Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Resource allocation processes at multilateral organizations working in global health

Resource allocation processes at multilateral organizations working in global health International institutions provide well over US$10 billion in development assistance for health (DAH) annually and between 1990 and 2014, DAH disbursements totaled $458 billion but how do they decide who gets what, and for what purpose? In this article, we explore how allocation decisions were made by the nine convening agencies of the Equitable Access Initiative. We provide clear, plain language de- scriptions of the complete process from resource mobilization to allocation for the nine multilateral agencies with prominent agendas in global health. Then, through a comparative analysis we illumin- ate the choices and strategies employed in the nine international institutions. We find that resource al- location in all reviewed institutions follow a similar pattern, which we categorized in a framework of five steps: strategy definition, resource mobilization, eligibility of countries, support type and funds al- location. All the reviewed institutions generate resource allocation decisions through well-structured and fairly complex processes. Variations in those processes seem to reflect differences in institutional principles and goals. However, these processes have serious shortcomings. Technical problems in- clude inadequate flexibility to account for or meet country needs. Although aid effectiveness and value for money are commonly referenced, we find that neither performance nor impact is a major cri- terion for allocating resources. We found very little formal consideration of the incentives generated by allocation choices. Political issues include non-transparent influence on allocation processes by donors and bureaucrats, and the common practice of earmarking funds to bypass the normal alloca- tion process entirely. Ethical deficiencies include low accountability and transparency at international institutions, and limited participation by affected citizens or their representatives. We find that recipi- ent countries have low influence on allocation processes themselves, although within these processes they have some influence in relatively narrow areas. Keywords: Aid, international health policy, resource allocation, global health Key Messages Extensive review of resource allocation at multilateral organisations working on global health finds the existence of well-structured and organized systems. Core principles of each institution are reflected in their design choices of allocation cycles (e.g. emphasis on fairness, health needs, efficiency). Although processes are documented, confidential or non-transparent elements such as qualitative adjustments and ear- marking agreements undermine the transparency and legitimacy of allocation decisions and restrict the knowledge and participation of recipient countries and their citizens. V The Author(s) 2018. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. i4 Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i5 Introduction analysed where DAH goes. Long time-series data have been used to analyse the patterns of DAH, but the results of those studies are not International institutions provide well over US$10 billion in devel- conclusive. For instance, Piva and Dodd (2009), Hotez et al. (2014) opment assistance for health (DAH) annually, and between 1990 and Dieleman et al. (2015) find wide variation in funding across and 2014 DAH disbursements totalled $458 billion (IHME 2014)— regions and disease areas that are not explained by differences in but how do they decide who gets what, and for what purpose? The disease burden or income. Estimation problems stemming from importance of these questions is difficult to overstate for the six bil- model misspecification, unobserved variables and measurement lion people in Low- and Middle-Income Countries (LMIC) whose problems have constrained empirical approaches (McGillivray health can be directly affected by DAH. Allocation decisions affect 2003; Hoeffler and Outram 2008). Others have discussed the pol- all three pillars of public health: the political, the technical, and the itics of DAH, particularly the haphazard process in which issues rise ethical (Roberts et al., 2003). Politically, allocation processes deter- and fall on the global health agenda (Shiffman & Smith, 2007; mine which countries get what assistance, raising distributional Hafner & Shiffman, 2012; Parkhurst & Vulimiri, 2013; Reich, issues at every level from the international down to individuals. 1995; Bump, Reich, & Johnson, 2013). Technical matters include the choice of objectives, for instance the Aid effectiveness has attracted lively commentary, attention suf- diseases to be addressed and the strategies that will be employed, the ficient to produce The Paris Declaration on Aid Effectiveness population to be targeted, and the effectiveness of interventions and (2005), and a robust literature, if not yet a consensus answer programs. Ethical considerations start with the fairness of the allo- (Bourguignon & Sundberg, 2007; Easterly, 2008; Piva and Dodd cation processes, and continue through the equity of their outcomes 2009; Hristos and Paldam 2009; Clemens et al. 2011). Adequate and consequences. and transparent resource allocation systems are often mentioned as The Equitable Access Initiative (EAI) was designed to explore desirable, for instance as Point 17 in the Paris Declaration, but as far some of these issues—primarily the technical aspects. The nine con- as we were able to detect, only Ottersen et al. (2014) have directly vening international agencies were particularly concerned with addressed the question of allocation criteria. Their work details the understanding the consequences of using gross national income criteria in allocation formulae in 10 bilateral and 5 multilateral insti- (GNI) per capita as a primary indicator of need, and interested in tutions working in global health and discusses their distributional exploring alternatives. Although GNI had long been used for that impact. Our aims differ in that we seek to understand the complete purpose, they noted that rising inequality means that most of the decision-making process leading to allocation decisions. Moreover, world’s poor now live in middle-income countries. Hence, the EAI unlike previous works, we supplemented a review of literature with was convened to investigate how processes might better reflect dis- interviews at each institution. ease burdens, national capacity to intervene, government health There is a very large literature on distributive justice and other budgets and other factors (EAI 2015a, b) primarily the technical aspects of ethics applied to global health. Much of this has centred aspects such as classifying country needs and capacities in health. on priority setting, which can be taken loosely as a synonym for al- Under the aegis of the EAI, academic teams were engaged to de- location in this context. Scholars have emphasized ‘accountability velop fresh thinking on classification methods for capturing country for reasonableness’ in such decisions (Daniels 2000, 2007), the im- needs and capacities in health, the products of which were the basis portance of ‘making fair choices’ of what to provide to whom in for papers in this special issue. These colleagues analysed the deter- health (Ottersen and Norheim 2014), and the centrality of ethics in minants of health outcomes to explore the value of GNI per capita all public health rationing decisions (Roberts and Reich 2002). Our as a proxy (Sterck et al.), modelled government ability to spend on inquiry sheds light on an important aspect of priority setting—how health under various conditions over time (Haakenstad et al.) and international agencies allocate DAH. conducted experiments to discover the values that participants felt We organize this article as follows. Section 2 The following section should be prioritized in allocation (Grepin et al.) and then applied presents our methods and data. Section 3 contains a descriptive table these ideas to develop new country rankings (Ottersen et al.). of the allocation process in each institution and summarizes our com- As an associated activity, we explored how allocation decisions parative analysis. Section 4 provides discussion and conclusions. were made by the EAI’s nine convening agencies themselves—Gavi, the Vaccine Alliance (hereafter Gavi), The Global Fund to fight AIDS, Tuberculosis, and Malaria (GFATM), The United Nations Methods and data Population Fund (UNFPA), The United Nations Children’s Fund (UNICEF), The United Nations Development Programme (UNDP), Institutions for analysis The World Bank Group’s International Development Agency, The Since our project was linked to the EAI, we focussed our analysis on World Health Organization (WHO), UNAIDS and UNITAID. We its convening agencies—Gavi, GFATM, UNFPA, UNICEF, UNDP, felt that examining allocation practices at these prominent institu- the World Bank’s IDA, WHO, UNAIDS (meaning the UNAIDS tions would be helpful to discussions about the future of allocation Secretariat) and UNITAID. Each of these institutions has significant and of great interest to recipient countries and their citizens. objectives in global health and a worldwide mandate, as we sum- To support our larger objective of fostering greater transparency, marize in Table 1. At present there are no other large international more informed discussion, and better allocation, we provide clear, institutions meeting these two criteria. Regional development banks, plain language descriptions of the complete process from resource bilateral agencies and foundations are also important objects of mobilization to allocation for each multilateral. Then, through a study, but these were not considered in the EAI and were beyond the comparative analysis we illuminate the choices and strategies em- scope of this article. Collectively, the EAI convening agencies pro- ployed in the nine international institutions. vided US $11.7 billion in DAH in 2014 (UNITAID 2013; Institute Our investigation of allocation methods at international agencies for Health Metrics and Evaluation 2014). In that year, the largest connects to several important literatures in global health, including sources were GFATM, WHO and Gavi (respectively US $4.1 billion, DAH and allocation, aid effectiveness and ethics. Some have US$2.1 billion and US$1.8 billion). Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i6 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 1. Institutional objectives Definition of the strategy Institution Objectives Gavi Save children’s lives and protecting people’s health by increasing access to immunisation in poor countries Allocation of Resource GFATM Accelerate the end of HIV/AIDS, tuberculosis, and mal- funds mobilization aria as epidemics UNAIDS Formulate a plan of action for all the institutions work- ing on HIV/AIDS to end the epidemic by 2030 UNDP Help eradicate poverty and reduce inequalities and exclusion Type of Eligibility of UNFPA Deliver a world where every pregnancy is wanted, every support countries birth is safe, and every young person’s potential is fulfilled UNICEF Promote the rights and wellbeing of every child Figure 1. Allocation cycle in multilateral organizations UNITAID Contribute to scale-up access to treatment for HIV/ AIDS, malaria, and TB for people in developing allocations. Step 3 is an eligibility determination; step 4 (type of sup- countries by leveraging price reductions of quality port) is the determination of what funds, services, in kind-support or drugs and diagnosis other resources will be made available; step 5 is the allocation of WHO The attainment of the highest possible standard of specific resources to specific recipients, such as programmes or health by all people. countries. Based on our document review and interview transcripts, World Bank End extreme poverty by decreasing the percentage of we categorized the collected information following this framework people living on less than $1.25 a day to less than 3%; promote shared prosperity by fostering the in- of analysis. For each step, we documented the institutional mechan- come growth of the bottom 40% for every country ism and gathered information on the actor(s) responsible. We recognize that in practice the five steps do not necessarily occur in order and are not always easily distinguished from one an- Data collection other. For instance, the determination of the strategy is connected In March 2015, we conducted a literature review on resource alloca- very closely to resource mobilization; typically these two things hap- tion in multilateral organizations using Google Scholar and the key- pen in dialogue with one another. Further, some of these steps are words ‘resource allocation” combined with either ‘multilateral negotiated on different cycles, as with replenishment activities or organisations’ or and institution name, e.g. ‘Gavi.’ To capture grey during programmatic strategy reviews. Nonetheless, we believe this literature we repeated the search using generic Google and browsing model is useful for clarifying the elements of the allocation process the publication section of each institution’s website. The results and facilitating a comparison across the institutions we study. included board decisions, budget documents, financial reports, re- ports on replenishments and other public documents. Documents Limitations from 1998 to 2015 were included in this review. Our study relies mainly on documentary evidence as made available We found that all institutions published some information on al- by the institutions in our sample and on information collected dur- location, although public disclosure varied substantially over the ing our interviews and literature search. As a result, this approach sample. To complement this literature search, for each institution presents several limitations. First of all, it is possible that inter- we conducted semi-structured interviews with at least two senior viewees overlooked important information, consciously or by mis- managers working on resource allocation or related policies or proc- take; or didn’t consider informal processes that shape allocation esses (for instance heads of policy, senior programme managers, dir- decisions. We could not gather information about the politics that ectors of data and information and the like) between April and could lead to deviations from the formal processes as neither author November 2015. The interviews lasted 1 h, and were conducted in had intimate access to board discussions or diplomatic channels. In person for the five institutions headquartered in Geneva, and by addition, we might not have discovered all aspects of resource allo- phone or Skype for the remaining four institutions. For each one, we cation in the interview and review process because we did not ask began by asking for a succinct account of the main resource alloca- the right questions or find the appropriate documents in our litera- tion steps without providing further guidance. After the interviews ture search process. We also do not know how much of the process we compared notes to produce a consensus account. Where there is shaped by the decisions that we attempted to document. It is pos- were uncertainties or we found discrepancies with published sources sible that very large amounts of resources are not accounted for in we asked additional questions. We then provided the completed ac- official documents, and are allocated through undocumented paral- count to each interviewee for verification, making additional adjust- lel systems. ments, if needed. Finally, our study was conducted mainly in 2015, but allocation processes are dynamic and in some cases were already changing by the time this article was drafted. Where we are aware of such Framework for analysis changes we say so in the text, but in none of the cases do we intend To structure our analysis we developed a model allocation cycle to suggest that processes will remain as we found them in 2015. based on five common themes that we judged to be prominent in our review of documents and in our interviews, presented below as Figure 1. The first of the five steps is the definition of institutional Results and discussion strategy, in which its governing body and/or senior leadership decide organizational goals. The second step is resource mobilization. The resource allocation cycle was used to provide a summary of the From there, we identified the sequence of decisions that lead to allocation cycle in each institution (Table 2), as well as to structure Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i7 Table 2. Summary description of resource allocation cycles Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds Gavi Formulated every four years Replenishment after the strat- Only countries with a GNI per - Funding of 11 vaccine The resources are allocated separately for health sys- by the Executive Board (in- egy is defined and adopted, capita below US$1580 þ programmes (in current tems support and vaccine programmes. cludes Gavi’s CEO, 28 resources are mobilized dur- DPT3 immunization rate vaccine portfolio) For HSS, size of annual birth cohort is multiplied by $5 members of international ing a pledging conference. above 70% of the eligible - Technical support for for countries with a GNI per capita lower than $365, organizations, independent Between 2000 and 2014: 69% population. health system strengthening and by $2.5 for countries with a GNI per capita com- experts, vaccine industry) from direct contributions Countries have to apply for (HSS) prised between $365 and $1580. from governments, founda- support (three windows of For vaccine programmes, the size of the birth cohort is tions and private donors. application per year), and the multiplied by the price of the vaccine minus the level 31% generated through in- independent Review of co-financing. Level of co-financing is based on novative financing Committee and the Gavi country’s income (countries with an income below mechanisms. Secretariat decide on the out- $1045 in 2015 contributed $0.20 per vaccine dose). Gavi’s funding is mainly come of the application. In addition, for newly approved vaccine programmes, a received in the form of open Countries benefit from support vaccine introduction grant is available ($0.80 per contributions (i.e. not until they are no longer eli- child in the birth cohort for all vaccines except earmarked) gible (e.g. GNI pc exceeds Human papilloma virus vaccine, for which countries threshold) receive $2.40 per girl in the birth cohort) GFATM New strategy defined every Replenishment after the strat- The GFATM applies income- The GFATM supports coun- The bulk of resources is allocated through an allocation four years by the Strategic egy is defined and adopted, based and disease-burden tries in three disease areas formula, except for countries with higher income and Investment and Impact resources are mobilized dur- based eligibility criteria. (HIV/AIDS, Malaria, and lower disease burden. For those countries, allocations Committee, led by the ing a pledging conference. High-income countries are Tuberculosis). are calculated using population size. In addition, Executive Director, strat- 95% of GFATM funding not eligible, regardless of dis- Countries can also apply for funds are set aside for innovative projects. Allocation egy leads, and consultants. comes from national govern- ease burden. Depending on HSS support. is calculated every four years following the ments and the European income, further eligibility cri- replenishment. Union. The remaining share teria are applied (e.g. focus of Allocations are calculated using an allocation formula of funding comes from pri- application, counterpart that relies mainly on a country’s disease burden (indi- vate foundations, corpor- financing, G-20 cators selected through consultation with partner ex- ations, and faith based membership). pert organizations) and country’s ability to pay organizations (5%). Countries have to apply for (which is derived from the GNI per capita). Countries The GFATM receives very lim- receive a score from the multiplication of these two funding, with the support of ited earmarked funding. the Secretariat through the elements, which corresponds to the country share, Country Coordination relative to the total funding envelope. Mechanism, and validated by Allocations are further adjusted using the following in- the Technical Review Panel. dicators: performance, impact, increasing rates of in- fection, absorptive capacity and other considerations. In addition, minimum and maximum caps are applied IDA/WB The strategic development of Every three years, resources are Only works with low-income Loans to finance development After each replenishment, a resource envelope is defined the institution is discussed replenished through a large countries, although some activities, at a zero interest for all eligible countries. This resource envelope is on a three year cycle, but a pledging conference. middle-income countries with rate and with a grant elem- allocated based on an allocation formula ‘strategy’ is not thoroughly IDA’s resources come from poor credit ratings exception- ent depending on country’s (Performance Based Allocation) that includes the fol- developed. This discussion loan repayment, income gen- ally benefit from support. risk of debt distress lowing indicators: country performance rating, popu- takes place during a meet- erated by other parts of the The eligibility for support is lation and GNI per capita. The most important ing with 52 IDA Deputies WB Group, and contribu- therefore defined by two cri- component is the country performance rating, which and 10 borrowing repre- tions made exclusively by teria: income (threshold set at is itself defined using the Country Policy and sentatives. IDA Deputies governments. $1205) and lack of access to Institutional Assessment (CPIA) and the Country (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i8 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds are appointed from the IDA does not receive ear- capital markets. Countries Portfolio Performance score, which is calculated using member states of the WB marked contributions also need to be part of the the occurrence of problems in previous IDA loans. International Monetary Allocations are adjusted as follow: a minimum alloca- Fund. tion is set (4 million SDR), as well as a maximum allo- To access funding, countries cation. A grant element is defined based on a need to undergo a systematic country’s risk of debt distress (for countries in high assessment and a country risk, the allocation is provided entirely in the form of partnership framework is a grant). defined to decide on the use Countries undergo a systematic assessment to determine of funds the constraints and opportunities to growth and pov- erty alleviation. Based on this assessment, countries formulate an application that details the programmes and interventions to be supported by the loans UNAIDS The strategy defines a set of Contributions are raised from UNAIDS does not allocate Supports the work of About one-third of the resources are allocated to coherent activities and ‘re- governments, Co-sponsors, funds to countries but to Cosponsors on HIV/AIDS cosponsors to strengthen their own programmes and sults areas’ for UNAIDS private partners and founda- Cosponsor organizations. to ensure that the global resource mobilization for HIV/AIDS. The remaining and its co-sponsors tions through an on-going There are 11 Cosponsors (10 response to HIV/AIDS is two-thirds of resources are used for ‘development organizations. financing dialogue. UN organizations and the coordinated. activities’ conducted by the Secretariat. It is approved by UNAIDS’ National governments contrib- World Bank). Funding allocated to each cosponsor is determined governing body, the ute 94% of the UNAIDS based on “epidemic priorities, performance of the Programme Coordinating budget, followed by 3% from Cosponsors and the funds that each Cosponsor Board (PCB), which in- Cosponsors, and 3% from raises” (UNAIDS, 2011) cludes representatives of 22 other partners (including pri- countries, the 11 cospon- vate sector partners and sors, and 5 NGOs. foundations). UNDP Every four years the A budget is prepared following All countries are eligible for Country Policy Support is Core resources (26% of all resources) are divided into Administrator of UNDP each strategy, and resources support, except those with a available to all countries three tiers under Targets for Resource Assignment and the Executive Board are mobilized from on-going GNI per capita of $12 475 or where need is perceived from the Core (TRAC) (UNDP, 2013a): supervise the development fund raising. more. Service delivery through its TRAC-1 (60% of regular funds) for programs of a new strategy. The Funding mainly come from Resources are allocated to offices TRAC-2 (31% of regular funds) is a flexible fund Executive Board is jointly governments, other multilat- country offices. Research that rewards projects that are well performing, high shared with UNOPS and eral organizations and private impact, or innovative UNFPA and composed of organizations. 74% of contri- TRAC-3 (8% of regular funds) for conflicts or emer- member countries butions are earmarked to spe- gencies, such as natural disasters representatives. cific activities. TRAC-1 resources are allocated using GNI per capita Only 26% of resources are core and total population. Allocation are raised if a coun- resources. try is categorized as an LDC (UN definition). Allocations are at least $350 000 to $500 000 depend- ing on whether they have a country office. Additionally, UNDP applies the following targets: 85 to 91 per cent of resources should be disbursed to low-income countries, 9 to 15 per cent to middle-in- come countries and at least 60 per cent to LDCs. Core resource allocations are complemented using non- core resources that are raised specifically to fund country programmes. These resources are raised based on the UNDAF. (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i9 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds UNFPA Prepared every four years, From the strategy, the budget UNFPA works with all coun- Policy and advocacy Regular resources are allocated through three systems: and approved by the for the institution as a whole tries based on an engagement Capacity development (i) the Resource Allocation System (RAS) (ii) global Executive Board (see cell is determined. Resources are framework that reflects coun- Research and regional programme and (iii) institutional budget. above). collected through an on- try needs and domestic Norms and standards setting The RAS assesses country needs using the following six going fund raising process. financing abilities (ranging Interventions, and technical indicators: (i) skilled birth attendance for the poorest Resources mainly come from from policy dialogue and ad- support quintile of the population, (ii) proportion of met de- voluntary contributions from vocacy to service delivery and mand for modern contraception, (iii) adolescent fertil- governments, non-govern- interventions (in low-income ity rate, (iv) maternal mortality ratio, (v) Gender mental organizations, foun- countries with needs judged Inequality Index, (vi) HIV prevalence among 15– dations, and private high or very high). 24 year olds. In addition, needs take into account the institutions. Resources are allocated to risk of humanitarian crises and inequality. Countries 53% of collected resources are country offices. are grouped in four categories of need. Each category earmarked to a specific activ- of countries (based on needs) receives a share of the ity or a thematic fund. The total resources, which is set during the definition of non-earmarked funds, 47% UNFPA’s strategy. For instance, countries with the of total contributions, are highest needs and lower income receive 53–63% of pooled as core resources the total envelope, whereas countries with the lowest (often referred to as ‘regular needs and highest income receive between 9–13%. resources’) A minimum allocation is also set between $300 000 and $500 000 depending on the income group. UNFPA also receives resources earmarked to one of its thematic fund (e.g. maternal health), which are dis- bursed in a similar fashion. Similarly to UNDP, non- core resources are also raised to fund country pro- grammes as set out in the UNDAF. UNICEF New strategy every four years Resources are raised through Works with all countries where Policy and advocacy Core resources are allocated using a formula that in- that includes an integrated on-going financing dialogue UNDAF assessment indicates Research cludes the following criteria: under five mortality plan and budget. based on strategy, budget, need, irrespective of income. Norms and standards setting rates, GNI per capita, and the population of children Strategy is approved by the and country-specific UNDAF. Resources are allocated to Interventions, and technical aged 5 or less. For each indicator, countries receive a Executive Board, composed Governments, non-governmen- country offices. support weight/point ranging from 0 to 1 based on their rank- of 36 members elected tal agencies and foundations ing (e.g. 0 attributed to the country with the lowest every three years by the UN contribute to the budget. child mortality, and 1 to the country with the high- Social and Economic UNICEF also receives re- est). An index is obtained by multiplying all three Council, and represents all sources raised by national weights, and is then used to calculate the allocation. regions. committees (in 36 countries). The following adjustments to the calculated allocation 26% are core resources; the re- are made: cap for the difference between two alloca- mainder is earmarked. tion periods (10%), minimum allocation (of $750 000 for all countries but high income), spending target of at least 60% to LDC, and of 50% to Sub-Saharan Africa. Similarly to UNFPA and UNDP, core resources do not represent the majority of resources disbursed. Non- core resources are earmarked resources to specific activities, and they are raised to support specific coun- try programmes (based on the UNDAF). (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i10 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds UNITAID The strategy is defined every Raises funding through on- The institution does not work Provides large scale multi- Funding to projects is allocated on a rounds-based sys- four years by the Executive going resource mobilization directly with countries, but country grants to improve tem, which starts with a call for proposal. Board, composed of 12 from national governments with implementing partners and accelerate the access of Organizations submit a letter of intent and if members, including one and innovative financing. (e.g. UNICEF, GFATM, the drugs and supplies. approved, a full proposal that includes the descrip- member appointed jointly In 2014, 50% of the funding Clinton Health Access tion of the equipment, drug or supply, timeline, by the five founding gov- was raised from the solidarity Initiative). Implementing budget, organisational details, policies on ethics, ernments (Brazil, Chile, levy on airline tickets, and partners are then in charge of anti-discrimination and the environment. A proposal France, Norway and the the remaining contributions distributing supplies to coun- review committee and the Secretariat review the pro- United-Kingdom) and were received by national tries, which are selected posal, and submit recommendations to the Executive Spain, representatives of governments and two based on a dialogue with the Board. the African and Asia re- foundations. Secretariat. gion, of civil society, of In general, at least 85% of WHO and of the constitu- funded supplies must be dis- ency of foundations. tributed to low-income coun- tries. No more than 10% in lower middle-income coun- tries or 5% to upper middle- income countries. WHO The World Health Assembly The WHO does not raise funds All member countries (without Research Resources are allocated to major offices (regional and (composed of 194 member through on-going resource regard to income) and work Norm and standard setting headquarter), which then allocate to countries every state representatives, usu- mobilization efforts. with partners (NGOs, civil Policy dialogue two years, based on life expectancy and GDP per cap- ally ministers) meets annu- 25% of total resources come society organizations). Programme support ita. Using these two indicators (which are scaled and ally to define and approve from ‘assessed contributions’ Resources are allocated to Capacity building and tech- multiplied to obtain a country score), countries are the program of work, set paid by member states based country offices. nical support ranked into deciles. Country shares are calculated by major policy directions on income. Emergency interventions multiplying the needs index (assigned to each decile) and approve the budget. Additional funding is raised with the log of population squared. through a financing dialogue Regional allocations are made by aggregating country with governments, other UN weightings in a given region. organisations, other intergov- It is worth noting that this resource allocation method- ernmental organizations, ology only applies for core resources. A working foundations, NGOS and the group was set up in 2014 to revisit this methodology, private sector. and a decision in April 2016 was made to add more The vast majority of these con- indicators and not rank countries into deciles (using tributions (93%) are ear- the country score instead). In addition to the existing marked to specific activities indicators, under-5 mortality and non-communicable (only 7% are core funding) disease prevalence, poverty headcount, and indicators of access (health workforce density, political instabil- ity, and DTP3 coverage) will be included to calculate allocations for each country A Supplementary Annex is published with this paper and contains a detailed summary of the process for each institution, as well as all bibliographical information for the information presented in this table. Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i11 the discussion on the trends across our group of institutions (below). Allocation For further information about individual institutions, as well as the Resources are allocated to specific programmatic and country activ- corresponding references, an extended account is available in ities within the bounds determined by the eligibility decision. Here, Supplementary Annex. we review how these allocation decisions are made by describing the processes that are used and the factors that are considered. We struc- ture our review by first discussing allocation systems at each institu- Strategy tion, and in the six cases where there is more than one, how funds All of the reviewed institutions had mechanisms for setting organiza- are divided between them. Second, we examine the primary vari- tional strategies, which typically include various goals and specify ables used in the main allocation system. Third, we discuss adjust- corresponding activities. In our sample, the strategies were updated ments, which may reflect qualitative factors or targets. The fourth on a cycle that ranged from 4 to 7 years. In each case, the highest step is country engagement, when the results of the allocation pro- governing body of the institution held final authority for approving cess are negotiated to arrive at a final figure. the strategy. UNITAID and UNAIDS do not make allocations to countries and We found no evidence to suggest that the frequency of strategy follow different processes. UNITAID makes allocations to implement- setting was a significant variable in explaining allocation decisions. ing partners based on submitted proposals, and UNAIDS allocates re- sources to Co-sponsors via a consultative process that reflects Resource mobilization each institution’s mandate, performance and capacity to mobilize re- We found that resources are mobilized either through periodic re- sources. We do not discuss these two further (see Supplementary plenishment, as at GFATM, Gavi, and IDA, or via continuous fund- Annex for more details), concentrating instead on the seven institu- raising, as by UNITAID and UN system institutions. For periodic tions that allocate to countries and programmes. replenishment, donors make multiyear commitments that are pooled, and collectively comprise the budget. As practiced by Resource allocation systems UNITAID and UN institutions, continuous fundraising engages Multiple resource allocation systems are used by each institution, many sources, including membership fees, voluntary contributions with the exception of IDA. GFATM uses a formula and subsequent and other activities such as selling products. adjustments to allocate the bulk of its resources. In parallel, ‘innova- We find that these two resource mobilization approaches gener- tive and impactful’ projects are funded from the Incentive Quality ate substantially different outcomes in our sample. Institutions em- Fund. Gavi provides support for vaccine programmes through a ploying periodic replenishment exercised complete, or nearly rounds-based system, and separately, allocates funding for health complete autonomy over their own budgets because contributed systems strengthening interventions using a resource allocation for- funds were pooled and not earmarked, except in rare circumstances. mula. In these two cases the use of two allocation systems corres- But institutions reliant on continuous fundraising received relatively ponds to two separate activities. At UNDP, UNFPA, UNICEF and few untied contributions. UNDP, UNICEF and WHO receive WHO, different systems are used to disburse different types of re- around 75% of total resources as earmarked contributions, whose sources. These institutions use a formula to allocate core resources. use is negotiated on an ad hoc basis with each donor and subject to Non-core resources—the majority of total resources—are allocated whatever restrictions are agreed. These earmarked funds are allo- via fragmented and poorly documented processes. This represents a cated as negotiated, and are not subject to the normal institutional significant limitation for our analysis of allocation systems because allocation process. the processes we describe are used for only 25–50% of resources for those institutions. Eligibility We do not find that the number of systems corresponds to differ- The most common eligibility metric was GNI per capita. Five of the ences in allocation. Similarly, we find no systematic difference be- seven institutions used a threshold based on GNI per capita, al- tween the results of processes that use an allocation formula vs those though these were set at different levels to align with different insti- that use application rounds. tutional principles. For instance, only countries classified as low- income by the World Bank are eligible for support from Gavi and IDA, which reflects a prioritization of the poorest countries. Gavi Indicators used in the primary resource allocation system then emphasizes absorptive capacity by including vaccine distribu- Table 3 provides an overview of indicators used to drive decisions tion performance requirement as measure of health system capacity. on the largest share of resources for Gavi, GFATM and IDA, and al- IDA includes measures of access to capital because it wants to chan- location of core resources for institutions of the UN system. In UN nel resources to countries that have the fewest alternatives. GFATM system institutions, the primary allocation system is not used for the emphasizes health needs and is willing to work with all countries ex- majority of resources because those are handled according to ad hoc cept those classified as high-income. At UNFPA, UNDP and agreements that we were unable to discover. UNICEF, eligibility is also determined through the UNDAF, regard- Our comparison of indicators used for allocation finds that GNI less of income level. Similarly, WHO also works with all countries per capita is used by all institutions, although not in the same way where need is identified from the Country Cooperation Strategy. or with the same weight (and WHO uses GPD rather than GNI). At Although ‘eligibility’ implies a binary decision, we find substan- GFATM and UNICEF, GNI per capita is considered using a sliding tial nuance in two dimensions—where engagement is sought, and scale that gives more weight to poorer countries and smooths thresh- then what type of support is provided. Among the seven institutions olds at higher levels. In contrast, in Gavi, GNI per capita is used to that provide support directly to countries we find that variation in define the co-payment on the vaccine drug ($0.20 per vaccine dose eligibility determinations largely follows different conceptualiza- for countries with a GNI per capita below $1045 in 2015). At IDA, tions of need (UNAIDS and UNITAID are not covered here). The GNI per capita is used directly in the allocation formula, although rules that define eligibility are nevertheless very influential because more emphasis is given to Country Performance Ratings (mainly the they do circumscribe subsequent allocation decisions. Country Policy and Institutional Assessment). Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i12 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 3. Summary of indicators used in allocation formulae Conclusion DAH has a critical impact on health services and health outcomes Institution Types of indicators for many of the world’s people, including the poorest and most vul- Gavi Size of the birth cohort, price of vaccine, GNI per capita nerable. Yet how exactly these resources are allocated has escaped (to calculate co-financing element) too many scholars and citizens. This article was motivated by GFATM Disease burden (calculated separately for each disease), the conviction that careful and transparent discussion of alloca- GNI per capita tion processes promotes equity and leads to better outcomes. UNDP GNI per capita, population size Documenting allocation processes proved difficult and it is possible UNFPA Skilled birth attendance for the poorest population or even probable that some details we report will require revision. quintile, proportion of demand for modern contra- This hazard is created by the absence of transparency on allocation ception satisfied, adolescent fertility rate, maternal mortality ratio, Gender Inequality Index, HIV preva- processes and speaks to the value of fostering transparency and pub- lence in 15–24 year olds, GNI per capita lic discussion, both of which promote fairness and accountability. UNICEF Under five mortality rate, GNI per capita, and child Considering technical factors, we reached mixed conclusions. At population all institutions we found systematic allocation processes incorporating WHO Life expectancy and GDP per capita many factors; none relied solely on GNI per capita. We also found that World Bank CPIA, Country Portfolio Performance, Population size, allocation formulae represent only a relatively narrow part of the allo- GNI per capita cation process, which is constrained far more by choices relating to strategy, type of support, eligibility, and qualitative adjustments. Applicable for the 2016–2017 budget period, but WHO is undergoing a Country participation in these aspects is very limited. Allocation deci- reform of its resource allocation formula for core resources sions cannot be understood by focusing on individual components, such as an allocation formula or the indicators it considers. Even taken Overall, we found that most institutions adjust allocation with as a whole, the formal allocation cycle does not explain all alloca- some health indicator. This is simplest at WHO, which uses life ex- tion—it is bypassed where earmarked resources are concerned, and pectancy, and most complex at UNFPA which uses five health statis- subject to post-hoc adjustments, as well. Moreover, despite the exist- tics and a gender inequality index in addition to GNI. Only IDA and ence of many monitoring and evaluation systems, country performance UNDP do not include a health indicator, although both use popula- seems to be considered only qualitatively in determining allocations tion size, and in the case of IDA, indicators of country performance. (except at the GFATM and IDA, and at the former the main perform- Surprisingly, we find that effectiveness or results are not included in ance measure is whether previous funds were spent). In a field preoccu- the primary resource allocation model; despite the widespread con- pied with measurement and effectiveness, this struck us as odd. cern in those institutions on results or issues such as ‘value for We were puzzled also by the lack of attention given to the incen- money’. tives generated by resource allocation processes. Most institutions This step seems to account for a great proportion of the observed emphasize absolute need, meaning that the relationship between differences in resource allocation. We were not able to collect infor- need and allocation is positive. At face value, this means that coun- mation on how institutions chose those indicators. From this result, tries are given fewer resources when they perform well, and, in ex- we postulate that, to some extent, those indicators are in line with pectation, it would drive resources to ineffective actors and institutional mandate. programs—the ones with the greatest need and lowest performance. We find this problematic, especially in countries where DAH ac- counts for a large share of health resources. Further research should analyse the incentives created by allocation systems and how they Types of adjustments and spending targets might impact country performance in health. Qualitative adjustments are subsequently applied to determine the Politically, we find that resource allocation activities are sensitive final allocation figures. Some adjustments are clear, the most com- and have been hidden from view as a result. Not all aspects of the pro- mon of which were minimum and maximum allocation limits, and cess were made public, and many available descriptions were difficult caps on variance between successive allocations. In addition, some to decipher. Qualitative adjustments to the results of formulae were institutions define spending targets to ensure that allocation deci- commonly mentioned, but very challenging to document. sions are aligned with institutional principles. For instance, UNICEF Negotiations around earmarking were not disclosed, either, and such has a target for programme allocations of 50% to Sub-Saharan funds are not subject to the normal allocation process at all. At Africa and 60% to countries classified as LDCs (UNICEF, 2012b). UNICEF, UNDP and WHO, >60% of total funds earmarked. We However, as a few informants explained to us, other adjustments recognize that confidentiality may be appropriate in stages, but we are used to account for important factors that resist quantification, argue that there should be clarity about all parts of the process, even for instance, absorptive capacity, the likelihood of corruption, past if the operational details of some steps are kept confidential. We con- performance or current political issues. Those adjustments rely on clude that further efforts are needed to improve the accountability internal data that are not made public. At UNDP, qualitative adjust- and the transparency of the decisions that shape resource allocation. ments are defined internally and approved by the Administrator. At From an ethical perspective, this study raises concerns about the the GFATM, qualitative adjustments are made to take into account representation of countries in resource allocation, which relied al- a wide range of considerations including past program performance, most exclusively on internal institutional processes. In some organ- risk and absorptive capacity. izations, countries are consulted individually about different parts of In most cases, such qualitative adjustments are not easily charac- the process, such as potential funding priorities, or implementation terized, although it is clear from our review that they can be import- details. However, country participation is limited during the wider ant, as at UNDP where 30% of core resources are allocated this decision making process. At the GFATM, countries are engaged at way. We could not document how these adjustments are made or the end of the process to negotiate their own allocation, although understand in detail what they reflect. Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i13 Dieleman JL, Graves C, Johnson E et al. 2015. Sources and focus of health de- only at the margin and in exceptional cases. As a result, countries’ velopment assistance, 1990–2014. JAMA 313: 2359–68. expressions of need and inputs seem to have only a limited effect on EAI (2015a). Updated–Terms of Reference for the Equitable Access Initiative. how overall funding envelopes are split between activities or coun- Equitable Access Initiative. https://www.theglobalfund.org/media/1320/ tries. A resource allocation cycle that is more inclusive of country eai_equitableaccessinitiative_tors_en.pdf. participation could foster better alignment of needs and allocation. EAI (2015b). Meeting Report: Equitable Access Initiative 23 February 2015, We also observe that allocation processes embody ethical prin- Geneva. Equitable Access Initiative. https://www.theglobalfund.org/media/ ciples, which we believe should be made more explicit. For instance, 1319/eai_2015-02-23-meeting_report_en.pdf. in its allocation process, Gavi prioritizes impact; in contrast Easterly W. (2008). Reinventing Foreign Aid. Vol. 1. MIT Press. GFATM prioritizes equity and stability, and the UN system institu- Hafner T, Shiffman J. 2012. The emergence of global attention to health sys- tions emphasize health need. Such discussion is important to pro- tems strengthening. Health Policy and Planning 28: 41–50. Hoeffler A, Outram V. (2008). What Determines the Allocation of Aid? The mote fairness, and could be conducted on a participatory basis. At Centre for the Study of African Economies Working Paper Series, 299. present, the types of support are constrained by politics, and none of Hotez PJ, Alvarado M, Basa ´ nez ~ MG et al. (2014). The global burden of disease the institutions are fully accountable to the countries or citizens they study 2010: interpretation and implications for the neglected tropical dis- intend to benefit. As a personal conviction, we believe that actual eases. PLoS neglected tropical diseases 8: e2865. and potential recipients have a right to know how allocation deci- Hristos D, Paldam M. 2009. The aid effectiveness literature: the sad results of sions are made and a right to representation in the process. We con- 40 years of research. Journal of Economic Surveys 23: 433–61. tribute this article as a step towards realizing these rights. Institute for Health Metrics and Evaluation (2014). Financing Global Health Visualization. Seattle, WA: IHME, University of Washington. http://vizhub. healthdata.org/fgh/. Supplementary Data McGillivray M. 2003. Modelling aid allocation: issues, approaches and re- Supplementary data are available at HEAPOL online. sults. Journal of Economic Development 28: 171–88. Ottersen T, Kamath A, Moon S, Røttingen JA. (2014). Development assist- ance for health: quantitative allocation criteria and contribution norms. Acknowledgements Centre on Global Health Security Working Group Papers, London: Chatham House. The authors would like to gratefully acknowledge the support of the Ottersen T, Norheim OF. 2014. Making fair choices on the path to Wellcome Trust in the funding this study, as well as the help of the many con- universal health coverage. Bulletin of the World Health Organization tacts and informants within each institution. 92: 389. Parkhurst JO, Vulimiri M. 2013. Cervical cancer and the global health agenda: Funding insights from multiple policy-analysis frameworks. Global Public Health 8: 1093–108. This work was supported by the Wellcome Trust [099114/Z/12/Z]. Piva P, Dodd R. 2009. Where did all the aid go? An in-depth analysis of Conflict of interest statement. None declared. increased health aid flows over the past 10 years. Bulletin of the World Health Organization 87: 930–39. Reich MR. 1995. The politics of agenda setting in international health: child References health versus adult health in developing countries. Journal of International Development 7: 489–502. Bourguignon F, Sundberg M. 2007. Aid effectiveness: opening the black box. Roberts M, Hsiao W, Berman P, Reich M. (2003). Getting Health Reform American Economy Review 97: 316–21. Right: A Guide to Improving Performance and Equity. Oxford University Bump JB, Reich MR, Johnson AM. 2013. Diarrhoeal diseases and the global Press. health agenda: measuring and changing priority. Health Policy and Roberts MJ, Reich MR. 2002. Ethical analysis in public health. The Lancet Planning 28: 799–808. 359: 1055–9. Clemens MA, Radelet S, Bhavnani RR, Bazzi S. 2011. Counting chickens Shiffman J, Smith S. 2007. Generation of political priority for global health when they hatch: Timing and the effects of aid on growth. 122: 590–617. initiatives: a framework and case study of maternal mortality. The Lancet Daniels N. 2000. Accountability for reasonableness: Establishing a fair process 370: 1370–9. for priority setting is easier than agreeing on principles. BMJ: British The Paris Declaration on Aid Effectiveness. (2005). http://www.oecd.org/dac/ Medical Journal 321: 1300. effectiveness/45827300.pdf. Daniels N. (2007). Just Health: Meeting Health Needs Fairly. Cambridge UNITAID. (2013). Strategy 2013–2016. UNITAID. University Press. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Health Policy and Planning Oxford University Press

Resource allocation processes at multilateral organizations working in global health

Health Policy and Planning , Volume 33 (suppl_1) – Feb 1, 2018

Loading next page...
1
 
/lp/ou_press/resource-allocation-processes-at-multilateral-organizations-working-in-wS8vcmU8GL

References (28)

Publisher
Oxford University Press
Copyright
Copyright © 2022 The London School of Hygiene and Tropical Medicine and Oxford University Press
ISSN
0268-1080
eISSN
1460-2237
DOI
10.1093/heapol/czx140
pmid
29415239
Publisher site
See Article on Publisher Site

Abstract

International institutions provide well over US$10 billion in development assistance for health (DAH) annually and between 1990 and 2014, DAH disbursements totaled $458 billion but how do they decide who gets what, and for what purpose? In this article, we explore how allocation decisions were made by the nine convening agencies of the Equitable Access Initiative. We provide clear, plain language de- scriptions of the complete process from resource mobilization to allocation for the nine multilateral agencies with prominent agendas in global health. Then, through a comparative analysis we illumin- ate the choices and strategies employed in the nine international institutions. We find that resource al- location in all reviewed institutions follow a similar pattern, which we categorized in a framework of five steps: strategy definition, resource mobilization, eligibility of countries, support type and funds al- location. All the reviewed institutions generate resource allocation decisions through well-structured and fairly complex processes. Variations in those processes seem to reflect differences in institutional principles and goals. However, these processes have serious shortcomings. Technical problems in- clude inadequate flexibility to account for or meet country needs. Although aid effectiveness and value for money are commonly referenced, we find that neither performance nor impact is a major cri- terion for allocating resources. We found very little formal consideration of the incentives generated by allocation choices. Political issues include non-transparent influence on allocation processes by donors and bureaucrats, and the common practice of earmarking funds to bypass the normal alloca- tion process entirely. Ethical deficiencies include low accountability and transparency at international institutions, and limited participation by affected citizens or their representatives. We find that recipi- ent countries have low influence on allocation processes themselves, although within these processes they have some influence in relatively narrow areas. Keywords: Aid, international health policy, resource allocation, global health Key Messages Extensive review of resource allocation at multilateral organisations working on global health finds the existence of well-structured and organized systems. Core principles of each institution are reflected in their design choices of allocation cycles (e.g. emphasis on fairness, health needs, efficiency). Although processes are documented, confidential or non-transparent elements such as qualitative adjustments and ear- marking agreements undermine the transparency and legitimacy of allocation decisions and restrict the knowledge and participation of recipient countries and their citizens. V The Author(s) 2018. Published by Oxford University Press in association with The London School of Hygiene and Tropical Medicine. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. i4 Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i5 Introduction analysed where DAH goes. Long time-series data have been used to analyse the patterns of DAH, but the results of those studies are not International institutions provide well over US$10 billion in devel- conclusive. For instance, Piva and Dodd (2009), Hotez et al. (2014) opment assistance for health (DAH) annually, and between 1990 and Dieleman et al. (2015) find wide variation in funding across and 2014 DAH disbursements totalled $458 billion (IHME 2014)— regions and disease areas that are not explained by differences in but how do they decide who gets what, and for what purpose? The disease burden or income. Estimation problems stemming from importance of these questions is difficult to overstate for the six bil- model misspecification, unobserved variables and measurement lion people in Low- and Middle-Income Countries (LMIC) whose problems have constrained empirical approaches (McGillivray health can be directly affected by DAH. Allocation decisions affect 2003; Hoeffler and Outram 2008). Others have discussed the pol- all three pillars of public health: the political, the technical, and the itics of DAH, particularly the haphazard process in which issues rise ethical (Roberts et al., 2003). Politically, allocation processes deter- and fall on the global health agenda (Shiffman & Smith, 2007; mine which countries get what assistance, raising distributional Hafner & Shiffman, 2012; Parkhurst & Vulimiri, 2013; Reich, issues at every level from the international down to individuals. 1995; Bump, Reich, & Johnson, 2013). Technical matters include the choice of objectives, for instance the Aid effectiveness has attracted lively commentary, attention suf- diseases to be addressed and the strategies that will be employed, the ficient to produce The Paris Declaration on Aid Effectiveness population to be targeted, and the effectiveness of interventions and (2005), and a robust literature, if not yet a consensus answer programs. Ethical considerations start with the fairness of the allo- (Bourguignon & Sundberg, 2007; Easterly, 2008; Piva and Dodd cation processes, and continue through the equity of their outcomes 2009; Hristos and Paldam 2009; Clemens et al. 2011). Adequate and consequences. and transparent resource allocation systems are often mentioned as The Equitable Access Initiative (EAI) was designed to explore desirable, for instance as Point 17 in the Paris Declaration, but as far some of these issues—primarily the technical aspects. The nine con- as we were able to detect, only Ottersen et al. (2014) have directly vening international agencies were particularly concerned with addressed the question of allocation criteria. Their work details the understanding the consequences of using gross national income criteria in allocation formulae in 10 bilateral and 5 multilateral insti- (GNI) per capita as a primary indicator of need, and interested in tutions working in global health and discusses their distributional exploring alternatives. Although GNI had long been used for that impact. Our aims differ in that we seek to understand the complete purpose, they noted that rising inequality means that most of the decision-making process leading to allocation decisions. Moreover, world’s poor now live in middle-income countries. Hence, the EAI unlike previous works, we supplemented a review of literature with was convened to investigate how processes might better reflect dis- interviews at each institution. ease burdens, national capacity to intervene, government health There is a very large literature on distributive justice and other budgets and other factors (EAI 2015a, b) primarily the technical aspects of ethics applied to global health. Much of this has centred aspects such as classifying country needs and capacities in health. on priority setting, which can be taken loosely as a synonym for al- Under the aegis of the EAI, academic teams were engaged to de- location in this context. Scholars have emphasized ‘accountability velop fresh thinking on classification methods for capturing country for reasonableness’ in such decisions (Daniels 2000, 2007), the im- needs and capacities in health, the products of which were the basis portance of ‘making fair choices’ of what to provide to whom in for papers in this special issue. These colleagues analysed the deter- health (Ottersen and Norheim 2014), and the centrality of ethics in minants of health outcomes to explore the value of GNI per capita all public health rationing decisions (Roberts and Reich 2002). Our as a proxy (Sterck et al.), modelled government ability to spend on inquiry sheds light on an important aspect of priority setting—how health under various conditions over time (Haakenstad et al.) and international agencies allocate DAH. conducted experiments to discover the values that participants felt We organize this article as follows. Section 2 The following section should be prioritized in allocation (Grepin et al.) and then applied presents our methods and data. Section 3 contains a descriptive table these ideas to develop new country rankings (Ottersen et al.). of the allocation process in each institution and summarizes our com- As an associated activity, we explored how allocation decisions parative analysis. Section 4 provides discussion and conclusions. were made by the EAI’s nine convening agencies themselves—Gavi, the Vaccine Alliance (hereafter Gavi), The Global Fund to fight AIDS, Tuberculosis, and Malaria (GFATM), The United Nations Methods and data Population Fund (UNFPA), The United Nations Children’s Fund (UNICEF), The United Nations Development Programme (UNDP), Institutions for analysis The World Bank Group’s International Development Agency, The Since our project was linked to the EAI, we focussed our analysis on World Health Organization (WHO), UNAIDS and UNITAID. We its convening agencies—Gavi, GFATM, UNFPA, UNICEF, UNDP, felt that examining allocation practices at these prominent institu- the World Bank’s IDA, WHO, UNAIDS (meaning the UNAIDS tions would be helpful to discussions about the future of allocation Secretariat) and UNITAID. Each of these institutions has significant and of great interest to recipient countries and their citizens. objectives in global health and a worldwide mandate, as we sum- To support our larger objective of fostering greater transparency, marize in Table 1. At present there are no other large international more informed discussion, and better allocation, we provide clear, institutions meeting these two criteria. Regional development banks, plain language descriptions of the complete process from resource bilateral agencies and foundations are also important objects of mobilization to allocation for each multilateral. Then, through a study, but these were not considered in the EAI and were beyond the comparative analysis we illuminate the choices and strategies em- scope of this article. Collectively, the EAI convening agencies pro- ployed in the nine international institutions. vided US $11.7 billion in DAH in 2014 (UNITAID 2013; Institute Our investigation of allocation methods at international agencies for Health Metrics and Evaluation 2014). In that year, the largest connects to several important literatures in global health, including sources were GFATM, WHO and Gavi (respectively US $4.1 billion, DAH and allocation, aid effectiveness and ethics. Some have US$2.1 billion and US$1.8 billion). Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i6 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 1. Institutional objectives Definition of the strategy Institution Objectives Gavi Save children’s lives and protecting people’s health by increasing access to immunisation in poor countries Allocation of Resource GFATM Accelerate the end of HIV/AIDS, tuberculosis, and mal- funds mobilization aria as epidemics UNAIDS Formulate a plan of action for all the institutions work- ing on HIV/AIDS to end the epidemic by 2030 UNDP Help eradicate poverty and reduce inequalities and exclusion Type of Eligibility of UNFPA Deliver a world where every pregnancy is wanted, every support countries birth is safe, and every young person’s potential is fulfilled UNICEF Promote the rights and wellbeing of every child Figure 1. Allocation cycle in multilateral organizations UNITAID Contribute to scale-up access to treatment for HIV/ AIDS, malaria, and TB for people in developing allocations. Step 3 is an eligibility determination; step 4 (type of sup- countries by leveraging price reductions of quality port) is the determination of what funds, services, in kind-support or drugs and diagnosis other resources will be made available; step 5 is the allocation of WHO The attainment of the highest possible standard of specific resources to specific recipients, such as programmes or health by all people. countries. Based on our document review and interview transcripts, World Bank End extreme poverty by decreasing the percentage of we categorized the collected information following this framework people living on less than $1.25 a day to less than 3%; promote shared prosperity by fostering the in- of analysis. For each step, we documented the institutional mechan- come growth of the bottom 40% for every country ism and gathered information on the actor(s) responsible. We recognize that in practice the five steps do not necessarily occur in order and are not always easily distinguished from one an- Data collection other. For instance, the determination of the strategy is connected In March 2015, we conducted a literature review on resource alloca- very closely to resource mobilization; typically these two things hap- tion in multilateral organizations using Google Scholar and the key- pen in dialogue with one another. Further, some of these steps are words ‘resource allocation” combined with either ‘multilateral negotiated on different cycles, as with replenishment activities or organisations’ or and institution name, e.g. ‘Gavi.’ To capture grey during programmatic strategy reviews. Nonetheless, we believe this literature we repeated the search using generic Google and browsing model is useful for clarifying the elements of the allocation process the publication section of each institution’s website. The results and facilitating a comparison across the institutions we study. included board decisions, budget documents, financial reports, re- ports on replenishments and other public documents. Documents Limitations from 1998 to 2015 were included in this review. Our study relies mainly on documentary evidence as made available We found that all institutions published some information on al- by the institutions in our sample and on information collected dur- location, although public disclosure varied substantially over the ing our interviews and literature search. As a result, this approach sample. To complement this literature search, for each institution presents several limitations. First of all, it is possible that inter- we conducted semi-structured interviews with at least two senior viewees overlooked important information, consciously or by mis- managers working on resource allocation or related policies or proc- take; or didn’t consider informal processes that shape allocation esses (for instance heads of policy, senior programme managers, dir- decisions. We could not gather information about the politics that ectors of data and information and the like) between April and could lead to deviations from the formal processes as neither author November 2015. The interviews lasted 1 h, and were conducted in had intimate access to board discussions or diplomatic channels. In person for the five institutions headquartered in Geneva, and by addition, we might not have discovered all aspects of resource allo- phone or Skype for the remaining four institutions. For each one, we cation in the interview and review process because we did not ask began by asking for a succinct account of the main resource alloca- the right questions or find the appropriate documents in our litera- tion steps without providing further guidance. After the interviews ture search process. We also do not know how much of the process we compared notes to produce a consensus account. Where there is shaped by the decisions that we attempted to document. It is pos- were uncertainties or we found discrepancies with published sources sible that very large amounts of resources are not accounted for in we asked additional questions. We then provided the completed ac- official documents, and are allocated through undocumented paral- count to each interviewee for verification, making additional adjust- lel systems. ments, if needed. Finally, our study was conducted mainly in 2015, but allocation processes are dynamic and in some cases were already changing by the time this article was drafted. Where we are aware of such Framework for analysis changes we say so in the text, but in none of the cases do we intend To structure our analysis we developed a model allocation cycle to suggest that processes will remain as we found them in 2015. based on five common themes that we judged to be prominent in our review of documents and in our interviews, presented below as Figure 1. The first of the five steps is the definition of institutional Results and discussion strategy, in which its governing body and/or senior leadership decide organizational goals. The second step is resource mobilization. The resource allocation cycle was used to provide a summary of the From there, we identified the sequence of decisions that lead to allocation cycle in each institution (Table 2), as well as to structure Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i7 Table 2. Summary description of resource allocation cycles Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds Gavi Formulated every four years Replenishment after the strat- Only countries with a GNI per - Funding of 11 vaccine The resources are allocated separately for health sys- by the Executive Board (in- egy is defined and adopted, capita below US$1580 þ programmes (in current tems support and vaccine programmes. cludes Gavi’s CEO, 28 resources are mobilized dur- DPT3 immunization rate vaccine portfolio) For HSS, size of annual birth cohort is multiplied by $5 members of international ing a pledging conference. above 70% of the eligible - Technical support for for countries with a GNI per capita lower than $365, organizations, independent Between 2000 and 2014: 69% population. health system strengthening and by $2.5 for countries with a GNI per capita com- experts, vaccine industry) from direct contributions Countries have to apply for (HSS) prised between $365 and $1580. from governments, founda- support (three windows of For vaccine programmes, the size of the birth cohort is tions and private donors. application per year), and the multiplied by the price of the vaccine minus the level 31% generated through in- independent Review of co-financing. Level of co-financing is based on novative financing Committee and the Gavi country’s income (countries with an income below mechanisms. Secretariat decide on the out- $1045 in 2015 contributed $0.20 per vaccine dose). Gavi’s funding is mainly come of the application. In addition, for newly approved vaccine programmes, a received in the form of open Countries benefit from support vaccine introduction grant is available ($0.80 per contributions (i.e. not until they are no longer eli- child in the birth cohort for all vaccines except earmarked) gible (e.g. GNI pc exceeds Human papilloma virus vaccine, for which countries threshold) receive $2.40 per girl in the birth cohort) GFATM New strategy defined every Replenishment after the strat- The GFATM applies income- The GFATM supports coun- The bulk of resources is allocated through an allocation four years by the Strategic egy is defined and adopted, based and disease-burden tries in three disease areas formula, except for countries with higher income and Investment and Impact resources are mobilized dur- based eligibility criteria. (HIV/AIDS, Malaria, and lower disease burden. For those countries, allocations Committee, led by the ing a pledging conference. High-income countries are Tuberculosis). are calculated using population size. In addition, Executive Director, strat- 95% of GFATM funding not eligible, regardless of dis- Countries can also apply for funds are set aside for innovative projects. Allocation egy leads, and consultants. comes from national govern- ease burden. Depending on HSS support. is calculated every four years following the ments and the European income, further eligibility cri- replenishment. Union. The remaining share teria are applied (e.g. focus of Allocations are calculated using an allocation formula of funding comes from pri- application, counterpart that relies mainly on a country’s disease burden (indi- vate foundations, corpor- financing, G-20 cators selected through consultation with partner ex- ations, and faith based membership). pert organizations) and country’s ability to pay organizations (5%). Countries have to apply for (which is derived from the GNI per capita). Countries The GFATM receives very lim- receive a score from the multiplication of these two funding, with the support of ited earmarked funding. the Secretariat through the elements, which corresponds to the country share, Country Coordination relative to the total funding envelope. Mechanism, and validated by Allocations are further adjusted using the following in- the Technical Review Panel. dicators: performance, impact, increasing rates of in- fection, absorptive capacity and other considerations. In addition, minimum and maximum caps are applied IDA/WB The strategic development of Every three years, resources are Only works with low-income Loans to finance development After each replenishment, a resource envelope is defined the institution is discussed replenished through a large countries, although some activities, at a zero interest for all eligible countries. This resource envelope is on a three year cycle, but a pledging conference. middle-income countries with rate and with a grant elem- allocated based on an allocation formula ‘strategy’ is not thoroughly IDA’s resources come from poor credit ratings exception- ent depending on country’s (Performance Based Allocation) that includes the fol- developed. This discussion loan repayment, income gen- ally benefit from support. risk of debt distress lowing indicators: country performance rating, popu- takes place during a meet- erated by other parts of the The eligibility for support is lation and GNI per capita. The most important ing with 52 IDA Deputies WB Group, and contribu- therefore defined by two cri- component is the country performance rating, which and 10 borrowing repre- tions made exclusively by teria: income (threshold set at is itself defined using the Country Policy and sentatives. IDA Deputies governments. $1205) and lack of access to Institutional Assessment (CPIA) and the Country (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i8 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds are appointed from the IDA does not receive ear- capital markets. Countries Portfolio Performance score, which is calculated using member states of the WB marked contributions also need to be part of the the occurrence of problems in previous IDA loans. International Monetary Allocations are adjusted as follow: a minimum alloca- Fund. tion is set (4 million SDR), as well as a maximum allo- To access funding, countries cation. A grant element is defined based on a need to undergo a systematic country’s risk of debt distress (for countries in high assessment and a country risk, the allocation is provided entirely in the form of partnership framework is a grant). defined to decide on the use Countries undergo a systematic assessment to determine of funds the constraints and opportunities to growth and pov- erty alleviation. Based on this assessment, countries formulate an application that details the programmes and interventions to be supported by the loans UNAIDS The strategy defines a set of Contributions are raised from UNAIDS does not allocate Supports the work of About one-third of the resources are allocated to coherent activities and ‘re- governments, Co-sponsors, funds to countries but to Cosponsors on HIV/AIDS cosponsors to strengthen their own programmes and sults areas’ for UNAIDS private partners and founda- Cosponsor organizations. to ensure that the global resource mobilization for HIV/AIDS. The remaining and its co-sponsors tions through an on-going There are 11 Cosponsors (10 response to HIV/AIDS is two-thirds of resources are used for ‘development organizations. financing dialogue. UN organizations and the coordinated. activities’ conducted by the Secretariat. It is approved by UNAIDS’ National governments contrib- World Bank). Funding allocated to each cosponsor is determined governing body, the ute 94% of the UNAIDS based on “epidemic priorities, performance of the Programme Coordinating budget, followed by 3% from Cosponsors and the funds that each Cosponsor Board (PCB), which in- Cosponsors, and 3% from raises” (UNAIDS, 2011) cludes representatives of 22 other partners (including pri- countries, the 11 cospon- vate sector partners and sors, and 5 NGOs. foundations). UNDP Every four years the A budget is prepared following All countries are eligible for Country Policy Support is Core resources (26% of all resources) are divided into Administrator of UNDP each strategy, and resources support, except those with a available to all countries three tiers under Targets for Resource Assignment and the Executive Board are mobilized from on-going GNI per capita of $12 475 or where need is perceived from the Core (TRAC) (UNDP, 2013a): supervise the development fund raising. more. Service delivery through its TRAC-1 (60% of regular funds) for programs of a new strategy. The Funding mainly come from Resources are allocated to offices TRAC-2 (31% of regular funds) is a flexible fund Executive Board is jointly governments, other multilat- country offices. Research that rewards projects that are well performing, high shared with UNOPS and eral organizations and private impact, or innovative UNFPA and composed of organizations. 74% of contri- TRAC-3 (8% of regular funds) for conflicts or emer- member countries butions are earmarked to spe- gencies, such as natural disasters representatives. cific activities. TRAC-1 resources are allocated using GNI per capita Only 26% of resources are core and total population. Allocation are raised if a coun- resources. try is categorized as an LDC (UN definition). Allocations are at least $350 000 to $500 000 depend- ing on whether they have a country office. Additionally, UNDP applies the following targets: 85 to 91 per cent of resources should be disbursed to low-income countries, 9 to 15 per cent to middle-in- come countries and at least 60 per cent to LDCs. Core resource allocations are complemented using non- core resources that are raised specifically to fund country programmes. These resources are raised based on the UNDAF. (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i9 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds UNFPA Prepared every four years, From the strategy, the budget UNFPA works with all coun- Policy and advocacy Regular resources are allocated through three systems: and approved by the for the institution as a whole tries based on an engagement Capacity development (i) the Resource Allocation System (RAS) (ii) global Executive Board (see cell is determined. Resources are framework that reflects coun- Research and regional programme and (iii) institutional budget. above). collected through an on- try needs and domestic Norms and standards setting The RAS assesses country needs using the following six going fund raising process. financing abilities (ranging Interventions, and technical indicators: (i) skilled birth attendance for the poorest Resources mainly come from from policy dialogue and ad- support quintile of the population, (ii) proportion of met de- voluntary contributions from vocacy to service delivery and mand for modern contraception, (iii) adolescent fertil- governments, non-govern- interventions (in low-income ity rate, (iv) maternal mortality ratio, (v) Gender mental organizations, foun- countries with needs judged Inequality Index, (vi) HIV prevalence among 15– dations, and private high or very high). 24 year olds. In addition, needs take into account the institutions. Resources are allocated to risk of humanitarian crises and inequality. Countries 53% of collected resources are country offices. are grouped in four categories of need. Each category earmarked to a specific activ- of countries (based on needs) receives a share of the ity or a thematic fund. The total resources, which is set during the definition of non-earmarked funds, 47% UNFPA’s strategy. For instance, countries with the of total contributions, are highest needs and lower income receive 53–63% of pooled as core resources the total envelope, whereas countries with the lowest (often referred to as ‘regular needs and highest income receive between 9–13%. resources’) A minimum allocation is also set between $300 000 and $500 000 depending on the income group. UNFPA also receives resources earmarked to one of its thematic fund (e.g. maternal health), which are dis- bursed in a similar fashion. Similarly to UNDP, non- core resources are also raised to fund country pro- grammes as set out in the UNDAF. UNICEF New strategy every four years Resources are raised through Works with all countries where Policy and advocacy Core resources are allocated using a formula that in- that includes an integrated on-going financing dialogue UNDAF assessment indicates Research cludes the following criteria: under five mortality plan and budget. based on strategy, budget, need, irrespective of income. Norms and standards setting rates, GNI per capita, and the population of children Strategy is approved by the and country-specific UNDAF. Resources are allocated to Interventions, and technical aged 5 or less. For each indicator, countries receive a Executive Board, composed Governments, non-governmen- country offices. support weight/point ranging from 0 to 1 based on their rank- of 36 members elected tal agencies and foundations ing (e.g. 0 attributed to the country with the lowest every three years by the UN contribute to the budget. child mortality, and 1 to the country with the high- Social and Economic UNICEF also receives re- est). An index is obtained by multiplying all three Council, and represents all sources raised by national weights, and is then used to calculate the allocation. regions. committees (in 36 countries). The following adjustments to the calculated allocation 26% are core resources; the re- are made: cap for the difference between two alloca- mainder is earmarked. tion periods (10%), minimum allocation (of $750 000 for all countries but high income), spending target of at least 60% to LDC, and of 50% to Sub-Saharan Africa. Similarly to UNFPA and UNDP, core resources do not represent the majority of resources disbursed. Non- core resources are earmarked resources to specific activities, and they are raised to support specific coun- try programmes (based on the UNDAF). (Continued) Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i10 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 2. (Continued) Strategy Resource mobilization Eligibility of countries Type of support Allocation of funds UNITAID The strategy is defined every Raises funding through on- The institution does not work Provides large scale multi- Funding to projects is allocated on a rounds-based sys- four years by the Executive going resource mobilization directly with countries, but country grants to improve tem, which starts with a call for proposal. Board, composed of 12 from national governments with implementing partners and accelerate the access of Organizations submit a letter of intent and if members, including one and innovative financing. (e.g. UNICEF, GFATM, the drugs and supplies. approved, a full proposal that includes the descrip- member appointed jointly In 2014, 50% of the funding Clinton Health Access tion of the equipment, drug or supply, timeline, by the five founding gov- was raised from the solidarity Initiative). Implementing budget, organisational details, policies on ethics, ernments (Brazil, Chile, levy on airline tickets, and partners are then in charge of anti-discrimination and the environment. A proposal France, Norway and the the remaining contributions distributing supplies to coun- review committee and the Secretariat review the pro- United-Kingdom) and were received by national tries, which are selected posal, and submit recommendations to the Executive Spain, representatives of governments and two based on a dialogue with the Board. the African and Asia re- foundations. Secretariat. gion, of civil society, of In general, at least 85% of WHO and of the constitu- funded supplies must be dis- ency of foundations. tributed to low-income coun- tries. No more than 10% in lower middle-income coun- tries or 5% to upper middle- income countries. WHO The World Health Assembly The WHO does not raise funds All member countries (without Research Resources are allocated to major offices (regional and (composed of 194 member through on-going resource regard to income) and work Norm and standard setting headquarter), which then allocate to countries every state representatives, usu- mobilization efforts. with partners (NGOs, civil Policy dialogue two years, based on life expectancy and GDP per cap- ally ministers) meets annu- 25% of total resources come society organizations). Programme support ita. Using these two indicators (which are scaled and ally to define and approve from ‘assessed contributions’ Resources are allocated to Capacity building and tech- multiplied to obtain a country score), countries are the program of work, set paid by member states based country offices. nical support ranked into deciles. Country shares are calculated by major policy directions on income. Emergency interventions multiplying the needs index (assigned to each decile) and approve the budget. Additional funding is raised with the log of population squared. through a financing dialogue Regional allocations are made by aggregating country with governments, other UN weightings in a given region. organisations, other intergov- It is worth noting that this resource allocation method- ernmental organizations, ology only applies for core resources. A working foundations, NGOS and the group was set up in 2014 to revisit this methodology, private sector. and a decision in April 2016 was made to add more The vast majority of these con- indicators and not rank countries into deciles (using tributions (93%) are ear- the country score instead). In addition to the existing marked to specific activities indicators, under-5 mortality and non-communicable (only 7% are core funding) disease prevalence, poverty headcount, and indicators of access (health workforce density, political instabil- ity, and DTP3 coverage) will be included to calculate allocations for each country A Supplementary Annex is published with this paper and contains a detailed summary of the process for each institution, as well as all bibliographical information for the information presented in this table. Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i11 the discussion on the trends across our group of institutions (below). Allocation For further information about individual institutions, as well as the Resources are allocated to specific programmatic and country activ- corresponding references, an extended account is available in ities within the bounds determined by the eligibility decision. Here, Supplementary Annex. we review how these allocation decisions are made by describing the processes that are used and the factors that are considered. We struc- ture our review by first discussing allocation systems at each institu- Strategy tion, and in the six cases where there is more than one, how funds All of the reviewed institutions had mechanisms for setting organiza- are divided between them. Second, we examine the primary vari- tional strategies, which typically include various goals and specify ables used in the main allocation system. Third, we discuss adjust- corresponding activities. In our sample, the strategies were updated ments, which may reflect qualitative factors or targets. The fourth on a cycle that ranged from 4 to 7 years. In each case, the highest step is country engagement, when the results of the allocation pro- governing body of the institution held final authority for approving cess are negotiated to arrive at a final figure. the strategy. UNITAID and UNAIDS do not make allocations to countries and We found no evidence to suggest that the frequency of strategy follow different processes. UNITAID makes allocations to implement- setting was a significant variable in explaining allocation decisions. ing partners based on submitted proposals, and UNAIDS allocates re- sources to Co-sponsors via a consultative process that reflects Resource mobilization each institution’s mandate, performance and capacity to mobilize re- We found that resources are mobilized either through periodic re- sources. We do not discuss these two further (see Supplementary plenishment, as at GFATM, Gavi, and IDA, or via continuous fund- Annex for more details), concentrating instead on the seven institu- raising, as by UNITAID and UN system institutions. For periodic tions that allocate to countries and programmes. replenishment, donors make multiyear commitments that are pooled, and collectively comprise the budget. As practiced by Resource allocation systems UNITAID and UN institutions, continuous fundraising engages Multiple resource allocation systems are used by each institution, many sources, including membership fees, voluntary contributions with the exception of IDA. GFATM uses a formula and subsequent and other activities such as selling products. adjustments to allocate the bulk of its resources. In parallel, ‘innova- We find that these two resource mobilization approaches gener- tive and impactful’ projects are funded from the Incentive Quality ate substantially different outcomes in our sample. Institutions em- Fund. Gavi provides support for vaccine programmes through a ploying periodic replenishment exercised complete, or nearly rounds-based system, and separately, allocates funding for health complete autonomy over their own budgets because contributed systems strengthening interventions using a resource allocation for- funds were pooled and not earmarked, except in rare circumstances. mula. In these two cases the use of two allocation systems corres- But institutions reliant on continuous fundraising received relatively ponds to two separate activities. At UNDP, UNFPA, UNICEF and few untied contributions. UNDP, UNICEF and WHO receive WHO, different systems are used to disburse different types of re- around 75% of total resources as earmarked contributions, whose sources. These institutions use a formula to allocate core resources. use is negotiated on an ad hoc basis with each donor and subject to Non-core resources—the majority of total resources—are allocated whatever restrictions are agreed. These earmarked funds are allo- via fragmented and poorly documented processes. This represents a cated as negotiated, and are not subject to the normal institutional significant limitation for our analysis of allocation systems because allocation process. the processes we describe are used for only 25–50% of resources for those institutions. Eligibility We do not find that the number of systems corresponds to differ- The most common eligibility metric was GNI per capita. Five of the ences in allocation. Similarly, we find no systematic difference be- seven institutions used a threshold based on GNI per capita, al- tween the results of processes that use an allocation formula vs those though these were set at different levels to align with different insti- that use application rounds. tutional principles. For instance, only countries classified as low- income by the World Bank are eligible for support from Gavi and IDA, which reflects a prioritization of the poorest countries. Gavi Indicators used in the primary resource allocation system then emphasizes absorptive capacity by including vaccine distribu- Table 3 provides an overview of indicators used to drive decisions tion performance requirement as measure of health system capacity. on the largest share of resources for Gavi, GFATM and IDA, and al- IDA includes measures of access to capital because it wants to chan- location of core resources for institutions of the UN system. In UN nel resources to countries that have the fewest alternatives. GFATM system institutions, the primary allocation system is not used for the emphasizes health needs and is willing to work with all countries ex- majority of resources because those are handled according to ad hoc cept those classified as high-income. At UNFPA, UNDP and agreements that we were unable to discover. UNICEF, eligibility is also determined through the UNDAF, regard- Our comparison of indicators used for allocation finds that GNI less of income level. Similarly, WHO also works with all countries per capita is used by all institutions, although not in the same way where need is identified from the Country Cooperation Strategy. or with the same weight (and WHO uses GPD rather than GNI). At Although ‘eligibility’ implies a binary decision, we find substan- GFATM and UNICEF, GNI per capita is considered using a sliding tial nuance in two dimensions—where engagement is sought, and scale that gives more weight to poorer countries and smooths thresh- then what type of support is provided. Among the seven institutions olds at higher levels. In contrast, in Gavi, GNI per capita is used to that provide support directly to countries we find that variation in define the co-payment on the vaccine drug ($0.20 per vaccine dose eligibility determinations largely follows different conceptualiza- for countries with a GNI per capita below $1045 in 2015). At IDA, tions of need (UNAIDS and UNITAID are not covered here). The GNI per capita is used directly in the allocation formula, although rules that define eligibility are nevertheless very influential because more emphasis is given to Country Performance Ratings (mainly the they do circumscribe subsequent allocation decisions. Country Policy and Institutional Assessment). Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 i12 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 Table 3. Summary of indicators used in allocation formulae Conclusion DAH has a critical impact on health services and health outcomes Institution Types of indicators for many of the world’s people, including the poorest and most vul- Gavi Size of the birth cohort, price of vaccine, GNI per capita nerable. Yet how exactly these resources are allocated has escaped (to calculate co-financing element) too many scholars and citizens. This article was motivated by GFATM Disease burden (calculated separately for each disease), the conviction that careful and transparent discussion of alloca- GNI per capita tion processes promotes equity and leads to better outcomes. UNDP GNI per capita, population size Documenting allocation processes proved difficult and it is possible UNFPA Skilled birth attendance for the poorest population or even probable that some details we report will require revision. quintile, proportion of demand for modern contra- This hazard is created by the absence of transparency on allocation ception satisfied, adolescent fertility rate, maternal mortality ratio, Gender Inequality Index, HIV preva- processes and speaks to the value of fostering transparency and pub- lence in 15–24 year olds, GNI per capita lic discussion, both of which promote fairness and accountability. UNICEF Under five mortality rate, GNI per capita, and child Considering technical factors, we reached mixed conclusions. At population all institutions we found systematic allocation processes incorporating WHO Life expectancy and GDP per capita many factors; none relied solely on GNI per capita. We also found that World Bank CPIA, Country Portfolio Performance, Population size, allocation formulae represent only a relatively narrow part of the allo- GNI per capita cation process, which is constrained far more by choices relating to strategy, type of support, eligibility, and qualitative adjustments. Applicable for the 2016–2017 budget period, but WHO is undergoing a Country participation in these aspects is very limited. Allocation deci- reform of its resource allocation formula for core resources sions cannot be understood by focusing on individual components, such as an allocation formula or the indicators it considers. Even taken Overall, we found that most institutions adjust allocation with as a whole, the formal allocation cycle does not explain all alloca- some health indicator. This is simplest at WHO, which uses life ex- tion—it is bypassed where earmarked resources are concerned, and pectancy, and most complex at UNFPA which uses five health statis- subject to post-hoc adjustments, as well. Moreover, despite the exist- tics and a gender inequality index in addition to GNI. Only IDA and ence of many monitoring and evaluation systems, country performance UNDP do not include a health indicator, although both use popula- seems to be considered only qualitatively in determining allocations tion size, and in the case of IDA, indicators of country performance. (except at the GFATM and IDA, and at the former the main perform- Surprisingly, we find that effectiveness or results are not included in ance measure is whether previous funds were spent). In a field preoccu- the primary resource allocation model; despite the widespread con- pied with measurement and effectiveness, this struck us as odd. cern in those institutions on results or issues such as ‘value for We were puzzled also by the lack of attention given to the incen- money’. tives generated by resource allocation processes. Most institutions This step seems to account for a great proportion of the observed emphasize absolute need, meaning that the relationship between differences in resource allocation. We were not able to collect infor- need and allocation is positive. At face value, this means that coun- mation on how institutions chose those indicators. From this result, tries are given fewer resources when they perform well, and, in ex- we postulate that, to some extent, those indicators are in line with pectation, it would drive resources to ineffective actors and institutional mandate. programs—the ones with the greatest need and lowest performance. We find this problematic, especially in countries where DAH ac- counts for a large share of health resources. Further research should analyse the incentives created by allocation systems and how they Types of adjustments and spending targets might impact country performance in health. Qualitative adjustments are subsequently applied to determine the Politically, we find that resource allocation activities are sensitive final allocation figures. Some adjustments are clear, the most com- and have been hidden from view as a result. Not all aspects of the pro- mon of which were minimum and maximum allocation limits, and cess were made public, and many available descriptions were difficult caps on variance between successive allocations. In addition, some to decipher. Qualitative adjustments to the results of formulae were institutions define spending targets to ensure that allocation deci- commonly mentioned, but very challenging to document. sions are aligned with institutional principles. For instance, UNICEF Negotiations around earmarking were not disclosed, either, and such has a target for programme allocations of 50% to Sub-Saharan funds are not subject to the normal allocation process at all. At Africa and 60% to countries classified as LDCs (UNICEF, 2012b). UNICEF, UNDP and WHO, >60% of total funds earmarked. We However, as a few informants explained to us, other adjustments recognize that confidentiality may be appropriate in stages, but we are used to account for important factors that resist quantification, argue that there should be clarity about all parts of the process, even for instance, absorptive capacity, the likelihood of corruption, past if the operational details of some steps are kept confidential. We con- performance or current political issues. Those adjustments rely on clude that further efforts are needed to improve the accountability internal data that are not made public. At UNDP, qualitative adjust- and the transparency of the decisions that shape resource allocation. ments are defined internally and approved by the Administrator. At From an ethical perspective, this study raises concerns about the the GFATM, qualitative adjustments are made to take into account representation of countries in resource allocation, which relied al- a wide range of considerations including past program performance, most exclusively on internal institutional processes. In some organ- risk and absorptive capacity. izations, countries are consulted individually about different parts of In most cases, such qualitative adjustments are not easily charac- the process, such as potential funding priorities, or implementation terized, although it is clear from our review that they can be import- details. However, country participation is limited during the wider ant, as at UNDP where 30% of core resources are allocated this decision making process. At the GFATM, countries are engaged at way. We could not document how these adjustments are made or the end of the process to negotiate their own allocation, although understand in detail what they reflect. Downloaded from https://academic.oup.com/heapol/article/33/suppl_1/i4/4835243 by DeepDyve user on 16 July 2022 Health Policy and Planning, 2018, Vol. 33, Suppl. 1 i13 Dieleman JL, Graves C, Johnson E et al. 2015. Sources and focus of health de- only at the margin and in exceptional cases. As a result, countries’ velopment assistance, 1990–2014. JAMA 313: 2359–68. expressions of need and inputs seem to have only a limited effect on EAI (2015a). Updated–Terms of Reference for the Equitable Access Initiative. how overall funding envelopes are split between activities or coun- Equitable Access Initiative. https://www.theglobalfund.org/media/1320/ tries. A resource allocation cycle that is more inclusive of country eai_equitableaccessinitiative_tors_en.pdf. participation could foster better alignment of needs and allocation. EAI (2015b). Meeting Report: Equitable Access Initiative 23 February 2015, We also observe that allocation processes embody ethical prin- Geneva. Equitable Access Initiative. https://www.theglobalfund.org/media/ ciples, which we believe should be made more explicit. For instance, 1319/eai_2015-02-23-meeting_report_en.pdf. in its allocation process, Gavi prioritizes impact; in contrast Easterly W. (2008). Reinventing Foreign Aid. Vol. 1. MIT Press. GFATM prioritizes equity and stability, and the UN system institu- Hafner T, Shiffman J. 2012. The emergence of global attention to health sys- tions emphasize health need. Such discussion is important to pro- tems strengthening. Health Policy and Planning 28: 41–50. Hoeffler A, Outram V. (2008). What Determines the Allocation of Aid? The mote fairness, and could be conducted on a participatory basis. At Centre for the Study of African Economies Working Paper Series, 299. present, the types of support are constrained by politics, and none of Hotez PJ, Alvarado M, Basa ´ nez ~ MG et al. (2014). The global burden of disease the institutions are fully accountable to the countries or citizens they study 2010: interpretation and implications for the neglected tropical dis- intend to benefit. As a personal conviction, we believe that actual eases. PLoS neglected tropical diseases 8: e2865. and potential recipients have a right to know how allocation deci- Hristos D, Paldam M. 2009. The aid effectiveness literature: the sad results of sions are made and a right to representation in the process. We con- 40 years of research. Journal of Economic Surveys 23: 433–61. tribute this article as a step towards realizing these rights. Institute for Health Metrics and Evaluation (2014). Financing Global Health Visualization. Seattle, WA: IHME, University of Washington. http://vizhub. healthdata.org/fgh/. Supplementary Data McGillivray M. 2003. Modelling aid allocation: issues, approaches and re- Supplementary data are available at HEAPOL online. sults. Journal of Economic Development 28: 171–88. Ottersen T, Kamath A, Moon S, Røttingen JA. (2014). Development assist- ance for health: quantitative allocation criteria and contribution norms. Acknowledgements Centre on Global Health Security Working Group Papers, London: Chatham House. The authors would like to gratefully acknowledge the support of the Ottersen T, Norheim OF. 2014. Making fair choices on the path to Wellcome Trust in the funding this study, as well as the help of the many con- universal health coverage. Bulletin of the World Health Organization tacts and informants within each institution. 92: 389. Parkhurst JO, Vulimiri M. 2013. Cervical cancer and the global health agenda: Funding insights from multiple policy-analysis frameworks. Global Public Health 8: 1093–108. This work was supported by the Wellcome Trust [099114/Z/12/Z]. Piva P, Dodd R. 2009. Where did all the aid go? An in-depth analysis of Conflict of interest statement. None declared. increased health aid flows over the past 10 years. Bulletin of the World Health Organization 87: 930–39. Reich MR. 1995. The politics of agenda setting in international health: child References health versus adult health in developing countries. Journal of International Development 7: 489–502. Bourguignon F, Sundberg M. 2007. Aid effectiveness: opening the black box. Roberts M, Hsiao W, Berman P, Reich M. (2003). Getting Health Reform American Economy Review 97: 316–21. Right: A Guide to Improving Performance and Equity. Oxford University Bump JB, Reich MR, Johnson AM. 2013. Diarrhoeal diseases and the global Press. health agenda: measuring and changing priority. Health Policy and Roberts MJ, Reich MR. 2002. Ethical analysis in public health. The Lancet Planning 28: 799–808. 359: 1055–9. Clemens MA, Radelet S, Bhavnani RR, Bazzi S. 2011. Counting chickens Shiffman J, Smith S. 2007. Generation of political priority for global health when they hatch: Timing and the effects of aid on growth. 122: 590–617. initiatives: a framework and case study of maternal mortality. The Lancet Daniels N. 2000. Accountability for reasonableness: Establishing a fair process 370: 1370–9. for priority setting is easier than agreeing on principles. BMJ: British The Paris Declaration on Aid Effectiveness. (2005). http://www.oecd.org/dac/ Medical Journal 321: 1300. effectiveness/45827300.pdf. Daniels N. (2007). Just Health: Meeting Health Needs Fairly. Cambridge UNITAID. (2013). Strategy 2013–2016. UNITAID. University Press.

Journal

Health Policy and PlanningOxford University Press

Published: Feb 1, 2018

Keywords: world health; resource allocation

There are no references for this article.