Institutional features that make the model work
›Structural separation
›Credible price regulation
›Universal service obligations
›Public ownership as alternative
Supporting cases
1989 Electricity Act separated generation (competitive) from transmission (regulated National Grid) and distribution (regulated regional DNOs). Wholesale prices fell in the first decade; grid investment remained adequate under RPI-X.
Singapore's Public Utilities Board operates integrated water supply as a public enterprise with full cost recovery, NEWater reuse, and among the lowest non-revenue water losses globally.
Statnett (Norway), Svenska kraftnät (Sweden), Fingrid (Finland) run transmission as publicly-owned TSOs while generation and retail compete in Nord Pool spot market. Reliability and prices have been strong over three decades.
Disconfirming cases
1996 privatisation of track infrastructure (Railtrack) as a for-profit firm led to underinvestment, Hatfield 2000 crash, and re-nationalisation as Network Rail in 2002. Illustrates that privatising the non-contestable layer without credible investment regulation fails.
Partial deregulation that kept retail price caps while liberalising wholesale markets produced the 2000-01 crisis with Enron manipulation. Ill-designed market architecture in a natural-monopoly sector is worse than either pole.
Post-1989 English water companies paid out large dividends while underinvesting in sewage treatment, producing the 2020s river-pollution scandal. Weak regulator (Ofwat) and leveraged financial engineering undermined the regulated-private model.
What this condition is NOT
- A general argument for nationalising competitive industries
- A claim that public ownership always beats regulated private
- A claim that regulated private always beats public ownership
- An endorsement of state-run generation, retail, or content layers, which are contestable
- A blanket condemnation of privatisation — the UK generation and retail segments have mostly worked
Policy implications
Identify the non-contestable layer precisely and do not attempt competition there. Apply competition aggressively to the layers that are contestable (generation, retail, train operations). Choose between well-regulated-private and well-run-public based on the jurisdiction's regulatory capacity: weak regulators should prefer public ownership; strong regulators can make regulated private work. Avoid the worst of both worlds: private ownership with weak regulation and without universal service obligations.
Framework position
Natural monopolies are one of the clearest conditions favouring intervention: economies of scale mean single-provider is efficient, marginal-cost pricing is unsustainable for private owners, and unregulated private monopoly extracts rents. The framework endorses either credible price regulation with structural separation, or direct public ownership with ring-fenced financing, depending on regulatory capacity. It rejects both naive privatisation of network layers and nationalisation of contestable layers. The design detail — where the unbundling line is drawn and how the regulator is empowered — determines whether the institutional solution succeeds or reproduces the market failure under a different name.