30 December 2010

Lies, damned lies, and policymaking

An ethical doctor decides to examine, discreetly, a patient who is being looked at by other doctors:
[S]he’s concerned that, like many patients, he’ll end up with prescriptions for multiple drugs that will do little to help him, and may well harm him. “Usually what happens is that the doctor will ask for a suite of biochemical tests—liver fat, pancreas function, and so on,” she tells me. “The tests could turn up something, but they’re probably irrelevant. Just having a good talk with the patient and getting a close history is much more likely to tell me what’s wrong.” Of course, the doctors have all been trained to order these tests, she notes, and doing so is a lot quicker than a long bedside chat. They’re also trained to ply the patient with whatever drugs might help whack any errant test numbers back into line. What they’re not trained to do is to go back and look at the research papers that helped make these drugs the standard of care. “When you look the papers up, you often find the drugs didn’t even work better than a placebo. And no one tested how they worked in combination with the other drugs,” she says. “Just taking the patient off everything can improve their health right away.” David H Freedman, quoting Dr Athina Tatsioni in Lies, Damned Lies, and Medical Science, 'The Atlantic', November 2010
A potential difficulty with Social Policy Bonds is that they rely, almost entirely, on meaningful correlations between measured variables and that which society wants to target: most likely, some component of well-being. It's a difficulty, because people can game the system, complying with the letter, but not the spirit, of any defined target-setting.

What's not so obvious is that it's an even bigger problem under current policymaking regimes. In our industrial societies, with their large, complex economies, government bodies and non-governmental organizations have extremely complicated tasks. Increasingly, and of necessity, government already relies on numerical indicators to manage its resource allocation. and largely supplanted families, extended families, and communities in supplying a range of welfare services to a large proportion of their populations. .

But this use of indicators is relatively recent, unsystematic, unsophisticated and incoherent. Indicators such as the number of medical tests performed, or the size of hospital waiting lists don’t measure what matters to people or are prone to manipulation. Even when numerical goals are clear and meaningful they are rarely costed, they are almost always too narrow, and they are largely chosen to mesh in with the goals and capabilities of existing institutional structures. Those broad targets that are targeted with some degree of consistency tend to be economic aggregates, such as the inflation rate, or the rate of growth of Gross Domestic Product — which has come to be the de facto indicator par excellence of rich and poor countries alike. But GDP’s shortcomings as a single indicator of the health of an economy are serious, and widely known. Government would do better to target ends rather than means: social and environmental outcomes that are meaningful to natural persons, as against government agencies and corporate bodies, rather than growth rates or other abstract economic indicators.

It would appear that the choice will increasingly be between (a) the current de facto targeting of per capita GDP along with an almost random array of narrow, easily manipulated indicators that have no necessary relationship to societal goals, and (b) the targeting of consistent, transparent, mutually supportive indicators that represent meaningful social outcomes, under something like a bond regime.

Social Policy Bonds are not perfect, but they still, I believe, would be better than the current system.

No comments: