The Public’s Foreign Aid Priorities: Evidence from a Conjoint Experiment

Abstract

Foreign aid is one of the few areas where Americans say the government should spend less. We leverage a unique conjoint experiment to assess how characteristics of an aid package, as well as characteristics of the targeted country, affect public support. We find that people are far more inclined to support economic aid than military aid and are disinclined to provide aid to undemocratic countries. We also find that people are more averse to providing aid—particularly economic aid—to countries in the “greater Middle East” than those countries’ other characteristics would suggest. These effects are comparable to those associated with substantial increases in the cost of the aid package, suggesting that public wariness of foreign aid is not rooted in a fundamental aversion to spending in this domain. Our findings offer new insights into the contours of public opinion regarding foreign aid.

Keywords

foreign aid conjoint experiment public opinion foreign policy

Foreign aid has the potential to improve conditions in targeted countries (e.g., Burnside & Dollar, 2000; Dreher et al., 2008; Mekasha & Tarp, 2013, but see De Mesquita & Smith, 2009), advance donor countries’ geopolitical interests (Alesina & Dollar, 2000; Dreher & Sturm, 2012; Nowak-Lehmann, 2009; Wang, 1996), and burnish the donor country’s image among foreign publics (Dietrich & Winters, 2015; Goldsmith et al., 2014). In line with this evidence, elites in the United States—which spends more on foreign aid than any other country in absolute terms (OECD, 2018)—tend to view foreign aid as money that is, on balance, well spent.¹ Yet foreign aid is one of a handful of policy domains where a substantial share of the American public believes the government should be spending less (Kates et al., 2019). These attitudes are likely to be consequential given that domestic audiences shape elites’ foreign policy choices (Fearon, 1994; Schultz, 2001; Slantchev, 2006; Tomz, 2007; Trager & Vavreck, 2011)—including those made regarding foreign aid (Eisensee & Strömberg, 2007; Heinrich, 2013; Heinrich et al., 2018; Nielsen, 2013; Tingley, 2010). Indeed, the U.S. public’s apparent aversion to foreign aid spending may, in part, explain why U.S. spending, although high in absolute terms, is quite low relative to other developed countries when calculated as a percentage of GDP.

One possibility is that the public is fundamentally opposed to this type of spending—many Americans may simply believe that government dollars should be spent domestically. However, it is also possible that the public does not understand how foreign aid dollars are spent or how much is spent. Surveys find that the public drastically overestimates the amount spent on foreign aid. Although foreign aid accounts for about 1 percent of the United States’ federal budget, in a 2014 poll the average respondent estimated that 26 percent of the budget is devoted to foreign aid—only 5 percent of respondents correctly gave a response of 1 percent. That same survey found that support for spending increased substantially when these misconceptions were corrected (DiJulio et al., 2015). Similarly, the public may fundamentally misunderstand what types of spending elites are referring to when they use the term “foreign aid” (Williamson, 2019) or may believe foreign aid dollars are spent on the wrong types of projects or countries.

Although researchers have examined the individual and contextual factors that affect broad support for foreign aid spending, we know far less about the features of aid proposals that make foreign aid spending more or less attractive to citizens. In this article we leverage a unique conjoint experiment included on a national survey that presented respondents with a series of pairs of hypothetical foreign aid packages and asked them which they preferred. We make two contributions that offer new insights into the contours of public attitudes about foreign aid.

First, our conjoint design allows us to construct a detailed description of the contours of the public’s foreign aid preferences (Bansak et al., 2016; Hainmueller et al., 2014). We assess how the features of an aid package (cost and purpose), as well as the attributes of the targeted country affect public support for that aid. Consistent with some recent work, our evidence casts doubt on the notion that the public harbors a blanket opposition to foreign aid spending—people care about more than the cost of an aid package (Blackman, 2018; Heinrich et al., 2018; Milner & Tingley, 2010). Our findings build on that literature by simultaneously assessing how much each of an array of factors affects support for specific aid proposals and how these effects compare to the effect associated with the cost of the aid. By way of preview, we find that people are substantially more inclined to support aid to countries where political freedoms are respected. People also exhibit a strong preference for packages that provide economic aid—especially economic aid that targets fundamental human needs (e.g., water sanitation and food assistance)—to those that provide military aid. Notably, consistent with recent findings suggesting a cross-party consensus on immigration priorities (Hainmueller & Hopkins, 2015), there appears to be substantial bipartisan consensus on foreign aid priorities.

Second, we leverage a novel aspect of our conjoint design to assess whether attitudes about foreign aid are shaped by citizens’ feelings about targeted countries that go beyond economic considerations and governance practices. For example, does the public feel that the U.S. has a special obligation to assist countries like Iraq and Afghanistan, where the United States pursued preemptive military strategies that destabilized those countries? Are there regions of the world where the public is particularly willing to provide aid? We answer these questions by assessing the extent to which public support for aid to top foreign aid recipients diverges from what would be expected given the actual characteristics of those countries. Our evidence indicates that people are distinctively averse to aid that targets countries in the “greater Middle East.”² Conversely, people appear to be more supportive of aid to poor countries in Africa and Latin American than those country’s characteristics would otherwise imply. Our data do not provide an avenue for identifying the mechanisms that drive these patterns. However, we note that they are consistent with existing evidence that large swaths of the U.S. public harbor unfavorable views towards Muslims, as well as with recent work that finds that the American public tends to view aid to Africa paternalistically, rather than through a lens of intergroup conflict.

Public Support for Foreign Aid

Recent work finds that public opinion regarding foreign policy is rooted in coherent principles (Aldrich et al., 2006; Hurwitz & Peffley, 1987; Jentleson, 1992; Kertzer & Zeitzoff, 2017, but see Zaller, 1992; Page et al., 1987) and that the relationship between public opinion on foreign affairs and policy-maker behavior is a two-way interaction where public opinion places certain constraints on leaders while elite attitudes play a reciprocal role, shaping public opinion (Heinrich et al., 2016, 2018; Jacobs & Page, 2005; Milner & Tingley, 2013b; Tingley, 2010). Studies of the correlates of support for foreign aid find that these attitudes often track cleavages on domestic issues (e.g., Bodenstein & Faust, 2017; Diven & Constantelos, 2009; Holsti, 2004; Milner & Tingley, 2010; Otter, 2003). Those who prefer a larger role for government in alleviating domestic poverty and spurring domestic economic growth tend to support the same sort of policies internationally. Foreign aid attitudes also appear to be shaped by economic conditions including personal income (Chong & Gradstein, 2008) and macro conditions (Heinrich et al., 2016). Individual-level factors including interpersonal trust, satisfaction with government, religiosity, media consumption, and a belief that the poor can escape poverty are also associated with variation support for foreign aid (Bayram, 2017; Chong & Gradstein, 2008; Paxton & Knack, 2012).

The question we pursue here is related to, but distinct from, this previous work. We are interested in the public’s foreign aid priorities rather than broad support for spending. Assuming some money is going to be spent on foreign aid, how does the public think those resources should be allocated? Given that surveys typically find low levels of support for foreign aid spending, all else equal we expect support for an aid proposal to decline as the price tag increases. We benchmark the effects of an array of factors to those associated with the cost of a proposal. Our approach builds on findings reported by Heinrich and Kobayashi (2020) and Blackman (2018). Like us, these studies use experimental designs to assess how various facets of foreign aid proposals affect people’s willingness to support specific aid packages. The Heinrich and Kobayashi (2020) study focused on willingness to provide aid to countries that engage in “nasty,” illiberal practices; the Blackman (2018) study explored how religious considerations affect support for aid. We corroborate and build on these findings in a number of ways.

First, we assess whether some of the core findings from these studies replicate using alternative approaches to describing political and economic conditions in recipient countries. Second, we consider how the purpose of the aid—what the money is to be used for—affects support, as well as whether the purpose of the package conditions the effects of the cost of the package or the characteristics of the recipient country. Third, our design allows us to assess whether the American public is more or less averse to providing aid to specific countries in the world than those countries’ characteristics would lead us to expect. Finally, we present descriptive analysis that examines the extent to which Democrats and Republicans differ in their foreign aid priorities.

Characteristics of Recipient Countries

We assess the effects of four attributes of recipient countries. The first three of these were chosen to reflect commonly cited justifications for foreign aid spending: promoting desirable political practices, promoting the donor country’s strategic interests, and providing aid to those in need (e.g., Griffin & Enos, 1970). The fourth country characteristic we present—population—was included to provide respondents with basic contextual information about the size of the country. Absent this information a respondent may have difficulty interpreting the importance of, for example, the fact that a country imports $200 million in U.S. goods in a year. This may strike respondents as a trivial amount if they are imagining a country like Pakistan (population 200 million), but substantial if they have a small country like Haiti (population 11 million) in mind.³

The first theoretically important country-level factor we consider is political freedom in the recipient country. Public support for democracy is high in the United States—as it is around the world (Wike et al., 2017). Consistent with existing evidence that foreign aid is often used a tool to promote democracy (Boschini & Olofsgård, 2007; Yuichi Kono & Montinola, 2009), there is some evidence that Americans oppose allocating foreign aid to countries with undemocratic or corrupt governments (Allendoerfer, 2017; Blackman, 2018; Boone, 1996; Paxton & Knack, 2012). For example, Heinrich and Kobayashi (2020) find that people are less willing to support aid to “nasty regimes” (though they find that this aversion is attenuated if the donor country takes steps to address illiberal aspects of the regime’s behavior). This said, there are at least two reasons levels of democracy in a targeted country may not substantially shape support for an aid proposal. First, other survey evidence suggests that few Americans view promoting and defending human rights in other countries as a priority (Brancati, 2014; Drezner, 2008; Wittkopf, 1986). Second, although reported support for democracy is generally high, recent work suggests that people’s commitment to democracy and democratizing may be soft and waning (Wike et al., 2017).

Our sample—which is descriptively representative of the U.S. adult population—also allows us to assess whether Democrats and Republicans give different weight to political freedoms in their assessments of foreign aid proposals. Existing observational work offers reason to expect such a difference. Specifically, Bodenstein and Faust (2017) find that those on the “right” are more likely to say that aid should be conditioned on recipient countries following “certain rules regarding democracy, human rights and governance.” We assess whether Republicans attach greater weight to political freedoms in prospective recipient countries than their Democratic counterparts.

The second country characteristic we consider is strategic interests. Heinrich and Kobayashi (2020) examine the effects of the recipient country’s cooperation in combatting terrorism and money laundering and find little evidence that these strategically relevant considerations affect public support. In contrast, Blackman (2018) finds that people are more willing to support aid to countries that are described as U.S. allies in the War on Terror. Here we instead focus on shared economic interests. We operationalize these shared interests in terms of annual U.S. exports to the recipient nation. There is some evidence that patterns of trade relations affect foreign aid policy. For example, there is some evidence that donor countries view importing goods from a developing country as a substitute for direct aid and that donors use aid to reinforce commercial ties with recipient countries (Lundsgaarde et al., 2010, 2007; Nowak-Lehmann et al., 2009).

There is also some evidence that economic considerations shape the public’s foreign policy attitudes. For example, many Americans appear to view protecting the jobs of American workers as a foreign policy priority (Wittkopf, 1986). Other work finds that union members and those who live in blue collar areas tend to view foreign aid as supporting economic competitors (rather than customers), while those living in areas where the economy is dependent on capital or white-collar jobs tend to be more supportive of foreign economic aid (Paxton & Knack, 2012). In a similar vein, recent work finds that foreign aid attitudes are affected by the local economic consequences of that aid (Christiansen et al., 2019; Heinrich et al., 2018). Finally, in her experiment, Blackman (2018) found that people were more likely to support aid to countries described as trade partners.

There is reason to suspect a partisan divide in how shared economic interests affect support for foreign aid proposals. For example, Milner and Tingley (2013b) asked respondents how they would allocate aid between two countries, one with low income that imported few U.S. goods and another with medium income that imports many U.S. goods. Republicans were more inclined to support aid to countries that import many U.S. goods, suggesting that Republicans are more inclined to support aid to countries where the U.S. has clear economic interests. However, this evidence does not offer an avenue for disentangling the effects of imports from the effects of income in the targeted country. Our experiment manipulates these two characteristics independently.

Finally, we consider the possibility that the public is more willing to support foreign aid that might be thought of as “charitable.” Specifically, we assess whether people are more inclined to support aid to poorer countries. Foreign aid spending is often justified on the premise that wealthy countries like the U.S. are well positioned to assist countries that are poorly positioned to “help themselves.” Thus, we might expect the public to be reluctant to provide aid to countries where standards of living are relatively high. This said, some evidence suggests that the public tends not to view assisting the foreign poor or improving living standards or abroad as a priority (Bansak et al., 2016; Drezner, 2008), perhaps because they fail to understand the nature of extreme poverty (van Heerde & Hudson, 2010). There is also some evidence that some view American military spending (and service as the “world’s policeman”) as a substitute for development aid (Paxton & Knack, 2012).

In terms of party differences, to the extent that attitudes about foreign aid track domestic policy attitudes we expect Republicans to be less enthusiastic about using direct government assistance to address poverty. A large body of literature finds sharp partisan differences in terms of support for government spending to alleviate poverty (e.g., Jacoby, 1994). Tingley (2010) further supports this intuition, finding that as national governments become more conservative, support for aid to low-income nations decreases. In short, we expect any negative relationship between income in the recipient country and support for aid to be weaker among Republicans than it is among Democrats.

Type of Aid

Turning to the purpose of the aid package, we focus primarily on the distinction between military (security) aid and economic aid. There are reasons to think each type of aid may appeal to the public. To the extent that the public views foreign policy matters through a “realist” lens (Drezner, 2008), military aid may be attractive in that it offers an avenue for addressing strategic security concerns without directly putting American lives at risk (Mueller, 1996). However, an alternative possibility is that the public is reluctant to provide military equipment, training, and supplies to foreign governments that may choose to use them in ways that fail to serve American interests. Consistent with this notion, survey evidence suggests that the public is substantially more supportive of spending on “efforts to improve health for people in developing countries” than spending on “foreign aid” (DiJulio et al., 2015).

We note that, as with the effects of economic ties, there is reason to expect Democrats and Republicans to respond differently to variation in the purpose of an aid package. Given that Republicans are less inclined to support government spending to address domestic social problems, there is reason to suspect that they will be disinclined to support spending to address problems that are similar in nature (if not in scale) abroad. In addition, Republicans tend to be more broadly hawkish than Democrats (Kane & Norpoth, 2017) and, thus, may be more attracted to spending on military aid.

Willingness to Aid Specific Countries

A novel feature of our experimental design is that we randomly assigned some participants to make choices between proposals that provided information about the features of the aid package and attributes of the targeted country discussed above, but did not name the recipient countries (No Names Condition). Others completed similar tasks that varied the cost and purpose of the aid packages, but presented the attributes of real countries (drawn from top recipients of U.S. aid), as well as the names of these targeted countries. Thus, we can leverage findings from respondents in the No Names Condition to assess the extent to which people’s inclination or aversion to aid packages that target a specific country diverges from what would be expected given that country’s characteristics. For example, Guatemala has a population of 15 million, a gross national income (GNI) per capita of $7,760, a Freedom House aggregate score of 54, and the U.S. exported roughly $6.5 billion dollars’ worth of goods to this country in 2017. The Names Condition allows us to assess if people are more or less willing to support aid to a country with these characteristics when we specify that the country is Guatemala, as opposed to an unnamed “generic” country with these same characteristics. Thus, we can shed light on whether there are some countries or regions of the world where people are distinctively eager (or averse) to providing aid—that is, where they are more or less willing to support aid than the country’s characteristics and the features of the aid package would otherwise suggest.

Our design does not allow us to unpack the mechanisms that may explain such patterns. However, one of two competing patterns seems particularly likely to emerge. First, Americans may feel obliged to provide support to regions and countries where the U.S. has intervened. This notion was popularized by Secretary of State Colin Powell, who said: “. . .if you get yourself involved—if you break a government, if you cause it to come down, by invading or other means, remember that you are now the government. You have a responsibility to take care of the people of that country.” (Gilsinan, 2015). If Americans adhere to this “you break it, you bought it” principle we would expect them to be distinctively inclined to provide aid to Iraq and Afghanistan, and perhaps to other countries in the “greater Middle East” that have been affected by the American interventions in Iraq and Afghanistan.

A competing—and perhaps more likely—possibility is that in the domain of foreign aid, attitudes about providing support to other countries are shaped by intergroup attitudes. Baker (2015) offers striking evidence of this tendency in an experiment that found that people are more supportive of aid when the recipients are of African descent than comparable targets in Eastern Europe. He shows that this pattern stems from a paternalistic outlook that is fostered by media coverage of black foreigners which tends to minimize signals of intergroup conflict. If foreign aid attitudes are shaped by intergroup attitudes, rather than concerns about fulfilling international obligations, we would expect people to be distinctively disinclined to support aid to countries in the “greater Middle East.” This pattern could be tied to high levels of anti-Muslim sentiment in the U.S (Panagopoulos, 2006; Doherty & Stancliffe, 2017), perceptions that Muslims are a threatening out-group (Davis, 2006), or a tendency to support policies—including foreign aid spending—that aid members of one’s in-group (e.g., Bansak et al., 2016; Blackman, 2018). In contrast, consistent with Baker’s evidence, we would expect people to be distinctively inclined to support aid to African countries (excluding those in North Africa that may be tied to the War on Terror, or Islam, in the public’s mind).

Conjoint Design

We test our expectations using a conjoint experiment. The design allows us to simultaneously consider how the array of factors discussed above affect support for a given aid package. Participants in our study were presented with 10 forced choice decision tasks. The tasks were introduced with the following text:

Next we’d like to ask you to think about your priorities when it comes to U.S. spending on foreign aid. On the next few pages we will present you with a series of ten (10) pairs of hypothetical foreign aid proposals. For each pair, please tell us which proposal you support more, even if you either like or dislike both proposals.

In each of the tasks that followed, respondents were presented with a choice between two foreign aid proposals. The proposals were presented in tabular format that included the cost and purpose of the package, as well as characteristics of the targeted countries. We present screenshots of example tasks from the No Names and Names Conditions in Figure 1.

Figure 1.

Example conjoint tasks. (a) No Names Condition. (b) Names Condition.

Three-fourths of respondents (n = 1,404) were randomly assigned to evaluate aid proposals where the targeted countries were not named (No Names Condition). In this condition, the characteristics of the countries were randomized (as we describe in more detail below). The remaining one-fourth of respondents (n = 483) completed tasks where the targeted countries were identified by name in one row of each task (Names Condition). Information about the purpose and cost of the package was randomized for these respondents, but they were presented with actual information about the named countries (e.g., Proposal A in Figure 1b lists the actual amount of U.S. exports to Ethiopia in 2017, as well as Ethiopia’s actual average annual income, population, and level of political freedom). Among those assigned to this latter condition, the twenty countries were sampled without replacement (i.e., each respondent saw a package targeting each country one time across the 10 tasks). These countries were selected because they were top overall recipients of U.S. aid in 2016 and received substantial military and economic aid.⁴ Thus, attitudes about aid to these countries may be particularly consequential. In addition, respondents are likely to view aid proposals targeting these countries as credible and relevant.

Each task listed six attributes of each of the two proposals. Two of these were tied to the features of the aid proposal: the cost and purpose. The purposes fall under one of the two types of aid that correspond to USAID coding practices: military and economic.⁵ In order to ensure that the costs we use reflect real-world funding practices, we consulted the U.S. Overseas Loans and Grants (Greenbook) file, available through USAID.⁶ This file lists individual grants and loans dating back to 1946. We restricted the data to positive funding obligations⁷ from fiscal years 2011 to 2016 that targeted the 20 top-recipient countries discussed above. In the conjoint tasks we draw costs at random from a uniform distribution ranging from $1 million (44th percentile of grants recorded by USAID) to $100 million (93rd percentile).⁸

The remaining four attributes provided information about the characteristics of the targeted country: the country’s population, level of political freedom, per capita income, and trade with the U.S. In the Names Condition we present country populations (in millions) as recorded in the CIA World Factbook. We operationalize political freedom using Freedom House’s Aggregate civil liberties scores (0–100). The wealth attribute is operationalized in terms of 2016 purchasing power parity (PPP) adjusted Gross National Income (GNI) per capita figures drawn from the World Bank.⁹ Finally, we vary the amount of U.S. exports to the country in 2017.¹⁰ These values are reported in Supplemental Appendix Table A1.

We emphasize that in the Names Condition, rather than receiving randomized values for the four attributes of the country, respondents saw real values for that country on population, political freedom, per capita income, and trade with the U.S. Thus, below we are able to leverage data from this experimental branch, in concert with data from the No Names Condition to assess how people’s feelings about specific countries (or regions of the world) affect their willingness to support aid to a particular country. For example, to what extent does willingness to support a $10 million economic aid package for Iraq deviate from willingness to support an identical package to an unnamed country with characteristics that mirror those of Iraq?

In the No Names Condition we draw country attributes at random. One approach would be to simply draw values from uniform distributions ranging from the minimum values we observe in the top 20 aid recipient countries discussed above to the maximums. The problem with this approach is that the distributions of country attributes used in the Names Condition are not at all uniform. Of particular concern are the distributions of the income and trade variables. For example, on the trade variable if we drew values from a uniform distribution ranging from the minimum observed value ($11 million [South Sudan]) to the highest ($12 billion [Colombia]), half of our cases would present respondents with recipient countries where U.S. imports exceeded $6 billion in 2017. Only three of the 20 top recipient countries meet this criterion (Colombia, Israel [$11bil], and Guatemala [$6.5bil]). In fact, median exports to these top 20 countries was slightly over $1 billion. The problem is similarly acute with the GNI per capita attribute: GNI per capita in Israel is $37,330, but the next highest GNI is $17,240 (Iraq). Half of the top recipient countries we use have a GNI per capita of less than $5,000 per year.

An alternative to drawing from a uniform distribution would be to use the values observed among the top 20 recipient countries from the Names Condition in the No Names Condition. The key limitation of this approach is that, as the above discussion suggests, these values are distributed in a way that leaves substantial gaps. Thus, simply using these values would undermine our ability to consider how people respond to a more continuous range of country attributes. We adopt a middle ground, starting with the distribution of values observed among the top 20 recipient countries and adding noise to create distributions of values with densities that closely track the contours of the values observed in the country data, but better cover the full range of potential values (see Supplemental Appendix Figure A1 and note for further details).

Analysis

Participants were recruited by Survey Sampling International to complete the study online in late February, 2018. Our sample includes 1,880 respondents who completed all ten conjoint tasks (only 3 respondents failed to complete all ten tasks; see Supplemental Appendix Table A2 for descriptive characteristics of the sample). Our approach to analyzing our data follows guidelines proposed by Hainmueller, Hopkins and Yamamoto (2014). We stacked the data to treat each proposal as our unit of analysis. This yields a sample of 37,600 observations (1,880 respondents × 10 tasks × 2 proposals per task)—28,040 in the No Names Condition and 9,560 in the Names Condition. We use Ordinary Least Squares models and cluster standard errors at the respondent level.

We begin by focusing on respondents in the No Names Condition. Our first model regresses our outcome measure—set to 1 if the respondent chose the proposal and 0 if they did not—on our experimentally manipulated proposal characteristics, restricting the sample to those in the No Names Condition. Following Bansak, Bechtel and Margalit (2019) we bin our continuous treatment variables in our analysis. These estimated treatment effects are indicated by the markers in Figure 2 (regression models are presented in column [1] of Supplemental Table A3 in the Appendix).

Figure 2.

Primary treatment effects (No Names Condition): Markers indicate estimated treatment effects. Estimates are from regression model presented in Table A3 of the Supplemental Appendix. Whiskers are 95% confidence intervals.

Four of the six attributes significantly affected support for the proposals: cost, purpose of the aid, level of political freedom in the targeted country, and the population of the targeted country. However, the effects tied to population were trivial in size and barely reach conventional thresholds of statistical significance. Relative to the baseline category of countries with populations of up to 25 million, respondents were 0.9% points more likely to support an aid package that targeted a country with a population ranging from 25 to 80 million people (p = .189) and 1.6% points more likely to support a package targeting a country with a population larger than 80 million (p = .046).

In contrast, the other effects we find are each substantial and easily meet conventional thresholds for statistical significance (p < .001 in each case). Respondents were approximately 4.5% points less likely to select a proposal that cost $26–50 million as opposed to the reference group range of $1 to $25 million. The effect grows in a seemingly linear fashion as the cost increases: the estimated effect of the cost being $51 to $75 million is 7.8% points and 12.2% points for packages that cost between $76 and $100 million.

Our bins for levels of political freedoms correspond roughly to Freedom House’s “Not Free,” “Partly Free,” and “Free” categories.¹¹ Respondents were approximately 5.9% points more likely to select a packages that targeted a “Partly Free” country (score from 36–70) than an otherwise-similar, but “Not Free” country. The effect of a country being “Free” (71–100) rather than “Not Free” was 9.6% points, approaching the size of the effect of a package costing $76 to $100 million, rather than $1–25 million.

The most pronounced effect is tied to the purpose of the aid package: respondents were 18.7% points less likely to support proposals that would fund military spending, as opposed to economic aid. As we show in Figure A2 of the Supplemental Appendix, unlike the effects of cost and levels of freedom, the type of aid appears to have been dispositive for a sizeable share of respondents—about 30 percent chose the economic aid package over a military package across all tasks where the packages differed on this dimension.¹²

The effects of the remaining two attributes are not statistically significant in spite of our large sample. The coefficients on the U.S. exports indicators are each small and fall well short of conventional thresholds of statistical significance (p = .395 and .535), as are the coefficients on the annual income indicators (p = .854 and .646)—null findings that fail to corroborate the statistically significant effects reported by Blackman (2018). One explanation for these null findings may be that the effects of these predictors were non-linear. For example, respondents may have been sensitive to variation in annual income among poor countries, but not among wealthier countries.

In Figure A4 of the Supplemental Appendix we consider this possibility by reporting kernel-weighted local polynomial regression lines—which do not require us to specify a particular functional form—that illustrate the bivariate relationship between each of our continuous treatment dimensions and the probability that a proposal would be selected. We find little evidence of pronounced nonlinear effects.¹³ Given the patterns evidence in Figure A4, for clarity and ease of exposition, in the additional analysis we present below, we use the full, continuous treatment variables rather than collapsing cases into discrete bins as we did for the analysis reported in Figure 2.¹⁴

Before proceeding, we more closely consider the effects of the treatment that most strongly affected choices: the purpose of the aid package. Rather than relying on the single indicator for Military Aid, we estimated a model that includes indicators for 11 of the 12 purpose descriptors used in the choice tasks (as well as our linear treatment measures). We report the findings graphically in Figure 3. The figure shows that each of the three forms of military spending we used led to significantly lower levels of support than any form of economic aid. Responses to the various types of economic aid appear to fall roughly into three tiers. The pattern suggests that support for spending grows as we move from proposals that would fund projects that may improve the structural health of a country’s economy to proposals that directly target individuals’ basic needs.

Figure 3.

Purpose effects (No Names Condition): Estimates from regression model presented in Table A6 of the Supplemental Appendix. Whiskers are 95% confidence intervals. Reference category = Military equipment.

In the first tier are proposals that would fund transportation and communications infrastructure. Each was approximately 10% points more likely to be selected than their military aid counterparts. In the second tier are disaster preparedness, energy generation, and farming supplies. These were associated with approximately 19% point increases in the likelihood that a proposal would be selected (relative to proposals that would fund military equipment). We note that these packages target facets of society that are arguably more critical to survival than communications or transportation projects, but less fundamental than, say, food and water. Finally, in the third tier are proposals that would directly target basic human needs: food, water, education, and health. These proposals were each approximately 25% points more likely to be selected than otherwise-similar proposals that would fund military equipment.

Country Preferences

Next, we consider data from respondents in the Names Condition, adding those cases to those from the No Names Condition. In column (2) of Table A7 of the Supplemental Appendix we estimate a model predicting whether a respondent chose a given aid package, adding these cases. We use the continuous core treatment variables and add indicators for each country used in the Names Condition (cases in the No Names Condition serve as the reference group). As a point of comparison, in column (1) we present a model excluding cases from the No Names Condition. The coefficients on the core treatment variables are essentially unchanged by the inclusion of the cases from the Names Condition, suggesting that the weight people assigned to each consideration was not substantially different in the Names Condition. In Figure 4 we present the coefficients on each of the 20 country indicators. These estimates capture how the country names affected choices above and beyond what would be expected based on the country characteristics we provided and the features of the aid proposal.¹⁵

Figure 4.

Country effects: Estimates from regression model presented in column (2) of Table A7 of the Supplemental Appendix. Whiskers are 95% confidence intervals. Reference category = Country not named.

The findings run counter to the expectation that Americans feel a special obligation to Iraq and Afghanistan. Respondents were substantially less likely to select proposals that targeted these countries than those in the No Names Condition were to select proposals targeting countries with similar characteristics. The other two country names that led to significantly lower levels of support were Lebanon and Pakistan. Lebanon is located in the Middle East and, although Pakistan is not technically in the Middle East, like Afghanistan it is part of the “greater Middle East” and has played a central role in U.S. foreign policy in the region. We suspect, though cannot demonstrate, that most Americans would consider both Afghanistan and Pakistan to be part of the Middle East. The coefficients on the three other Middle Eastern countries we used in our design—Egypt, Israel, and Jordan—are also negative, though not statistically significant.¹⁶

At the other end of the spectrum are a mix of poor African countries (Uganda, Democratic Republic of Congo, and Kenya), as well as two fairly poor countries located in the Americas (Guatemala and Haiti). People expressed significantly more support for aid to these countries than their economic and political characteristics would predict.¹⁷ This pattern is consistent with existing evidence that the U.S. public tends to view aid to Africa paternalistically Baker (2015).

Differences by Party

Finally, we consider the possibility that Democrats and Republicans respond differently to the features of foreign aid packages by estimating a model regressing our outcome measure on the six core treatment dimensions and the vector of country name indicators, including indicators for respondents who, through the standard branching party identification questions, identified as Democrats (including “leaners”) and “pure” Independents. We interacted each of these party indicators with the full vector of treatment variables. In Figure 5 we present estimated treatment effects from this model for Democratic and Republican respondents.¹⁸

Figure 5.

Conditioning effects of partisanship: Markers in Panel A indicate the estimated effect of a change from the 25th to 75th percentile in terms of the observed characteristics of the top foreign aid recipients used in the Names Condition (except for cost which shows estimated effect of a change from $1 to $100 million). Estimates from model interacting treatments with partisanship (see Table A8). Whiskers are 95% confidence intervals. Reference category = Country not named.

In Panel A we find that, consistent with the notion that Republicans are generally less inclined to support foreign aid spending (e.g., Bodenstein & Faust, 2017), Republican respondents were more sensitive to the cost of the aid package. The other point of partisan divergence is tied to the purpose of the aid. Although, respondents from both parties were substantially less likely to support military—as opposed to economic—aid, the estimated effect is larger among Democrats (–22.8) than among Republicans (–12.2). Beyond this, we find substantial consensus across parties. Democrats and Republicans attach similar value to levels of political freedom in targeted countries and we find only scattered evidence of differences regarding whether specific countries are particularly worthy (or unworthy) of aid. Of the 20 country name treatments, we only find significant party differences in the effects of two: Pakistan and South Sudan (see Panels B–D). In each case, Democrats (and pure independents) responded more favorably to aid proposals targeting these countries than their Republican counterparts. Given the number of tests conducted, as well as the fact that these two countries appear to share little in common, we are reluctant to over-interpret this pattern. In short, consistent with recent findings suggesting a cross-party consensus on immigration priorities (Hainmueller & Hopkins, 2015), our evidence suggests that in spite of the intense party polarization that pervades the contemporary political landscape in the contemporary United States, Democrats and Republicans priorities in the realm of foreign aid are remarkably similar.

Discussion

Elites in the United States tend to support foreign aid as a relatively low cost way to promote American interests abroad. However, surveys consistently show that a large share of the American public supports cutting foreign aid. Recent work suggests that the public’s foreign policy attitudes are rooted in stable principles and have the power to constrain policy-makers under certain circumstances. We presented fine-grained evidence regarding the contours of the public’s foreign aid preferences using a research design that existing work finds has the potential to mirror real-world decision-making dynamics remarkably effectively (Hainmueller et al., 2015).

Consistent with the evidence reported in some recent work, our findings paint a picture of a public that views foreign aid priorities through a lens that is more idealistic than transactional (Blackman, 2018; Heinrich & Kobayashi, 2020). Although we might expect military aid to more directly serve strategic national interests, the public is far more supportive of aid that addresses economic concerns. Similarly, we find no evidence that the public is more inclined to support aid to countries that import large quantities of U.S. goods. Instead, they are concerned with whether money is going to countries where political freedoms are protected. Democrats and Republicans appear to largely agree on these priorities. We emphasize that the magnitudes of the effects we find are comparable to those associated with the cost of the proposed package, reinforcing the claim that people care about more than total spending in this domain. Our findings offer little support for the expectation that the public is more willing to support aid to impoverished countries.

We also leveraged a novel feature of our research design that allowed us to assess whether people’s willingness to support aid to some particular countries diverges from what would be expected given salient characteristics of that country. We find that, in spite of recent U.S. interventions in the region, the public is averse to providing aid to countries in the greater Middle East. This aversion is particularly pronounced when it comes to economic aid (see endnote 17). Instead the patterns that emerge are consistent with a dynamic where public support for foreign aid is tied to a paternalistic desire to help others, so long as they are not perceived as members of a threatening out-group (Baker, 2015). Finally, our analysis yields only scattered and modestly-sized partisan differences in terms of foreign aid priorities.

Like all research, the present work has limitations. For example, our design also does not offer insight into the consequences of all of the dimensions of foreign aid. For example, we did not vary which entity would control spending of the foreign aid packages we presented (for evidence on this front see Milner & Tingley, 2013a). Our data also do not offer an avenue for identifying the specific mechanisms underlying the patterns we found. For example, while our evidence is broadly consistent with the findings reported by Baker (2015), we cannot demonstrate that the patterns we find regarding which countries people are most (and least) inclined to aid are driven by intergroup attitudes. In addition, as discussed above, our efforts to assess the independent effects of providing a country’s name are limited by the fact that the data generating process used in the No Names Condition did not mirror the forces that determine the features of countries in the real world. In a similar vein, our design only allows us to assess the effects of naming a subset of countries. Future work could assess the effects of naming a broader range of countries, perhaps including wealthy countries where we would expect the public to view providing foreign aid as particularly inappropriate or unnecessary.

These limitations aside, our results offer new insight into the American public’s foreign aid attitudes. In some ways, the public’s priorities appear to be reflected in actual policy. For example, countries that squelch democratic freedoms do tend to receive less aid than freer countries (Alesina & Weder, 2002; Reinsberg, 2015). The fact that people are far more averse to military aid than to economic aid is also reflected in existing policy practices. With the exceptions of a surge of post-World War II military aid to Europe and a burst of military aid to Asian countries in the Vietnam War era, the U.S. has consistently allocated more aid to economic projects (see Supplemental Appendix Figure A6). In concert with recent work (Williamson, 2019), our findings suggest that some of the public’s reported preference for reducing foreign aid may stem from a pattern where many Americans conflate spending on foreign aid and military spending. This said, our findings also show that the public is most averse to aiding many countries in the greater Middle East that currently receive substantial aid. The public’s aversion to spending in a region that many associate with U.S. military intervention may serve as a cautionary tale: although many elites may feel bound by Colin Powell’s “Pottery Barn principle,” the public shows little indication that they feel obliged to provide open-ended support to countries where the U.S. military chooses to engage.

Supplemental Material

Supplemental_material – Supplemental material for The Public’s Foreign Aid Priorities: Evidence from a Conjoint Experiment

Supplemental material, Supplemental_material for The Public’s Foreign Aid Priorities: Evidence from a Conjoint Experiment by David Doherty, Amanda Clare Bryan, Dina Hanania and Matthew Pajor in American Politics Research

Footnotes

Declaration of Conflicting Interests

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

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

David Doherty

Amanda Clare Bryan

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biographies

David Doherty is an associate professor of Political Science at Loyola University Chicago where he teaches courses on American politics, public opinion, campaigns, and statistics. His research addresses an array of questions about the forces that affect political attitudes and behaviors.

Amanda Clare Bryan is an associate professor of Political Science at Loyola University Chicago. She teaches courses on American politics, constitutional law, and the politics of the courts. Her research focuses primarily on the United States Supreme Court and how the Court is influenced by, and influences, American political attitudes.

Dina Hanania is a graduate of Loyola University Chicago. She is a humanitarian global advocacy practitioner at CARE International. She has a MA in conflict tesolution from the Georgetown University School of Government where she focused on the intersections between international humanitarian aid and the role of the UN system in conflict transformation. She currently focuses her work and research on gender in emergencies.

Matthew Pajor is a graduate of Loyola University Chicago and a Loyola President’s Medallion recipient.

References

Aldrich

J. H.

Gelpi

Feaver

Reifler

Sharp

K. T.

(2006). Foreign policy and the electoral connection. Annual Review of Political Science, 9(1), 477–502.

Alesina

Dollar

(2000). Who gives foreign aid to whom and why? Journal of Economic Growth, 5(1), 33–63.

Alesina

Weder

(2002). Do corrupt governments receive less foreign aid? American Economic Review, 92(4), 1126–1137.

Allendoerfer

M. G.

(2017). Who cares about human rights? Public opinion about human rights foreign policy. Journal of Human Rights, 16(4), 428–451.

Baker

(2015). Race, paternalism, and foreign aid: Evidence from U.S. public opinion. American Political Science Review, 109(1), 93–109.

Bansak

Bechtel

M. M.

Margalit

(2019). Why austerity? The mass politics of a contested policy. Mimeo, Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3486227

Bansak

Hainmueller

Hangartner

(2016). How economic, humanitarian, and religious concerns shape European attitudes toward asylum seekers. Science, 354(6309), 217–222.

Bayram

A. B.

(2017). Aiding strangers: Generalized trust and the moral basis of public support for foreign development aid. Foreign Policy Analysis, 13(1), 133–153.

Bjørnskov

(2014). Types of foreign aid. SSRN Working Paper.

10.

Blackman

A. D.

(2018). Religion and foreign aid. Politics and Religion, 11(3), 522–552.

11.

Bodenstein

Faust

(2017). Who cares? European public opinion on foreign aid and political conditionality. JCMS: Journal of Common Market Studies, 55(5), 955–973.

12.

Boone

(1996). Politics and the effectiveness of foreign aid. European Economic Review, 40(2), 289–329.

13.

Boschini

Olofsgård

(2007). Foreign aid: An instrument for fighting communism? The Journal of Development Studies, 43(4), 622–648.

14.

Brancati

(2014). The determinants of US public opinion towards democracy promotion. Political Behavior, 36(4), 705–730.

15.

Brenan

(2018). Americans like Canada most, North Korea least. Gallup.com. https://news.gallup.com/poll/228293/americans-canada-north-korea-least.aspx

16.

Burnside

Dollar

(2000). Aid, policies, and growth. American Economic Review, 90(4), 847–868.

17.

Chong

Gradstein

(2008). What determines foreign aid? The donors’ perspective. Journal of Development Economics, 87(1), 1–13.

18.

Christiansen

Heinrich

Peterson

T. M.

(2019). Foreign policy begins at home: The local origin of support for US democracy promotion. International Interactions, 45(4), 595–616.

19.

Davis

D. W.

(2006). Negative liberty: Public opinion and the terrorist attacks on America. New York: Russell Sage Foundation.

20.

de Mesquita

B. B.

Smith

. (2009). A political economy of aid. International Organization, 63(2), 309–340.

21.

Dietrich

Winters

M. S.

(2015). Foreign aid and government legitimacy. Journal of Experimental Political Science, 2(2), 164–171.

22.

DiJulio

Firth

Brodie

(2015). Data note: Americans’ views on the US role in global health. Kaiser Family Foundation, 23. https://www.kff.org/global-health-policy/poll-finding/data-note-americans-views-on-the-u-s-role-in-global-health/

23.

Diven

P. J.

Constantelos

(2009). Explaining generosity: A comparison of US and European public opinion on foreign aid. Journal of Transatlantic Studies, 7(2), 118–132.

24.

Doherty

Stancliffe

(2017). Interpreting and tolerating speech: The effects of message, messenger, and framing. American Politics Research, 45(2), 224–255.

25.

Dreher

Nunnenkamp

Thiele

(2008). Does aid for education educate children? Evidence from panel data. The World Bank Economic Review, 22(2), 291–314.

26.

Dreher

Sturm

J.-E.

(2012). Do the IMF and the world bank influence voting in the UN general assembly? Public Choice, 151(1–2), 363–397.

27.

Drezner

D. W.

(2008). The realist tradition in American public opinion. Perspectives on Politics, 6(1), 51–70.

28.

Eisensee

Strömberg

(2007). News droughts, news floods, and US disaster relief. Quarterly Journal of Economics, 122(2), 693–728.

29.

Fearon

J. D.

(1994). Domestic political audiences and the escalation of international disputes. American Political Science Review, 88(3), 577–592.

30.

Gilsinan

(2015). The pottery barn rule: Syria edition. The Atlantic, September 30. https://www.theatlantic.com/international/archive/2015/09/the-pottery-barn-rule-syria-edition/408193/

31.

Goldsmith

B. E.

Horiuchi

Wood

(2014). Doing well by doing good: The impact of foreign aid on foreign public opinion. Quarterly Journal of Political Science, 9(1), 87–114.

32.

Griffin

K. B.

Enos

J. L.

(1970). Foreign assistance: objectives and consequences. Economic development and cultural change, 18(3), 313–327.

33.

Hainmueller

Hangartner

Yamamoto

(2015). Validating vignette and conjoint survey experiments against real-world behavior. Proceedings of the National Academy of Sciences of the United States of America, 112(8), 2395–2400.

34.

Hainmueller

Hopkins

D. J.

(2015). The hidden American immigration consensus: A conjoint analysis of attitudes toward immigrants. American Journal of Political Science, 59(3), 529–548.

35.

Hainmueller

Hopkins

D. J.

Yamamoto

(2014). Causal inference in conjoint analysis: Understanding multidimensional choices via stated preference experiments. Political Analysis, 22(1), 1–30.

36.

Heinrich

(2013). When is foreign aid selfish, when is it selfless? Journal of Politics, 75(2), 422–435.

37.

Heinrich

Kobayashi

(2020). How do people evaluate foreign aid to ‘nasty’ regimes? British Journal of Political Science, 50(1), 103–127.

38.

Heinrich

Kobayashi

Bryant

K. A.

(2016). Public opinion and foreign aid cuts in economic crises. World Development, 77, 66–79.

39.

Heinrich

Kobayashi

Long

(2018). Voters get what they want (when they pay attention): Human rights, policy benefits, and foreign aid. International Studies Quarterly, 62(1), 195–207.

40.

Holsti

O. R.

(2004). Public opinion and American foreign policy. Ann Arbor, MI: University of Michigan Press.

41.

Hurwitz

Peffley

(1987). How are foreign policy attitudes structured? A hierarchical model. American Political Science Review, 81(4), 1099–1120.

42.

Jacobs

L. R.

Page

B. I.

(2005). Who influences U.S. foreign policy? American Political Science Review, 99(1), 107–123.

43.

Jacoby

W. G.

(1994). Public attitudes toward government spending. American Journal of Political Science, 38(2), 336–361.

44.

Jentleson

B. W.

(1992). The pretty prudent public: Post post-Vietnam American opinion on the use of military force. Inter-national Studies Quarterly, 36(1), 49–73.

45.

Kane

J. V.

Barabas

(2019). No harm in checking: Using factual manipulation checks to assess attentiveness in experiments. American Journal of Political Science, 63(1), 234–249.

46.

Kane

J. V.

Norpoth

(2017). No love for doves? Foreign policy and candidate appeal. Social Science Quarterly, 98(5), 1659–1676.

47.

Kates

Michaud

Kirzinger

Brodie

(2019). Where does public opinion stand on the U.S. role in global health? Kaiser Family Foundation. https://www.kff.org/global-health-policy/poll-finding/where-does-public-opinion-stand-u-s-role-global-health/

48.

Kertzer

J. D.

Zeitzoff

(2017). A bottom-up theory of public opinion about foreign policy. American Journal of Political Science, 61(3), 543–558.

49.

Lundsgaarde

Breunig

Prakash

(2007). Trade versus aid: Donor generosity in an era of globalization. Policy Sciences, 40(2), 157–179.

50.

Lundsgaarde

Breunig

Prakash

(2010). Instrumental philanthropy: Trade and the allocation of foreign aid. Canadian Journal of Political Science, 43(3), 733–761.

51.

Mekasha

T. J.

Tarp

(2013). Aid and growth: What meta-analysis reveals. The Journal of Development Studies, 49(4), 564–583.

52.

Milner

H. V.

Tingley

D. H.

(2010). The political economy of U.S. foreign aid: American legislators and the domestic politics of aid. Economics and Politics, 22(2), 200–232.

53.

Milner

H. V.

Tingley

(2013a). The choice for multilateralism: Foreign aid and American foreign policy. The Review of International Organizations, 8(3), 313–341.

54.

Milner

H. V.

Tingley

(2013b). Public opinion and foreign aid: A review essay. International Interactions, 39(3), 389–401.

55.

Mueller

(1996). Fifteen propositions about American foreign policy and public opinion in an era free of compelling threats, National Convention of the International Studies Association. International Studies Association, April 16–20.

56.

Nielsen

R. A.

(2013). Rewarding human rights? Selective aid sanctions against repressive states. International Studies Quarterly, 57(4), 791–803.

57.

Nowak-Lehmann

D. F

Martínez-Zarzoso

Klasen

Herzer

(2009). Aid and trade–a donor’s perspective. The Journal of Development Studies, 45(7), 1184–1202.

58.

OECD. (2018). Net ODA (indicator). Online Resource. https://data.oecd.org/oda/net-oda.htm

59.

Ottaway

Carothers

(2004). The greater middle east initiative: Off to a false start. Policy Brief No. 29, (Technical Report). Carnegie Endowment for International Peace, Washington, DC.

60.

Otter

(2003). Domestic public support for foreign aid: Does it matter? Third World Quarterly, 24(1), 115–125.

61.

Page

B. I.

Shapiro

R. Y.

Dempsey

G. R.

(1987). What moves public opinion? The American Political Science Review, 81(1), 23–43.

62.

Panagopoulos

(2006). The polls-trends: Arab and Muslim Americans and Islam in the aftermath of 9/11. Public Opinion Quarterly, 70(4), 608–624.

63.

Paxton

Knack

(2012). Individual and country-level factors affecting support for foreign aid. International Political Science Review, 33(2), 171–192.

64.

Reinsberg

(2015). Foreign aid responses to political liberalization. World Development, 75, 46–61.

65.

Schultz

K. A.

(2001). Looking for audience costs. Journal of Conflict Resolution, 45(1), 32–60.

66.

Slantchev

B. L.

(2006). Politicians, the media, and domestic audience costs. International Studies Quarterly, 50(2), 445–477.

67.

Tingley

(2010). Donors and domestic politics: Political influences on foreign aid effort. The Quarterly Review of Economics and Finance, 50(1), 40–49.

68.

Tomz

(2007). Domestic audience costs in international relations: An experimental approach. International Organization, 61, 821–840.

69.

Trager

R. F.

Vavreck

(2011). The political costs of crisis bargaining: Presidential rhetoric and the role of party. American Journal of Political Science, 55(3), 526–545.

70.

van Heerde

Hudson

. (2010). ‘The righteous considereth the cause of the poor’? Public attitudes towards poverty in developing countries. Political Studies, 58(3), 389–409.

71.

Wang

T. Y.

(1996). U.S. foreign aid and UN voting: An analysis of important issues. International Studies Quarterly, 43(1), 199–210.

72.

Wike

Simmons

Stokes

Fetterolf

(2017). Globally, broad support for representative and direct democracy. Technical report Pew Research Center.

73.

Williamson

(2019). Public ignorance or Elitist Jargon? Recon-sidering Americans’ overestimates of government waste and foreign aid. American Politics Research, 47(1), 152–173.

74.

Wittkopf

E. R.

(1986). On the foreign policy beliefs of the American people: A critique and some evidence. International Studies Quarterly, 30(4), 425–445.

75.

Wong

Karni

Cochrane

(2019). Trump administration drops proposal to cut foreign aid after intense debate. New York Times, August 22.

76.

Yuichi Kono

Montinola

G. R

. (2009). Does foreign aid support autocrats, democrats, or both? The Journal of Politics, 71(2), 704–718.

77.

Zaller

J. R.

(1992). The nature and origins of mass opinion. Cambridge University Press.

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.

0.00 MB

0.31 MB