The Serco/Capita/G4S/Carillion-style model exposes some of the hardest structural problems in B2G, because these firms don't merely sell government a product; they can become part of the machinery through which government actually operates.

That makes them analytically different from, say, Microsoft selling software licences to a ministry or a construction company building a bridge.

Consider what a classic outsourcer can be contracted to operate: prisons, immigration detention, prisoner transport, welfare/employment services, military-base support, healthcare administration, facilities management and other citizen-facing services. When one of these arrangements fails, government generally cannot respond as an ordinary customer would—terminate the supplier and simply stop buying the service. The service still has to exist tomorrow morning.

That creates several mutually reinforcing problems.

First, outsourcing can create dependency rather than genuine risk transfer. The state nominally transfers operational responsibility to a private company, but ultimately retains responsibility for continuity. The collapse of Carillion illustrates this unusually clearly: at liquidation it was involved in roughly 420 UK public-sector contracts, spanning hospitals, schools, prisons, transport and the armed forces. Government therefore had to ensure essential services continued despite the company's insolvency. The NAO subsequently emphasized the dependencies created when a company becomes a strategic supplier.

Second, competition can be strongest before the contract and weakest afterward. Imagine three companies bidding to operate a complex government service for ten years. Serco wins, hires/transfers the workforce, builds the systems, acquires years of operational knowledge and integrates itself into the department. Five years later, "we'll just switch supplier" is much less credible than it sounded during procurement. The Institute for Government specifically identifies weak competition as a condition that makes outsourcing failures more likely.

Third, governments can progressively lose the expertise required to challenge the contractor. If you haven't operated a service internally for 15 years, how do you know what it should cost? How many staff are really necessary? Whether the contractor's proposed change is reasonable? Or whether bringing it back in-house is feasible? The NAO has documented substantial weaknesses in government contract-management capability, while the Institute for Government has similarly argued for stronger commercial expertise inside government.

That produces a nasty feedback loop:

outsource expertise → internal capability declines → government becomes more dependent on suppliers → switching becomes harder → supplier bargaining power increases → further outsourcing becomes easier than rebuilding capability.

And there is another problem I think is particularly important: the contractor and government can both rationally prefer a bad contract to admitting that the underlying economics don't work.

Suppose government estimates a service costs £120m annually. A supplier promises £90m and wins. Several years later it becomes apparent that providing the required service properly really costs £110m.

There isn't an easy answer anymore. The contractor can cut quality, reduce staffing, seek contractual variations, absorb losses temporarily, renegotiate, or eventually exit. Government can pay more, tolerate deteriorating performance, retender the service, or rebuild an organization capable of running it itself. None is painless.

The UK Parliament's post-Carillion inquiry documented examples where government itself had inadequate information about the services being outsourced. In one NHS/Capita case, NHS England did not sufficiently understand the inherited services to establish realistic specifications initially, while Capita underestimated the task.

So it's important not to interpret every failure as "greedy contractor exploits government." Sometimes the state is an extremely difficult customer: unrealistic specifications, politically imposed savings targets, poor data, changing requirements and procurement rules that reward an attractive bid rather than a realistic one can generate dysfunctional contracts. The government's 2013 review of major Serco and G4S contracts, for example, identified significant weaknesses in government contract management while not finding deliberate acts by those firms causing billing irregularities across the 28 contracts reviewed.

The really problematic subset

I'd therefore distinguish ordinary B2G from what we might call privatized state capacity.

A company selling police departments laptops is B2G.

A company operating the government's prison, immigration, welfare-processing or defence-support infrastructure is something economically different. Government has effectively delegated part of its operational capacity while retaining ultimate political and legal responsibility for the outcome.

That produces an unusual combination:

Private management + public funding + limited competition + difficult-to-measure outputs + long contracts + high switching costs + government as buyer of last resort.

Normal market discipline works less effectively under exactly those conditions.

And that explains why Serco, Capita, G4S and the former Carillion became such prominent UK examples. The Institute for Government noted that those four alone earned about £6.6bn from UK government and the wider public sector in 2012–13, while simultaneously observing how difficult it was to determine their overall performance across contracts.

It doesn't mean outsourcing necessarily performs worse than public provision—the relevant counterfactual matters enormously, and government agencies have their own incentive and performance problems. But the further you move from purchasing a clearly specified commodity toward outsourcing an essential, complex public function, the weaker the normal economic case for simple "competition will discipline the supplier" reasoning becomes.

That's also why Serco/Capita/G4S/Carillion are a more interesting category to analyze than the much broader list I gave earlier. They're close to the purest examples of the large-scale public-service-outsourcing model and expose its institutional weaknesses particularly clearly.