During the coffee break at a seminar I once attended, I let slip that I thought two of my fellow panellists seemed 'quite switched on'. 'Yes,' a delegate replied, 'they've both got a Nobel Prize'. Umm, amazing powers of insight, Stevens.

Presumably a similar feeling of idiocy has just overcome the elderly residents of a care home in Minneapolis as they learnt that their somewhat bedraggled 90-year-old co-resident, Leonid Hurwicz, has just won this year's Nobel Prize for Economics. Born in Moscow, educated in Warsaw before escaping the Nazis, studying at the LSE and then emigrating to the US, Hurwicz has the distinction of being the oldest ever recipient of a Nobel and - the bit I like best - an economist who never got an economics degree.

Mechanism design, the area of study that Hurwicz founded, is the science of structuring allocation mechanisms and aligning incentives to produce socially desirable results. It has been successfully applied to problems as diverse as optimal tax structures (think about the current political row over capital gains tax); so-called 'matching algorithms' (for example how to properly allocate junior doctors to training posts) and how to create markets even where the standard assumptions about perfect information do not apply. Hence its applicability to healthcare.

Just as Al Gore is topical on the greenery debate, so the branch of economics that Hurwicz kicked off has a lot to say about the current kerfuffle over the cancellation of new independent sector treatment centres.

Part of its insight is that government makes a major mistake when it views the health, education or criminal justice sectors as unitary institutions rather than (eco)systems. In part I think this happens for sociological reasons. Having come up through the ranks first managing a hospital then, say, a region, by the time many of us come to work at the Department of Health, there is an unconscious tendency to regard the health sector as a whole as just one very big hospital; one that can be managed using roughly the same toolkit as would apply back in a district general hospital. In that sense perhaps, managerialism can sometimes be the enemy of effective policy design (an accusation often muttered sotto voce by the remaining members of the civil servant mandarin class at the DoH).

Anyway, Hurwicz's mechanism design theory helps us to understand why Whitehall's attempts to get results often fail.

First, government departments tend to take a static rather than dynamic view of the world. Static thinking leads you to think that ISTCs are mainly about surgical capacity, while underplaying the effects that alternative suppliers have on the behaviour of incumbent providers. Static policy design suggests that the way to get GPs to open for longer hours is to cost the extra time it would take and then pay them a grant, not spotting that - in the absence of any competitive threat - GPs might (and indeed have) used their extra income to work shorter hours not longer.

Second, in the absence of well-designed incentives, Whitehall's approach to target setting - Treasury public service agreements and the like - is incapable of getting participants to reveal information about how much cash it will really take to deliver a given objective. As a result government constantly spends more than would be really needed on some objectives and less than is needed to get the job done on others.

These twin failings became obvious, for example, in the DoH's pre-2002 approach to cutting waiting. At the start of the year, senior officials (with the noble exception of the DoH economics division) would walk in to the health secretary's office to get sign-off for an approach that simply budgeted an additional£500m or so of new cash to fund extra elective activity for the year ahead. Then a year or so later the same group would embarrassedly shuffle in to try and explain why cash plus 'performance management' had not produced extra elective output at all (statistics show electives were essentially flat for both 2000-01 and 2001-02).

It was this experience of failure that led directly to the obvious insight that getting the incentives right was as important as the cash, the 'capacity', or the target. Instead it meant overturning perverse incentives to keep waiting times long: for managers because it was rewarded by more cash under the old 'waiting list initiatives' and for surgeons because it sustained private practice. Hence, through trial and error, the current inter-related policies of tariff payments, patient choice, and new providers. And with it a realistic shot at getting waits down from 18 months to 18 weeks.

So as Gordon Brown turns to writing his first major speech on health since becoming prime minister, he needs to do two things.

First, if he is frustrated that the press are criticising him for equivocation, he has to come off the fence about whether he is for or against the NHS reform programme (which the government he now leads was elected on two years ago).

And second, in doing so, he should give a rather more nuanced account of the role that incentives can play in healthcare delivery than when he last set out his stall in a speech to the Social Market Foundation several years ago. There, he largely stuck to the rudimentary claim that since 'perfect competition' in healthcare is not possible because of information asymmetries, instead the only real alternatives were targets and government direction or letting the producers decide.

This time round his speechwriters need to demonstrate a greater understanding of the work of this year's Nobel winner. It proves that, even in complex social systems such as the NHS, success is more likely if improvement is driven by incentives. Mr Brown can take it on good authority - from a 90-year-old in a Minnesota care home.