n.V1-4.01 | WHAT "BOTTOM-UP" CERTIFIES

01| Bottom-up, community-led, locally led, participatory, community-driven. Five words for the same claim, and the claim carries weight: work described this way is presumed better than work described as top-down. Funders ask for it, applicants promise it, evaluations record it. Before arguing about whether it works, there is a prior question almost nobody asks. What does the label actually certify?

02| The field is not short of instruments for answering that. It built two of them, and it built them early.

03| Sherry Arnstein published her ladder of citizen participation in 1969. Eight rungs in three bands. Manipulation and therapy she called nonparticipation — putting residents on advisory bodies in order to educate them or engineer their consent. Informing, consultation and placation she called tokenism: arrangements that look like power-sharing and are not. Only the top three rungs — partnership, delegated power, citizen control — involve citizens actually deciding anything. Her opening sentence has been quoted for fifty-six years and ignored for fifty-six years:

"There is a critical difference between going through the empty ritual of participation and having the real power needed to affect the outcome of the process."— Sherry R. Arnstein (A Ladder of Citizen Participation, Journal of the American Institute of Planners, Vol. 35, No. 4, 1969, p. 216)

04| Jules Pretty's typology, from 1995, runs seven categories from passive participation — people are told what has already been decided — up to self-mobilisation, where people act independently of external institutions. The ordering principle is how much decision-making authority the external agent keeps. Consultation, in his definition, concedes none: outsiders define the problem, control the analysis, and are under no obligation to act on what they hear. Functional participation — the kind justified on efficiency grounds, the kind that gets the project built — happens after the main decisions have been taken elsewhere.

05| Both instruments locate the distinction in the same place: decision rights over resources. Not who implements. Not who is employed. Not where the receiving organisation is registered.

06| Now the operational standard. The Grand Bargain, agreed in 2016, committed its signatories to routing 25 percent of global humanitarian funding to local and national responders "as directly as possible" — a phrase given the working meaning of reaching them through up to one intermediary. USAID's localisation policy set two targets: 25 percent of funding direct to local partners by the end of FY2025, and at least half of programmes empowering local actors to lead on priority setting, design, implementation and evaluation by 2030.

07| Read the USAID pair slowly, because it contains both kinds of standard sitting side by side. The 25 percent target counts contracts — who receives money. The 50 percent target counts authority — who decides. The first can be reported this quarter. The second requires exactly the judgement Arnstein's ladder was built to make. Only the first is routinely reported.

08| The results against the countable one are the central fact of this chapter. In 2023, local and national actors directly received $174 million from Grand Bargain donors: 0.6 percent of first-level funding. In 2022 it was 1.2 percent, $485 million. In 2021 direct funding fell by almost two-thirds, to $302 million — the lowest volume and share in five years. Seven years after a public commitment to a quarter, delivery has stayed at or below one percent. The target was missed by a factor of about twenty-five, with the commitment restated at every summit in between.

09| USAID reports better: $2.1 billion in FY2024, or 12.1 percent. The denominator is contested. Independent analysts note the figure is calculated against roughly 43 percent of the agency's total resources rather than all of it, and that the definition of "local" does substantial work. Change either choice and the number moves materially. The body reporting the number controls both choices.

10| So what does the label certify? On this evidence, a contracting relationship. An organisation registered in the country of operation, receiving a subgrant to deliver activities designed elsewhere, against outcomes defined elsewhere, reported against metrics defined elsewhere, counts as local funding. On Arnstein's ladder that sits in the tokenism band whatever the recipient's address. In Pretty's scheme it is functional participation: people brought in to deliver someone's objectives more efficiently, after the decisions.

11| This is not an accusation of bad faith, and it should not be read as one. It is what happens when a question about power gets operationalised as a question about money flow, because money flow is what can be counted. The proxy then becomes the definition, and after a while nobody remembers it was a proxy.

12| The last part is the most useful. Does the label predict outcomes? The World Bank supports roughly 400 projects it classifies as community-driven, in more than 90 countries, worth close to $30 billion. Casey, Glennerster and Miguel found short-run gains in local public goods from such a programme in Sierra Leone and no sustained effect on collective action, decision-making, or the inclusion of marginalised groups. Mansuri and Rao's review documents elite capture across the induced-participation portfolio. The label does not predict success. It does not predict failure either. It predicts nothing, which is worse than either, because a signal that carries no information still gets used as one.

13| Monique Deveaux borrows a test from Srilatha Batliwala that does more work in one line than the whole certification apparatus: these are not organisations that mobilised a constituency, their constituents created them. Applied to almost any list of community-led organisations, it separates two or three from the rest.

14| Anyone using the word owes a specification. Which decision rights are held locally, by whom, over what money, and who can take them back. Unspecified, "bottom-up" certifies a registered address. Specified, it becomes a claim someone can check — and a claim someone can check is a claim that can be wrong, which is the whole point.