Shifting official development assistance during COVID-19: earmarking, donor concentration and loans

Introduction

COVID-19 has had a profound impact on the need for official development assistance (ODA), including humanitarian aid. On the demand side, recipient health systems faced an increased burden globally, as they fought the pandemic and worked to maintain routine health services; requiring significant domestic and alternative funding streams. A consequence has been reversed progress on the number of countries graduating from low- to middle-income status. On the supply side, donors also managed the pandemic ‘at home’, stretching wealthier countries’ disbursement portfolios. Taken together, there is significant potential for ODA changes to disrupt the achievement of national and global universal health coverage (UHC) agendas by 2030, the Sustainable Development Goals (SDGs) and broader health system strengthening efforts. In line with this concern, data compiled by the International Aid Transparency Initiative suggest overall bilateral donor disbursements (those provided directly from donor countries to recipients) remained relatively constant or fell during the pandemic. However, in contrast, multilateral aid (ODA pooled and distributed by entities like the World Bank and United Nations (UN) agencies) increased in both relative and absolute terms.1 This may represent a positive shift: multilateral agencies offer a unique vantage for coordination and prioritisation of aid that transcends national priorities.

In 2016, the Overseas Development Institute characterised bilateral channels as highly politicised compared with multilateral; leading recipients to prefer multilateral aid.2 Multilaterals can, in theory, focus on cross-national development outcomes, while bilateral aid is often seen as a mechanism to achieve donors’ strategic interests.3 4 Moreover, due to pooling functions of multilaterals, ODA dispersed through these entities may be less fragmented and better structured to achieve common global causes or support international governing frameworks, such as the implementation of International Health Regulations.5 6 Evidence suggests multilateral aid is more efficient in reaching stated objectives7 and following Organisation for Economic Co-operation and Development (OECD) ‘best practice’ guidelines; including serving recipient needs as opposed to donor priorities.8 Consequently, the data on a shifting ODA landscape during COVID-19 were promising: a higher share of ODA disbursements flowing through multilaterals may be uniquely relevant in addressing global threats and achieving cross-national global agendas. Yet, the pandemic revealed an aid landscape of donor agencies that struggle with coordination in fundamental ways: conflicting prioritisation, delays in disbursement and a failure of multilateralism with regard to global public goods and vaccines.9 10 An increase in multilateral expenditure, through actors whose disbursement practices are not subject to more democratic processes, therefore raises questions regarding global health governance, multilateral power and influence in the wake of the pandemic. For example, voting at the World Bank Group (WBG) is often characterised as ‘one dollar, one vote’, with influence concentrated among countries most able to pay.

The increase in multilateral ODA also coincides with an unprecedented surge in public debt.11 To expand health provision during the pandemic while mitigating long-term damage to the economy and facilitating recovery, governments pursued ‘expansionist’ fiscal policies, leading to budget deficits and compounding debt stress.12 In addition, the debt-to-gross domestic product ratio in developing countries is expected to rise, threatening the well-being of individuals. In light of this, the Government of Barbados has called for liquidity to stop the debt crisis, asking the G20 to agree to an ambitious Debt Service Suspension Initiative that includes all multilateral development bank (MDB) loans.13 Debt-related strains may compound broader social fractures surrounding the unequal burden associated with post-COVID recovery.12 Thus, the potential of multilateral entities as a neutral third party through which ODA can be distributed is increasingly relevant as countries strive to achieve non-pandemic-related health goals.14 Despite these concerns, we lack a critical examination of ODA from multilaterals and how this type of aid has shifted in light of COVID-19. We hypothesise that while there has been a shift towards multilateral financing, functionally this transition may mimic issues of a bilateral-dominated market: reaffirming, rather than addressing, issues of donor influence.

We address three primary research questions motivated by recent increases in multilateral ODA: (1) How is earmarking borne out through different multilateral agencies and for different types of aid? (2) We then examine market concentration—how diverse is the ODA landscape, to what extent is the multilateral system concentrated and has this changed over time? (3) Finally, implications for financing allocation vehicles—what does an increase in multilateral ODA mean for financing modalities used, and the potential for low- and middle-income country debt sustainability? Specifically, we look at ODA repayment expectations and distribution of ODA outflows in the form of grants versus loans. Accordingly, we consider what these might mean for donor influence across the global health landscape.

Background

This section outlines a contextual framework for examining aid contingencies, concentration and allocation. The framework extends beyond a binary distinction of ‘bilateral’ versus ‘multilateral’ by focusing on the aid market and practical attributes of aid itself. This diverges from prior examinations that centre donor motivation (eg, ideational, political, commercial, security related or cultural) or recipient capacity.15

Methods

Results

Looking at multilateral entities providing ODA, we first examine the amount each multilateral disbursed in total ODA, showing the top five multilaterals across three categories: UN, non-UN multilaterals and RDBs (table 2). Examining trends in ODA over time, in 2020 US$ constant prices, we see high variability from year to year (not shown) but a steady increase in multilateral ODA over time. We observe a $31 425 million increase in multilateral ODA (gross disbursements) from 2011 to 2021 across all three categories. Non-UN multilaterals contributed most both in absolute and relative terms: with approximately $41 000 million disbursed in 2021 (+20 273 million, from 20 953 to 41 226). UN multilaterals and RDBs both disbursed approximately $5000 million in 2021, with RDBs representing a more significant increase ($2899 million) from 2011. Of RDBs, the ADB disbursed the most at $3049 million in 2021, an order of magnitude lower than the highest non-UN multilateral, the WBG at $29 913 million.

We find that the earmarked share of contributions to multilateral entities has been rising steadily over the past decade (figure 1), whereas core contributions have remained constant. General, core multilateral contributions (based on gross disbursements) increased by approximately 7% (+7.3%, from 29.9% to 37.7%). This aligns with a broader increase in overall multilateral inflows from just over $50 billion in 2012 to nearly $80 billion in 2020; with most of the increase composed of earmarked funds. This increase, and the concurrent increase in the share of multilateral contributions earmarked at the global or regional level, may reflect an increase in global financing mechanisms in the form of managed trusts, such as the COVID-19 Response and Recovery Trust Fund.11

Request permission

In line within findings from the OECD, we see that the share of DAC members’ non-core (earmarked) contributions in their total ODA is channelled through the multilateral development system and earmarked for specific objectives; core functions are receiving less funding proportionally.11 Looking at this by UN entity (figure 2), we see that the WHO, the UN agency tasked with the directing and coordinating authority on international health, has one of the highest rates of earmarked inflow (74%). While the WHO has a large average annual financing volume as compared with other UN agencies, we observe a positive and statistically significant relationship between financing volume and the share of inflow earmarked (R²=0.272, p<0.05).

Request permission

Looking at how inflow is earmarked (table 3), we see an increase in earmarking for social programming (online supplemental table 2) but decreases in humanitarian ODA and ODA related to governance. The social sector, which comprises health and social protection measures, had an increase in earmarked contributions from DAC members between 2019 and 2020 (+5.6%, from 18.5% to 24.1%). This increase was largely driven by a rise in contributions earmarked for health and other social infrastructure. Conversely, the focus on COVID-19 appears to have crowded out financing to some other health-related issues, such as disaster prevention (humanitarian) and rural development.

Regarding multilateral market concentration, we observe an increase in total multilateral actors from 2011 to 2021 (table 4). This corresponded with an increase in both measures of diversity: an increase in the Shannon-Weiner Function (+0.422, from 3.258 to 3.680) and the Gini-Simpson Index (+0.012, from 0.962 to 0.974), indicating increased market diversity. However, we find the HHI for the multilateral market increased by 391 from 2011 to 2021 (+391, from 1132 to 1523). This suggests higher market concentration in 2021 than prior to the onset of the COVID-19 pandemic, with the WBG making up 37.4% of all funding disbursed (compared with 32.5% in 2011). When including trusts as part of their managing entities (model II), the HHI is higher in every year with a more pronounced increase from 2011 to 2021 (+1366, from 2756 to 4122). Increases in the HHI indicate a decrease in competition and an increase in market power. The HHI results for model I suggest the ODA multilateral marketplace would be categorised as an ‘oligopoly’ or moderately concentrated with a small number of actors all serving as influential players. However, when accounting for trusts, it is a highly concentrated market (all values over 2500). After the WBG, the donor entity with the second highest market share in 2021 (excluding trusts) was the IMF at 5.0%, whereas RDBs made up only 7.0% collectively, of which the ADB disbursed the highest absolute amount of ODA with a market share of 3.8%.

In examining trends in grants versus loans (figure 3), we find a decrease in grants as a share of all ODA commitments from multilaterals at 54% in 2020 (−14%, from 68% to 54%). This trend coincides with a rise in non-UN multilateral agencies’ share of the multilateral marketplace. However, in looking more granularly at funding from the WBG, we see an increase in the share of ODA provided as grants for social sector programming, which includes health and population-related ODA, over time (+16%, from 19% to 35%).

Request permission

The increase in multilateral ODA in the form of loans that we observe coincides with a growing number of low-income countries that are at risk of debt distress. Rates of indebtedness are unprecedentedly high for non-conflict periods with 90% of low-income countries designated as moderate, high or distressed regarding risk of debt distress (online supplemental figure 1).

This work is subject to several limitations. First, we do not attempt to make causal claims or attribute changes in ODA to the pandemic. As the OECD has noted, that the pandemic period did not correspond with a more significant deviation from previous trends suggests the existence of rigidities and path dependencies that constrain the ability of DAC members to swiftly adjust and balance ODA allocations. Second, the functional attributes of ODA that we examine are inherently limited by data availability. For example, while earmarking is one mechanism to exert donor prioritisation, as demonstrated by Khan et al, donors influence recipient counties at multiple stages within health policy processes, such as priority setting, policy formulation, policy implementation and monitoring and evaluation.16 Even when reducing conditionality, donors may employ alternative strategies to influence national policy.34 As a result, donor priority setting must be understood within the context of more entrenched norms of donor–recipient relations.16 Third, the data we use for this paper are highly aggregated at the national-annual level. While this allows for cross-donor comparison—is also means we cannot examine more granular issues, that is, allocative fairness of earmarked ODA within a country. Fourth, there is limited, or incomplete, information on trusts and we may undercount the role of these funds.32 Finally, China is not a member of the DAC and does not report into the OECD-DAC or other international aid repositories.35 However, China does hold approximately 6% of shares of the ADB and just short of 27% of shares for the Asian Infrastructure Investment Bank.36

Discussion

Aid disbursement from multilateral institutions increased significantly at the onset of COVID-19. Yet, functionally, we find ODA from multilaterals is mimicking the attributes of bilateral aid. This was borne out in three important, although not comprehensive, areas during COVID-19: (1) an increasing prevalence of earmarking of aid inflows to the multilateral system; (2) higher concentration of the donor ‘market’ driven by the WBG; and (3) a rise in loan-based disbursements. While the absolute number of multilateral actors has been increasing since 2012, and in turn market diversification, there has been a concurrent consolidation led by the WBG at 37% of market share in 2021; and just under 50% if we consider trusts held by the WBG. Consolidation may streamline revenue pooling and, in theory, allow for a more collective approach to achieving global development goals, but prioritisation exercised through increased earmarking also has the potential to sideline counties’ democratic manifesto pledges and orient ODA towards donor-driven priorities.

While we explore the role of earmarking as it relates to donor control,37 we recognise the use cases for earmarked funding. For example, earmarking can be used to protect high-priority programmes or issues from budget cuts, or shifting political content—such as support for HIV/AIDS-related programming.17 28 In addition, this may generate support from donors that would otherwise not be allocated to ODA, the proceeds of which can be used for broader system-wide health provision and strengthening, such as occurs in the Global Polio Eradication Initiative. However, the overarching argument against excessive earmarking is that it explicitly removes power from recipient countries to control their own health and/or development programmes.17 Moreover, donor-driven initiatives through earmarking contributions tend to prioritise vertical interventions that can be easily measured using standard performance metrics, often in contrast to more horizontal, system-based or rights-based approaches to health.38 This may include the number of bed nets delivered or antiretrovirals administered at the cost of broadened system-strengthening health activities, such as workforce capacity building.38

In addition, if we think of earmarked contributions to multilaterals as eclipsing core contributions (ie, zero-sum, at least in relative terms), this could lead to a gradual erosion of critical functions relevant to global health governance. Providing strategic and long-term oversight of key reforms requires cross-national coordination and adapting to the evolving and expanding nature of global development challenges.14 This is all too real for the WHO, which as an institution only controls approximately 20% of their budget. The remaining ~80% comes in the form of earmarked funding from a small number of state and non-state donors, with donor-determined projects resulting in a contracting-like dynamic. The result has been described as a vicious cycle: if countries do not trust the WHO to effectively deliver on core mandates, they increasingly turn to earmarked contributions where they can further national health-related development goals.39 Recent reforms at WHO have sought to increase the assessed contributions of WHO both in real terms, and as part of their overall budget share.40 While these have been agreed in principle, it is unclear whether member states will adhere, given historical actual financing deficits, and how this might alter earmarking.41 While the OECD has recently sought to standardise accounting practices for earmarked contributions, there is a pressing need to be candid regarding impact on institutional authority.28 In addition, the measures used in this piece are highly aggregated. In reality, there is a more complex landscape of aid contingencies with inflexibility (eg, directly delivered technical assistance) to more flexible allocation (eg, budget support).17

The argument for multilateralism has often centred liberal democratic principles of governance, including transparency, representation and participation. Yet, the growing prevalence of earmarking in multilateral aid suggests a less rigid distinction between bilateral and multilateral ODA, the major concern (in both cases) is that someone other than the recipient is deciding how aid should be allocated. Bird and Aninat categorise concentration in the following forms: (1) a lower number of total donors or (2) an increasing share of total donation amounts controlled by fewer donors.22 We find there has been an increase in the absolute number of official donors over time, as evidenced by the proliferation of actors in global health,42 43 but a simultaneously more concentrated marketplace in terms of the volume of aid controlled by fewer actors. This could be viewed as the worst of both scenarios: as the number of entities increases, each controls a very small share of the market, leading to fragmentation of financing, governance and authority; and simultaneously increasing concentration provides large funders with undue influence in public goods creation.22 This is particularly notable for the WBG, who has a controversial history in global health provision, and whose governance is linked to that of the US government.34 44 Indeed, unlike the governance and decision-making model in the World Health Assembly, based on one state, one vote, decision-making in the WBG is determined by financial contribution. In addition, our estimates regarding concentration are more stark when taking into account the WBG’s trusteeship of entities, such as the Global Fund and GAVI the Vaccine Alliance.32 These relationships may provide a veneer of diversification across the sector, but in practice demonstrate greater consolidation. The WBG helped set up both institutions and remains a major financial supporter, serving as board member for both GAVI and the Global Fund, as well as fiduciary agent, financial and treasury manager of GAVI. As such, the WBG maintains considerable influence over the activities of these institutions.

On the surface, increased ODA disbursements from multilateral actors are something to be lauded during a pandemic: meaningful commitments for managing transnational health issues. Yet, the data demonstrate an outsized, and increasing, role of certain actors, which merits re-examination after pandemic. A stark example is the increasing role of the WBG in global health governance in contrast to the WHO, which is the institution mandated to direct and coordinate international health work. Principal-agent theory is one lens through which to consider the WBG’s increasing market share: principals may be less willing to give power to multilateral agents they deem ineffective and in turn take steps to maintain control of their ODA funding or target funding to agents deemed effective, for example, the WBG and its trustee institutions. While allocation to trusted agents is rational, these data come midst broader efforts to decolonise global health motivated by complex and long-standing power dynamics. Despite these efforts, trends in ODA suggest development and health governance increasingly sit with the WBG and in turn the USA.45 The perceived effectiveness of GAVI and the Global Fund may incentivise national investments that would otherwise not be allocated towards ODA. However, consolidation more broadly demonstrates the concentration of power within global health governance.46 States that do engage in ODA increasingly appear to do so when they are able to dictate terms (ie, earmarking), potentially due to the ability to demonstrate meaningful results for their own electorates.

Moreover, WBG dominance specifically raises specific concerns for internal governance and debt. First, WBG has historically approached health from a development perspective, whereby health is largely seen as a mechanism for achieving production value within labour markets.47 Health may, in turn, be treated as a commodity, rather than a right, informing concepts such as human capital, cost-effectiveness and disability-adjusted life-years,48 each of which ascribe value to health as it relates to labour force contribution.49 Second, the governance model employed within the WBG challenges accepted liberal democratic norms of governance: voting at the WBG is not one state, one vote as seen elsewhere in the global health architecture, but determined by volume of financial contribution.44 The US government has the single largest voting share, can veto any major decision and always appoints the president of WBG through an unwritten agreement. For many, this makes WBG a proxy of the US government, challenging independent health and development policy where the most prolific principal actor is in effect simultaneously a proxy agent.50 Third, WBG has been widely criticised for its role in global health programmes over the last half century. For example, structural adjustment programmes have ‘set back healthcare in low and middle income countries more in one decade than anything states themselves could have done since colonialism’.51 Compounding these issues is an enduring and historically contentious issue regarding how aid is distributed. Reliance on loans means that many countries entered the pandemic in debt. While the WBG does adjust future financing from loans to grants for countries at high risk of debt distress (and an increasing share of the WBG’s social portfolio distributed in grants) offers some promise, this approach will be tested as more countries enter debt distress (online supplemental table 3).52

Conclusion

COVID-19 confirmed the significant role of multilateral ODA in times of global crisis. However, the data presented in this paper highlight the importance of looking beyond aggregate disbursements and examining the changing nature of ODA: an increase in earmarking, decrease in grant-based disbursements and a larger, though more concentrated, donor landscape. These findings raise important questions regarding global health governance and the ability to respond to global crises. For example, many governments departed from their financial and legal obligations to UN institutions.53 After pandemic, we have the opportunity to re-examine these systems and the extent to which they align with the promise of multilateral development assistance: a coherent, and representative, set of organisations that can mobilise on shared goals. However, this requires clear examination of how the current system operates, for whom and why.