Affordable Access: The Unit Economics of the Last Mile
Why the final ten percent of coverage costs more than the first ninety, and which financing structures have actually closed that gap in practice rather than in policy documents.
The shape of the cost curve
Network economics are unusually brutal at the margin. Dense urban coverage is profitable because fixed costs are spread across many subscribers. Each successive tranche of coverage reaches fewer people per unit of infrastructure, and the people reached typically have lower ability to pay.
The result is a cost curve that rises steeply exactly where the social return is highest. The last ten percent of a population routinely costs more to connect than the first ninety, and generates a small fraction of the revenue. No amount of exhortation changes this arithmetic.
Why most subsidy schemes underperform
Universal service funds are the standard instrument and frequently disappoint. The common failure modes are consistent across jurisdictions.
Funds accumulate faster than they are disbursed, because disbursement requires project pipelines that regulators are not staffed to build. Subsidies are awarded for capital expenditure but not operating expenditure, so towers are built and then switched off when the operating loss becomes clear. And coverage obligations are specified geographically rather than by service quality, which is satisfied by a signal that technically exists and practically cannot carry a video call.
Each of these is a design flaw rather than an inevitability.
What has actually worked
Four structures show a better record.
Reverse auctions for the lowest subsidy per connected household, which surface the true cost rather than the claimed one. Infrastructure sharing mandates that let multiple operators use one tower, which changes the underlying economics rather than compensating for them. Anchor-tenant arrangements where a school, clinic or government office provides baseline revenue that makes the site viable. And demand-side support — subsidising the service rather than the infrastructure — which avoids paying for capacity nobody uses.
The common feature is that each attacks the cost structure rather than transferring the loss to the public balance sheet.
Affordability is not only price
A final point that policy frequently misses: connectivity that is technically affordable can remain practically unusable. If the device costs three months of income, if the interface is unavailable in the user's language, or if there is no reason to go online because no relevant service exists, coverage statistics will improve while usage does not.
Access is a bundle — network, device, language, relevance and skills. Funding one component and declaring the problem solved is the most common error in the field.