Does transnational terrorism stimulate foreign assistance?

Abstract

This paper quantifies how past transnational terrorist attacks against a potential donor's assets result in enhanced foreign aid flows to a country hosting the responsible terrorist group. Given the reversed causality between foreign aid and terrorism, our empirical analysis puts forward an instrumental variable. Both conflict and governance assistance are shown to stem from transnational terrorist incidents involving recipient–donor dyads during 1974–2013 for a global sample. For recipient-related terrorism, lagged transnational terrorist events against a donor's assets display a robust positive impact on conflict and governance aid. Placebo or falsification tests support the exogeneity of the instrument.

Keywords

transnational terrorism conflict and governance foreign aid instrument war on terror falsification test

Introduction

Ever since the four US skyjackings on 11 September 2001 (henceforth, 9/11), researchers have studied the allocation of foreign aid to countries hosting terrorist groups as a means to curb transnational terrorism¹ (see, e.g. Azam and Delacroix, 2006; Azam and Thelen, 2008, 2010; Bandyopadhyay et al., 2011; Bapat, 2011; Savun and Tirone, 2018; Young and Findley, 2011). In some articles, the perspective is on how foreign aid to recipient countries affects the survival of resident terrorist groups that target the interests of donor countries (Bapat, 2011, 2019; Boutton, 2019; Kim et al., 2021; Kim and Sandler, 2021). Contrary to a donor's intentions, the resident terrorist group's survival may be enhanced as the recipient nation responds to a moral hazard dilemma because eliminating the resident group, as required, will lose the host country its aid (Bapat, 2011). Other articles investigate whether or not terrorism induces greater foreign aid from at-risk countries threatened by resident terrorist groups in potential aid-recipient countries (Boutton and Carter, 2014; Dreher and Fuchs, 2011; Fleck and Kilby, 2010). Foreign aid as a counterterrorism tool implies a two-way causality: namely, foreign aid can affect the level of transnational terrorism, and transnational terrorist attacks can influence the amount of foreign aid. This endogeneity concern is not always adequately addressed in the literature.

A main purpose of the current study is to apply an instrument to re-examine the extent to which transnational terrorist incidents perpetrated by foreign terrorists stimulate a targeted country to provide foreign aid to the country hosting the perpetrators. To accomplish this task, we investigate transnational terrorist attacks associated with recipient–donor dyads during 1974–2013. In so doing, we have a global sample of donor countries, unlike some previous studies (e.g. Fleck and Kilby, 2010). For instance, such recipient–donor dyads consist of Afghanistan and the United States, Pakistan and the United States, Yemen and the United States, and Morocco and Spain. A transnational terrorist incident relevant to the first dyad was a Taliban suicide terrorist attack on 30 December 2009, which killed and injured American civilians in the Khost Province of Afghanistan. For the second dyad, a germane transnational terrorist event was the kidnapping and subsequent murder of Daniel Pearl, an American journalist for the Wall Street Journal, in Pakistan in early 2002. An incident associated with the third dyad is the plot by al-Qaida in the Arabian Peninsula in Yemen to send printer-cartridge bombs aboard two US cargo planes with the intent of them exploding over US cities. The terrorist plots in late October 2010 were foiled when the planes were searched at intermediate stops in the UK and Dubai, based on an informant. The fourth dyad corresponds to the 11 March 2004 commuter train bombings during the morning in Madrid, whose victims included Spaniards and foreign nationals, some of whom were from donor countries to Morocco. In response to transnational terrorist attacks, our study is interested in increases in two forms of foreign aid to recipient host countries—conflict aid and government and civil society aid (henceforth, governance aid). Conflict aid is primarily concerned with encouraging conflict resolution, including reduced terrorism in the recipient country. Governance aid is intended to bolster governing structures, civil liberties, political rights and civilian protections in the recipient country. Enhanced government institutions can stabilize the recipient country in light of internal challenges, while greater civil liberties and political rights can reduce the grievances that are often behind terrorism (Abadie, 2006; Gaibulloev et al., 2017).

The paper contains a number of findings. For recipient-based terrorists, the empirical analysis demonstrates that lagged transnational terrorist incidents against a donor country's assets induce a robust positive effect on the amount of conflict aid for the terrorist-hosting country. Additionally, lagged transnational terrorist attacks directed at a donor's interests encourage more governance aid to the country hosting the terrorist perpetrators. In short, recipient–donor dyads are associated with terrorism-induced increased conflict and governance aid. Both positive relationships account for standard controls. Robustness tests support our findings when transnational terrorist attacks are only tied to the recipient venue country rather than the perpetrators’ nationalities in the aid-recipient country. Other robust tests account for dyadic alliances, recipients’ civil war and other controls. Furthermore, placebo or falsification tests support the use of our instrument based on the total number of domestic terrorist attacks in countries adjacent to the aid-recipient country.

Literature review

The literature on foreign aid and terrorism falls into three camps. One camp focuses on how terrorism, especially transnational terrorism, increases foreign aid flows. A second camp looks at the reversed causality in terms of how foreign aid effectively limits terrorism, and a third camp investigates how foreign aid influences the survival of resident terrorist groups that attack donor countries’ interests.

The pioneering article on how 9/11 affected foreign assistance solely concentrates on US aid. By considering pre- and post-9/11 US aid, Fleck and Kilby (2010) show that US aid increased greatly after 9/11, but this increase did not address the core or poverty needs of recipients, leading the authors to conclude that US aid was redirected to bolster the US-led War on Terror. Their conjecture is supported circumstantially by the discovery that recipients’ share of GDP devoted to military expenditure was an important positive indicator of US bilateral aid (Fleck and Kilby, 2010: 187–188). However, their study does not address the endogeneity of foreign aid and terrorism. In a key follow-up study, Dreher and Fuchs (2011) analyze the bilateral-aid behavior of 22 donor and 140 recipient countries in response to donor-directed transnational terror attacks during the War on Terror. Although bilateral aid rose during the post-9/11 era, Dreher and Fuchs (2011) find that foreign assistance did not respond to the number of terrorist events. Based on a Heckman selection model, they show that countries from which terrorists originated were not more inclined to receive foreign aid; however, those terrorist-hosting countries that received assistance got more aid after 9/11. To address endogeneity, Dreher and Fuchs (2011) use the system generalized method of moments rather than an instrument, as we employ. Like our study, they examine recipient–donor dyads, but unlike our study, they do not uncover a significant relationship between transnational terrorism and governance aid. Also, in terms of terrorism-induced foreign aid, Boutton and Carter (2014) consider US aid going to countries serving as a venue for terrorist attacks against US interests during 1976–2006. Those authors find that US military and economic assistance was more apt to be given to venue countries from which resident terrorists attack US assets. Moreover, the amount of US aid was positively related to the level of anti-American terrorism. Their study does not address the likely endogeneity concern between aid and terrorism beyond lagging independent variables.

A second set of terrorism-aid articles examine the reversed relationship: the influence of foreign aid as a means for limiting transnational terrorist attacks. In contrast to the current study, most articles on aid and terrorism take this vantage point. In an early empirical analysis, Azam and Delacroix (2006) show that foreign aid may increase, rather than decrease, the number of terrorist attacks in the aid-recipient venue country. They trace this positive relationship to aid going to venue countries with the most difficult terrorism problems, suggesting that the identified positive relationship is not really capturing the effect of aid on home terrorist attacks. Their pioneering study presents a game-theoretic analysis of the multi-stage interaction between the donor and recipient countries, and the subsequent response of the resident terrorist groups to these countries’ counterterrorism measures. Azam and Delacroix’s (2006) analysis is also important for highlighting the endogeneity between foreign aid and terrorism. A subsequent empirical piece by Azam and Thelen (2010) demonstrates that tied foreign aid reduces transnational terrorist incidents, consistent with the theoretical model, as the actions of resident terrorists in the recipient nation are inhibited. Contrary to general foreign aid and especially education aid, Azam and Thelen (2010) find that US military intervention increases transnational terrorist events originating from the recipient country, thereby favoring aid over direct military actions.² In the spirit of Azam and co-authors, Bandyopadhyay et al. (2011, 2014) formulate a three-stage, game-theoretic model of the efficacy of foreign aid in controlling terrorism. However, their model acknowledges potential backlash as participants in the recipient country react negatively to a perceived foreign intervention embodied by tied counterterrorism aid. The backlash can create regime instability and terrorist attacks, not associated with general aid.

Young and Findley (2011) study the effectiveness of foreign aid against recipient-based transnational terrorism when the aid is broken down by sectors. They demonstrate that aid supporting education, health, civil society and conflict prevention curtailed transnational terrorism coming from the recipient country. However, aid given to improve the capacity of the recipient's government institutions – so-called governance aid – did not reduce the number of transnational terrorist incidents coming from the aid recipient. In contrast, Savun and Tirone (2018) demonstrate that governance or civil society aid dampened the effectiveness of resident terrorist groups and limited domestic terrorist attacks in countries not embroiled in civil conflicts. To address aid endogeneity, they employ a two-period lag.

A third line of empirical investigations concerns resident terrorist groups’ survival in response to foreign aid given by targeted developed countries to host developing countries. Generally, such studies establish that this aid increased the longevity of resident terrorist groups as the recipient countries respond to perverse incentives that do not favor the elimination of the resident terrorist groups (Bapat, 2011; Kim et al., 2021; Kim and Sandler, 2021). The donors may be inclined to overlook the recipients’ inaction against the resident terrorists because the aid strengthens the recipient governments sufficiently not to succumb to the terrorists nor to reach an accommodation with them (Bapat, 2011). This is yet another instance where the ability of foreign aid to annihilate terrorist groups abroad is by no means assured. Given such mixed findings, the ability of transnational terrorism to encourage foreign aid flows to developing countries hosting terrorist groups remains quite topical.

Theoretical considerations

In the wake of 9/11, the War on Terror induced targeted developed countries to employ foreign aid to reduce terrorist attacks against their interests, coming from foreign-based terrorist groups. If targeted developed countries can use aid to shift more proactive measures abroad to emasculate a resident foreign terrorist threat, then the developed country can save on its own defensive and proactive measures against the terrorist threat (Bandyopadhyay and Sandler, 2011, 2021; Bandyopadhyay et al., 2011, 2014).³ Targeted developed countries are expected to raise their foreign aid going to countries harboring terrorists as terrorist attacks against these developed countries’ people or property increase at home or abroad. The aid is a substitute for direct military intervention by the donor countries, which studies show raised transnational terrorism contrary to intentions (e.g. Azam and Thelen, 2008, 2010; Danzell et al., 2019; Piazza and Choi, 2018). In this scenario, the targeted developed countries are the donors, while the developing countries with resident terrorists or terrorist groups are the aid recipients. The terrorists and victims’ nationalities associated with each incident allow for the identification of recipient countries and recipient–donor dyads.

In particular, we focus on transnational terrorist attacks as stimulating conflict and governance aid flowing to developing countries hosting terrorists, who engage in attacks against donor countries’ assets. Such attacks motivate conflict aid as means to protect against donor-directed terrorism and its root causes. Conflict aid addresses more than terrorism by promoting conflict resolution, supporting enhanced security, facilitating peacebuilding efforts, bolstering post-conflict peace, reintegrating soldiers and rebels into domestic life, reducing small-arms caches and clearing land mines (AidData, 2017). Such actions are intended, in part, to limit the grievances and resources of foreign-based terrorist groups. By promoting a more stable environment, conflict aid reduces acts of violence that can ignite new cycles of politically motivated killings and retribution that spill overs to potential donors’ interests. Ideally, conflict aid limits conflict-related externalities that adversely affect the donors’ interests. Such externalities concern refugee flows, reduced trade, foreign direct investment (FDI) losses and impeded resource supplies that may destabilize recipient countries and create grievances, which may erupt into terrorist attacks. In the form of money, technical support or intelligence, conflict aid can be used to augment host-country defenses, thereby providing the recipient with an enhanced ability to keep its resident terrorists in check (Kim and Sandler, 2021). Moreover, conflict aid can inhibit the host country from reaching an accommodation with resident terrorists (Bapat, 2011). In sum, we have:

Hypothesis 1: Donor countries provide conflict aid to countries whose resident terrorists attack donors’ assets at home or abroad. Moreover, a greater number of donor-targeted terrorist attacks increases conflict aid.

Next, we turn to how transnational terrorist attacks, stemming from developing countries’ terrorists, induce targeted donor countries to raise their governance aid as a counterterrorism measure. Governance aid strengthens the recipient government's institutions, administrative capacity, public-sector financing, law enforcement and judicial system. Additionally, governance assistance bolsters civil liberties, rule of law and physical integrity rights in the recipient country (AidData, 2017). Better governmental institutions and administrative abilities in recipients reduce corruption and allow for the better utilization of foreign aid, which makes the giving of aid more attractive to donor countries. Enhanced public-sector financing enables the host developing government to raise funds for its military expenditure and counterterrorism actions to confront the resident terrorists, thus motivating the donors’ largesse. An improved judicial system in the host-developing country fosters its ability to ensure that arrested resident terrorists face the consequences of their attacks.

The terrorism literature links the protection of civil and physical integrity rights to addressing political grievances, which, if not addressed, increase domestic and transnational terrorism (see, e.g. Abadie, 2006; Piazza, 2017; Savun and Tirone, 2018; Walsh and Piazza, 2010; Young and Findley, 2011). Increased transnational terrorism against donors’ assets motivates donors to provide more governance aid to address the drivers of this terrorism. Gaibulloev et al. (2017) provide empirical support that greater civil and political rights curb domestic and transnational terrorism, while Choi (2010) demonstrates that improved rule of law curtails domestic and transnational terrorism. An increase in transnational terrorist attacks may lead targeted donor countries to provide governance aid as a way to provide terrorism-alleviating rights, thereby ratcheting down terrorism that harms donors’ assets. Thus, we have the second hypothesis:

Hypothesis 2: Donor countries provide governance aid to countries whose resident terrorists attack donors’ assets at home or abroad. Moreover, a greater number of donor-targeted terrorist attacks increases governance aid.

We are concerned with transnational terrorist attacks where the victims are from the donor countries and the perpetrators are from the recipient countries so that at least two nationalities are involved. When testing Hypotheses 1 and 2, we investigate recipient–donor dyads for which logged conflict or governance aid is regressed in period t on logged and lagged transnational terrorist incidents, directed at the donor's interests and perpetrated by recipient-country terrorists. For these hypotheses, the terrorist incident may occur in recipient i, donor j, or a third country, but all three instances involve a perpetrator from recipient i.

We are also interested in a narrower hypothesis that only relates to those terrorist attacks on donors’ assets that occur in the recipient country and not elsewhere. As before, conflict and governance aid for venue recipient countries’ incidents are motivated by the same charitable considerations as that for attacks against donors’ interests in donor, recipient, or third-country venues, e.g. to limit grievances and to bolster institutional integrity. The third related compound hypothesis is:

Hypothesis 3: Donor countries provide conflict and governance aid to venue recipient countries, whose resident terrorists attack donors’ assets. Moreover, a greater number of donor-targeted terrorist attacks increases these two forms of aid.

For the first two hypotheses, regression tables are presented in the Results section. Regression tables supporting the third hypothesis are provided in the Online Appendix but are carefully discussed in the Results section. In both cases, the dependent variable is either logged conflict aid or logged governance aid. Since each form of aid may affect the underlying causes of transnational terrorism differently, each is analyzed separately. The key independent variable is logged and lagged transnational terrorist incidents based on the nationalities of the terrorists for the first two hypotheses. However, for the third hypothesis, the key independent variable is logged and lagged transnational terrorist incidents taking place in the venue recipient countries. In both cases, the terrorist attacks are against the donor countries’ interests for each recipient–donor dyad. Given that up to three perpetrators and victims’ nationalities may be identified with an incident, multiple recipient–donor dyads may apply for a single attack.

Empirical methods

Because we concentrate on terrorist incidents tied to recipient–donor dyads, we are naturally drawn to transnational terrorist attacks as our primary independent variable. As such, those attacks concern donor assets harmed by perpetrators from the recipient nation. To link recipient and donor dyads, we must ideally know the nationalities of terrorists that target the donors’ people or property. Unlike the Global Terrorism Database (GTD) (National Consortium for the Study of Terrorism and Responses to Terrorism, 2019), the International Terrorism: Attributes of Terrorist Events (ITERATE) dataset provides up to three nationalities of terrorists (when relevant and known) for each transnational terrorist attack (Mickolus et al., 2021). For the sample events during 1974–2013, ITERATE identifies one or more perpetrators’ nationalities in 95% of the cases (i.e. 11,429 of 12,077 incidents). Thus, we rely on ITERATE for the main independent variable – logged and lagged transnational terrorist attacks by perpetrators from the recipient country against donors’ interests.

To estimate the impact of transnational terrorism involving recipient–donor dyads on conflict or governance foreign aid, we employ ordinary least squares (OLS) regressions with donor-country fixed effects.⁴ The models are specified as:

\ln (aid)_{i j t} = α \ln (transnational terrorism)_{i j t - 1} + x^{'} β + ϕ countr y_{j} + e_{i j t}

(1)

where

\ln (aid)_{i j t}

denotes logged conflict or governance aid in recipient country i, given by donor country j at time t, and

\ln (transnational terrorism)_{i j t - 1}

represents the logged and lagged transnational terrorist incidents involving dyad ij at time t – 1.

Countr y_{j}

denotes donor-country fixed effects, which control for any time-invariant, donor-specific factors.

x

is a vector of control variables, while

α

β

, and

ϕ

are corresponding coefficients.

In this line of research, endogeneity is a key concern owing to reversed causality between foreign aid and terrorist attacks, $\ln (transnational terrorism)_{i j t - 1}$ . “While military aid may affect the likelihood of a terrorist group's collapse, lengthy existence of terrorist groups may prompt a donor country to provide more conflict aid to a recipient country” (Kim et al., 2021: 5). That is, conflict aid impacts the length of terrorist campaigns and, thus, the number of terrorist attacks (also, see Azam and Thelen, 2008, 2010; Savun and Tirone, 2018; Young and Findley, 2011). Alternatively, the number of transnational terrorist attacks influences targeted countries’ contribution of foreign aid to countries harboring terrorists during the War on Terror (Dreher and Fuchs, 2011; Fleck and Kilby, 2010). Hence, our main independent variable may be correlated with the error term, which results in a potential endogeneity problem. To address the issue, we use a two-stage least squares method with an instrumental variable (IV). In the first stage, we estimate a linear regression of transnational terrorist attacks on the IV and all of the exogenous variables. From that regression, we get the predicted value of the transnational terrorist incidents, which is then used as the independent variable in the second stage.

To obtain consistent parameter estimates, the IV must satisfy two conditions: (a) relevance (inclusion) that the IV is correlated with the independent variable; and (b) exogeneity (exclusion) that the IV is uncorrelated with the error term. To abide by these two assumptions, we use the lagged number of domestic terrorist incidents (drawn from GTD) in all countries contiguous (adjacent) to the recipient country as our IV (see, e.g. Berrebi and Ostwald, 2013, 2015). First, our IV of domestic terrorism in adjacent countries to the recipient is probably related to the incidence of transnational terrorism in the recipient country. Previous studies show that domestic terrorism in neighboring countries may spill over or raise transnational terrorism in contiguous neighbors (Berrebi and Ostwald, 2013; Enders et al., 2011). In particular, Enders et al. (2011) show that domestic terrorism Granger-caused transnational terrorism in other countries but not the other way around. This was true for a host of different terrorism series involving all incidents, casualty incidents, and specific types of attacks. Logically, potential terrorism spillover is anticipated to be the greatest between adjacent countries, where neighboring domestic incidents have a better chance to affect terrorism abroad. The hypothesized correlation can come from neighboring terrorist groups sharing tactics, operatives, propaganda, and intelligence with neighboring groups (Berrebi and Ostwald, 2013, 2015). Neighboring domestic attacks can have a demonstration effect on terrorists operating in next-door countries. In the reported first-stage regressions in the Online Appendix, domestic terrorism in adjacent countries is a significant determinant of our lagged dyadic measure of transnational terrorism.

Second, the proposed IV should satisfy the exogeneity conditions that the number of domestic terrorist incidents in adjacent countries is not a determinant of the governance or conflict aid given by donor countries to the recipient countries. When donor countries decide to supply aid, they generally focus on economic, security, institutional, civil liberties, and political rights in the recipient countries, not in their contiguous neighbors. Domestic terrorism in adjacent countries should not be a factor affecting the governance/civil society or the need for governance aid in politically distinct neighbors. Civil liberties, rule of law, political rights, and physical integrity rights in the aid recipient are not anticipated to be influenced by nearby domestic terrorist attacks, which by definition do not involve the recipient's people or property. Moreover, neighboring domestic terrorism is not anticipated to affect adjacent recipients’ conflict instabilities and, thus, the need for conflict assistance. Domestic terrorism involves perpetrators and victims from the venue neighboring countries, which isolates such events from myriad conflict considerations in adjacent aid recipients. Neighboring domestic terrorism does not decrease the political stability nor the vulnerability of adjacent countries. Simply put, domestic terrorism in a contiguous country is not expected to influence governance or conflict aid in neighboring countries. To test the inclusion, we use the F-statistics on the IV in the first stage and report them in the empirical tables, where large F-statistics are indicative of a strong IV (Stock and Yogo, 2005). Exclusion is always subject to doubt so we must go further.

As well as the IV models, we also employ falsification or placebo tests to support the robustness of our results (Berrebi and Ostwald, 2015). This test implies that future transnational terrorist attacks against the donor's interest associated with terrorists’ nationalities, or the venue recipient country, should not influence past aid. This test relates transnational terrorist attacks in period t + 2 (future) to past conflict or governance aid in period t. The idea of this exercise is that if our assumed causal direction is correct, then our “false” setup of including future transnational terrorist attacks for each recipient–donor dyad should seldom reveal any statistically significant and positive effects on past allocation of conflict or governance aid. Following Berrebi and Ostwald (2015), we continue to include our main independent variable, lagged transnational terrorism, to capture the hypothesized relationship.

Next, we consider the likely influence of some controls that can affect the amount of conflict and governance aid for recipient–donor dyads. The geographical distance between recipient and donor countries’ capitals may negatively affect both forms of aid. The hypothesized negative relationship follows because the potential externalities coming from recipient countries’ conflicts or from their weak governing institutions dilute with space. For example, refugee inflows, spillover skirmishes, resource disruptions, and other conflict-related externalities are less noticeable or disruptive when the recipient is further from the donor. This is also true for externalities arising from corruption and poor administration in recipient countries. As those and other externalities dissipate, the need for conflict and governance aid also lessens.

Political affinity between recipient and donor dyads may affect various aid flows. For instance, Bapat (2011) views greater foreign-policy affinity between a recipient and donor as more conducive to conflict aid when the United States is the donor. A similar argument is made by Kim et al. (2021) for other major donors, i.e. France, Germany, and the UK. If, however, the donor is willing to base its aid amount on keeping the recipient from reaching an accommodation with the resident terrorists, then reduced affinity may lead to more aid because an accommodation with resident terrorists reached by a recipient with a much different foreign policy agenda than the donor has more dire consequences. As such, the donor is then more willing to overlook foreign policy differences when giving aid. We rely on the Ideal Points Distance measure of Bailey et al. (2017), which quantifies the political proximity of two countries based on a single indicator of the two countries’ overall foreign policy agenda during each sample year. The sought-after distance is the absolute difference of the two countries’ ideal positions, where a smaller (larger) distance indicates more (less) similar foreign policy stances. A zero distance implies that the two countries share the same stance.

Another control involves an interactive term between geographical distance and ideal points distance for recipient–donor dyads. We view geographical distance as a potential moderator variable affecting the strength of the relationship between aid and political affinity. A positive coefficient on the interaction variable indicates that the donor gives more aid to a recipient with less political affinity when the two countries are more geographically separated. That is, a positive sign on the interaction term means that greater geographical separation raises aid for a given level of political affinity. In this scenario, geographical distance softens the consequences of different foreign policy stances between the dyads, making aid more desirable.

Given the rise in all forms of counterterrorism aid following 9/11 (Fleck and Kilby, 2010; Dreher and Fuchs, 2011), we include a dummy variable, Year₂₀₀₂, equal to 1 for 2002 on, and 0 otherwise. A positive sign is anticipated. We also employ dummy variables for the seven World Bank (2021) designated regions – sub-Saharan Africa (SSA), Latin America and Caribbean (LAC), South Asia (SAS), Europe and Central Asia (ECA), East Asia and Pacific (EAP), North America (NA), and Middle East and North Africa (MENA). Gaibulloev and Sandler (2019) and Kim and Sandler (2020) display shifts in transnational terrorism away from ECA, NA, and LAC to MENA, SA, and SSA over the last four decades. These shifts are apt to affect aid flows negatively for the first three regions and positively for the last three regions.

For the main runs, we include three controls regarding recipient countries and one control concerning the donor countries. The log of lagged military expenditure in the recipient countries is anticipated to have a negative effect on both conflict and governance aid, because recipient countries are better able to fend off resident terrorists. We include a lagged democracy dummy for the recipients, based on the polity scores drawn from Marshall et al. (2018). These scores vary from −10 (full autocracy) to 10 (full democracy). Following much of the literature, we devise a democracy dummy equal to 1 if the recipients’ polity score is 6 or above, and 0 otherwise. Donor countries are expected to supply more conflict and governance aid to democracies. Another control is recipients’ lagged trade (imports and exports) as a share of GDP because donors are likely to be more generous to countries with greater overall trade. Such generosity may benefit donors directly with more trade tied to aid-induced peace and stability. Also, we include lagged GDP per capita of donors because demand for peace and stability is an income-normal good (Sandler and Hartley, 1995). As such, richer donors are more inclined to provide aid to terrorism-plagued recipients whose resident terrorists threaten donors.

In the main robustness runs, we include two additional controls. An indicator for an alliance for the donor–recipient dyad in period t – 1, if relevant, may affect conflict aid negatively or positively. When the presence of the alliance means that the donor can easily come to the recipient country's assistance in times of conflict, the alliance may exert a negative impact as less conflict aid is needed. In contrast, a positive relationship between alliance and conflict aid indicates that a formal security linkage may stimulate more conflict aid. Such an alliance may curb the need for governance aid as the recipient's allies can come to the recipient's assistance in times of institution-induced instability. A second robustness variable concerns the recipient experiencing a civil war. Such wars are likely to curb the donor's enthusiasm for supporting the recipient's conflict or governance aid during such times of instability, so that a negative sign is anticipated for lagged civil war in the recipient. There should be a more marked fall in governance aid than in conflict aid when the recipient is embroiled in a civil war.

Data

Our empirical dataset includes variables for 1974–2013. AidData (2017) provides observations through 2013 on conflict and governance aid, which we convert into 2010 US dollars using World Bank (2021). ITERATE contains counts of transnational terrorist events for the sample period (Mickolus et al., 2021), which we use to identify attacks based on the perpetrators’ nationalities or on the venue recipient countries that involved donors’ interests. For the two forms of aid considered here, AidData (2017) identifies recipient and donor countries. During 1974–2013, there were 99 recipient and 30 donor countries of conflict aid, and 104 recipient and 33 donor countries of governance aid. By identifying perpetrators and victims’ nationalities for most of the recorded attacks, ITERATE allows us to match incidents with recipient countries harboring terrorists and donor countries sustaining attacks on their interests. For domestic terrorist attacks in recipient-adjacent countries, we use domestic attacks given by Enders et al. (2011), who decomposed GTD terrorist event data into domestic and transnational attacks.

The distances in thousands of kilometers between recipient and donor capitals are taken from Gleditsch and Ward (2001), while ideal points distances are drawn from Bailey et al. (2017). Regional designations are those from the World Bank (2021). Recipients’ military expenditure and donors’ GDP per capita are in 2010 US dollars (World Bank, 2021). In percentage terms, the trade shares of GDP are also drawn from World Bank (2021). The recipients’ democracy indicator is a dummy variable constructed from Polity (Marshall et al., 2018).

The lagged dummy indicator for an alliance between the donor–recipient dyad comes from the Correlates of War Project (2021) and Gibler (2009). Also, the lagged civil war dummy for recipient countries is from the Correlates of War Project (2021), based on Dixon and Sarkees’ (2016) analysis of intrastate wars, which includes civil wars involving the government and a non-state entity. Any additional controls for robustness tests are indicated below.

Table 1 displays the count, mean, standard deviation (SD), minimum (min), and maximum values for the study's variables. We highlight a few select values. By taking the exponential transformation of mean ln(conflict aid), we get 1.813 million US dollars in a sample year. Similarly, the exponential transformation of mean ln (governance aid) is 4.158 million US dollars in a sample year. On average, recipient countries experience just over one transnational terrorist incident per year.

Table 1.

Summary statistics.

	Count	Mean	SD	Minimum	Maximum
ln(Conflict aid)_t	72,242	0.597	2.789	0	20.222
ln(Government and civil society aid)_t	72,242	1.432	4.258	0	21.449
ln(transnational terrorism)_t−1	70,480	0.051	0.226	0	4.511
ln(Adjacent domestic terrorism)_t−1	70,480	2.27	1.928	0	8.142
Distance	71,832	5.361	4.008	0.078	18.779
Ideal points distance_t−1	60,914	1.352	1.035	0	5.037
Year₂₀₀₂	72,242	0.293	0.455	0	1
Sub-Saharan Africa	72,242	0.198	0.398	0	1
Latin America and Caribbean	72,242	0.17	0.375	0	1
South Asia	72,242	0.075	0.263	0	1
Europe and Central Asia	72,242	0.183	0.386	0	1
East Asia and Pacific	72,242	0.093	0.29	0	1
North America	72,242	0.036	0.187	0	1
Middle East and North Africa	72,242	0.246	0.431	0	1
ln(Military expenditure)_t−1 in recipient	58,591	20.773	2.317	10.543	27.175
Democracy_t−1 in recipient	64,497	0.413	0.492	0	1
(Trade/GDP)_t−1 (%) in recipient	59,694	57.145	33.514	0.021	437.327
ln(GDP per capita) _t−1 in donor	63,643	9.069	1.463	4.915	11.492
Alliance_t−1	70,480	0.166	0.372	0	1
Civil war_t−1 in recipient	70,480	0.214	0.41	0	1

Notes: Distance is measured in thousands of kilometers.

Results and robustness

In Table 2, we investigate support for Hypothesis 1 as it pertains to conflict aid from donors during 1974–2013. The table is based on the number of domestic terrorist attacks in adjacent countries serving as the instrument for lagged transnational terrorist attacks involving donors’ assets. The latter transnational incidents are based on the perpetrators’ nationalities in aid-recipient countries, hosting the terrorists. The first-stage F-statistics on the instrument are significant at the 0.01 level, suggesting that we have a strong instrument (Stock and Yogo, 2005). In the Online Appendix, Table A1 displays the first-stage estimates for conflict and governance aid, where logged and lagged adjacent domestic terrorism is a significant determinant of logged and lagged transnational terrorism at the 0.01 level for all four displayed models.

Table 2.

Determinants of conflict aid, 1974–2013, based on terrorists’ nationalities.

	Model 1	Model 2	Model 3	Model 4
	Coefficients	Coefficients	Coefficients	Coefficients
ln(Transnational terrorism)_t−1	5.133***	9.986***	5.193**	9.788***
ln(Transnational terrorism)_t−1	(1.971)	(3.665)	(2.036)	(3.704)
Distance	−0.024*	−0.002	−0.025**	−0.007
Distance	(0.012)	(0.017)	(0.013)	(0.017)
Ideal points distance_t−1	0.094*	0.149**	0.075	0.134*
Ideal points distance_t−1	(0.053)	(0.069)	(0.060)	(0.078)
Distance × Ideal points distance_t−1	0.030***	0.031***	0.031***	0.034***
Distance × Ideal points distance_t−1	(0.008)	(0.011)	(0.009)	(0.012)
Year₂₀₀₂	1.684***	1.671***	1.627***	1.612***
Year₂₀₀₂	(0.096)	(0.135)	(0.102)	(0.143)
Sub-Saharan Africa	0.751***	0.833***	0.794***	0.857***
Sub-Saharan Africa	(0.102)	(0.186)	(0.108)	(0.196)
Latin America and Caribbean	0.140	−0.098	0.150	−0.081
Latin America and Caribbean	(0.100)	(0.178)	(0.105)	(0.184)
South Asia	0.565***	0.250	0.601***	0.241
South Asia	(0.120)	(0.163)	(0.131)	(0.171)
Europe and Central Asia	−0.216**	−0.435***	−0.233**	−0.419**
Europe and Central Asia	(0.087)	(0.154)	(0.095)	(0.164)
East Asia and Pacific	0.315***	0.251	0.328***	0.259
East Asia and Pacific	(0.119)	(0.172)	(0.127)	(0.181)
North America	−0.658***	−0.485**	−0.599***	−0.367
North America	(0.121)	(0.227)	(0.124)	(0.238)
ln(Military expenditure)_t−1 in recipient		−0.112***		−0.118***
ln(Military expenditure)_t−1 in recipient		(0.030)		(0.031)
Democracy_t−1 in recipient		0.435***		0.434***
Democracy_t−1 in recipient		(0.111)		(0.115)
(Trade/GDP)_t−1 (%) in recipient		0.003		0.003
(Trade/GDP)_t−1 (%) in recipient		(0.002)		(0.002)
ln(GDP per capita) _t−1 in donor			0.429***	0.352***
ln(GDP per capita) _t−1 in donor			(0.100)	(0.117)
Constant	−0.093	0.568	−4.513***	−2.896**
Constant	(0.470)	(0.880)	(1.030)	(1.246)
First-stage F on instrument	60.34***	27.14***	57.27***	26.13***
N	60,912	48,054	56,656	45,250

Notes: * p < 0.1, ** p < 0.05 and *** p < 0.01. Cluster-robust standard errors based on recipient–donor countries are given in parentheses. Fixed effects based on donor countries are included.

Returning to Table 2, it shows that lagged transnational terrorist attacks by recipients’ nationals against donors’ interests are significant determinants of conflict aid from donors in Models 1–4. Baseline Model 1 excludes the three recipient-specific controls and the single donor-specific control. Model 2 includes the baseline variables and the three recipient-specific variables, while Model 3 includes the baseline variables and the donor-specific control. Finally, Model 4 contains all controls.

Lagged transnational terrorism against donors’ assets is significant at the 0.01 level for Models 1, 2, and 4, and at the 0.05 level for Model 3, thus supporting Hypothesis 1 that the level of donor-provided conflict aid increases with the number of transnational terrorist attacks by terrorists from the aid recipients. In Model 1 of Table 2, a 1% rise in transnational terrorism in the previous year increases conflict aid by 5.133%. In Models 2–4, a 1% rise in transnational terrorist attacks increases conflict aid by 9.986, 5.193 and 9.788%, respectively.

We next consider the other significant coefficients in Table 2. Geographical distance between recipient–donor dyads has the anticipated negative sign, where a greater distance resulted in less conflict aid. In Model 1, if distance increased by 1000 km, then conflict aid fell by 2.4% at the 0.10 level of significance. A similar negative decrease is associated with Model 3 but at the 0.05 level of significance. Distance is not a significant determinant of conflict aid for Models 2 and 4. For three of the four models, an increase in the ideal points distance for recipient–donor dyads raised conflict aid, but the effect is only significant at the 0.05 level for Model 2, where a one-point increase in the ideal points distance led to a 14.9% increase in conflict aid. In Models 1 and 4, the resulting increase in conflict aid is marginally significant at the 0.10 level. Thus, there is some evidence that donors were willing to support recipient countries with resident terrorists even when the aid recipients’ foreign policy agenda differed from that of the donors. The interactive term between geographical distance and lagged ideal points distance is positive, which means that the larger the geographical distance, the more positive is the effect of ideal points distance on conflict aid. For Model 1, the marginal impact of the interaction variable corresponds to an increase of 3% of conflict aid. For Models 2–4, the marginal increase in conflict aid is 3.1, 3.1 and 3.4%, respectively.

In Table 2, conflict aid rose by 168.4% after 9/11, with similar increases in Models 2–4. For all four models, conflict aid increased from 75.1 to 85.7% in SSA relative to MENA. SAS and EAP are associated with increased conflict aid in Models 1 and 3 compared with MENA. In contrast, ECA and NA, which experienced a decrease in transnational terrorism over the last few decades (Gaibulloev and Sandler, 2019), lost conflict aid compared with MENA. Conflict aid in LAC did not change relative to MENA over the sample period despite falling transnational terrorism. For Models 2 and 4, a 1% increase in lagged military expenditure in the recipient countries resulted in decreased conflict aid of 0.11 and 0.12%, respectively, as the recipient had more capacity to confront proactively the resident terrorists on its own. As anticipated, recipient democracies received more conflict aid than non-democracies on the order of 43% in Models 2 and 4. In Table 2, lagged trade shares did not affect recipients’ conflict aid. Finally, a 1% increase in donors’ lagged GDP per capita resulted in a 0.43% and 0.35% increase in conflict aid in Models 3 and 4, so that the demand for conflict is income normal but inelastic.

Table 3 extends our investigation to governance aid and a test of Hypothesis 2. Although the magnitudes of the marginal effects differ from those on corresponding controls in Table 2, all significant coefficients display the same signs as before. We focus on the key differences between the determinants of the two forms of aid. First, logged and lagged transnational terrorism had a large impact on governance aid, where a 1% increase in attacks in the previous year raised aid by between 14% and 23.5%, which supports Hypothesis 2. The impact on governance aid exceeded that for conflict aid. Thus, donors put much aid effort into improving recipients’ institutional structures, rule of law, physical integrity rights, and civil liberties. By so doing, donors addressed the sources of terrorism-inducing grievance through their aid. Second, geographical distance between recipient–donor dyads did not play a role in determining governance aid, which differs from conflict aid. Third, the marginal influences of the determinants of governance aid are much greater than those for conflict aid. For instance, governance aid increased by between 236.7 and 300.3% following 9/11. For all four models, regional governance aid flows changed a good deal, e.g. aid flows to SSA rose by between 161.1 and 181.4% relative to MENA. Generally, regions that experienced more transnational terrorism acquired more governance aid compared with those that experienced less transnational terrorism (e.g. NA and ECA). The exceptions are LAC and EAP, which received more governance aid despite falls in transnational terrorism. Fourth, donors viewed their demand for governance aid as income normal and highly elastic, i.e. a 1% increase in the donors’ per capita income raised governance aid by 1.68 to 1.70%.

Table 3.

Determinants of government and civil society aid, 1974–2013, based on terrorists’ nationalities.

	Model 1	Model 2	Model 3	Model 4
	Coefficients	Coefficients	Coefficients	Coefficients
ln(Transnational terrorism)_t−1	14.080***	23.497***	12.586***	20.368***
ln(Transnational terrorism)_t−1	(3.261)	(6.194)	(3.267)	(5.902)
Distance	−0.027	0.004	−0.029	−0.008
Distance	(0.022)	(0.032)	(0.022)	(0.030)
Ideal points distance_t−1	0.425***	0.621***	0.430***	0.659***
Ideal points distance_t−1	(0.107)	(0.154)	(0.117)	(0.162)
Distance × Ideal points distance_t−1	0.044***	0.049**	0.043**	0.051**
Distance × Ideal points distance_t−1	(0.017)	(0.023)	(0.018)	(0.023)
Year₂₀₀₂	2.875***	3.003***	2.367***	2.438***
Year₂₀₀₂	(0.140)	(0.222)	(0.140)	(0.216)
Sub-Saharan Africa	1.611***	1.814***	1.654***	1.725***
Sub-Saharan Africa	(0.192)	(0.349)	(0.200)	(0.347)
Latin America and Caribbean	1.326***	0.872**	1.389***	0.955***
Latin America and Caribbean	(0.214)	(0.340)	(0.217)	(0.328)
South Asia	1.312***	1.498***	1.402***	1.503***
South Asia	(0.180)	(0.334)	(0.189)	(0.330)
Europe and Central Asia	−0.323*	−0.651**	−0.346*	−0.550*
Europe and Central Asia	(0.180)	(0.294)	(0.185)	(0.292)
East Asia and Pacific	1.001***	0.853***	1.034***	0.903***
East Asia and Pacific	(0.206)	(0.288)	(0.215)	(0.290)
North America	−1.260***	−1.116***	−1.126***	−0.851**
North America	(0.243)	(0.430)	(0.243)	(0.417)
ln(Military expenditure)_t−1 in recipient		−0.192***		−0.215***
ln(Military expenditure)_t−1 in recipient		(0.052)		(0.052)
Democracy_t−1 in recipient		0.688***		0.655***
Democracy_t−1 in recipient		(0.196)		(0.191)
(Trade/GDP)_t−1 (%) in recipient		0.006		0.003
(Trade/GDP)_t−1 (%) in recipient		(0.004)		(0.004)
ln(GDP per capita) _t−1 in donor			1.680***	1.702***
ln(GDP per capita) _t−1 in donor			(0.215)	(0.254)
Constant	−0.687	−0.059	−17.663***	−16.305***
Constant	(0.793)	(1.701)	(2.243)	(2.756)
First-stage F statistics on instrument	60.34***	27.14***	57.27***	26.13***
N	60,912	48,054	56,656	45,250

Notes: * p < 0.1, ** p < 0.05 and *** p < 0.01. Cluster-robust standard errors based on recipient–donor countries are given in parentheses. Fixed effects based on donor countries are included.

Since the administrative distribution of conflict aid in response to transnational terrorist events may require more than a single period, we include alternative lag structures of two and three periods as a robustness test in Table A2 in the Online Appendix. In Table A2, twice-lagged and thrice-lagged transnational terrorist incidents are significant positive determinants of conflict and governance aid. As a longer lag is allowed, the marginal effect of past transnational terrorist incidents grows in size.

For the second set of robustness runs, we examine transnational terrorist attacks against donors’ assets, where the venue countries are those of the conflict-aid recipients. The number of domestic terrorist attacks in countries adjacent to the recipient country serves as the instrument for lagged transnational terrorist attacks in the venue recipient country.

In the Online Appendix, Table A3 displays summary statistics for the runs based on venue countries. The count of variables is now larger because venues have more observations than perpetrators’ nationalities. Table A4 contains the results for conflict aid, and Table A5 includes the results for governance aid. In both tables, the first-stage F-statistics suggest that the instrument is strong. Lagged transnational terrorist attacks are positive and significant determinants of conflict or governance aid, thus supporting Hypothesis 3. When comparing the corresponding coefficients between Tables 2 and A4, we see that all of the significant coefficients have identical signs. The coefficients on lagged transnational terrorist incidents are larger when the venue country, rather than terrorist nationalities, defines those incidents. For completeness, Table A6 provides the first-stage estimates for conflict and governance aid, where lagged adjacent domestic terrorism is a significant determinant of lagged transnational terrorism when venue based.

Online Table A5 investigates the determinants of governance aid in which lagged transnational terrorism corresponds to attacks in the venue recipients, leveled at donors’ interests. In support of Hypothesis 3, lagged transnational terrorism is a significant inducer of governance aid during the sample period. There is a marked similarity in how transnational terrorism affects governance aid under the two alternative perspectives.

Another set of robustness runs involves adding two additional variables – lagged alliance dyad and lagged recipient civil war dummies – to Model 4 from Tables 2 and 3 for conflict and governance aid, respectively. In Table 4, Models 1 and 3 have conflict aid as the dependent variable, while Models 2 and 4 have governance aid as the dependent variable. Robustness follows because logged and lagged transnational terrorism remains a significant positive determinant of the two forms of aid, where marginal values are in the range of earlier results. Moreover, the other controls display significance patterns largely in keeping with those in Tables 2 and 3. In Table 4, the alliance dyad indicator is a negative determinant of conflict and governance aid, consistent with donors substituting their alliance pledge for those forms of aid. The negative sign of the lagged civil war indicator in the recipient is consistent with reduced donor enthusiasm to support conflict efforts and governance structure in conflict-challenged recipient countries. That lack of enthusiasm is more marked for governance aid.

Table 4.

Robustness check with alliance and civil war.

	Model 1	Model 2	Model 3	Model 4
	Conflict	Government and civil society	Conflict	Government and civil society
	Coefficients	Coefficients	Coefficients	Coefficients
ln(Transnational terrorism)_t−1	10.277***	21.248***	9.424***	19.438***
ln(Transnational terrorism)_t−1	(3.858)	(6.168)	(3.478)	(5.499)
Distance	−0.033*	−0.056	−0.009	−0.014
Distance	(0.019)	(0.036)	(0.016)	(0.028)
Ideal points distance_t−1	0.046	0.500**	0.134*	0.660***
Ideal points distance_t−1	(0.095)	(0.196)	(0.077)	(0.158)
Distance × Ideal points distance_t−1	0.038***	0.058**	0.035***	0.052**
Distance × Ideal points distance_t−1	(0.013)	(0.024)	(0.012)	(0.023)
Year₂₀₀₂	1.626***	2.462***	1.596***	2.394***
Year₂₀₀₂	(0.145)	(0.221)	(0.135)	(0.201)
Sub-Saharan Africa	0.893***	1.790***	0.901***	1.838***
Sub-Saharan Africa	(0.202)	(0.358)	(0.205)	(0.361)
Latin America and Caribbean	0.072	1.230***	−0.045	1.049***
Latin America and Caribbean	(0.180)	(0.327)	(0.176)	(0.316)
South Asia	0.213	1.452***	0.277	1.594***
South Asia	(0.172)	(0.331)	(0.176)	(0.337)
Europe and Central Asia	−0.368**	−0.457	−0.395**	−0.489*
Europe and Central Asia	(0.171)	(0.307)	(0.164)	(0.291)
East Asia and Pacific	0.283	0.944***	0.278	0.950***
East Asia and Pacific	(0.179)	(0.289)	(0.180)	(0.288)
North America	−0.060	−0.300	−0.335	−0.770*
North America	(0.304)	(0.541)	(0.247)	(0.425)
ln(Military expenditure)_t−1 in recipient	−0.111***	−0.203***	−0.109***	−0.193***
ln(Military expenditure)_t−1 in recipient	(0.030)	(0.051)	(0.029)	(0.048)
Democracy_t−1 in recipient	0.459***	0.701***	0.420***	0.620***
Democracy_t−1 in recipient	(0.119)	(0.199)	(0.112)	(0.184)
(Trade/GDP)_t−1 (%) in recipient	0.003	0.003	0.002	0.002
(Trade/GDP)_t−1 (%) in recipient	(0.002)	(0.004)	(0.002)	(0.004)
ln(GDP per capita) _t−1 in donor	0.356***	1.710***	0.359***	1.721***
ln(GDP per capita) _t−1 in donor	(0.118)	(0.257)	(0.114)	(0.250)
Alliance_t−1	−0.577***	−1.037**
Alliance_t−1	(0.211)	(0.406)
Civil war_t−1 in recipient			−0.289*	−0.738***
Civil war_t−1 in recipient			(0.174)	(0.264)
Constant	−2.628**	−15.824***	−3.018**	−16.618***
Constant	(1.264)	(2.790)	(1.249)	(2.748)
First-stage F statistics on instrument	24.94***	24.94***	29.33***	29.33***
N	45,250	45,250	45,250	45,250

Notes: * p < 0.1, ** p < 0.05 and *** p < 0.01. Cluster-robust standard errors based on recipient–donor countries are given in parentheses. Fixed effects based on donor countries are included.

We also add three additional controls in a further set of robustness tests (available upon request). First, we include a dummy for lagged civil wars in recipient-adjacent countries, where such neighboring civil wars should curb the donors’ enthusiasm for conflict and governance aid, given neighboring instability. That prediction is borne out. Second, we include a democratization variable indicating the annual increase, if any, in the recipient's democracy (based on polity scores), with the expectation that enhanced democratization increases the donors’ conflict and governance assistance. This expectation is only marginally met for governance aid. Lastly, we include a dyadic trade variable (Correlates of War Project, 2021), where we have no strong priors. An increase in dyadic trade between the recipient and donor decreased conflict aid but had no impact on governance aid. Enhanced dyadic trade may reflect less of a perceived need for conflict assistance. For all three controls, logged and lagged transnational terrorism encourages both forms of aid, in keeping with robustness.

The sought-after strong causal relationship between transnational terrorism and the two forms of aid can be further supported by a placebo or falsification variable. In particular, transnational terrorism in the future should not have an effect on current conflict or governance aid (Berrebi and Ostwald, 2013, 2015). If such a link between future terrorism and current aid were found, then there is an omitted variable or endogeneity concern. This follows because such a relationship would indicate that the terrorism–aid relationship presupposed here is not fully determined by our empirical models.

Table 5 represents the falsification test based on terrorists’ nationalities and venue countries, where the key new variable is $\ln (transnational terrorism)_{i j t + 2}$ for future terrorist attacks. Models 1 and 2 are for terrorists’ nationalities, while Models 3 and 4 are for recipient attack venues. For all four models, future transnational terrorist attacks are not significant determinants of current conflict and governance aid. In addition, the pattern of significant coefficients remains the same as that before the inclusion of future transnational terrorist attacks. This falsification test supports our identification strategy.

Table 5.

Falsification test based on terrorists’ nationalities and venue countries.

	Nationality		Venue
	Model 1	Model 2	Model 3	Model 4
	Conflict	Government and civil society	Conflict	Government and civil society
	Coefficients	Coefficients	Coefficients	Coefficients
ln(Transnational terrorism)_t+2	−0.105	0.162	−0.115	0.173
ln(Transnational terrorism)_t+2	(0.087)	(0.166)	(0.080)	(0.140)
ln(Transnational terrorism)_t−1	0.248**	0.611***	0.215***	0.644***
ln(Transnational terrorism)_t−1	(0.104)	(0.193)	(0.082)	(0.155)
Distance	−0.031***	−0.068***	−0.023**	−0.060***
Distance	(0.011)	(0.020)	(0.009)	(0.017)
Ideal points distance_t−1	0.177***	0.736***	0.263***	0.908***
Ideal points distance_t−1	(0.061)	(0.126)	(0.052)	(0.107)
Distance × Ideal points distance_t−1	0.025***	0.038**	0.013	0.016
Distance × Ideal points distance_t−1	(0.009)	(0.019)	(0.008)	(0.016)
Year₂₀₀₂	1.208***	1.688***	1.147***	1.659***
Year₂₀₀₂	(0.079)	(0.103)	(0.067)	(0.087)
Sub-Saharan Africa	0.618***	1.172***	0.484***	1.332***
Sub-Saharan Africa	(0.150)	(0.249)	(0.127)	(0.222)
Latin America and Caribbean	0.080	1.171***	0.089	1.271***
Latin America and Caribbean	(0.149)	(0.269)	(0.119)	(0.217)
South Asia	0.087	1.212***	0.244**	1.456***
South Asia	(0.131)	(0.262)	(0.109)	(0.227)
Europe and Central Asia	−0.328***	−0.461**	−0.101	−0.191
Europe and Central Asia	(0.117)	(0.208)	(0.092)	(0.167)
East Asia and Pacific	0.338**	1.068***	0.500***	1.375***
East Asia and Pacific	(0.159)	(0.271)	(0.132)	(0.229)
North America	−0.461***	−1.202***	−0.161	−0.654**
North America	(0.157)	(0.308)	(0.132)	(0.256)
ln(Military expenditure)_t−1 in recipient	−0.089***	−0.160***	−0.121***	−0.206***
ln(Military expenditure)_t−1 in recipient	(0.024)	(0.041)	(0.021)	(0.034)
Democracy_t−1 in recipient	0.329***	0.500***	0.210***	0.402***
Democracy_t−1 in recipient	(0.092)	(0.149)	(0.074)	(0.123)
(Trade/GDP)_t−1 (%) in recipient	−0.000	−0.004	−0.002***	−0.006***
(Trade/GDP)_t−1 (%) in recipient	(0.001)	(0.003)	(0.001)	(0.002)
ln(GDP per capita) _t−1 in donor	0.503***	2.112***	0.474***	1.915***
ln(GDP per capita) _t−1 in donor	(0.096)	(0.223)	(0.076)	(0.179)
Constant	−2.701***	−16.563***	−1.554*	−13.704***
Constant	(0.985)	(2.279)	(0.802)	(1.864)
N	42,480	42,480	54,076	54,076

Notes: * p < 0.1, ** p < 0.05 and *** p < 0.01. Cluster-robust standard errors based on recipient–donor countries are given in parentheses. Fixed effects based on donor countries are included.

Concluding remarks

For a global sample of recipient–donor dyads, the current study finds that transnational terrorist attacks against donors’ interests by terrorists from recipient countries increase both conflict and governance aid. This a novel finding where transnational terrorism stimulates two important forms of foreign assistance. Unlike earlier studies on this issue, we account for the two-way causality between foreign aid and transnational terrorism with the use of an instrument, consisting of domestic terrorist attacks in countries adjacent to the recipient. In contrast to two of the three earlier studies, we include all major donor countries, not just the United States. Relative to the interesting study of Dreher and Fuchs (2011), the current study includes more donors, more post-9/11 years, and an alternative methodology. We uncover a strong positive relationship between transnational terrorist attacks and the receipt of governance aid, which differs from Dreher and Fuchs (2011). In fairness to them, we must point out that they draw their sample donors from Development Assistance Committee data, while our sample is drawn from AidData (2017). Moreover, our methodologies differ – they employ a Heckman selection model, and we utilize a two-stage least squares regression.

Some of our findings deserve highlighting. First, past transnational terrorist attacks involving recipient–donor dyads generally result in a greater impact on governance aid relative to conflict aid. Thus, donors appear more interested in using aid to get at some of the root causes of terrorism rather than bolstering the recipients’ conflict-fighting abilities. This makes sense especially in light of studies that show that aid given to eliminate resident terrorist groups results in a moral hazard problem leading to expanded longevity for groups (Bapat, 2011 in particular). Second, donor countries view their demand for governance aid as more income elastic relative to conflict aid. As recipients become wealthier over time, their interest in giving governance aid as a counterterrorism tool should increase in importance. Third, both the perpetrators’ nationalities and venue or place of the attack affect the flow of aid to recipient countries. Fourth, the influence of transnational terrorist attacks on aid flows responds to the standard controls in the anticipated direction. Fifth, because the post-9/11 era displays a large effect on conflict and governance aid, studies, like ours, that include more post-9/11 years are apt to find a strong stimulation of transnational terrorism on foreign aid.

Supplemental Material

sj-docx-1-cmp-10.1177_07388942221081103 - Supplemental material for Does transnational terrorism stimulate foreign assistance?

Supplemental material, sj-docx-1-cmp-10.1177_07388942221081103 for Does transnational terrorism stimulate foreign assistance? by Wukki Kim and Todd Sandler in Conflict Management and Peace Science

Supplemental Material

sj-zip-2-cmp-10.1177_07388942221081103 - Supplemental material for Does transnational terrorism stimulate foreign assistance?

Supplemental material, sj-zip-2-cmp-10.1177_07388942221081103 for Does transnational terrorism stimulate foreign assistance? by Wukki Kim and Todd Sandler in Conflict Management and Peace Science

Footnotes

Acknowledgements

We thank three anonymous reviewers for their very helpful comments.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

ORCID iDs

Wukki Kim

Todd Sandler

Supplemental material

Supplemental material for this article is available online.

Notes

References

Abadie

(2006) Poverty, political freedom and the roots of terrorism. American Economic Review 96(2): 50–56.

AidData (2017) AidDataCore_ResearchRelease_Level1_v3.1 Research Releases Dataset. Williamsburg, VA: AidData. Available at: http://aiddata.org/research-datasets (accessed 8 March 2021).

Azam

J-P

Delacroix

(2006) Aid and the delegated fight against terrorism. Review of Development Economics 10(2): 330–344.

Azam

J-P

Thelen

(2008) The roles of foreign aid and education in the War on Terror. Public Choice 135(3–4): 375–397.

Azam

J-P

Thelen

(2010) Foreign aid versus military intervention in the War on Terror. Journal of Conflict Resolution 54(2): 237–261.

Bailey

Strezhnev

Voeten

(2017) Estimating dynamic state preferences from United Nations voting data. Journal of Conflict Resolution 61(2): 430–456.

Bandyopadhyay

Sandler

(2011) The interplay between preemptive and defensive counterterrorism measures: A two-stage game. Economica 78(3): 546–564.

Bandyopadhyay

Sandler

(2021) Counterterrorism policy: Spillovers, regime solidity, and corner solutions. Journal of Economic Behavior and Organization 188: 811–827.

Bandoyopadhyay

Sandler

Younas

(2011) Foreign aid as counterterrorism policy. Oxford Economic Papers 63(3): 423–447.

10.

Bandoyopadhyay

Sandler

Younas

(2014) Foreign direct investment, aid, and terrorism. Oxford Economic Papers 66(1): 25–50.

11.

Bapat

(2011) Transnational terrorism, US military aid, and the incentive to misrepresent. Journal of Peace Research 48(3): 303–318.

12.

Bapat

(2019) Monsters to Destroy: Understanding the War on Terror. New York: Oxford University Press.

13.

Berrebi

Ostwald

(2013) Exploiting the chaos: Terrorist target choice following natural disasters. Southern Economic Journal 79(4): 793–811.

14.

Berrebi

Ostwald

(2015) Terrorism and fertility: Evidence for a causal influence of terrorism on fertility. Oxford Economic Papers 67(1): 63–82.

15.

Boutton

(2019) Of terrorism and revenue: Why foreign aid exacerbates terrorism in personalist regimes. Conflict Management and Peace Science 3(4): 359–384.

16.

Boutton

Carter

(2014) Fair-weather allies? Terrorism and the allocation of US foreign aid. Journal of Conflict Resolution 58(7): 1144–1173.

17.

Choi

S-W

(2010) Fighting terrorism through the rule of law? Journal of Conflict Resolution 54(6): 940–966.

18.

Correlates of War Project (2021) Formal Alliances v4.1 and Intra-State Wars Dataset v.5.1. Available at: https://correlatesofwar.org/data-sets.

19.

Danzell

Kisangani

Pickering

(2019) Aid, intervention, and terror: The impact of foreign aid and foreign military intervention on terror events. Social Science Quarterly 100(3): 951–964.

20.

Dixon

Sarkees

(2016) A Guide to Intra-State Wars: An Examination of Civil, Regional, and Intercommunal Wars, 1816–2014. Washington, DC: CQ Press.

21.

Dreher

Fuchs

(2011) Does terror increase aid? Public Choice 149(3–4): 337–363.

22.

Enders

Sandler

Gaibulloev

(2011) Domestic versus transnational terrorism: data, decomposition, and dynamics. Journal of Peace Research 48(3): 319–337.

23.

Fleck

Kilby

(2010) Changing aid regimes? U.S. foreign aid from the Cold War to the War on Terror. Journal of Development Economics 91(2): 185–197.

24.

Gaibulloev

Sandler

(2019) What we have learned about terrorism since 9/11. Journal of Economic Literature 52(2): 275–328.

25.

Gaibulloev

Piazza

Sandler

(2017) Regime types and terrorism. International Organization 71(3): 491–522.

26.

Gibler

(2009) International Military Alliances, 1648–2008. Washington, DC: CQ Press.

27.

Gleditsch

Ward

(2001) Measuring space: A minimum-distance database and applications to international studies. Journal of Peace Research 38(6): 739–758.

28.

Kim

Sandler

(2020) Middle East and North Africa: terrorism and conflicts. Global Policy 11(4): 424–438.

29.

Kim

Sandler

(2021) Foreign aid and terrorist groups: Incidents, ideology, and survival. Public Choice 189(1–2): 139–160.

30.

Kim

Sandler

(2021) Resident terrorist groups, military aid, and moral hazard: Further empirical analysis. Defence and Peace Economics 32(1): 1–17.

31.

Marshall

Jaggers

Gurr

(2018) Polity IV Project: Dataset and Users’ Manual. Available at: http://www.systemicpeace.org/inscrdata.html.

32.

Mickolus

Sandler

Murdock

, et al. (2021) International Terrorism: Attributes of Terrorist Events, 1968–2020 (ITERATE). Ponte Vedra, FL: Vinyard Software.

33.

National Consortium for the Study of Terrorism and Responses to Terrorism (2019) Global Terrorism Database (GTD). Available at: www.start.umd.edu/gtd/.

34.

Piazza

(2017) Repression and terrorism: a cross-national empirical analysis of types of repression and domestic terrorism. Terrorism and Political Violence 29(1): 102–118.

35.

Piazza

Choi

S-W

(2018) International military intervention and transnational terrorist backlash. International Studies Quarterly 62(3): 686–695.

36.

Sandler

Hartley

(1995) The Economics of Defense. Cambridge: Cambridge University Press.

37.

Savun

Tirone

(2018) Foreign aid as a counterterrorism tool: More liberty, less terror. Journal of Conflict Resolution 62(8): 1607–1635.

38.

Stock

Yogo

(2005) Testing for weak instruments in linear IV regression. In: Andrews

WKD

Stack

(eds) Identification and Inference for Econometric Models: Essays in Honor of Thomas Rothenberg. Cambridge: Cambridge University Press, 80–108.

39.

Walsh

Piazza

(2010) Why respecting physical integrity rights reduces terrorism. Comparative Political Studies 43(5): 551–577.

40.

World Bank (2021) World Development Indicators (WDI). Available at: http://databank.worldbank.org/ddp/home.do

41.

Young

Findley

(2011) Can peace be purchased? A sectoral-level analysis of aid’s influence on transnational terrorism. Public Choice 149(3–4): 365–381.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

11.26 MB

0.00 MB

0.05 MB