Terrorism and counterterrorism: An overview

This introduction sets the stage for the articles collected in this special issue of Oxford Economic Papers. It begins by introducing essential concepts including domestic terrorism, transnational terrorism, defensive actions, proactive countermeasures, and guerrilla warfare. Three terrorist event databases, used by seven of the articles, are briefly introduced. These data sets are then used to display some stylized facts about domestic and transnational terrorism during the past four decades. Next, some essential strategic distinctions are drawn between defensive and proactive measures in the case of transnational terrorism when multiple countries are confronted by a common terrorist group. These strategic concerns vanish for domestic terrorism as a central government is able to internalize potential externalities. Finally, the key findings of the articles in the special issue are highlighted in two tables.


Introduction
Terrorism is the premeditated use or threat to use violence by individuals or subnational groups to obtain a political or social objective through the intimidation of a large audience beyond that of the immediate noncombatant victims (Enders and Sandler, 2012, p.4). The two essential ingredients of terrorism are its violence and its political or social motive. Terrorists tend to employ shockingly violent acts, such as beheadings, downing of commercial airlines, bombings in public markets, and armed attacks in public places, to intimidate an audience. Their unpredictable and horrific attacks are meant to make everyone feel at risk even though the true likelihood of falling victim to a terrorist incident is rather minuscule, roughly equivalent to that of drowning in one's bathtub (Mueller, 2006). Terrorists seek to circumvent normal channels for political change by traumatizing the public with brutal acts so that governments feel compelled to either address terrorist demands or divert public funds into hardening potential targets. Terrorist campaigns are more prevalent in liberal democracies, where the government's legitimacy hinges on its ability to protect the lives and property of its citizens (Eubank and Weinberg, 1994). of the eight articles in this issue, are briefly presented in Section 3. In Section 4, two of these data sets are used to display some recent trends and aspects of domestic and transnational terrorism during the last four decades. Essential concepts of counterterrorism are then presented in Section 5, where proactive measures are distinguished from defensive actions. Key findings of the four terrorism and four counterterrorism articles contained in this special issue are highlighted in two summarizing tables in Section 6. Concluding remarks follow in Section 7.

On terrorism
I now return to the definition of terrorism, given at the outset of this article. Any definition of terrorism involves much debate (Hoffman, 2006;Enders and Sandler, 2012). The research community is converging to a consensus based on an operational definition on which to construct event data sets to test theoretical propositions. The article puts forward a definition that is consistent with that used by the main event data sets and relied on by researchers. Also, this definition possesses the main ingredients that are agreed on by economists, political scientists, and political economists.
The three stakeholders in this definition are the perpetrators, the victims, and the audience. By limiting terrorism to subnational agents including individuals or a 'lone wolf', my definition rules out state terror in which a government terrorizes its own people. The definition, however, does not rule out state-sponsored terrorism where a government clandestinely assists a terrorist group through various means, including supplying weapons, safe haven, intelligence, training, funding, or safe passage (Mickolus, 1989;Bapat, 2006). There was a lot of state sponsorship of terrorism during the final decade of the Cold War with groups such as the Abu Nidal Organization serving as a terrorist group for hire (Hoffman, 2006). 2 The most controversial element of my definition is the victim, since some definitions exclude combatants, so that attacks against an occupying army, such as US forces in Iraq or Afghanistan, are not viewed as terrorism. Generally, an attack against peacekeepers, such as the 23 October 1983 suicide bombing of the US Marines barracks at Beirut International Airport, is considered an act of terrorism. The barracks' bombing had the political objective of removing peacekeepers from Lebanon, which happened in February 1984. Attacks against US soldiers and their dependents stationed in Germany constitute terrorist incidents, because these targeted individuals were noncombatants when attacked. 'Audience' refers to the collective that terrorists seek to intimidate through their wanton brutality. With sufficient and sustained intimidation, the audience will apply pressures on the besieged government to concede to the terrorist group's political demands or alternatively to take decisive action to annihilate the group. 3 In the latter case, the Italian authorities dismantled the Italian Red Brigades in the 1980s.  (Enders and Sandler, 2012). This group is credited with the first simultaneous terrorist attacks. 3 On alternative ways that terrorist groups end, see Carter (2012), Gailbulloev and Sandler (2014), and Phillips (2014a).
There are some crucial distinctions to draw between terrorism and related concepts. For instance, there is the distinction between terrorism and crime. A kidnapping for ransom is a criminal act of extortion when the kidnappers are not pursuing or financing a political agenda. If a political motive is tied to the kidnapping, then it is a terrorist incident even with ransom demands being made. The hijacking of a commercial airliner by a deranged person is a crime but not terrorism. In the absence of a political motive, an armed attack by a student on fellow students or teachers is a criminal action. Next consider an insurrection, which 'is a politically based uprising intended to overthrow the established system of governance and to bring about a redistribution of income' (Sandler and Hartley, 1995, p. 307). Leaders of insurrections recruit from the peasantry and general population in the hopes of challenging the government's hold on power (Grossman, 1991). Successful rebel operations can generate new recruits and may ideally cause the government to impose draconian measures on its citizens, which subsequently create more support for the insurgency. If a tipping point is attained, then the government may be sufficiently challenged to lose its power to the rebels.
In distinction to insurrections, guerrilla warfare generally involves a band of rebel forces (e.g., the Revolutionary Armed Forces of Colombia [FARC], Abu Sayyaf in the Philippines, or Shining Path in Peru) that controls a sector of the country, from which to dispatch its operatives to confront government forces. Some guerrilla wars take place in urban centers. In contrast to most terrorist groups, guerrilla groups are larger in number and organized like a military force. Some guerrilla groups engage in terrorist acts, such as the three justmentioned groups, to raise funds to secure their operations and pursue their political aims. For example, FARC kidnaps government officials and others for ransoms. Unlike an insurrection, guerrilla groups are not bent on overthrowing the government or engaging in propaganda to gain popular support (Hoffman, 2006). Shining Path and FARC apply threats and harsh measures to gain the compliance of the people in the territory that they control. Guerrilla groups rely on surprise and cover to harass numerically superior government forces. Terrorism is a tactic employed by both insurrections and guerrilla movements. As a consequence, many guerrilla groups are listed as terrorist groups despite their control of territory. Often, countries with jungle cover or mountainous terrain provide remote areas where guerrillas can conduct training and operations. In this special issue, Carter (2015) is interested in the interaction between a guerrilla group and the government, as the former chooses between terrorism and the control of territory and the latter chooses between defensive counterterrorism actions and proactive military responses to influence the group's decision.

Domestic versus transnational terrorism
Domestic terrorism is homegrown and home-directed, and represents the most common form of terrorism. For domestic acts of terrorism, the perpetrators, victims, and audience hail from the venue country, where the attack takes place. Domestic terrorist incidents include Timothy McVeigh's bombing of the Alfred P. Murrah Federal Building in Oklahoma City on 19 April 1995 or Eric Rudolph's anti-abortionist bombing of Centennial Olympic Park in Atlanta, Georgia, on 27 July 1996. Civil wars often involve numerous domestic terrorist attacks before and during the conflict by the adversaries (Findley and Young, 2012). 4 These terrorist acts are more apt to be domestic when an intervention by a third party from 4 On the relationship between civil wars and terrorism, see Sambanis (2008). outside the country is not involved. Boko Haram's kidnapping of more than 200 female students is a domestic terrorist incident, which involves victims and perpetrators from the venue country of Nigeria. Boko Haram is an Islamic jihadist terrorist group that controls territory in the northeast portion of Nigeria. Given the country's limited military capabilities, its government sought some assistance from the USA in terms of military advisors and intelligence in addressing the significant threat that Boko Haram poses. At times, Boko Haram crosses into Chad. If one or more of the schoolgirls are moved into a neighboring country, then the kidnapping becomes a transnational terrorist incident. In general, poor countries may request foreign assistance if they cannot properly confront an indigenous terrorist group that may attack at home or abroad (Azam and Thelen, 2010;Fleck and Kilby, 2010;Bandyopadhyay et al., , 2014Young and Findley, 2011).
Terrorism is transnational when an incident in the venue country concerns perpetrators or victims from another country. If a terrorist attack in the UK is perpetrated by terrorists from Yemen, then the incident is one of transnational terrorism. When a terrorist attack in France harms Dutch citizens, the attack is transnational. If one or more victims or perpetrators are not citizens of the venue country, then the terrorist attack is transnational. The kidnapping in January 2002 and subsequent murder of US reporter Daniel Pearl in Pakistan is classified as a transnational terrorist incident. The same is true of the near-simultaneous bombings of the US embassies in Kenya and Tanzania on 7 August 1998. Terrorist attacks against another country's embassy, even when perpetrated by citizens of the venue country, are transnational terrorist events because an embassy's grounds represent foreign soil. Similarly, terrorist attacks against international organizations' personnel or property are considered to be transnational terrorist acts. An important transnational terrorist incident is the August 2006 plot to use liquid explosives to blow up 10 or more transatlantic flights departing the UK for the USA and Canada. A skyjacking originating in one country that is diverted to another country for political purposes is a transnational terrorist event. If a politically motivated hijacked plane has citizens from more than one country, the event is transnational terrorism even if the flight is domestic and ends in the country of origin. On 9/11, the four skyjackings are transnational terrorist acts since the victims were citizens from upward of 80 nations and the perpetrators were foreigners. The kidnapping of US journalist James Foley in Syria on 22 November 2012 and his beheading on 19 August 2014 is a transnational terrorist act. The victim was American, whereas the murderer is allegedly a British citizen from the Islamic State in Iraq and Syria (ISIS), also known as Islamic State (IS).
Transnational terrorist incidents frequently imply transnational externalities-for example, perpetrators from one country impose uncompensated costs on the victims of another country. If a country provides safe haven to a transnational terrorist group that attacks other countries' interests, then transnational externalities ensue. 5 The Taliban in Afghanistan had given safe haven to Osama bin Laden's al-Qaida, which planned and executed the events of 9/11. When the Taliban would not surrender bin Laden to the USA following 9/11, the USA led an invasion of Afghanistan on 7 October 2001 (Enders and Sandler, 2012). In this extreme case, the transnational externality resulted in a military invasion with the intent to defeat the Taliban and al-Qaida. Transnational externalities also arise from counterterrorism policies of targeted countries, which result in inefficient 5 How terrorist groups choose their home base is discussed in this special issue by Gaibulloev (2015). On the definition of terrorist groups, see Phillips (2014b).
T. SANDLER levels of these policies (Sandler and Lapan, 1988;Sandler and Siqueira, 2006;Bueno de Mesquita, 2007). Actions by one targeted country to secure its borders and ports of entry may merely transfer the attack abroad, where borders are more porous (see Section 5).
Since 9/11, few transnational terrorist incidents occur on US soil but 35% to 40% of such incidents involve US people or property in other countries (Enders andSandler, 2006, 2012). Hoffman (2006, pp.63-5) traces the modern era of transnational terrorism to the 22 July 1968 hijacking of an Israeli El Al flight en route from Rome to Athens by three armed members of the Popular Front for the Liberation of Palestine (PFLP) terrorist group. This skyjacking was motivated by the intention of the PFLP terrorist to trade its hostages for Palestinian prisoners held by Israel. This event is noteworthy for a number of reasons. First, through its protracted 40-day negotiations, the Israelis were forced to negotiate with the Palestinian terrorists, which the Israelis had hitherto vowed they would never do (Hoffman, 1998, p.68). Second, the media coverage demonstrated to terrorists worldwide that such incidents could garner worldwide attention for their cause. Not surprisingly, transnational terrorist attacks increased greatly in numbers during the years following this incident (see the figures in Section 4). Third, there was evidence of state sponsorship after the diverted plane landed in Algiers as Algerian forces secured the hostages and held some Israeli hostages until 1 September 1968 when a deal was concluded (Mickolus, 1980, pp.93-4). Fourth, Israel eventually traded 16 Arab prisoners from the 1967 Arab-Israeli War for the remaining Israeli hostages. This trade showed terrorists that hostage taking could yield significant concessions. Transnational terrorist groups were primarily nationalists/separatists or leftists (socialists) during the late 1960s until the late 1980s (Rapoport, 2004). Even the Palestinian terrorists were secular until the end of the 1980s with the rise of Hamas, Palestinian Islamic Jihad, and other groups. After the mid-1990s, the religious fundamentalists came to dominate and increased the carnage (Enders and Sandler, 2000;Gaibulloev and Sandler, 2014). The phenomenon of religious-based transnational terrorism is not novel and can be traced back to the Sicarii or Zealots, a Jewish sect that conducted a terror campaign against the Romans and their Jewish collaborators in Judea from CE 48 to 73 (Rapoport, 1984;Bloom, 2005). Sicarii terrorists engaged in daytime assassinations in public places that typically resulted in the death of the assassin. As such, their dagger attacks were an early form of suicide terrorism, since the perpetrator had little chance of escape. 6 From 1090 to 1256, the Islamic Assassins opposed Sunni rule in Persia and Syria, with the intent to set up their own community and state of believers in the region (Bloom, 2005). Although the Assassins' terrorist campaign was on a much smaller scale, their goal was similar to that of ISIS. Like the Sicarii, the Assassins relied on politically motivated assassinations, performed with a dagger. Perpetrators usually sacrificed their own lives by making no efforts to escape after the deed.

Some historical considerations of transnational terrorism
6 In this special issue, Choi (2015) investigates suicide terrorism, which may be domestic or transnational in nature. Suicide terrorist attacks grew in numbers after 1988 (Santifort-Jordan and Sandler, 2014).

Event data sets
In the beginning of the 1980s, the first terrorist event data set-International Terrorism: Attributes of Terrorist Events (ITERATE)-was made available to researchers. ITERATE only includes transnational terrorist attacks. Coverage starts in 1968, the beginning of the modern era of transnational terrorism, and runs until the end of 2012, with annual updates in August (Mickolus et al., 2013). ITERATE codes many variables-for example, incident date, country start location, country end location, attack type, target entity, terrorist group, perpetrators' nationalities, number of deaths, number of injuries, victims' nationalities, logistical outcome, US victims, state sponsorship, and scene of attack-in its Common File of over 40 variables. In addition, there is a Fate File indicating the fate of the terrorists-for example, the number of terrorists captured, the number of terrorists sentenced, and their length of incarceration. There is also a Hostage File, which has invaluable observations used by researchers to analyze logistical and negotiation success of hostage taking (Santifort and Sandler, 2013). If an attack is completed as planned, then it is a logistical success. For hostage missions, securing one or more hostages is deemed a logistical success. The Hostage File of ITERATE is currently updated through 2010. Finally, there is a Skyjacking file with additional observations, and variables on skyjacking missions such as the duration of the incidents, airline involved, and negotiation strategies used. ITERATE, like the other event data sets, relies on the news media-print, broadcast, and digital-for the observations of its variables. An initial focus of empirical studies was on transnational terrorism because ITERATE was the most extensive data set available throughout the 1980s and 1990s. Its lengthy series of daily data were ideal for time-series studies, which dominated the research landscape except for a few survival studies, the first being Atkinson et al.'s (1987) study of the duration of hostage-taking incidents. Today, panel studies are prevalent including those in this special issue- Berrebi and Ostwald (2015), Choi (2015), Egger and Gassebner (2015), Gries et al. (2015), and Younas (2015).
Another competing event data set, modeled after ITERATE, is the RAND (2012) data set, which currently codes incidents for 1968-2009 and is not being updated. Gaibulloev (2015) uses the RAND event data in conjunction with Jones and Libicki's (2008) classification of terrorist groups' ideologies in his study of groups' location decisions. For 1968-97, RAND event data only include transnational terrorist attacks; after 1998, RAND data distinguish between transnational and domestic terrorist attacks in a manner consistent with my early definitions. Compared to ITERATE, RAND data code fewer variables and, for transnational terrorist incidents, have more limited coverage than ITERATE as demonstrated by Enders (2007).
The third event data set, germane to this special issue, is that of the Global Terrorism Database (GTD), which records both domestic and transnational terrorist incidents (La Free and Dugan, 2007; National Consortium for the Study of Terrorism and Response to Terrorism, 2013). Although GTD recorded both kinds of incidents, until 2013 it did not distinguish between the two kinds of incidents. Enders et al. (2011) devised a five-step procedure for distinguishing between domestic and transnational terrorist incidents in GTD for 1970-2007 and made their breakdown available to researchers. This division is now applied to 2008-2012 (see Enders et al., 2014). A breakdown of terrorism into its two components is essential because the two types of terrorism may affect economic variables and counterterrorism differently-for example, economic growth or foreign direct T. SANDLER investment is more influenced by transnational terrorism (Gaibulloev and Sandler, 2008). Moreover, the influence of other variables on domestic and transnational terrorism may differ (Sandler, 2014;Choi, 2015). The Enders et al. (2011) procedure does a much better job in distinguishing between the two types of events than recent GTD efforts, based on the authors' method without attribution. GTD has tens of thousands of unclassified incidents compared to Enders et al. (2011).
There are some things to note about GTD. First, it has changed its coding conventions a few times, most recently for the 2012 data. Coding was also changed after the National Consortium for the Study of Terrorism and Responses to Terrorism (2013), based at the University of Maryland, took charge of the data around 2005. As shown by Enders et al. (2011), there are periods of undercounting and overcounting of incidents, which can be addressed by these authors' calibration methods. Second, GTD data for 1993 are very incomplete because in an office move by Pinkerton, which originated the database, the box containing 1993 fell off of the truck! Third, perpetrators' nationalities are not identified. Fourth, GTD does not contain any hostage negotiation variables, making the study of hostage-taking incidents impossible with this database. Fifth, GTD coverage of some kinds of domestic incidents, such as kidnappings, is virtually nonexistent prior to the late 1990s (Enders et al., 2011). Table 1 indicates key empirical or theoretical aspects of the terrorism and counterterrorism articles contained in this special issue. For each study, the table identifies the type of terrorism studied, the method employed, the unit of analysis used (if applicable), the terrorism data source applied, and the years and countries covered. As shown, there are two theoretical studies that address aspects of counterterrorism. Currently, there are no global counterterrorism data sets available with a wide range of counterterror variables. ITERATE has some counterterror variables, such as negotiation responses of the authorities. Specific countries-for example, Israel-maintain data sets of their counterterrorism responses; these data sets are difficult to acquire.
A variety of empirical techniques are applied by the six empirical articles. In his cross-sectional study of the location choice of terrorist groups, Gaibulloev (2015) relies on a conditional logit estimator. To account for the count nature of the dependent terrorism variable, Choi (2015) and Gries et al. (2015) apply a negative binomial panel estimator. Endogeneity between the dependent variable and one or more independent variables is addressed in various ways: the Berrebi and Ostwald (2015) article employs instrumental variables (IV), consisting of lagged domestic terrorist attacks in neighboring countries, whereas the Younas (2015) article uses system generalized method of moments (SGMM) to augment his feasible generalized least squares (FGLS) estimates. The Gries et al. (2015) article addresses endogeneity with lagged values of the independent variables and also SGMM estimates. The Choi (2015) article primarily uses lagged independent variables to partly handle the endogeneity concern.
4. Domestic and transnational terrorism: some data plots  T. SANDLER interplay of the terrorists and the authorities (Faria, 2003;Enders and Sandler, 2012). Transnational terrorism attained its highest annual totals during 1973-93. The end of the era of state sponsorship of terrorism in the 1980s and the end of the Cold War in 1991 roughly coincided with the precipitous drop in transnational terrorism. With the fall of communism, many left-wing terrorist groups stopped operations or were defeated by the authorities. These left-wing groups were replaced by more ruthless and bloodthirsty religious fundamentalist terrorist groups that went for more 'bang for the buck' with fewer incidents containing greater casualties (deaths or injuries) (Enders and Sandler, 2000;Hoffman, 2006;Gaibulloev and Sandler, 2014). The dashed series in Fig. 1 measures annual casualties on the right-hand vertical axis for transnational terrorism. These casualties are generally higher after 1999 despite the huge fall in incidents. Sandler (2014) shows that the proportion of transnational terrorist incidents with casualties was much greater after 1990 and the rising dominance of religious fundamentalist terrorists. Figure 2 employs GTD data to indicate the annual number of total, domestic, and transnational terrorist incidents. Note that the domestic terrorism series lends its shape to the total terrorism series insofar as domestic terrorism constitutes the overwhelming component of total terrorism. The huge drop in 1993 in all three series is due to the lost box of data and does not denote a respite or vacation on the part of the terrorists. The steep rise in 2012 corresponds to a GTD change in coding methods. Also, Enders et al. (2011) demonstrate that there was an undercounting of events during much of the 1970s as GTD trained coders. Discontinuities in GTD coding pose significant concerns and challenges for time series, cross-sectional, and panel studies. In a recent study, Enders et al. (2014) broke their analysis into two periods-1970-92 and 1994-2010-to address the lost box of 1993 data and calibrated the data to handle coding inconsistencies. Fortunately, these two periods roughly corresponds to left-wing and religious fundamentalist terrorist dominance, so that the split is conceptually interesting.
In Fig. 3, the transnational terrorism series involving US interests is displayed annually for 1968 to 2012, based on ITERATE data. Peaks and troughs are evident. Annually, about  35% to 40% of transnational terrorist incidents are against US interests even though relatively few such incidents occurred on US soil during the past two decades. This is a clear indication of the transference of attacks from secure US borders to foreign venues, which constitutes a transnational externality. The declining series in Fig. 3 is consistent with the declining transnational terrorism series in Fig. 1 because a fairly constant proportion of such incidents are against US interests. In this special issue, the Gries et al. (2015) article investigates some drivers of these anti-US attacks abroad. Figure 4 returns to the issue of casualties and depicts the annual number of casualties per terrorist attack for transnational and domestic terrorism using ITERATE and GTD data, respectively. Since 1999 and the overwhelming dominance of the religious fundamentalist terrorist groups, the casualties per attack is generally higher for transnational terrorism. Thus, on a per attack basis, transnational terrorist attacks are more worrisome than domestic terrorist attacks. However, because domestic terrorist attacks far outnumber transnational terrorist attacks each year, there are more annual casualties with domestic than with transnational terrorism. This is also bolstered by a higher proportion of domestic terrorist incidents, compared to transnational terrorist attacks, that results in casualties (Sandler, 2014).

Counterterrorism
This special issue contains articles on terrorism and counterterrorism. Given my discussion of terrorism and its concepts, it is now instructive to introduce some basic concepts and concerns that are associated with the practice of counterterrorism.
Quite simply, counterterrorism corresponds to actions to ameliorate the threat and consequences of terrorism. These actions can be taken by governments, military alliances, Defensive countermeasures protect potential targets by making attacks more costly for terrorists or reducing their likelihood of success. When, however, successful terrorist attacks ensue, defensive actions also serve to limit the resulting losses to the target. Defensive measures have generally been reactive, instituted after some successful or innovative terrorist attacks. In the USA, airline passengers are now required to remove their shoes when being screened, following the innovative, but fortunately unsuccessful, attempt by Richard Reid to bring down American Airlines flight 63 en route from Paris to Miami on 22 December 2001 with explosives hidden in his shoes. Before the installation of metal detectors to screen passengers at US airports on 5 January 1973, there were on average over 25 skyjackings each year in the USA (Enders et al., 1990). After their installation, attempted US skyjackings dropped to fewer than four a year. The success of these metal detectors in US airports led to their installation worldwide over the next six months. Following the downing of Pan Am flight 103 over Lockerbie, Scotland, on 21 December 1988 and the downing of UTA flight 772 over Niger on 19 September 1989, bombdetecting devices were used to screen checked luggage. Defensive or protective counterterror actions may involve more than technological barriers. Other instances of defensive measures include target hardening, such as defensive perimeters around government buildings or embassies, or guards at key points of a country's infrastructure. Defensive measures can also take the form of issuing terrorism alerts, enacting stiffer penalties for terrorism offenses, enhancing first-responder capabilities, and stockpiling antibiotics and antidotes for biological and chemical terrorist attacks. This list of defensive actions is by no means exhaustive.
By contrast, proactive measures are offensive as a targeted government directly confronts the terrorist group or its supporters. Proactive measures may destroy terrorists' resources (e.g., training camps), curb their finances, eliminate their safe havens, or kill and capture their members. In recent years, the Obama administration has relied on drone attacks to assassinate terrorist leaders and operatives. Proactive operations may assume myriad other forms, including a retaliatory raid against a state sponsor that provides resources, training, sanctuary, logistical support, or intelligence to a terrorist group. On 15 April 1986, the USA launched a retaliatory bombing raid on targets in Libya for its alleged support in the terrorist bombing of the La Belle discotheque in West Berlin on 4 April 1986, where 3 died and 231 were wounded, including 62 Americans (Mickolus et al., 1989, vol. 2, pp.365-7). Another proactive measure takes the form of a preemptive attack against a terrorist group or a harboring country, such as the US-led invasion of Afghanistan four weeks after 9/11. A preemptive strike differs from a retaliatory raid because the former is more sustained and meant to severely compromise the capabilities of the terrorists to conduct future missions. Such strikes or raids concern transnational terrorism where a targeted country confronts the foreign threat. Other proactive measures include infiltrating terrorist groups, engaging in military action, conducting propaganda campaigns against the terrorists, and gathering intelligence to foil terror plots (Kaplan, 2015). 'Military action' generally refers to operations by the host government against a resident terrorist group as in Carter (2015). Actions that improve the economy, which in turn reduces grievances, can also be proactive (Choi, 2015). Younas (2015) demonstrates empirically that increased globalization of a country's economy may also be proactive by limiting harmful and therefore grievance-causing effects of terrorism on economic growth. In this special issue, the four counterterrorism articles primary address aspects of proactive measures. Only Carter (2015) considers both defensive and proactive counterterrorism responses.
The game-theoretic literature on counterterrorism draws some fascinating strategic contrasts between defensive and proactive countermeasures. Suppose that two or more countries are confronted by the same Islamic jihadist terrorist group-for example, al-Qaida in the Arabian Peninsula (AQAP). In this transnational terrorism scenario, each at-risk country is inclined to work at cross-purposes by engaging in a defensive race in the hopes of transferring the terrorist attacks to other targeted countries (Sandler and Lapan, 1988;Arce and Sandler, 2005). In the process, the countries engage in too much defense since the negative transference externality is not internalized. The only check on this adverse 'defense race' stems from the countries having large interests abroad, which may be hit when attacks are transferred abroad-recall Fig. 3 and attacks against US interests .
Next consider proactive measures in these multicountry scenarios. Any country's actions to confront the common terrorist threat confer purely public (nonrival and nonexcludable) benefits to all potential target countries. As a result, there is too little proactive response as each country tries to free ride on the actions of other countries. Thus, defensive measures are strategic complements as one country's actions encourage those of other countries (i.e., reaction paths are upward-sloping), whereas proactive measures are strategic substitutes as one country's actions inhibit those of other countries (i.e., reaction paths are downward-sloping) (Eaton, 2004;Sandler and Siqueira, 2006). Moreover, leadership can be shown to curb the defensive race, whereas leadership exacerbates the free-riding underprovision of offensive measures (Sandler and Siqueira, 2006). Thus, there is no simple fix for these concerns among sovereign targeted nations.
For domestic terrorism, countries possess the proper incentives to choose defensive and proactive measures judiciously, because all associated costs and benefits are internalized (Enders and Sandler, 2012). Thus, Kaplan's (2015) finding that the USA staffs about the right number of intelligence analysts to intercept terror plots is consistent with past game-theoretic findings regarding domestic counterterrorism optimality. Carter's (2015) article is also geared to domestic terrorism, whereas Choi (2015) and Younas (2015) may involve economic-based countermeasures for either type of terrorism.

Findings of the articles of the special issue
Tables 2 and 3 summarize some of the most important findings of the articles of this special issue. Table 2 does so for the terrorism articles, and Table 3 does so for the counterterrorism articles. Every article is rich in results, so those listed in the tables are not exhaustive.
This special issue contains many noteworthy advances in the study of terrorism. Gaibulloev (2015) is the first article to address terrorist groups' base location choice; currently, the study of terrorist group survival is an active area of research. Berrebi and Ostwald (2015) is the initial study to examine the impact of terrorism on fertility. Younas (2015) and Choi (2015) refine the relationship between terrorism and growth; Egger and Gassebner (2015) show that terrorism has less of an effect on trade than conventionally believed. Kaplan (2015) investigates the social efficiency of intelligence staffing in intercepting terror plots. Carter's (2015) analysis is novel because the terrorist group must decide between terrorist attacks and holding territory in reaction to the government's countermeasures. The interface between terrorism and other forms of political violence is seldom

Concluding remarks
Terrorism remains an active research area in economics, political economy, and political science. Given the microeconomic and macroeconomic consequences of terrorism, its study is a fitting topic for a general economics journal such as Oxford Economic Papers. Following 9/11, industrial countries engaged in a huge reallocation of resources toward counterterrorism, whose efficacy can be best studied and understood with the theoretical and empirical tools drawn from economics.
The rise of ISIS and the considerable threat that its Western fighters pose for their home countries means that terrorism will remain a policy concern. AQAP's war on the USA • Terrorist groups are attracted to base their operations in countries with other terrorist groups that possess a similar ideology. This is particularly true for left-wing terrorist groups.
• Risk of state failure and/or political instability is conducive for terrorist groups to establish their home base.
• Generally, terrorist groups locate closer to where they conduct their attacks abroad. However, the distance consideration is nonlinear.
• Cooperation is needed between host and venue countries to address a common terrorist threat.

Egger and Gassebner
• There is little or no immediate or short-run effect of transnational terrorism on international trade when monthly data are used.
• Transnational terrorism impacts bilateral and unilateral trade, if at all, in the medium term, about one and a half years after incidents.
• The analysis intends to remove measurement errors when terrorism is aggregated over a year or else general equilibrium effects are ignored.
• Authorities should look beyond trade as to where to counter the adverse economic effects of terrorism.

Berrebi and Ostwald
• Terrorist attacks reduce both total fertility and crude birth rates owing to stress and intervening factors.
• The author's findings account for endogeneity with an instrumental variable accounting for lagged domestic terrorist attacks in neighboring countries.
• The findings are robust over myriad model specifications, accounting for fixed effects and first differences.
• This article shows that terrorism can have consequences beyond economic considerations. Gries et al.
• US aid-recipient countries' characteristics (e.g., human rights violations) can affect or motivate US-directed attacks.
• A positive association between anti-American terrorism and US-associated economic and military aid is somewhat attenuated by local oppression in the client (recipient) state.
• There is no evidence that US assistance curbs anti-American attacks even in oppressive regimes.
• These results hold when accounting for potential endogeneity between US aid and anti-American terrorist attacks.

T. SANDLER
and its allies is yet another terrorist threat. Just as Africa appears to be achieving sustained economic growth in select countries, significant terrorist challenges arise from Boko Haram and al-Shabaab in Nigeria and Kenya, respectively. Failed states-for example, Afghanistan, Iraq, Somalia, Syria, and Yemen-supply safe havens for terrorist groups that threaten Western interests. It is my hope that this special issue on terrorism and counterterrorism will stimulate further research in these two topics.

Acknowledgements
The present collection arises from a set of papers presented at the sixth conference on Terrorism and Counterterrorism Policy at the University of Texas, Dallas. Held between 21 May and 24 May 2014, this conference invited many leading contributors to the study of terrorism and counterterrorism. Past conferences resulted in special issues published in Economics & Politics While assuming full responsibility for any remaining shortcomings, this article has profited from comments from Anindya Banerjee, who also advised me on the special issue. Khusrav Gaibulloev Table 3. Key findings of counterterrorism studies Kaplan • Assuming a proportional hazard function, the article computes how many agents are necessary to detect an acceptable percentage of terrorist plots.
• Terrorists select their optimal number of plots given the socially optimum staff. Thus, the terrorists move first, followed by the government's staffing choice.
• Based on event data for the USA, the article computes the socially efficient number of detections.
• A numerical example shows that the USA is close to a socially efficient optimum based on calculated benefits and costs.

Carter
• State first allocates between defensive and proactive countermeasures, followed by the group's choice between terrorism and controlling territories in guerrilla operations.
• Under weak assumptions, the more capable governments confront more terrorism as they employ more boots on the ground to protect against guerrilla warfare.
• When it is optimal for a group to employ both terrorist incidents and guerrilla attacks, the state generally focuses on limiting the group's capture of territory in guerrilla operations.
• Corner solutions with the government protecting against guerrilla war are likely, which can fuel more serious terrorist attacks. Younas • Globalization lessens the negative impact of total, domestic, and transnational terrorism on economic growth in developing countries.
• The author finds, when relevant, the critical value where globalization completely offsets the harmful growth effects of terrorism.
• The analysis is robust to the use of disaggregate measures of openness.
• Globalization can be a counterterrorism tool.

Choi
• High industrial economic growth reduces domestic and transnational terrorist incidents. • High agricultural economic growth has little or no impact on domestic and transnational terrorist incidents.
• High industrial economic growth can, however, be associated with more suicide terrorism, which seeks out hard targets.
• Generally, high industrial economic growth is favorable to reducing terrorism, especially since few countries suffer suicide terrorism.

16
TERRORISM AND COUNTERTERRORISM Downloaded from https://academic.oup.com/oep/article-abstract/67/1/1/2913015 by guest on 30 July 2018 read and commented on an earlier version. Finally, I thank all of this special issue's reviewers, most of whom are major contributors to the study of terrorism. These reviewers provided rigorous reviews in a timely fashion. All articles went through a careful and demanding prescreening process, followed by two rounds of reviews by two to three anonymous referees. I appreciate the understanding of those authors whose papers did not make it into the special issue.

Funding
The Vibhooti Shukla Endowment and the Center for Global Collective Action at the University of Texas at Dallas funded the May 2014 Conference on Terrorism and Counterterrorism Policy, where these articles along with others were initially presented.