Donor Government Ideology and Aid Bypass

Donor Government Ideology and Aid Bypass Abstract How do domestic politics in the donor country affect foreign aid policy? Scholars have analyzed the impact that political parties and domestic institutions play in shaping aid effort. Building on this previous work, we examine the impact of domestic politics on the channels chosen for aid distribution. In this paper, we explore whether and how the ideological composition of donor governments affects the method of delivering aid to recipient states. Specifically, we argue that more liberal governments are more inclined to provide aid via nongovernmental organizations (NGOs), while more conservative governments prefer to give aid through direct bilateral government-to-government channels. These choices, we argue, reflect the different goals that right and left parties pursue through their allocation of foreign aid. Our results indicate that left governments are more inclined to channel aid through NGOs in an attempt to directly implement poverty alleviation strategies. Alternatively, right wing governments place a greater emphasis on channeling aid through recipient state governments to promote the economic interests of their constituents and the geopolitical interests of the state. There is an extensive literature on the factors that drive donor states to give foreign economic aid. Much of this work focuses on the characteristics of recipient states and relationships with donor states to explain how much aid a state receives (see Meernik, Krueger, and Poe 1998; Alesina and Dollar 2000; Lai and Morey 2006; Bueno de Mesquita and Smith 2007). By comparison, less work explores how donor states’ domestic political environments shape the aid allocation process. Prior work has largely focused on the influence of donor government ideology and support for aid in aggregate terms (e.g., Thérien and Nöel 2000; Tingley 2010; Milner and Tingley 2010), and almost no work has been aimed at understanding the link between donor government ideology and the specific methods of aid delivery. Considering the effects of government ideology on aid bypass highlights a critical source of variation in aid flows—both within and across countries—that accounts for how billions of dollars in aid is allocated each year. It also suggests that foreign aid is a more flexible and valuable political tool for governments than exiting literature has recognized, the manipulation of which requires no change in aggregate aid budgets—only in aid delivery channels. We argue that the ideology of donor governments affects the method of aid delivery. Parties across the left-right spectrum cater to fundamentally different societal constituencies with different interests, goals, and worldviews (e.g., Hibbs 1977; Carney et al. 2008; Graham, Haidt, and Nosek 2009; Hirsh et al. 2010; Milner and Tingley 2013a).1 The desire to retain office should lead governing parties to use the tools at their disposal, like foreign aid, to pursue policy goals that benefit key constituents. The left-right orientation of a party necessarily relates to the sorts of policies and goals that constituents want their representatives to pursue. Because some aid delivery channels are more or less effective at achieving certain goals, the electoral benefits of allocating aid through particular channels will vary according to the party and its goals. Earlier work has explored the use of multilateral organizations as a way to “bypass” donor governments (Hoadley 1980; Milner 2006). Here we examine another bypass channel, focusing on how ideology affects whether aid is delivered through NGOs versus public-sector channels.2 We argue that liberal governments should allocate a greater proportion of foreign aid through NGOs, specifically. Development NGOs are more likely to prioritize direct poverty alleviation, focusing on the needs of citizens in the recipient country rather than on the priorities of their governments. Conservative governments should be more likely to favor government-to-government transfers, which are thought to be a better tool for advancing the immediate economic and political goals of the donor country, despite being less directly focused on long-term poverty reduction (Nunnenkamp, Weingarth, and Weisser 2009). We find that more liberal donor governments tend to allocate greater shares of aid through NGOs, while more conservative governments appear to favor government-to-government transfers. We also find that the effects of government ideology are conditioned by the nature of the donor state's relations with the recipient state. Specifically, we find that more conservative governments allocate relatively less aid through NGOs when recipient states have close relations with donor states. However, ideology has no effect on the allocation of aid when the donor and recipient have poorer relations. The substantive implications of this research are great. First, our study contributes to the growing literature on how domestic politics influence donor state decisions over the allocation of aid (e.g., Fleck and Kilby 2006; Tingley 2010; Milner and Tingley 2013a; Dietrich 2016). We depart from most existing research by linking partisan and ideological interests to how aid is allocated across public and private sector channels. This is the central way in which we build on previous work that has examined how donor government politics affect bypass decisions. For example, whereas Dietrich (2016) focuses on variation across countries, our analysis explores how variation in governments’ political preferences within and across countries affects the relative allocation of aid between government-to-government channels and private sector channels. We find that the left-right orientation of donor governments has a clear effect on the channels through which aid is distributed. Somewhat paradoxically, we show that right parties actually prefer public sector channels, while left governments have a greater preference for allocating aid through NGOs in the private sector. These findings highlight the fact that while conservative governments tend to allocate less money to foreign aid, they do provide some aid. Conservatives and liberals both see political value in foreign aid; they simply use aid differently. Understanding the donor-side factors that drive the allocation of aid to NGOs can also be of enormous import for both NGOs and recipient states. Over the twelve-year period of our analysis, Canada allocated an average of 20 percent of its ODA. The United States allocated approximately 10 percent (OECD 2015). This translates into billions of dollars in the aggregate, much of which is allocated to NGOs and private contractors (Dietrich 2013, 690). If the ideology of donor governments influences the methods of aid delivery, then these values may shift considerably over time. A change from a liberal to a more conservative government may translate into millions in lost funding for NGOs, even as aggregate aid levels remain unchanged, given conservative governments’ preference for public sector channels. By extension, this may have profound effects on grassroots development projects in the recipient state. The Domestic Politics of Donor States The conventional wisdom is that liberals support aid and conservatives oppose it. While there is some truth to this, a more nuanced look demonstrates that governments across the political spectrum utilize aid differently and provide aid in distinct ways. Political ideology should certainly be influential in matters of foreign assistance. Ideology, often conceptualized as a single underlying dimension, reflects beliefs and attitudes about the role of government in the economy. Ideology acts as a constraint on policy choices by “bundling” various issues together (Gabel and Huber 2000; Laver and Budge 1992). Attitudes about the role of government in the economy help place political parties on the right-left spectrum (Laver and Garry 2000). Thérien (2002) argues that the conflicting priorities of right and left parties have been integral to the development of foreign assistance programs. Given their support of social welfare programs and government intervention in their domestic economies, scholars often suggest that liberals should be more supportive of foreign aid. Preferring less government intervention in the economy, conservatives should be opposed to aid programs. Studies of foreign aid flows from Organization for Economic Cooperation and Development (OECD) donor states concluded that leftist governments correlate indirectly with higher levels of foreign aid; the effect largely felt through the expansion of the welfare state and the externalization of liberal norms through foreign policy (Nöel and Thérien 1995; Thérien and Nöel 2000). Tingley (2010) observes that among OECD donor states, conservative governments are characterized by a lower aid effort than liberal governments, but this effect is limited to aid flows to developing countries and multilateral organizations. Brech and Potrafke (2014) provide evidence of a partisan divide in bilateral grant aid. These partisan divisions appear to exist in the general public as well. Cross-national public opinion polls have found that liberals are more supportive of foreign economic aid, and conservatives less so (Nöel, Thérien, and Dallaire 2004; Paxton and Knack 2012; Milner and Tingley 2013a). Studies examining the United States illustrate similar preference differences between liberals and conservatives but also that the parties have different goals for giving aid. Fleck and Kilby (2010) demonstrate that conservative administrations have allocated less money to foreign aid than liberal administrations. The priorities of US aid also appear to vary by ideology. Fleck and Kilby (2006) show that Republicans emphasize commercial factors like the promotion of US exports, while Democrats emphasize humanitarian and developmental factors. In addition, analyses of congressional voting behavior have also found that conservative legislators are more likely to oppose measures that would increase economic aid spending (Milner and Tingley 2010, 2011), which is likely driven by the conservative belief that foreign aid is an obstacle to market efficiency and welfare improvement in both donor and recipient states (Thornton 2002). Such patterns hold at the mass level as well—Milner and Tingley (2013a, 395) find that “more conservative voters respond more positively when aid is helpful to the US economy, and more liberal voters respond more favorably to aid when it targets the neediest groups abroad.” This literature provides a useful basis for our theoretical argument, but the specific electoral mechanisms are worth expanding upon. Assuming leaders in office wish to stay in office, aid provides them with a tool that can be used to generate benefits for constituents. Left and right governments tend to represent different core economic constituencies, with the left tending to represent the interests of lower income groups and labor, and the right representing higher income groups and capital (Hibbs 1977). Drawing on the Heckscher-Ohlin and Stolper-Samuelson models of trade, Milner and Tingley (2010) argue that foreign aid transfers from wealthy donor countries to poorer countries represent a transfer of capital that can have distributional effects on individuals within the donor country that vary according to individuals’ factor endowments—holders of capital will see increased incomes while holders of labor will see no or negative returns from aid transfers. As described by Milner and Tingley (2010, 207), “The economics of aid . . . constitute a part of donor nations’ commercial strategy to secure larger trade benefits . . . The economic benefits . . . are largest when recipient countries import goods in which donor countries have a comparative advantage in production.” Thus aid transfers can increase the returns to the producers of capital-intensive goods in wealthy developed states—groups that tend to be represented by rightist parties—but only if aid promotes greater consumption of goods produced by the donor state. The economic incentive for right parties to use aid to promote commercial goals is clear—the benefits go to constituents typically associated with right parties. However, this dynamic does not help us to understand the preferences of more liberal individuals who should suffer economic losses in the form of the higher tax rates used to fund aid. Leftist parties are associated with support for larger aid budgets, but research on public attitudes towards aid indicates that, whereas conservatives are more likely to want aid to go to promoting economic and geopolitical goals, liberals tend to want aid to go to promoting development among the poor and needy. Conservative respondents demonstrate a preference for giving a larger share of the aid to states where the United States has strong commercial ties, and liberal respondents are more inclined to allocate a greater proportion of aid to needier states (Milner and Tingley 2013a). Paxton and Knack (2012) similarly find that conservatives and individuals expressing more critical attitudes towards the poor are less likely to support aid. Studies in political psychology help explain these different priorities. Carney, Jost, Gosling et al. (2008) find that liberals and conservatives differ on several key personality traits, with liberal ideology correlating with an increased emphasis on compassion and egalitarianism (Hirsh et al. 2010), as well as less acceptance of inequality (Graham, Haidt, and Nosek 2009). Thus, while liberals may not reap direct economic benefits from aid, they may be willing to endure some marginally higher tax burdens if aid is used to promote humanitarian and development goals abroad. If we expect politicians to represent the preferences of their constituents, it follows that left (right) governments will funnel more aid to development (commercial) goals. Aid Delivery Channels, Delegation, and Ideology To understand how ideology affects government decisions to delegate through NGOs, we first draw on previous research looking at aid delegated through another bypass channel—multilateral organizations. We then link this literature to more recent work on aid delegation to NGOs. Research on the choice to delegate to multilateral organizations can help us to better understand government choices to allocate aid through bypass channels and the role of ideology in this process. As with multilateral aid, delegating to development NGOs involves some loss of control over how aid is implemented, thereby depriving governments of the ability to leverage aid into policy concessions. These literatures can help us understand how government ideology affects the choice to allocate aid through bypass channels. Why do states delegate control over the distribution of foreign aid to multilateral organizations? Assuming multilateral organizations have a degree of independence from donor governments, how they choose to administer aid will likely diverge to some extent from a donor's preference.3 One possibility is that delegation to multilateral organizations may benefit some domestic economic interests in the donor state (McLean 2015). Another explanation grounded in donor domestic politics is that governments delegate to send signals to their domestic audiences (Hoadley 1980; Milner 2006). Studies have shown that even where publics tend to support foreign aid in general, citizens are often skeptical aid is effective at promoting development and reducing poverty in recipient states or believe that it is used for corrupt purposes (Nöel, Thérien, and Dallaire 2004; Milner 2006).4 This skepticism risks constraining policymakers’ access to a valuable foreign policy tool, as popular support is rarely on the side of increasing aid budgets (Nöel, Thérien, and Dallaire 2004). Given the focus of multilateral organizations on promoting development, delegating aid to multilateral organizations can serve as a signal to donor governments’ publics that aid is promoting development rather than being siphoned off by corrupt bureaucrats or repressive regimes (Hoadley 1980; Milner 2006). This argument is useful but assumes that all governments and donor audiences are motivated to improve development in the recipient state. However, politicians in donor states are often motivated by a variety of goals beyond the promotion of development, such as fostering commercial expansion or rewarding allies—goals that have been linked to partisanship and ideology (e.g., Fleck and Kilby 2006; Milner and Tingley 2013b,a). This research tends to indicate that ideology does affect these decisions. Milner and Tingley (2013b) examine public support in the United States for delegating aid to multilateral organizations, arguing that donors make a tradeoff between greater control of aid and greater burden-sharing. The development-oriented focus of multilateral organizations implies lower agency loss for liberals, who prefer to allocate aid to development-oriented ends. Given their focus on promoting commercial outcomes and security goals, the agency loss from delegating should be greater for conservatives. If the goal is to obtain concessions from the recipient government, the management of aid funds by multilateral organizations severely diminishes states’ capacity to manipulate aid to that end. As discussed above, liberal and conservative voters do emphasize different goals that are associated with aid. There are strong parallels between the work on delegating through multilateral organizations and more recent work on aid bypass. Dietrich (2013) argues that the quality of governance in the recipient state influences how donors choose to channel aid. Dietrich finds that donors are indeed sensitive to governance issues in the recipient state, allocating more aid to nonstate actors where the quality of governance is poor and more aid directly to governments where governance is better. By bypassing poorly governed states, Dietrich argues that development aid can be allocated to more productive ends. Using several indicators of recipient state quality of governance, Acht, Mahmoud, and Thiele (2015) similarly find that poor governance correlates with higher levels of bypass aid. Dietrich (2016) asserts that the effect of governance is conditioned by cross-national differences in donor states’ emphasis on the role of the state in the delivery of public goods and services; In some recipient countries, the state plays a larger role in the delivery of services, while other recipient countries place a greater emphasis on the role on private markets in providing aid. Where governance is poor and the possibility of aid capture is high, donor governments that would otherwise allocate more aid through government-to-government channels should shift their allocation more heavily towards more market-oriented channels. Work on aid bypass shows that when donors are concerned about poor governance, aid allocated through NGOs can function similar to aid allocated through multilateral organizations.5 For example, Fruttero and Gauri (2005) model the relationship between donor governments and NGOs as a principal-agent game. The official aid providers (the principals) have incomplete information about the projects being implemented by the NGOs (the agents), but future NGO funding is contingent on perceptions concerning the success or failure of current projects. The belief is that NGOs are less bound by the policy agendas of donor governments, and by working on the ground in recipient countries, NGOs are more aware of the needs of the population and the ways in which aid funds can be utilized to meet those needs. If implementation of aid by NGOs is less distorted by self-interest than those made directly by donor nations, then aid may be more effective at poverty alleviation (Nancy and Yontcheva 2006). NGOs may also be able to circumvent corrupt or inefficient local governments to deliver aid directly to those in need (Riddell, Bebbington, and Peck 1995). Thus, delegating to NGOs may also alleviate problems of credibility with domestic publics, if private organizations are perceived as being less corrupt and wasteful than their government counterparts. Given their preference for pursuing development-oriented outcomes, liberal governments should allocate relatively more aid through NGOs. By definition, development NGOs are concerned with promoting economic growth and human development. NGOs are also viewed as having more direct contact with the people that aid is intended to help, and are thought to be more effective and efficient vehicles for aid delivery (Bebbington and Riddell 1995; Nancy and Yontcheva 2006; Koch et al. 2009; Tender 1982). The specific means of accomplishing these goals may vary across organizations, but many development NGOs approach developmental issues through grass roots activities, like providing micro-credit loans and grants to needy populations.6 Thus NGOs should be better equipped to pursue the types of projects that liberal voters and governments prefer. On the other hand, given their desire to advance commercial and geopolitical goals, we expect conservative governments to allocate less aid through NGOs. Foreign aid can be used to buy policy concessions and can help stabilize the political leadership in the recipient state (Bueno de Mesquita and Smith 2007). The fact that they are private organizations and not a part of the recipient state's government, combined with the emphasis that many NGOs place on grass roots development projects, suggests that development NGOs are not positioned to translate aid dollars into substantial influence over recipient state policies. Critics suggest that these sorts of financial transfers allow “bad” governments to remain in power longer than they otherwise would be able to do (Easterly 2001; Radelet 2006). However, if the goal is to buy favorable policies with aid, then the quality of recipient state governance should matter less and the incentives to pursue bypass channels should be mitigated.7 A possible objection to this argument is that it contradicts the attitudes we typically associate with left and right parties. Given their emphasis on free markets, one might argue that conservative governments should favor NGOs. There are a few key reasons why we believe this is not the case. Our argument is that the relative allocation of aid between public and private sector channels is a means to an end for politicians, not the end itself. Simply put, NGOs are not in a position to advance the kinds of economic and geopolitical goals that conservative governments prioritize. There are indeed numerous conservative NGOs to choose from, most NGOs are closely associated with more liberal political and economic goals. Eade (2004, 12) notes that while a range of NGOs exists, “more often than not, development NGOs are in some way involved in transferring resources from societies which have plenty to those who have little.” For example, Catholic Relief Services and FHI-360, two groups regularly listed among USAID's top vendors, have dramatically different attitudes about HIV/AIDS prevention and sex education (USAID 2013). While many faith-based development groups have more conservative values, the objectives of these groups do not square neatly with the commercial and geopolitical interests of conservative governments. Clarke (2007) argues that faith-based organizations, while highly active in more rural areas of developing states, were largely disconnected from development policy debates in donor and recipient states until the 1990s. In the case of the United States, President George W. Bush oversaw numerous changes intended to better integrate the activities of faith-based organizations into US development policy (Clarke 2007, 82–83). These organizations have largely remained dedicated to poverty alleviation, with the added dimension of conducting missionary work in the process. The fact that they are still primarily concerned with pursuing development-oriented goals makes them of less use to conservative governments, who are pursuing economic and geopolitical goals with aid. Evidence indicates that development NGOs tend to focus their efforts on the needy and that these efforts have a positive effect on development outcomes. Studies of NGO aid-giving find that development NGOs respond to need in determining where to deploy their resources (Nancy and Yontcheva 2006; Koch et al. 2009). Evidence indicates that their presence correlates positively with the development of human capital and economic growth (Murdie and Kakietek 2012). Though there is some risk in inferring preferences from outcomes, it would appear that development NGOs are investing resources with the foremost priority of promoting development. Accordingly, we expect conservative governments to allocate more aid to public-sector channels relative to NGO channels. What aid conservative governments do allocate to NGOs may be funneled to conservative NGOs over more liberal organizations, but this has no bearing on the more general weighting of public-sector versus NGO aid. Table 1. Summary relationships for ideology and aid Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  View Large Table 1. Summary relationships for ideology and aid Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  View Large Table 1 provides a brief summary of the expected relationships between ideology, the aggregate amount, and the type of aid. Our primary hypothesis is as follows: Hypothesis 1:Conservative governments should allocate less aid through bypass channels as compared with liberal governments. While ideology may have an independent effect on aid allocation, there are reasons to expect characteristics of the relationship between the donor and the recipient, or characteristics of the recipient state, to condition how ideology affects aid bypass decisions. Similar to Dietrich (2016), we argue that the inclinations of donor governments towards public versus private allocation channels may be conditional upon a variety of factors. We focus herein on six different policy areas/variables: 1) Exports; 2) Foreign policy similarity; 3) Alliance relations; 4) Recipient human rights conditions; 5) Recipient development; and 6) Recipient governance. Variables 1–3 capture aspects of how salient the recipient state is to the donor state in terms of economic and strategic relations. Variables 4–6 capture aspects of what we can broadly conceive of as indicators of recipient state governance. These variables speak to the salience of the underlying goals that motivate the basic differences between left and right parties that we discussed above. First, we expect elements of the dyadic relationship in question to condition the effect of ideology on aid allocation. Consider exports from the donor to the recipient, for example. Let us assume there are two recipients, State A and State B, and that the donor exports goods worth $10 billion per year to State A and $10 million per year to State B. Our basic theory suggests that a shift from a liberal to conservative government should lead to an increase in government-to-government aid relative to NGO aid but does not differentiate between these two cases. Given that commercial interests are stronger between the donor and State A versus State B, we expect the shift to a conservative government will produce a larger relative shift towards government-to-government aid in the case of State A than State B. Because conservatives prefer to allocate aid in such a way as to advance commercial interests, it stands to reason that this shift should be greater where the value of those commercial relationships is greater and the potential returns to government-to-government aid are higher. We expect similar dynamics when considering security-related issues. Since aid can be used as a tool to buy policy concessions or support friendly governments, conservative governments should seek to allocate more funding through public sector channels to states with which they share alliance ties or states that have similar policy preferences. However, when the donor and recipient are not allied, or when they are not closely aligned on global issues, liberal and conservative governments should be more constrained in their willingness to allocate aid to the recipient government. For example, the differences between liberals and conservatives in the United States over aid channels should be dampened when the would-be recipient is a country like Iran as compared to a country with less antagonistic relations. This leads us to our first conditional hypothesis: Hypothesis 2:The negative effect of more conservative governments on bypass aid should be stronger where economic and security ties between the donor and recipient states are stronger. The quality of governance in the recipient state should also lead liberal governments to allocate even greater amounts of aid through bypass channels (Acht, Mahmoud, and Thiele 2015; Dietrich 2013, 2016). However, given the different underlying attitudes towards inequality and suffering, we expect liberals and conservatives to respond differently (e.g., Graham, Haidt, and Nosek 2009). Given that liberal governments are more concerned with ensuring that aid is used to promote development and poverty alleviation, any indication that the government is going to use aid for corrupt or abusive purposes may give liberal governments pause when allocating aid through public sector channels, leading them to reduce the amount of public sector aid that they are willing to give as compared to a less corrupt or less abusive state. Similarly, liberal governments’ emphasis on poverty reduction should also lead them to allocate more aid through bypass channels as need within the recipient state increases as it indicates an increased urgency of direct poverty-alleviation efforts. Low income may also indicate that recipient state policies are largely ineffective at promoting growth and meeting the needs of the populous. On the other hand, conservative governments, with their focus on promoting economic and security goals, which are better achieved by appealing to governmental sources, may be less concerned with matters of poor governance or need, since providing funds directly to the people does little to purchase greater government support for donor policies. This leads us to our second conditional hypothesis: Hypothesis 3:The negative effect of more conservative governments on bypass aid should be larger where the recipient state's domestic environment is characterized by poor governance. Data and Research Design The data for our dependent variable come from the OECD's Creditor Reporting System (CRS) database (OECD 2015).8 Information regarding foreign aid channels is available for all DAC countries from 2003 to 2014.9 The aid data include total aid as well as aid flows by the following channels: 1) traditional government-to-government aid, 2) NGO aid, 3) multilateral aid, 4) public-private partnerships, and 5) a catch-all “other” category. For the purposes of this project, our attention is focused on NGO aid and traditional government-to-government ODA.10 The dependent variable, defined in Equation 1, is the ratio of aid from donor i to recipient j in time t that is channeled through NGO and civil society organizations to aid channeled in the more traditional government-to-government manner. As the numerator grows, the donor is channeling more aid through NGOs. As the denominator grows, the donor is channeling more aid directly to the recipient government. The aid values used to create the ratio are raw dollar amounts, not budgetary percentages. Accordingly, the ratio is highly skewed, and so we use the natural log of this ratio as the dependent variable in the following analyses.11  \begin{equation} {Y_{ijt}} = {\rm ln} ({\it Ratio})_{ijt} = \left( \frac{\hbox{NGO Aid}_{ijt}}{\hbox{Government-to-government Aid}_{ijt}} \right) \end{equation} (1) We opt for this ratio measure because we conceptualize these policy choices as kinds of substitutes. Figure 1 provides some cursory support for this approach. There is a strong negative correlation between the mean proportion of each DAC member's aid budget allocated to NGO aid and the proportion allocated to public sector aid (r = −0.64). The fact that there are five possible budget categories suggests that this relationship is not simply imposed by the presentation of budgetary proportions. For example, countries like Australia, Canada, Norway, Switzerland, and the Netherlands have relatively low levels of both bilateral and NGO aid. The balance often goes to multilateral aid. The results remain unchanged when we use alternative measures of the dependent variable (see online appendix). Figure 1. View largeDownload slide Mean proportion of aid by channel and donor state. Donor states are members of the OECD Development Assistance Committee (DAC) and are ordered according to least NGO aid to most NGO aid as a proportion of aid given. Figure 1. View largeDownload slide Mean proportion of aid by channel and donor state. Donor states are members of the OECD Development Assistance Committee (DAC) and are ordered according to least NGO aid to most NGO aid as a proportion of aid given. Our primary independent variables are created using the Comparative Manifestos Project (CMP) data (Volkens et al. 2011). The CMP-coded manifestos provide information about political parties in all donor countries. Information from CMP is used in two different ways to measure ideology. First, following Tingley (2010), we calculate ideological scores (using the method described in Gabel and Huber 2000) for party attitudes on the economy. These scores range from 0–10. The CMP data include information on parties’ positions on free enterprise, market regulation, and various forms of government involvement in the economy. The economic scores enable us to clearly delineate between the left and right preferences on economic issues, while the full left-right index includes a broader range of issues. In the following models we include the left-right economic ideology measure for only those parties who are currently in government. Second, we follow Tingley (2010) in calculating the left-right measure using party attitudes across all issues, not just economic issues.12 We also control for the economic circumstances of the donor country. We include a measure of economic growth (World Bank 2014). Countries experiencing economic growth are more likely to provide foreign aid than those with struggling economies. Previous research has also linked the expansiveness of the donor state's welfare institutions to foreign aid (Nöel and Thérien 1995; Thérien 2002). Larger social welfare institutions may generally indicate a higher baseline rate of government intervention in the economy, apart from the ideology of a particular governing party, which may correlate with more public sector aid. Accordingly, we include the percentage of GDP allocated to domestic welfare spending (OECD 2013). Lastly, we control for the total amount of aid the donor gives. Larger donors may be better able to leverage large aid budgets into political concessions and may be more likely to emphasize public-sector aid (Milner 2006; Hoadley 1980; Milner and Tingley 2013b). These data come from the OECD CRS (OECD 2015).13 In addition, we consider characteristics of the recipient country. The economic state of the recipient country is an important predictor of the type of aid a country receives. Leftist governments are more likely to focus on need, and so we believe needier countries may be targeted with NGO aid more frequently. For this reason, we include GDP per capita in order to account for recipient need (World Bank 2014). We include the natural log of this variable given its skewed distribution. The political characteristics of the recipient country are also relevant. In Assessing Aid, the World Bank suggests that aid effectiveness can be strengthened by targeting poor countries with strong institutions. Following (Dietrich 2013), there are also reasons to expect that regime type will influence the aid channel chosen. We include the Polity2 variable to control for the recipient state's regime type (Marshall, Jaggers, and Gurr 2014). DAC donors may be inclined to give more assistance to more democratic countries. The human rights record of a recipient may also influence how aid is distributed. We utilize the physical integrity rights scores from the CIRI Human Rights Data Project (Cingranelli, Richards, and Clay 2014). Dietrich (2013) has also shown that the quality of governance in the recipient state influences decisions over how aid is allocated. We include an indicator of the recipient state's quality of governance using data obtained from the World Governance Indicator's project (Kaufmann, Kraay, and Mastruzzi 2015). Lower values represent poorer governance and higher values represent better quality governance. We also include a dichotomous indicator of civil conflict in the recipient state (Gleditsch et al. 2002). Donor governments may be wary of channeling aid to recipient governments that are engaged in a civil conflict, as aid may be misappropriated and used to bolster the government's security forces. The relationship between the donor and recipient also influences the aid relationship. We control for the level of political alignment measured by similarity in UNGA voting (Strezhnev and Voeten 2012). We also control for defense pacts (Gibler 2009). Where there are strong political and security relationships we anticipate that donors will allocate less aid through NGOs and more aid through traditional government-to-government channels. We also include a dichotomous indicator for whether or not the recipient state is a former colony of the donor state.14 We expect donor and recipient government ties to be stronger in such cases, prompting donors to allocate more aid through government-to-government channels. As with security ties, donors may channel more aid directly to recipient governments where the donor and recipient have strong economic ties. We include a measure of exports from the donor the recipient, $$( \frac{{\it Exports}_{ijt}}{GDP_{it}} ) \times 100$$, where i is the donor country, j is the recipient country, and t denotes the year. Given that the volume of donor exports varies, this measure allows us to capture the relative importance of recipient markets across donor countries. These data are drawn from the IMF Direction of Trade Statistics (International Monetary Fund 2014).15 Summary statistics can be seen in Table A2. We estimate our primary models using OLS and a generalized linear model with random effects. Independent variables are lagged by one time period as the anticipated effects are not instantaneous. We also include a lagged dependent variable to control for serial correlation.16 The basic models are presented in Table 2.17 Table 2. Ratio of NGO aid to public sector aid   (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998      (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Table 2. Ratio of NGO aid to public sector aid   (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998      (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Analysis and Discussion Table 2 shows the results of our base models. We first estimate a simple OLS model (Model 1) to demonstrate that there is a clear negative bivariate correlation between more conservative governments (those scoring higher on the ideology scale) and the ratio of NGO aid to public-sector aid, in-line with our primary hypothesis.18 Models 2 and 3 add the control variables. Model 2 presents an additional full OLS model, and Model 3 is estimated using GLS with random effects. Across both models the donor government's ideology score continues to have a strong and statistically significant negative coefficient, indicating that more conservative governments give less money to NGOs relative to government-to-government aid. Of the variables capturing characteristics of the donor country, only the size of the donor's welfare state has a consistent and statistically significant coefficient. As the overall amount of the donor state's social welfare spending increases, we see relatively less aid allocated through NGOs and more aid allocated through traditional government-to-government channels. A larger welfare state may indicate that the donor government has greater institutional capacity to manage aid distribution. This finding may be picking up a dynamic similar to that discussed by Dietrich (2016). Of the variables measuring the relationship between the donor and the recipient, none are statistically significant. Exports, political similarity, colonial history, and defense pacts do not appear to be systematically related to the choices made about how aid is channeled. We discuss this issue further below. More democratic recipient states receive more aid channeled through NGOs and civil society groups relative to public-sector aid. In some ways, this is surprising, given the findings of (Dietrich 2013). Presumably democratic governments would be more trustworthy and capable than their autocratic counterparts, which might translate into more aid channeled directly to the government. However, we see here that democratic governments receive relatively less public sector aid. The quality of governance in the donor state correlates negatively with the relative amount of NGO aid and is in line with Dietrich (2013, 2016). Presumably donor states are more comfortable channeling aid through the recipient state's government when that government has a proven track record of responsible behavior. Physical integrity rights correlate negatively with the dependent variable. This suggests that as government respect for physical integrity rights improves, that state receives more public sector aid relative to NGO aid. Governments with less respect for physical integrity rights see more of their country's aid allocated through bypass/NGO channels. This finding makes sense, and is similar to the governance measure, as NGOs give donor states a way to continue to support development projects in countries where governments may misappropriate aid funds. The models in Table 2 provide a useful first cut. However, while the coefficient on ideology is negative as we expect, this can result from a number of scenarios: 1) conservative governments may cut NGO aid while leaving public sector aid untouched; 2) NGO aid may be untouched while public-sector aid is increased; 3) both aid types may be increased or decreased, but these changes are less pronounced for NGO aid than for public-sector aid. To address these possibilities, we use seemingly unrelated regression models to estimate two separate equations for the logged aggregate values of NGO aid and public sector aid in Table 3. To conserve space we present only the coefficient for the ideology variable.19 Model 1 is the basic model while Model 2 includes recipient fixed-effects. Table 3. Seemingly unrelated models   (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858      (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858    Standard errors in parentheses. *p ≤ 0.10 **p ≤ 0.05, ***p ≤ 0.01. View Large Table 3. Seemingly unrelated models   (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858      (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858    Standard errors in parentheses. *p ≤ 0.10 **p ≤ 0.05, ***p ≤ 0.01. View Large The results of these models support our initial findings. In both models the donor ideology variable has a negative and statistically significant coefficient, indicating that more conservative governments correlate negatively with the level of NGO aid. In the public sector aid models we find a positive and significant coefficient between donor ideology and aid. The magnitudes of these coefficients are also informative. Using Model 1 as an example, a one-unit increase in the donor ideology score translates into a 19 percent cut in NGO aid but leads to a 15 percent increase in public sector aid. The mean bilateral flows in this estimation sample are $5.51 million for NGO aid and $37.36 million for public sector aid. These coefficients would translate into a cut of $1.05 million in NGO aid and an increase of $5.60 million in public sector aid. Conditional Effects As we discuss previously, there is reason to believe that characteristics of the recipient state or its relationship to the donor state condition the effect of changing donor ideology. We evaluate these possibilities in Table 4. Using the ratio measure defined in 1, we examine the results of six additional models where we interact donor ideology with 1) donor exports to the recipient, 2) political similarity between the donor and the recipient, 3) security ties between the donor and the recipient, 4) the recipient state's quality of governance, 5) the recipient state's human rights record, and 6) the recipient state's level of development. Table 4. Ratio models with conditioning variables   (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998      (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Table 4. Ratio models with conditioning variables   (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998      (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large In general, these models resemble those in Table 2. The factors that have the clearest conditioning effect on decisions concerning aid channels are those associated with the relationship between the donor and recipient. Specifically, the only statistically significant interactions are for the factors tied to political and security relationships between the donor and recipient states.20 To better understand these interactive effects, we plot the marginal effects below. Figure 2 shows the marginal effect of donor ideology across the range of the UN Affinity score (Panel A) and the marginal effect of an increase in the UN Affinity score variable across the range of the ideology variable (Panel B). Plotting the marginal effect of both constituent variables helps us to better understand how the interaction of these variables shapes aid allocation patterns.21 Figure 2. View largeDownload slide Marginal effect of ideology (Panel A) and UN Affinity scores (Panel B). 95 percent confidence intervals shown. The ranges on the X-axes reflect the ranges observed in the estimation sample. The dashed line shows the distribution of the given variable within the estimation sample. Figure 2. View largeDownload slide Marginal effect of ideology (Panel A) and UN Affinity scores (Panel B). 95 percent confidence intervals shown. The ranges on the X-axes reflect the ranges observed in the estimation sample. The dashed line shows the distribution of the given variable within the estimation sample. In Panel A there is no statistically significant effect associated with an increase in the ideology variable at the low end of the X-axis. However, as the Affinity score increases we see a statistically significant negative effect emerge, indicating that a negative effect for ideology emerges at higher values of Affinity. When the donor and recipient state have poor relations, there is no evidence that ideology affects the balance between NGO aid and public-sector aid. However, when dealing with friendlier states we see that conservative governments allocate less aid through NGOs relative to public-sector channels. Furthermore, Panel A indicates that there is a statistically significant increase in the magnitude of this effect within the significant range (i.e., between Affinity = –0.2 and Affinity = 1). Panel B shows the marginal effect of an increase in the Affinity score across the range of the donor ideology variable. When looking at more liberal governments (i.e., low ideology score) we see a positive and statistically significant effect, indicating that an increase in the Affinity score produces an increase in NGO aid under liberal governments. However, as governments become more conservative (i.e., higher ideology scores) increases in Affinity eventually have a statistically significant negative effect, indicating that more conservative governments respond to improved relations by increasing government-to-government aid relative to NGO aid. These results are consistent with our expectations. When donor and recipient governments are more closely aligned it may be safer for conservative donor governments to choose to allocate more aid directly to the recipient government. When the donor has poor relations with a recipient state to begin with there is likely little incentive to alter aid channels, as concessions may not be possible. Governments are also likely to be wary of giving funds directly to a hostile government. Any development aid that is allocated to the recipient will likely go through NGOs so as not to risk funds falling into a potentially hostile government's hands. Similarly, if liberal governments are genuinely more concerned with development-oriented goals, then it seems reasonable that they would not respond to changes in political alignment in the same way that conservative governments would, given that the latter relies more on government concessions. The interaction between ideology and the alliance variable was the only other interaction with a statistically significant coefficient in Table 4. To conserve space we present the plot of these effects in the online appendix. Briefly, when looking at nonallied dyads, we find more conservative governments decrease aid to NGOs in favor of government-to-government aid. When looking at allied dyads, we see that an increase in the ideology score correlates with an increase in the relative proportion of aid that is channeled through NGOs. These results are reinforced when we consider the marginal effect of the alliance variable. Here, the effect is negative and significant for liberal governments, indicating liberals give allies less NGO aid and more direct government-to-government aid. These result runs counter to our argument. We expect conservative governments to reward allies by allocating more aid directly to the recipient government. We will refrain from speculating further as to why this might be, as there are good reasons to be skeptical of these findings. It is possible that this result represents a degree of bias resulting from the specific observations included in the estimation sample for the alliance grouping. Of the 3,998 observations, only 173 (4.3 percent) observations are allied dyads. Of those, 113 (65 percent) contain the United States as the donor state. Furthermore, many of the recipient states are in Latin America (excepting Turkey and Pakistan). When also factoring in the limited temporal domain, it seems likely that these results reflect the peculiarities of the United States during this time period, rather than any generalizable effects that we may attribute to ideology. In spite of the lack of statistically significant interaction coefficients in Table 4, these models still produce results that are generally consistent with our primary hypothesis. We present the figures showing the marginal effects for the remaining independent variables in the online appendix. In the case of exports, physical integrity rights, recipient state governance, and recipient state development, we continue to find a general negative correlation between ideology and NGO aid. However, in none of these models is there evidence of a conditioning effect between the constituent terms. We also find that physical integrity rights and recipient state governance yield negative and significant effects across most of the range of the ideology variable, suggesting that as the recipient state's governance and respect for physical integrity rights improve, we should expect to see less aid allocated through NGO channels and relatively more allocated through public sector channels. Overall, we find consistent support for the idea that conservative governments tend to allocate less money through NGOs relative to direct government-to-government transfers. Conclusion In this paper, we have explored how the ideologies of donor governments influence decisions on how foreign aid is delivered to recipient states. The analysis presented here ties together previous scholarship on aid channels (e.g., Milner 2006; Dietrich 2013, 2016) and research on ideological preferences associated with aid (Tingley 2010; Milner and Tingley 2010; Thérien and Nöel 2000). We find that more liberal governments are more likely to allocate larger shares of their aid budgets to NGOs and other civil society organizations relative to direct transfers to recipient governments. These choices are a reflection of the belief that development NGOs have priorities that more closely align with liberal governments and that assist needy populations with less money siphoned off by government kleptocrats. On the other hand, while conservative governments tend to be more skeptical of aid in general (Tingley 2010), the aid they do allocate is more heavily weighted towards the public sector as compared to their more liberal counterparts. Despite the belief that foreign aid distorts markets in the recipient country (Thornton 2002), conservative governments appear to be more interested in meeting short-term self-interested goals like building economic and political ties. With the exception of political similarities between the donor and the recipient, we find little evidence that economic ties, or quality of governance (broadly conceived) within the recipient state, have any conditioning effect on how ideology shapes aid allocation decisions. Still, our finding that political similarities, as captured by the UN Affinity measure, do condition the effect of ideology is important, as it suggests some constraints on how governments can manipulate aid allocation. Only when the donor and recipient state are closely aligned do we see ideological differences emerge in how donor governments allocate aid. The insignificant effect at the low end of the affinity scale does not mean that those states receive no aid but does suggest that the domestic political differences within the donor state are muted in their influence on aid channel allocation decisions. Overall, our work helps clarify when and how conservative and liberal governments employ foreign aid. Future research should continue to explore these relationships. Due to sporadic reporting of NGO aid data by donor states, most of the variation we see in the dependent variable is cross-sectional in nature—that is, varying by donors and dyads. Where we do have more complete data over time, we do indeed see variation in the relative amounts of NGO aid allocated within a given donor-recipient dyad. As disaggregated data become more regularly and consistently reported, future studies will be better able to evaluate the relative influence of within-country variation in ideology as compared to cross-national differences.22 The aid policy choices made by both liberal and conservative governments reflect their political interests and those of their constituents. While conservative governments tend to allocate less aid, these conservative governments allocate aid in such a way that may produce more immediate benefits for the donor state. For liberal governments, it may also be the case that the promotion of development-oriented goals is partly intended to generate long-term benefits for the donor state. Reducing poverty, increasing incomes, and promoting economic and political stability in developing states certainly yields potential benefits to donors through increasing consumption of donor-produced goods and services. Even if ODA does not bring meaningful poverty alleviation, perhaps aid can help buy an increasingly positive image for the donor country within the recipient state. Our findings also dovetail nicely with Dietrich's (2013) work, which suggests that when donors are concerned about governance quality, they are more likely to utilize bypass channels. The liberal interest in strengthening economic and political development of recipient states is also associated with bypass. These findings also offer insight into the future of foreign aid distribution. The average portion of DAC donors’ aid moneys being channeled through NGOs in 2004 was approximately eight percent. By 2013, that number had more than tripled to just under 28 percent (OECD 2015). While the majority of foreign aid is not being delivered by NGOs, the influence of these organizations in international development is increasing. Recognizing the conditions under which these NGO agents are most likely to be empowered by national principals, as well as understanding when NGOs can (and potentially cannot) enhance the effectiveness of foreign aid, will be an important step for anticipating the global future of aid and poverty alleviation. Changes in donor ideology and the donor's political ties with potential recipient states may be important factors in determining when and where NGOs play a greater role in the aid delivery process. Supplemental Information Supplemental information is available at the Foreign Policy Analysis data archive. 1 We refer to leftist parties as “liberal” and rightist parties as “conservative.” Our usage of liberal is not synonymous with the classical sense (i.e., free markets, smaller government, etc.). 2 Note that we use the terms “public-sector aid” and “government-to-government aid” interchangeably. 3 Schneider and Tobin (2013) demonstrate that when member states have heterogeneous preferences, aid allocation goals may more closely reflect the unique interests of the institutions, but when member preferences are more similar, aid allocation will more closely reflect national interests. 4 Skepticism is not unfounded. Kono and Montinola (2012) find that autocratic regimes are more likely to divert resources to build military capacity. 5 The correlation between the NGO and multilateral channels is 0.54 in our primary estimation sample, whereas the correlation between these channels and public sector aid is only 0.19 and 0.15, respectively. 6 Left-wing governments are not necessarily behaving altruistically or nonstrategically by funding poverty alleviation projects. Liberal governments may see this type of project as a long–term investment in soft power (Nye 2004). Beyond that, these projects may benefit donor countries economically, but the time horizon for those benefits is longer than the types of foreign aid projects favored by their conservative counterparts. In recent years, USAID has funded programs based on both ending extreme poverty and expanding trade. Both of these goals are related to creating economic growth and economic opportunity, but the approaches to these goals are different. 7 Empirically verifying this claim is difficult, but ODA does appear to lengthen the tenure of leaders, particularly in nondemocratic regimes (Licht 2010). 8 We use aid commitments rather than disbursements. 9 We focus only on individual state donors and exclude multilateral donors from our analysis. See the online appendix for more detail on case selection and data coding. 10 We focus on NGOs and public sector aid for several reasons. As we discuss above, previous studies have focused on government decisions to allocate aid through multilateral organizations (see Milner 2006; Milner and Tingley 2013b). Further, data on public-private partnerships is extremely limited. Utilizing public sector and NGO aid channels, our primary models include 3,998 complete observations. By comparison, there are 3,827 missing observations for the public-private partnership variable—approximately 96% of our estimation sample. Lastly, given the ambiguity of the “other” category we do not include it in our analyses as we have no clear expectations as to how ideology would affect decisions related to this category. 11 The unlogged variable ranges from 0.000018 to 6,705.562 with a mean of 11.82. 12 See online appendix. 13 This variable is measured as ln(x + 1). 14 This variable comes from Dietrich (2013, 704), which she coded from the CIA World Factbook. We forward coded it for years beyond her study. 15 Data were adjusted to 2005 dollars and weighted using GDP figures from the World Bank (World Bank 2014). 16 A test for the presence of serial correlation is significant at the 0.01 level (Wooldridge 2010). 17 A Hausman test of Model 3 in Table 2 yields a significant coefficient, but we opt to use random effects over fixed effects for a few reasons. First, because of the large number of cross-sectional units (i.e., 1,191 dyads) relative to time periods, fixed effects soak up the majority of within-unit variation. Random effects allow us to control for the possibility that various dyadic relationships may have different baseline ratios of aid allocation. Estimating individual parameters for individual dyad fixed effects also reduces the model's degrees of freedom. Using random effects allows us to control for heterogeneity across units without the costs associated with fixed effects. Second, some variables, like alliances, would be perfectly collinear with dyad fixed effects, preventing us from analyzing theoretically important variables. Lastly, Clark and Linzer (2015) argue that the Hausman test is often insufficient to evaluate the appropriateness of random effects. They find that when the number of units and the number of observations for each unit are small, and when the correlation between the key independent variables and the unit effects are relatively low, random effects models outperform fixed effects models. Excepting the alliance dummy, the correlations between our independent variables and the unit effect and the independent variables is low. We have included robustness checks using recipient fixed effects in the online appendix (see A7 and A8). 18 We present the results from our models using the full ideology measure in Tables A4 and A5. 19 See Table A3 for full table. 20 We follow Brambor, Clark, and Golder (2006) and include all constituent terms in the interactive models. 21 See Berry, Golder, and Milton (2012) on symmetry in interaction terms. 22 See online appendix and Table A1. References Acht Martin, Mahmoud Toman Omar, Thiele Rainer. 2015. “Corrupt Governments Do Not Receive More State-To-State Aid: Governance and the Delivery of Foreign Aid Through Non-State Actors.” Journal of Development Economics  114: 20– 33. Google Scholar CrossRef Search ADS   Alesina Alberto, Dollar David. 2000. “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth  5 ( 1): 33– 63. Google Scholar CrossRef Search ADS   Bebbington Anthony, Riddell Roger. 1995. “The Direct Funding of Southern NGOs by Donors: New Agendas and Old Problems.” Journal of International Development  7 ( 6): 879– 93. Google Scholar CrossRef Search ADS   Berry William D., Golder Matt, Milton Daniel. 2012. “Improving Tests of Theories Positing Interaction.” Journal of Politics  74 ( 3): 653– 71. Google Scholar CrossRef Search ADS   Brambor Thomas, Clark William Roberts, Golder Matt. 2006. “Understanding Interaction Models: Improving Empirical Analyses.” Political Analysis  14 ( 1): 63– 82. Google Scholar CrossRef Search ADS   Brech Viktor, Potrafke Niklas. 2014. “Donor Ideology and Types of Foreign Aid.” Journal of Comparative Economics  42 ( 1): 61– 75. Google Scholar CrossRef Search ADS   Bueno de Mesquita Bruce, Smith Alastair. 2007. “Foreign Aid and Policy Concessions.” Journal of Conflict Resolution  51 ( 2): 251– 84. Google Scholar CrossRef Search ADS   Carney Dana R., Jost John T., Gosling Samuel D., Potter Jeff. 2008. “The Secret Lives of Liberals and Conservatives: Personality Profiles, Interaction Styles, and the Things They Leave Behind.” Political Psychology  29 ( 6): 807– 40. Google Scholar CrossRef Search ADS   Cingranelli David L., Richards David L., Chad Clay K.. 2014. “CIRI Human Rights Data Project (v.2014-04-14).” http://www.humanrightsdata.com/. Clark Tom S., Linzer Drew A.. 2015. “Political Science Research and Methods Should I Use Fixed or Random Effects? Should I Use Fixed or Random Effects?” Political Science Research and Methods  3( 2): 399– 408. Google Scholar CrossRef Search ADS   Clarke Gerard. 2007. “Agents of Transformation? Donors, Faith-Based Organizations and International Development.” Third World Quarterly  28 ( 1): 77– 96. Google Scholar CrossRef Search ADS   Dietrich Simone. 2013. “Bypass or Engage? Explaining Donor Delivery Tactics in Foreign Aid Allocation.” International Studies Quarterly  57 ( 4): 698– 712. Google Scholar CrossRef Search ADS   ——— 2016. “Donor Political Economies and the Pursuit of Aid Effectiveness.” International Organization  70: 65– 102. Google Scholar CrossRef Search ADS   Eade Deborah. 2004. “Preface.” In Development, NGOs, and Civil Society , edited by Eade Deborah, 9– 14. Oxfam GB. Easterly William. 2001. The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics . Cambridge, MA: MIT Press. Fleck Robert K., Kilby Christopher. 2006. “How Do Political Changes Influence U.S. Bilateral Aid Allocations? Evidence from Panel Data.” Review of Development Economics  10 ( 2): 210– 23. Google Scholar CrossRef Search ADS   ——— 2010. “Changing Aid Regimes? U.S. Foreign Aid from the Cold War to the War on Terror.” Journal of Development Economics  91: 185– 97. Google Scholar CrossRef Search ADS   Fruttero Anna, Gauri Varun. 2005. “The Strategic Choices of NGOs: Location Decisions in Rural Bangladesh 1.” Journal of Development Studies  41 ( 5): 759– 87. Google Scholar CrossRef Search ADS   Gabel Matthew J., Huber John D.. 2000. “Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions From Party Manifestos Data.” American Journal of Political Science  44 ( 1): 94– 103. Google Scholar CrossRef Search ADS   Gibler Douglas M. 2009. International Military Alliances, 1648–2008 . CQ Press. Google Scholar CrossRef Search ADS   Gleditsch Nils Petter, Wallensteen Peter, Eriksson Mikael, Sollenberg Margareta, Strand Havard. 2002. “Armed Conflict 1946 –  2001: A New Dataset.” Journal of Peace Research  39 ( 5): 615– 37. Graham Jesse, Haidt Jonathan, Nosek Brian A.. 2009. “Liberals and Conservatives Rely on Different Sets of Moral Foundations.” Journal of Personality and Social Psychology  96 ( 5): 1029– 46. Google Scholar CrossRef Search ADS PubMed  Hibbs Douglas A. 1977. “Political Parties and Macroeconomic Policy.” American Political Science Review  71 ( 4): 1467– 87. Google Scholar CrossRef Search ADS   Hirsh Jacob B., DeYoung Colin G., Xu Xiaowen, Peterson Jordan B.. 2010. “Compassionate Liberals and Polite Conservatives: Associations of Agreeableness with Political Ideology and Moral Values.” Personality & Social Psychology Bulletin  36 ( 5): 655– 64. Google Scholar CrossRef Search ADS PubMed  Hoadley J. Stephen. 1980. “Small States as Aid Donors.” International Organization  34 ( 01): 121. Google Scholar CrossRef Search ADS   International Monetary Fund. 2014. “Direction of Trade Statistics (DOTS).” http://data.imf.org/?sk=9D6028D4-F14A-464C-A2F2-59B2CD424B85. Kaufmann Daniel, Kraay Aart, Mastruzzi Massimo. 2015. World Governance Indicators . Tech. rep., World Bank. Koch Dirk-Jan, Dreher Axel, Nunnenkamp Peter, Thiele Rainer. 2009. “Keeping a Low Profile: What Determines the Allocation of Aid by Non-Governmental Organizations?” World Development  37 ( 5): 902– 18. Google Scholar CrossRef Search ADS   Kono D. Y., Montinola G. R.. 2012. “The Uses and Abuses of Foreign Aid: Development Aid and Military Spending.” Political Research Quarterly  66 ( 3): 615– 29. Google Scholar CrossRef Search ADS   Lai Brian, Morey Daniel S.. 2006. “Impact of Regime Type on the Influence of US Foreign Aid.” Foreign Policy Analysis  2 ( 4): 385– 404. Google Scholar CrossRef Search ADS   Laver Michael, Budge Ian. 1992. Party Policy and Government Coalitions . St. Martin's Press. Google Scholar CrossRef Search ADS   Laver Michael, Garry John. 2000. “Estimating Policy Positions from Political Texts.” American Journal of Political Science  44( 3): 619– 34. Google Scholar CrossRef Search ADS   Licht Amanda A. 2010. “Coming into Money: The Impact of Foreign Aid on Leader Survival.” Journal of Conflict Resolution  54 ( 1): 58– 87. Google Scholar CrossRef Search ADS   Marshall Monty G., Jaggers Keith, Gurr Ted Robert. 2014. “Polity IV Project.” http://www.systemicpeace.org/polity/polity4.htm. McLean Elena V. 2015. “Multilateral Aid and Domestic Economic Interests.” International Organization  69 ( 01): 97– 130. Google Scholar CrossRef Search ADS   Meernik James, Krueger E. L., Poe S. C.. 1998. “Testing Models of US Foreign Policy: Foreign Aid During and After the Cold War.” The Journal of Politics  60 ( 1): 63– 85. Google Scholar CrossRef Search ADS   Milner Helen V. 2006. “Why Multilateralism? Foreign Aid and Domestic Principal-Agent Problems.” In Delegation and Agency In International Organizations , edited by Hawkins Darren G., Lake David A., Nielson Daniel N., Tierney Michael J., chap. 4, 107– 39. New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Milner Helen V., Tingley Dustin. 2011. “Who Supports Global Economic Engagement? The Sources of Preferences in American Foreign Economic Policy.” International Organization  65 ( Winter): 37– 68. Google Scholar CrossRef Search ADS   ——— 2010. “The Political Economy of U.S. Foreign Aid: American Legislators and the Domestic Politics of Aid.” Economics and Politics  22 ( 2): 200– 32. ——— 2013a. “Public Opinion and Foreign Aid: A Review Essay.” International Interactions  39 ( 3): 389– 401. Google Scholar CrossRef Search ADS   ——— 2013b. “The Choice for Multilateralism: Foreign Aid and American Foreign Policy.” Review of International Organizations  8 ( 3): 313– 41. Google Scholar CrossRef Search ADS   Murdie Amanda Marie, Kakietek Jakub. 2012. “Do Development INGOs Really Work? The Impact of International Development NGOs on Human Capital and Economic Growth.” Journal of Sustainable Society . 1( 1): 1– 12. Nancy Gilles, Yontcheva Boriana. 2006. Does Ngo Aid Go to the Poor? Empirical Evidence from Europe . International Monetary Fund. Nöel Alain, Thérien Jean-Philippe. 1995. “From Domestic to International Justice: The Welfare State and Foreign Aid.” International Organization  49 ( 3): 523– 53. Google Scholar CrossRef Search ADS   Nöel Alain, Thérien Jean-Philippe, Dallaire Sebastian. 2004. “Divided Over Internationalism: The Canadian Public and Development Assistance.” Canadian Public Policy  30 ( 1): 29– 45. Google Scholar CrossRef Search ADS   Nunnenkamp Peter, Weingarth Janina, Weisser Johannes. 2009. “Is NGO Aid Not so Different After All? Comparing the Allocation of Swiss aid by Private and Official Donors.” European Journal of Political Economy  25 ( 4): 422– 38. Google Scholar CrossRef Search ADS   Nye Joseph S. 2004. Soft Power: The Means to Success in World Politics . Public Affairs. OECD. 2013. Evaluating Development Activities: 12 Lessons from the OECD DAC . Tech. rep., Organization for Economic Cooperation and Development. ——— 2015. Creditor Reporting System Database . Tech. rep., Organization for Economic Cooperation and Development (OECD). Paxton P., Knack S.. 2012. “Individual and Country-Level Factors Affecting Support for Foreign Aid.” International Political Science Review  33 ( 2): 171– 92. Google Scholar CrossRef Search ADS   Radelet Steven. 2006. “A Primer on Foreign Aid.” Center for Global Development Working Paper 92. Riddell Roger C., Bebbington Anthony, Peck Lennart. 1995. “Promoting Development by Proxy. An Evaluation of the Development Impact of Government Support to Swedish NGOs.” SIDA Evaluation Report (Sweden) . Schneider Christina J., Tobin Jennifer L.. 2013. “Interest Coalitions and Multilateral Aid Allocation in the European Union.” International Studies Quarterly  57 ( 1): 103– 14. Google Scholar CrossRef Search ADS   Strezhnev Anton, Voeten Erik. 2012. “United Nations General Assembly Voting Data.” URL: http://hdl.handle.net/1902.1/12379. Tender Judith. 1982. Turning Private Voluntary Organizations into Development Agencies: Questions for Evaluation . USAID, Bureau of Food and Voluntary Assistance. Thérien Jean-Philippe. 2002. “Debating Foreign Aid: Right Versus Left.” Third World Quarterly  23 ( 3): 449– 66. Google Scholar CrossRef Search ADS   Thérien Jean-Philippe, Nöel Alain. 2000. “Political Parties and Foreign Aid.” American Political Science Review  94 ( 1): 151– 62. Google Scholar CrossRef Search ADS   Thornton Mark. 2002. Corruption and Foreign Aid . Online Publications on Ludwig von Mises Institute. Tingley Dustin. 2010. “Donors and Domestic Politics: Political Influences on Foreign Aid Effort.” The Quarterly Review of Economics and Finance  50 ( 1): 40– 49. Google Scholar CrossRef Search ADS   USAID. 2013. Where Does the Money Go?  Tech. rep., United States Agency for International Development. Volkens Andrea, Lacewell Onawa, Lehmann Pola, Regel Sven, Schultze Henrike, Werner Annika. 2011. The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR) , Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB). Wooldridge Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data . Cambridge: The MIT Press, 2nd edition. World Bank. 2014. World Development Indicators 2014 . © The Author (2017). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Foreign Policy Analysis Oxford University Press

Donor Government Ideology and Aid Bypass

Loading next page...
 
/lp/ou_press/donor-government-ideology-and-aid-bypass-loy4F0hvc4
Publisher
Wiley-Blackwell
Copyright
© The Author (2017). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
ISSN
1743-8586
eISSN
1743-8594
D.O.I.
10.1093/fpa/orx005
Publisher site
See Article on Publisher Site

Abstract

Abstract How do domestic politics in the donor country affect foreign aid policy? Scholars have analyzed the impact that political parties and domestic institutions play in shaping aid effort. Building on this previous work, we examine the impact of domestic politics on the channels chosen for aid distribution. In this paper, we explore whether and how the ideological composition of donor governments affects the method of delivering aid to recipient states. Specifically, we argue that more liberal governments are more inclined to provide aid via nongovernmental organizations (NGOs), while more conservative governments prefer to give aid through direct bilateral government-to-government channels. These choices, we argue, reflect the different goals that right and left parties pursue through their allocation of foreign aid. Our results indicate that left governments are more inclined to channel aid through NGOs in an attempt to directly implement poverty alleviation strategies. Alternatively, right wing governments place a greater emphasis on channeling aid through recipient state governments to promote the economic interests of their constituents and the geopolitical interests of the state. There is an extensive literature on the factors that drive donor states to give foreign economic aid. Much of this work focuses on the characteristics of recipient states and relationships with donor states to explain how much aid a state receives (see Meernik, Krueger, and Poe 1998; Alesina and Dollar 2000; Lai and Morey 2006; Bueno de Mesquita and Smith 2007). By comparison, less work explores how donor states’ domestic political environments shape the aid allocation process. Prior work has largely focused on the influence of donor government ideology and support for aid in aggregate terms (e.g., Thérien and Nöel 2000; Tingley 2010; Milner and Tingley 2010), and almost no work has been aimed at understanding the link between donor government ideology and the specific methods of aid delivery. Considering the effects of government ideology on aid bypass highlights a critical source of variation in aid flows—both within and across countries—that accounts for how billions of dollars in aid is allocated each year. It also suggests that foreign aid is a more flexible and valuable political tool for governments than exiting literature has recognized, the manipulation of which requires no change in aggregate aid budgets—only in aid delivery channels. We argue that the ideology of donor governments affects the method of aid delivery. Parties across the left-right spectrum cater to fundamentally different societal constituencies with different interests, goals, and worldviews (e.g., Hibbs 1977; Carney et al. 2008; Graham, Haidt, and Nosek 2009; Hirsh et al. 2010; Milner and Tingley 2013a).1 The desire to retain office should lead governing parties to use the tools at their disposal, like foreign aid, to pursue policy goals that benefit key constituents. The left-right orientation of a party necessarily relates to the sorts of policies and goals that constituents want their representatives to pursue. Because some aid delivery channels are more or less effective at achieving certain goals, the electoral benefits of allocating aid through particular channels will vary according to the party and its goals. Earlier work has explored the use of multilateral organizations as a way to “bypass” donor governments (Hoadley 1980; Milner 2006). Here we examine another bypass channel, focusing on how ideology affects whether aid is delivered through NGOs versus public-sector channels.2 We argue that liberal governments should allocate a greater proportion of foreign aid through NGOs, specifically. Development NGOs are more likely to prioritize direct poverty alleviation, focusing on the needs of citizens in the recipient country rather than on the priorities of their governments. Conservative governments should be more likely to favor government-to-government transfers, which are thought to be a better tool for advancing the immediate economic and political goals of the donor country, despite being less directly focused on long-term poverty reduction (Nunnenkamp, Weingarth, and Weisser 2009). We find that more liberal donor governments tend to allocate greater shares of aid through NGOs, while more conservative governments appear to favor government-to-government transfers. We also find that the effects of government ideology are conditioned by the nature of the donor state's relations with the recipient state. Specifically, we find that more conservative governments allocate relatively less aid through NGOs when recipient states have close relations with donor states. However, ideology has no effect on the allocation of aid when the donor and recipient have poorer relations. The substantive implications of this research are great. First, our study contributes to the growing literature on how domestic politics influence donor state decisions over the allocation of aid (e.g., Fleck and Kilby 2006; Tingley 2010; Milner and Tingley 2013a; Dietrich 2016). We depart from most existing research by linking partisan and ideological interests to how aid is allocated across public and private sector channels. This is the central way in which we build on previous work that has examined how donor government politics affect bypass decisions. For example, whereas Dietrich (2016) focuses on variation across countries, our analysis explores how variation in governments’ political preferences within and across countries affects the relative allocation of aid between government-to-government channels and private sector channels. We find that the left-right orientation of donor governments has a clear effect on the channels through which aid is distributed. Somewhat paradoxically, we show that right parties actually prefer public sector channels, while left governments have a greater preference for allocating aid through NGOs in the private sector. These findings highlight the fact that while conservative governments tend to allocate less money to foreign aid, they do provide some aid. Conservatives and liberals both see political value in foreign aid; they simply use aid differently. Understanding the donor-side factors that drive the allocation of aid to NGOs can also be of enormous import for both NGOs and recipient states. Over the twelve-year period of our analysis, Canada allocated an average of 20 percent of its ODA. The United States allocated approximately 10 percent (OECD 2015). This translates into billions of dollars in the aggregate, much of which is allocated to NGOs and private contractors (Dietrich 2013, 690). If the ideology of donor governments influences the methods of aid delivery, then these values may shift considerably over time. A change from a liberal to a more conservative government may translate into millions in lost funding for NGOs, even as aggregate aid levels remain unchanged, given conservative governments’ preference for public sector channels. By extension, this may have profound effects on grassroots development projects in the recipient state. The Domestic Politics of Donor States The conventional wisdom is that liberals support aid and conservatives oppose it. While there is some truth to this, a more nuanced look demonstrates that governments across the political spectrum utilize aid differently and provide aid in distinct ways. Political ideology should certainly be influential in matters of foreign assistance. Ideology, often conceptualized as a single underlying dimension, reflects beliefs and attitudes about the role of government in the economy. Ideology acts as a constraint on policy choices by “bundling” various issues together (Gabel and Huber 2000; Laver and Budge 1992). Attitudes about the role of government in the economy help place political parties on the right-left spectrum (Laver and Garry 2000). Thérien (2002) argues that the conflicting priorities of right and left parties have been integral to the development of foreign assistance programs. Given their support of social welfare programs and government intervention in their domestic economies, scholars often suggest that liberals should be more supportive of foreign aid. Preferring less government intervention in the economy, conservatives should be opposed to aid programs. Studies of foreign aid flows from Organization for Economic Cooperation and Development (OECD) donor states concluded that leftist governments correlate indirectly with higher levels of foreign aid; the effect largely felt through the expansion of the welfare state and the externalization of liberal norms through foreign policy (Nöel and Thérien 1995; Thérien and Nöel 2000). Tingley (2010) observes that among OECD donor states, conservative governments are characterized by a lower aid effort than liberal governments, but this effect is limited to aid flows to developing countries and multilateral organizations. Brech and Potrafke (2014) provide evidence of a partisan divide in bilateral grant aid. These partisan divisions appear to exist in the general public as well. Cross-national public opinion polls have found that liberals are more supportive of foreign economic aid, and conservatives less so (Nöel, Thérien, and Dallaire 2004; Paxton and Knack 2012; Milner and Tingley 2013a). Studies examining the United States illustrate similar preference differences between liberals and conservatives but also that the parties have different goals for giving aid. Fleck and Kilby (2010) demonstrate that conservative administrations have allocated less money to foreign aid than liberal administrations. The priorities of US aid also appear to vary by ideology. Fleck and Kilby (2006) show that Republicans emphasize commercial factors like the promotion of US exports, while Democrats emphasize humanitarian and developmental factors. In addition, analyses of congressional voting behavior have also found that conservative legislators are more likely to oppose measures that would increase economic aid spending (Milner and Tingley 2010, 2011), which is likely driven by the conservative belief that foreign aid is an obstacle to market efficiency and welfare improvement in both donor and recipient states (Thornton 2002). Such patterns hold at the mass level as well—Milner and Tingley (2013a, 395) find that “more conservative voters respond more positively when aid is helpful to the US economy, and more liberal voters respond more favorably to aid when it targets the neediest groups abroad.” This literature provides a useful basis for our theoretical argument, but the specific electoral mechanisms are worth expanding upon. Assuming leaders in office wish to stay in office, aid provides them with a tool that can be used to generate benefits for constituents. Left and right governments tend to represent different core economic constituencies, with the left tending to represent the interests of lower income groups and labor, and the right representing higher income groups and capital (Hibbs 1977). Drawing on the Heckscher-Ohlin and Stolper-Samuelson models of trade, Milner and Tingley (2010) argue that foreign aid transfers from wealthy donor countries to poorer countries represent a transfer of capital that can have distributional effects on individuals within the donor country that vary according to individuals’ factor endowments—holders of capital will see increased incomes while holders of labor will see no or negative returns from aid transfers. As described by Milner and Tingley (2010, 207), “The economics of aid . . . constitute a part of donor nations’ commercial strategy to secure larger trade benefits . . . The economic benefits . . . are largest when recipient countries import goods in which donor countries have a comparative advantage in production.” Thus aid transfers can increase the returns to the producers of capital-intensive goods in wealthy developed states—groups that tend to be represented by rightist parties—but only if aid promotes greater consumption of goods produced by the donor state. The economic incentive for right parties to use aid to promote commercial goals is clear—the benefits go to constituents typically associated with right parties. However, this dynamic does not help us to understand the preferences of more liberal individuals who should suffer economic losses in the form of the higher tax rates used to fund aid. Leftist parties are associated with support for larger aid budgets, but research on public attitudes towards aid indicates that, whereas conservatives are more likely to want aid to go to promoting economic and geopolitical goals, liberals tend to want aid to go to promoting development among the poor and needy. Conservative respondents demonstrate a preference for giving a larger share of the aid to states where the United States has strong commercial ties, and liberal respondents are more inclined to allocate a greater proportion of aid to needier states (Milner and Tingley 2013a). Paxton and Knack (2012) similarly find that conservatives and individuals expressing more critical attitudes towards the poor are less likely to support aid. Studies in political psychology help explain these different priorities. Carney, Jost, Gosling et al. (2008) find that liberals and conservatives differ on several key personality traits, with liberal ideology correlating with an increased emphasis on compassion and egalitarianism (Hirsh et al. 2010), as well as less acceptance of inequality (Graham, Haidt, and Nosek 2009). Thus, while liberals may not reap direct economic benefits from aid, they may be willing to endure some marginally higher tax burdens if aid is used to promote humanitarian and development goals abroad. If we expect politicians to represent the preferences of their constituents, it follows that left (right) governments will funnel more aid to development (commercial) goals. Aid Delivery Channels, Delegation, and Ideology To understand how ideology affects government decisions to delegate through NGOs, we first draw on previous research looking at aid delegated through another bypass channel—multilateral organizations. We then link this literature to more recent work on aid delegation to NGOs. Research on the choice to delegate to multilateral organizations can help us to better understand government choices to allocate aid through bypass channels and the role of ideology in this process. As with multilateral aid, delegating to development NGOs involves some loss of control over how aid is implemented, thereby depriving governments of the ability to leverage aid into policy concessions. These literatures can help us understand how government ideology affects the choice to allocate aid through bypass channels. Why do states delegate control over the distribution of foreign aid to multilateral organizations? Assuming multilateral organizations have a degree of independence from donor governments, how they choose to administer aid will likely diverge to some extent from a donor's preference.3 One possibility is that delegation to multilateral organizations may benefit some domestic economic interests in the donor state (McLean 2015). Another explanation grounded in donor domestic politics is that governments delegate to send signals to their domestic audiences (Hoadley 1980; Milner 2006). Studies have shown that even where publics tend to support foreign aid in general, citizens are often skeptical aid is effective at promoting development and reducing poverty in recipient states or believe that it is used for corrupt purposes (Nöel, Thérien, and Dallaire 2004; Milner 2006).4 This skepticism risks constraining policymakers’ access to a valuable foreign policy tool, as popular support is rarely on the side of increasing aid budgets (Nöel, Thérien, and Dallaire 2004). Given the focus of multilateral organizations on promoting development, delegating aid to multilateral organizations can serve as a signal to donor governments’ publics that aid is promoting development rather than being siphoned off by corrupt bureaucrats or repressive regimes (Hoadley 1980; Milner 2006). This argument is useful but assumes that all governments and donor audiences are motivated to improve development in the recipient state. However, politicians in donor states are often motivated by a variety of goals beyond the promotion of development, such as fostering commercial expansion or rewarding allies—goals that have been linked to partisanship and ideology (e.g., Fleck and Kilby 2006; Milner and Tingley 2013b,a). This research tends to indicate that ideology does affect these decisions. Milner and Tingley (2013b) examine public support in the United States for delegating aid to multilateral organizations, arguing that donors make a tradeoff between greater control of aid and greater burden-sharing. The development-oriented focus of multilateral organizations implies lower agency loss for liberals, who prefer to allocate aid to development-oriented ends. Given their focus on promoting commercial outcomes and security goals, the agency loss from delegating should be greater for conservatives. If the goal is to obtain concessions from the recipient government, the management of aid funds by multilateral organizations severely diminishes states’ capacity to manipulate aid to that end. As discussed above, liberal and conservative voters do emphasize different goals that are associated with aid. There are strong parallels between the work on delegating through multilateral organizations and more recent work on aid bypass. Dietrich (2013) argues that the quality of governance in the recipient state influences how donors choose to channel aid. Dietrich finds that donors are indeed sensitive to governance issues in the recipient state, allocating more aid to nonstate actors where the quality of governance is poor and more aid directly to governments where governance is better. By bypassing poorly governed states, Dietrich argues that development aid can be allocated to more productive ends. Using several indicators of recipient state quality of governance, Acht, Mahmoud, and Thiele (2015) similarly find that poor governance correlates with higher levels of bypass aid. Dietrich (2016) asserts that the effect of governance is conditioned by cross-national differences in donor states’ emphasis on the role of the state in the delivery of public goods and services; In some recipient countries, the state plays a larger role in the delivery of services, while other recipient countries place a greater emphasis on the role on private markets in providing aid. Where governance is poor and the possibility of aid capture is high, donor governments that would otherwise allocate more aid through government-to-government channels should shift their allocation more heavily towards more market-oriented channels. Work on aid bypass shows that when donors are concerned about poor governance, aid allocated through NGOs can function similar to aid allocated through multilateral organizations.5 For example, Fruttero and Gauri (2005) model the relationship between donor governments and NGOs as a principal-agent game. The official aid providers (the principals) have incomplete information about the projects being implemented by the NGOs (the agents), but future NGO funding is contingent on perceptions concerning the success or failure of current projects. The belief is that NGOs are less bound by the policy agendas of donor governments, and by working on the ground in recipient countries, NGOs are more aware of the needs of the population and the ways in which aid funds can be utilized to meet those needs. If implementation of aid by NGOs is less distorted by self-interest than those made directly by donor nations, then aid may be more effective at poverty alleviation (Nancy and Yontcheva 2006). NGOs may also be able to circumvent corrupt or inefficient local governments to deliver aid directly to those in need (Riddell, Bebbington, and Peck 1995). Thus, delegating to NGOs may also alleviate problems of credibility with domestic publics, if private organizations are perceived as being less corrupt and wasteful than their government counterparts. Given their preference for pursuing development-oriented outcomes, liberal governments should allocate relatively more aid through NGOs. By definition, development NGOs are concerned with promoting economic growth and human development. NGOs are also viewed as having more direct contact with the people that aid is intended to help, and are thought to be more effective and efficient vehicles for aid delivery (Bebbington and Riddell 1995; Nancy and Yontcheva 2006; Koch et al. 2009; Tender 1982). The specific means of accomplishing these goals may vary across organizations, but many development NGOs approach developmental issues through grass roots activities, like providing micro-credit loans and grants to needy populations.6 Thus NGOs should be better equipped to pursue the types of projects that liberal voters and governments prefer. On the other hand, given their desire to advance commercial and geopolitical goals, we expect conservative governments to allocate less aid through NGOs. Foreign aid can be used to buy policy concessions and can help stabilize the political leadership in the recipient state (Bueno de Mesquita and Smith 2007). The fact that they are private organizations and not a part of the recipient state's government, combined with the emphasis that many NGOs place on grass roots development projects, suggests that development NGOs are not positioned to translate aid dollars into substantial influence over recipient state policies. Critics suggest that these sorts of financial transfers allow “bad” governments to remain in power longer than they otherwise would be able to do (Easterly 2001; Radelet 2006). However, if the goal is to buy favorable policies with aid, then the quality of recipient state governance should matter less and the incentives to pursue bypass channels should be mitigated.7 A possible objection to this argument is that it contradicts the attitudes we typically associate with left and right parties. Given their emphasis on free markets, one might argue that conservative governments should favor NGOs. There are a few key reasons why we believe this is not the case. Our argument is that the relative allocation of aid between public and private sector channels is a means to an end for politicians, not the end itself. Simply put, NGOs are not in a position to advance the kinds of economic and geopolitical goals that conservative governments prioritize. There are indeed numerous conservative NGOs to choose from, most NGOs are closely associated with more liberal political and economic goals. Eade (2004, 12) notes that while a range of NGOs exists, “more often than not, development NGOs are in some way involved in transferring resources from societies which have plenty to those who have little.” For example, Catholic Relief Services and FHI-360, two groups regularly listed among USAID's top vendors, have dramatically different attitudes about HIV/AIDS prevention and sex education (USAID 2013). While many faith-based development groups have more conservative values, the objectives of these groups do not square neatly with the commercial and geopolitical interests of conservative governments. Clarke (2007) argues that faith-based organizations, while highly active in more rural areas of developing states, were largely disconnected from development policy debates in donor and recipient states until the 1990s. In the case of the United States, President George W. Bush oversaw numerous changes intended to better integrate the activities of faith-based organizations into US development policy (Clarke 2007, 82–83). These organizations have largely remained dedicated to poverty alleviation, with the added dimension of conducting missionary work in the process. The fact that they are still primarily concerned with pursuing development-oriented goals makes them of less use to conservative governments, who are pursuing economic and geopolitical goals with aid. Evidence indicates that development NGOs tend to focus their efforts on the needy and that these efforts have a positive effect on development outcomes. Studies of NGO aid-giving find that development NGOs respond to need in determining where to deploy their resources (Nancy and Yontcheva 2006; Koch et al. 2009). Evidence indicates that their presence correlates positively with the development of human capital and economic growth (Murdie and Kakietek 2012). Though there is some risk in inferring preferences from outcomes, it would appear that development NGOs are investing resources with the foremost priority of promoting development. Accordingly, we expect conservative governments to allocate more aid to public-sector channels relative to NGO channels. What aid conservative governments do allocate to NGOs may be funneled to conservative NGOs over more liberal organizations, but this has no bearing on the more general weighting of public-sector versus NGO aid. Table 1. Summary relationships for ideology and aid Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  View Large Table 1. Summary relationships for ideology and aid Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  Ideology  Goals  Aggregate Aid  Public-Sector Aid  NGO Aid  Conservative  Commercial/geopolitical  −  +  −  Liberal  Development-oriented  +  −  +  View Large Table 1 provides a brief summary of the expected relationships between ideology, the aggregate amount, and the type of aid. Our primary hypothesis is as follows: Hypothesis 1:Conservative governments should allocate less aid through bypass channels as compared with liberal governments. While ideology may have an independent effect on aid allocation, there are reasons to expect characteristics of the relationship between the donor and the recipient, or characteristics of the recipient state, to condition how ideology affects aid bypass decisions. Similar to Dietrich (2016), we argue that the inclinations of donor governments towards public versus private allocation channels may be conditional upon a variety of factors. We focus herein on six different policy areas/variables: 1) Exports; 2) Foreign policy similarity; 3) Alliance relations; 4) Recipient human rights conditions; 5) Recipient development; and 6) Recipient governance. Variables 1–3 capture aspects of how salient the recipient state is to the donor state in terms of economic and strategic relations. Variables 4–6 capture aspects of what we can broadly conceive of as indicators of recipient state governance. These variables speak to the salience of the underlying goals that motivate the basic differences between left and right parties that we discussed above. First, we expect elements of the dyadic relationship in question to condition the effect of ideology on aid allocation. Consider exports from the donor to the recipient, for example. Let us assume there are two recipients, State A and State B, and that the donor exports goods worth $10 billion per year to State A and $10 million per year to State B. Our basic theory suggests that a shift from a liberal to conservative government should lead to an increase in government-to-government aid relative to NGO aid but does not differentiate between these two cases. Given that commercial interests are stronger between the donor and State A versus State B, we expect the shift to a conservative government will produce a larger relative shift towards government-to-government aid in the case of State A than State B. Because conservatives prefer to allocate aid in such a way as to advance commercial interests, it stands to reason that this shift should be greater where the value of those commercial relationships is greater and the potential returns to government-to-government aid are higher. We expect similar dynamics when considering security-related issues. Since aid can be used as a tool to buy policy concessions or support friendly governments, conservative governments should seek to allocate more funding through public sector channels to states with which they share alliance ties or states that have similar policy preferences. However, when the donor and recipient are not allied, or when they are not closely aligned on global issues, liberal and conservative governments should be more constrained in their willingness to allocate aid to the recipient government. For example, the differences between liberals and conservatives in the United States over aid channels should be dampened when the would-be recipient is a country like Iran as compared to a country with less antagonistic relations. This leads us to our first conditional hypothesis: Hypothesis 2:The negative effect of more conservative governments on bypass aid should be stronger where economic and security ties between the donor and recipient states are stronger. The quality of governance in the recipient state should also lead liberal governments to allocate even greater amounts of aid through bypass channels (Acht, Mahmoud, and Thiele 2015; Dietrich 2013, 2016). However, given the different underlying attitudes towards inequality and suffering, we expect liberals and conservatives to respond differently (e.g., Graham, Haidt, and Nosek 2009). Given that liberal governments are more concerned with ensuring that aid is used to promote development and poverty alleviation, any indication that the government is going to use aid for corrupt or abusive purposes may give liberal governments pause when allocating aid through public sector channels, leading them to reduce the amount of public sector aid that they are willing to give as compared to a less corrupt or less abusive state. Similarly, liberal governments’ emphasis on poverty reduction should also lead them to allocate more aid through bypass channels as need within the recipient state increases as it indicates an increased urgency of direct poverty-alleviation efforts. Low income may also indicate that recipient state policies are largely ineffective at promoting growth and meeting the needs of the populous. On the other hand, conservative governments, with their focus on promoting economic and security goals, which are better achieved by appealing to governmental sources, may be less concerned with matters of poor governance or need, since providing funds directly to the people does little to purchase greater government support for donor policies. This leads us to our second conditional hypothesis: Hypothesis 3:The negative effect of more conservative governments on bypass aid should be larger where the recipient state's domestic environment is characterized by poor governance. Data and Research Design The data for our dependent variable come from the OECD's Creditor Reporting System (CRS) database (OECD 2015).8 Information regarding foreign aid channels is available for all DAC countries from 2003 to 2014.9 The aid data include total aid as well as aid flows by the following channels: 1) traditional government-to-government aid, 2) NGO aid, 3) multilateral aid, 4) public-private partnerships, and 5) a catch-all “other” category. For the purposes of this project, our attention is focused on NGO aid and traditional government-to-government ODA.10 The dependent variable, defined in Equation 1, is the ratio of aid from donor i to recipient j in time t that is channeled through NGO and civil society organizations to aid channeled in the more traditional government-to-government manner. As the numerator grows, the donor is channeling more aid through NGOs. As the denominator grows, the donor is channeling more aid directly to the recipient government. The aid values used to create the ratio are raw dollar amounts, not budgetary percentages. Accordingly, the ratio is highly skewed, and so we use the natural log of this ratio as the dependent variable in the following analyses.11  \begin{equation} {Y_{ijt}} = {\rm ln} ({\it Ratio})_{ijt} = \left( \frac{\hbox{NGO Aid}_{ijt}}{\hbox{Government-to-government Aid}_{ijt}} \right) \end{equation} (1) We opt for this ratio measure because we conceptualize these policy choices as kinds of substitutes. Figure 1 provides some cursory support for this approach. There is a strong negative correlation between the mean proportion of each DAC member's aid budget allocated to NGO aid and the proportion allocated to public sector aid (r = −0.64). The fact that there are five possible budget categories suggests that this relationship is not simply imposed by the presentation of budgetary proportions. For example, countries like Australia, Canada, Norway, Switzerland, and the Netherlands have relatively low levels of both bilateral and NGO aid. The balance often goes to multilateral aid. The results remain unchanged when we use alternative measures of the dependent variable (see online appendix). Figure 1. View largeDownload slide Mean proportion of aid by channel and donor state. Donor states are members of the OECD Development Assistance Committee (DAC) and are ordered according to least NGO aid to most NGO aid as a proportion of aid given. Figure 1. View largeDownload slide Mean proportion of aid by channel and donor state. Donor states are members of the OECD Development Assistance Committee (DAC) and are ordered according to least NGO aid to most NGO aid as a proportion of aid given. Our primary independent variables are created using the Comparative Manifestos Project (CMP) data (Volkens et al. 2011). The CMP-coded manifestos provide information about political parties in all donor countries. Information from CMP is used in two different ways to measure ideology. First, following Tingley (2010), we calculate ideological scores (using the method described in Gabel and Huber 2000) for party attitudes on the economy. These scores range from 0–10. The CMP data include information on parties’ positions on free enterprise, market regulation, and various forms of government involvement in the economy. The economic scores enable us to clearly delineate between the left and right preferences on economic issues, while the full left-right index includes a broader range of issues. In the following models we include the left-right economic ideology measure for only those parties who are currently in government. Second, we follow Tingley (2010) in calculating the left-right measure using party attitudes across all issues, not just economic issues.12 We also control for the economic circumstances of the donor country. We include a measure of economic growth (World Bank 2014). Countries experiencing economic growth are more likely to provide foreign aid than those with struggling economies. Previous research has also linked the expansiveness of the donor state's welfare institutions to foreign aid (Nöel and Thérien 1995; Thérien 2002). Larger social welfare institutions may generally indicate a higher baseline rate of government intervention in the economy, apart from the ideology of a particular governing party, which may correlate with more public sector aid. Accordingly, we include the percentage of GDP allocated to domestic welfare spending (OECD 2013). Lastly, we control for the total amount of aid the donor gives. Larger donors may be better able to leverage large aid budgets into political concessions and may be more likely to emphasize public-sector aid (Milner 2006; Hoadley 1980; Milner and Tingley 2013b). These data come from the OECD CRS (OECD 2015).13 In addition, we consider characteristics of the recipient country. The economic state of the recipient country is an important predictor of the type of aid a country receives. Leftist governments are more likely to focus on need, and so we believe needier countries may be targeted with NGO aid more frequently. For this reason, we include GDP per capita in order to account for recipient need (World Bank 2014). We include the natural log of this variable given its skewed distribution. The political characteristics of the recipient country are also relevant. In Assessing Aid, the World Bank suggests that aid effectiveness can be strengthened by targeting poor countries with strong institutions. Following (Dietrich 2013), there are also reasons to expect that regime type will influence the aid channel chosen. We include the Polity2 variable to control for the recipient state's regime type (Marshall, Jaggers, and Gurr 2014). DAC donors may be inclined to give more assistance to more democratic countries. The human rights record of a recipient may also influence how aid is distributed. We utilize the physical integrity rights scores from the CIRI Human Rights Data Project (Cingranelli, Richards, and Clay 2014). Dietrich (2013) has also shown that the quality of governance in the recipient state influences decisions over how aid is allocated. We include an indicator of the recipient state's quality of governance using data obtained from the World Governance Indicator's project (Kaufmann, Kraay, and Mastruzzi 2015). Lower values represent poorer governance and higher values represent better quality governance. We also include a dichotomous indicator of civil conflict in the recipient state (Gleditsch et al. 2002). Donor governments may be wary of channeling aid to recipient governments that are engaged in a civil conflict, as aid may be misappropriated and used to bolster the government's security forces. The relationship between the donor and recipient also influences the aid relationship. We control for the level of political alignment measured by similarity in UNGA voting (Strezhnev and Voeten 2012). We also control for defense pacts (Gibler 2009). Where there are strong political and security relationships we anticipate that donors will allocate less aid through NGOs and more aid through traditional government-to-government channels. We also include a dichotomous indicator for whether or not the recipient state is a former colony of the donor state.14 We expect donor and recipient government ties to be stronger in such cases, prompting donors to allocate more aid through government-to-government channels. As with security ties, donors may channel more aid directly to recipient governments where the donor and recipient have strong economic ties. We include a measure of exports from the donor the recipient, $$( \frac{{\it Exports}_{ijt}}{GDP_{it}} ) \times 100$$, where i is the donor country, j is the recipient country, and t denotes the year. Given that the volume of donor exports varies, this measure allows us to capture the relative importance of recipient markets across donor countries. These data are drawn from the IMF Direction of Trade Statistics (International Monetary Fund 2014).15 Summary statistics can be seen in Table A2. We estimate our primary models using OLS and a generalized linear model with random effects. Independent variables are lagged by one time period as the anticipated effects are not instantaneous. We also include a lagged dependent variable to control for serial correlation.16 The basic models are presented in Table 2.17 Table 2. Ratio of NGO aid to public sector aid   (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998      (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Table 2. Ratio of NGO aid to public sector aid   (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998      (1)  (2)  (3)    Base Model  OLS Model  Random Effects  Donor Government Ideology Score  −0.836***  (0.072)  −0.370***  (0.056)  −0.436***  (0.064)  Donor Welfare      −0.046***  (0.008)  −0.066***  (0.012)  Donor Growth Rate      0.005  (0.010)  0.002  (0.009)  UN Affinity      0.014  (0.104)  0.017  (0.132)  Alliance      −0.100  (0.172)  0.031  (0.244)  Exports as % of Donor GDP      0.236  (0.153)  0.156  (0.174)  Former Colony      −0.009  (0.071)  0.025  (0.096)  Recipient Polity Score      0.025***  (0.007)  0.033***  (0.009)  Recipient GDP Per Capita      −0.038  (0.036)  −0.066  (0.050)  Recipient Governance      −0.309***  (0.082)  −0.442***  (0.109)  Recipient Civil Conflict      0.142*  (0.082)  0.100  (0.091)  Recipient Physical Integrity Rights      −0.066***  (0.020)  −0.100***  (0.024)  ln(All Aid Channels)      −0.165***  (0.021)  −0.206***  (0.026)  Lagged ln(Aid Ratio)      0.580***  (0.017)  0.368***  (0.021)  Observations  3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Analysis and Discussion Table 2 shows the results of our base models. We first estimate a simple OLS model (Model 1) to demonstrate that there is a clear negative bivariate correlation between more conservative governments (those scoring higher on the ideology scale) and the ratio of NGO aid to public-sector aid, in-line with our primary hypothesis.18 Models 2 and 3 add the control variables. Model 2 presents an additional full OLS model, and Model 3 is estimated using GLS with random effects. Across both models the donor government's ideology score continues to have a strong and statistically significant negative coefficient, indicating that more conservative governments give less money to NGOs relative to government-to-government aid. Of the variables capturing characteristics of the donor country, only the size of the donor's welfare state has a consistent and statistically significant coefficient. As the overall amount of the donor state's social welfare spending increases, we see relatively less aid allocated through NGOs and more aid allocated through traditional government-to-government channels. A larger welfare state may indicate that the donor government has greater institutional capacity to manage aid distribution. This finding may be picking up a dynamic similar to that discussed by Dietrich (2016). Of the variables measuring the relationship between the donor and the recipient, none are statistically significant. Exports, political similarity, colonial history, and defense pacts do not appear to be systematically related to the choices made about how aid is channeled. We discuss this issue further below. More democratic recipient states receive more aid channeled through NGOs and civil society groups relative to public-sector aid. In some ways, this is surprising, given the findings of (Dietrich 2013). Presumably democratic governments would be more trustworthy and capable than their autocratic counterparts, which might translate into more aid channeled directly to the government. However, we see here that democratic governments receive relatively less public sector aid. The quality of governance in the donor state correlates negatively with the relative amount of NGO aid and is in line with Dietrich (2013, 2016). Presumably donor states are more comfortable channeling aid through the recipient state's government when that government has a proven track record of responsible behavior. Physical integrity rights correlate negatively with the dependent variable. This suggests that as government respect for physical integrity rights improves, that state receives more public sector aid relative to NGO aid. Governments with less respect for physical integrity rights see more of their country's aid allocated through bypass/NGO channels. This finding makes sense, and is similar to the governance measure, as NGOs give donor states a way to continue to support development projects in countries where governments may misappropriate aid funds. The models in Table 2 provide a useful first cut. However, while the coefficient on ideology is negative as we expect, this can result from a number of scenarios: 1) conservative governments may cut NGO aid while leaving public sector aid untouched; 2) NGO aid may be untouched while public-sector aid is increased; 3) both aid types may be increased or decreased, but these changes are less pronounced for NGO aid than for public-sector aid. To address these possibilities, we use seemingly unrelated regression models to estimate two separate equations for the logged aggregate values of NGO aid and public sector aid in Table 3. To conserve space we present only the coefficient for the ideology variable.19 Model 1 is the basic model while Model 2 includes recipient fixed-effects. Table 3. Seemingly unrelated models   (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858      (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858    Standard errors in parentheses. *p ≤ 0.10 **p ≤ 0.05, ***p ≤ 0.01. View Large Table 3. Seemingly unrelated models   (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858      (1)  (2)    No Fixed Effects  Recipient FE  NGO Aid Equation          Donor Government Ideology Score  −0.205***  (0.037)  −0.227***  (0.037)  Public Sector Aid Equation          Donor Government Ideology Score  0.144***  (0.047)  0.141***  (0.047)  Observations  3998    3998    χ2  16.729    23.858    Standard errors in parentheses. *p ≤ 0.10 **p ≤ 0.05, ***p ≤ 0.01. View Large The results of these models support our initial findings. In both models the donor ideology variable has a negative and statistically significant coefficient, indicating that more conservative governments correlate negatively with the level of NGO aid. In the public sector aid models we find a positive and significant coefficient between donor ideology and aid. The magnitudes of these coefficients are also informative. Using Model 1 as an example, a one-unit increase in the donor ideology score translates into a 19 percent cut in NGO aid but leads to a 15 percent increase in public sector aid. The mean bilateral flows in this estimation sample are $5.51 million for NGO aid and $37.36 million for public sector aid. These coefficients would translate into a cut of $1.05 million in NGO aid and an increase of $5.60 million in public sector aid. Conditional Effects As we discuss previously, there is reason to believe that characteristics of the recipient state or its relationship to the donor state condition the effect of changing donor ideology. We evaluate these possibilities in Table 4. Using the ratio measure defined in 1, we examine the results of six additional models where we interact donor ideology with 1) donor exports to the recipient, 2) political similarity between the donor and the recipient, 3) security ties between the donor and the recipient, 4) the recipient state's quality of governance, 5) the recipient state's human rights record, and 6) the recipient state's level of development. Table 4. Ratio models with conditioning variables   (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998      (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large Table 4. Ratio models with conditioning variables   (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998      (1)  (2)  (3)  (4)  (5)  (6)    Exports  Affinity  Alliance  Governance  Human Rights  Development  Donor Government Ideology Score  −0.459***  (0.068)  −0.297***  (0.079)  −0.495***  (0.068)  −0.432***  (0.089)  −0.412***  (0.124)  −0.838**  (0.426)  Exports as % of Donor GDP  −0.079  (1.013)  0.136  (0.177)  0.167  (0.171)  0.146  (0.174)  0.141  (0.176)  0.142  (0.174)  UN Affinity  0.086  (0.131)  2.652***  (0.830)  0.074  (0.131)  0.086  (0.131)  0.088  (0.131)  0.082  (0.132)  Alliance  0.026  (0.242)  0.021  (0.240)  −5.430***  (1.486)  0.026  (0.241)  0.027  (0.242)  0.023  (0.241)  Recipient Governance  −0.431***  (0.109)  −0.435***  (0.109)  −0.431***  (0.109)  −0.660  (0.618)  −0.431***  (0.109)  −0.435***  (0.109)  Recipient Physical Integrity Rights  −0.101***  (0.024)  −0.100***  (0.024)  −0.101***  (0.024)  −0.101***  (0.024)  −0.041  (0.140)  −0.101***  (0.024)  Recipient GDP Per Capita  −0.077  (0.050)  −0.073  (0.050)  −0.076  (0.050)  −0.078  (0.050)  −0.077  (0.050)  −0.345  (0.296)  Donor Government Ideology Score × Exports as % of Donor GDP  0.044  (0.183)                      Donor Government Ideology Score × UN Affinity      −0.511***  (0.165)                  Donor Government Ideology Score × Alliance          1.083***  (0.301)              Donor Government Ideology Score × Recipient Governance              0.046  (0.123)          Donor Government Ideology Score × Recipient Physical Integrity Rights                  −0.012  (0.028)      Donor Government Ideology Score × Recipient GDP Per Capita                      0.054  (0.060)  Donor Welfare  −0.071***  (0.012)  −0.075***  (0.012)  −0.072***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  −0.071***  (0.012)  Donor Growth Rate  0.033**  (0.013)  0.029**  (0.013)  0.031**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  0.033**  (0.013)  Former Colony  0.032  (0.096)  0.028  (0.096)  0.028  (0.095)  0.031  (0.096)  0.031  (0.096)  0.031  (0.096)  Recipient Polity Score  0.032***  (0.009)  0.032***  (0.009)  0.033***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  0.032***  (0.009)  Recipient Civil Conflict  0.019  (0.095)  0.026  (0.094)  0.027  (0.094)  0.020  (0.094)  0.019  (0.095)  0.019  (0.094)  Post–Financial Crisis  0.279***  (0.087)  0.286***  (0.087)  0.282***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  0.279***  (0.087)  ln(All Aid Channels)  −0.206***  (0.026)  −0.202***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  −0.206***  (0.026)  Lagged ln(Aid Ratio)  0.365***  (0.021)  0.362***  (0.021)  0.364***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  0.365***  (0.021)  Constant  4.189***  (0.649)  3.422***  (0.681)  4.378***  (0.656)  4.059***  (0.704)  3.956***  (0.842)  6.059***  (2.128)  Observations  3998    3998    3998    3998    3998    3998    Robust standard errors in parentheses. *p ≤ 0.10, **p ≤ 0.05, ***p ≤ 0.01. View Large In general, these models resemble those in Table 2. The factors that have the clearest conditioning effect on decisions concerning aid channels are those associated with the relationship between the donor and recipient. Specifically, the only statistically significant interactions are for the factors tied to political and security relationships between the donor and recipient states.20 To better understand these interactive effects, we plot the marginal effects below. Figure 2 shows the marginal effect of donor ideology across the range of the UN Affinity score (Panel A) and the marginal effect of an increase in the UN Affinity score variable across the range of the ideology variable (Panel B). Plotting the marginal effect of both constituent variables helps us to better understand how the interaction of these variables shapes aid allocation patterns.21 Figure 2. View largeDownload slide Marginal effect of ideology (Panel A) and UN Affinity scores (Panel B). 95 percent confidence intervals shown. The ranges on the X-axes reflect the ranges observed in the estimation sample. The dashed line shows the distribution of the given variable within the estimation sample. Figure 2. View largeDownload slide Marginal effect of ideology (Panel A) and UN Affinity scores (Panel B). 95 percent confidence intervals shown. The ranges on the X-axes reflect the ranges observed in the estimation sample. The dashed line shows the distribution of the given variable within the estimation sample. In Panel A there is no statistically significant effect associated with an increase in the ideology variable at the low end of the X-axis. However, as the Affinity score increases we see a statistically significant negative effect emerge, indicating that a negative effect for ideology emerges at higher values of Affinity. When the donor and recipient state have poor relations, there is no evidence that ideology affects the balance between NGO aid and public-sector aid. However, when dealing with friendlier states we see that conservative governments allocate less aid through NGOs relative to public-sector channels. Furthermore, Panel A indicates that there is a statistically significant increase in the magnitude of this effect within the significant range (i.e., between Affinity = –0.2 and Affinity = 1). Panel B shows the marginal effect of an increase in the Affinity score across the range of the donor ideology variable. When looking at more liberal governments (i.e., low ideology score) we see a positive and statistically significant effect, indicating that an increase in the Affinity score produces an increase in NGO aid under liberal governments. However, as governments become more conservative (i.e., higher ideology scores) increases in Affinity eventually have a statistically significant negative effect, indicating that more conservative governments respond to improved relations by increasing government-to-government aid relative to NGO aid. These results are consistent with our expectations. When donor and recipient governments are more closely aligned it may be safer for conservative donor governments to choose to allocate more aid directly to the recipient government. When the donor has poor relations with a recipient state to begin with there is likely little incentive to alter aid channels, as concessions may not be possible. Governments are also likely to be wary of giving funds directly to a hostile government. Any development aid that is allocated to the recipient will likely go through NGOs so as not to risk funds falling into a potentially hostile government's hands. Similarly, if liberal governments are genuinely more concerned with development-oriented goals, then it seems reasonable that they would not respond to changes in political alignment in the same way that conservative governments would, given that the latter relies more on government concessions. The interaction between ideology and the alliance variable was the only other interaction with a statistically significant coefficient in Table 4. To conserve space we present the plot of these effects in the online appendix. Briefly, when looking at nonallied dyads, we find more conservative governments decrease aid to NGOs in favor of government-to-government aid. When looking at allied dyads, we see that an increase in the ideology score correlates with an increase in the relative proportion of aid that is channeled through NGOs. These results are reinforced when we consider the marginal effect of the alliance variable. Here, the effect is negative and significant for liberal governments, indicating liberals give allies less NGO aid and more direct government-to-government aid. These result runs counter to our argument. We expect conservative governments to reward allies by allocating more aid directly to the recipient government. We will refrain from speculating further as to why this might be, as there are good reasons to be skeptical of these findings. It is possible that this result represents a degree of bias resulting from the specific observations included in the estimation sample for the alliance grouping. Of the 3,998 observations, only 173 (4.3 percent) observations are allied dyads. Of those, 113 (65 percent) contain the United States as the donor state. Furthermore, many of the recipient states are in Latin America (excepting Turkey and Pakistan). When also factoring in the limited temporal domain, it seems likely that these results reflect the peculiarities of the United States during this time period, rather than any generalizable effects that we may attribute to ideology. In spite of the lack of statistically significant interaction coefficients in Table 4, these models still produce results that are generally consistent with our primary hypothesis. We present the figures showing the marginal effects for the remaining independent variables in the online appendix. In the case of exports, physical integrity rights, recipient state governance, and recipient state development, we continue to find a general negative correlation between ideology and NGO aid. However, in none of these models is there evidence of a conditioning effect between the constituent terms. We also find that physical integrity rights and recipient state governance yield negative and significant effects across most of the range of the ideology variable, suggesting that as the recipient state's governance and respect for physical integrity rights improve, we should expect to see less aid allocated through NGO channels and relatively more allocated through public sector channels. Overall, we find consistent support for the idea that conservative governments tend to allocate less money through NGOs relative to direct government-to-government transfers. Conclusion In this paper, we have explored how the ideologies of donor governments influence decisions on how foreign aid is delivered to recipient states. The analysis presented here ties together previous scholarship on aid channels (e.g., Milner 2006; Dietrich 2013, 2016) and research on ideological preferences associated with aid (Tingley 2010; Milner and Tingley 2010; Thérien and Nöel 2000). We find that more liberal governments are more likely to allocate larger shares of their aid budgets to NGOs and other civil society organizations relative to direct transfers to recipient governments. These choices are a reflection of the belief that development NGOs have priorities that more closely align with liberal governments and that assist needy populations with less money siphoned off by government kleptocrats. On the other hand, while conservative governments tend to be more skeptical of aid in general (Tingley 2010), the aid they do allocate is more heavily weighted towards the public sector as compared to their more liberal counterparts. Despite the belief that foreign aid distorts markets in the recipient country (Thornton 2002), conservative governments appear to be more interested in meeting short-term self-interested goals like building economic and political ties. With the exception of political similarities between the donor and the recipient, we find little evidence that economic ties, or quality of governance (broadly conceived) within the recipient state, have any conditioning effect on how ideology shapes aid allocation decisions. Still, our finding that political similarities, as captured by the UN Affinity measure, do condition the effect of ideology is important, as it suggests some constraints on how governments can manipulate aid allocation. Only when the donor and recipient state are closely aligned do we see ideological differences emerge in how donor governments allocate aid. The insignificant effect at the low end of the affinity scale does not mean that those states receive no aid but does suggest that the domestic political differences within the donor state are muted in their influence on aid channel allocation decisions. Overall, our work helps clarify when and how conservative and liberal governments employ foreign aid. Future research should continue to explore these relationships. Due to sporadic reporting of NGO aid data by donor states, most of the variation we see in the dependent variable is cross-sectional in nature—that is, varying by donors and dyads. Where we do have more complete data over time, we do indeed see variation in the relative amounts of NGO aid allocated within a given donor-recipient dyad. As disaggregated data become more regularly and consistently reported, future studies will be better able to evaluate the relative influence of within-country variation in ideology as compared to cross-national differences.22 The aid policy choices made by both liberal and conservative governments reflect their political interests and those of their constituents. While conservative governments tend to allocate less aid, these conservative governments allocate aid in such a way that may produce more immediate benefits for the donor state. For liberal governments, it may also be the case that the promotion of development-oriented goals is partly intended to generate long-term benefits for the donor state. Reducing poverty, increasing incomes, and promoting economic and political stability in developing states certainly yields potential benefits to donors through increasing consumption of donor-produced goods and services. Even if ODA does not bring meaningful poverty alleviation, perhaps aid can help buy an increasingly positive image for the donor country within the recipient state. Our findings also dovetail nicely with Dietrich's (2013) work, which suggests that when donors are concerned about governance quality, they are more likely to utilize bypass channels. The liberal interest in strengthening economic and political development of recipient states is also associated with bypass. These findings also offer insight into the future of foreign aid distribution. The average portion of DAC donors’ aid moneys being channeled through NGOs in 2004 was approximately eight percent. By 2013, that number had more than tripled to just under 28 percent (OECD 2015). While the majority of foreign aid is not being delivered by NGOs, the influence of these organizations in international development is increasing. Recognizing the conditions under which these NGO agents are most likely to be empowered by national principals, as well as understanding when NGOs can (and potentially cannot) enhance the effectiveness of foreign aid, will be an important step for anticipating the global future of aid and poverty alleviation. Changes in donor ideology and the donor's political ties with potential recipient states may be important factors in determining when and where NGOs play a greater role in the aid delivery process. Supplemental Information Supplemental information is available at the Foreign Policy Analysis data archive. 1 We refer to leftist parties as “liberal” and rightist parties as “conservative.” Our usage of liberal is not synonymous with the classical sense (i.e., free markets, smaller government, etc.). 2 Note that we use the terms “public-sector aid” and “government-to-government aid” interchangeably. 3 Schneider and Tobin (2013) demonstrate that when member states have heterogeneous preferences, aid allocation goals may more closely reflect the unique interests of the institutions, but when member preferences are more similar, aid allocation will more closely reflect national interests. 4 Skepticism is not unfounded. Kono and Montinola (2012) find that autocratic regimes are more likely to divert resources to build military capacity. 5 The correlation between the NGO and multilateral channels is 0.54 in our primary estimation sample, whereas the correlation between these channels and public sector aid is only 0.19 and 0.15, respectively. 6 Left-wing governments are not necessarily behaving altruistically or nonstrategically by funding poverty alleviation projects. Liberal governments may see this type of project as a long–term investment in soft power (Nye 2004). Beyond that, these projects may benefit donor countries economically, but the time horizon for those benefits is longer than the types of foreign aid projects favored by their conservative counterparts. In recent years, USAID has funded programs based on both ending extreme poverty and expanding trade. Both of these goals are related to creating economic growth and economic opportunity, but the approaches to these goals are different. 7 Empirically verifying this claim is difficult, but ODA does appear to lengthen the tenure of leaders, particularly in nondemocratic regimes (Licht 2010). 8 We use aid commitments rather than disbursements. 9 We focus only on individual state donors and exclude multilateral donors from our analysis. See the online appendix for more detail on case selection and data coding. 10 We focus on NGOs and public sector aid for several reasons. As we discuss above, previous studies have focused on government decisions to allocate aid through multilateral organizations (see Milner 2006; Milner and Tingley 2013b). Further, data on public-private partnerships is extremely limited. Utilizing public sector and NGO aid channels, our primary models include 3,998 complete observations. By comparison, there are 3,827 missing observations for the public-private partnership variable—approximately 96% of our estimation sample. Lastly, given the ambiguity of the “other” category we do not include it in our analyses as we have no clear expectations as to how ideology would affect decisions related to this category. 11 The unlogged variable ranges from 0.000018 to 6,705.562 with a mean of 11.82. 12 See online appendix. 13 This variable is measured as ln(x + 1). 14 This variable comes from Dietrich (2013, 704), which she coded from the CIA World Factbook. We forward coded it for years beyond her study. 15 Data were adjusted to 2005 dollars and weighted using GDP figures from the World Bank (World Bank 2014). 16 A test for the presence of serial correlation is significant at the 0.01 level (Wooldridge 2010). 17 A Hausman test of Model 3 in Table 2 yields a significant coefficient, but we opt to use random effects over fixed effects for a few reasons. First, because of the large number of cross-sectional units (i.e., 1,191 dyads) relative to time periods, fixed effects soak up the majority of within-unit variation. Random effects allow us to control for the possibility that various dyadic relationships may have different baseline ratios of aid allocation. Estimating individual parameters for individual dyad fixed effects also reduces the model's degrees of freedom. Using random effects allows us to control for heterogeneity across units without the costs associated with fixed effects. Second, some variables, like alliances, would be perfectly collinear with dyad fixed effects, preventing us from analyzing theoretically important variables. Lastly, Clark and Linzer (2015) argue that the Hausman test is often insufficient to evaluate the appropriateness of random effects. They find that when the number of units and the number of observations for each unit are small, and when the correlation between the key independent variables and the unit effects are relatively low, random effects models outperform fixed effects models. Excepting the alliance dummy, the correlations between our independent variables and the unit effect and the independent variables is low. We have included robustness checks using recipient fixed effects in the online appendix (see A7 and A8). 18 We present the results from our models using the full ideology measure in Tables A4 and A5. 19 See Table A3 for full table. 20 We follow Brambor, Clark, and Golder (2006) and include all constituent terms in the interactive models. 21 See Berry, Golder, and Milton (2012) on symmetry in interaction terms. 22 See online appendix and Table A1. References Acht Martin, Mahmoud Toman Omar, Thiele Rainer. 2015. “Corrupt Governments Do Not Receive More State-To-State Aid: Governance and the Delivery of Foreign Aid Through Non-State Actors.” Journal of Development Economics  114: 20– 33. Google Scholar CrossRef Search ADS   Alesina Alberto, Dollar David. 2000. “Who Gives Foreign Aid to Whom and Why?” Journal of Economic Growth  5 ( 1): 33– 63. Google Scholar CrossRef Search ADS   Bebbington Anthony, Riddell Roger. 1995. “The Direct Funding of Southern NGOs by Donors: New Agendas and Old Problems.” Journal of International Development  7 ( 6): 879– 93. Google Scholar CrossRef Search ADS   Berry William D., Golder Matt, Milton Daniel. 2012. “Improving Tests of Theories Positing Interaction.” Journal of Politics  74 ( 3): 653– 71. Google Scholar CrossRef Search ADS   Brambor Thomas, Clark William Roberts, Golder Matt. 2006. “Understanding Interaction Models: Improving Empirical Analyses.” Political Analysis  14 ( 1): 63– 82. Google Scholar CrossRef Search ADS   Brech Viktor, Potrafke Niklas. 2014. “Donor Ideology and Types of Foreign Aid.” Journal of Comparative Economics  42 ( 1): 61– 75. Google Scholar CrossRef Search ADS   Bueno de Mesquita Bruce, Smith Alastair. 2007. “Foreign Aid and Policy Concessions.” Journal of Conflict Resolution  51 ( 2): 251– 84. Google Scholar CrossRef Search ADS   Carney Dana R., Jost John T., Gosling Samuel D., Potter Jeff. 2008. “The Secret Lives of Liberals and Conservatives: Personality Profiles, Interaction Styles, and the Things They Leave Behind.” Political Psychology  29 ( 6): 807– 40. Google Scholar CrossRef Search ADS   Cingranelli David L., Richards David L., Chad Clay K.. 2014. “CIRI Human Rights Data Project (v.2014-04-14).” http://www.humanrightsdata.com/. Clark Tom S., Linzer Drew A.. 2015. “Political Science Research and Methods Should I Use Fixed or Random Effects? Should I Use Fixed or Random Effects?” Political Science Research and Methods  3( 2): 399– 408. Google Scholar CrossRef Search ADS   Clarke Gerard. 2007. “Agents of Transformation? Donors, Faith-Based Organizations and International Development.” Third World Quarterly  28 ( 1): 77– 96. Google Scholar CrossRef Search ADS   Dietrich Simone. 2013. “Bypass or Engage? Explaining Donor Delivery Tactics in Foreign Aid Allocation.” International Studies Quarterly  57 ( 4): 698– 712. Google Scholar CrossRef Search ADS   ——— 2016. “Donor Political Economies and the Pursuit of Aid Effectiveness.” International Organization  70: 65– 102. Google Scholar CrossRef Search ADS   Eade Deborah. 2004. “Preface.” In Development, NGOs, and Civil Society , edited by Eade Deborah, 9– 14. Oxfam GB. Easterly William. 2001. The Elusive Quest for Growth: Economists’ Adventures and Misadventures in the Tropics . Cambridge, MA: MIT Press. Fleck Robert K., Kilby Christopher. 2006. “How Do Political Changes Influence U.S. Bilateral Aid Allocations? Evidence from Panel Data.” Review of Development Economics  10 ( 2): 210– 23. Google Scholar CrossRef Search ADS   ——— 2010. “Changing Aid Regimes? U.S. Foreign Aid from the Cold War to the War on Terror.” Journal of Development Economics  91: 185– 97. Google Scholar CrossRef Search ADS   Fruttero Anna, Gauri Varun. 2005. “The Strategic Choices of NGOs: Location Decisions in Rural Bangladesh 1.” Journal of Development Studies  41 ( 5): 759– 87. Google Scholar CrossRef Search ADS   Gabel Matthew J., Huber John D.. 2000. “Putting Parties in Their Place: Inferring Party Left-Right Ideological Positions From Party Manifestos Data.” American Journal of Political Science  44 ( 1): 94– 103. Google Scholar CrossRef Search ADS   Gibler Douglas M. 2009. International Military Alliances, 1648–2008 . CQ Press. Google Scholar CrossRef Search ADS   Gleditsch Nils Petter, Wallensteen Peter, Eriksson Mikael, Sollenberg Margareta, Strand Havard. 2002. “Armed Conflict 1946 –  2001: A New Dataset.” Journal of Peace Research  39 ( 5): 615– 37. Graham Jesse, Haidt Jonathan, Nosek Brian A.. 2009. “Liberals and Conservatives Rely on Different Sets of Moral Foundations.” Journal of Personality and Social Psychology  96 ( 5): 1029– 46. Google Scholar CrossRef Search ADS PubMed  Hibbs Douglas A. 1977. “Political Parties and Macroeconomic Policy.” American Political Science Review  71 ( 4): 1467– 87. Google Scholar CrossRef Search ADS   Hirsh Jacob B., DeYoung Colin G., Xu Xiaowen, Peterson Jordan B.. 2010. “Compassionate Liberals and Polite Conservatives: Associations of Agreeableness with Political Ideology and Moral Values.” Personality & Social Psychology Bulletin  36 ( 5): 655– 64. Google Scholar CrossRef Search ADS PubMed  Hoadley J. Stephen. 1980. “Small States as Aid Donors.” International Organization  34 ( 01): 121. Google Scholar CrossRef Search ADS   International Monetary Fund. 2014. “Direction of Trade Statistics (DOTS).” http://data.imf.org/?sk=9D6028D4-F14A-464C-A2F2-59B2CD424B85. Kaufmann Daniel, Kraay Aart, Mastruzzi Massimo. 2015. World Governance Indicators . Tech. rep., World Bank. Koch Dirk-Jan, Dreher Axel, Nunnenkamp Peter, Thiele Rainer. 2009. “Keeping a Low Profile: What Determines the Allocation of Aid by Non-Governmental Organizations?” World Development  37 ( 5): 902– 18. Google Scholar CrossRef Search ADS   Kono D. Y., Montinola G. R.. 2012. “The Uses and Abuses of Foreign Aid: Development Aid and Military Spending.” Political Research Quarterly  66 ( 3): 615– 29. Google Scholar CrossRef Search ADS   Lai Brian, Morey Daniel S.. 2006. “Impact of Regime Type on the Influence of US Foreign Aid.” Foreign Policy Analysis  2 ( 4): 385– 404. Google Scholar CrossRef Search ADS   Laver Michael, Budge Ian. 1992. Party Policy and Government Coalitions . St. Martin's Press. Google Scholar CrossRef Search ADS   Laver Michael, Garry John. 2000. “Estimating Policy Positions from Political Texts.” American Journal of Political Science  44( 3): 619– 34. Google Scholar CrossRef Search ADS   Licht Amanda A. 2010. “Coming into Money: The Impact of Foreign Aid on Leader Survival.” Journal of Conflict Resolution  54 ( 1): 58– 87. Google Scholar CrossRef Search ADS   Marshall Monty G., Jaggers Keith, Gurr Ted Robert. 2014. “Polity IV Project.” http://www.systemicpeace.org/polity/polity4.htm. McLean Elena V. 2015. “Multilateral Aid and Domestic Economic Interests.” International Organization  69 ( 01): 97– 130. Google Scholar CrossRef Search ADS   Meernik James, Krueger E. L., Poe S. C.. 1998. “Testing Models of US Foreign Policy: Foreign Aid During and After the Cold War.” The Journal of Politics  60 ( 1): 63– 85. Google Scholar CrossRef Search ADS   Milner Helen V. 2006. “Why Multilateralism? Foreign Aid and Domestic Principal-Agent Problems.” In Delegation and Agency In International Organizations , edited by Hawkins Darren G., Lake David A., Nielson Daniel N., Tierney Michael J., chap. 4, 107– 39. New York: Cambridge University Press. Google Scholar CrossRef Search ADS   Milner Helen V., Tingley Dustin. 2011. “Who Supports Global Economic Engagement? The Sources of Preferences in American Foreign Economic Policy.” International Organization  65 ( Winter): 37– 68. Google Scholar CrossRef Search ADS   ——— 2010. “The Political Economy of U.S. Foreign Aid: American Legislators and the Domestic Politics of Aid.” Economics and Politics  22 ( 2): 200– 32. ——— 2013a. “Public Opinion and Foreign Aid: A Review Essay.” International Interactions  39 ( 3): 389– 401. Google Scholar CrossRef Search ADS   ——— 2013b. “The Choice for Multilateralism: Foreign Aid and American Foreign Policy.” Review of International Organizations  8 ( 3): 313– 41. Google Scholar CrossRef Search ADS   Murdie Amanda Marie, Kakietek Jakub. 2012. “Do Development INGOs Really Work? The Impact of International Development NGOs on Human Capital and Economic Growth.” Journal of Sustainable Society . 1( 1): 1– 12. Nancy Gilles, Yontcheva Boriana. 2006. Does Ngo Aid Go to the Poor? Empirical Evidence from Europe . International Monetary Fund. Nöel Alain, Thérien Jean-Philippe. 1995. “From Domestic to International Justice: The Welfare State and Foreign Aid.” International Organization  49 ( 3): 523– 53. Google Scholar CrossRef Search ADS   Nöel Alain, Thérien Jean-Philippe, Dallaire Sebastian. 2004. “Divided Over Internationalism: The Canadian Public and Development Assistance.” Canadian Public Policy  30 ( 1): 29– 45. Google Scholar CrossRef Search ADS   Nunnenkamp Peter, Weingarth Janina, Weisser Johannes. 2009. “Is NGO Aid Not so Different After All? Comparing the Allocation of Swiss aid by Private and Official Donors.” European Journal of Political Economy  25 ( 4): 422– 38. Google Scholar CrossRef Search ADS   Nye Joseph S. 2004. Soft Power: The Means to Success in World Politics . Public Affairs. OECD. 2013. Evaluating Development Activities: 12 Lessons from the OECD DAC . Tech. rep., Organization for Economic Cooperation and Development. ——— 2015. Creditor Reporting System Database . Tech. rep., Organization for Economic Cooperation and Development (OECD). Paxton P., Knack S.. 2012. “Individual and Country-Level Factors Affecting Support for Foreign Aid.” International Political Science Review  33 ( 2): 171– 92. Google Scholar CrossRef Search ADS   Radelet Steven. 2006. “A Primer on Foreign Aid.” Center for Global Development Working Paper 92. Riddell Roger C., Bebbington Anthony, Peck Lennart. 1995. “Promoting Development by Proxy. An Evaluation of the Development Impact of Government Support to Swedish NGOs.” SIDA Evaluation Report (Sweden) . Schneider Christina J., Tobin Jennifer L.. 2013. “Interest Coalitions and Multilateral Aid Allocation in the European Union.” International Studies Quarterly  57 ( 1): 103– 14. Google Scholar CrossRef Search ADS   Strezhnev Anton, Voeten Erik. 2012. “United Nations General Assembly Voting Data.” URL: http://hdl.handle.net/1902.1/12379. Tender Judith. 1982. Turning Private Voluntary Organizations into Development Agencies: Questions for Evaluation . USAID, Bureau of Food and Voluntary Assistance. Thérien Jean-Philippe. 2002. “Debating Foreign Aid: Right Versus Left.” Third World Quarterly  23 ( 3): 449– 66. Google Scholar CrossRef Search ADS   Thérien Jean-Philippe, Nöel Alain. 2000. “Political Parties and Foreign Aid.” American Political Science Review  94 ( 1): 151– 62. Google Scholar CrossRef Search ADS   Thornton Mark. 2002. Corruption and Foreign Aid . Online Publications on Ludwig von Mises Institute. Tingley Dustin. 2010. “Donors and Domestic Politics: Political Influences on Foreign Aid Effort.” The Quarterly Review of Economics and Finance  50 ( 1): 40– 49. Google Scholar CrossRef Search ADS   USAID. 2013. Where Does the Money Go?  Tech. rep., United States Agency for International Development. Volkens Andrea, Lacewell Onawa, Lehmann Pola, Regel Sven, Schultze Henrike, Werner Annika. 2011. The Manifesto Data Collection. Manifesto Project (MRG/CMP/MARPOR) , Berlin: Wissenschaftszentrum Berlin für Sozialforschung (WZB). Wooldridge Jeffrey M. 2010. Econometric Analysis of Cross Section and Panel Data . Cambridge: The MIT Press, 2nd edition. World Bank. 2014. World Development Indicators 2014 . © The Author (2017). Published by Oxford University Press on behalf of the International Studies Association. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com

Journal

Foreign Policy AnalysisOxford University Press

Published: Oct 18, 2017

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off