Can Domestic Budgets Carry Africa’s Health Systems?
What the first year of evidence means for the Accra Reset.
On this page
- The scale of the contraction
- Is domestic revenue replacing external financing?
- Which functions absorb the cut?
- When does a cut to prevention become a deferred liability?
- What can efficiency and earmarked taxes contribute?
- Are the institutions and the workforce ready?
- What financiers owe the transition
- What an accountability mechanism should measure
- Conclusion
Between the final quarters of PEPFAR’s 2024 and 2025 fiscal years, the number of people on HIV treatment in programmes it supports barely moved. Over the same comparison, new pre-exposure prophylaxis starts fell by 41 per cent and HIV testing by 17 per cent. That pattern, shaped partly by which services the United States chose to keep funding, is an early signal of where a steep contraction in external health financing bites first.
The Accra Reset High-Level Panel (2026), whose report was presented at the UN General Assembly in September, asks governments to make domestic resources the foundation of their health systems. External financing would be redirected towards managed transitions, capacity development, fragile settings, emergencies, regional functions and global public goods. This issue looks at what the first year of the contraction suggests about the conditions for that transition to succeed. It looks first at whether domestic revenue is replacing external financing and which health functions absorb the cut. It then asks when a cut to prevention becomes a deferred liability, what efficiency can contribute, and whether the institutions the transition relies on are ready.
The scale of the contraction
Estimates differ by method and period, but all point the same way. Apeagyei and colleagues (2026) estimate that development assistance for health worldwide fell from US$81.3 billion in 2021 to US$44.9 billion in 2025. Their study is an Institute for Health Metrics and Evaluation preprint that has not been peer reviewed. IHME’s earlier preliminary estimates, reported by Sriram (2026), put the fall between 2024 and 2025 at about 21 per cent, most of it from a 67 per cent cut in United States funding.
For Africa the fall appears steeper. IHME’s July 2025 estimates put the 2024–25 decline for sub-Saharan Africa at 25 per cent. The Panel, citing an estimate by Africa CDC’s Director-General published in The Lancet, records that external financing for health in Africa contracted by close to 70 per cent between 2021 and 2025; that figure uses a different method and is not directly comparable with IHME’s. Both multi-year estimates start from 2021, when pandemic-era financing peaked, so each overstates the fall relative to pre-pandemic levels. Even so, IHME’s July 2025 estimates placed global development assistance for health in 2025 at its lowest level since 2009.
Exposure is uneven. Asamani and colleagues (2026) report that external sources financed 23.9 per cent of current health expenditure in the WHO African Region in 2022. In at least five countries, external financing accounted for 45 to 65 per cent.
Is domestic revenue replacing external financing?
On the evidence so far, not quickly. Westgard (2026), in a preprint covering 71 aid-dependent countries worldwide over 2010 to 2023, finds no evidence that government health spending replaced development assistance when it fell. Its confidence intervals are wide, but even the upper bound falls far short of what would offset the 2025 contraction, and the period ends before the 2025 cuts. A Center for Global Development review (Clements, Gupta and Ward, 2026) of budget documents in 18 aid-exposed sub-Saharan African countries found policy responses to the aid cuts in only two of them. In-year reallocations may not appear in those documents.
Two decades of data give little reason to expect a rapid shift in budget priority. Uzor and colleagues (2026), examining 53 African Union member states, find that the average share of government expenditure going to health rose by 0.5 percentage points between 2001 and 2022. South Africa was the only member state meeting the Abuja target of 15 per cent in 2022. The indicator measures priority within the budget, not the level of spending.
Some replacement has occurred. The Panel cites Ghana’s removal of the cap on its earmarked National Health Insurance Levy, which it credits with a GHS 3.4 billion increase in the health budget, and a US$200 million increase in Nigeria’s health budget. For HIV across low- and middle-income countries, UNAIDS (2026) reports that in 2025 international resources fell by 18 per cent while domestic funding rose by 4 per cent. Domestic sources now provide about 60 per cent of the US$17.6 billion available. These increases are real but smaller than the external decline, and budget outturns for 2026 and 2027 will be the clearer test.
Part of the explanation is fiscal. Herrera and Olaberría (2020), writing on Latin America and the Caribbean, describe wages, pensions and debt service as naturally inflexible components of public spending. The IMF’s April 2026 Regional Economic Outlook reports that fiscal deficits in 21 sub-Saharan African countries exceed the level required to stabilise debt. Its October 2026 update estimates that interest payments absorb a median of about 12 per cent of government revenue, excluding grants, over 2025–27. In the short run, reprioritisation towards health therefore has to come from a margin already narrowed by interest costs, wage commitments and, in many countries, further fiscal adjustment.
Which functions absorb the cut?
PEPFAR data for the fourth quarter of its 2025 fiscal year (July to September 2025) show a broadly consistent pattern, though KFF’s analysis (2026) describes several results as mixed. It reports 20.3 million people on antiretroviral treatment, against 20.4 million a year earlier. New enrolments on treatment fell by 16 per cent and HIV tests by 17 per cent, although testing remained above its 2021–23 levels. Pre-exposure prophylaxis initiations dropped by 41 per cent and completions of DREAMS, a programme for adolescent girls and young women, by 86 per cent.
The Global Fund (2026) reports the same direction across the more than 100 countries where it invests, not only in Africa. People on treatment rose to 26.9 million, while recipients of pre-exposure prophylaxis fell to 1.1 million from 1.4 million. The Global Fund attributes the fall in prophylaxis to reductions in both domestic and international funding. Because the two programmes overlap in many countries, these are not independent confirmations.
Three caveats apply. Treatment is a stock while testing and prophylaxis are flows, so some divergence is expected in the short run. DREAMS reporting became optional during PEPFAR’s 2025 fiscal year, so part of that fall may reflect reduced reporting rather than fewer services. And under the limited waiver of 1 February 2025, treatment, HIV testing, prevention of vertical transmission and prophylaxis for pregnant and breastfeeding women could continue, while other prophylaxis and most other prevention could not. Part of the decline in prophylaxis and DREAMS therefore reflects a donor decision about which services to fund. It is not, on its own, evidence of country-level reallocation.
As domestic budgets take over these functions, there is a risk that they reproduce part of the pattern through a different channel. When revenue falls short, ministries of finance ration cash in a predictable order. Ally and Piatti-Fünfkirchen (2021) describe it for Tanzania: wages and debt are paid first, and development expenditure and goods and services “then bear the brunt of the burden”. Barroy and colleagues (WHO, 2019) record execution rates in the Democratic Republic of the Congo between 2011 and 2015 of 94 per cent for staff costs and 32 per cent for non-staff spending, and describe discretionary spending as “more fragile to low execution”. This order protects wages, not treatment as such, since medicines and test kits are also non-staff spending. The risk falls most on outreach, supervision, sample transport and community programmes, which depend heavily on fuel, allowances and other operating costs.
Preventive services also account for a small share of spending. Fuady and colleagues (2024), in a study of Indonesia, report that preventive care accounts for about 11 per cent of health spending in low- and middle-income countries, against 54 per cent for curative care, citing the WHO Global Health Expenditure Database.
When does a cut to prevention become a deferred liability?
Where transmission can rebound, a cut can create a larger cost later. Cohen and colleagues (2012) reviewed 75 malaria resurgence events across 61 countries. They attribute 91 per cent at least in part to the weakening of control programmes, most often because of resource constraints. The review examined resurgences that occurred, so it describes how resurgence happens. It cannot show how often cuts cause it.
Mauritius is an extreme but well-documented case. Tatarsky and colleagues (2011) record elimination in 1969 and reintroduction in 1975, which a WHO review linked partly to relaxed case detection after certification. A second elimination followed in 1998. The 1982–88 campaign cost an average of US$4.43 per person per year, against US$2.06 for the prevention-of-reintroduction programme in 2008, both in 2008 US dollars.
A government under a binding budget constraint can reasonably give less weight to returns that arrive late. Haacker, Hallett and Atun (2020) argue that economic evaluations in low- and lower-middle-income countries should use a discount rate of at least 5 per cent, above the conventional 3 per cent. Higher rates reduce the present value of late benefits, as many preventive benefits are, but they equally reduce the weight of costs pushed into the future. The case against a cut therefore rests on the later cost being much larger, not merely later. Not all prevention is equally valuable, and antiretroviral treatment itself prevents transmission. The case for protection is strongest where reversal is costly, as with transmissible disease close to control.
What can efficiency and earmarked taxes contribute?
Efficiency gains come in two kinds. Some release cash, for example a lower delivered price for a commodity. Others raise the services obtained from the same budget, for example better attendance or prescribing. The second kind matters, but it creates no money a ministry of finance can move.
Evidence that efficiency releases cash is thin, even where it is best documented. Finlay and colleagues (2026) reviewed 337 UK sources from 2019 to 2025 on efficiency interventions aligned with NHS England’s priorities. None contained enough information to show, against five criteria the authors set, that an intervention had released cash. Cash savings are also visible to a treasury. Without a rule allowing the health sector to retain them, a successful procurement reform can reduce the next health allocation.
Payroll clean-up is one African example of efficiency that can release cash. Reconciling payrolls against staff records removes ghost workers. Ayehu and colleagues (2026) cite 26,589 removed from Ghana’s government payroll in 2017 and about 30,000 identified in Mozambique between 2015 and 2017.
The record on earmarking is mixed. Cashin, Sparkes and Bloom (WHO, 2017) note that budgets are fungible, so an earmarked revenue source is likely to be offset by cuts elsewhere. Cashin (World Bank, 2016) records that Kazakhstan introduced a 3 per cent earmarked payroll tax for health in 1996, yet government health spending fell from 3.0 to 1.5 per cent of GDP between 1995 and 1998, during a wider post-Soviet fiscal collapse. Ghana’s recent experience, cited above, suggests that removing a cap on transfers can allow an earmark to add to the health budget.
Are the institutions and the workforce ready?
The Panel’s recommendations on pooled procurement and regional regulation depend on institutions that are still being built. The usual benchmark is the Pan American Health Organization’s Revolving Fund for vaccines, which has operated for more than 40 years. PAHO states that its member states would otherwise have paid at least 75 per cent more for the region’s 13 most-used routine vaccines. That Fund rests on working capital: Seraj Ahmad and colleagues (2022) put it above US$120 million, against annual purchases of about US$1.1 billion. The African Pooled Procurement Mechanism, endorsed by the African Union in February 2024, announced in May 2026 the results of its first tender for reproductive, maternal and newborn health medicines, covering ten products for ten member states. Africa CDC reports prices 30 to 90 per cent below national benchmarks. Like PAHO’s figure, this is the institution’s own estimate; contract volumes have not yet been published.
Regulatory capacity is improving from a low base. Okereke (2025) records that the number of African regulators assessed by WHO at maturity level 3, which denotes a stable, well-functioning system, rose from two in 2021 to nine by September 2025. Mozambique’s listing brings the total on WHO’s list as of August 2026 to ten, counting South Africa for vaccines only. The African Medicines Agency became operational only in 2025. In August 2026, health ministers at the WHO Regional Committee for Africa endorsed a regional strategy (AFR/RC76/7) that expects the Agency to implement at least 70 per cent of maturity level 3 sub-indicators by 2029, from a 2026 baseline of zero that is expected for so new an agency. None of the 41 WHO-Listed Authorities designated by December 2025, the status WHO now uses in place of “stringent regulatory authority”, is in Africa.
The workforce data point in two directions. The WHO Regional Office for Africa (2026) estimates, by extrapolation from nine countries, about 943,000 potentially unemployed health workers in the WHO African Region in 2024. It also projects a needs-based shortage of 5.85 million by 2030. The explanation is contested. Nwadiuko, Doua and Nyawira (2025) point to restricted public wage bills alongside overproduction of graduates. The WHO Regional Office attributes unemployment mainly to fiscal constraints, recruitment ceilings and slow public recruitment rather than oversupply. An earlier World Bank study, Labor Market for Health Workers in Africa (2013), argued that wage-bill ceilings were rarely binding and pointed instead to rigid public employment policy and under-investment in training. Either way, many trained workers are not being absorbed into funded posts despite large needs-based shortages.
The final two sections set out proposals.
What financiers owe the transition
Donor transition policies place most of their explicit conditions on countries. The Panel asks countries for a single national health plan with a costed, multi-year investment plan built into the national budget, and for time-bound domestic financing targets. Financiers add their own requirements: the Global Fund, for example, releases at least 15 per cent of a country’s allocation only if co-financing commitments are made and met. The United States has signed bilateral health agreements, most covering 2026–2030, with 36 countries; according to KFF, these set US funding 34 per cent below the previous five years and pair it with country co-investment commitments.
The Panel asks a great deal of financiers too: to align fully behind national plans, reform their operating models, avoid abrupt withdrawal, give countries a say over the pace of transition and submit to independent tracking. It calls for transitions that are “paced, jointly managed and supported”, and the Lusaka Agenda (2023) asks global health initiatives for transitions that are gradual and “clearly planned, communicated and coordinated”. Building on both, this issue proposes that the financier’s side be specified with the same precision as the country’s. That means a minimum notice period before a funding line is reduced, and a published schedule of reductions in place of mid-year shortfalls. It also means handing over, in working order, the surveillance, laboratory, supply chain and data systems that external programmes built.
What an accountability mechanism should measure
The Panel’s tenth recommendation calls for an independent follow-up and accountability mechanism, to be established within six months of the report’s launch. The indicators it adopts will decide what it can see. The domestic share of health expenditure rises mechanically whenever external financing falls faster than domestic spending, even if government spending is flat or falling, or if households are paying more out of pocket. Most of the following are established measures, and several give operational form to the Panel’s own call for tracking from budgets and commitments through to results. This issue proposes them as the mechanism’s core set:
- Government health expenditure per capita in constant prices, so that a rising domestic share is not mistaken for a rising level. The African Union’s Africa Scorecard on Domestic Financing for Health already reports a version of this.
- The share of spending on prevention, and volumes of new treatment enrolments, testing and prevention services, so that the pattern seen in 2025 is measured directly.
- Budget execution by programme, at a stated stage of spending, since figures based on commitments can look healthy while payments lag. The stock and age of unpaid bills, which PEFA assessments already examine, should be reported alongside.
- Delivered unit prices for a defined basket of commodities, with stock-out days, to test whether pooled procurement saves money.
- Filled posts against approved posts and staffing norms, with the number of qualified health workers not in employment, to track whether trained workers are being absorbed.
- Catastrophic health spending (SDG indicator 3.8.2) and unmet need, by income and place, since a transition can preserve national averages while shifting costs onto poorer and rural households. These data come from household surveys every few years, so they sit alongside the annual measures.
- Disbursements against commitments for each external financier, with the notice given before any reduction, so that the mechanism tracks financiers as well as governments, as the Panel intends.
Conclusion
The first year of the contraction shows the direction of adjustment, not yet its size. Domestic substitution has so far been limited, with notable exceptions such as Ghana and Nigeria. Treatment numbers held while new enrolments, testing and prevention fell, partly by donor design. The regional institutions the transition depends on are at an early stage, though some are building quickly. The Panel’s direction is sound, and the independent accountability mechanism it calls for, covering financiers as well as governments, would provide the means to track progress. The measures above are offered in that spirit, to show what happens to services and to household spending as the transition proceeds.
Systems and Sovereignty examines the financing, institutions, and evidence behind health policy in Africa. Published by EPIC Health Systems.
Systems and Sovereignty by email
New issues, sent when they are published.













