Skip to content

Structural Separation or Unbundling

Structural remedy — instantiates Bottleneck Power Governance

Splits the controller so the bottleneck is owned and run separately from the businesses that depend on it — removing the incentive to self-preference rather than merely policing it.

Structural Separation or Unbundling is the strongest remedy in the archetype: instead of writing rules the controller must follow, it changes who owns what. It severs the bottleneck asset from the competitive businesses that ride on it — by ownership separation, a legally independent operator, or unbundling the chokepoint into a stand-alone product rivals can buy — so that the entity running the access point no longer has any stake in favouring one user over another. Its defining move is that it eliminates the incentive to abuse rather than monitoring the behaviour: a grid operator that owns no generators has no reason to rig dispatch; a track authority that runs no trains has no reason to hoard the good slots. Where a firewall polices a channel and a tariff governs terms, separation removes the conflict of interest at its root and, by making the bottleneck element separately purchasable, opens the downstream market to entry.

Example

A national railway owns both the track — which no rival can economically duplicate — and the passenger trains that run on it. Whenever an independent operator asks for a slot on a profitable route, the incumbent, wearing its infrastructure hat, finds reasons to give itself the peak departures and offer the rival the 5 a.m. path. No conduct rule fully fixes this, because the same company profits from both roles.

Structural separation splits the roles: an independent infrastructure manager owns and runs the track and sells access slots, while train operations become separate companies competing for those slots — the shape of Britain's rail restructuring, where track (Network Rail) sits apart from the train operating companies. Now the track authority earns the same fee whichever operator takes a slot, so it allocates by neutral rules; and because track access is a product one can simply buy, an open-access operator can enter a route with three trains a day without owning a metre of rail. The bottleneck still exists — there is one track — but controlling it no longer confers power over the market that uses it.

How it works

  • Draw the separation line. Identify the true bottleneck asset and cut it away from the contestable activities — cleanly enough that the separated operator has no residual stake in downstream outcomes.
  • Choose the depth. Set how far the split goes: accounting separation, a legally distinct subsidiary, or full ownership divestiture, trading enforceability against disruption.
  • Make the element purchasable. Turn the unbundled bottleneck into a stand-alone product with a price and terms, so rivals can buy just that input and build competing offers on top.
  • Manage the transition. Sequence the divestiture and stand up the successor entities so service continuity and dependent parties survive the cut.

Tuning parameters

  • Separation depth — accounting → functional → legal → full ownership. Deeper cuts make neutrality self-enforcing but destroy more of whatever efficiencies the integration provided and are harder to reverse.
  • Unbundling granularity — how finely the bottleneck is sliced into separately-sellable elements. Fine unbundling maximizes entry but multiplies interface and coordination costs between the pieces.
  • Scope of the split — only the pure bottleneck, or adjacent assets too. A narrow cut preserves value but may leave residual leverage; a wide cut removes leverage but sacrifices genuine synergies.
  • Transition horizon — a fast, disruptive divestiture versus a phased one. Speed ends the conflict of interest sooner; phasing protects dependents but leaves the incentive to self-preference alive during the handover.

When it helps, and when it misleads

Its strength is durability: by removing common ownership it removes the motive to discriminate, so it needs far less ongoing supervision than a firewall or a conduct rule — the neutrality holds even when the regulator looks away. It is the remedy of choice when conduct rules have been tried and predictably gamed, or when the integration exists mainly to leverage the bottleneck.

But it is the bluntest and most irreversible tool, and its costs are real. Separating functions that genuinely belonged together sacrifices economies of scope and coordination,[1] and a badly drawn line leaves the successor pieces stranded — an infrastructure manager starved of the operating knowledge that used to sit next door. It is slow, contested, and hard to undo if it turns out wrong, and it can be run for show or for spite — a nominal legal separation that leaves control intact, or a break-up urged by rivals for whom it is simply convenient rather than warranted. The discipline is to reserve it for bottlenecks where the conflict of interest is structural and conduct remedies have demonstrably failed, to cut where the true asset boundary lies rather than where it is politically easy, and to weigh the lost synergies honestly rather than assuming separation is free.

How it implements the components

  • structural_separation_boundary — the mechanism is the drawing and enforcing of this boundary: the ownership or operational line that severs the bottleneck from the businesses depending on it.
  • substitute_and_entry_creation_path — unbundling the chokepoint into a purchasable stand-alone input lets rivals buy it and enter the downstream market, manufacturing the substitutes the bottleneck otherwise suppresses.

It changes structure but does not itself set the price or terms on which the separated element is sold — those fall to Price-Cap or Rate Review and the Non-Discrimination Access Tariff — and where breaking the firm is unwarranted, the lighter Self-Preferencing Firewall governs the same conflict without the cut.

Notes

Separation and the firewall govern the same conflict of interest at opposite intensities; the firewall keeps the firm whole and watches the channel, separation removes the firm's stake entirely. Because it is costly and near-irreversible, structural separation is best read as the escalation point — what a jurisdiction reaches for once tariffs, firewalls, and access mandates have been tried and gamed — rather than the first response.

References

[1] Economies of scope: cost savings that arise from producing related activities together. Structural separation deliberately forgoes them to eliminate the conflict of interest, which is why the remedy is weighed against the synergies it destroys rather than treated as costless. ↩