Faith and Foreign Aid? The Effects of Islam on Development Finance. Josh Loud. Daniel Nielson.

Transcription

1 Faith and Foreign Aid? The Effects of Islam on Development Finance Josh Loud Daniel Nielson Christopher O Keefe Chris.okeefe.79@gmail.com Abstract:

2 2 Between 1980 and 2000 donor governments gave a disproportionate share of their foreign aid to Muslim countries. We find this pattern puzzling. It seems unlikely that donor countries mostly Western liberal Christian democracies feel a secret affinity for Islam. Instead, we explore the competing possibilities that either (A) voters (principals ) interests in security from terrorism or reducing immigration or (B) governments (agents ) preferences for oil, trade, or alliances drive aid allocations to Muslim states. The scholarly community has generally concluded that recipient need and cultural similarities do not drive bilateral aid allocation. A few scholars claim that development concerns affect aid (Lumsdaine 1993, Noel and Therien 1995), but several others have challenged this conclusion (Alesina and Dollar 2000; Schraeder, Hook and Taylor 1998). Instead, most scholars see donors as self-interested and venal, seeking opportunities to secure strategic goals in economics and politics, rather than improve the development levels of recipients (Bueno de Mesquita and Smith 2006). Our findings support this majority view that aid flows for purposes other than recipient need. If aid were allocated solely according to developmental criteria, Muslim nations given their higher socioeconomic status would receive disproportionately less aid, not more (see tables A, B). Since it seems highly unlikely that Europeans and North Americans regard for Islam dictates aid allocation, we explore the security and economic factors that might contribute to this unexpectedly high level of aid to Muslim countries. We argue that several strong concerns of both voters and their governments are highly collinear with majority-muslim countries. First, voters in most donor countries consistently rate immigration as among their most important concerns. Especially in

3 3 Europe, immigrants are predominantly Muslim. As immigration flows increase, agency theory suggests that governments should anticipate voter concern and may seek to address it through foreign aid. Increased aid to immigrant-sending countries may buy more restrained immigration from recipients or at the least better educated and healthier immigrants (Teitelbaum 1984, 446; Todaro and Maruszko 1987, 111). Second, fear of terrorism may be the only issue that outpaces voter concern about immigration. And much of the terror targeting Western populations originates in Islamic countries. Governments anticipating voter interests may then use foreign aid to attempt to buy more aggressive anti-terror policy in states that produce terrorists. At the least, targeted aid may alleviate the poverty that many believe provokes terror or help to build government institutions that can prevent it. Aid may forestall states from failing and many analysts believe that failed states enable terrorism (see Krueger and Maleckova 2003, 140). Of course, the above two arguments presume that governments are faithful agents of voters. It may well be the case instead that governments as agents possess significant autonomy from voters, which enables slack, shirking, or both. Governments may instead use foreign aid to pursue aims that only indirectly relate to voter interests. First, in pursuit of their strategic interests, governments may use aid to buy favorable terms of trade with oil producers. Second, governments may wish to use aid to bolster strategic alliances (independent of concerns about oil). And third, governments may use aid to build better relations with their trading partners (McKinlay and Little 1977, Maizels and Nissanke 1984, Alesina and Dollar 2000). While all of these concerns ultimately affect voter interests in indirect ways, we argue that none of them provoke the same kind of

4 4 response in public opinion as immigration and terrorism. Thus, aid allocation focused on oil, trade, and alliances ought to reflect more on government preferences than voter preferences. We proceed by spelling out the empirical puzzle, discussing recent literature on aid allocation, and developing our argument. Using data from the Project Level Aid Database (PLAID) we construct a dyadic dataset of donors and recipients. We test the different causal pathways we have specified and conclude. Uncovering a Puzzle: Muslim Countries Get More Aid We find, unlike Alesina and Dollar, that religion does matter in the allocation of foreign aid (see Table B). Our initial results show that even when controlling for oil exports, arms imports, terrorist attacks, and total imports/exports, as well as a set of other controls, Muslim states receive 53.1% more aid than non Muslim states, which is significant at the 10% level. Muslim states also significantly receive more projects. We expect to resolve this puzzle by considering issues that represent important policy concessions that would also be collinear with the Muslim-ness of aid recipients: oil, immigration, and terrorism. Ten of OPEC s twelve members are Muslim-majority states (OPEC 2007). Seven of the fourteen countries that produce more than 2 million barrels per day of oil do not belong to OPEC, including the United States, Russia, Mexico, Canada, Norway and the United Kingdom (U.S. Department of Energy 2005). Immigrants, especially to the European donors, tend to come from Muslim states. Lastly, Arab-Islamic terrorism has been on the news at least since the Palestinian attacks at the Munich games in Islamic-ness correlates with so many other issues that it may have little to do with aid allocation itself. We attempt to control for several of these

5 5 variables to determine whether the initial relationship between Islam and aid allocation is spurious, or legitimate. Foreign Aid: Intergovernmental Bribe or Philanthropy? The host of articles published about the effectiveness of foreign aid assumes that foreign aid is allocated by developmental criteria. More recent studies have begun to disaggregate foreign aid, examining the relationship between aid that targets project or program aid (Mavrotas and Ouattara 2006), and the impact of aid that should be beneficial in the short or long term (Clemens et al 2004, Rajan and Subramanian forthcoming, Bermeo 2007). These studies still assume that aid is targeted to have some impact on development. However, other studies indicate that aid is more often allocated with an eye to strategic and commercial concerns than developmental aims (McKinlay and Little 1977, Maizels and Nissanke 1984, Alesina and Dollar 2000). How does this explain the higher overall allocation of aid to Muslim countries? We argue, first, that these countries do not receive more aid on account of their dominant religion. The sole fact of a positive correlation between Islamic-ness and aid allocation lacks a causal mechanism. Instead, we suggest that specific measures of strategic or commercial concern offer potential explanations. That is, aid is used as a foreign policy tool, consistent with the aid allocation literature. Unfortunately, this conclusion is too easy. Aid may indeed be allocated as a foreign policy tool, but this neglects the roots of foreign policy behavior. An analysis of the relationship between aid allocation and specific issues, beyond traditional analysis that emphasizes recipient need or donor commercial or strategic interests, helps us address these roots. Do voter (principal) concerns matter in aid allocation?

6 6 Foreign aid allocation can be analyzed within the principal-agent framework (Milner 2006, Martens et al. 2002). A principal, in this case, the voters, delegates aid allocation to the government. Once a task is delegated, principal and agent preferences can diverge (Hawkins et al 2006). Agents can take advantage of information asymmetries and hide their actions from the principal, either generating agency slack (not working hard enough) or shirking (pursuing different goals from the principal s). Asymmetric information and hidden action can allow agents to work against their principal s preferences. Given a limited foreign aid budget, the policymakers would prefer to allocate the entire budget to strategic concerns (Milner 2006) building trade parternships, strengthening bilateral alliances, and, we argue, maintaining stable oil flows. On the other hand, voters have a separate set of concerns, which include fear of terrorism and immigration. Voters prefer that the government address these concerns, and faithful governments may then choose to use foreign aid as a tool either to thwart terrorism, stem immigration, or both. Although the literature on agency theory argues that various mechanisms are necessary to ensure that the agent follow the principal s will (Kiewiet and McCubbins 1991, Hawkins et al. 2006), we simply evaluate the influence that independent principal and agent preferences have on aid outcomes. Screening and selection, police patrols and fire alarms, among other mechanisms as well as the salience of the principal s preferences and the nature of principals whether single or collective all complicate the act of delegation and the principal s ability to ensure that the agent does what he or she wants (Nielson and Tierney 2003, Hawkins et al 2006).

7 7 Foreign aid in itself is a low salience issue, easily trumped by other voter concerns (Cooper and Verloren van Themaat 1989; Lancaster 2000). Although the public generally supports increased allocation, especially for development (Noel and Therien 2002, Milner 2006, McDonnell 2003, Smillie et al 1998), this low salience may mean that politicians can use aid for their aims, without concern for electoral repercussion (Lundsgaarde 2007). However, politicians may be less free to ignore crucial hot-button issues, such as terrorism and immigration -- problems that voters might readily link to Muslim nations. Foreign aid can play an important role in each of these issues. Azam and Delacroix (2006) find that aid flows from to terrorist source states, suggesting that aid may be linked to a prominent argument made recently by the Bush Administration (cited in Krueger and Maleckova 2003, 140) that aid ought to support the development of institutions that can be used to counter terrorism. Gaytan-Fregoso and Lahiri (2000) develop a model of aid and illegal immigration and find that under one set of conditions, aid has no effect on illegal immigration; under another, it can actually increase illegal immigration. Similarly, Schiff (1994) argues that aid increases migration, although policy makers think it may reduce it. However, aid may staunch immigration that results from humanitarian crises, since aid is less expensive than social programs for new immigrants (Teitelbaum 1984, 445). Aid also has the potential of improving development in the countries where immigration orginates (446; Todaro and Maruszko 1987, 111). This may either stem the flow or at least improve the human capital of immigrants. Policy makers in receiving states might also condition bilateral relations including the receipt of aid on immigration policy (Cornelius and

8 8 Rosenblum 2005, 109). None of these authors test their arguments quantitatively, so we attempt to fill that void here. We thus assume that policy makers may believe that aid can either slow immigration or improve immigrant quality. Unfortunately, few studies include oil flows as a determinant of aid allocation. However, the oil-dependent economies of most donor states generate incentives to ensure that oil continues to flow. Aid might be a balanced exchange that appears unbalanced to the observer. That is, following Keohane (1984, 128), aid may flow to oil-exporting states as insurance that oil will continue to flow in the future. Divergent Preferences and Foreign Aid Allocation Agency theory becomes interesting when principals and agents have different preferences. In our model, we argue that voters are primarily concerned with terrorism and immigration in foreign policy. If electoral systems work, then voters ought to be able to sanction politicians who do not follow their preferences. And politicians, anticipating this sanctioning, should match their behavior to voter preferences. As we have argued, we expect counter-terrorism and immigration are issue areas we expect to see driving aid allocation to Muslim states, in addition to the strategic interests of oil, alliances, and trade. Voter Preferences: Terrorism and Immigration Specific voter preferences ought to influence the allocation of foreign aid, as well as other aspects of economic statecraft. While voters do not desire aid allocation to Muslims because of a preference for the Islamic faith, their concerns with immigration

9 9 and terrorism, though not specifically tied to aid, might induce allocation to slow immigration and fight terrorism. Voters do not harbor a secret affinity for Muslims. In recent polls, Muslims are seen as the least favored religious group in the United States. One study indicates that affinity for American-Muslims in the US is at least what it was in the months before September 11 th, while Muslims who are not identified as Americans are less favorably viewed than before (Pew Research Center for the People and the Press/Pew Forum on Religion and Public Life 2003). A 2004 poll shows that 50% of Western Europeans view Muslims with suspicion. A 2006 poll showed mixed results, with a majority of respondents in France and Britain responding favorably, although Spanish and German respondents expressed negative views (EUMC 2006, 10). Opinion polls from before September 11 th in the United States indicate a stable concern with terrorism prior to the attacks, at least as far back as 1995 and Aldrich et al (2006, 493) argue that September 11 th represented less a turning point in American public opinion regarding terrorism as a realization among elites that the public viewed terrorism as a threat. Europeans were much more familiar with terrorism, and hence, we suspect that anxiety among European publics was higher. Pape (2003) finds that suicide terrorism exclusively targets democratic states, with the view that democratic citizens have a low tolerance for casualties and a high degree of influence on policy making. Although suicide terrorism is demonstrably more destructive than other forms (Pape 2003, 346), we argue that publics will be concerned with the threat of terrorism overall, failing to differentiate between suicide and nonsuicide terror. Public concerns about terrorism should drive public policy on the matter,

10 10 as politicians anticipate voter fears. At the same time, terrorism becomes a major concern for governments only when voters influence their tenure significantly. Failure to respond to terrorism constitutes an electoral threat. But except in the instance of weapons of mass destruction-armed terrorists does it constitute an existential one and should thus be a smaller concern for governments than for voters. Immigration, like terrorism, spurs voter fears more than government fears. There is, at the very least, a popular unease with immigrants generally in Europe and the United States. Although Lahav (2004) argues that the data disagrees, several scholars who study immigration and public opinion make a case that voters and their governments diverge with respect to preferences for immigration policy (Fetzer 2000, Freeman 1995). Public opposition of immigrants can stem from multiple sources, including fears of economic and cultural marginalization, economic self-interest, and negative personal contact with immigrants (Fetzer 2000, 3). We are agnostic as to why voters oppose immigration. On the other hand, governments may prefer to encourage immigration in order to maintain flows of workers in immigration-dependent sectors of the economy (Cornelius 1998). Like terrorism, failure to address immigration amounts to an electoral threat, but not an existential threat. Indeed, independent of voter concerns, politicians and policy makers may encourage immigration, especially in the face of labor shortages. Government Preferences: Oil, Trade and Alliances According to realist theory, the state is a rational, unitary organization that is socialized by the competitive pressures of the international system to seek its survival as a first priority (Waltz 1979). Essentially, realism provides a set of independent government level preferences (Milner 1997). This state is primarily concerned with

11 11 power, both military and economic. Meernik, Krueger and Poe (1998) argue that aid will be given to support allies or states with significant donor military presence. Schraeder, Hook and Taylor (1998) argue, on the other hand, that aid will favor security relationships or nations with large military establishments, so as to ensure more bang for the foreign aid dollar. Developing and maintaining alliances and trade relationships both tend to be low salience issues with voters, thus politicians should be able to target aid to these tasks. In all three cases we expect to see aid as an intergovernmental bribe, in the same way that Bueno de Mesquita and Smith (2006) see aid. States should be primarily concerned with securing access to oil to ensure both military and economic power. Likewise, concern with maintaining alliances stems from the same roots (Schraeder, Taylor and Hook 1998; Meernik, Krueger and Poe 1998). Maintaining and strengthening trade partnerships should likewise tie into realist concerns with power. Hypotheses Following the agency theory literature, since aid overall is a low salience issue for voters and control mechanisms over its allocation are weak, we expect agents to have the advantage in policy making. Thus, we derive the following hypotheses: Principal (Voter) Preferences: H1 Immigration: An increase in immigration flows to a donor should have no significant effect on aid allocation. H2 Terrorism: An increase in terrorism against the donor country will have no significant effect on aid allocation.

12 12 Agent (Government) Preferences: oil exporter. H3 Oil: An increase in oil flows to the donor country should increase aid to the H4 Alliances: Aid allocation will flow to donors alliance partners. H5 Trade: Aid allocation will flow to donors trade partners. H6 (Religion): Islam should not be a significant predictor of aid flows. The last hypothesis does not derive from our theoretical analysis, but the expectation that the inclusion of immigration, terrorism, oil, trade, and alliance data will be collinear with religion and thus resolve the puzzle that began this paper. Data and Methods We evaluate the argument that government concerns drive aid allocation over voter concerns using a dyadic database that includes directed instances of terrorism, immigration, oil flows, trade flows, security alliances, and bilateral foreign aid allocation. We also include a large set of controls. We take the log of foreign aid commitments as our dependent variable. Our data on aid come from the Project Level Aid (PLAID) database, 1 designed to capture the universe of aid projects. 1 Funded by the National Science Foundation grant SES

13 13 Our main independent variables include instances of terrorism and immigration, as voter concerns. We use ITERATE data (Mickolus 2006), creating dyads based on the nationality of the terrorist and victim, respectively. Aid might flow from the nation (or nations) whose citizens were attacked to the nation whose nationals commit the act of terrorism in an effort to strengthen the terrorist-sending nation s capacity to fight terrorism. We also include a variable that indicates whether the attack was statesponsored or not. Data on immigration comes from the OECD (2006a) and Eurostat (2006). We constructed a common dataset that reduces the problems of data coverage present in these individual sources. In cases where the OECD and Eurostat data overlapped, we defaulted to the Eurostat data, which had better coverage overall. Again, we used these data to construct directed immigration dyads for immigration to donor countries. We coded the recipient of immigration as the donor country, and the immigrant s previous country of residence 2 as the aid recipient. Given the high degree of missingness in the immigration data, we used stepwise imputation to fill the gaps. As covariates for the imputation process, we included lagged values (one and two periods) of the immigration variable, directed imports and exports, total imports and exports, population, per capita GDP, distance between the two states, and surface area. Data on directed oil flows comes from the OECD (OECD 2006b), with the exporter coded as the aid recipient and the importer as the donor. We took our data on trade from the IMF (IMF 2006) and included trade flows in both directions to encompass the complete nature of the relationship. Both trade variables are included as a percentage 2 We were unable to find good cross-national time series data on immigrant nationalities.

14 14 of the total imports and exports to approximate the importance of the relationship to each party in the dyad. Lastly, we obtained our data on alliances from the ATOP data on security agreements (Leeds et al. 2002). We also include several control variables. First, we use the same religion data with which we discovered the puzzle that motivates this paper. We expect the positive relationship between Islam and aid allocation to disappear once we control for the effects of immigration, terrorism, oil flows, alliances, and trade. As further controls, we include population, per capita GDP, distance between state A and state B, total years of independence for former colonies, land area, arable land, long term debt, exchange rate, polity score, winning coalition size, and colonial ties. Summary statistics for these variables can be found in the appendix (see table C). As dependent variables we include the natural log of aid, adjusted to allow for zeros (ln(aid + 1)) and the total number of projects each recipient received from each donor in each year. Given the large number of zeros in the aid amount variable, we estimate the first models using time series cross sectional Tobit regressions. We estimate this model twice, once with an interaction term between Muslim and immigration to account for the possibility of a unique effect for immigration from Muslim countries. With the number of projects as the dependent variable, we opt for time series cross sectional negative binomial regressions to account for both the discrete nature of count data and the overdispersion of the dependent variable. Again, this model is estimated twice: once with, and once without the interaction term. The dataset used for estimation contains 44,163 observations and 2,415 dyads.

15 15 Analysis and Results In the first regression with the log of dyadic aid flows as the dependent variable, several interesting patterns arise. First, both imports and exports have a significant effect on aid flows. When donors import goods from or export goods to a developing nation, they give disproportionate amounts of aid to this country. For example, donors whose exports comprise 10% of the recipient s imports will give 16.7 percent more aid than donors who do not export to the recipient at all. Also of interest is that donors give 61.6% more aid to recipients they share alliances with than recipients they do not, which is significant at the.01 level. Surprisingly, donors give less aid to countries they transfer arms to, which is also significant at the.01 level. Terrorist attacks also appear to have an effect on aid flows, with each attack leading to an 11.2% increase in aid. Surprisingly, Muslim states still receive disproportionate amounts of aid; states whose populations are greater than 50 percent Muslim receive 32.9% more aid than those states that do not. Interestingly, while immigration is not significant, the Muslim immigration interaction is, but is negative. Muslim states whose citizens immigrate to donor nations receive less aid than other states whose citizens move to donor countries. The models with the number of projects as the dependent variable tell a similar story. Trade ties are a significant factor in determining aid flows; a ten percent increase in the proportion of a nation s imports attributable to a given donor increases the number of aid projects received by that donor by.5. Again, Muslim states receive more projects

16 16 than other states; a state with a Muslim majority will receive.28 more projects which is significant at the.01 level. The presence of an alliance likewise increases the number of projects by.525. Again, arms transfers decrease the number of projects a recipient receives at a statistically significant level, but the substantive significance of the coefficient is small. Terrorist attacks also significantly increase the number of aid projects a nation receives, but again, this number is not substantively significant. Interestingly, even after controlling for trade, alliances, arms flows, oil flows, and several other dyadic variables, Islamic states still receive more aid, both in terms of amounts and number of projects. The results therefore are mixed. Immigration does not seem to have an effect on aid flows, and Muslim immigration has a significant, yet meager effect. Terrorism has a significant effect on aid flows, yet terrorist attacks are so seldom (mean =.03) that it does not appear that terrorist attacks are driving the flow of aid to Muslim countries. Oil flows do not affect the amount of aid that a state receives, but they do affect the number of projects, although this effect is in the opposite direction of the hypothesized relationship. Alliances seem to be an important factor and increase both the amount of aid and the number of projects, which is also true of trade relationships. We find the persistent effect of Islam puzzling. Further research will be required to uncover whether Islam truly affects aid allocations or whether this variable is still proxying for another unobserved variable. Conclusions

17 17 This research has shown that strategic factors play a large role in foreign aid allocations. In particular, government (agent) preferences seem to have a large effect on allocations, but terrorism (a principle preference) does seem to play a role. Of particular interest is the result that even when controlling for other factors thought to be highly collinear with Islam, Muslim states still appear to receive both more projects and more aid than their counterparts.

18 18 WORKS CITED Aldrich, John, Christopher Gelpi, Peter Feaver, Jason Reifler and Kristin Thompson Sharp Foreign policy and the electoral connection. Annual Review of Political Science 9: Alesina, Alberto and David Dollar Who gives foreign aid to whom and why? Journal of Economic Growth 5(1): Azam, Jean-Paul and Alexandra Delacroix Aid and the delegated fight against terrorism. Review of Development Economics 10(2): Bermeo, Sarah Foreign aid, foreign policy, and development: Sector allocation in bilateral aid. Presented at the Annual Meeting of the International Studies Association, Chicago, IL. Bueno de Mesquita, Bruce and Alastair Smith Foreign aid and policy concessions. Journal of Conflict Resolution 51(2): Clemens, Michael, Steven Radelet and Rikhil Bhanavi Counting chickens when they hatch: The short-term effect of aid on growth. Working paper 44. Washington, DC: Center for Global Development.

19 19 Cornelius, Wayne and Marc Rosenblum Immigration and politics. Annual Review of Political Science 8: European Monitoring Center of Racism and Xenophobia (EUMC) Muslims in the European Union: Discrimination and Islamophobia. Austria: EUMC. Eurostat Eurostat Immigration by sex and country of previous residence. Available at < 0, ,0_ &_dad=portal&_schema=PORTAL>. 23 August, Faini, R., Venturini, A., Trade, aid and migrations: some basic policy issues. European Economic Review 37: Fetzer, Joel Public attitudes toward immigration in the United States, France, and Germany. Cambridge: Cambridge University Press. Freeman, Gary Modes of immigration politics in liberal democratic states. International Migration Review 29: Gaytan-Fregoso, Helena and Sajal Lahiri Foreign aid and illegal immigration. Journal of Development Economics 63: IMF Direction of Trade Statistics. Washington, DC: IMF.

20 20 Keohane, Robert After hegemony: Cooperation and discord in the world political economy. Princeton, NJ: Princeton University Press. Krueger, Alan and Jitka Maleckova Education, poverty and terrorism: Is there a causal connection? Journal of Economic Perspectives 17: Leeds, Brett Ashley, Jeffrey M. Ritter, Sara McLaughlin Mitchell, and Andrew G. Long Alliance Treaty Obligations and Provisions, International Interactions 28: Lumsdaine, David Moral vision in international politics: The foreign aid regime, Princeton, NJ: Princeton University Press. Lundsgaarde, Erik The domestic politics of foreign aid: Societal interests, institutions and bureaucratic politics. Paper presented at annual meetings of the International Studies Association, Chicago, IL. Maizels, Alfred and Machiko Nissanke Motivations for aid to developing countries. World Development 12(9): Mavrotas, George and Bazoumana Ouattara Aid disaggregation and the public sector in aid-recipient economies: Some evidence from Cote D Ivoire. Review of Development Economics 10(3):

21 21 McDonnell, Ida, Henri-Bernhard Solignac Lecomte and Liam Wegimont, eds Public opinion and the fight against poverty. Paris, OECD. McKinlay, Robert and Richard Little A foreign policy model of US bilateral aid allocations. World Politics 30(1): Meernik, James, Eric Krueger and Steven Poe Testing models of foreign policy: Foreign aid during and after the Cold War. Journal of Politics 60(1): Meyers Eytan International Immigration Policy: A Theoretical and Comparative Analysis. Basingstoke, UK: Palgrave Macmillan. Mickolus, Edward International terrorism: Attributes of terrorist events, Dunn Loring, VA: Vinyard Software. Milner, Helen Why multilateralism? Foreign aid and domestic principal-agent problems. In Delegation and Agency in International Organizations. Hawkins, Darren, David Lake, Daniel Nielson and Michael Tierney, eds. Cambridge:Cambridge University Press. Noel, Alain and Jean-Phillipe Therien From domestic to international justice: The welfare state and foreign aid. International Organization 49(3):

22 22 Noel, Alain and Jean-Phillipe Therien Public opinion and global justice. Comparative Political Studies 35 (6): OECD. 2006a. Immigration data [get better citation]. OECD. 2006b. Oil data [get better citation]. Organization of Petroleum Exporting Countries (OPEC) About Us. At < Accessed 21 August Pape, Robert The strategic logic of suicide terrorism. American Political Science Review 97(3): Pew Research Center for the People and the Press/Pew Forum on Religion and Public Life Religion and Politics: Contention and Consensus. Washington, DC: Pew Research Center for the People and the Press/Pew Forum on Religion and Public Life. Rajan and Subramanian. Forthcoming. Aid and growth: What does the cross-country evidence really show? Review of Economics and Statistics.

23 23 Schiff, Maurice How trade, aid and remittances affect migration. Policy Research Working Paper Washington, DC: World Bank. Schraeder, Hook and Taylor Clarifying the foreign aid puzzle: A comparison of American, Japanese, French and Swedish aid flows. World Politics 50(2): Smillie, Ian, Henny Helmich, Tony German, Judith Randel, eds Public attitudes and international development coordination. Paris: OECD. Stanton, R., Teitelbaum, M., International migration and international trade. Discussion Paper No Washington, DC: World Bank. Teitelbaum, MS Immigration, refugees, and foreign policy. International Organization 38(3): Todaro, M.P., Maruszko, L., Illegal immigration and US immigration reform: A conceptual framework. Population and Development Review 13, U.S. Department of Energy Non-OPEC Fact Sheet. At < Accessed on 21 August World Bank World Development Indicators. Washington, DC: World Bank.

24 24 Appendix Table A Muslim Non Muslim Average Aid Total 252,000, ,000,000 Average Number of Projects Table B Undirected Regressions (1) (2) Dependent Variable ln(aid) project count Total Imports *** ( ) ( ) Total Exports *** ( ) ( ) Muslim majority 0.531* 0.130* (0.32) (0.072) Christian Majority *** (0.28) (0.066) Buddhist Majority (0.58) (0.13) Population *** ( ) ( ) Per capita GDP *** *** ( ) ( ) Years of Independence *** (0.0017) ( ) Arms imports ( ) ( ) Oil exports *** *** ( ) ( )

25 25 Terrorism *** (0.013) (0.0036) State sponsor (0.56) (0.12) Land area *** *** ( ) ( ) Arable land (% total) *** (0.010) (0.0024) Long term debt 0*** 0*** (0) (0) Exchange rate *** ( ) ( ) Polity 0.101*** *** (0.013) (0.0038) Selectorate size (0.30) (0.088) British Colony *** (0.34) (0.13) French Colony * (0.33) (0.13) Spanish/Portuguese colony ** *** (0.38) (0.18) Immigration * *** ( ) ( ) Constant 19.36*** 1.593*** (0.36) (0.083) Observations Number of countries Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Table C: Descriptive Statistics Aid Project Imports Exports Muslim Count (% total) (% total) min mean p

26 26 max 6.67E sd 6.99E Christian Buddhist Population Per capita Distance (miles) GDP min mean p max sd Alliances Arable land Long Term Exchange Polity (% total) debt Rate min E mean E p E max E sd E Years of Independence Arms transfers Oil flows Terrorism State sponsored min mean p max sd Colony Immigration Land Area Winning Coalition size min mean p max E+07 1 sd Table D Tobit Models (1) (2) Dependent variable ln(aid) ln(aid) Directed imports (% total) 1.671*** 1.791*** (0.33) (0.33) Directed exports (% total) 1.894*** 1.969*** (0.50) (0.51)

27 27 Muslim majority 0.329* (0.18) (0.19) Christian majority *** *** (0.17) (0.16) Buddhist majority * (0.40) (0.28) Population *** *** ( ) ( ) Per capita GDP *** *** ( ) ( ) Distance (miles) ( ) ( ) Alliance 0.616*** 0.738*** (0.23) (0.23) Years of independence *** *** (0.0010) ( ) Arms transfers *** *** ( ) ( ) Oil flows ( ) ( ) Terrorist attacks 0.112** 0.113** (0.044) (0.044) State sponsored (1.44) (1.44) Land area ** *** ( ) ( ) Arable land (% total) (0.0064) (0.0059) Long term debt 5.09e-11*** 5.05e-11*** (0) (0) Exchange rate * ** ( ) ( ) Polity 0.151*** 0.150*** (0.0077) (0.0077) Winning coalition size * (0.18) (0.18) Directed former colony * * (0.51) (0.51) Directed Immigration

28 28 ( ) ( ) Muslim*Immigration ** ( ) Constant 7.899*** 7.575*** (0.27) (0.27) Observations Number of dyads Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Table E Negative Binomial models (1) (2) Dependent variable project_count project_count Directed imports (% total) 0.493*** 0.492*** (0.039) (0.039) Directed exports (% total) 0.566*** 0.558*** (0.086) (0.086) Muslim majority 0.276*** 0.245*** (0.035) (0.033) Christian majority *** *** (0.031) (0.031) Buddhist majority 0.117** 0.120** (0.058) (0.058) Population *** *** ( ) ( ) Per capita GDP *** *** ( ) ( ) Distance (miles) *** *** ( ) ( ) Alliance 0.525*** 0.524*** (0.045) (0.045) Years of independence *** *** ( ) ( ) Arms transfers *** ***

29 29 ( ) ( ) Oil flows *** *** ( ) ( ) Terrorist attacks *** *** (0.0059) (0.0059) State sponsored (0.22) (0.22) Land area *** *** ( ) ( ) Arable land (% total) *** *** (0.0011) (0.0011) Long term debt 0*** 0*** (0) (0) Exchange rate *** *** ( ) ( ) Polity *** *** (0.0019) (0.0019) Winning coalition size * * (0.044) (0.044) Directed former colony *** *** (0.15) (0.15) Directed Immigration ( ) ( ) Muslim*Immigration ** ( ) Constant *** *** (0.053) (0.052) Observations Number of dyads Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1