Foreign Aid Composition and Sustaining Peace After War

Abstract

This article examines the relationship between aggregate foreign aid and conflict relapse or sustained peace in postconflict settings. We use Organisation for Economic Co-operation and Development (OECD) data for all country allocable aid to postconflict contexts in the 1990–2022 period, and break down aid into humanitarian, development and peacebuilding types. The findings confirm a statistical relationship between total aid levels and sustained peace. Results also highlight the composition of the aid portfolio as important for avoiding conflict relapse—the peacebuilding support share in particular. Our contribution is to take a necessary first step to analyse the collective behavior of donors in postconflict environments across different aid types. The findings lay the necessary groundwork for future research on the causal relationship between peacebuilding assistance and conflict. At a time where aid budgets are shrinking and the number of armed conflicts is high, traditional and new donors alike are forced to consider aid effectiveness and how to prioritise between different forms of support.

Introduction

In countries coming out of war, needs are overwhelming, and risks are high. Many postconflict countries relapse into organised violence (Collier et al., 2003; Walter, 2011). While national actors hold the keys to postconflict peace, efforts to sustain peace can be supported from the outside. Most research on the connection between foreign aid, peace and conflict tends to focus on aggregate aid (de Ree & Nillesen, 2009) or on one particular form of foreign aid (Findley, 2018; Narang, 2014; Zürcher, 2017). However, donors have different aid instruments to choose from—humanitarian, development, and peacebuilding assistance.

The existing literature is not clear on how donors allocate different types of aid in postconflict environments, or how aid-allocation patterns shape postconflict outcomes. We follow Campbell and Spilker (2022) and Findley (2018) by empirically exploring both the level and composition of total aid, and their association with the duration of peace after conflict. Donor decision-making on aid is a function of demand by national counterparts and of donor preferences. Using data from the Organisation for Economic Co-operation and Development Creditor Reporting System (OECD-CRS) codes, we cover all country-allocable OECD-DAC aid in the 1990–2022 period, and disaggregate it into humanitarian, development, and peacebuilding assistance. As such, this study is largely centered on self-reported data from OECD donors, which is a limitation. However, several non-OECD donors also report their aid through OECD CRS. ¹

Our statistical analysis yields important pertinent findings. The aggregate level of official development assistance (ODA) is correlated to the survival of peace. However, while countries receiving more aid are less likely to relapse into conflict, the effect disappears when controlling for the composition of aid. The share of peacebuilding assistance is positively correlated with sustained peace. There is an early divergence in peacebuilding assistance between countries that keep the peace and those that relapse into conflict.

The article makes several contributions. First, it confirms a statistical relationship between aggregate donor behavior and peace outcomes, suggesting that aid matters for durable peace. Second, it highlights the novel finding that the composition of total aid in a given country could be important for sustained peace. The relationship between the composition of total aid, collectively provided by donors across all aid domains, and postconflict peace has not been analysed before. Third, while the paper does not establish causality, the results lay the necessary groundwork for future research on the relationship between peacebuilding assistance and conflict.

Foreign Aid in Postconflict Settings

In postconflict settings, needs are often overwhelming. There is a strong sense of urgency and prioritisation is challenging (The World Bank, 2011). Postconflict situations are high-risk environments. A recent history of conflict is a strong predictor of future conflict (Hegre et al., 2013). Conflict can become a driver of aid dependency (Marktanner & Merkel, 2019). DAC donors allocate the majority of their aid to countries affected by conflict (Campbell & Spilker, 2022).

Various considerations underlie donor decisions on aid allocations in postconflict environments. Aid decisions are often the product of dialogue with recipient country governments and other national counterparts, through which priorities are identified. Each donor makes individual decisions on the level of ODA it provides. Donors also decide on how much aid is allocated to humanitarian, development, and peacebuilding assistance. While humanitarian assistance often follows (exogenous) shocks of man-made and natural disasters, nearly 40% of humanitarian appeals have been unmet in recent years (Development Initiatives, 2023). Donors provide development and peacebuilding assistance in places where humanitarian needs remain unmet, which suggests that humanitarian assistance is not completely exogenous.

Aside from the semi-exogenous decision on humanitarian assistance, a major donor decision is whether and how much to contribute to development- and peacebuilding aid. While each donor sets their aid level and composition individually, they typically do so alongside many other donors. An individual postconflict state may have as many as 20 or 30 donors in any given year. The aid level and composition in a given postconflict country represents donor behavior “in aggregate” and thus the complex outcome of aggregate decision-making by multiple donors.

Previous research on aid-allocation either aggregates all ODA or focuses on one type of aid in isolation (Findley, 2018; Narang, 2014). Some authors differentiate between aid that bypasses or goes directly to government (Dietrich, 2013). However, donors may also use particular aid instruments strategically, which is only reflected in recent studies (Mross, 2019, 2022). Donors do not typically analyse the effects of the parallel allocation of different aid types on conflict and peace (Campbell et al., 2017). The literature does not tell how donors allocate different types of aid or how these aid-allocation patterns shape postconflict transitions (Campbell & Spilker, 2022).

Development assistance and sustaining peace

Development needs after conflict are often massive (The World Bank, 2022a, 2022b, 2023). Development aid encompasses a range of sectors such as transport, energy, education, health, social protection, agriculture, urban-, social-, and community development. The majority of ODA is development assistance. Overall goals for development aid tend to concern improving living conditions, reducing poverty, and improving education and healthcare through service provision, infrastructure, and material support. Development often targets the immediate needs of a population and does not necessarily seek to address the underlying political or social issues (Davenport et al., 2018).

The literature suggests that postconflict aid follows a certain cycle. After the end of conflict, donors reduce humanitarian and transitional aid, and instead increase development aid, thus increasing total country-level assistance. This is particularly prevalent in countries that seem to be moving toward an inclusive political settlement (Campbell & Spilker, 2022). Countries weakly committed to peace tend to receive relatively less support from donors (Smith et al., 2020).

There is mixed evidence on the relationship between conflict and development aid (Findley, 2018; Zürcher, 2020). The literature highlights various causal mechanisms through which aid increases or decreases conflict. One is to make the prize of state control more valuable and, therefore, worth violent capture (Blattman, 2023). Another is exclusion and grievance, when, for example, development projects lead to an unequal distribution of aid across communities leading to increased insurgent activity (Karell & Schutte, 2018). The effect of development assistance in postconflict situations may also depend on the development sectors chosen. Aid has been found to be effective in improving social infrastructure, but ineffective in improving economic infrastructure (Donaubauer et al., 2019).

Many studies highlight the predicaments of development assistance in postconflict situations. One study finds that one quarter of all development projects had negative unintended consequences due to donor failures to understand domestic political power dynamics (Davidson et al., 2022). To successfully undertake conflict sensitive approaches, an understanding of the formal and informal political systems and processes that determine the political and economic behaviors in a postconflict setting is needed (Paudel et al., 2023). Other studies highlight the need to counter elite corruption in efforts to reestablish macroeconomic structures during reconstruction (O’Driscoll, 2018).

A significant body of research points to the need for donors to apply contextually tailored and politically informed approaches to development, to avoid unintentionally promoting conflict. Given the risks of unintended negative consequences, development cooperation needs to be marked by close partnerships with local and national actors (Kolk & Lenfant, 2015).

It is important to note that much of the research either addresses aggregate development flows, or the effectiveness of individual development projects. Less is known about the interaction effects, that is, whether different donor activities are complementary, irrelevant to-, or aggravating each other.

Peacebuilding assistance and sustaining peace

Peacebuilding assistance directly targets violence reduction, underlying drivers of conflict, and social cohesion (Hartzell et al., 2001; Hoddie & Hartzell, 2003; Walter, 2002). This type of aid encompasses, for example, support to disarmament, the reintegration of former combatants, dialogue, reconciliation, governance, legal development, accountability mechanisms, political party support, inclusive participation, and civil society.

Peacebuilding assistance can either support “negative” or “positive” peace. Negative peace is seen as the absence of violence, or war (Galtung, 1964). Positive peace can be viewed more broadly as the conditions that allow people and societies to realise their full potential (Galtung, 1969). Peacekeeping operations are aimed at affecting negative peace, through reducing or preventing violence. ODA funds peacekeeping efforts to a small but significant degree. ² Research suggests a strong relationship between peacekeeping and negative peace, that is, the reduction or absence of violence, after civil war (Walter et al., 2021). In contrast to conventional development aid, positive peacebuilding goals are to create long-term, sustainable peace by preventing future violence. Positive peace thus emphasises political inclusion, promoting social justice, economic equity, and human rights. Positive peacebuilding addresses the root causes of conflict, through transforming the relationships and systems that have led to violence (Davenport et al. 2018).

Research shows mixed effects of peacebuilding aid on positive peace objectives. This may go back to overly ambitious goal setting by donors who want to achieve liberal democracy in postwar countries that have never previously experienced it (Cooper et al., 2011). International efforts with limited local, nonelite ownership have proven less effective. Peacebuilding donors find it challenging to embrace diverse and authentic local forms of mobilisation in pursuit of peace (Bojicic-Dzelilovic & Martin, 2018). There are limits to what external peacebuilding efforts postconflict can do to redistribute power in domestic state formation processes (Krause & Jütersonke, 2005). Recent evidence indicates that aid effectiveness in fragile settings is improved by support to state institutional reforms that promote legitimacy (Carment & Samy, 2023).

While sustained peace after conflict necessarily depends on the drive and willingness of domestic actors, the international community can play a role in peacebuilding (Adedokun, 2017). Fiedler et al. (2020) and Mross (2019) show that negative peace and positive peace can be enhanced simultaneously. Mross (2022) specifically demonstrates that postconflict aid to democratic institutions, processes, and actors can foster support for cooperation, and for furthering institutional constraints and competition.

Mross et al. (2022) highlight the importance of combining different types of peacebuilding support. The authors disaggregate international peacebuilding support into five subtypes: Peacekeeping, nonmilitary security support, politics and governance support, socioeconomic development, and societal conflict transformation. They find peacekeeping to be an effective postconflict support type, but that the largest share of sustained peace cases is explained by support for politics and governance (Mross et al., 2022).

A growing body of research suggests that the donor choice for peacebuilding assistance reflects the political will of leaders in postconflict countries. Sustainable peace can only happen when domestic elites promote transitional justice, tolerance, and reconciliation (Mohammed, 2019). Such initiatives signal national or local commitment to build a lasting peace. However, if there is weak domestic demand for political reform and addressal of grievances, donors may steer away from peacebuilding assistance out of respect for national sovereignty (Cohen, 2014; Graben & Fitz-Gerald, 2013).

Aside of supporting national or subnational peace aspirations, some donors may also have intrinsic strategic interests in keeping a certain country or region peaceful and stable. The degree to which donors will increase aid could depend upon their geostrategic interests in the recipient state (Flynn, 2017). Major powers may be more likely to get engaged if the region in which the country is situated is strategically relevant, motivated by the drive to establish or expand influence (Clare & Danilovic, 2022).

Formulating hypotheses

As mentioned above, the literature has, thus far, not investigated how or why donors allocate different types of aid or how aid-allocation patterns shape postconflict outcomes (Campbell & Spilker, 2022). Aid may both reduce and exacerbate the risk of conflict relapse. We take a deductive approach to empirically examining the aggregate level and composition of different types of aid.

As a point of departure, we assume that if donors significantly increase total aid in the postconflict period, they must either believe in the national leadership´s commitment to peace or have geostrategic or other interests in the stability of the recipient country. Regardless of the direction of causality, there will be a positive relationship between total ODA levels and sustained peace after conflict. This is formulated in Hypothesis 1:

Hypothesis I: Postconflict countries that receive more aid are less likely to relapse into conflict.

Next, we turn to the composition of postconflict ODA and its relationship to sustained peace drawing on research pointing to the usefulness of politically informed transitional aid efforts to address grievances and underlying drivers of conflict:

Hypothesis II: Postconflict countries with more peacebuilding assistance are less likely to relapse into conflict.

We thereafter differentiate between “negative” peacebuilding and “positive” peacebuilding support. We posit that negative peace measures, focused on basic safety and security (see Table 1), will be critical in the years immediately following conflict termination, but that continued reliance on negative peace support could increase the risk of relapse, in accordance with Tikuisis et al. (2015). Sustained peace thus depends on sustained positive peacebuilding measures in line with Mross et al. (2022):

Hypothesis III: Postconflict countries that receive more “positive peacebuilding” assistance are less likely to relapse into conflict.

Table 1.

OECD-CRS Codes for Negative Peacebuilding and Positive Peacebuilding Assistance Used in the Analysis.

Negative peacebuilding

15210: Security system management and reform 15220: Civilian peacebuilding, conflict prevention, and resolution 15230: Participation in international peacekeeping operations 15240: Reintegration and small arms and lights weapons (SALW) control 15250: Removal of land mines and explosive remnants of war 15261: Child soldiers (prevention and demobilisation)

Positive peacebuilding

15110: Public-sector policy and administrative management 15111: Public finance management 15112: Decentralisation and support to subnational government 15113: Anticorruption organisations and institutions 15130: Legal and judicial development 15150: Democratic participation and civil society 15152: Legislatures and political parties 15153: Media and free flow of information 15160: Human rights 15170: Women's rights organisations, movements, and government institutions 15180: Ending violence against women and girls 15190: Facilitation of orderly, safe, regular, and responsible migration and mobilisation

Note. OECD-CRS = Organisation for Economic Co-operation and Development Creditor Reporting System.

Method

We now introduce the data, definitions and coding rules, and outline our methodological approach. Only two variables of interest are in focus: Aid (levels and composition) and conflict (presence of minor or major conflict). The event studies and the survival model analysis that will be described below are intentionally parsimonious to avoid overidentifying any of our models. The event study is a purely descriptive approach. The survival model is used to statistically test relationships between key variables.

Aid data

We use OECD-CRS aid data, aggregating the deflated gross disbursed amounts of ODA per recipient country, year, and CRS purpose code, for the period 1990 to 2022 (OECD Creditor Reporting System, 2024). The aggregate- and composition of aid is for all donor countries in a year.

The total donor ODA envelope in each country is then disaggregated into humanitarian, development and peacebuilding assistance for every year in the sample.

Humanitarian aid (H) is computed using sector code 700, that is, all CRS Codes 700xx.

Peacebuilding assistance (P) is computed using the OECD 2022 States of Fragility report 18 CRS purpose codes: 15110, 15111, 15112, 15113, 15130, 15150, 15152, 15153, 15160, 15170, 15180, 15190, 15210, 15220, 15230, 15240, 15250, 15261 (see Table 1).

Development aid (D) is all other sector codes not included in humanitarian and peacebuilding. ³

Following this schema, the composition of aid during a year/period is defined by the percentage of all aid that is humanitarian assistance (H%) and percentage of aid that is peacebuilding assistance (P%). The development share of assistance (D%) is then the remainder, D% = 100%-H%-P%.

For each ODA category aggregate (H, D, P) we also calculate per capita (pc) assistance based on population data from the World Bank World Development Indicators dataset, resulting in three per capita measures for every year (H_pc, D_pc, P_pc).

Conflict data

We use conflict data from the UCDP Georeferenced Event Dataset Global version 23.1 for the period 1990–2022. Countries are grouped according to their conflict profile.

Conflict: A country is considered “in conflict” if it experiences 25 battle deaths or more in a year in a state-based conflict (i.e., a conflict in which one side is the government). During the sample period, a few countries were in conflict all the time, “always conflict.” (For countries that became independent during the sample period and have data preceding independence, we use conflict data corresponding to the conflict with the country from which independence was sought.)

Postconflict: When a country experiences its first-year battle-death drop below 25 per year, a “postconflict” period starts.

 ○ No relapse: If a country experiences no conflict again during the sample period, the country is considered a “no relapse” country.

 ○ Relapse: Countries that again experience a minimum of 25 battle deaths in a year before the end of the sample period are considered “relapse” countries in that year. (Some countries relapse multiple times over the sample period.)

Control group (no conflict): Using the 25-battle death definition, a “control group” of countries that do not exceed the threshold necessary to be considered “in conflict” (above) are included. ⁴ (The control group data is used for the event study analysis, but not in the survival analysis).

Undetermined: There are also unclassified country-year observations for countries before they experience conflict onset. They can be considered “preconflict” and their closest comparator are the “control group” country years. (Underdetermined category cases are included in the descriptive statistics only and not in subsequent analysis.)

Methodological approach

Based on the data we generate descriptive statistics on ODA levels and ODA composition of postconflict countries and compare that to countries in the other conflict categories.

To examine the link between the ODA profile of postconflict countries and the prevalence of conflict relapse or no relapse, we use both an event study approach and a Cox proportional hazards survival model, treating each postconflict period for each country as a separate observation. The event study approach, defines conflict end as the last year of a conflict, and thereafter plots the average aid per capita for all postconflict countries (Chen et al., 2008). The Cox model is used to statistically test the relationship between aid levels and composition and the survival of peace after a conflict ends. This statistical approach assumes that the chance of relapse into conflict is the same every year. We test whether independent variables (level of aid, level of certain types of aid, composition of aid) are related to conflict relapse. We intentionally keep the model parsimonious. Though there are 1366 country-years in our sample, there are only 80 unique postconflict countries—the inclusion of too many independent variables may overidentify the model.

Empirical analysis

The level and composition of aid in postconflict countries, and how these compare with countries in the other categories, are described in Table 2. There are 178 countries with aid data for the sample period, 1990–2022, in the OECD-CRS Database from which data was extracted on 25 January 2024 (The OECD, n.d.). Only the 87 countries in the “No Conflict-Control group” and five countries, in the “Conflict” category, that experienced conflict throughout the entire period belong exclusively to one category. The remaining 86 countries move between the other categories over the 1990–2022 sample period.

Table 2.

Descriptive Statistics by Conflict Category.

Country type	Country-years	No of periods	Countries	Average period length	Composition measures			Level measures
					H % of aid	D % of aid	P % of aid	Hpc py	Dpc py	Ppc py	ODA pc py
Postconflict	1366	158	80	8.6	5.4	83.3	11.4	4.6	73.2	12.8	90.6
Conflict	968	184	85	5.3	16.3	75.3	8.5	12.3	35.2	7	54.5
Control/no conflict	2113	87	87	24.3	3	88.8	8.3	6.6	225.3	26.5	258.4
Undetermined	427	37	37	11.5	1.9	93.9	4.2	1	33.7	2.3	37
Total	4874	466	178	10.5	6.2	85	8.8	6.7	123	16.3	146

ODA = official development assistance; OECD-CRS = Organisation for Economic Co-operation and Development Creditor Reporting System.

Source. Authors’ calculations following the methodology described above, using original data from OECD-CRS and UCDP. The ODA composition is measured by the percentage of aid that is humanitarian (H%), development (D%) and peacebuilding (P%). The level is measured by the per capita per year (H_pc, D_pc, P_pc) aid for each of these categories. The sum of these three is the total per capita ODA per year, ODA_pc.

Aid level and sustaining peace

In postconflict country-years, total ODA per capita (ODA_pc) per year is $90.60, which is higher than in conflict ($54.50 ODA_pc), but also markedly lower than the in no conflict control group ($258.40 ODA_pc). The control no-conflict group receives nearly 3 times as much ODA per capita per year than the postconflict countries. This effect is evident when using an event study approach, plotting the average aid per capita for all postconflict countries after the end of conflict in Figure 1. For completeness, Figure 1 also includes the average for the five countries that experienced conflict throughout the sample period.

Figure 1.

Event study: Postconflict ODA per capita with comparator groups.

Figure 1 shows that ODA per capita in the first postconflict year is similar to aid during conflict. As expected, as the postconflict country proceeds further away from the conflict in time, aid increases, moving toward countries in the control (no conflict) group, increasing at about $4.2 per year, on average. After about 5 years, total aid in postconflict settings diverges significantly from conflict countries. The increasing average reflects both the increasing postconflict aid and the average of those that have survived without relapse to year t. The gray confidence band shows significant variance within the postconflict group.

Testing hypothesis 1

We test our first hypothesis if postconflict countries that receive more aid are less likely to relapse into conflict. Our model results show that postconflict countries that receive more aid are indeed less likely to relapse into conflict (Annex Table 1, column a). The statistically significant ODA per capita coefficient at −.0052 suggests that a dollar increase in average ODA per capita, is associated with roughly a half percentage point less chance of relapse in a comparator country. These results are reflected in the event study shown in Figure 2. ⁵ Countries that do not relapse into conflict receive increasing amounts of aid, reaching approximately $100 per capita in year 5 ⁶ while average aid per capita in the relapse countries hovers around $50. The annual confidence intervals accommodate the within-class variance due to country size, ensuring that countries of different sizes can be compared. ⁷

Figure 2.

Event study: Divergence in total aid per capita, no relapse and relapse.

Table 3 shows the descriptive statistics of the postconflict countries that relapse into conflict and those that do not. An ANOVA test confirms that the levels of ODA per capita, peacebuilding per capita and peacebuilding % of aid are significantly different for the relapse and no relapse countries. These results might suggest that the no relapse countries are preferentially treated by donors.

Table 3.

Descriptive Statistics Decomposed for Post-Conflict Relapse and No-Relapse Cases.

Country type	Country-years	No of periods	Countries	Average period length	Composition measures			Level measures
					H % of aid	D % of aid	P % of aid	H pc py	D pc py	P pc py	ODA pc py
Post Conflict	1366	158	80	8.6	5.4	83.3	11.4	4.6	73.2	12.8	90.6
No Relapse	883	53	53	16.7	4.1	82.8	13.1	4.8	87.1	16.9	108.8
Relapse	483	105	49	4.6	7.7	84.1	8.2	4.3	47.7	5.4	57.4

Note. Same methodology as in Table 2, postconflict is disaggregated into relapsing and nonrelapsing cases.

Aid composition and sustaining peace

We next examine the composition of aid. Table 2 showed that the peacebuilding assistance share was higher in postconflict situations (11.4%) compared to the others (around 8% for conflict and no conflict countries). Within the postconflict category, peacebuilding was higher in the no relapse countries (Table 3). The average postconflict peacebuilding assistance level was $12.80 per capita, about twice the level of conflict countries at $7.00 but less than half the allocation of control group (no conflict) countries at $26.50. Table 2 demonstrated that within the postconflict category, the percentage of peacebuilding aid was 13.1% in no relapse countries (average peacebuilding per capita was $16.90 per year), significantly higher than 8.2% in relapse countries (equivalent to $5.40 because total aid is lower, as well). Figure 3 graphically illustrates this—despite higher variance, peacebuilding per capita levels in nonrelapsing countries are consistently higher than the relapsing countries for the first 5 postconflict years.

Figure 3.

Event study: Divergence in peacebuilding per capita, no relapse and relapse.

Note that relapse countries receive 7.7% of their aid as humanitarian support, whereas nonrelapse countries receive only 4.1%. There is only a negligible difference in the level of humanitarian aid per capita between countries that relapsed ($4.30) and those that did not ($3.90), suggesting that the overall aid envelope is increasing for the no-relapse countries, but that the underlying humanitarian needs are not significantly different postconflict.

Testing hypothesis II

Informed by the descriptive statistics and results above, we test our second hypothesis, whether postconflict countries with higher peacebuilding assistance are less likely to relapse into conflict.

The results (reported in Annex Table 1, Column b) demonstrate a significant and strong relationship between peacebuilding aid per capita and avoiding conflict relapse at −.0325. For $1 additional average peacebuilding per capita, the associated chance of conflict relapse is 3.2% lower. We test whether this relationship holds if both total ODA and peacebuilding levels are considered simultaneously, but collinearity between the variables muddies the explanatory value for each (Column c). To overcome collinearity, the model is run again, using a level effect for total aid (ODA per capita) and a composition effect for the share of peacebuilding (P%) (Column d).

The results in Column d strongly support our second hypothesis. While countries that receive more aid are less likely to relapse into conflict (Column a), the explanatory power now moves off of total aid and to the percentage of aid that takes the form of peacebuilding assistance. In Column d, a $1 increase in average aid per capita is related to a lower likelihood of conflict relapse by only .36%, a smaller magnitude than observed in Column a, and not statistically significant. Meanwhile, the composition of aid matters statistically: For every 1% of aid that is peacebuilding (rather than humanitarian or development assistance), the related likelihood of conflict relapse is around 4% lower. The attribution and identification issues discussed previously still hold.

To visualise the probability of conflict relapse, we divide the full sample into two groups, those above and those below the peacebuilding per capita median. The Kaplan–Meier survival curve in Figure 4 shows two (roughly parallel) trends with a higher survival probability, that is, the probability of avoiding conflict relapse, for countries in the group with higher peacebuilding per capita.

Figure 4.

Kaplan–Meier survival curve: Stratified by level of peacebuilding per capita.

Figure 5 visualises the “composition of aid” effect—that is, the peacebuilding share of total aid and conflict relapse—through an event study. From Year 1, nonrelapse countries start receiving significantly more of their aid in the form of peacebuilding support, with an initial push above 15% in the first 5 years and averaging well over 10%. However, relapse countries receive on average markedly less than 10%.

Figure 5.

Event study: Divergence in peacebuilding per capita, no relapse and relapse.

It is possible that countries predisposed to peace are more likely to receive peacebuilding assistance. To examine this further, we compare the level of total aid, the level of peacebuilding, and the percentage of peacebuilding between the relapsing and nonrelapsing countries for “Year 0,” that is the year before the postconflict period starts. An ANOVA test reveals that the level of aid and peacebuilding per capita are not significantly different for the relapse and no relapse countries, while the peacebuilding percentage in Year 0 is significantly different at the 5% level. These findings suggest that the relapse and no relapse countries are not vastly different immediately before the postconflict period starts but that there are still issues that would merit further research.

There is a marked difference in peacebuilding per capita levels between the relapse and nonrelapse countries postconflict. This is demonstrated in the box plot, which shows countries that have similar levels of total aid per capita (see Figure 6). We divide the sample into categories of total ODA per capita: “Low” is countries that receive $0–$20 per capita, per year, “medium” is $20–$100 pc, annually, and “High” is greater than $100 per capita, per year. Within each category, nonrelapsing countries receive more peacebuilding per capita than countries that did relapse. To compare like with like, we only consider cases that last over 5 years in our sample.

Figure 6.

Box plot, countries with similar total ODA levels per capita.

There are notable comparisons. Among the countries in the category receiving medium levels of aid, for example, nonrelapsing Guatemala, Papua New Guinea, Sierra Leone, and Nepal, they have an order of magnitude higher level of peacebuilding per capita compared to relapsing Niger, Cameroon, Bangladesh, Nigeria, and Egypt. Within the category of high ODA, the Western Balkans region stands out. Kosovo, Bosnia-Herzegovina and Croatia received an order of magnitude higher peacebuilding per capita than Lebanon, Senegal, and the Congo. While there are outliers in both directions, the groupwise comparisons visually demonstrate how the peacebuilding effect dominates the total aid effect.

These findings give rise to challenging questions on inference. It is not clear whether peacebuilding aid is contributing to stability or whether peacebuilding is being directed toward countries already more inclined toward peace, thus implying endogeneity. Going forward, more qualitative and comparative approaches could help explore the direction of causality. Another aspect emerging from the findings that would merit further research has to do with the relevance of lead donors in given geographies, such as European donors in the Western Balkans, or Australia in Timor Leste.

Positive peacebuilding aid and sustaining peace

Testing hypothesis III

We move on to our third hypothesis that postconflict countries that receive more positive peacebuilding are less likely to relapse into conflict.

We divide peacebuilding aid into two types: “negative” peacebuilding and “positive” peacebuilding for each country-year period. While negative peacebuilding assistance, focused on stability and force, is necessary immediately at the end of conflict, it may be insufficient to secure long-term peace (Mross et al., 2022; Tikuisis et al., 2015). Positive peacebuilding addresses underlying grievances related to governance, political inclusion, social cohesion, and consequences of violence, that need to be addressed for peace to sustain. We test whether the type of peacebuilding received over the postconflict period matters for sustained peace. In line with Mross (et al., 2022) and OECD (2022), we use the OECD-CRS codes starting with 151 as positive peacebuilding and the codes starting with 152 as negative peacebuilding as shown in Table 1.

The results support our third hypothesis, as seen in Annex Table 1 Column e. For every dollar increase in average “positive” peacebuilding per capita since the end of conflict, the country is 8% less likely to relapse into conflict. However, the “negative” peacebuilding per capita has a positive coefficient, which suggests that a dollar increase is related to a 10% higher likelihood of conflict. To ensure that it is not the “donor darling-donor orphan” effect that dominate in the no relapse countries, we include total ODA per capita but find no significant relationship (Column f). Concordance and the information criteria suggest that the inclusion of ODA in Column f has not increased the explanatory value of the model, so Column e should be interpreted as our main results. ⁸ To ensure that our results are not driven by definitional issues, we carry out robustness checks using alternative specifications of peacebuilding assistance. The results confirm that the peacebuilding share is significantly higher for no relapse countries regardless of how peacebuilding is defined. ⁹

It needs to be noted, however, that not all support to negative peacebuilding is captured in the OECD-CRS data. A more in-depth assessment of the relative merits of negative peacebuilding compared to positive peacebuilding support, and the relationship to peace or conflict prevalence, would also require factoring in nonaid data, for example, on country allocable financing for UN peacekeeping or postconflict demining.

Conclusion and Discussion

To understand how donors allocate aid in postconflict recipient countries and how that correlates with sustained or ruptured peace, we disaggregate all country allocable ODA into humanitarian, development, and peacebuilding assistance over the 1990–2022 period. This yields 1366 postconflict country-year observations.

Our statistical results confirm that postconflict countries receiving more aid are less likely to relapse into conflict and highlight the novel finding that the composition of total aid is important. As the aid composition is added as a variable in the statistical analysis, the total ODA level is no longer statistically significant. Instead, the explanatory power fully shifts to the peacebuilding share of total ODA. Where more foreign aid takes the form of peacebuilding support, peace is more likely to last. The clear divergence in both the peacebuilding per capita level and the peacebuilding share of total aid between the nonrelapse and relapse countries, is notable almost immediately after conflict termination. The negligible difference in the level of humanitarian aid per capita between the no relapse and relapse countries suggests that the underlying humanitarian needs are not significantly different. The main difference between the no relapse and relapse states is the proportion of long-term—non-humanitarian—aid that goes to peacebuilding.

Our contribution is to take a necessary and fundamental first step to analyse the collective behavior of donors in postconflict settings across all aid types—humanitarian, development, and peacebuilding support—in terms of levels and percentage breakdowns. This has not been done before and is especially important to policy. Our findings thus lay the groundwork for further research that can help establish causation.

Our results empirically support the core notion of the peacebuilding-development nexus (McCandless, 2021) and speak to the wider debate on whether foreign aid ameliorates conflict or not (Findley, 2018). That peacebuilding ODA is associated with favorable outcomes in conflict-to-peace transitions aligns with recent research based on perceptions data (Campbell & Spilker, 2022; Mross et al., 2022). The low peacebuilding assistance in the conflict relapse countries conform with earlier findings that donors are less effective at supporting post-conflict states regressing back toward war (Matanock, 2020; Walter et al., 2021). We find that positive peacebuilding assistance is significantly correlated to sustained peace, in line with Mross et al. (2022).

Our results also show that peacebuilding takes time, thus problematising the notion of “postconflict” duration. Previous literature has proposed postconflict periods of 5–7 years, based on the time it takes for an economy to recover to “preconflict” levels (Chen et al., 2008; Pani, 2008). It is difficult to compare the length of “no relapse” to “relapse” periods as the former do not have an end. The disparity between an almost 20-year average for postconflict no relapse countries and a 10-year average for the relapse countries validate the notion of perpetual peacebuilding where impacts of conflict need to be addressed several decades after conflict (Hoeffler, 2012; Paffenholz, 2021).

The main limitation of this study is that we do not establish causation. Factors to be examined in future research aimed at proving causation concern motives and opportunities on the part of both domestic and external actors. The political will and commitment to peace by recipient governments matter greatly. To assess domestic commitment, measures of domestic expenditure and other policy actions significant for peacebuilding could potentially be used. The role of lead donors and their interests in particular geographies is also worth looking into. Opportunities related to postconflict aid allocation may concern the preexisting presence and structures of international agencies and international NGOs during conflict, which might affect aid patterns after conflict termination. External security support to back up peace agreements and provide security guarantees, such as non-OECD CRS expenditures on international peacekeeping, would need to be factored in as well. A causal model would also require including external military support that might counter peace promotion efforts—for example, through arms transfers to conflict parties.

Given the various challenges with inference, an alternative way forward would be to employ a qualitative approach using matching cases. That would enable in-depth comparisons of postconflict countries that are similar in terms of income levels and conflict profiles, but that differ on peace or conflict prevalence. Such an approach could help to identify possible explanatory factors and, in particular, lay bare the role of peacebuilding assistance.

As alternative peacebuilding conceptions emerge, including those focused more on short-term deals and coercive military power (Lewis, 2022), evidence based approaches are much needed. At a time when aid budgets are shrinking and the number of armed conflicts is high, traditional and new donors alike are forced to consider aid effectiveness and how to prioritise between different types of support. Our results speak to the utility of support to peacebuilding aid to governance, judicial development, inclusive participation, and human rights promotion as a critical complement to other development assistance in postconflict settings.