagged Investment, t–1 (% of GDP) Govt. consumption, t–1 (% of GDP) Ln (Initial real GDP p.c.) Political rights and civil liberties, t–1 Ln (Inflation), t–1 Ln (School enrollments), t–1 Constant Time effects Country fixed effects Limit on # of instruments Windmeijer standard errors 0.090 (0.967) Yes Yes # of observations # of countries, n # of instruments, i countries/instruments ratio, r ¼ n/i Hansen J testa Autocorrelation testb 773 120 @growth/@T ¼ b1 þ b4G Threshold level of G 0.0032 632 120 0.0018 Notes: Dependent variable: growth rate real GDP per capita. FGLS regressions are allowed for the presence of heteroskedasticity across panels and autocorrelation within panels, which provides panel-corrected robust standard errors. Two-step procedure for the SGMM is used, which is asymptotically efficient and robust to all kinds of heteroskedasticity. Column (5) derives estimates using Windmeijer finite-sample corrected standard errors. Superscripts ***, ** and * indicate significance at the 1%, 5%, and 10% levels, respectively. p-values are in parentheses, as well as for Hansen J and autocorrelation tests. a The null hypothesis is that the instruments are not correlated with the residuals. b The null hypothesis is that the error term exhibits no second-order serial correlation. 142 GLOBALIZATION, TERRORISM, AND GROWTH effects on the economy (e.g., Gaibulloev and Sandler, 2008; Dreher et al., 2011; Bandyopadhyay et al., 2014). The size of its coefficient indicates that one incident of total terrorism is associated with a 0.026% decline in growth rate (column (2)). Since the average country in the sample experienced 11.41 terrorist incidents a year, this finding is quite disconcerting. The interaction term of total terrorism incidents and globalization is positive and statistically significant in both columns. This suggests that globalization alters the relationship between terrorism and growth. To elucidate this results, I evaluate the partial effect, [@growth/@T], at the average value of globalization index in the sample, which stands at 41.01. As reported in Table 3, the negative effect of terrorism declines from 0.0196 to 0.0032 (0.0196 þ 0.0004  41.01) and from 0.0264 to 0.0018 (0.0264 þ 0.0006  41.01) for results in columns (1) and (2), respectively. This implies that greater openness can reduce the negative effects of terrorism on growth. In both columns, the coefficients of globalization and its squared terms are not statistically significant. Because FGLS results do not address the reverse causation problem per se, the inference based on them cannot be reliable. Therefore, as discussed in section 3, I also report the results of SGMM regressions (columns (3)–(5)). In column (3), the sign and significance of total terrorism incidents and its interaction term show that although terrorism has growthrepressing effect, globalization mitigates this effect. One incident of terrorism reduces growth by 0.0327%, but the partial effect, [@growth/@T], evaluated at the average level of globalization index suggests that international openness reduces this effect of terrorism on growth to 0.004%. Indeed, this encourages terrorism prone countries to implement reforms for international openness. I offer an example to further clarify these results. Consider Bangladesh, the fourth least globalized economy in the sample with an average value of globalization at 24.13. It experienced an average of 16.82 total terrorist incidents per year. Suppose that the number of total terrorist incidents in Bangladesh increase by 1 sample standard deviation (SD ¼ 42.47, see Table 2). Then, all else equal, this will decrease its growth rate by about 0.67% (@growth/@T ¼ (0.0327 þ 0.0007  24.13)  42.47 ¼ 0.67) for results in column (3). Now let us increase the level of globalization in Bangladesh to the level of Panama, the fourth most globalized country in our sample with an average value of globalization at 59.93, and it experienced an average of only 3.66 total terrorist incidents per year. Then, all else equal, raising the level of globalization of Bangladesh to the level of Panama will not only offset the negative effect of terrorism, it will result in a net effect of positive growth (@growth/@T ¼ (0.0327 þ 0.0007  59.93)  42.47 ¼ 0.39). The independent effect of terrorism on growth is negative, but the positive effects of the international exchange of goods and services as well as ideas and technologies appear to alleviate the growth-repressing effects of terrorism. The negatively significant coefficient of globalization and positively significant coefficient of its squared term suggest that the relationship between globalization and growth is U-shaped—the positive effect of globalization shows only when it exceeds a certain threshold. In Table 3, I also calculate and report this threshold level of globalization index (b2 þ 2  b3  (G) ¼ 0). In column (3), b2 ¼ 0.223 and b3 ¼ 0.002, so the value of G at this level is 55.75 (0.223 þ 2  0.002  (G) ¼ 0). Given that the median value of G is 40.24 in the sample, this calculation suggests that the majority of countries lie below this threshold level. J. YOUNAS 143 Now I briefly discuss the effects of other variables. Notice that the coefficients of political rights and civil liberties, log initial GDP per capita, and log school secondary enrollments that were insignificant in FGLS regression are statistically significant in SGMM regression. Their signs suggest that institutional improvements and higher human capital stock increase growth, whereas the negative sign on log initial GDP per capita confirms the convergence effect. The signs on other control variables also agree with the findings in the growth literature. In columns (4) and (5), I check the robustness of results by reducing the number of instrument counts and deriving estimates using the finite sample correction of standard errors, respectively. The sign and significance of the main variables of interest, that is, total terrorism incidents and its interaction term with globalization, remain intact. Also, notice that in all SGMM regressions, the p-values of the Hansen J test and autocorrelation test confirm the validity of instruments and the absence of second-order serial correlation in the residuals, respectively. 4.2 Does the mitigating role of globalization vary for both types of terrorism? To answer this question, I run regressions for the number of domestic and transnational terrorist incidents, separately. Table 4 reports the results of the former, and Table 5 presents the results of the latter. I follow the same estimation strategy as in Table 3 and report results for both FGLS and SGMM models. However, I only discuss the results of SGMM regressions in columns (3) in Tables 4 and 5. Both domestic and transnational terrorism negatively influence the economy, but openness reduces their damaging effects on growth. However, there is one notable difference. The direct harmful effect of transnational terrorism is substantially larger than that of domestic terrorism. One incident of the former reduces growth rate by 0.2465%, and one incident of the latter declines growth rate by 0.0371%. Interestingly, the partial effect, @growth/@T, evaluated at the average level of globalization index, suggests that openness can reduce these negative effects. Note that the average number of domestic and transnational terrorist incidents stands at 8.725 and 1.459 in our sample, respectively. Although one incident of transnational terrorism causes more damaging effects, the average country experienced 5.9 times more incidents of domestic terrorism, which also results in substantial losses to the economy. This finding agrees with the recent study of Bandyopadhyay et al. (2014), who quantify the consequences of these two types of terrorism on net foreign direct investment (FDI) position of a country. The effects of all other variables including that of U-shaped relationship of globalization remain the same as for total terrorism in Table 3. 4.3 At what level does globalization offset the negative effects of terrorism? So far I have calculated the marginal effect of terrorism at the mean level of the globalization index, G. However, it is important to know whether this effect is significant across the observed range of G in our sample. Therefore, I evaluate, @growth/@T, at the 10th, 25th, 50th, 75th, and 90th percentiles and mean levels of G in the sample, while all other variables are fixed at their mean values. The 10th, 25th, 50th, 75th, and 90th percentiles and means level of G largely correspond to the average values of G for ‘Bhutan, Sierra Leone’; ‘Maldives, Mali’; ‘Paraguay, Togo’; ‘Brazil, Brunei’; ‘Bahrain, Panama’; ‘Indonesia, Suriname’; and ‘Indonesia, Suriname,’ over the sample period, respectively. For this purpose, 144 GLOBALIZATION, TERRORISM, AND GROWTH Table 4. Domestic terrorist incidents and globalization Estimation technique! Independent variables ; FGLS (1) FGLS (2) SGMM (3) SGMM (4) SGMM (5) Domestic terrorist incidents, t–1 Globalization, t–1 0.0255*** (0.000) 0.056 (0.285) 0.0004 (0.411) 0.0005*** (0.009) 0.009 (0.682) 0.0317*** (0.000) 0.010 (0.847) 0.0001 (0.941) 0.0007*** (0.000) 0.064*** (0.006) 0.109*** (0.000) 0.083*** (0.000) 1.057* (0.091) 0.062 (0.157) 0.498*** (0.000) 0.407 (0.321) 8.774** (0.035) Yes Yes 0.0371*** (0.000) 0.207*** (0.000) 0.002*** (0.000) 0.0008*** (0.000) 0.104*** (0.000) 0.175*** (0.000) 0.155*** (0.000) 1.390*** (0.000) 0.134*** (0.000) 0.716*** (0.000) 2.451*** (0.000) 9.683*** (0.000) Yes 0.0446*** (0.000) 0.044 (0.336) 0.001** (0.046) 0.0010*** (0.000) 0.071*** (0.000) 0.142*** (0.000) 0.099*** (0.000) 1.266*** (0.000) 0.058* (0.087) 0.383*** (0.000) 2.259*** (0.000) 5.112*** (0.000) Yes 0.0371** (0.040) 0.207 (0.191) 0.002 (0.178) 0.0008* (0.077) 0.104 (0.211) 0.175*** (0.005) 0.155*** (0.005) 1.390** (0.026) 0.134 (0.343) 0.717*** (0.002) 2.451*** (0.011) 9.683*** (0.003) Yes No No Yes No No Yes 632 120 118 1.02 632 120 109 1.10 632 120 118 1.02 0.340 0.399 0.325 0.310 0.340 0.418 0.0043 51.75 0.0036 0.0043 (Globalization, t–1)2 Domestic terrorist incidents, t–1  Globalization, t–1 Growth rate real GDP p.c., lagged Investment (% of GDP), t–1 Govt. consumption (% of GDP), t–1 Ln (Initial real GDP p.c.) Political rights and civil liberties, t–1 Ln (Inflation), t–1 Ln (School enrollments), t–1 Constant Time effects Country fixed effects Limit on # of instruments Windmeijer standard errors 0.001 (1.000) Yes Yes # of observations # of countries, n # of instruments, i countries/instruments ratio, r ¼ n/i Hansen J test Autocorrelation test 773 120 @growth/@T ¼ b1 þ b4 (G) Threshold level of G 0.005 632 120 0.003 Note: Dependent variable: Growth rate real GDP per capita. Other notes are same as for Table 3. I use the estimates of b1 and b4 in column (2) for FGLS and column (3) for SGMM in Tables 4 and 5.8 FGLS results are reported in columns (1) and (2), and SGMM results are presented in columns (3) and (4) of Table 6. To save space, I only discuss results in columns (1) and 8 Since the results of total and domestic terrorist incidents are similar, I only report the findings of the latter from this point onward. J. YOUNAS 145 Table 5. Transnational terrorist incidents and globalization Estimation technique! Independent variables ; FGLS (1) Transnational terrorist incidents, t–1 Globalization, t–1 0.2389*** (0.000) 0.036 (0.469) 0.0001 (Globalization,t–1)2 (0.769) Transnational terrorist incidents, 0.0052*** t–1  Globalization, t–1 (0.000) Growth rate real GDP p.c., 0.008 lagged (0.743) Investment (% of GDP), t–1 Govt. consumption (% of GDP), t–1 Ln (Initial real GDP p.c.) Political rights and civil liberties, t–1 Ln (Inflation), t–1 Ln (School enrollments), t–1 Constant Time effects Country fixed effects Limit on # of instruments Windmeijer standard errors 0.354 (0.868) Yes Yes # of observations # of countries, n # of instruments, i countries/instruments ratio, r ¼ n/i Hansen J test Autocorrelation test 773 120 @growth/@T ¼ b1 þ b4 (G) Threshold level of G 0.0256 FGLS (2) SGMM (3) SGMM (4) SGMM (5) 0.2730*** (0.000) 0.032 (0.511) 0.0001 (0.785) 0.0061*** (0.000) 0.077*** (0.001) 0.109*** (0.000) 0.105*** (0.000) 1.070* (0.087) 0.072* (0.093) 0.541*** (0.000) 0.508 (0.208) 8.956** (0.032) Yes Yes 0.2465*** (0.000) 0.206*** (0.000) 0.002*** (0.000) 0.0060*** (0.000) 0.160*** (0.000) 0.209*** (0.000) 0.135*** (0.000) 1.282*** (0.000) 0.093*** (0.000) 0.673*** (0.000) 2.076*** (0.000) 8.939*** (0.000) Yes 0.1572*** (0.000) 0.005 (0.875) 0.0003 (0.204) 0.0041*** (0.000) 0.079*** (0.000) 0.111*** (0.000) 0.078*** (0.000) 1.283*** (0.000) 0.095*** (0.002) 0.391*** (0.000) 2.179*** (0.000) 5.328*** (0.000) Yes 0.2465** (0.060) 0.206 (0.140) 0.002 (0.102) 0.0060** (0.047) 0.160** (0.054) 0.209*** (0.005) 0.135** (0.029) 1.282** (0.029) 0.093 (0.476) 0.673*** (0.007) 2.076*** (0.017) 8.939*** (0.004) Yes No No Yes No No Yes 632 120 118 1.02 632 120 109 1.10 632 120 118 1.02 0.259 0.518 0.252 0.344 0.259 0.537 0.0004 51.5 0.0109 0.0004 632 120 0.0228 Notes: Dependent variable: growth rate real GDP per capita. Other notes are same as for Table 3. (2) for domestic and transnational terrorism, respectively. Out of 12 coefficients, 11 are statistically significant. The negative effect of domestic terrorist incidents on growth declines as G increases. Interestingly, the partial effect turns positive when G lies between the 50th and 75th percentile levels. The critical value of G that offsets the negative effect of domestic terrorist incidents stands at 45.29 (0.0317 þ 0.0007  G ¼ 0). There are a total of 76 countries 146 GLOBALIZATION, TERRORISM, AND GROWTH Table 6. Marginal effect at different levels of globalization (i.e., @growth/@T ¼ b1 þ b4 (G)) FGLS model SGMM model Level of G Percentile of G Corresponding countries Domestic terrorism (1) Transnational Domestic terrorism terrorism (2) (3) Transnational terrorism (4) 24.46 10th 31.75 25th Bhutan, Sierra Leone Maldives, Mali 40.02 50th Paraguay, Togo 49.63 75th 58.80 90th 41.01 Mean Brazil, Brunei Panama, Bahrain Indonesia, Suriname 0.015*** (0.000) 0.010*** (0.000) 0.004*** (0.006) 0.003 (0.574) 0.009** (0.046) 0.003** (0.019) 0.124*** (0.000) 0.079*** (0.000) 0.029** (0.026) 0.030* (0.080) 0.086*** (0.001) 0.023* (0.086) 0.018*** (0.000) 0.012*** (0.000) 0.005*** (0.000) 0.003*** (0.000) 0.010*** (0.000) 0.004*** (0.000) 0.099*** (0.000) 0.056*** (0.000) 0.006 (0.159) 0.051*** (0.000) 0.106*** (0.000) 0.001 (0.791) 45.29 76 44 44.75 73 47 46.38 81 39 41.08 60 60 Critical value of G No. of countries below critical value No. of countries above critical value Notes: G ¼ globalization index. ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. p-values are in parentheses. that lie below this critical value, whereas 44 countries are above this level in the sample. Similarly, the critical value of G that offsets the negative effect of transnational terrorist incidents on growth stands at 44.75 (0.2730 þ 0.0061  G ¼ 0). A total of 73 countries are below this critical value, whereas 47 countries are above this level. These results suggests that (i) only a significantly high level of openness can offset the negative consequences of terrorism on growth; and (ii) since the majority of countries lie below the critical level of openness, a commitment to reforms for openness can benefit a large number of developing countries in overcoming the negative effects of terrorism. 4.4 Transforming terrorism variables The impact of terrorism and the mitigating role of globalization may be different for a large versus a small country. Thus, I transform our terrorism variables into the number of incidents per one million persons to account for terrorism relative to the country’s population. Bandyopadhyay et al. (2014) argue that this transformation of the terrorism variable provides a better reflection of the terror-threat perception in a country. Analyzing the effect of terrorism on per capita growth rates in Asia over the period 1970–2004, Gaibulloev and Sandler (2009) also used their transnational terrorism measure in terms of incidents per million persons. The results in Table 7 suggest that for the lower estimates, one domestic terrorist incident per one million persons decreases growth by 0.46% (column (2)), while one transnational terrorist incident per one million persons lowers growth by 3.02% (column (5)). Gaibulloev and Sandler (2009) find that an additional transnational terrorist incident per million persons lowers per capita growth by about 1.5%. This difference in the magnitude of the impact of transnational terrorism can result from differences in the terrorism data set sources, number of countries, J. YOUNAS 147 Table 7. Terrorist incidents per million persons and globalization Domestic terrorism Transnational terrorism Estimation technique! Independent variables ; FGLS (1) SGMM (2) SGMM (3) FGLS (4) SGMM (5) SGMM (6) Terrorist incidents p.m.p., t–1 Globalization, t–1 0.644*** (0.002) 0.019 (0.714) 0.0001 (0.928) 0.015*** (0.004) 0.464*** (0.000) 0.129*** (0.001) 0.001*** (0.000) 0.011*** (0.000) 0.464** (0.029) 0.129 (0.355) 0.001 (0.292) 0.011** (0.028) 5.051*** (0.000) 0.022 (0.668) 0.0001 (0.325) 0.121*** (0.000) 3.023*** (0.000) 0.130*** (0.001) 0.001*** (0.000) 0.076*** (0.000) 3.023** (0.028) 0.130 (0.383) 0.001 (0.324) 0.076** (0.026) 0.064*** (0.006) 0.106*** (0.000) 0.066*** (0.005) 1.222* (0.051) 0.040 (0.362) 0.482*** (0.000) 0.560 (0.172) 9.814** (0.018) Yes Yes 0.097*** (0.000) 0.151*** (0.000) 0.135*** (0.000) 1.363*** (0.000) 0.089*** (0.007) 0.716*** (0.000) 2.194*** (0.000) 8.756*** (0.000) Yes 0.097 (0.196) 0.151** (0.014) 0.135*** (0.009) 1.363** (0.023) 0.089 (0.493) 0.716*** (0.001) 2.194*** (0.009) 8.756*** (0.002) Yes 0.072*** (0.004) 0.105*** (0.000) 0.073*** (0.005) 1.229** (0.046) 0.040 (0.326) 0.498*** (0.000) 0.558 (0.166) 11.147*** (0.006) Yes Yes 0.149*** (0.000) 0.180*** (0.000) 0.121*** (0.000) 1.024*** (0.000) 0.111*** (0.000) 0.627*** (0.000) 1.410*** (0.000) 8.110*** (0.000) Yes 0.149** (0.050) 0.180*** (0.002) 0.121** (0.021) 1.024 (0.125) 0.111 (0.275) 0.627*** (0.009) 1.410 (0.120) 8.110** (0.014) Yes No Yes No Yes 632 120 118 1.02 632 120 118 1.02 632 120 118 1.02 632 120 118 1.02 0.315 0.395 0.315 0.411 0.444 0.460 0.444 0.477 (Globalization, t–1)2 Total terrorist incidents p.m.p., t–1  Globalization, t–1 Growth rate real GDP p.c., lagged Investment, t–1 (% of GDP) Govt. consumption, t–1 (% of GDP) Ln (Initial real GDP p.c.) Political rights and civil liberties, t–1 Ln (Inflation), t–1 Ln (School enrollments), t–1 Constant Time effects Country fixed effects Windmeijer standard errors # of observations # of countries, n # of instruments, i countries/instruments ratio, r ¼ n/i Hansen J test Autocorrelation test 632 120 632 120 Notes: Dependent variable: growth rate real GDP per capita. p.m.p. ¼ per one million persons. All other notes are the same as for Table 3. and sample period.9 In all regressions in Table 7, the interaction term strongly conform to the findings that globalization mitigates the negative consequences of terrorism on growth. 9 Gaibulloev and Sandler (2009) use the International Terrorism: Attributes of Terrorist Events (ITERATE) data set of Mickolus et al. (2010), whereas I use GTD data set, which offers information 148 GLOBALIZATION, TERRORISM, AND GROWTH Table 8. Further robustness checks Domestic terrorism Estimation technique SGMM (2) SGMM (3) FGLS (4) SGMM (5) SGMM (6) Panel A: Excl. conflict ridden countries Terrorist incidents, t–1 0.037** (0.016) Terrorist incidents, 0.001** t–1  Globalization, t–1 (0.048) 0.047*** (0.000) 0.001*** (0.000) 0.047** (0.036) 0.001* (0.066) 0.498*** (0.000) 0.012*** (0.000) 0.352*** (0.000) 0.009*** (0.000) 0.352* (0.091) 0.009* (0.069) # of observations 576 countries/instruments ratio Windmeijer standard errors Hansen J test 576 0.94 No 0.585 576 0.94 Yes 0.585 576 576 0.94 No 0.451 576 0.94 Yes 0.451 Panel B: Excl. OPEC countries Terrorist incidents, t–1 0.035*** (0.000) Terrorist incidents, 0.001*** t–1  Globalization, t–1 (0.000) 0.042*** (0.000) 0.001*** (0.000) 0.042** (0.014) 0.001** (0.033) 0.273*** (0.000) 0.006*** (0.000) 0.228*** (0.000) 0.006*** (0.000) 0.228** (0.019) 0.006*** (0.008) # of observations 590 countries/instruments ratio Windmeijer standard errors Hansen J test Autocorrelation test 590 0.93 No 0.622 0.485 590 0.93 Yes 0.622 0.506 590 590 0.93 No 0.597 0.534 590 0.93 Yes 0.597 0.556 0.046*** (0.000) 0.001*** (0.000) 0.046** (0.013) 0.001** (0.026) 0.318*** (0.000) 0.007*** (0.000) 0.452*** (0.000) 0.011*** (0.000) 0.452*** (0.003) 0.011*** (0.002) 556 0.85 No 0.886 0.559 556 0.85 Yes 0.886 0.581 556 556 0.85 No 0.880 0.616 556 0.85 Yes 0.880 0.629 Panel C: Excl. transition economies Terrorist incidents, t–1 Terrorist incidents, t–1  Globalization, t–1 FGLS (1) Transnational terrorism 0.034*** (0.000) 0.001*** (0.000) # of observations 556 countries/instruments ratio Windmeijer standard errors Hansen J test Autocorrelation test Notes: Dependent variable: growth rate real GDP per capita. All regressions include a full set of control variables as in Table 3. All other notes are the same as in Table 3. 4.5 Further robustness checks I conduct a number of further robustness checks. In Tables 8 and 9, I only report the results of the main variables of interest for domestic and transnational terrorism that I derive using FGLS and SGMM. The full set of results is available on request. During the sample period, some countries experienced significantly more terrorist incidents than others. They were also marred with internal civil conflicts. These countries are Angola, Colombia, Ethiopia, Guatemala, India, Indonesia, Philippines, Sri Lanka, and on both domestic and transnational terrorism. Enders et al. (2011) offers a detailed comparison between GTD and ITERATE datasets. J. YOUNAS 149 Table 9. Further robustness checks Domestic terrorism Estimation technique FGLS (1) Transnational terrorism SGMM (2) SGMM (3) FGLS (4) SGMM (5) SGMM (6) 0.028*** (0.000) 0.001*** (0.000) 0.028** (0.030) 0.001* (0.068) 0.212*** (0.000) 0.005*** (0.000) 0.086*** (0.002) 0.003*** (0.000) 0.086 (0.373) 0.003 (0.179) 554 0.94 No 0.661 0.803 554 0.94 Yes 0.661 0.818 554 554 0.94 No 0.557 0.980 554 0.94 Yes 0.557 0.981 0.059* Panel B: Foreign direct investment Terrorist incidents, t–1 0.010*** (0.000) Terrorist incidents, 0.007*** t–1  FDI, t–1 (0.000) 0.009*** (0.000) 0.006*** (0.000) 0.009*** (0.030) 0.006** (0.029) 0.078*** (0.000) 0.040*** (0.000) 0.077*** (0.000) 0.050*** (0.000) 0.077*** (0.010) 0.050*** (0.003) # of observations 609 countries/instruments ratio Windmeijer standard errors Hansen J test Autocorrelation test 609 0.99 No 0.373 0.319 609 0.99 Yes 0.373 0.330 609 609 0.99 No 0.346 0.313 609 0.99 Yes 0.346 0.326 0.011*** (0.001) Terrorist incidents, 0.0001 t–1  Trade openness, t–1 (0.205) 0.028*** (0.000) 0.001*** (0.000) 0.028* (0.070) 0.001* (0.092) 0.104*** (0.000) 0.002*** (0.000) 0.030 (0.324) 0.001 (0.118) 0.030 (0.782) 0.001 (0.658) # of observations 634 countries/instruments ratio Windmeijer standard errors Hansen J test Autocorrelation test Test of joint significance: 0.000*** T&(T  G) 634 1.10 No 0.609 0.214 634 1.10 Yes 0.609 0.238 634 634 1.10 No 0.256 0.297 0.067* 634 1.10 Yes 0.256 0.318 0.839 Panel A: Economic globalization Terrorist incidents, t–1 0.022*** (0.000) Terrorist incidents, 0.001*** t–1  Economic (0.000) Globalization, t–1 # of observations 554 countries/instruments ratio Windmeijer standard errors Hansen J test Autocorrelation test Test of joint significance: T&(T  G) Panel C: Trade openness Terrorist incidents, t–1 Notes: Dependent variable: growth rate real GDP per capita. All regressions include a full set of control variables as in Table 3. All other notes are the same as in Table 3. p-values are reported for the test of joint significance. Sudan. Panel A of Table 8 report results by excluding them from our regressions. Oil Producing and Exporting Countries (OPEC) generate high foreign exchange reserves from the sale of oil in international market, which allows them to expend resources to preempt terrorist threats more effectively. Panel B shows results without including OPEC countries. 150 GLOBALIZATION, TERRORISM, AND GROWTH Transition economies experienced significant economic transition and change in political regime starting early 1990s. Moreover, they faced the least incidents of terrorism over the sample period. Panel C reports results by excluding these economies. All these result support the main findings already presented. The composite globalization index used so far incorporates information on all three indicators of openness—economic, political, and social. I examine these findings by including them separately in the regressions. However, I report the results for economic globalization only. Moreover, I also checked whether the findings are robust to using two individual indicators of openness—net FDI position of a country and trade openness (imports plus exports), both as percentage of GDP.10 FDI is debated as the most prominent feature of globalization, and its growth over the past few decades has far outpaced growth in income and trade (e.g., Pica and Mora, 2011).11 These results are presented in Table 9. With the exception of SGMM regression, where transnational terrorist incidents and its interaction term with trade openness is not statistically significant, the results strongly support that open economies are less influenced by the negative effects of terrorism.12 5. Concluding remarks Should countries aim for greater openness in a bid to counter the damaging effects of terrorism? For reasons outlined in the introduction, the effect of globalization on the economy, at least in the short run, is not clear. Moreover, the effect of globalization on the ease with which terrorist attacks may be planned or executed is also ambiguous, in general. Therefore, I rely on empirical evidence to determine whether globalization helps in containing the damages from acts of terrorism. The robust findings suggest that although all types of terrorism depress growth, globalization dissipates these consequences of terro...
Purchase answer to see full attachment