Lending Democracy: How Governance Aid May Affect Freedom

Transcription

1 Lending Democracy: How Governance Aid May Affect Freedom Richard Nielsen and Daniel Nielson! Abstract: Recent studies of the political impacts of aid on the democracy level of recipient countries have come to mixed conclusions. While these studies often propose specific mechanisms through which aid of various types might positively or negatively impact the democracy level of recipients, most studies fail to test these hypotheses. Instead, past research has primarily attempted to establish an overall direction of correlation between aggregated aid and democracy. We develop several potential mechanisms through which aid of various types might have positive or negative impacts on the democracy level of recipient countries. We test five hypotheses linking foreign aid to democratic outcomes: (1) that democracy aid improves democracy, (2) that aid channeled through NGOs increases democracy by promoting civil society (3) that aid for economic growth and human capital provides an environment conducive to democracy, and (4) that fungible aid is analogous to other natural resource windfalls and may decrease democratic accountability. Using an original foreign aid dataset which allows us to differentiate aid by purpose, we test the hypotheses associated with these mechanisms and find that education and democracy aid lead to incremental increases in democracy, while aid-in-cash is associated with larger bursts of democratization. Introduction Foreign aid is heterogeneous. Given by bilateral, multilateral, and private lenders, both in cash and in kind, foreign assistance provides funding for an entire spectrum of activities related to development, ranging from humanitarian food aid and coal-fired electricity plants to telecommunication satellites and human rights conferences ( with lunch ) (OECD 2002). Moreover, it seems likely that aid has changed over time, responding to structural changes such as the end of the Cold War (Lai 2003), changes in state and development bank priorities (Nielson and Tierney 2003), and the bureaucratic and organizational changes such as the projectization of aid. Some aid is given in the form of loans at nearly market rates, while much is given well below competitive interest rates and may consist partially or wholly of grants. The sheer variety in sources,! Richard Nielsen is a PhD student at Harvard University. Daniel Nielson is Associate Professor at Brigham Young University. Prepared for delivery at the 2008 Annual Meeting of the American Political Science Association, August 28-31,

2 purposes, and funding of aid suggests that the political and economic effects of aid in recipient countries may vary significantly with the type of aid received. Nevertheless, most scholars attempting to identify the effects of aid have used highly aggregated measures of assistance, considering relationships between aid and growth (Burnside and Dollar 2000), democracy (Knack 2004), and corruption (Alesina 2002) as if all aid were likely to have similar effects on these outcomes (see Clemens et al. 2004, for an important exception). Here, we consider the effects of development assistance on democracy and governance when we subdivide aid by sector. This disaggregation is vital because virtually all theories about the domestic political effects of assistance rely on mechanisms implying that some types of aid will lead to different outcomes than other types. However, despite suggesting sector-specific mechanisms, previous studies have largely failed to test these hypotheses, instead attempting to establish a general correlation between aggregated assistance and democracy levels. The few studies that have disaggregated aid (Finkel et al. 2007; Bermeo 2007), have nevertheless failed to test competing mechanisms of how aid might promote or discourage democratization. Additionally, these studies have often ignored the possibility of reverse causation and have suffered from incomplete datasets of foreign assistance. We are not immune to many of these same difficulties, but we address them to the extent possible given the observational nature of our data. In this manuscript, we explore the mechanisms through which aid influences democratic outcomes in recipient countries. We first review theoretical arguments linking aid to changes in the polity of recipient countries and consider previous findings regarding the salience of each of these causal mechanisms. We then consider how the derivative hypotheses of these theories might be tested. Using several original datasets, we then test these hypotheses to see which are supported when we disaggregate aid in the ways suggested by the competing theories. We conclude by 2

3 discussing our findings and their implications for democracy and democratization in the developing world. Foreign Aid as a Tool of Democracy Promotion Since the Marshall Plan, Western foreign aid donors have attempted to use aid to shape the politics and political institutions of recipient countries. Of course, donors pursued these objectives alongside developmental and humanitarian goals. Throughout its modern history, foreign aid has almost always been highly politicized, flowing disproportionately to developing countries with great strategic and political importance to donor states. In post-war Europe, the U.S. leaders were vitally concerned with curbing the spread of communism and believed that allocating substantial resources to rebuild Europe would counter what they saw as a looming threat. Thus, even from its earliest hours, foreign aid was viewed as a way of potentially supporting and promoting democracy abroad. As the Cold War deepened, the political competition between the United States and the Soviet Union continued to make aid strongly political, although Western aid was certainly not intended to universally promote democracy between the 1960s and the early 1990s. Instead, both superpowers used aid to buy allies and promote regime changes that would increase their respective spheres of influence. The fall of the Soviet Union ushered in a decade of foreign aid aimed squarely at promoting democracy. Bolstered by the belief that democracies were more peaceful and better economic partners, Western donors viewed the end of the Cold War as an opportunity to reshape the world toward like-minded institutions. The U.S. was particularly vigorous in its democracy promotion efforts, claiming democratic transitions in Ukraine, Georgia, and Serbia as clear successes for its foreign aid policies. European aid for democracy increased dramatically as well. Although several 3

4 donors, notably the U.S., had maintained human rights as a condition for aid since the mid-1970s, donors began to allocate aid intended to improve human rights on a large scale only as the Cold War was visibly winding down. The Nordic countries in particular officially adopted human rights conditions in their aid policies in 1988, and other Western donors followed suit through the 1990s. The promotion of democracy became a centerpiece of the new thinking about foreign aid. Aid s Sector-Specific Effects on Democracy The effect of aggregate foreign aid on democracy seems to be either insignificant (Knack 2004) or negative (Kalyvitis and Vlachaki 2008; Djankov et al. 2006), but this leaves questions. The sectoral make-up of aid has changed from , the period of most previous studies, and if different sectors have different political effects in the recipient countries, then we have to qualify our understanding of aggregate aid to mean aggregate aid as it was in the period of our sample. This problem is especially acute because studies of aggregate aid still tend to rely on sectorbased explanations to link assistance flows to political outcomes. Studies arguing that aid might improve democracy have asserted that education, economic growth, democracy aid, and aid to NGOs are all possible causal mechanisms, while studies arguing that aid decreases democracy have focused on the role of fungible aid in propping up dictatorships. In the end, testing these sectorlevel theories with aggregated aid data provides very little evidence on which to evaluate theories linking aid and democracy, with results indicating aggregate correlations without confirming specific causal pathways. Below, we draw on the existing literature on the politics of aid to identify potential causal mechanisms linking aid and democracy and evaluate the somewhat sparse existing evidence for and against them. Of the many possible mechanisms through which aid might impact democracy, we choose to evaluate four: (1) the possibility that democracy aid positively influences democracy, (2) 4

5 the possibility that aid flowing through NGOs promotes democracy by strengthening civil society, (3) the possibility that aid promoting economic growth and human capital development strengthens democracy, and (4) the possibility that highly fungible aid and aid flowing through the government undermines democracy by propping up the existing regime. The yearly flows of aid to these sectors, categorized according to the discussion which follows, is shown in Figure 1. Figure 1: Aid Flows for Selected Sectors ( ) Aid Amount (millions of US 2000 dollars) Education Aid Democracy Aid Growth Aid Cash Aid Aid through NGOs Year For each, we develop testable hypotheses intent on discerning what evidence exists for each mechanism. We expect that aid will impact democracy and democratization through a variety of mechanisms associated with different types of assistance, meaning that the effect of aggregate aid on democracy will depend crucially on the types of aid that make up the aggregate. It thus seems a bit nonsensical to us to analyze the effect of aggregate aid on democracy without considering aid s 5

6 constituent parts the net impact of aid will depend on the relative amounts of each type of aid that a developing country receives. Democracy Aid First, we consider aid that donors claim is intended to promote democratization. This aid ostensibly funds activities that directly strengthen the democratic practices and institutions within recipient states: election monitoring, training for parliamentarians, funding of human rights organizations, etc. If effective, democracy aid should directly improve the constituent features of democracy, which in turn should be captured in the measures of democracy we use here. Moreover, if this mechanism is at work, then subsectors of democracy aid should be directly effective as well: funding for free media should lead to increases in measures of the freedom of the media, etc. While the most direct democracy aid is funded bilaterally, international financial institutions like the World Bank often fund projects intended to promote good governance and support the rule of law, despite clauses in their charters requiring them to be politically neutral. Finkel et al. (2007) find that USAID democracy aid is positively correlated with democratization. While impressive, their analysis may not adequately deal with reverse causality the possibility that the magnitude of democracy aid flows are affected by the democracy level of the recipient. More generally, it is unclear whether democracy aid from other OECD donors will have similar effects since the US democracy aid program is unique in many respects. Other studies have come to less promising conclusions about the role of democracy aid in democracy promotion. Ethier (2003), for example, argues that democracy aid has not proven to be very effective, but her evidence is mostly secondary and relies on qualitative evidence that does not appear to have been selected systematically, perhaps revealing more nuanced details at the expense of generalization. 6

7 Finkel et al (2007) make the important step of linking subsectors of democracy aid to particular outcomes, showing that aid for free speech, political participation, etc, leads to increases in democratic rights in those particular areas. We agree that disaggregating the outcome is potentially as important as disaggregating aid. Aid generally aims at strengthening elections, increasing the institutional capacity of legislative and judicial branches, fostering civil society groups and the like; it is important to determine whether this democracy aid is having its intended specific effects as well as a more general impact on overall democracy. Likewise, Kalyvitis and Vlachaki (2007) also find evidence that democracy aid programs are correlated with increased democracy. Aid Through NGOs Some scholars propose an additional mechanism through which aid promotes democracy. Noting that a strong civil society seems to be an important element in establishing democracy, Carothers and others have argued that aid which flows through NGOs and civil society groups will have the effect of strengthening these groups in society, perhaps even if the aid is not intended to promote democracy directly. By providing revenue for these groups, aid donors strengthen their ability to provide voices of opposition in closed societies. Much of this aid is intended to directly promote democratization, suggesting that this mechanism is compatible with the previous theory linking aid directed specifically at democracy with political reform. In particular, USAID has spent significant funds assisting civil society by supporting labor unions, women s organizations, and other local NGOs. Perhaps all aid that goes through NGOs, regardless of sector, builds civil society and is thus effective (Carothers and Ottaway 2000, 13). One might also believe that democracy aid through NGOs will be particularly effective because NGOs have incentives to see that the government 7

8 liberalizes, whereas the government may face countervailing incentives. Thus, democracy aid that goes through the government may not accomplish much, but democracy aid through NGOs may. However, others are doubtful about the positive effect of aid to NGOs. In particular, Carapico (2002) argues that democracy aid channeled through NGOs to promote democracy in the Middle East actually undermined the legitimacy of the NGOs and led state authorities to crack down on them. Ottoway in Carothers and Ottoway (2000) discusses the case of Zambia, where the regime directly threatens NGOs, which have struggled to survive because incumbent leaders believe the NGOs seek chiefly to weaken regime power. So perhaps democracy aid works best when given to NGOs that are in regimes that are not repressive enough to suppress all NGO activity. Similar repression is currently occurring in Zimbabwe, where Robert Mugabe has shut down all foreign aid operations he suspects of having supported his political opponents. Aid for Education and Growth Aid may increase democracy levels by increasing literacy, health, and wealth (Knack 2004; Goldsmith 2001), factors which may lead to democratization (Lipset 1959; Barro 1997). While these mechanisms may be related, they are also distinct in important ways. Each suggests that aid to a particular sector, either education aid, health aid, or aid for long-term capital growth, should lead to increased levels of democracy. This implies a two-stage mechanism aid has a direct impact on education (health, growth), which in turn increases democracy. As Goldsmith (2001) notes, better educated and healthier people, in turn, may make better informed and more active citizens, who are the lifeblood for democratic institutions (137). Gazibo (2005) takes a related but distinct tack on the effects of aid-induced growth on political outcomes. Sustained aid is particularly important [to emerging democracies] in a context of severe scarcity, because in situations of social unrest, it allows political elites to address the popular demands that threaten the viability of fragile democracies (69). 8

9 Others argue that aid cannot influence democratization via economic growth because aid has not been consistently linked with subsequent economic growth and Knack (2004) finds that overall aid flows appear to have no effect, positive or negative, on democratization. Does Aid Weaken Democracy? In contrast to the generally positive expectations of the effects of aid summarized above, other scholars argue that aid of various types retards democratization. Although this argument takes many forms, the basic idea is that aid acts like a natural resource, providing a source of income for a regime independent of its ability to raise taxes. This liberates the government from the need to pay attention to domestic public opinion and creates incentives for rentier behavior. In support of this argument, work by Djankov, Montalvo, and Reynol-Querol (2006) finds that aid significantly decreases the democracy level of recipient countries. The authors argue that this finding results from the ability of aid to strengthen non-democratic regimes, a finding similar to Bueno de Mesquita et al (2003), who present evidence that economic aid increases the tenure of leaders in authoritarian regimes. This was certainly the thesis of early work on foreign aid, which argued that assistance merely propped up authoritarian regimes. Danaher, Berryman, and Benjamin (1987) argue that aid simply supports elites in developing countries, extending their tenure in office and the brutal means with which they can suppress rebellion. Chomsky and Herman (1979) concur, arguing that increased aid lends direct support to military dictators and indirectly funds increased torture and other violations of physical integrity. A very dramatic expression of this negative effect of aid could occur if aid flows increase the probability of civil conflicts and coups (Grossman 1992). These types of mechanisms would rely on the fungibility of aid: it is either directly seized by recipient governments or it substitutes for expenses the recipient would have otherwise undertaken and thus frees up funds to be spent elsewhere at the recipient s discretion. Sogge s (2002) findings 9

10 corroborate this claim, presenting evidence that that democracy assistance has failed largely because donors have pursued their own interests via aid rather than allocate it with recipient needs in mind. A related set of arguments holds that aid dependence induces poor governance, although the results of aid may cut both ways. Brautigam and Knack (2004) argue that in sub-saharan Africa, large influxes of aid may undermine the ability of governments to govern, undercutting the democracy that aid is often intended to promote. However, Goldsmith (2001) shows that for subsaharan Africa, aid leads to net gains in governance quality, albeit the gains are so small as to be easily overwhelmed by other factors. Complementing the mechanism above where aid through NGOs is effective in causing democratization, it may be that aid which flows through the central government is bad for democracy, regardless of purpose, because it weakens accountability and encourages rentier behavior (Friedman 1958). Aid Selectivity and Democratic Reform Other possible mechanisms may also account for aid s impact on the level of democracy in a recipient country. We mention them here but, for reasons of space and scope, we do not test in the empirical section. In addition to giving aid specifically intended to promote democracy, donors have also started selectively withholding economic aid from countries that violate human rights or fail to make democratic reforms. There is significant debate about whether donors really do this or not Nielsen (2007) finds that many do, but not all. Crawford (2001) looks seriously at whether withholding aid leads to political improvements and provides the only test we have seen (188). He finds some support that when donors withhold aid because of human rights abuses and nondemocratization, regimes respond with improvements. 10

11 Dunning (2004) attempts to show that the dynamics of aid for democracy assistance fundamentally changed at the end of the Cold War. Still using aggregate aid, he reproduces Goldsmith s analysis showing that the positive correlation between aid and improved governance only holds for Africa after 1987, corresponding with the thawing of the Cold War and eventual break-up of the Soviet Union. Like Knack and Goldsmith, Dunning only looks at aggregated aid and explains its democratizing effects in the post-cold War era as the result of increased reliance on conditionality once the Cold War rivalry for client states has ended, donors are more willing to risk losing a client in order to make aid conditional on political liberalization (known as aid selectivity). This finding is corroborated by Nielsen (2007), who shows that countries where human rights are violated receive less infrastructure and production aid (the kind used to buy support) but only after the end of the Cold War. Testing the Mechanisms Some of these mechanisms have already been partially tested. Knack has already found that there is no correlation between overall aid and democracy, although Djankov et al (2006) and Kalyvitas and Vlachaki (2008) find a negative correlation. However, as noted above, Finkel et al. (2007) and Kalyvitis and Vlachaki (2007) find that democracy aid increases democracy. In this manuscript we attempt to resolve some of this ambiguity and add value to the existing literature. We test previously untested hypotheses. And we subject previously tested hypotheses to tests on for the previously tested ones, we use the detail of an improved dataset of foreign aid to provide more detailed tests of their observable implications. 11

12 Endogeneity: Do Democracies Receive Different Kinds of Aid? Identifying the effects of sector aid on democracy is an extremely difficult case for causal inference given the behavior of donor countries aid for these sectors is almost certainly dependent upon the recipients current level of democracy the donor s perceptions of a recipient s likelihood of future democratization. Svensson (1999) finds that aid on average is not channeled to more democratic countries, even though there are large cross-country differences between major donors (abstract), but Alesina and Dollar (2000) find contrary evidence that aggregate aid flows disproportionately to new democracies. In any case, there is plausibly reciprocal causation between the aid sectors we use and democracy and unfortunately for our analysis, in most cases we have not clear theoretical prediction about direction the endogenous relationship. Most obviously, it is unlikely that democracy aid is given irrespective of the democracy level of the recipient. If the donor s primary goals are to promote democracy where it is weakest and thus needed the most, then democracy aid will flow disproportionately to countries where democracy measures are weakest. If donors believe that democracy aid is most helpful in transitioning polities, then they will give their democracy aid disproportionately to countries with middling democracy scores (and recent changes in democracy levels). The same is true if donors are trying to display the effectiveness of their aid by picking winners. If this happens then we would expect them to pick middling polities that are already showing signs of increased liberalization. Aid through non-governmental organizations my be similarly endogenous. NGOs are likely to be more repressed in autocratic polities, so there are more NGOs in more democratic countries; as the number of NGOs increases, donors will be able to use NGOs for more aid projects, so more aid will flow through NGOs. However, if the government is particularly undemocratic, then donors may be less willing to hand democracy aid directly to the government because the government probably can t be trusted to use it in the way the donor would like (several major donors including 12

13 the US claim to do this. See Crawford 2001). If this is the case, then they will give more aid through NGOs in autocratic countries. These two possibilities are cross-cutting, again leaving the overall direction of the potentially endogenous relationship unclear. Aid for education or growth, and aid that comes directly as cash is likely to be endogenous with democracy and democratization if donors think that democracies use these types of aid more effectively. Bermeo s (2007) work shows that donors heavily favor recipients with good policies (including aspects of democracy) when they allocate growth aid and we suspect that donors may allocate education aid or aid in cash similarly. With all of these types of aid, there may also be endogeneity between aid and changes in democracy, which we use below as a proxy for democratization. Thus, simply differencing our measures is not the answer to the problem of reverse causation. The problem is that aid policymakers may not only base their allocation on a recipient s current level of democracy but instead also attempt to predict future changes in democracy. To the extent that policy-makers expectations about future democratization are correlated with actual moves toward democracy, endogeneity may persist. If we were able to accurately measure the degree to which policy-maker expectations about future democratization factor into aid flows, we might avoid this problem, but such a measure would be very difficult to construct. Addressing Endogeneity The most satisfactory solution to potential reverse causation is to convince donors of the importance of randomizing aid allocation to determine the effects of aid via experimentation. Unfortunately, this is unfeasible, leaving us with observational data on which to rely for causal inferences. One approach to identifying the causal effect of aid on democracy related outcomes is to undertake detailed case studies to understand what exactly aid accomplished in terms of political 13

14 outcomes within a specific recipient state. This approach, while useful, has several potential drawbacks. Most importantly, such work is very difficult to generalize it is excellent for producing hypotheses about how aid might influence democracy, but it is inadequate for testing whether these possible links are operative outside of the particular case in question. Notably, many of the hypotheses we test in this paper were developed from observation of particular cases. Also, case studies tend to do very well at identifying the direct effects of aid how many pamphlets were produced, whether election monitoring occurred as planned, and whether aid reached women s organizations and human rights NGOs. However, qualitative case studies may be less useful for identifying the overall effect of aid on the political situation of a country pamphlets, election monitoring, and vibrant NGOs may or may not translate into increased democracy and higher quality governance. At the other extreme, researchers have attempted to identify the effects of aid on democracy using a variety of econometric approaches to correct for reverse causality. Our study is closer to this body of literature, but we note several significant problems with previous work in this tradition and attempt to solve them below. The most common solution to the reciprocal causation between aid and democracy is to find a suitable instrument a variable that is strongly correlated with aid flows but otherwise uncorrelated with democracy and use this to perform two-stage least-squares regression (2SLS). This method relies crucially on the exogeneity of the instrument, which is impossible to test empirically and thus must be argued on theoretical grounds. However, as we discuss in greater detail below, we find the arguments supporting the use of existing instruments to be questionable, and as we divide aid at sector level it becomes even more difficult to identify potential instruments. Although we attempt instrumentation, we retain a much more critical view of our findings than many previous studies. 14

15 Additionally, the previous practice of most econometric studies has been to simply enter some measure of democracy on the left hand side and some measure of aid on the right, glossing over the details that qualitative studies suggest underlie any possible links between aid and democracy. While the methodology we use here retains most of the clumsiness of a large-n study, we feel that acknowledging and testing the sector-level links between aid and democracy moves us in the direction of appreciating the potentially nuanced links between democracy and aid. Dependent Variables: Democracy and Democratization We measure the political impact of aid on levels of democracy and on democratization. While these two concepts are closely related, we believe that there may also be significant differences between them. We use the Polity IV measure of democracy to measure the democracy level and democratization of recipient countries (Marshall and Jaggers 2002). We note that this variable captures primarily institutional features of democracy, in contrast with measures such as Freedom House scores which place a stronger emphasis on democratic rights. To measure levels of democracy, we simply use the polity scores. These have received some criticism and some claim that it is possible only to classify polities as democratic or undemocratic, so we also use Pzerworski s (2000) binary measure of regime type. It is also not clear that an aggregate measure of democracy is completely adequate, especially because the causal mechanisms we test predict specific aspects of democracy which should be affected. Democracy is multi-faceted and it may be that aid liberalizes some aspects of a polity without affecting others. In future work we hope to measure the impact of aid on specific aspects 15

16 of democracy the competitiveness of elections, the probability of executive turn-over, and respect for various aspects of political and civil rights but that is beyond the scope of this paper. To measure democratization, we estimate the probability that a regime will experience at least a 3 point improvement in polity scores in 3 years. This measure captures regime liberalization rather than transitions to full democracy which is considered to be a score of 7 or higher on the Polity scale. Additionally, we are interested in the possibility that aid affects the probability of a democracy reverting to autocracy (Gazibo 2005, 69). We therefore estimate the impact of aid on the probability that a polity will experience autocratization a 3 point decrease in Polity scores over a period of 3 years. Key Independent Variables: The PLAID database We expand on previous studies by including the effective universe of foreign aid; we include assistance from the 21 OECD bilateral donors, the World Bank, the all of the major regional development banks, UN organizations such as the UNDP, and non-regional development banks such as the Nordic Investment Bank and the Islamic Development Bank. This data is drawn from the Project Level Aid (PLAID) database, which combines lending from all of these sources between 1970 and 2005 and codes each multilateral project with a CRS purpose code, comparable to those used by the OECD to determine project sector allocation. Because these data are self-reported, some of it is imperfect, especially in comparison with the USAID data used by Finkel et al (2007). They note that their data is the more extensive and finely grained than that used by any previous study, and they suggest that previous work has been hampered by very coarse estimations of foreign assistance world-wide (18). This may be the case, and we support their recommendation 16

17 that other international donors should undertake studies isolating democracy-building assistance from other types of assistance in order to evaluate the real impact of the aid provided (87). However, such studies are likely to be slow in coming and require intense cooperation with each of the bilateral donors. Thus, we take what we believe to be the second best option and use the data reported by the OECD as an estimate of the aid from each bilateral donor by sector. This solution is admittedly imperfect. Finkel et al note that their enhanced USAID data is only correlated with the OECD data for U.S. democracy aid at a.62 level (53-54). 1 However, we believe that even these flawed measures can tell us something about the relationship between aid and democracy and as better data become available, we expect our result to be either corroborated or refuted. However, the multilateral lending we use has been consistently coded from the annual reports of the various development banks and we can trust that these data are more consistently and accurately assigned to the correct purpose code. Aid Data: We divide our aid data into the sectors/categories discussed below. More generally however, we follow the OECD definition of aid, which is that flows are only counted if they are given expressly for development purposes (as opposed to military aid) and are given as grants or lent highly concessional, generally meaning that they have at least a 25 percent grand equivalent (OECD). We measure sector aid using the natural log of per capita amount received by sector in US 2000 dollars. Because aid may need several years to take effect, particularly in sectors such as education, or aid for economic growth, we use a moving averages of sector aid for the previous five years. 1 We found many apparently miscoded democracy projects which had received purpose codes for other sectors. We identify these via word searches. Such mistakes may account for the low correlation between the two measures that Finkel et al compared. 17

18 Democracy Aid We separate democracy aid from other types of aid activities using the purpose codes that donors themselves assign to indicate projects which target government administration, aid intended to improve judicial and legislative capacity, election monitoring, and aid for civil society. Our data Aid Through NGOs: In order to determine whether aid through NGOs is effective for fostering democracy by strengthening civil society and grassroots organizations, we use information provided by foreign aid donors on the channel by which their aid was delivered. Not all donors report this information evenly and donors generally report only the name of the organization in charge of delivering the aid, so constructing a measure of aid through NGOs requires a serious effort and a non-trivial amount of guesswork. From the name for the channel of delivery reported by the donor, we coded these projects by whether the aid was channeled directly through the government, through an NGO, etc. We coded aid that flowed through NGOs (local and international) and multilateral organizations as non-governmental aid. Coding details are available on request. Unfortunately, the data are very incomplete; about half the 80,000 projects are missing information on whether they were delivered to the government or through non-government channels. The data are clearly not missing completely at random; certain donors tend to have better or worse reporting standards. However, there is no obvious evidence that the likelihood of observing the channel through which aid was delivered to a recipient is correlated with a recipient s level of democracy. This allows us to assume that the missingness in the data is ignorable and we can use the existing data as a proxy for the missing aid flows. Thus, while our data do not provide accurate point estimates of the absolute level of non-governmental aid flowing into each country, we argue that our data is fairly accurate with respect to relative levels of non-governmental aid, both 18

19 across countries and across different years for the same recipient. Moreover, the breakdown by sector conforms with conventional wisdom about how aid of various kinds is delivered; less than 5% of infrastructure aid is channeled through non-governmental sources in our dataset, but over 30% of human rights aid flows through NGOs. Aid for Education: We measure aid for education using the education purpose codes assigned to education of all levels primary, secondary, and tertiary. This also category also includes aid for vocational schools. Growth Aid: Following Clemens et al (2004) we argue that growth aid is aid intended to cause growth in the short term primarily aid for infrastructure and public capital investments. While we do not have space here to list the detailed breakdown of our categorization, we note that we exactly follow the classification used by Clemens et al, who find strong evidence that aid to the growth sectors they identify is strongly correlated with growth in the recipient country. Fungible Aid/Aid that flows through the Government: While there are many ways of conceptualizing how fungible aid is and this is contested territory in the literature we adopt a very simple proxy for fungible aid: the amount of aid given to each recipient directly in cash. This money, which flows directly to the government, is likely to be highly fungible because it can easily be diverted to any purpose the recipient government wishes. We doubt that donors have significant control over aid given as cash once it leaves their hands. 19

20 Control Variables: Other Causes of Democratization We control for a variety of relevant factors noted in the literature on democratization. Noting the role of domestic economic and demographic factors, we include population, real GDP per capita, and GDP growth. In alternative specifications, we account for severe economic downturns, the land area of each recipient, and the distance from the normalized latitude of each recipient (La Porta et al. 1999). Education, religious tradition and legal heritage may be important for democratization so we include the adult literacy rate, the percentage of Protestant, Catholic, and Muslim adherents in each recipient (separately), and a dummy variable for recipients with a legal system patterned after British law. Noting the possibility that natural resources affect democratization, we include a oil exports as a proportion of GDP (Energy Information Administration 2008). We measure ethnic fractionalization using ethno-linguistic fractionalization (Fearon and Laitin 2003). International factors may also affect democratization. We account for the openness of a country to international economic flows by measuring the total sum of trade (imports and exports) for each recipient in a given year, and consider terms of trade shocks in alternative specifications. We include a dummy variable equal to 1 after 1991 to account for period effects of the Cold War. To account for democracy diffusion, we include the average democracy score (measured using Polity IV) for the states contiguous to each recipient, with contiguity defined and coded by the Correlates of War project (Stinnett et al. 2002). We also include (but do not report) regional fixed effects using the World Bank regional designations. Finally, we control for international non-governmental organizations using data from Landman (2005). 20

21 Methodology To identify the effects of sector-level aid on democracy and democratization, we use twostage least-squares regression (2SLS) to account for the potential reciprocal causality between aid and democracy. Of critical importance to this procedure are valid instruments variables which are correlated with the endogenous regressor and not causally related to the independent variable which we discuss below. However, we first note a few other minor issues. We use time-series cross-sectional data, with a dataset that covers 100 countries between 1975 and 2001, although not all countries are observed in all time periods. This requires that we use time-series corrections in our estimation; we estimate the models using an AR1 correction. We also lag all time-varying independent variables by one year. Additionally, we use several binary dependent variables one of our primary measures of regime type is a binary indicator for democracies, and we measure democratization as a binary variable indicating whether the polity score of a country increased by 3 points over 3 years. For these models we use a combination of 2SLS with time-series logistic regression. Analytically calculating the error on the point estimates of this non-standard model is complicated so we bootstrap the standard errors and use the 2.5 th and 97.5 th percentiles as a non-parametric 95% confidence interval for our coefficients. 2 Instruments for Aid The main strategy for identifying the effects of aid on democracy in the face of possible reverse causation is to use instrumental variables, a path which we follow here. To date, a variety of instruments have been used in the extant literature: (1) Foreign Policy Priority, measured as the number of times that a secretary or assistant secretary of state was mentioned by the New York Times in relation to a particular country in any given year (Finkel et al. 2007), (2) strategic interest variables 2 Actually, we had some trouble with the computers, so we are unable to present proper standard errors in the models we present here. 21

22 such as colony dummies, dummies for Africa, the Franc zone, Egypt, and Central American countries (Djankov et al. 2006; Kalyvitis and Vlachaki 2007; Knack 2004), and (3) measures of recipient need such as initial income in the previous period (Djankov et al. 2006; Kalyvitis and Vlachaki 2008), initial life expectancy (Kalyvitis and Vlachaki 2008), initial infant mortality (Knack 2004), and initial population {Knack, 2004, Kalyvitis, 2008). We do not find the majority of these instruments to be convincing and we feel that even the best of them are not particularly helpful for our situation, in large part because we believe that aid flows for different sectors require different instruments. A good instrument must fulfill at least two requirements, being (1) strongly correlated with the endogenous independent variable, and (2) theoretically unrelated to the outcome variable except through its relation with the endogenous independent variable (which is empirically unverifiable and must be argued theoretically). We argue that the instruments listed above clearly fail to satisfy at least one of these conditions. A proxy for US foreign policy interest our favorite instrument among those mentioned above may work well for predicting growth aid or cash aid, we are less convinced that it will predict education aid or nongovernmental aid, thus failing the first requirement. The others are all highly suspect when we attempt to imagine that they are truly exogenous to democratization We therefore adopt what we argue are a better set of instruments, especially for the sectorlevel nature of the aid variables in our analysis. Aid decisions in developed Western democracies are made by governments and bureaucrats that make allocation decisions. We expect, and find evidence, that aid flows are regionally clustered more than they would be if we aid were allocated purely on the basis of individual country characteristics. Indeed, this is the type of effect that Kalyvitas and Vlachaki (2007) attempt to capture by using regional dummies as instruments, but unfortunately for the econometrics, democracy and a variety of other factors are also regionally clustered so dummies might be picking up any number of things which could be correlated with 22

23 democracy and thus make regions inadvisable as instruments. We circumvent this problem by using a measure of regional aid flows excluding the recipient in question, which are highly correlated with the aid flows of each recipient yet theoretically uncorrelated with democracy or democratization in the recipient country. To instrument for different sectors of aid, we measure the regional flows of each sector; we use regional education aid to instrument for education aid in a single recipient, etc. To calculate regional aid flows to recipient A, we first use the COW contiguity data to identify the states that are contiguous with recipient A and then measure the average of aid receipts among the contiguous states by sector and year. For example, in order to instrument for education aid to Ethiopia, we measure the yearly average of education aid to Kenya, Somalia, Sudan, Eritrea, and Djibouti. As with sector aid, we use 5 year moving averages. These instruments are all strongly correlated with the endogenous independent variable and we argue that they are also exogenous to democracy or democratization in the country in question. Once we have accounted for sector-level aid in recipient A, we find it convincing that sector-level aid in contiguous countries will have no effect on democracy level of recipient A. Estimation and Results: We first estimate naïve models of the impact of sector-level aid on recipient democracy levels in which we assume that sector aid flows are not endogenously related to democracy. Taking care to be clear that we are not interpreting these naïve coefficients as causal effects, it is nevertheless interesting to note that without instrumentation (Model 1.1) growth aid, aid-in-cash, and aid through NGOs are correlated with higher democracy scores while aid for education and democracy have no clear relationship. 23

24 Table 1: The Effects of Sector Aid on Democracy (Polity IV scores) 1.1 OLS on Polity scores 1.2 OLS on Democracy Dummy 1.3 2SLS on Polity scores 1.4 2SLS on Democracy Dummy Education Aid * (0.0735) (0.108) (0.667) (0.979) Democracy Aid * (0.0785) (0.0994) (0.544) (0.642) Growth Aid 0.151** 0.202** (0.0696) (0.0950) (0.679) (0.907) Cash Aid 0.201* 0.281* (0.0453) (0.0658) (0.331) (0.406) Non-gov. Aid 0.311* 0.291* (0.0659) (0.0860) (0.362) (0.502) ln(population) * (0.325) (0.407) (0.765) (0.565) ln(gdp per capita) (0.414) (0.623) (1.192) (1.449) GDP growth * * * (0.0110) (0.0167) (0.0160) (0.0185) Adult Literacy Rate (0.0211) (0.0236) (0.0577) (0.0453) Ethnic Frac (1.776) (1.587) (3.949) (1.840) Oil Production/GDP * (33347) (58923) (53945) (61922) Percent Catholic (2.227) (2.251) (5.387) (2.906) Percent Protestant (6.193) (5.589) (15.35) (8.239) Percent Muslim (1.943) (1.498) (4.635) (1.992) Legal Origin: UK (1.235) (1.003) (2.894) (1.299) ln(total Trade) (0.155) (0.252) (0.314) (0.385) Regional Democracy 0.491* 0.269* 0.363* (0.0445) (0.0561) (0.0900) (0.0956) INGOs per capita 3.418* 9.931* * (1.615) (2.914) (2.391) (3.273) Constant * * (6.353) (8.357) (20.99) (21.33) Observations Number of Countries Instrumental Variables? No No Yes: For all aid variables Yes: For all aid variables Standard errors in parentheses. * p < Regional fixed effects included but not reported. The standard errors for model 1.4 are too small we intended to bootstrap them but ran into computational problems that were not yet solved when this paper hit the presses. 24

25 Once we instrument for each sector of aid, we obtain drastically different estimates of the impact of sector aid on recipient democracy. First, we find that education aid now has a positive and statistically significant effect on democracy; a ten percent increase in education aid per capita is associated with roughly a 0.22 increase in polity score. Although this may seem substantively small, note that a ten percent increase is relatively modest doubling the amount of education aid received over the past 5 years would increase a recipient s democracy level by 2 points on the Polity IV scale, an increase that could be considered serious democratization. Strikingly, we find that democracy aid has a significant negative impact on democracy levels once we use 2SLS, a sharp contrast to the finding of Finkel et al (2007) regarding democracy aid from USAID. On average, we find that a ten percent increase in democracy aid leads to a decrease of 0.15 in a recipient s democracy score. At present, we have few answers as to why this might be the case it appears that via some mechanism, democracy aid actually harms more than it helps. We do not find that growth aid, aid disbursed as cash, or aid through non-governmental channels has any significant effect on recipient democracy. Although the coefficients are all positive, the error around each of these point estimates is too large to confidently conclude that these types of aid have positive effects on democracy levels. When we re-estimate logistic regression models predicting whether a recipient will have a democratic regime or not as coded by Pzerworski (2000), we find that aid growth aid, cash aid, and NGO aid are again positively correlated with democracy in the model without instrumentation. However, once we use instruments and perform two-stage least-squares, none of the aid variables are statistically significant. 3 This suggests that the promising findings of model 1.2 are dependent to some extent on the measure of democracy we use. 3 We were unable to bootstrap the errors as planned because of a computation issue late in the game. The standard errors reported in Table 1, model 1.4 are too small, but the insignificance of the aid variables would remain insignificant even if the standard errors were corrected for the added variance in the first stage of the 2SLS process. 25

26 The main finding from Table 2 is that while aid for education and democracy seem to lead to higher levels of democracy, they don t appear to lead to large spurts of political liberalization. However, note that Cash aid is positively correlated with the probability of democratization and negatively impacts the probability of autocratization, a finding directly contrary to what Djankov et al (2006) predict about aid undermining democratic politics. We do find that growth aid increases the probability of autocratization, although this relationship is only barely statistically significant. 4 Table 2: The Effects of Sector Aid on Democratization and Autocratization 2.1. Democratization (3 point increase in Polity score) 2.2. Autocratization (3 point decrease in Polity score) Coefficients Standard Error Coefficient Standard Error Education Aid Democracy Aid Growth Aid * Cash Aid 0.692** ** Non-gov. Aid ln(population) * * ln(gdp per capita) GDP growth Adult Literacy Rate * Ethnic Frac Oil Production/GDP * ** Percent Catholic 3.138** Percent Protestant 8.464* ** Percent Muslim Legal Origin: UK * ln(total Trade) 0.383* *** Regional Democracy INGOs per capita ** * Constant * Observations Number of Countries Instrumental Variables? Yes: all aid variables Yes: all aid variables Note that the standard errors reported here are somewhat smaller than they should be. We intended to bootstrap them. * p < 0.1, ** p < 0.05, *** p < Keep in mind that the standard errors reported in Table 2 are not quite as large as they should be the bootstrap we hoped to use to account for the added variance of the first estimation stage ran into problems. 26