The Meeting Nobody Writes About
Picture yourself in a government ministry, third floor, fluorescent light humming over a conference table. Across from you sits the technical team that spent two years designing a rail rehabilitation project. The economic case is sound. The engineering is solid. Someone has flown in from the regional development bank's headquarters, and she is explaining, carefully, that the project will need to be revised and resubmitted. She does not say it will be rejected. She doesn't need to. You have been through this before.
You never hear about the road that wasn't built. The port expansion that died in a subcommittee. The grid upgrade that made it through three rounds of technical review and then, without ceremony, stopped appearing on any agenda. Somewhere in a development bank's internal structure, a risk committee reviewed that project, weighed something nobody outside the room fully understood, and moved on. The project's proponents were told to revise and resubmit. They rarely do.
This is how a large share of infrastructure investment decisions actually get made, and it has almost nothing to do with the press releases about "transformative partnerships" and "catalytic capital." The real mechanism is governance: who sits on which committee, what mandate that committee holds, how it weighs sovereign risk against development mandate, and whether the institution's internal incentives reward approvals or protect against losses. Get those structures wrong, and even the most technically sound project quietly disappears.
Who Actually Holds the Veto
Development banks, whether multilateral institutions like the African Development Bank or bilateral vehicles like Germany's KfW, share a broadly similar internal architecture. A board of governors, typically representing shareholder governments, sits at the top. Below it, a board of directors handles operational oversight. Then comes the executive layer: a president or chief executive, and beneath that a constellation of vice-presidents, managing directors, and, critically, the risk function.
The risk committee is where attrition happens. It is usually chaired by a chief risk officer who reports directly to the board, not to the president. That reporting line matters enormously. It means the risk committee cannot easily be overruled by operational management excited about a flagship project. A president who wants to announce a landmark port deal in a fragile state still has to clear a committee whose chair owes her job to the board, not to him. The architecture is deliberately adversarial.
In practice, the committee reviews projects at a stage called the credit or investment approval memo, a document that can run to several hundred pages for a major piece of infrastructure. The memo must satisfy a checklist that typically includes sovereign creditworthiness ratings, environmental and social compliance (often benchmarked against the Equator Principles or the IFC Performance Standards), procurement integrity assessments, and a debt sustainability analysis. Each of those categories has a threshold. A project that trips one threshold faces a mandatory remediation process before it can advance.
Consider what this looks like in practice. A regional development bank is evaluating a 400-million-dollar rail rehabilitation project in a lower-middle-income country. The technical team rates it highly. The economic rate of return models out at around fourteen percent. The country's public debt has crossed seventy percent of GDP, the risk committee's internal threshold for heightened scrutiny, and the national railway authority carries two unresolved audit findings from a previous project with the same bank. Under the bank's operational policies, both flags trigger an automatic escalation. The project doesn't die at that moment. It enters a holding pattern: the borrower is asked to restructure the financing, bring in a co-lender to reduce the bank's exposure, and close out the prior audit findings. That process takes, on average, between eighteen months and three years at most large multilaterals. Think of it as a bureaucratic amber light that almost nobody has the patience to wait out. Many projects don't survive the wait.
The Incentive Problem Nobody Likes to Admit
Here is where governance theory meets institutional psychology, and where the picture turns genuinely uncomfortable. Development banks are not commercial banks, but they are not charities either. They carry credit ratings, and those ratings depend on maintaining a healthy portfolio. A Triple-A rating, which institutions like the European Investment Bank and the World Bank's IBRD have held for decades, is not a point of institutional pride so much as the load-bearing wall of the entire development model. It is what allows the bank to borrow cheaply in capital markets and on-lend at concessional rates. Lose the rating, and the model becomes more expensive for everyone downstream.
That creates a structural tension with no clean resolution. The mandate says: take risks that private capital won't. The incentive says: protect the balance sheet. Risk officers live in that tension every working day, and the resolution, in most institutions, tilts toward protection. Not because anyone is cynical. Because the career consequences of approving a project that later defaults are far more visible than the career consequences of shelving a project that would have worked. The road not built doesn't show up in the portfolio review.
Two colleagues who joined the same multilateral in the same cohort illustrate the divergence plainly. One moves into the risk function and builds a reputation for rigorous screening; her projects carry a near-zero non-accrual rate. The other stays on the operational side, closes more deals, has one restructuring on her record. Which career path does the institution's promotion data historically reward? The risk function answers that question clearly enough that talented people self-select toward it, and the operational side gradually loses its most ambitious officers to a function whose entire purpose is to say no.
This isn't corruption or negligence. It is incentive architecture, the kind that accumulates over decades without anyone designing it maliciously. Reforming it is genuinely difficult, because the same conservatism that kills promising projects also prevented several major multilaterals from taking catastrophic losses during the sovereign debt crises of the nineteen-eighties and the commodity collapses that followed. Caution has a real track record. So does paralysis.
What Shareholder Pressure Actually Changes
Shareholder governments, particularly the larger-quota members who hold effective veto power on major decisions, can and do shift these internal dynamics, though rarely through direct intervention in individual project decisions. The more common lever is policy reform: a shareholder coalition pushes for a revision to the bank's risk appetite framework, raising the debt sustainability threshold, or carving out an explicit high-impact exemption category that allows the risk committee to approve projects above normal exposure limits if the development case clears a separate qualitative review.
The World Bank Group's evolution of its approach to fragile and conflict-affected states is one documented example of exactly this kind of shareholder-driven recalibration. Pressure from borrower-country shareholders and advocacy from donor governments produced a revised operational framework that explicitly acknowledged the paradox: the countries most in need of infrastructure investment are precisely the countries that fail the standard credit screens. The policy response was a parallel track with different risk parameters, staffed separately and governed by a modified committee structure. It is the institutional equivalent of building a side door when the main entrance keeps slamming shut.
Does it work? Partially. The parallel track approves projects that the main committee would not, and it carries, by design, a higher loss rate, which the bank's shareholders have agreed to absorb as the price of the mandate. The governance innovation is real. So is the ceiling: even the revised framework has thresholds, and a project in a country with active armed conflict near the proposed infrastructure corridor will still, quietly, not make it through.
Ask yourself what that means at scale. Across dozens of institutions making thousands of decisions over decades, the cumulative shape of global infrastructure is not the product of need, or even of available capital. It is the product of governance structures, thresholds, reporting lines, and career incentives that most people affected by the outcomes have never had occasion to examine. The projects that get built are a reflection of the structures that decided them. That is not an argument against caution. It is an argument for being honest about what governance choices cost, and honest accounting of that kind remains, at most development institutions, conspicuously overdue.