Download Globalization, Democratization and Terrorism: Impact on Source and Target Countries and more Study notes Market economy in PDF only on Docsity!
The Lexus and the Olive Branch:
Globalization, Democratization and Terrorism
S. Brock Blomberg Claremont McKenna College Gregory D. Hess Claremont McKenna College and CESifo First version: September 2005 This version: November 2005
âThis paper was written for the World Bank Workshop on Security and Development, Washington D.C., May 26-27, 2005. We thank the participants at this workshop for their helpful comments, and Abdoulaye Tall andZamira Simkins for excellent research assistance. Address correspondence to Brock Blomberg, Department of Economics, Claremont McKenna College, Claremont, CA 91711. E-mail: bblomberg@mckenna.edu, tel:(909) 607-2654, fax: (909) 621-8249. The opinions expressed are those of the authors and do not necessarily reflect the views of the World Bank.
The Lexus and the Olive Branch Abstract
This paper provides an original study into how democratization and globalization influence terrorism â examining the motives of terrorists and how democratic institutions and in- ternational integration influence non-state economic actors. We employ a gravity model to investigate the relative importance of globalization and democratization on transnational terrorism. We construct an original database of over 200,000 observations from 1968- for 179 countries, to examine the extent to which economic, political and historical factors influence the likelihood of citizenâs from one country to engage in terrorist activities against another. We find that the advent of democratic institutions, high income and more open- ness in a source country significantly reduces terrorism. However, the advent of these same positive developments in targeted countries actually increases terrorism. Ceteris paribus, the impact of being a democracy or participating in the WTO for a source country decreases the number of terrorist strikes by about 2 to 3 per year, which is more than two standard deviations greater than the average number of strikes between any two countries in a given year.
JEL Codes: E6, H1, H5, D74, O Keywords: Growth, Conflict, Terrorism, Globalization, Democratization
leads to a fall in trade and, in turn, a decline in aggregate economic activity. Put differently, an increase in terrorism in country A increases the cost to doing business with country A so that country B will either purchase goods or services domestically or from another more peaceful country. Thus, violence acts as a distorting tax or tariff that limits the attainment of the benefits from free trade. Anderson and Marcouiller (2002) have pursued this angle employing corruption and imperfect contract enforcement as impediments to international trade. They find that omit- ting indexes of institutional quality obscures the negative relationship between per capita income and the share of total expenditure devoted to traded goods. Their paper, however, does not consider direct measures of conflict.^1 Blomberg and Hess (2005a) calculate that, for a given country year, the presence of terrorism, as well as internal and external con- flict is equivalent to as much as a 30 percent tariff on trade. This is larger than estimated tariff-equivalent costs of border and language barriers and tariff-equivalent reduction through GSPs and WTO participation. In a complementary study, Glick and Taylor (2004) consider the direct effect of very large external wars on trade from a broader historical perspective. To estimate the quantitative implications of violence and globalization on interna- tional investment, Blomberg and Mody (2005) use a gravity model of bilateral FDI flows. Three findings emerge from Blomberg and Modyâs (2005) analysis. First, violence at home tends to move investment abroad. Second, violence in the host country deters both trade and FDI flows. Host country violence hurts inflows of investment with particular force in developing countries. And, third, they find a strong positive impact of WTO membership on bilateral FDI flows. Their results, therefore suggest that while violence raises political (^1) Nitsch and Schumacher (2004) also analyze some aspects of conflictâs impact on trade but over a signif- icantly shorter time horizon.
risk and discourages investment flows, WTO membership acts as a commitment device that, by limiting the possibility of arbitrary policy changes, lowers country risk. These results are robust to a variety of specifications checks. While these papers provide important evidence on violenceâs impact on globaliza- tion, they fail to consider the opposite effect â namely globalizationâs impact on terrorism. Moreover, they also do not formally examine the impact of democratization. The central contribution of our paper is to do just that. There are other papers that do examine globalizationâs and democratizationâs role in terrorism. Li and Schaub (2004) employ a sample of 112 countries from 1975 to 1997 and find that neither trade nor investment has a positive impact on terrorism. Li (2005) uses the same data to analyze the impact of democracy on terrorism. He finds that democracy can reduce terrorism. There are serious limitations to this line of research. In particular, by using standard panel estimation in the analysis, these papers are unable to separate globalization or de- mocratizationâs impact on terrorism from the host and the source country perspective. For example, suppose increased economic integration has the consequence of harming individ- uals in import competing industries. Further, suppose these individuals join forces with a terrorist organization and express their displeasure through a terrorist attack on its trading partner. This attack on the host country from a neighboring source country will not be appropriately taken into account. In fact, to truly understand the impetus for any transna- tional event, one must understand the motivation from the point of view of both the host of the attack and the source of the attack. The standard treatment of the data, is unable to get at this crucial issue.
the characteristics of a country that might make it a likely target country may indeed be very different from the characteristics that make a country a likely source of international terrorism. The features of the polity that make a country a terrorist-producer may be dif- ferent from the political structures, institutions and environment that make a state terrorist target. To analyze the importance of both democratization and globalization in determining terrorism, we embed the analysis in the workhorse model of trade and finance â the gravity model. The gravity model is also useful because it allows researchers to examine the net flow of activity between countries while ignoring the domestic terrorist activities. This is useful in our case in that there is no comprehensive data set that includes country level measures of domestic terrorism over a long time horizon. In its simplest form, a gravity model postulates that bilateral activity, usually trade or investment, is positively related to the size of the two economies and negatively influenced by the distance between them. We extend this analysis, by considering terrorism as the bilateral activity between each country- year-pair. Besides including the size and distance variables in basic gravity equations, our baseline specification includes other control variables commonly used. Importantly, they rely on estimates that include bilateral country-pair dummies, which control not only for distance but for all unobserved common relationships between the countries. The purpose of estimating a gravity equation for terrorism is to estimate the im- portance of democratization and globalization on terrorism and to compare these relative magnitudes to other factors previously highlighted as relevant in explaining terrorism e.g. GDP or GDP per capita. In this way, we begin with a baseline terrorism model in which development is the main engine in determining terrorism. Then we add globalization and
democratization to separately ferret out the significance of each. In addition, we add new variables and consider specifications suggested by recent advances in the interpretation of gravity models.^2 Our central hypotheses are these:
- H1: The effects of democracy and globalization on terrorism differ for source and target countries
- H2: Terrorism falls with democracy and globalization in the source countries
- H3: Terrorism rises with democracy and globalization in the target countries. We find that differences in income, democracy and openness go a long way into explaining transnational terrorism. We find the advent of democratic institutions in a source country significantly reduces terrorism. However, the advent of these same institutions in host countries actually increases terrorism, providing more support for our above conjecture. We also find that source-country openness has a negative and statistically significant impact on terrorism. Once again, however, host-country openness often has a positive and statistically significant on terrorism. Ceteris paribus, the impact of being a democracy or participating in the WTO for a source country decreases the number of terrorist strikes by about 2 to 3, which is more than two standard deviations greater than the average number of strikes between any two countries in a given year. (^2) For examples in the trade literature, see among others, Anderson (1979) who championed use of the gravity equation in structural trade models. Blomberg and Hess (2005a) focus on trade, especially on com-paring the costs of conflict to measures for trade promotion. Alternatively, Blomberg, Hess and Orphanides (2004) investigate the impact of various forms of conflict such as terrorism, internal wars and external warson a countryâs economic growth.
country, h, from attackers whose nationality comes from source country s. In addition, we measure T as the number of victims rather than number of incidents in a given year. We present several caveats before we proceed. First, one may be concerned that the nationality of the source attacker may not represent the views of the country for which he is associated. While a possibility, this problem is no less severe than what we encounter when we try to measure any international variable â e.g., how do we properly account for the nation of origin of a Mercedes-Benz manufactured in Alabama using parts imports from Asia, for example? Second, one may be concerned that there could be more than one nationality included in the attacking force, making it hard to determine the source country of the terrorist incident. This concern turns out to be less of an issue in practice for the following reason: 98 percent of attacks are reported with only one source country.^3 Finally, one may be concerned that we could be under-counting the number of incidents as not all attacks are identified with a particular group. Even so, the vast majority of attacks do have an identified source country, amounting to over 8,000 incidents. It is also likely that incidents that are reported without association to any particular group are not successful terrorist attacks and should be ignored. As the definition of terrorism in ITERATE requires knowledge of a political agenda, the events without associated countries are likely to have any direct impact on the relationship between any two countries in particular.
2.1 Globalization, Democratization and Terrorism
As shown in our companion piece, Blomberg and Hess (2005b), rich countries have had about four times as many incidents and incidents per capita as poor countries and democracies (^3) Experimenting with different classifications for source country had no discernable impact on the results. Hence, we did not include source countries for multiple country attacks.
(globalizers) have had about four times as many incidents per capita as compared to non- democratic regimes (non-globalizers). Why might this be so? Krug and Reinmoeller (2004) argue that globalization is an important determinant of terrorism. In their paper, they build a model to explain the internationalization of terrorism as a natural response to a globalizing economies. As countries become more economically integrated and market-oriented, there is no discrimination between what certain terrorist groups might see as âbadâ products and âgoodâ products or investments. Moreover, the same advances in technology that allow for easy access of goods and services also allow for easy access to military hardware and technology. In the short run, globalization may have the consequence of creating a series of winners and losers. These same losers will have easier access to retaliate in response to their loses thereby multiplying the affect of globalization on terrorism. An alternative view put forth by Crenshaw (2001) is that it is naive to believe that globalization is encouraging international terrorism. So that while globalization and ter- rorism may be seemingly impacting one another, there is something more complicated at work. The latest incidence of terrorism is not necessarily driven by globalization. Instead, the latest wave of terrorism should be seen as a series of civil wars which may be motivated by a strategically unified reaction to American power, rather than directly to globalization. It is an empirical matter to determine which hypothesis best supported by the exist- ing evidence. Tables 1 - 2 report the total number of terrorist incidents and incidents per capita parsed by globalization, democratization and development.^4 If globalization, democ- (^4) Our definitions for high or low globalization, democratization and development are standard measures. High (low) growth is defined as average growth per capitafor countries with polity > (<)7 and/or eiec+liec> (<)14; High (low) globalized is defined as countries with > (<)1.5 percent; High (low) democratic is defined trade as a percentage of GDPcutoff values. > (<)30 percent. The general qualitative results are not sensitive to different
Tables A1-A5 provide three interesting facts. First, terrorists are more likely to originate from countries in and around the Middle East and certain Western Europe countries than from other regions around the world. Iran, Iraq, Israel, Jordan, Kuwait, Lebanon, Liberia, Oman, and Yemen are each responsible for more than the average number of incidence per capita. European countries such as Cyprus, Greece, and Ireland have some of the highest incidence of both terrorism and terrorism per capita. Second, there is little correlation between measures of globalization, democracy, development and terrorism as these high incidence source countries are not particularly democratic/non-democratic, developed/developing or open/closed. For example, for the 12 countries with high levels of terrorism mentioned above: 6 have higher than average scores for democracy and 6 have lower than average scores; 6 have higher than average incomes and 6 have lower than average incomes; 6 have higher than average openness and 6 have lower than average values. In summary, to best assess the impact of globalization, democratization and develop- ment on terrorism, researchers much not only account for the changes in these variables but must also account for the relative size of these variables. Hence, a bi-lateral model may best describe the economic motives of terrorist groups. This can be seen in the gravity model described in the following section.
3 The Gravity Model
For several decades, the gravity model has been the workhorse of empirical trade research and more recently empirical FDI literature. One reason is that the model is relatively intuitive. The gravity equation simply states that there is a positive relationship between
trade/financial flows and the sizes of countries and a negative relationship between trade/financial flows and distance. A central contribution of our paper is to introduce terrorism T as the dependent variable into these various gravity models. To include T in the aforementioned approaches, consider the following gravity equation for log trade (xhst) for country pair h, s at time t and its determinants: xhst = f (Yhst, Zhst, phst) (1)
where y is log of real GDP per capita, Y is log of real GDP, Z is a vector of observables to include trade costs Ï (e.g. distance and language barriers), and p are multilateral resistance terms such as prices that refer to the bilateral barrier between countries relative to the average trade barrier each country faces with all trading partners.^5 These multilateral resistance terms may be thought of as product price variables that may create wedges to trade. For traditional trade gravity models, one representation of equation (1) is:
xhst = α 0 + α 1 yht + α 2 yst + α 3 Yht + α 4 Yst + ΎZhst + Δhst (2)
where Z is a vector of variables including distance (both physical and technological mea- sures), income per capita, etc. and language barriers and the error may be specified to control for random or time/country fixed effects. We modify equation (2) by specifying Z and redefining the left-hand-side variable as T, so that we have:
Thst = α 0 +α 1 ·yht+α 2 ·yst+α 3 ·Yht+α 4 ·Yst+α 5 ·logdistancehs+α 6 ·+Comlanghs+α 7 ·areahs (3) (^5) For notational convenience, we have written the country pair as yhs, but we switch to yh and ys to refer to an individual country.
To correctly estimate the elasticities, then, it is necessary to consider the bias on account of censoring. We employ the Tobit model that estimates the coefficients through a maximum likelihood procedure.
4 Empirical Results
4.1 Cross-Country Empirical Motivation
We motivate our discussion by considering the link between terrorism that occurs within a countyâs borders from outsiders and terrorism that occurs by the citizens of a country in other countries. Crudely speaking, what we are doing is examining bilateral terrorist net imports. The purpose of this preliminary exercise is to see if the same countries that experience significant international terrorism are those whose citizens are terrorizing other countries abroad. This is useful because it may shed some light into some of the causes of terror: whether terrorism is driven by civil strife between countries who may have been given arbitrary borders by colonial powers; whether terrorism is linked to particular countries such as the United States that may have very strong or polarizing international policies; whether terrorism is due to globalization/democratization/development such that those countries are more apt to be net importers of terrorism rather than net exporters. Figure 2A plots countries by the number of terrorist exports versus the number of terrorist imports and a line of best fit. If countries import and export the same amount of terrorism, we would expect the data to fall along the 45 degree line. In fact, the line of best fit is measured at 43 degreesâin line with such a hypothesis. This result merely demonstrates an identityâin equilibrium, net exports and net imports must be equated. However, there are several important differences. First, there are notable net im-
porters of terrorismâthey include Israel, the United States, France and Great Britain. There are also several notable net exporters of terrorismâIreland, Iran, and Cuba. While there may be many factors that shift countries away from the diagonal line, it is interesting to note that the net importers mentioned are clearly more democratic and developed than the net exporters. We denote the least democratic/developed/open countries with dots. Most appear to be net exporters of terrorism. Hence, when developing our gravity model, it would appear that the traditional variables included in gravity models would also apply to terrorismânamely income, trade and institutions. This can be seen once we do the same experiment controlling for democracy, openness and income. In this case, there does not appear to be such a difference in estimated imports or exports from terrorism. Figure 2B plots this conditional regression. Notice that in this case there are just as many dots below and above the estimated line. Interestingly, it is still estimated to be a 45 degree line. While these figures may be illuminating, they do not provide any direct evidence re- garding the relationship between globalization, democratization and terrorism. The purpose of the next subsection is to directly address these issues.
4.2 Baseline Results
We begin by explaining the results from estimating the gravity model, (3). In Table 4, columns 1-7 include variables that do not change over time. These include distance, land mass, as well as dummy variables for language. Column 8 estimates the model to include controls for time. Column 9 estimates the model to control for random effects by country- pair. Each of these models are estimated using the Tobit estimator with standard errors clustered by the income level of each country-pair.
have become more globalized and democratized, some countries have been âleft behindâ while others have flourished. Perhaps, terrorists in these âleft behindâ economies has chosen to strike against those countries that have become more advantaged during the period in question. We directly address this point as we consider the effect of these dynamic forcesâ globalization and democratizationâon terrorism. There are two main results from this estimation. First, the advent of democratic institutions in a source country significantly reduces terrorism. However, the advent of these same institutions in host countries actually increases terrorism, providing more support for our above conjecture. Second, source-country openness has a negative and statistically significant impact on terrorism. Once again, however, host-country openness often has a positive and statistically significant on terrorism. Ceteris paribus, the impact of being a democracy or participating in the WTO for a source country decreases the number of terrorist strikes by about 1 to 2, which is more than two standard deviations greater than the average number of strikes between any two countries in a given year. How should one interpret our findings? Do the forces of modernization (democracy, globalization, growth) lead terrorists to attack other countries, or do terrorists from poor countries attack rich neighbors because itâs a low cost method of voicing their discontent? The effects of democracy on a countryâs likelihood of being a source for transna- tional terrorism are not firmly established. Non-democracies create fewer outlets for polit- ical grievances to be addressed, making violent means of political action more likely. This might lead to increased domestic terrorism, but doesnât speak to the country as a source of transnational terrorism. When the autocratic government is perceived to have its authority
bolstered by its foreign relations with democracies however, we might expect the terrorist group advocating the removal of the illegitimate autocrat may indeed target its foreign allies, some of whom may be democracies. We might expect therefore that non-democracy abroad could increase transnational terrorism at home. As to what makes a country a source of terrorists, there is little evidence of any kind. Discussion in this regard has rarely distinguished between domestic and transnational terrorism. Where political conflict is domestic, the lack of outlets for political discontent make violent means of protest more likely. Where a wider variety of groups get to participate in the political process, non violent means are at least attempted first. Others have argued that in a more democratic regime more political action of all kinds, violent and non-violent alike, is likely. Overall, the lack of clarity on the issue stems, in our view, from treating the source and target countries in the same manner; when the effects of democracy are permitted to differ conditional on whether the observation is a source or target, allows a more precise view on the determinants of transnational terrorism. Globalization also affects the costs, benefits and resources available for terrorist activ- ities. Firstly if terrorism emerges from a sense of relative deprivation, then globalization, in so far that it encourages economic growth, may mitigate terrorist tendencies. On the other hand, if globalization is associated with increased inequality across countries and groups, then we might expect globalization to lead to more violence. On the costs side of the equa- tion, the lowered barriers to flows of goods, money, people and ideas, makes the networks of terrorist operations cheaper to operate. Terrorist themselves find it easier to move across increasingly permeable borders; resource flows across borders necessary to finance terrorist
