Rising Military Spending in the Context of Falling Development Assistance

This is the second in a series of three publications produced under the project ‘Development Aid as a Security Priority’ funded by the Gates Foundation. This paper examines the growing divergence between military spending and development aid among OECD DAC members, finding that cuts to aid will have significant humanitarian and development consequences while delivering only modest savings compared with planned increases in military spending. The final paper will examine the evolving relationship between development and security in Europe

I. Introduction

The 34 member and associate states of the Development Assistance Committee (DAC) of the Organisation for Economic Co-operation and Development (OECD) are the most significant source of humanitarian and development aid. Yet, at a time when global development and humanitarian needs are reaching unprecedented levels, the DAC states’ official development assistance (ODA) is entering a period of significant decline.¹ The number of armed conflicts has reached record levels, humanitarian and health crises have become more frequent and protracted, and many developing countries face persistent poverty, weak economic growth and mounting debt burdens.² Following record-high levels of ODA in 2022 and 2023, driven largely by exceptional support for Ukraine and the costs of hosting refugees in donor states (known as in-donor refugee costs, IDRC), aid levels have fallen sharply (see figure 1).

This retreat is taking place against the backdrop of the largest shift in the global security environment since the end of the cold war. Military expenditure by many of the same states that are members of the DAC is increasing rapidly, driven by rearmament in Europe and renewed spending commitments since the Russian Federation’s full-scale invasion of Ukraine and amid generally heightened geopolitical tensions.³ Although ODA and military expenditure have historically moved broadly in parallel, in recent years there has been a clear divergence in spending trends; and, as outlined below, in some cases governments have explicitly justified aid cuts as a means of financing higher military expenditure.

While the scale of the recent divergence has attracted considerable attention, emphasizing only the latest ODA cuts and military spending increases risks obscuring a longer-term structural shift. This paper examines the complexities underlying these two trends and finds that even a substantial gap between ODA needs and ODA funding is small relative to the scale of planned increases in military spending among DAC countries. This raises questions about the extent to which aid reductions can meaningfully contribute to rearmament while potentially carrying significant development and humanitarian consequences. The paper first (in section II) reassesses recent developments in ODA by distinguishing between temporary crisis-related assistance over the last few years and core development assistance. This provides a more nuanced picture than total ODA figures alone. It then (in section III) analyses the growing divergence between military expenditure and ODA across DAC states, including case studies of Germany and the United Kingdom, to explore whether development assistance is increasingly being crowded out by higher military spending. Finally, it considers ODA relative to rapidly increasing global development and humanitarian needs by using econometrics to create a needs-adjusted estimated level of aid and the resulting financing gap. The paper concludes (in section V) by highlighting the core findings. Appendix A follows with details of the methodology used for empirical analysis.

II. Trends in official development assistance

There was a sharp reduction in official development assistance in 2025. According to OECD data, ODA by the DAC states totalled US$174 billion in 2025, 23 per cent less than in 2024.⁴ The total accounted for only 0.26 per cent of these states’ combined gross national income (GNI), well below the target of 0.7 per cent of GNI that the United Nations General Assembly adopted in 1970.⁵ The decline in 2025 was the largest annual contraction in ODA ever recorded and brought total ODA commitments by DAC states back to 2015 levels in real terms.

While these headline figures suggest a sudden and significant reduction in ODA, the decline followed the largest annual increase in ODA by DAC states on record, in 2022. That increase was driven by overlapping crises, including the continuing effects of the Covid-19 pandemic and Russia’s full-scale invasion of Ukraine, which led to sharp rises in aid to Ukraine and in in-donor refugee costs.

Since ODA encompasses a range of activities with different objectives and time horizons, the OECD separates ODA for development, ODA for humanitarian assistance and ODA for IDRC. While humanitarian assistance provides immediate life-saving relief in response to crises, development programmes support longer-term poverty reduction, economic development and sustainable development. In-donor refugee costs are incurred within donor states during the initial period of refugee arrival and support and therefore do not represent resources transferred to developing countries.

In 2022, aid to Ukraine and IDRC accounted for almost all of the increase (see figure 1). This distinction between, on the one hand, support for Ukraine and IDRC and, on the other, what can be called ‘core’ ODA provides a clearer view of ODA more closely directed towards its development purpose: reducing poverty, promoting economic welfare and strengthening institutional resilience in developing countries and least developed countries (LDCs).

The increase in total ODA in years prior to 2022 was driven primarily by humanitarian assistance in response to immediate needs. IDRC also contributed to the increase, but to a lesser extent, particularly in 2016 when European states faced large inflows of refugees from Afghanistan and Syria.⁶ The higher level of total ODA in 2022 was sustained in 2023 but was followed by substantial reductions in 2024 and 2025, highlighting the high volatility of ODA over the past five years. Yet, volatility is inherent to humanitarian assistance given the need to respond to ongoing crisis.

Narrowing the focus to core ODA reveals more stability. Excluding bilateral aid to Ukraine and IDRC, average annual ODA in 2021–25 was 5.7 per cent higher than in 2016–20 (see figure 1). However, ODA (both total and core) declined in both 2024 and 2025, and announced future cuts suggest that this may mark the beginning of a broader and sustained decline in development assistance across multiple categories.⁷

Disaggregating ODA into different sectors and modalities points to trends similar to that for core ODA. By 2025, bilateral assistance by DAC states for education and health had fallen to below the levels that preceded the Covid-19 pandemic (see figure 2); the largest relative reduction has been in aid to states in sub-Saharan Africa.⁸ Education and health outcomes are fundamental to long-term human development and form two of the three cornerstones of the Human Development Index (HDI) of the UN Development Programme (UNDP); the sustained reduction in developmental assistance to these two sectors could therefore undermine progress in human development.⁹

In addition to ODA for education and health, a similar trend is shown in other components of ODA—ranging from agriculture and water and sanitation, via support for economic infrastructure and services to conflict prevention and peace and security. In earlier years, increases or decreases lasted no more than four consecutive years before reversing. Since 2022–23, however, a persistent downward trend has emerged (see figure 2).

According to one estimate, the ongoing and projected reductions in ODA funding could result in 9.4–22.6 million more deaths due to communicable diseases across all ages by 2030 in low- and middle-income countries, depending on the severity of the ODA reduction.¹⁰ In this context, the latest OECD projections for 2026–28 offer a stark and grim outlook: further reductions in ODA for education, health, peace and security, and food all continue to at least 2028.¹¹

The projected reductions in development assistance are occurring at the intersection of rising needs, domestic financial constraint and renewed drive for military expansion among DAC states. As many DAC donor governments seek to strengthen national security and build military deterrence, ODA is increasingly being weighed against military spending. In this trade-off context, cutting ODA is usually politically easier than reducing domestic social spending, raising taxes or borrowing.¹²

III. Parallel reductions in ODA and increases in military spending by DAC states

As total ODA has faced an unprecedented level of cuts since 2023–24, the military expenditure of many donor states—which are also some of the biggest spenders on the military—has been going in the opposite direction. Military spending by DAC members has reached the highest level ever recorded, and the year-on-year increase in 2025 was the highest recorded since the end of the cold war.¹³

Historically, DAC states’ spending on the military and spending on development have broadly followed a similar long-term trajectory. They have been driven by similar macroeconomic and geopolitical factors, although ODA has faced higher short-term volatility (see figure 3 and figure 1 above). Both increased during periods of sustained economic growth and increasing public expenditure in the early 2000s and then stagnated and decreased around 2010 after the global financial crisis and the budgetary pressure under austerity measures across many advanced economies.

From the mid-2010s, ODA started to recover gradually amid renewed economic growth and donor commitments to the Sustainable Development Goals (SDGs) adopted in 2015, as well as humanitarian crises in Afghanistan and Syria which increased support for IDRC.¹⁴ From 2015, military expenditure also started to rebound, especially by European members of the DAC following Russia’s annexation of Crimea in 2014.

More recently, however, the two indicators have diverged. Military expenditure has risen sharply following Russia’s full-scale invasion of Ukraine in 2022, reflecting European rearmament ambitions and higher spending commitments by member states of the North Atlantic Treaty Organization (NATO).¹⁵ As noted above, ODA also reached historically high levels in 2022–23, driven by exceptional spending related to Ukraine, refugees and the Covid-19 pandemic, before declining markedly from 2024 as the temporary expenditure subsided. Further reductions are projected as donor governments face mounting fiscal pressures and competing priorities. The recent divergence comes amid a longer-term evolution of the ways in which many donor governments frame the interconnection between development and security.¹⁶ While development has increasingly been linked to security objectives since at least the early 2000s, recent geopolitical tensions appear to be placing greater emphasis on military capabilities and deterrence than on development cooperation.¹⁷ The historically positive correlation between military spending and ODA is turning into a trade-off between expanding military budgets and maintaining development assistance within constrained public finances.¹⁸

Country case studies: Germany and the United Kingdom

For almost all of the period studied here, the United States, Germany and the United Kingdom were the top three spenders on both ODA and military expenditure among DAC states. They represent around half of total DAC ODA and about two-fifths of world military expenditure, and therefore largely shape the total trends for both. While the sharp decline in ODA in 2025 was driven primarily by cuts in US aid, US military expenditure also fell following reductions in military assistance to Ukraine. In contrast, Germany and the UK provide clearer examples of the budgetary trade-off between military and development spending. Both the changes in their own spending and their influence on other DAC states mean that these two states’ decisions give an indication of DAC states’ future spending trends.

Germany

Germany’s spending trends provide a good example of the recent divergence between ODA and military spending. Both remained relatively stable until the mid-2010s before rising from 2015 (see figure 4), reflecting reaction to Russia’s annexation of Crimea as well as increased spending on refugee costs and expanded development commitments.¹⁹ Military expenditure accelerated after Russia’s full-scale invasion of Ukraine in 2022, reaching record levels as Germany implemented its biggest military build-up plan since the end of World War II. The German government has explicitly linked reductions in development spending in recent budgets to its prioritization of military security, citing the changed European security environment and the need to strengthen military capabilities.²⁰

Based on the most recent budget announcements, this divergence is expected to grow even further, at least over the rest of the decade. According to Germany’s medium-term financial plans, ODA as a share of GNI is expected to fall from 0.56 per cent in 2025 to approximately 0.43 per cent by 2029.²¹ In contrast, Germany’s military spending is expected to roughly double by 2030 from its 2025 level of $114 billion, with the increase financed by a combination of borrowing, a special fund and budget reallocations.²² As a share of gross domestic product (GDP), it is expected to reach 3.5 per cent by 2029, up from 2.3 per cent in 2025. However, the increase in military spending will be more than 10 times larger than the planned ODA reduction—illustrating that cuts to development aid make only a marginal contribution to the financing of Germany’s military expansion while substantially reducing resources available for international development.

The contrast is equally apparent when compared with individual procurement programmes. Germany has placed arms orders worth tens of billions of dollars since 2020, including major acquisitions of F-35 combat aircraft and of air defence and land systems.²³ Annual procurement costs reached around $96 billion in 2025, more than three-and-a-half times Germany’s total ODA that year.²⁴ While the aircraft procured in 2022 are only expected to be operational around 2029, ODA cuts can have immediate effects on humanitarian assistance and development programmes, as shown by the sudden withdrawal of international assistance in Gaza and sub-Saharan Africa.²⁵ The planned annual reductions in Germany’s development budget are modest relative to these procurement programmes—they would finance only a small share of a single acquisition. This underscores that ODA cuts have a limited financial impact on government budgets but have a significant development impact.

The United Kingdom

The United Kingdom’s ODA increased gradually until the late 2010s before declining after the reduction of the aid target—initially presented as ‘temporary’—from 0.7 to 0.5 per cent of GNI in 2021 (see figure 5).²⁶ Following a short-lived increase associated with support for Ukraine and refugee costs, ODA fell again in 2024 and 2025. Meanwhile military expenditure continued to rise in response to Russia’s full-scale invasion of Ukraine and the British government’s objective that military spending would reach 2.5 per cent of GDP by 2027.²⁷

In 2025 the British government explicitly linked these trends by announcing that ODA would be reduced further—from 0.5 per cent of GNI to 0.3 per cent by 2027—to finance higher military spending.²⁸ Based on forecast GNI, this was forecast to free up around $7.5 billion for the military. The government described the shift as a necessary response to a deteriorating security environment while maintaining financial discipline.²⁹ This represents one of the clearest examples among DAC states of an explicit policy trade-off between development assistance and military expenditure. However, the reduction in ODA—projected to be $0.6 billion between 2025 and 2026—is again dwarfed by the planned increase in military spending of $5.6 billion in 2026.

Germany and the UK could represent a broader trend among major DAC donors, with several states announcing significant ODA reductions amid rising military and security priorities.³⁰ Their experiences provide two examples of how this broader shift is being justified and implemented. If these plans materialize, not only would future reductions in ODA exceed current projections, but they would also mark a paradigm shift in donor mentality by framing development and security as competing—rather than complementary—priorities. Although the money released from aid reductions is small compared with planned increases in military expenditure, the implications for sustainable development, humanitarian response and long-term global security could be significant.

IV. ODA relative to rising global needs

As shown above, by 2025 total ODA commitments by DAC states had fallen to levels comparable with those last seen in 2015. Once aid for Ukraine and in-donor refugee costs are excluded, the level of core ODA is even lower, at $141 billion. On this narrower measure, commitments have returned to a level not seen since 2008 (see figure 1).³¹At the same time, the level of development and humanitarian needs that ODA is expected to address has changed markedly since the late 2000s, with needs outstripping supply.³²

Indicators of the rising need for ODA

A range of indicators point to a more severe and deteriorating global environment, and consequent growing need for ODA, in 2025 than in either 2008 or 2015. The clearest deterioration is in conflict-related needs. Since 2011 the number of state-based conflicts, as defined by the armed conflict data set of the Uppsala Conflict Data Program (UCDP), has increased, with more than 50 conflicts active each year over the decade 2016–25.³³ In 2025 the number of state-based conflicts reached 65, the highest level ever recorded and 6 more than the second highest, reached in 2024.

Armed conflicts increase need for ODA in numerous sectors such as humanitarian needs, displacement, infrastructure destruction, weaker public services, food insecurity and lower economic growth.³⁴ They also place additional pressure on states that are already fragile or have limited fiscal capacity.³⁵ The regional distribution of conflicts is also significant: 29 of the 65 armed conflicts recorded in 2025 were in Africa and 19 in Asia. This concentration closely overlaps with patterns of fragility, with 52 of the 61 states classified as facing high or extreme fragility located in these two regions.³⁶

Additionally, many LDCs and low-income countries face overlapping pressures of weak economic growth, high and rising inequality, and rates of poverty in 2025 as high as they were 10 years before.³⁷ The World Bank estimates that about 45 per cent of the population of sub-Saharan Africa fell below the poverty line of $3 per day in 2025; if the threshold is increased to $8.3 per day—the benchmark for upper-middle income countries set by the World Bank—the proportion increases to 89 per cent.³⁸ As a result, due to both inflation eroding purchasing power and rising humanitarian and development needs, an ODA level in 2025 that may appear similar in historical terms to 2008 or 2015 will not have the same development or humanitarian impact.

Political violence is a well-established obstacle to economic and social development.³⁹ Data on political violence from the Armed Conflict Location and Event Data (ACLED) project points to the scale in recent years:⁴⁰ the number of recorded events of political violence—defined as the use of force by a group with a political purpose⁴¹—reached more than 210 000 across 163 states, up from about 42 000 events in 72 states in 2021.⁴² The growing scale and geographic spread of this violence in recent years indicate a potentially greater need for policies and strategies aimed at preventing, managing and reducing such violence. ODA by DAC states has historically played a crucial role in peace- and security-related activities, including peacebuilding, institutional resilience and social stability. For example, the European Union (EU) provided development assistance in Colombia to support the implementation of the 2016 peace agreement through budget support and rural reforms in ex-combat areas.⁴³

For developing countries that are exposed to the effects of climate change, shocks from natural disasters add another layer of need for ODA.⁴⁴ The number of natural disasters recorded each year by the EM-DAT database has been around 20 per cent higher in the 2020s than in the 2010s.⁴⁵ These events affect hundreds of millions of people in sub-Saharan Africa and the Asia-Pacific each year. As well as generating immediate humanitarian needs for food, water, shelter and health services, the destruction of infrastructure leads to loss of output and employment and so also weakens longer-term development prospects.⁴⁶ There is currently both a continued and increasing need for both humanitarian and development assistance for states caught in a climate–poverty trap.⁴⁷

Moreover, stagnating economic and human development outcomes add further pressure on development assistance. Poverty reduction has slowed markedly since the late 2010s; the World Bank notes that progress has stalled amid low economic growth, the setbacks of the Covid-19 pandemic and increased fragility. Poverty rates in low-income countries are higher than before the Covid-19 pandemic.⁴⁸ At the same time, the UNDP’s Human Development Report 2025 describes a major slowdown since 2020 in human development progress, with sub-Saharan Africa particularly affected.⁴⁹ These trends increase the need for core ODA directed towards poverty reduction and sustainable and inclusive economic growth.

The high—and potentially increasing—need for ODA alongside expected further reductions in ODA raises the concern of a widening funding gap. Assessing this gap requires comparing projected ODA with the level implied by historical donor responses to global crises and development pressures. Comparing actual and projected ODA can reveal insight that year-on-year totals cannot capture, including whether funding is falling while needs increase. The rest of this section therefore develops a needs-adjusted estimate for ODA and compares this with actual and projected ODA commitments by DAC states. The resulting difference provides an estimate of the potential ODA funding gap in 2026–28.

Projecting needs-based ODA and the ODA gap, 2026–28

To construct estimates for needs-adjusted ODA levels, an empirical model was used (see appendix A). The model captures the historical relationship between measured global needs and ODA commitments. This provides benchmarks for ODA levels under different need scenarios. These estimates reflect historical donor responses to needs, rather than the level of financing required to fully address current needs.

Following the empirical model, a projection of a needs-based ODA commitment level can be created based on an ODA needs index (see table 1). This was conducted for three scenarios.

1. A baseline scenario. This scenario assumes that need for ODA remains broadly in line with the historical levels of 2017–21. This excludes exceptional, ‘black swan’ events such as the Covid-19 pandemic and Russia’s full-scale invasion of Ukraine.

2. A higher-needs scenario. In this scenario global pressures intensify. Existing conflicts continue, new conflicts emerge, displacement increases, and economic and human development outcomes deteriorate as lower-income countries face higher energy prices, weaker growth and reduced progress towards the SDGs. Under this scenario, need for ODA rises.

3. A lower-needs scenario. In this scenario there is an improvement in global economic, political and security conditions. Ceasefires and peace agreements reduce the number and intensity of armed conflicts, economic growth improves, human development outcomes and poverty levels recover, and fewer people are affected by natural disasters. Under this scenario, need for ODA falls but does not disappear as under-development, fragility, climate change-related vulnerabilities and structural poverty remain.

As hoped, the model provides an excellent fit to the data on actual ODA in 2005–22 (see figure 6). From 2023, the three scenarios fork; but each prong of the fork reveals a stark gap between the need for ODA and actual ODA committed by DAC states and the OECD projection of commitments to 2028.

Under the baseline scenario, need for ODA is projected to reach roughly $262 billion by 2028 (in constant 2024 US dollars); in contrast, the OECD forecasts total commitments by DAC states to be only about $147 billion in 2028.⁵⁰ This implies a baseline funding gap of roughly $115 billion in 2028 (see table 1). Cumulatively, the gap between projected ODA and the needs-based baseline over the period 2026–28 amounts to approximately $313 billion. This shortfall, while significant, remains modest relative to the GNI of donor states. The projected ODA needs of $262 billion would amount to 0.42 per cent of DAC states’ 2025 GNI.⁵¹ Notably, this is well below the internationally agreed target of 0.7 per cent GNI, and DAC donors’ combined ODA has in the past come close to 0.4 per cent of GNI.⁵² Meeting the ODA needs projected in the baseline scenario is thus achievable.

Under the higher-needs scenario, the commitment gap is substantially larger. Predicted ODA based on needs would need to reach approximately $326 billion by 2028, more than double the projected DAC commitments. This will lead to a funding gap of $179 billion in 2028. Cumulatively, the gap over the period 2026–28 will reach approximately $474 billion. Yet under this higher-needs scenario, the estimated total ODA by DAC states would only amount to 0.52 per cent of DAC states’ 2025 GNI, still below the 0.7 per cent target.

Even under the lower-needs scenario in more favourable global conditions, estimated actual ODA still falls $27 billion short of predicted ODA needs of $174 billion by 2028. Over the period 2026–28, the cumulative gap under the lower-needs scenario is estimated to be approximately $84 billion. The fact that there is such a gap even under the lower-needs scenario reflects the substantive scale of ODA retreat by DAC states.

Comparing the needs-based ODA scenarios with actual ODA suggests that the decrease in ODA levels represent more than a temporary decline in development assistance. There is a structural and worrying withdrawal of development assistance by DAC donors.

V. Conclusions

There has been a broad, structural weakening of donor support through official development assistance that is partly obscured by fluctuations in crisis-related spending. Headline ODA figures obscure a longer-term decline in development assistance. The sharp increases in 2022–23 were driven in significant part by temporary crisis-related spending, particularly assistance to Ukraine and in-donor refugee costs. Excluding these components provides a clearer picture of the resources directed towards longer-term sustainable development, which remained more stable before declining sharply in 2024–25.

Notably, ODA has declined relative to rising global needs. The decline in ODA is more significant when assessed against current development and humanitarian needs rather than historical spending levels alone. Even before the recent cuts, ODA was insufficient to meet global needs. The needs-adjusted analysis presented here suggests a structural and worrying withdrawal of development assistance by DAC donors. The retreat of ODA needs to be understood in the context where it is not simply a return to historical levels seen 10–15 years ago, but instead the decline is both more dramatic and the funding gap more consequential if assessed relative to need for ODA. Donors are providing less at a time when the need for development and humanitarian assistance remains high and is potentially increasing.

Meanwhile, DAC states have presented ODA cuts as necessary in order to increase military expenditure, but the aid cuts make only a limited contribution. The simultaneous decline in ODA and increase in military expenditure among major DAC donors does not mean that aid reductions are a major source of financing for rearmament. The cases of Germany and the United Kingdom show that the savings generated by ODA cuts are small relative to the scale of planned increases in military spending and therefore make only a limited contribution to long-term military build-up plans. However, these cuts can carry significant development and humanitarian consequences.

Moreover, the effects of development and military spending occur on different timelines. Military spending, particularly major procurement programmes, can take years to translate into actual capabilities. In contrast, reductions in ODA can affect development and humanitarian programmes much more immediately, while also generating long-term consequences for sustainable development and human security. These different timescales complicate further comparisons based only on their financial impact.

Finally, it can be concluded that the relationship between development and security is undergoing an important shift. The recent divergence between ODA and military spending reflects an acceleration of a longer-term trend in how security is being framed. Rather than viewing security as a collective objective, in which improved economic, social and political conditions in recipient states can also contribute to international stability, many DAC states are increasingly viewing security in the traditional sense of military security and deterrence. While this shift is understandable given the perception of more immediate military threats, the consequence is a growing tension between different narratives of security and competing uses of limited public resources. Instead, development should be seen as a complement to military spending that is part of a more holistic security agenda.

Appendix A. Methodology for empirical analysis

A needs-adjusted ODA estimate of commitments by DAC states under various needs scenarios can be constructed to estimate the gap between available ODA funding and humanitarian and development needs. The estimate of needs-adjusted ODA measures what aid commitments might have looked like if they had continued to respond to global needs in line with historical donor patterns—albeit insufficient to meet global needs—observed before the recent weakening of donor commitment. As the analysis covers ODA from all DAC states, there is no comparable group against which to compare actual ODA commitments.⁵³ The analysis therefore uses a needs-based model to estimate how ODA commitments might have evolved if they had continued to respond to global needs.

An index of ODA needs

The first step is the creation of an index of needs for ODA. The index combines indicators that capture different dimensions of need based on 10 variables that have been identified in the literature as driving the need for ODA.

Three of the 10 variables relate to armed conflict. The number of armed conflicts (all types) as defined by UCDP ( X_conflicts,t in year t) captures the impact of conflicts involving the state. The number of battle-related deaths from non-state conflicts (X_{battle-related deaths,t}) attempts to account for the numerous communal and organized conflicts without state involvement. ACLED’s data on numbers of violent political events (X_{political violence,t}) is used to capture violent and non-violent actions on civilians that are potentially missed by UCDP’s armed conflict data sets.

Two variables, based on data from EM-DAT on the number of natural disasters per year (X_disasters,t) and the number of displaced people (X_{displacement,t}), capture the short-term humanitarian assistance and spillover effects resulting from unexpected natural disasters.

In addition, the World Risk Index (WRI) of the Institute for International Law of Peace and Armed Conflict (Institut für Friedenssicherungsrecht und Humanitäres Völkerrecht, IFHV) captures the geopolitical, environmental, disaster and other risks not already captured by the five UCDP, ACLED and EM-DAT variables (X_WRI,t). While including the WRI brings the limitation of double measurement and multicollinearity, the index is constructed using over 100 variables. The positive effects of capturing more missing information can thus be seen as outweighing the disadvantage of a small amount of double counting.

For structural dimensions such as economic development, health and education outcomes, the HDIs for the two subregions that have the lowest relative HDI—sub-Saharan Africa and South Asia ( X_HDI SSA,t and X_HDI SA,t, respectively)—stand as proxies for the regions or states that need ODA the most. The UNDP’s Human Development Index is comprised of the geometric mean of GNI per capita, life expectancy at birth, and expected and mean years of schooling. These variables are supposed to represent knowledge, a decent standard of living, and a long and healthy life, which when aggregated is human development.

Two major events that had a global impact over the period 2005–21—the Covid-19 pandemic (X_Covid,t) and the Syrian refugee crisis (X_Syria,t)—are represented by two additional, dummy variables. These take a value of 1 in years of occurrence and 0 otherwise.

The 10 variables are normalized by calculating the standard deviation of each and constructing a z score such that different contributing factors are on the same scale. One drawback of this method is that, by standardizing all factors onto a single scale, by construction, each variable will be given an equal weight. However, the index provides a consistent indicator of how much need for ODA exists based on events and developments that ODA has historically been used to address. The equation for the ODA needs index is as follows:

where

denotes the standardized value of each of the 10 variables j in year t, calculated using its mean X̄_j and standard deviation s_j over the period. The interpretation of the indicators is that, apart from the HDIs for sub-Saharan Africa and South Asia (i.e. X_HDI SSA,t and X_HDI SA,t), higher values represent higher needs.

The relationship between ODA needs and ODA commitments

The second step is to estimate the relationship between the ODA needs index and actual ODA by DAC states. The model uses data from 2005 to 2021, a period before the sharp reprioritization of donor budgets following Russia’s full-scale invasion of Ukraine and the subsequent increase in military expenditure among many DAC states. A small sample size in empirical analysis affects the statistical power of estimates and outliers may have a large impact on the results. However, in this case, the interest is not in the explanatory power or statistical significance of the variables but the fit of the model (or R squared). That is because the use of this index is to determine if the ODA needs index is closely related to actual ODA.

The model regresses ODA commitments on the ODA needs index and a time trend using ordinary least squares (OLS):⁵⁴

where ODA committed_t is total ODA committed in year t, ODA needs index_t is as defined above, and T represents a linear time trend. The intercept is α, while β is the coefficient representing the relationship between ODA committed and the ODA needs index after controlling for the time trend, and ε_t is the error term.

The fit of the model suggests that the needs index is a useful predictor of historical ODA commitments. The adjusted R squared is 0.79, indicating that the needs index and time trend explain a substantial share of the variation in ODA commitments over the period 2005–21. The needs index is also statistically significant at the 5 per cent level, suggesting that measured need was associated with DAC donor commitments during this period.

Using the results from the OLS regression that estimated the relationship between the ODA needs index and actual ODA commitments, ‘fitted values’ of ODA commitments can be estimated. These values are the levels of ODA predicted by the regression model (i.e. the ODA needs index) for each year observation. The values provide a benchmark for ODA that would be expected based on the level of measured need. This benchmark can be used to compare with actual ODA committed and forms the basis for a future projection of needs-based ODA level (see table 1 above).

Projections for 2026–28

The ODA needs index is projected for 2026 to 2028 under three scenarios which account for different paths of humanitarian crisis and development pressures. The general specification is

where N_t^s represents the ODA needs index in year t under scenario s, which can be baseline (B), high (H) or low (L); c_s is a constant annual addition to the needs-adjusted ODA index capturing the change in humanitarian and development needs, expressed in index points, and p_s is the proportion of the previous year’s index carried forward. In the baseline and higher-needs scenarios, all of the past year’s needs index is carried forward (i.e. p_B=1 and p_H=1) because the need for ODA is not improving but is instead remaining the same or getting worse. In the lower-needs scenario, the past year’s needs index is proportionally decreasing, with p_L=0.9816. This figure was estimated in the empirical analysis when variables such as the number of conflicts reduced, the war in Ukraine reached a ceasefire and the HDIs of sub-Saharan Africa and South Asia improved.

Additionally, for the lower-needs scenario, the constant annual addition to the needs index is zero (i.e. c_L=0). Together with a p_L that is less than 1, this results in the gradual decline of the needs-adjusted ODA index. This decline represents an assumed easing of developmental and humanitarian needs. In this scenario, the needs index declines by approximately 1.84 per cent annually (i.e. each year’s value is 98.16 per cent of the preceding year’s value). The decrease in the index in absolute terms will slow as a constant relative reduction of 1.84 per cent will result in annual smaller absolute reductions over time.

For the baseline case, c_B is estimated to be 0.28 based on the linear trend of the ODA needs index for the years 2022–25; it thus captures the economic, geopolitical and natural disaster developments in the four years preceding 2026. This means a constant increase in the index, capturing a gradual increase in needs for ODA that corresponds to gradual worsening in humanitarian and development terms.

The higher-needs scenario compares the average ODA needs index for 2019–21 with that for 2022–25 to capture the deterioration in crisis and development conditions. The latter period was marked by more conflicts, worsening development outcomes, increased displacement and the effects of the wars in Ukraine and Gaza. The difference between the averages over the two periods was converted into an annual increment of c_H=0.855 index points. This increment was added each year to the 2025 index value to project needs for 2026–28 if the pace of deterioration between the two periods continues. This worsening is much more pronounced than the baseline scenario (with c_B=0.28) and captures the continued and increasingly worsening geopolitical environment that translates to higher needs for ODA by recipient states.

Endnotes

1. 1 OECD, ‘ODA projections for 2026 and the near-term: Implications for vulnerable countries and sectors’, OECD Policy Brief no. 59, 19 June 2026; and UN Office for the Coordination of Humanitarian Affairs (OCHA), ‘Global Humanitarian Overview 2026: At a glance’, 8 Dec. 2025, pp. 2–4.
2. 2 OECD, States of Fragility 2025 (OECD: Paris, 2025), pp. 13–14.
3. 3 Lopes da Silva, D. et al., ‘Military expenditure’, and Scarazzato, L. et al., ‘Arms production and military services’, SIPRI Yearbook 2026: Armaments, Disarmament and International Security (Oxford University Press: Oxford, 2026).
4. 4 OECD, Development Co-operation Directorate, ‘Preliminary official development assistance levels in 2025: Detailed summary note’, DCD(2026)8, 15 June 2026. All US dollar figures are in current US dollars unless otherwise stated (the exceptions being in the subsection on projecting needs-based ODA). All year-on-year comparisons are in real terms (i.e. taking account of inflation), based on 2024 constant prices.
5. 5 UN General Assembly Resolution 2626 (XXV), ‘International development strategy for the second United Nations Development Decade’, 24 Oct. 1970, para. 43; and OECD, ‘Hitting the 0.7% ODA/GNI target while protecting funding commitments’, 28 June 2021.
6. 6 Schiano Lomoriello, R. and Scott, R., ‘What official development assistance went to fragile contexts?’, OECD, States of Fragility 2018 (OECD: Paris, 2018).
7. 7 OECD (note 1); and UN Office for the Coordination of Humanitarian Affairs (note 1).
8. 8 OECD, ‘The latest 2026 ODA projections reveal aid for health is further at risk’, 29 June 2026.
9. 9 Ahmad, A. S., Trebosc, C. and Tarif, K., ‘Aid cuts and security risks: From health system disruption to human insecurity’, SIPRI Research Policy Paper, June 2026.
10. 10 da Silva A. F. et al., ‘Impact of two decades of humanitarian and development assistance and the projected mortality consequences of current defunding to 2030: Retrospective evaluation and forecasting analysis’, The Lancet Global Health, vol. 14, no. 5 (May 2026).
11. 11 OECD (note 1).
12. 12 Cattaneo, O., ‘Understanding and overcoming the aid crisis’, Elcano Royal Institute, 30 June 2025.
13. 13 Lopes da Silva et al. (note 3).
14. 14 UN General Assembly Resolution 70/1, ‘Transforming our world: The 2030 Agenda for Sustainable Development’, 25 Sep. 2015.
15. 15 Lopes da Silva et al. (note 3); and Scarazzato et al. (note 3).
16. 16 Kumar, C. et al., The Case for Development in 2025: Exploring New Narratives for Aid in the Context of the EU’s New Strategic Agenda (ODI Europe: Brussels, June 2025), pp. X–XIV; and Lazell, M. and Petrikova, I., ‘Is development aid securitized? Evidence from a cross-country examination of aid commitments’, Development Policy Review, vol. 38, no. 3 (May 2020).
17. 17 Lazell and Petrikova (note 16); and Olivié, I. and Pérez, A., ‘Whose and what aid securitisation? An analysis of EU aid narratives and flows’, Third World Quarterly, vol. 42, no. 8 (2021).
18. 18 Langlotz, S. and Potrafke, N., ‘Does development aid increase military expenditure’, Journal of Comparative Economics, vol. 47, no. 3 (Sep. 2019); Round, J. I. and Odedokun, M., ‘Aid effort and its determinants’, International Review of Economics & Finance, vol. 13, no. 3 (2004); and Mo Ibrahim Foundation, ‘Demystifying Africa’s dependence on foreign aid’, Research brief, Oct. 2025, p. 9.
19. 19 Drescher, S., ‘Verzerrte Quote’ [Distorted figures], Welt-Sichten, 13 Apr. 2016; and OECD, ‘Development co-operation profiles: Germany’, 16 June 2026.
20. 20 Fürstenau, M., ‘Germany slashes development aid to boost defense spending’, DW, 30 June 2025.
21. 21 ‘Germany publishes 2026 draft budget, signals decline in ODA’, Donor Tracker, SEEK Development, 30 July 2025.
22. 22 Nöstlinger, N., ‘Germany ramps up defense spending in new budget plan ahead of NATO summit’, Politico, 6 July 2026.
23. 23 Siebold, S., ‘Germany looking into buying eight additional F-35 jets’, 7 June 2024, Reuters; ‘Germany and Israel sign an agreement for Berlin to buy a US–Israeli missile defense system’, Associated Press, 28 Sep. 2023; and Rheinmetall, ‘Major order for the German army technology industry: Bundeswehr orders fifty more Puma infantry fighting vehicles’, 15 May 2023.
24. 24 Kiel Institute for the World Economy, ‘Big spending, little future: Germany is rearming differently from Poland and the UK’, 12 May 2026.
25. 25 Mercy Corps, ‘When aid disappears: The human cost of foreign aid cuts’, 1 Apr. 2025.
26. 26 British Foreign, Commonwealth and Development Office, ‘Changes to the UK’s aid budget in the spending review’, 26 Nov. 2020.
27. 27 Starmer, K., Statement to the House of Commons, 25 Feb. 2025.
28. 28 British Foreign, Commonwealth and Development Office, ‘Main estimates memorandum for the Foreign, Commonwealth and Development Office 2025–26’, 9 May 2025.
29. 29 British Foreign, Commonwealth and Development Office (note 28), section 1.8.
30. 30 E.g. Heinen, E., Dutch minister of finance, Transcript of speech at the presentation of the 2025 budget memorandum, 18 Sep. 2024; and Finnish Ministry for Foreign Affairs, ‘Budget proposal for the Ministry for Foreign Affairs and its branch of government for 2026’, 23 Sep. 2025.
31. 31 Welch, Z. and Laub, K., ‘The impact of defense spending on ODA: Outlook and trends’, Donor Tracker, SEEK Development, 27 May 2025.
32. 32 Kharas, H., ‘Towards purpose-driven official development assistance’, eds M. Ahmed, R. Calleja and P. Jacquet, The Future of Official Development Assistance: Incremental Improvements or Radical Reform? (Center for Global Development: Washington, DC, 2025).
33. 33 Rustad, S. A., Conflict Trends: A Global Overview, 1946–2025 (Peace Research Institute Oslo: Oslo, 2026), pp. 12–13.
34. 34 Vesco, P. et al., ‘The impacts of armed conflict on human development: A review of the literature’, World Development, vol. 187 (Mar. 2025).
35. 35 Chowdhury, A. R. and Murshed, S. M., ‘Conflict and fiscal capacity’, Defence and Peace Economics, vol. 27, no. 5 (Oct. 2016).
36. 36 OECD (note 2), p. 213.
37. 37 Chrimes, T. et al., The Great Reversal: Prospects, Risks, and Policies in International Development Association (IDA) Countries (World Bank: Washington, DC, 2024); and Helth Lønborg, J. et al., ‘September 2025 global poverty update from the World Bank: New data and regional classifications’, World Bank Data Blog, 30 Sep. 2025.
38. 38 Helth Lønborg et al. (note 37).
39. 39 Cotte Poveda, A. and Martinez Carvajal, J., ‘Economic development, political violence and socio-political instability in Colombia: An econometric analysis using panel data’, European Research Studies Journal, vol. 22, no. 1 (2019).
40. 40 On the expansion of ACLED’s geographical coverage over time, which requires caution in historical comparisons, see Armed Conflict Location and Event Data (ACLED), ‘Country/time-period coverage’, [n.d.].
41. 41 Armed Conflict Location and Event Data (ACLED), ‘ACLED codebook’, [n.d.].
42. 42 Armed Conflict Location and Event Data (ACLED), ‘Number of political violence events by country-year’, ACLED data, [n.d.].
43. 43 OECD, Peace and Official Development Assistance, OECD Development Perspectives no. 37 (OECD: Paris, Oct. 2023), pp. 9–21; and Ioannides, I., Peace and Security in 2019: Evaluating EU Efforts to Support Peace in Colombia (European Parliament, European Parliamentary Research Service: Brussels, May 2019).
44. 44 UN Economic and Social Commission for Asia and the Pacific (ESCAP), ‘The uncertain future of ODA in the Pacific: Why it matters more than ever’, 12 Feb. 2026.
45. 45 University of Louvain, Centre for Research on the Epidemiology of Disasters (CRED), EM-DAT. Since the annual number of natural disasters can fluctuate considerably, a 3-year moving average is used to provide a more stable measure of the trend. Improvements in EM-DAT’s disaster reporting over time, especially for smaller-scale events in lower-income countries, mean that there may be inconsistencies between the numbers for the two periods.
46. 46 Arezki, R. et al., ‘Natural disasters and foreign aid allocation: How much, how quickly, and to whom?’, World Development, vol. 203 (July 2026).
47. 47 UN Development Programme (UNDP) and Oxford Poverty & Human Development Initiative (OPHI), 2025 Global Multidimensional Poverty Index (MPI)—Overlapping Hardships: Poverty and Climate Hazards (UNDP: New York, 2025).
48. 48 World Bank, Poverty, Prosperity, and Planet Report 2024: Pathways out of the Polycrisis (World Bank: Washington, DC, 2024), p. 49.
49. 49 UN Development Programme (UNDP), Human Development Report 2025—A Matter of Choice: People and Possibilities in the Age of AI (UNDP: New York, 2025), pp. 4–5.
50. 50 OECD (note 1). Inflation forecasts are based on OECD projections. All predicted figures in this subsection and in table 1 are in constant 2024 US prices based on OECD inflation forecasts.
51. 51 OECD (note 1).
52. 52 OECD, ‘DAC1: Flows by provider’, OECD Data Explorer, 10 July 2026.
53. 53 Abadie, A., Diamond, A. and Hainmueller, J., ‘Synthetic control methods for comparative case studies: Estimating the effect of California’s tobacco control program’, Journal of the American Statistical Association, vol. 105, no. 490 (June 2010); and Bove, V., Elia, L. and Smith, R. P., ‘The relationship between panel and synthetic control estimators of the effect of civil war’, Birkbeck Centre for Applied Macroeconomics (BCAM) Working Papers no. 1406, Oct. 2014.
54. 54 Wooldridge, J. M., Introductory Econometrics: A Modern Approach, 6th edn (Cengage Learning: Boston, MA, 2016), pp. 19–104.