In all the reaction to Carillion's collapse and the NAO's report on PFI, one important perspective has for me been under-stated - that of transactions cost economics.

Let's start with an important distinction - between private finance and private provision.

There is no justification for private finance of government projects. One thing the state can do better than the private sector is borrow cheaply. Private finance is the result of daft accounting rules described by Laurie MacFarlane rather than of economic rationality.

Private provision, however, is another matter. In principle, competitive tendering by private firms could bid down the cost to the tax-payer of government services.

But there's a problem here. As Simon says, companies that win tenders by bidding low have an incentive to cut quality. The question is: is it possible to stop this happening?

It's here that transactions cost economics enters. This perspective began with Ronald Coase's famous essay, The Nature of the Firm (pdf). Whether we should do a job in-house or through the market depends upon the comparative costs. And, he said, "there is a cost of using the price mechanism."

In our context, this cost is the difficulty or even impossibility of writing contracts which ensure good quality provision. As Oliver Hart put it:

It may be prohibitively expensive to write a contract that conditions quantity, quality and price...This is not just because some of the variables are privately observed, but also because, even if publicly observable, the variables are inherently hard to specify in advance. (Firms, Contracts and Financial Structure, p24)

Whether outsourcing is a good idea hinges upon whether this is the case or not. If a contract controlling outcomes adequately can be written, then outsourcing might work. But only might, because a contract demanding decent quality would be an expensive one. The NAO says (pdf):

Our work on PFI hospitals found no evidence of operational efficiency: the costs of services in the samples we analysed were similar...More recent data from the NHS London Procurement Partnership shows that the cost of services, like cleaning, in London hospitals is higher under PFI contracts.

And if we can't write an adequate contract, the case for outsourcing is even weaker.

The answer here will differ from service to service. It should be possible to contract properly for building work, for example, because a competent inspector should be able to assess quality - though even this isn't always the case. In other cases, though, quality isn't so observable and controllable; this might well be true of social care (where much hinges upon the manner of the individual carer) or forensic science.

But why might in-house provision be cheaper? One consideration here is the public service ethos. If people are motivated by an idea of public service, they'll do good jobs without oversight or detailed contracts. Good culture is potentially a cheap solution to the problem of incomplete contracts.

But only potentially. Such an ethos should not be taken for granted. The Stafford Hospital affair - and perhaps the "failure of leadership at all levels" in the prison service - remind us that it is sometimes dangerously lacking. Martin Wolf makes an important point (in a different but applicable context) when he reminds us that nationalized industries " treated users with indifference."

Which brings me to another question that's overlooked: how can we build a true public sector ethos in which workers are, in Le Grand's words, knights rather than knaves? (The answer might well include less managerialism and more empowerment of workers and users.)

It might be a good idea to reduce the amount of outsourcing; that depends not upon ideology but a cool-headed assessment of transactions costs. Doing so, though, should be a means to a greater end, and not just an end in itself.