22 February 2014

Nobody's perfect

Freeman Dyson reviews Brilliant Blunders, by Mario Livio, "a lively account of five wrong theories proposed by five great scientists during the last two centuries." The examples Livio writes about:
give for nonexpert readers a good picture of the way science works. .... Wrong theories are not an impediment to the progress of science. They are a central part of the struggle. .... The five chief characters in Livio’s drama are Charles Darwin, William Thomson (Lord Kelvin), Linus Pauling, Fred Hoyle, and Albert Einstein. Each of them made major contributions to the understanding of nature, and each believed firmly in a theory that turned out to be wrong..... [W]rong ideas can be helpful or unhelpful to the search for truth. No matter whether wrong ideas are helpful or unhelpful, they are in any case unavoidable. The case for blunders, Freeman Dyson, 'New York Review of Books' dated 6 March
Even more so in social policy, where underlying relationships change over time and are rarely independent of the psychic makeup of the principal actors and stakeholders. We need to encourage diverse approaches to our social problems, and ones that can adapt when they are seen to be inefficient or counter-productive. As with science, though, social policy practitioners, be they politicians, bureaucrats, academics or members of think-tanks, frequebtly commit their egoes - and public funds - to deficient theories or ideologies.

The chief difference between science and other human enterprises such as warfare and politics is that brilliant blunders in science are less costly.
Quite: when great scientists commit themselves to wrong ideas the costs can be high, but when politicians do so they can be calamitous.

Social Policy Bonds would penalise failed or inefficient pseudo-solutions to our social problems, and reward only the most cost-effective ways of achieving our social goals. Bondholders would be motivated to terminate failing projects and divert funds into only the ones that are cost effective. If they don't do this quickly enough, others would bid more for the bonds than they are worth to the current holders. The bonds, being tradeable, would always be in the hands of people who are motivated to be efficient. Commitment to wrong theories would be penalised in immediate, pecuniary ways - a stark contrast with the current policymaking system, within which failed policies, instead of being terminated, often receive more and more funding in an effort to shore up vested interests.

17 February 2014

Greenhouse gases, recidivist rates, cholesterol, and the one percent

What do greenhouse gas emissions, recidivist rates and cholesterol readings have in common? They are all surrogate indicators; that is, they are things that governments target, thinking (or pretending to think) that by doing so they are benefiting society. They aren't. Whether the associated loose thinking - or just plain dishonesty - originates in government or in the people who pay governments to shape the regulatory environment in their favour, surrogate indicators have little to do with human well-being.

Perhaps we need to ask in whose interest it is that we target things like greenhouse gas emissions, or recividist rates or cholesterol readings? Surely, if we want to reduce the adverse impacts of climate change on humans and the environment, we'd be better off, with all the scientific uncertainties, to target reductions in those negative impacts? Similarly, if we actually want to reduce crime rates, why don't we target crime rates rather than recidivist rates, which have very little, if anything, to do with crime? And if we want to target physical health, why don't we reward improvements in physical health, rather than encourage the mass ingestion of statins, whose long-term effects are nebulous at best and dangerous at worst?

One reason that I am a less-than-enthusiastic supporter of Social Impact Bonds is that they are targeting recidivism rates. Their targeting of an indicator that has nothing to do with things that matter to ordinary people risks discrediting the whole idea of channeling the market's incentives and efficiencies into the public good. We have had plenty of recent and disastrous experience of financial instruments being gamed to death, with calamitous effects on ordinary hard-working citizens. So we need to be very careful about introducing new financial instruments. There is, unfortunately, every reason to be cynical. Bankers, consultants, the financial services sector, big corporations, government agencies and even non-governmental organizations all have made lots of money doing things that are purportedly in the public interest, but in fact have done nothing for ordinary people.

That is why I suggest that Social Policy Bonds target only metrics that are, or are inextricably linked to, indicators of societal well being. The bond mechanism allows for that sort of targeting, because it does not specify how our goals shall  be achieved, nor who shall achieve them. Unfortunately, without that sort of guarantee, there is every reason to expect that the well-meaning targeting of rhetorically persuasive but flawed indicators will continue to enrich only the one percent.

09 February 2014

The system's broken

 The Economist discusses the US Farm Bill, passed on 4 February:
[M]ore than 10,000 policyholders received over $100,000 from crop-insurance subsidies in 2011. The new bill tries to cap the amount that any one farmer can receive; but if the weather is bad, it could lead to higher payouts than planned. Taken together, these subsidies distort behaviour and trade in unhelpful ways. They have created products that make no economic sense in the rest of the world, such as making sugar from corn. As a penalty for keeping cotton subsidies in place, the World Trade Organisation’s rules require the American government to pay $147m a year to compensate farmers in Brazil. A trillion in the trough, 'The Economist', 8 February
It's the persistence for decades of these economically and environmentally disastrous policies that indict our entire policymaking system. Yes, policymakers will make mistakes; all the more reason why we should have systems in place to ensure that failed policies are terminated. But instead, we have the systems that ensure that appalling policies become more and more entrenched because of political inertia, because they subsidise resistance to their termination, or because they become capitalised into high asset values that would create genuine but temporary hardship if they were withdrawn. Governments have a long history of meddling in agriculture; they persist even though it's been known for decades that they are, to put it kindly, irrational.

And corrupt. The Economist continues:
How could Congress write such a law? One answer can be found in the register of political donations. The ten members of the House, nine Republicans and one Democrat, who accepted most money from agriculture lobbyists took in an average of $225,000 in political contributions during 2013, according to Open Secrets, which tracks donations—almost as much as some farmers received in return.
Not much is black or white in politics and policymaking, but as P J O’Rourke put it twenty-three years ago (in Parliament of Whores):
I spent two and a half years examining the American political process. All that time I was looking for a straight forward issue. But everything I investigated – election campaigns, the budget, lawmaking, the court system, bureaucracy, social policy – turned out to be more complicated than I had thought. There were always angles I hadn’t considered, aspects I hadn’t weighed, complexities I’d never dreamed of. Until I got to agriculture. Here at last is a simple problem with a simple solution. Drag the omnibus farm bill behind the barn, and kill it with an ax.

01 February 2014

Another useless indicator creeps in

Social Policy Bonds embody two main principles: targeting outcomes; and the use of markets to achieve these outcomes most efficiently. Even if discussion of Social Policy Bonds just leads to a rational discussion of which outcomes we want to achieve, then my work on the bonds would not be in vain.

I've discussed the futility of over-narrow objectives before, when discussing  Social Impact Bonds. (See also why I think the bonds must be tradeable.) But broad indicators too can, by default, become de facto targets, and they can be similiarly flawed; by which I mean that they are not inextricably linked to society's well-being. We have a tendency not to think through the implications of certain indicators: the biggest one is GDP. But another that is creeping into policymaking dicussion is five-year survival rates for cancer. The Economist casually slips it into an article about managing heatlh care:
Deciding where to seek treatment might seem simple for a German diagnosed with prostate cancer. The five-year survival rate hardly varies from one clinic to the next.... Need to know, 'the Economist', 2 February
But the five-year survival rate is meaningless:
[I]n the U.S. prostate cancer is being diagnosed earlier, a lead-time bias, and the cancer is being over diagnosed, that is, a pseudodisease is detected" in the form of screening-detected abnormalities that "meet the pathologic definition of cancer but will never progress to cause symptoms in the patient’s lifetime." Source (Scroll down to Incorrect metrics.)
The article in the Economist ends:
[D]octors ... have long focused on clinical outcomes such as infection and re-admission rates. But by thinking about what matters to patients, providers can improve care and lower costs at the same time.
Exactly so. We need to be focusing on broad, meaningful indictors of well-being, such as Quality Adjust Life Years, and target those, rather than casually accept the use of flawed measures such as five-year survival rates.