Hoverview

consortium · funding · localisation · logframe · sub-grantees

When your output is their outcome

Sub-granting produces two logframes over one delivery, sitting at different levels. What that costs, and who pays it.

A local organisation in northern Nigeria runs a nutrition screening programme. In its own results framework, the top-level result is a reduction in acute malnutrition among children under five in the target wards. That is what the organisation is accountable for, what its board reviews, and what its staff describe when asked what the programme achieves.

The same programme appears in an international NGO's logframe as an output: children screened for acute malnutrition, disaggregated by age and sex. The INGO's outcome sits a level above, covering nutrition outcomes across four states. And in the donor's results framework, the INGO's outcome is itself an intermediate result under a country-level objective.

Three organisations, three frameworks, one set of screenings. The child was screened once. The screening is a top-level outcome, a mid-level output, and a component of an intermediate result, depending on which document you are reading. Nobody is wrong, and nothing about the delivery changed between the three descriptions.

This is not a bug in one funding chain. It is the shape of most funding chains, and the method was designed to handle it. What has not been designed is the reporting infrastructure underneath, which means the cascade works on paper and fails in practice at the point where figures have to move up the chain with their meaning intact.

The cascade was in the original design

Rosenberg and Posner's 1979 formulation for Practical Concepts Incorporated, available now only through a third-party mirror, describes exactly this situation. When a programme has both a programme-level logframe and project-level logframes underneath it, each of the projects needed to achieve the programme's purpose appears as an output on the programme logframe, and that same output becomes the purpose on the project manager's own logframe.

The design intent is clear. The logframe was built for hierarchical management, where each level of management holds a frame appropriate to its own span of control. A programme manager's outputs are the things the programme delivers. A project manager working under that programme has one of those outputs as their purpose, because it is the highest result they are accountable for. The cascade is the mechanism by which accountability nests.

Des Gasper, writing at ISS The Hague in 2000, records that this was understood in practice: Norad recommended linked logframes for exactly this reason, and his broader observation that levels are contextual rather than inherent is what makes the cascade coherent rather than contradictory. A result is not inherently an output. It is an output in the frame that treats it as one.

So the conceptual work is done, and has been for four decades. The question is what happens when the figures have to move.

What the sub-grantee is actually asked to produce

The reporting obligations flow in one direction, and they accumulate.

Under US federal rules, a prime recipient that passes funding to another organisation becomes a pass-through entity, and 2 CFR 200.332 requires it to impose on the subrecipient every requirement necessary for the federal award to be used in accordance with federal statutes and the terms of the award, plus any additional requirements the pass-through entity imposes for its own compliance purposes. The subrecipient must submit final financial and performance reports to the pass-through entity no later than 90 calendar days after the period of performance ends, so that the pass-through entity can meet its own 120-day deadline to the federal agency.

The structure is a chain of deadlines, each one earlier than the one above it. The sub-grantee reports first, the prime consolidates, the donor receives. That is the correct sequence, and it means the sub-grantee's reporting calendar is set by an obligation two levels up that it has no visibility into.

The National Endowment for the Humanities' guidance for pass-through entities states the legal position plainly: NEH has a legal and financial relationship only with its recipients, not with subrecipients. The prime is legally responsible for ensuring that activities conducted by subrecipients comply with all federal regulations. The sub-grantee has no relationship with the donor at all, which means it cannot ask the donor what a column means, cannot see the frame its output is being folded into, and cannot check whether its figure arrived in the donor's report in the form it was submitted.

That opacity runs both ways. The prime cannot easily verify the sub-grantee's number either, beyond what the sub-award agreement lets it inspect, which is why the same regulation requires risk assessment before award and monitoring throughout.

The figures move; the meaning does not

Here is where the cascade breaks operationally.

The sub-grantee reports 12,400 children screened. That figure travels up to the prime, which adds it to figures from three other sub-grantees and reports 41,000 children screened as an output. The donor receives 41,000 against an output indicator, and rolls it into a country-level intermediate result.

Three things are lost in that journey.

The first is the sub-grantee's own framing. The 12,400 was, in the sub-grantee's results logic, evidence toward a malnutrition outcome. In the prime's logframe it is an output count. The causal claim the sub-grantee was making, that screening leads to earlier referral leads to reduced acute malnutrition, does not travel with the number, because the prime's frame has no slot for a sub-grantee's outcome logic. The number arrives stripped of the argument it was collected to support.

The second is provenance. Once the four sub-grantee figures are summed, the aggregate carries no record of which organisation contributed what, or under what definition. If one sub-grantee counted a child screened twice in the same quarter as two screenings and another counted it as one, the aggregate is internally inconsistent and nothing in the reported figure says so.

The third is the ability to correct. If the sub-grantee discovers an error three months later, the correction has to travel back up a chain that has already closed its reporting period. In most systems this means either a silent adjustment to a later period, which misstates both periods, or an unrecorded discrepancy between what the sub-grantee's records show and what the donor was told.

None of these are failures of care. They are consequences of aggregating through a chain where each link holds a different frame and no link holds the mapping between frames.

The transparency evidence

The scale of the problem is visible in the localisation data, where the sector has been trying to count exactly this kind of pass-through flow for a decade.

Development Initiatives' analysis found that Grand Bargain donors provided just 0.6 per cent of funding directly to local and national actors, and that transparency of intermediaries remains a challenge. The full picture of how much funding reaches local and national actors remains obscure, despite commitments made by intermediary organisations to improve data transparency. UNHCR's publication of granular partnership data that can be independently verified against Grand Bargain definitions is noted as an improvement, which tells you how unusual it is.

That is a funding-flow problem rather than a results-reporting problem, but it has the same shape and the same cause. Money passes through an intermediary and the record of where it went is held only by the intermediary, in the intermediary's own categories. Results pass through an intermediary and the record of what was claimed is held only by the intermediary, in the intermediary's own frame.

The Charter for Change practice paper on the intermediary role puts the consequence in terms of relationship rather than data: local actors are often treated as subcontractors rather than as equal partners. A subcontractor delivers against someone else's results logic. A partner has a results logic of its own that has to be reconciled with the prime's. The reporting infrastructure assumes the first arrangement, which makes the second one expensive to sustain even where both organisations want it.

What the reporting burden work has and has not solved

The Grand Bargain's 8+3 narrative reporting template is the sector's most serious attempt to reduce what sub-grantees are asked to produce. The 2022 independent review by HPG at ODI found that more than half of grant-giving signatories were using the template in at least some form for their civil society partners, and that there is confidence among signatories using it that it reduces the reporting burden on downstream partners. The review also notes that those benefits will be maximised only when the template is used at scale, which was not yet the case.

The 8+3 template addresses narrative reporting: the questions a sub-grantee has to answer in prose. It does not address the structural problem this post is about, which is that the sub-grantee's results framework and the prime's results framework are different objects at different levels, and the mapping between them exists nowhere.

A simplified narrative template still requires the sub-grantee to describe its work in terms that will fit the prime's output row. If the sub-grantee's own outcome is "reduced acute malnutrition" and the prime's output is "children screened," the sub-grantee writes about screening, files its own outcome logic separately for its board, and maintains two accounts of the same programme. That is Gasper's double book-keeping, arriving through the funding structure rather than through the method.

What it would take

The technical requirement is narrow and the governance requirement is not.

Technically, what is missing is a record that says: this result, in the prime's frame, is at output level, and the same result, in the sub-grantee's frame, is at outcome level. One mapping, held once, applied at reporting time. With it, a figure reported at the bottom of a chain can appear correctly at every level above, and a correction at the bottom can propagate upward as a restatement rather than a silent adjustment. Without it, every level re-enters the number by hand into its own frame and the chain is only as reliable as its least careful transcription.

The governance requirement is harder. Someone has to agree what the mapping is, and the two parties agreeing are not equals: the prime holds the money and the sub-grantee holds the delivery. In practice the prime's frame wins, because the prime's frame is the one the donor will read. Making the sub-grantee's own results logic visible in the chain means the prime accepting that its output row is a summary of something the sub-grantee understands differently, and there is no incentive structure in the funding relationship that rewards that acceptance.

There is also a real objection to solving this at all. Every additional field in a sub-award reporting requirement is a cost paid by the organisation least able to afford it. A mapping that has to be maintained by both parties, reviewed when either party's framework changes, and checked at each reporting cycle, is more work than a single output count. If the sub-grantee's own outcome logic is only ever read by the sub-grantee's own board, recording it in the funding chain adds burden without adding a reader.

Whether that objection holds depends on something nobody has measured: how often a figure reported at the bottom of a funding chain arrives at the top meaning what the person who collected it thought it meant. The sector has counted funding flows through intermediaries and found the picture obscure. It has not run the equivalent exercise on results, and until it does, the argument for leaving the mapping unrecorded rests on an assumption about accuracy that has never been tested.

Sources

The method's own text

Regulatory requirements on pass-through entities

  • Subrecipient Monitoring and Management, US Department of the Treasury (February 2024), summarising 2 CFR 200. Source for the pass-through entity's obligation to impose all federal requirements plus its own on the subrecipient, and for the 90-day and 120-day reporting deadlines
  • General Guidance for Pass-through Entities on Managing Subawards, National Endowment for the Humanities. Source for the position that the funder has a legal and financial relationship only with its recipients, not with subrecipients

Transparency and the intermediary relationship

Reporting burden reform