Unanswered Questions on the Global Fund’s New Allocation Methodology

This is a joint post with Victoria Fan.

The Global Fund’s New Funding Model (NFM) was approved by its Board more than a year ago, representing what the Fund’s Director Mark Dybul called “a new beginning” to “achieve greater impact in the lives of people affected by HIV and AIDS, TB and malaria.  A key piece of the NFM is an allocation methodology that aims to inform how the Global Fund distributes funding between countries and among disease based on objective criteria  of disease burden and country ability to pay [1].  This should demonstrate a marked improvement from past allocation measures which funded proposals on a first-come, first-served basis until the money ran out [2].

The NFM’s allocation methodology was intended to solve mismatches between disease burden and Global Fund money, and to increase the predictability of available funding to countries. In recent work, we find that in the past most top-25 HIV grant recipients received disproportionately high funding relative to their disease burden, ability to pay, and performance. There were enormous disparities in per-capita allocation between countries that would appear similar on a variety of criteria (including disease burden, population at risk, income, and performance capacity), with a number of countries receiving too much or too little than might be expected (see illustrative figure below based on one criteria).

Figure 1. Historical Global Fund HIV/AIDS spending (%) vs global HIV cases by country (%)

The new allocation methodology was intended to fix this imbalance, which in turn might drive quicker progress against disease. At the 30^th Board Meeting, the Strategy, Investment and Impact Committee (SIIC), as requested by the Board, approved the parameters to be used by the allocation methodology for the 2014-2016 period. Aidspan published a summary and a description of the new approach, and the Global Fund recently put out a Resource Book for Applicants.

Yet the justification for the choices made as part of the NFM allocation methodology remain unclear and sometimes perplexing, with uncertain effects on the money-disease mismatch.  Here’s how we see it:

The new allocation approach is composed of three major steps:

First, there is a “global disease split” that is “determined by the Board,” resulting in 50% HIV, 32% malaria and 18% TB. This split is not related to the relative burden of disease or the relative cost-effectiveness of the interventions to battle each disease, and is instead the historical distribution of Global Fund monies amongst diseases. This decision implies a missed opportunity to reduce mismatch between money and burden. 

Next, the allocation formula is calculated within each disease. In the diagrammed example below from Aidspan, the number of HIV cases (HIV burden) would be multiplied by the ability to pay factor (could be interpreted as the proportion of the HIV cases that will be externally funded, 95% in the case of low-incomes according to the guidance). The product generates a number of cases that is called a “country score for HIV” that is then divided by the sum of all country HIV cases. This generates some proportion of cases that is applied to total available HIV funding, termed the “notional HIV amount.” Basically, this is just taking HIV cases that will (might?) be externally funded in full or in part and dividing up the money among them.  This approach makes sense if the idea is to pay only for HIV treatment, if the share of externally funded cases is approximately accurate and if the unit cost of treatment is reasonably ascertained in low-income countries, none of which may be valid assumptions. 

The measures of disease burden to be used in the formula are inconsistent across diseases (see table below), and for TB and malaria, would lack the rationale that makes the HIV formula reasonable if treatment were our only objective. The disease burden indicators were based on recommendations from technical partners (WHO, STOP TB, RBM, UNAIDS).

Finally, after several other steps, there is a provision for “graduated reductions,” meaning that countries that are supposed to receive less given the new allocation formula will not actually receive less for some period of time, “2017-1019 and beyond” according to the Aidspan document, with the rationale of maintaining the gains made and the need to responsibly transfer programs to host country governments or other donors.

In the absence of an official explanation of the decisions that underlie this new approach, we have many unanswered questions:

• Why aren’t consistent measures of disease burden used to allocate amongst diseases in a transparent way? Doing so might move more money towards malaria, since mortality from malaria is concentrated among children under 5 – wouldn’t this be a good thing?
• Why should MDR-TB cases be prioritized at 8 times the number of drug-susceptible TB or HIV/AIDS cases?
• What kinds of incentives exist to improve performance on disease reduction in the new allocation approach?
• What explains the “kinky” sliding scale of national incomes that groups countries into bands for purposes of ability to pay, not distinguishing between the ability to pay of a country at $4,000 GDP per capita and one at $10,000 GDP per capita?
• Why are the base years for measuring the disease indicators different, such that malaria allocations depend on what was happening with the disease 13 years ago?
• Why put countries in “bands” adding another layer of complexity (as raised by some Board members in the 29^th Board Meeting[3])?
• How exactly should the allocation methodology based on objective criteria be reconciled with historical commitments for treatment? Or as implied by the Global Fund, how will it “ensure sufficient indicative funding to satisfy existing grant commitments”[4]? Is this the same or different than the “graduated reduction” strategy?

All of this adds up to an allocation approach that has not yet answered how the new method - in combination with graduated reductions - actually fixes the old problem of funding imbalances, or better value for money using the allocation of resources.

In our analyses, we find that the NFM allocation methodology could –if implemented without graduated reductions – reduce discretionary or “suboptimally” allocated funding by about 40%.  Importantly, we also find that a clearer, more theoretically justifiable allocation formula could reduce “suboptimally” allocated funding by 20-50% more than the NFM method.  But our analyses are based on guess work since we don’t really know how the allocation process works in detail. 

The new allocation methodology is a step forward. But the Global Fund Board and its partners can and should better explain the rationale for choices taken, and better document the impact of the new formula on recipients and money-burden mismatches. Most importantly, we hope that the Global Fund Board will revisit the allocation methodology in the coming years and adjust as needed – we’ve offered some options for different approaches to allocation in a draft paper.

Notes

1.       See p. 4 of the NFM Transition Manual

2.       See p. 23 of the high-level report

3.       See p. 13 of the Report of the 29^th Board Meeting

4.       See p. 22 of the Report of the 29^th Board Meeting