Limitations of the Payment by Results model

An interesting comment from Mr Toby Lowe on an article extolling the benefits of Payment by Results (PbR):

[I]n order for PbR to work, 'results' must be directly attributable to particular interventions .... In complex systems, results are never attributable to particular interventions, and so PbR cannot work. This is not a technical issue about measures, it is an inescapable consequence of the way that complex systems operate. Toby Lowe, commenting on The next step in payment by results by Rodney Schwartz, 28 September

You might think that this - valid - point undermines the Social Policy Bond concept. But I would disagree. PbR as practised today differs from Social Policy Bonds in that the bodies that are paid to perform better are (1) conventionally structured, and (2) generally known in advance.

Both these features multiply the opportunities for gaming and manipulation. Bodies that are paid by results under current regimes are service providers that have a persistent composition and identity, and mainly for that reason are committed to a fairly limited range of activities. The 'results' for which these bodies are paid, though termed 'outcomes' are therefore quite narrow. So a body can, for instance, game the PbR scheme by simply exporting a targeted problem to areas not covered by that scheme. More importantly, the very scope of the 'outcomes' targeted is greatly restricted in time and space by the constraints imposed by the structures and activities of existing organizations. I belive that to tackle broad social problems we need a new type of organization; one of protean structure and composition, so that at each point along the outcome-achieving path it forms a coalition of the bodies that will be most efficient at solving the targeted problem. Social Policy Bonds would bring about such organizations.

What about the difficulty of attributing results to interventions in complex societies? Again, I agree with Mr Lowe. But under a Social Policy Bond regime, there would rarely be a need for direct, deliberate (and manipulable) attribution. Take for example the goal of targeting a nation's health for sustained improvement over, say, thirty years, as measured by a combination of such indicators as longevity, infant mortality and quality adjusted life years. Under a Social Policy Bond regime, such Health Bonds would be valued not by some cash-doling bureaucracy, but by the market for the bonds. Any activities undertaken by holders of the bonds would raise the value of their bonds only if they, in the market's view, make the early achievement of society's health goal more likely. Bondholders can undertake, or finance, a vast range of health-improving activities, some of which might benefit from a PbR approach, others of which will not (and might even conflict with one). It will be for motivated bondholders to decide. Such broad outcomes, undertaken by bodies that come and go during the lifetime of the bonds and overseen by a motivated market, cannot be manipulated. The complexity of the interventions and their effects is matched by the complexity of the bondholders', their coalition and the vast range of approaches that they will try in their efforts to achieve the targeted goal. Direct attribution of payment to successful interventions is no more necessary in such a long-term, broad project than it is to employees of a hospital.

Most of the PbR schemes that are being talked about involve Social Impact Bonds (also known as Payment for Success bonds), which are non-tradeable versions of Social Policy Bonds. I have discussed the limitations of such bonds, and my ambivalence toward them, here and here and in many posts in this blog: here and here, for examples, or search this blog for Social Impact Bonds. Long ago I did a post on New Public Management, which is also relevant.