n.V1-9.02 | Why the Money Must Move

01| In April 2017, at the World Bank's Spring Meetings in Washington, the institution's president described how his own organisation rewards its staff.

"Inside the World Bank now, if you do so well that it begins to look commercially viable, the incentives are aligned so that you quickly get it through the board and get your credit for having got that loan out the door."— Jim Yong Kim, World Bank Spring Meetings, April 2017

02| He went on to describe what he wanted instead — that turning an idea into a bankable project should earn all the points, the celebration and the promotion — and then added the line that gives the whole thing its force: "It doesn't work like that right now."

03| Be precise about what he was arguing, because the quote gets borrowed for arguments he was not making. Kim's complaint was that the Bank lends money to projects private investors could have financed, instead of preparing projects and handing them to private capital. His reform was about mobilising private investment. It was not about whether a school still functions five years after the ribbon.

04| That distinction makes the story better, not worse. The mechanism he named in passing — you get your credit for having got the loan out the door — is the subject of this chapter. He described it accurately while campaigning for something else. Even when the head of the institution stood up to demand a change in its incentive culture, durability was not the change he was asking for.

05| So the question is not whether these institutions know they reward disbursement over outcomes. Some of them say so on the record, at their own annual meetings, to reporters. The question is why saying so has changed so little.

Unspent money is a career problem

06| The mechanism has been studied for over twenty years and the finding is uncomfortably simple: money that does not go out is punished.

07| Jakob Svensson set out the logic in 2003, in work on why conditional aid does not work. Aid agencies run on budgets that must be committed and spent within a period. A department that fails to spend its allocation does not keep it, and typically receives less the following year. Budget size determines a department's weight inside the organisation — its staffing, its standing, the seniority of the jobs in it. Underspending is therefore not a neutral outcome; it is a career problem. Which is why conditions attached to loans tend not to be enforced. Enforcing them means withholding money, and withholding money is the one thing the system reliably penalises. Peter Kilby's later empirical work on World Bank disbursements found the same pattern across large numbers of loans, driven by factors well outside the merits of any individual project.

08| If the logic holds, projects get shaped by the need to spend. Money moves faster through international consultants, imported equipment and centrally managed logistics than through slow local processes, so those are the shapes projects take. Those are also the shapes that disappear fastest once the funding stops, because nothing about them was ever owned by the people left behind.

09| That last step is an inference, not a measurement, and we mark it as one. The disbursement incentive is documented. The claim that it produces top-heavy projects that evaporate is a reasonable reading of it, argued by many practitioners, and it is not the same kind of statement.

Measuring where measuring is easy

10| If the incentive is to move money, evaluation effort should drift away from the places where most money moves. It has.

11| An expert group convened by the Center for Global Development examined how evidence is actually used in aid, drawing on more than 150 studies and consultations with close to a hundred policymakers and researchers. More than a third of official development assistance goes into transport, energy and civil society. Around seven percent of development impact evaluations look at those sectors. These are the large, slow, expensive investments where the ten-year question matters most, and they are close to unexamined.

12| The rest of the picture is consistent. Two-thirds of evaluations contain no gender or equity analysis, so the question of who among the intended beneficiaries actually gained is usually not asked either. And only around a tenth of aid assessments qualify as impact evaluations at all; the rest are something less demanding.

13| You can read this as institutional self-protection. The more careful reading is that it is what happens when nobody is required to look. Evaluation is expensive, difficult in infrastructure, slow to produce results, and no career depends on the answer. The gap between where the money goes and where the measuring happens does not need a conspiracy. It needs only a structure in which nobody is worse off for not knowing.

The name for it

14| Matt Andrews, Lant Pritchett and Michael Woolcock call it isomorphic mimicry: organisations adopting the outward forms of functional institutions — the plans, the units, the manuals, the procedures — without acquiring the function. The form is what outsiders can see and reward, so the form is what gets produced. Their related idea, the capability trap, describes an organisation taking on reform after reform, each adding procedure and appearing to be progress, while its actual ability to do anything stays flat.

15| This is the spine the sandcastle complaint is missing. A project that spends its budget, files its reports, cuts its ribbon and stops working is not a failed project. It is a system producing exactly what it selects for. The broken pumps and the empty schools are not evidence that something went wrong. They are evidence that the visible part was the part that counted.

The insider version

16| The strongest form of this critique comes from a man who ran the largest bilateral aid agency in the world. Andrew Natsios was administrator of USAID from 2001 to 2006. In 2010 he argued that the measurement and compliance apparatus built to make aid accountable actively damages the work — because the things easiest to count are the least transformational, and the things that matter most cannot be demonstrated inside a reporting cycle. Faced with a system that punishes documented failure and ignores undocumented decline, staff rationally choose work that documents well.

17| Natsios has no data in the ordinary sense. He has standing and an argument that holds together, and the argument predicts the pattern the CGD numbers later showed. In a chapter about what insiders think is wrong, that combination is the point.

What actually happened to the reforms

18| One claim does not survive the record: that outcome-based approaches were proposed repeatedly and adopted nowhere. Nancy Birdsall and William Savedoff developed Cash on Delivery Aid, in which donors pay a fixed sum for each independently verified unit of progress and nothing at all for inputs, plans or activities. The World Bank launched Program-for-Results lending in 2012, disbursing against verified results rather than expenditures. Results-based financing has run at scale in health in many countries.

19| So the honest question is not why nobody tried. It is what happened to the attempts — whether paying for verified outcomes changed institutional behaviour, or whether verification became one more layer of documentation, absorbed by the compliance machinery it was built to bypass. That is precisely what the capability trap predicts: a reform adopted in form, generating new procedures and new reports, leaving the incentive untouched.

20| What is strange here is not that these institutions are unaware. The disbursement literature was published in the field's leading journal. The evaluation numbers came from a Washington think tank at the centre of the establishment. The most damaging argument about measurement was written by a former head of USAID. The reform meant to break the input model came out of the same institution that hosted the criticism. None of it is hidden.

21| Which suggests — and it is a suggestion — that self-awareness is not the constraint. An institution can hold an accurate account of its own dysfunction and continue, because the account lives in the part of the organisation that writes reports and the incentive lives in the part that hands out promotions. Those are different rooms. And nothing in the mechanism requires billions. A foundation funding a neighbourhood in three-year cycles, asking at the end of each cycle what was delivered rather than what remained standing, is running the same system in miniature, with people who sincerely want it to work.