n.V1-5.05| THE MISSING DIVISOR

01| Every project has a budget. Every budget breaks into line items. Divide a line item by the number of things it delivered and you have a unit cost: what one latrine cost, one water connection, one land title, one metre of paved footpath. These numbers exist inside the files of every agency that has ever built anything.

02| So it should be possible to collect them. Gather enough project documents, extract the unit costs, and you would have something this sector has never had — a price list, against which any new proposal could be checked. The idea is obvious enough that it has occurred to a great many people. What happens when someone tries is the useful part.

03| Assemble ten unit costs from published sources — slum upgrading per household, a formal housing unit, a serviced plot, a land title, a latrine, simplified sewerage, the overhead on a cash transfer. Each figure came from a project document, a costing study or an evaluation. None of them was invented. The list looks authoritative.

04| It cannot be used to choose anything, and the reasons are worth being exact about.

05| The denominators differ. Some figures are per household, some per housing unit, some per household per year, some per dollar transferred. A one-off capital cost and an annual operating cost are not the same kind of number, and putting them in adjacent rows of a table implies a comparison the arithmetic does not support.

06| The dates differ. Figures from a 2004 assessment and a recent costing study are separated by two decades of inflation, currency movement and construction cost change. Without adjustment to a common year and a common basis, the table compares money that does not have the same value.

07| The scope differs. A slum upgrading figure of a few tens of dollars per household and one in the low thousands are both real and describe different things — one a minimal intervention, the other a fuller package. The gap between them is mostly definitional, not a finding about efficiency.

08| And the source populations differ. A careful sanitation costing in one South African township is good evidence about that township. It becomes something else entirely when it is presented as the cost of sanitation.

09| None of this makes the exercise worthless. It means the first honest output is not a price list. It is a demonstration that the sector has no shared unit of account: no agreed definition of a household reached, no convention on capital versus lifecycle cost, no standard price year. That absence is more interesting than any individual number in the table.

10| The obvious response is transparency — make agencies publish and comparison becomes possible. That was tried, deliberately and at scale. The International Aid Transparency Initiative was built to do exactly this: a common standard for publishing development activity data, adopted voluntarily by donors, agencies and implementers, with a large and growing body of published data behind it. Open contracting standards did comparable work on procurement.

11| It worked, and it did not produce a price list. What IATI publishes is activity and transaction data — who funded what, through whom, for how much, when. To get from a transaction record to a unit cost you need the divisor: how many latrines, of what specification, serving how many people, working for how long. That is precisely the part that is not standardised, not required, and often not recorded in a form anyone else could use.

12| So the obstacle was never concealment. Three decades of transparency reform did what they set out to do. The money is increasingly traceable. What is missing is the bottom half of the fraction.

13| There is one recent case where an agency forced the comparison instead of waiting for the data to assemble itself. USAID began commissioning cash benchmarking studies: rather than asking whether a programme worked, asking whether it worked better than handing the same money directly to the people it was meant to help. Cash makes a useful yardstick precisely because its administrative cost is low and its household effects are well documented, so it sets a floor any more elaborate programme ought to clear.

14| In Rwanda, a youth workforce readiness programme was tested against cost-equivalent cash grants, with results published by Craig McIntosh and Andrew Zeitlin. On most primary outcomes the programme did not beat simply giving people the money.

15| Two things about that matter more than the result. The comparison had to be commissioned as a bespoke research exercise, at considerable cost, because no existing data could answer the question. And it remains rare — the Center for Global Development, reviewing the effort, described cost-effectiveness benchmarking at USAID as new and niche, requiring a champion inside the agency to advance it. Which is another way of saying that asking whether the money would have done more elsewhere is not something the system does unless a particular person makes it happen.

16| What would produce comparability is duller and harder than transparency: an agreed definition of the unit, a convention for capital versus lifecycle cost, a common price year, and a requirement to report the divisor alongside the spend. None of it is technically difficult. All of it requires an organisation to make its costs comparable to a rival's. That is a different kind of ask, and it is the whole explanation for why we still have no price list.