Aid allocation across the humanitarian-development-peace nexus: the role of fragility as a donors' motive

The promise of leaving no one behind is at the heart of the Agenda 2030 for Sustainable Development. The world today still sees millions of people at grave risk of being left far behind, most markedly in fragile contexts. In 2020, before the pandemic, three-quarters of the world's extremely poor resided in fragile contexts, and none of these contexts were on track to achieve SDGs on hunger, health, gender equality and women empowerment. The war in Ukraine exacerbates economic instability in poorer, fragile countries. The combined impact of pandemic, climatic, and economic shocks is putting people in fragile contexts at the risk of being left even further behind. Official Development Assistance (ODA) can be a powerful instrument to tackle fragility, but coordination across sectors and among donors is paramount. Humanitarian action responds to needs but cannot prevent crises or end needs without development and political solutions. The Humanitarian-Development-Peace (HDP) nexus calls for coordinated, coherent interventions across humanitarian, development, and peace pillars. Fragility and the nexus are intertwined, as addressing fragility often requires support from all three pillars. Operationalising the nexus faces financing challenges; actors tend to work in silos and separate funding streams widen gaps and hamper collaboration. Donors are recommended to adopt funding mechanisms bridging the HDP spectrum. There is limited empirical evidence on how donors allocate aid across the nexus in response to fragility. This research aims to identify the role fragility plays as a donors' motive and investigate if and how the allocation of humanitarian, development, and peace ODA is tailored to address fragility. It quantitatively analyses the link between aid allocation across the nexus and indicators of fragility, and is structured to review literature, present methodology, report results and discuss policy implications.

The literature on aid allocation traditionally considers three categories of donor motives: self-interest, recipients' needs, and merit. Evidence shows donors’ self-interests influence allocation, including economic motives (trade and commercial relations) and geopolitical interests (e.g., colonial ties, UN voting alignment), with studies by Barthel et al. (2014), Furuoka (2017), Younas (2008), and Alesina and Dollar (2000) among others. Recipients’ needs often correlate with higher aid to poorer countries (negative relationship with GDP/GNI per capita), though a middle-income bias is sometimes observed due to absorptive capacity concerns; population size can drive per capita aid patterns, including population bias. Merit involves policy and institutional quality, democracy, human rights, and governance, with mixed findings on whether better governance or human rights records attract more aid (e.g., Burnside and Dollar, 2000; Neumayer, 2003a,b; Alesina and Weder, 2002). Fragility intersects with these motives but is conceptually distinct. Debates on state ownership and aid effectiveness in fragile contexts suggest links between fragility and merit/absorptive capacity, yet evidence is mixed, and fragility is multidimensional beyond institutions alone. Post-9/11, security concerns linked to fragility gained prominence, highlighting overlapping elements between donors’ self-interest and recipients’ needs and the cross-border ramifications of fragility (e.g., conflict spillovers, global threats). The HDP nexus emphasizes the need for coordinated approaches across humanitarian, development, and peace activities in fragile contexts. This study positions fragility as a separate factor, alongside self-interest, needs, and merit, examining its role in aid allocation across the HDP nexus. It adopts multidimensional measures of fragility (OECD States of Fragility classifications and the Fund for Peace Fragile States Index) and notes a gap in empirical research specifically linking donors’ HDP allocations to fragility measured in this way.

Data and variables: The study uses a three-dimensional panel (donor, recipient, year) covering ODA commitments from 23 DAC donors to 126 developing countries, 2009–2019. Dependent variables are ODA commitments disaggregated by HDP pillars: humanitarian, development, and peace, classified using OECD CRS purpose codes. Explanatory variables are grouped into donors’ self-interests (exports from donor to recipient; geographical distance between capitals; colonial tie dummy; recipient CO2 emissions per capita as a proxy for environmental interests), recipients’ needs (recipient GDP per capita, with quadratic term when significant; population), and merit (Freedom House Index combining political rights and civil liberties). Fragility is treated as a distinct factor in two forms: (1) state of fragility (binary, based on OECD States of Fragility/Fragile States Reports across five dimensions: economic, environmental, political, security, societal); (2) degree of fragility (continuous score from Fund for Peace Fragile States Index across twelve indicators in five categories). Time-variant variables (exports, CO2 per capita, GDP per capita, population, freedom, fragility measures) are lagged one year. Several variables (ODA, exports, distance, CO2 per capita, GDP per capita, population) are log-transformed; one is added to zero ODA values prior to log transformation to retain zeros. Empirical model: Due to the prevalence of zeros in ODA data (approx. 61% development, 86% humanitarian, 80% peace), a limited dependent variable approach is used. The study employs a random-effects Tobit model, as a one-step method that accounts for censoring, avoiding challenges of two-part and Heckman selection models (e.g., independence assumptions and exclusion restrictions). The model assumes the same sign and set of determinants for eligibility and level decisions and homoskedasticity (mitigated via log transformations). To reduce endogeneity concerns, lagged explanatory variables are used; remaining endogeneity risks are acknowledged. Fixed-effects in Tobit would raise incidental parameters issues, so recipient/donor fixed-effects are applied only in robustness checks using dummy variables. Model specification (core): ln(ODA) = α + β1*fragility(t-1) + β2*ln_export(t-1) + β3*ln_CO2_pc(t-1) + β4*ln_distance + β5*colony + β6*ln_gdp_pc(t-1) + β7*ln_population(t-1) + β8*freedom(t-1) + year dummies + ε. Quadratic ln_gdp_pc term is included where significant in tests. Sample size and censoring: Main DAC23 regressions include at least 28,476 observations; censored observations total 19,983 (humanitarian), 7,270 (development), 13,869 (peace). Negative ODA commitments (3 observations) are excluded. Robustness checks include models with donor and recipient fixed-effects, tests excluding/including fragility variables only, recalculating FSI excluding three overlapping dimensions, and interaction terms between the fragility state dummy and other variables. An additional UN General Assembly voting similarity variable was tested but omitted from main models due to methodological concerns and insignificance without fixed effects.

Aggregate DAC23 results (Table 2): - Fragility: - Humanitarian ODA: Positive and significant association with both measures of fragility. State of fragility coef ≈ 1.644 (SE 0.316, p<0.01); degree of fragility coef ≈ 0.253 (SE 0.021, p<0.01). - Development ODA: Fragility not significantly associated. State coef ≈ -0.110 (ns); degree coef ≈ -0.004 (ns). - Peace ODA: State of fragility insignificant (≈ 0.120, ns), but degree of fragility positive and significant (≈ 0.107, SE 0.013, p<0.01). - Self-interest and other controls: - Exports: Positive and highly significant across humanitarian, development, and peace in both models (e.g., ≈ 1.404 to 1.477 for humanitarian; ≈ 0.542 for development; ≈ 0.885 to 0.902 for peace; all p<0.01). - CO2 per capita: Negative and significant for humanitarian and development (e.g., ≈ -1.47 to -1.57 humanitarian; ≈ -0.358 development; p<0.01), insignificant for peace. - Distance: Negative and significant across the nexus, especially for peace (e.g., ≈ -1.836 humanitarian; ≈ -1.042 development; ≈ -2.001 peace; p<0.01). - Colonial ties: Positive and significant across all three pillars (e.g., ≈ 4.7–6.0 development; up to ≈ 8.56 peace; p<0.01). - GDP per capita (needs): Negative and significant overall (e.g., ≈ -1.239 development; negative for humanitarian and peace), with evidence of a middle-income bias in some models where quadratic term is significant (positive linear and negative quadratic). - Population: Positive and significant across the nexus (e.g., ≈ 1.398 to 1.757 humanitarian; ≈ 1.419 development; ≈ 1.593 to 1.731 peace; p<0.01). - Freedom (merit): Positive and significant for development (≈ 0.014, p<0.01) and peace (≈ 0.017 to 0.034, p<0.01); for humanitarian, negative and significant when using the state of fragility model (≈ -0.045, p<0.01) and not significant with the degree model. - Donor heterogeneity (Table 3, Table 4): - Humanitarian: Degree of fragility is positive and significant for virtually all donors (all except Greece significant). State of fragility is positive for all donors and significant in nine cases (e.g., Belgium, Canada, Switzerland, Germany, Japan, Luxembourg, New Zealand, Norway, Sweden). - Development: State of fragility is generally insignificant; Denmark provides significantly less to fragile states (state negative). Degree of fragility significantly correlates for 9/23 donors: positive for Ireland, Netherlands, United States, Denmark; negative for United Kingdom, Japan, Korea, Portugal, Sweden; insignificant for others (some cases may reflect sparse non-zero allocations). - Peace: State of fragility significant for Germany and France (positive) and Switzerland (negative); otherwise mostly insignificant. Degree of fragility positively and significantly associated with peace aid for eleven donors. Robustness checks: - Including donor and recipient fixed-effects preserves the directions and significance of fragility variables: both fragility measures positive and significant for humanitarian; degree of fragility positive and significant for peace; development remains not associated. With fixed-effects, exports, distance, and colonial ties remain significant; population loses significance in development and turns negative in peace; freedom negative and significant in humanitarian, insignificant in development, positive and significant in peace; CO2 becomes insignificant. - Models estimated with/without fragility variables separately show stability of other coefficients; fragility alone is always positive and significant. - Reduced FSI (excluding three dimensions overlapping with other regressors) yields identical signs and significance to the main degree-of-fragility results; development remains not associated. - Interaction terms with the fragility-state dummy indicate differential effects for fragile vs. non-fragile recipients, particularly in humanitarian and peace aid (e.g., weakened effects of exports/CO2/distance in fragile contexts, reinforced population effects; adjustments to middle-income bias). Data features: - Observations: ≈ 28,476–28,478 depending on pillar; censored (zero) shares: ≈ 61% development, 86% humanitarian, 80% peace.

The study set out to examine whether and how fragility motivates donors’ allocation of ODA across the humanitarian-development-peace (HDP) nexus. The findings show that DAC donors, in aggregate, systematically tailor humanitarian aid in response to fragility (both its state and degree) and peace aid to the degree of fragility, while development aid shows no robust association with fragility. This pattern suggests donors prioritize urgent needs and stability-related interventions as fragility intensifies, yet remain cautious in scaling development aid to fragile contexts, possibly due to perceived absorptive capacity constraints. The consistent positive association between exports and all three aid pillars indicates that donors’ economic interests are embedded in allocation decisions, even in humanitarian assistance where neutrality and independence are foundational principles. Distance and colonial ties further reveal geopolitical structuring of aid flows. Needs-based targeting is evident through negative relationships with GDP per capita and positive relationships with population, while merit considerations (freedom) are rewarded in development and peace but not in humanitarian contexts, aligning with the emergency nature of humanitarian responses often occurring where freedoms are compromised. Donor-level analyses highlight substantial heterogeneity: while nearly all donors increase humanitarian aid with higher fragility, development and peace allocations vary widely by donor and by whether fragility is captured as a state or as a degree. This heterogeneity underscores the complex blend of motivations and capacities shaping each donor’s approach across the nexus. Policy implications emerging from these results stress the importance of coherence and coordination across donors and pillars. Given that many donors adjust allocations in ways related to fragility, aligning strategies and bridging siloed funding mechanisms could enhance the collective impact on fragile contexts. The evidence supports strengthening nexus-oriented planning to address interconnected humanitarian, development, and peace needs in fragile settings.

The paper contributes to aid allocation research by: (1) disaggregating ODA into humanitarian, development, and peace pillars; (2) treating fragility as a distinct, multidimensional factor alongside self-interest, needs, and merit; and (3) operationalising fragility via both a state (binary) and degree (continuous) measure. Results show DAC donors reflect recipient fragility in ODA allocations, with strong associations for humanitarian aid (state and degree) and for peace aid (degree), but not for development aid in aggregate. Donor responses and the influence of traditional motives vary considerably across donors and nexus pillars. Addressing fragility requires collective, coherent, and coordinated efforts across donors and sectors to generate synergies; enhanced mutual understanding and coordination can improve the consistency and effectiveness of international cooperation across the HDP nexus.

Methodological and data limitations include: (1) Tobit model restrictions that determinants affect both eligibility and level decisions with identical signs, and the assumption of homoskedasticity (partly mitigated via log transforms); (2) potential endogeneity remains despite lagging explanatory variables; (3) inability to include fixed-effects within Tobit without incidental parameters problems—recipient and donor fixed-effects are only applied in robustness checks; (4) extensive zero observations and the need to add one to zero ODA before log-transform; (5) conceptual and methodological limitations of fragility measures (OECD SoF and FSI), and partial overlap with other regressors (though correlations are moderate); (6) heterogeneity in donors’ allocation patterns and sparse non-zero allocations for some donor–sector combinations may limit power and significance; (7) exclusion of a small number of negative ODA observations; (8) using a common model specification across all individual donors may not capture donor-specific determinants optimally; (9) UNGA voting similarity variable was tested but omitted from main models due to methodological guidance and inconsistent significance without fixed-effects.