On the effectiveness of democracy aid in post-civil war recipient countries

ABSTRACT Democracy aid has been a major foreign policy instrument over the past 30 years. However, it is not clear whether such aid has any meaningful effect on democratic development in recipient countries, as previous studies have yielded somewhat contradictory results. We contribute to the burgeoning literature on the effectiveness of democracy aid by arguing that recipient countries’ history of conflict constitutes an important moderator in the relationship between democracy aid and democratic development. Specifically, we develop a theoretical framework which examines the effects of two different types of democracy aid – top-down and bottom-up assistance – on democratic development in post-conflict recipient countries. Analysing data on 147 recipients over a period of 19 years (2002–2020), we find that while neither type has a significant general effect on democracy levels in recipient countries, bottom-up democracy assistance does have a positive and significant effect in post-conflict contexts, which suggests that post-conflict periods pose an opportunity for domestic pro-democracy actors and their international supporters to nudge their countries towards democratic development. These findings have implications for the literatures on democracy aid and conflict as well as policymakers interested in fostering democratic development abroad.


Introduction
For more than a decade, democratic quality around the world has been in decline, with the current average level of democracy being the lowest since 1989. 1 To counter this trend and advance global democracy levels, some are looking to democracy aid as a potential remedy. 2 Various donors including nation states, international organizations, and non-governmental organizations (NGOs) provide financial and technical support with the explicit aim of democratizing the recipient country. 3Despite its longstanding application as a foreign policy instrument, democracy aid's effectiveness in fostering recipients' democratic development is still contested, since previous research has produced ambiguous results.While some studies find no impact or even adverse consequences of democracy aid on democracy, 4 others report significant positive, albeit minor, effects. 5hese diverging findings can partly be explained by the fact that the effectiveness of democracy aid appears to vary across contexts.Studies have shown that certain donor characteristics, conditions in the recipient country, and even features of the aid itself can make democracy aid more or less effective at improving democracy levels.Examples of such factors include donor credibility, 6 regime type of the recipient country, 7 and the consistency of the aid disbursements. 8Another important yet somewhat underexplored condition that may influence the effectiveness of democracy aid is recipient countries' prior conflict experience.There has been an ongoing increase in the global level of conflict as well as the number of civil wars that make up the vast majority of all state-based conflicts since 1945. 9onsequently, foreign aid in general and democracy aid specifically is increasingly received by conflict-affected states.In 2020, for example, the members of the Development Assistance Committee (DAC) jointly disbursed 63% of the total Official Development Assistance (ODA), which includes democracy aid, to fragile states, a majority of which were affected by some type of state-based conflict. 10In these situations, democracy aid is usually not disbursed during the civil war but after the civil war has ended to utilize the window of opportunity for democratization post-conflict. 11cholars generally distinguish between two different types of democracy assistance, top-down and bottom-up aid, which are intended to foster recipient countries' democratic development in distinct ways. 12Top-down democracy assistance is given directly to foreign governments to promote order, good governance, and institutional capacity, which are themselves aspects of democracy and may in the long term induce further democratic development. 13Conversely, bottom-up democracy aid is disbursed to perceived pro-democracy actors in the recipient country, such as political parties, independent media, and civil society organizations, which use this assistance to hold the government accountable, protect individual rights, and otherwise support democratic development. 14Given these distinct mechanisms, we argue that the two types of democracy aid are moderated differently by post-conflict contexts.Due to the risk of renewed conflict and the resulting security and stability-democracy dilemmas for incumbents, we expect top-down democracy aid to be less effective post-conflict than in post-peace recipients.Conversely, we expect bottom-up democracy aid to be positively moderated by the post-conflict period as leaders are generally weakened after conflict, providing a window of opportunity for a more effective empowerment of pro-democracy actors and their ability to pressure their leaders.Examining data on 147 recipient countries between 2002 and 2020, we find support for the latter expectation, as our analyses suggest a positive and significant interaction between bottom-up aid and post-conflict settings.
Our study makes several contributions.Theoretically, we firstly contribute to the ongoing debate about consequential moderating factors for explaining the variation in democracy aid's effectiveness by introducing the post-civil war phase as one such contextual factor.Despite some contributions exploring the mechanisms of democracy aid in post-conflict recipient countries, research on the effectiveness of democracy aid thus far does not systematically include the post-civil war phase of a recipient as a contextual and potentially moderating factor. 15Our findings show that it is important to differentiate between recipients in a post-civil war context and those that are not, since the two contexts represent discrete and contrasting starting points for democracy aid.Second, we offer a possible explanation for the ambiguous and contradictory findings of previous research on the general effectiveness of democracy aid: by not accounting for the potential conflict experiences of recipients, scholars missed an important piece of the puzzle as to why democracy aid has positively affected democracy levels in some cases but not in others.Third, by differentiating between top-down and bottomup aid, we systematize their distinct mechanisms both post-peace and postconflict.Fourth, by combining two previously separate literature strandsdemocracy aid effectiveness and international conflict studieswe develop a novel theoretical framework that provides insights into an important but largely overlooked consequence of conflict.
Empirically, we contribute to the quantitative literature on democracy aid's effectiveness which has mostly focused on single donors, recipients, or regions. 16nstead, we go beyond a specific case and analyse the overall effect of OECD/ DAC democracy aid on all potential recipient countries while explicitly differentiating between top-down and bottom-up democracy aid.We thereby offer generalizable results and practical implications for the overall effectiveness of the two types of democracy aid as well as the impact of the post-civil war phase on their effectiveness.
The remainder of this study proceeds as follows: In the next section, we define the key concepts of our study, review previous literature on both the effectiveness of democracy aid and civil war legacies, and develop our theoretical framework from which we derive two hypotheses.Following this, we lay out our research design.Subsequently, we present the results of our main analyses as well as a series of robustness and endogeneity checks.In a concluding section, we discuss the theoretical and practical implications of our findings, address the limitations of our study, and present avenues for future research on the subject.

Literature review and hypotheses
Democracy aid is defined as the provision of financial and/or technical assistance from a donor to a recipient country with the explicit goal of strengthening democracy levels of the recipient.As a specific type of democracy promotion, democracy aid (which is also referred to as democracy assistance) exclusively describes financial and technical support of democratic development in recipient countries while the broader term democracy promotion captures a wider range of actions in support of democracy abroad, including rhetorical actions, economic coercion, and military activities. 17It should also be pointed out that democracy aid is distinct from general development aid due to their different goals, as the former is explicitly and directly aimed at strengthening democratic actors, institutions, and processes while the latter is directed at socioeconomic outcomes such as poverty alleviation as well as increased employment and educational opportunities. 18efore democracy aid can be disbursed, however, the recipient country must agree to this type of aid. 19Even though democracy aid in theory threatens the survival of non-democratic incumbents they may allow the aid's disbursement for two reasons.First, survival-driven incumbents employ cost-benefit-calculations to decide whether to allow the disbursement of democracy aid.Therefore, non-democratic leaders tend to allow aid only if they perceive it as not threatening to their survival.This is often the case since democracy aid programmes are not always all-encompassing and frequently only address minimal reform concessions such as holding elections or allowing opposition parties. 20Additionally, some forms of democracy aid may actually empower the incumbent, as will be illustrated below, and the leader has the opportunity to shape and prevent potentially harmful implementations since democracy aid is only disbursed in accordance. 21econd, incumbents may view democracy aid as an opportunity to credibly signal their willingness to democratize.Contemporary world politics are constituted in such a way that failure to show any progress towards democratization may result in negative sanctions, for example reductions in future foreign aid or the exclusion from international forums.Therefore, states have an incentive to be perceived as a democratizing state to reap the associated rewards in the international community and seek opportunities to credibly signal their commitment to democratization. 22In line with Hyde's (2011) argument on inviting international election observers, 23 we argue that inviting democracy aid acts as a credible signal of the recipient's willingness to democratize towards the international community.In turn, rejecting democracy aid to be disbursed signals an unwillingness of the incumbent to democratize.As a result, autocratic rulers tend to accept a certain amount of democracy assistanceand with it the (small) risk of democratizationrather than outright refuse such aid and risk drawing the ire of the international community.
Once democracy aid is allowed into the recipient country, there are two distinct was in which such aid may be disbursed, each with its own mechanism through which it may improve democracy levels.First, democracy aid may be given directly to government actors and state institutions of the recipient country via the top-down approach. 24he goal of this type of democracy assistance is to increase the recipient country's overall state capacity and to alter cost-benefit calculations of political elites to carry out democratic reforms. 25Such institution-building efforts include the strengthening of the administrative apparatus, training programmes for civil servants, and the development of monitoring and evaluation capabilities.Top-down democracy aid also includes support for legal and judicial development, for example by establishing lawyers' associations, providing professional legal education, and enhancing oversight for law enforcement. 26Overall, then, top-down democracy assistance aims at strengthening various government institutions, thereby reducing corruption, increasing efficiency, and decentralizing power through establishing checks and balances.This not only provides recipient states with more effective and capable institutions but also with legitimacy and credibility within their society, thus creating an incentive to implement further democratic reforms. 27econd, through the bottom-up approach, democracy aid empowers domestic prodemocracy forces by providing financial and/or technical assistance to non-government actors. 28This type of aid is implemented through donor agencies as well as international and local NGOs, bypassing the incumbent government.In the electoral dimension, donors may for example provide assistance for civic education programmes, fund the creation of opposition political parties, or provide financial resources to civil society and independent media outlets for monitoring the democratic code of conduct during elections. 29To assist human rights, democracy aid may for example fund human rights monitoring by local civil society organizations or education programmes. 30This type of aid strengthens the potential opposition and civil society, and puts them in a better position to hold undemocratic leaders accountable.As a result, the balance of power between regime and opposition within the recipient country changes, effectively weakening the incumbent leader(s). 31Therefore, it becomes too costly for authoritarian leaders to maintain an autocratic system and eventually, democratization reforms are implemented by the incumbent. 32hile these mechanisms seem intuitively convincing, the empirical evidence regarding the relationship between democracy aid and democratic development in recipient countries is still somewhat unclear, as previous research has produced ambiguous results. 33While some studies find no impact or even adverse consequences of democracy aid on democratic development, 34 others report significant positive, albeit minor, effects. 35To explain this apparent variation in democracy aid's effectiveness, researchers have turned their attention to various contextual factors that potentially moderate the relationship between democracy assistance and recipients' democracy levels.The moderators most commonly referred to can be grouped into donor-specific and recipient-specific factors. 36On the donor side, aspects such as fragmentation and coordination, 37 strategic and geopolitical considerations, 38 and the credibility of political conditionality have been shown to moderate the impact of democracy aid. 39On the recipient side, scholars have pointed to dependency on financial aid, 40 regime type, 41 and the size of the military as important contextual factors. 42In addition, it has been argued that features of the aid itself can make democracy aid more or less effective at improving democracy levels.For example, some studies find that bottom-up democracy aid is generally more effective than topdown assistance, 43 while others indicate the opposite. 44nother moderating factor that has recently started to receive attention from scholars interested in the effectiveness of democracy aid is recipient countries' history with conflict, particularly the previous occurrence of intrastate war. 45Post-civil war countries are defined as "societies in which intrastate conflict has ended and the international community has recognized the government as legitimate." 46An intrastate war, in turn, is defined as an armed conflict between a government and a non-governmental party. 47This interest in the post-civil war context originated in the ongoing debate on the prospects of democratization after conflict.Studies have repeatedly shown that it is more difficult to establish a process of democratization in a postconflict country, since the post-conflict context imposes adverse conditions on the usual determinants of democratization. 48For example, economic activity is stifled, governmental institutions have been compromised, 49 and there exists political mistrust between former belligerents 50 as well as polarization within society. 51Scholars have identified the risk of renewed violence and the thus arising dilemma between stability and democratization as a major factor impeding post-conflict democratization. 52To avert renewed conflict, local and international actors engage in peacebuilding efforts and seek short-term solutions to maintain peace and stability. 53Democratization, however, may increase the risk of violence in the short-term, for example through the introduction of party competition in post-conflict elections. 54Moreover, incomplete democratization processes have been linked to the onset of civil war. 55Overall, actors face the problem of providing credible commitments to a democratization process in a situation of uncertainty and risk of renewed conflict. 56xisting scholarship on the effectiveness of democracy aid to post-conflict recipient countries is mostly based on case studies of specific donors, recipients, or types of aid, often focusing only on electoral assistance. 57These case studies come to mixed conclusions: While some studies find democracy aid to make at least minor contributions to democratization in post-conflict recipients, others are more critical and see no merit in disbursing democracy aid under such conditions. 58Comparative studies that systematically analyse the post-conflict context as a potential moderating factor for the effectiveness of democracy aid are currently missing. 59Therefore, it is still unclear whether there exists a significant and generalizable difference between post-conflict and other recipients for the effectiveness of democracy aid.This is especially striking considering the advanced qualitative and quantitative scholarship on the difference in effectiveness of developmental aid in post-conflict contexts, reporting ambiguous results depending on the particular type of foreign aid. 60Additionally, authors have investigated the impact of state fragility on foreign aid effectiveness, defining fragility as deficits in at least one of the state's key responsibilities. 61While a post-conflict state is not necessarily a fragile state, their finding that the effectiveness of aid likely depends on the specific context of fragility 62 suggests a need for further research into the context of the recipient state if it experiences challenges to its capacity, as is the case post-conflict.For the specific realm of democracy aid, Zulueta-Fülscher has called for further research into possible determinants of democracy aid's effectiveness in different types of fragile environments. 63Overall, then, there is a clear need for a generalizable analysis of the effectiveness of democracy aid that is sensitive to the potential conflict history of recipient countries.
We provide such an analysis by arguing that post-conflict contexts moderate the effects of top-down and bottom-up democracy assistance differently due to their distinct mechanisms through which they foster democratic development.We acknowledge previous research which finds that variation in post-conflict contexts matters for the prospects of democratic development, for example whether the conflict ended in a decisive military victory or in a negotiated settlement. 64However, we do not include this variation in our theoretical and analytical framework, and instead aim to establish an overall account of post-conflict top-down and bottom-up democracy aid.Given the current lack of comparative studies that systematically analyse the post-conflict context as a potential moderating factor for the effectiveness of democracy aid, we believe such an account is warranted and can provide a useful starting point for future research investigating the potential role of different post-conflict contexts in the relationship between democracy assistance and democratic development.
Specifically, we posit that top-down democracy aid is less effective in post-conflict states relative to recipients that have not experienced a recent civil war, while we expect bottom-up democracy assistance to be more impactful under such circumstances.Seeing as top-down democracy aid is directly dispersed to the recipient's government, its effectiveness in raising democracy levels hinges upon incentives for the incumbent regime to implement reforms even though these reforms may diminish its power.We argue that the prevailing threat of renewed violence in a post-conflict environment changes the baseline expectations (i.e. the expected future position) of survivaldriven leaders, making post-conflict top-down democracy aid less effective at incentivising the incumbent to implement democratic reforms. 65In a post-conflict society, the risk of a renewed conflict and outbreak of violence threatens the survival of the incumbent, since renewed civil war could result in a change of leadership.Additionally, polarization, grievances, and mistrust within society and between previous warring parties are usually prominent. 66While autocratic leaders after a peaceful episode generally expect to remain in power in the future, post-conflict incumbents expect their future position to be conflictual and more uncertain given the threat of renewed violence.As a result, they find themselves in a security dilemma with regards to the implementation of democratization reforms.
The concept of the security-dilemma originally stems from realist international relations theory. 67In its essence, the security dilemma states that all countries want to be secure from possible attacks by others.To achieve this state of security, they aim to have more power than others to defend themselves.However, acquiring power unsettles other states with the same fear, starting an endless power competition that leaves all actors less secure about their position. 68While this concept has been used to explain multiple phases of war and peace, 69 we posit that its logic can also be applied to the process of democratization.To effectively implement democratization programmes funded by top-down democracy aid, political actors in power must give up some of their influence, for example by creating a strong and independent bureaucracy, by committing to the rule of law, or by decreasing their hold over the judiciary.However, due to the risk of renewed conflict and the resulting baseline expectations, there are strong incentives to retain political power just in case the conflict renews as to either stop the conflict decisively or shape it according to their own preferences. 70his is where the security dilemma comes in: To democratize and receive the benefits as incentivized by the democracy aid programme, political actors would have to give up a substantial portion of their power.However, if they give up too much, they are more vulnerable in case of a renewed conflict, especially considering the possible lack of credible commitments to a peaceful and democratic system by former belligerents within the society. 71As a result, post-conflict incumbents are less inclined than post-peace incumbents to fully implement democratization programmes funded by top-down democracy assistance, since these reforms might weaken their future position in relation to the general population and actors of the previous conflict, ultimately threatening their survival.Leaders may still agree to receive post-conflict top-down democracy aid since such aid can potentially be diverted in an attempt to increase their state capacity which is especially crucial after conflict where countries usually suffer from a lack of resources. 72In other words, due to the security dilemma, incumbents will refrain from effectively implementing the funded democratization programmes.Instead, leaders will only partially or superficially implement reforms to satisfy donors while maintaining or even strengthening their hold on power.Consequently, if reforms are pursued by the incumbent, they are less meaningful and top-down democracy aid is therefore less effective compared to post-peace situations where the recipient government on average is more secure in their future position and therefore more willing to democratize.
In addition to the security dilemma faced by post-conflict governments, a major concern for donors in post-conflict environments is creating stability.While the process of democratization promises to bring that stability in the long run, previous research has shown that temporarily, it may lead to instability and volatility, for example due to competitive elections. 73International donors targeting post-conflict countries therefore face the stability-democracy dilemma of deciding between immediate but potentially non-democratic stability and the long-term but temporarily more volatile goal of democratization. 74Usually, donors prioritize stability over democratization, focusing on strengthening the capacity of the state to keep order. 75Consequently, we argue that post-conflict top-down democracy aid will on average include less demanding democratization reforms to begin with, focusing more on maintaining order instead of decentralizing power within the government and democratizing state structures.Therefore, post-conflict top-down democracy aid affects democratic development in recipient countries to a lesser extent than when such aid is disbursed in post-peace settings.
In sum, due to the security dilemma as well as the stability-democracy dilemma, we expect top-down democracy aid to have a diminished effect on democracy levels in post-conflict settings.Our first hypothesis therefore states: H 1 : Top-down democracy aid is less effective at achieving its intended goals in post-intrastate war countries compared to recipients that have not experienced a recent intrastate war.
In contrast to top-down aid, bottom-up democracy aid's success hinges upon the effective empowerment of domestic pro-democracy actors, which is intended to shift the balance of power in favour of local democratizers and enable them to pressure their leaders into enacting democratic reforms.If a country has not experienced a recent civil war, the state capacity of the autocratic incumbent is relatively strong, severely tilting the balance of power towards the government.Given this strong imbalance, international donors will find it difficult to empower local pro-democracy actors to a point where they can effectively constrain autocratic leaders and pressure them to implement democratic reforms.While the incumbent regime may allow bottom-up democracy aid due to the aforementioned incentives, it may still hinder its successful implementation, for example through arbitrary interference in the internal affairs of civil society organizations or swaying public opinion against them. 76fter a civil conflict, however, the incumbent regime's resources and state capacity are usually weakened, 77 thus providing a window of opportunity for bottom-up democratization.For one, governmental institutions, infrastructure and control over law and order are usually weak or dysfunctional after their erosion or destruction that allowed the previous conflict to come about. 78Second, the limited monopoly on the use of force compared to post-peace autocracies acts as a significant constraint for the post-conflict incumbents' state capacity.During a post-conflict phase and while post-conflict democracy aid is administered, the demobilization of all previous combatants and the security sector reform is usually ongoing to facilitate the transition from civil war to lasting peace. 79Of course, this initially strengthens the incumbent's state capacity since the monopoly on the use of force is transferred towards the state away from several competing factions.However, not all parties will (yet) be effectively disarmed during a post-conflict phase and there is a high chance that previous combatants still hold military capabilities.
The capacity of former combatants to mobilize supporters along the lines of the preceded conflict during a post-conflict phase puts a further strain on the post-conflict state capacity. 80While this mobilization capacity also holds true for the incumbent if they are an ex-combatant that gained power as a result of the civil war, rebel groups with a similar mobilization capacity still affect the relative balance of power.Thus, a post-conflict incumbent does not hold a guaranteed monopoly on the use of force and does not fully control any societal mobilization against them while this is generally the case under a post-peace autocratic leader that has been able to shut down any attempts to initiate conflict by rebel groups.Due to the decrease in state capacity post-conflict, the difference in power between government and internal pro-democracy actors is smaller than if the country had not experienced a civil war before.Autocratic leaders find it more difficult to resist the demands of local prodemocracy actors, as the tactics used to supress them (such as the harassment by government officials or the establishment of parallel NGOs and media outlets that are loyal to the regime) require substantial resources. 81hus, even small amounts of bottom-up democracy aid may successfully change the balance of power in favour of civil society and the political opposition, allowing these actors to effectively pressure the incumbent regime into implementing democratic reforms.Our second hypothesis therefore states: H 2 : Bottom-up democracy aid is more effective at achieving its intended goals in post-intrastate war countries compared to recipients that have not experienced a recent intrastate war.

Research design
To test our hypotheses, we examine data on 147 countries potentially eligible for democracy aid 82 over a period of 19 years (2002-2020). 83Since we are interested in examining the effects of different kinds of democracy assistance, top-down and bottom-up aid, it would make little sense to measure our dependent variable, democratic development, via an aggregated democracy index such as Freedom House, Polity, or V-Dem's high-level democracy indices including its Electoral Democracy Index (EDI) and Liberal Democracy Index (LDI).Rather, it seems conceptually sensible to use two separate, disaggregated outcome measures which capture distinct aspects of democratic development that correspond closely with the intended effects of the different kinds of democracy aid. 84herefore, as top-down democracy assistance is intended to aid democratic development through the strengthening of various government institutions, we use the World Bank's Government Effectiveness Index (GEI), which captures "the quality of public service provision, the quality of the bureaucracy, the competence of civil servants, the independence of the civil service from political pressures, and the credibility of the government's commitment to policies." 85Further, as bottom-up democracy assistance aims to empower domestic actorspolitical parties, independent media, and civil society organizationswhich will in turn hold government actors accountable and advance democracy, we use V-Dem's Freedom of Expression and Alternative Sources of Information Index (FEASII) as a second dependent variable, which measures the extent to which governments "respect press and media freedom, the freedom of ordinary people to discuss political matters at home and in the public sphere, as well as the freedom of academic and cultural expression." 86For ease of interpretation, both variables have been transformed such that they range from 0 to 100, with higher values indicating more effective governance and greater government restraint, respectively.
To measure our independent variable, democracy assistance, we use annual total aid disbursements (not commitments) from the OECD/DAC Creditor Reporting System (CRS).The CRS provides data on all foreign aid activities under ODA from all DAC donors, reported in millions of constant 2020 USD.Top-down democracy aid is captured by combining OECD/DAC categories "15110: Public Sector Policy and Administrative Management" which includes all institution-building assistance given with the goal of strengthening "core public sector management systems and capacities," and "15130: Legal and Judicial Development" which, among other things, covers support to institutions, systems, and procedures of the justice sector; to the ministries of justice, the interior, and home affairs; and to law enforcement agencies, police, prisons, and their supervision. 87We measure bottom-up democracy aid by combining OECD/DAC categories "15150: Democratic Participation and Civil Society" which includes assistance given with the goal of strengthening "electoral management bodies and processes, legislatures and political parties, and independent media and the free flow of information," and "15160: Human Rights," which covers all financial and technical support given to "human rights defenders and human rights NGOs; human rights advocacy, activism, mobilisation; awareness raising and public human rights education." 88For both variables, we only use data that can be unequivocally attributed to a certain recipient country.That means we omit regional aid which refers to support aimed at more than one recipient country, since information is unavailable for how large the share of this regional aid is per recipient country.
Our theoretical framework posits that the effect of democracy aid on democracy levels is moderated by recipient countries' history of conflict, particularly the previous occurrence of intrastate war.We consider countries to be in a post-civil war situation if the termination of a country's intrastate war occurred between one and ten years ago.To account for this post-civil war context, we generate a dummy variable based on information obtained from the country-year transformed Uppsala Conflict Data Program (UCDP)/PRIO Armed Conflict Dataset.This dummy variable is coded 1 if a country-year observation lies within ten years after a civil war has ended, and 0 otherwise.If another intrastate war interrupts during the ten years after the first intrastate war has concluded, our post-war dummy is coded 0 during the second conflict and restarts after it has ended.The fact that the post-war dummy and the binary variable for intrastate conflict can never be coded 1 at the same time allows us to clearly distinguish the effects of post-war environments from those of actual intrastate conflicts.In addition, we control for the following conditions in recipient countries: GDP per capita, GDP per capita growth, trade openness, intrastate conflict, interstate war, regime duration, regime type, other types of aid, population size, and the presence of UN peacekeeping operations. 89n accordance with other recent studies on the drivers of democratic development, 90 we estimate OLS models with robust standard errors clustered by country.The baseline models include the main independent variable, the control variables, year dummies, region dummies, 91 and a one-year lag of the dependent variable.To test our hypotheses, we estimate interaction models which include the same variables as the baseline models plus the post-war dummy as well as its interaction with democracy assistance.The year dummies absorb any atypical year-to-year variation (i.e. they capture common shocks such as financial crises) while the region dummies control for regional developments and neighbourhood effects.The one-year lag of the dependent variable corrects for serial correlation and floor/ceiling effects.Note that inclusion of a lagged dependent variable means that our models are equivalent to estimating changes in the dependent variable.All time-varying independent variables are lagged by one year.

Results
Table 1 presents the descriptive statistics for the dependent, independent, and moderating variables as well as the controls.As can be seen, the share of missing values is below 15% for all variables and thus relatively low.Interestingly, the mean for government restraint (61.807) is substantially higher than that for government effectiveness (39.879), which indicates that the countries in our sample are more likely to show restraint in violating certain rights than to exhibit effective governance.Also of note is the fact that the mean for top-down democracy aid (25.148) is almost three times that for bottom-up assistance (9.185), meaning that the former is a much more prominent form of democracy assistance than the latter. 92Both measures of democracy aid exhibit a great deal of variation, which we address in the robustness section by performing outlier analysis.Roughly 16% of observations are classified by us as post-conflict periods.Another 16.9% and 1.4% of observations are coded as intrastate and interstate conflicts, respectively, making the former a much more common form of war than the latter.The average regime duration (i.e.regime age) in our sample is 21.3 years.The most common regime type in our sample is electoral autocracy with 42.42%, followed by 36.29%electoral democracies, 14.93% closed autocracies, and 6.36% democracies.In roughly 10% of observations, a UN peacekeeping mission is carried out.
Table 2 shows the results for our baseline models.As can be seen, top-down democracy aid does not have a significant effect on government effectiveness, echoing some recent studies claiming that this type of assistance is ineffective or even detrimental to democratic development. 93Most of the control variables are also insignificant, with a few noteworthy exceptions.GDP per capita and GDP per capita growth both increase government effectiveness, as does regime duration.Additionally, the coefficient for our regime type variable is positive and significant, which indicates that democratic governments are, on average, more effective than autocratic ones.Bottom-up democracy aid does not significantly impact government restraint in either direction, which is in line with findings by recent studies suggesting that this type of assistance has been largely ineffective post-2001. 94All control variables in this model are insignificant as well.Overall, our results do not confirm a difference in effectiveness between the two types of democracy aid, as some authors have claimed. 95able 3 illustrates the results for our interaction models.The interaction between the top-down democracy aid variable and the post-war dummy is statistically insignificant, meaning that top-down assistance is no more or less likely to further government effectiveness during post-conflict periods than it is during regular times, which leads us to reject H1.Interestingly, the interaction between bottom-up aid and the post-war dummy is positive and significant, which suggests that this type of democracy assistance is more effective in restraining governments during post-conflict periods than it is otherwise.To better visualize this interaction, Figure 1 illustrates the marginal effects for the model.Specifically, the x-axis shows bottom-up democracy aid in millions of constant 2020 US dollars while the y-axis should be interpreted as the predicted value of the dependent variable, government restraint.During regular times, the value for V-Dem's FEASII slightly decreases as bottom-up assistance increases, while for post-conflict periods there is a clear upwards slope.Even though the confidence intervals are quite large for high values of bottom-up democracy aid and therefore overlap, we consider this as evidence that this type of democracy assistance is more effective in post-conflict settings; therefore, H2 is supported by the evidence.It appears that times of change and upheaval offer a unique opening for domestic actors to move a country in a more democratic direction, and bottom-up assistance is a way for international actors to increase their capacity to do so.
One potential objection to our findings is that there may be selection bias at play, such that donors specifically target countries with differing past or expected democratic trajectories, 96 or that countries self-select into the pool of recipients at a given time. 97Therefore, we estimate a series of instrumental variable models that mitigate such concerns about endogeneity and provide us with considerable leverage for establishing causality.Specifically, we use the Arellano-Bond dynamic panel estimator, 98 which has the major advantage that it does not require the specification of external instruments because it utilizes lagged levels of the endogenous regressors as instruments instead. 99The results of these instrumental variable models, which are available in the appendix, are similar to those of our baseline models, which indicates that our findings are not driven by endogeneity but rather representative of a "true" relationship between democracy aid and democratic development.
In addition to estimating these instrumental variable models, we undertake five robustness checks to ensure that our findings are not sensitive to alternative model specifications.First, given the great deal of variation in our democracy aid variables, we employ the Hosmer and Lemeshow leverage statistic as well as Cook's D to identify potential outliers.Based on these statistics, two country-years (Iraq in 2005 and  2006) qualify as outliers and we consequently re-run our analyses without these particular observations.Second, we re-run all of our models using alternative dependent variables.For top-down democracy aid we use V-Dem's "Rule of Law Index," and for bottom-up aid we use V-Dem's "CSO Participatory Environment" variable.Third, we estimate our interaction models with a fixed-effects specification to control for potential confounders that affect both the likelihood of a civil war occurring and the effectiveness of (post-conflict) democracy aid.Fourth, given previous studies that find armed conflict to significantly affect a country up to ten years after it has ended regardless of the specific conflict trajectory, 100 we opted for a length of ten years for the postconflict period.To ensure that this choice does not drive our results, we re-run our interaction models with two alternative post-conflict lengths.Lastly, we estimated multiple interactions models to ensure that our findings are not limited to particular conflict outcomes.Our results hold across all alternative specifications, thus greatly increasing our confidence in their validity.Detailed results for all robustness checks are available in the appendix.

Concluding remarks
In this study, we have developed a theoretical framework arguing that recipient countries' post-intrastate war context moderates the effectiveness of top-down and bottom-up democracy aid in different ways.Due to the risk of renewed conflict and the resulting security and stability-democracy dilemmas, we expected top-down democracy aid to be less effective in post-conflict contexts.Conversely, we hypothesized bottom-up democracy aid to be more effective after civil wars, as incumbents are generally weakened after conflict, providing a window of opportunity for a more effective empowerment of pro-democracy actors and their ability to pressure their leaders.To test these arguments, we employed a series of panel data regressions, analysing the effect of the two types of democracy aid on the democratic development of potential recipient countries between 2002 and 2020.Doing so, we found that both types of democracy aid have no significant direct effect on democratic outcomes.Additionally, our findings show that top-down democracy aid remains insignificant when interacted with post-conflict contexts.Conversely, bottom-up democracy assistance has a significant and positive effect on recipients' democracy levels in postconflict settings.These results were robust to a series of alternative model specifications, particularly endogeneity checks, outlier analysis, alternative dependent variables and measures for our moderating variable, and a different estimator.
Theoretically, this study makes several contributions.First, we contribute to the growing literature on factors moderating the relationship between democracy aid and recipient countries' democratic development.Second, we offer a possible explanation for the ambiguous and contradictory findings of existing research regarding the general effectiveness of democracy aid, as previous studies have not sufficiently accounted for recipient countries' history of conflict.Third, by differentiating between top-down and bottom-up democracy aid, we systematize their distinct mechanisms both post-peace and post-conflict.Fourth, by combining two previously separate literature strandsdemocracy aid effectiveness and international conflict studieswe systematize the relationship between the two and introduce new concepts to the study of post-conflict democracy aid.Empirically, we contribute to the quantitative literature on democracy aid's effectiveness which has mostly focused on specific donors, recipients, or regions.Instead, we go beyond a specific case and analyse the overall effect of OECD/DAC democracy aid on all developing countries receiving this aid.We thereby offer generalizable results and practical implications for the overall effectiveness of democracy assistance as well as the impact of the post-civil war phase on democracy aid's effectiveness.
As for the practical implications, the empirical results of this study shed a negative light on democracy aid programmes which have long faced the critique of disbursing money without producing meaningful results.In line with other authors and popular concerns, this study shows that democracy aid generally is not effective in raising democracy levels in recipient countries for the post-2001 period.Consequently, donors and practitioners should continue to explore complementary strategies to improve their programmes' effectiveness as well as alternative democracy promotion strategies.Most importantly, we show that the conflict history of a recipient needs to be recognized as an important moderating factor for democracy aid's effectiveness.Our finding that bottom-up aid has a positive effect on the recipient's democracy levels post-conflict should restore some confidence in this type of aid.Currently, OECD donors focus their democracy aid disbursement primarily on top-down democracy aid.Following our results, donors should instead increase the disbursement of bottom-up democracy aid, especially to post-conflict recipients.
Our study is not without its limitations, chiefly among them the fact that our research design was unable to measure the proposed mechanismsthe security dilemma for incumbents in countries after conflict, the stability-democracy dilemma, and the window of opportunity for bottom-up pro-democratic empowermentdirectly.Future research can build on our framework and devise empirical strategies to measure and test these mechanisms.In addition, our study did not distinguish between different types of post-conflict contexts and instead opted for a general account of how post-conflict environments can moderate the effects of democracy aid.Future research can examine whether variation in post-conflict contexts, such as the type of conflict termination, matters for the effectiveness of democracy assistance.Findings from this research could then be used to develop adequate strategies to increase democracy aid's effectiveness in post-intrastate recipient countries.

Table 2 .
Democracy assistance and democratic development -Main effects.