Why Your City Government Is the Wrong Size

In 2007, Denmark merged 271 municipalities into 98. The stated rationale was familiar: economies of scale, administrative efficiency, a stronger tax base to fund services. International reform consultants were approving. What followed was not a service delivery revolution. What followed, when researchers studied the before-and-after carefully, was a measurable collapse in something harder to quantify: citizens’ sense that they could actually influence their local government.

That finding should bother anyone who designs municipal governance reforms for a living. It bothers me.

There is a settled assumption running through most city governance work, in Africa, in South Asia, in Eastern Europe, in virtually every context where central governments are trying to fix underperforming local government systems. The assumption is that bigger is better. Larger municipalities achieve economies of scale. Consolidated government is more professional, easier to partner with, simpler to monitor. When local governments fail, the prescription is: merge them, standardize them, and build their capacity.

The evidence says this prescription is, at best, partially right, and often badly wrong.

The shape of the efficiency curve

Cross-national research on municipal performance consistently finds an inverted U-shaped relationship between jurisdiction size and efficiency. Municipalities serving between 25,000 and 250,000 people perform best. Above 250,000, efficiency falls as coordination costs multiply and administrative distance from service delivery grows. The sweet spot is smaller than most reform programs assume.

But the size finding understates the problem, because efficiency is task-specific in ways the consolidation argument ignores.

Capital-intensive services, water infrastructure, sewerage, public works, do benefit from scale. The logic is straightforward: large pipes, large plants, and long supply chains reward large procurement and unified management. But labor-intensive services, policing, trash collection, primary education, are more efficiently delivered by smaller jurisdictions. And labor-intensive services account for 80 to 85 percent of what municipalities actually spend.

The dominant argument for consolidation therefore applies to the minority of municipal activity. The majority, by expenditure share, suffers from scale rather than benefiting from it. Nobody puts this in the reform proposal.

Brazil runs the experiment backwards

Between 1988 and 2000, Brazil did the opposite of what reform consultants usually recommend. Hundreds of municipalities were split in two, creating smaller, younger, poorer local governments. Conventional wisdom predicted disaster. New municipalities would lack fiscal capacity. They’d be captured by local strongmen. They’d be too inexperienced to manage service delivery contracts.

Household trash collection in the newly created municipalities rose by 4.4 percentage points. Literacy improved. The explanation was simpler than the critics expected: remote areas, long politically subordinate to the larger municipalities they had been carved from, finally received resources and administrative attention that the old headquarters had chronically withheld.

Large cities steal from their peripheries. This is almost a law of political economy. It rarely appears in the aggregate data, because the average masks the gradient. But if you’ve spent time in a large African city, you’ve seen it. The center gets paved roads, the peri-urban fringe gets promises.

South Africa’s manufactured crisis

South Africa illustrates a different dimension of the same problem. After 1994, South Africa built one of the most sophisticated local governance frameworks on the continent. Wall-to-wall municipalities covering every inch of territory. Intricate performance management requirements. Procurement rules, service level agreement templates, intergovernmental reporting frameworks. The system was designed to make local government legible, accountable, and professional.

What it produced instead was what researchers have started calling a manufactured capacity crisis.

The widely repeated diagnosis is that South African municipalities fail because they lack skilled staff. Twenty years of reform programs have tried to fix this: more training, better HR management, bursary schemes for engineers and financial managers. The diagnosis is repeated with such confidence that questioning it reads as contrarian.

But the actual constraint is different. South African municipalities spend enormous resources, financial and human, simply maintaining administrative legitimacy under a compliance architecture of crushing complexity. Reports that nobody reads. Audits that measure process and ignore outcome. Consultants hired to generate documentation rather than solve problems. The skills gap is real, but it’s partly an artifact of system design. Municipalities built to the specification of large metro complexity have been deployed across sparse rural territories that need something far simpler.

The state designed a system for Johannesburg and applied it to towns of 15,000. Then it diagnosed the resulting dysfunction as a human capital problem.

This is what happens when form doesn’t follow function. When a uniform governance model covers territories that range from dense metropolitan cores to dispersed rural areas with almost nothing in common, you produce a chronic institutional misfit. Standardization is not a neutral technical choice. It is a political choice that privileges the administrative preference of the center over the actual needs of the periphery.

More targets, less performance

Central governments typically respond to local government failure by adding measurement. The intuition is coherent: if you can’t see what’s happening, you can’t improve it. Track enough indicators and performance will follow.

The UK’s experience under Blair is instructive. The Public Service Agreement framework had grown to roughly 600 targets by 1998. By 2008, the government had cut this to 30. The reduction wasn’t a concession to lower ambition. It reflected the recognition that 600 targets had created a reporting industry rather than a performance improvement system. Norway tells a parallel story: national agencies invest substantially in formulating performance measurement frameworks that then drive very little of their actual management.

More targets produce better target-meeting. This is not the same as better service delivery.

The sprawl exception

None of this means smaller is always right. There’s strong evidence from Europe and North America that highly fragmented metropolitan governance, many small jurisdictions competing for the same tax base, drives urban sprawl. When dozens of municipalities race to attract high-income residents while offloading regional infrastructure costs onto neighbors, land-use becomes a competitive game with negative-sum consequences for the metropolitan area as a whole. In cities above a million people, fragmentation reliably correlates with sprawl.

So the constraint is not size but function. Capital investment needs scale. Routine service delivery needs proximity. Political accountability strengthens at smaller scale. Metropolitan land use needs coordination across jurisdictions. Applying one institutional model to all four simultaneously is the error.

The reform nobody wants

The reform community’s preference for consolidation is not irrational. Large jurisdictions are easier to manage from the center, easier for donors to engage with, easier to hold accountable under standardized frameworks. There’s an administrative convenience logic to bigness that operates independently of whether bigness delivers services.

But 80 to 85 percent of what municipalities spend goes to services that deliver less as distance increases. The peripheries of large municipalities are chronically underserved, as Brazil’s natural experiment showed. And citizens in large municipalities feel, with some justification, that their voice barely registers.

The reform that the evidence most clearly supports is also the most politically awkward: differentiate the model by function. Let the services that need scale operate at scale. Let the services that need proximity operate small. Accept that the administrative landscape will be messier and harder to monitor than a uniform model. Measure outcomes, not compliance.

Nobody has seriously tried this. The reasons are not mysterious. Differentiation requires central governments to give up control. Outcome measurement requires deciding what outcomes matter, which is a political question masquerading as a technical one. And acknowledging that your compliance architecture manufactured the capacity crisis you spent twenty years trying to fix is not a comfortable institutional admission.