Wednesday 10 July 2013

The problem with evidence-based policy

There was a bit of chatter about evidence-based policy yesterday. My blog post addressed the lack of it when it came to smoking bans, this chap from Stirling University said it was all about power and Ryan Bourne said in City AM that it was 'the most meaningless phrase in politics today'.

Here's my problem with it. Let's say that the government decides it's going to paint every house in Britain red. It could gather lots of wonks and experts together to chew over the best way to go about it. It could carry out surveys to find which shade of red the public liked best. It could conduct impact assessments. It could commission economic research to ascertain what would be the most efficient way of painting all the houses. Should it use crimson? Should it use brushes or rollers? Should it use scaffolding or ladders? Which type of paint would be most durable? Would it be better to do the work in winter or summer?

All of these questions could have empirically sound answers. Those answers could be published in peer-reviewed journals. It might transpire that it would be best to work North to South starting in April using rollers and a colour known as Dusky Rouge. The work would be outsourced to private companies.

This programme would clearly be evidence-based. Countless conferences, numerous studies and thousands of experts would have been consulted to discover the most efficacious way of painting the entire British housing stock red. If, after all this, you expressed the view that the job would be better done using paint-brushes and Salsa Red, you would be shown a ream of data showing that you were wrong. You might even be labelled a denier.

But what if you don't want to have your house painted? Even if you did want your house painted, perhaps you'd rather do it yourself than have the government do it for you. And even if the majority of Britons preferred Dusky Rouge, maybe you prefer Salsa Red or, god forbid, a shade of blue or green?

In short, the evidence base has not got us any closer to proving that the state should be painting people's houses, it has only shown the government how to do it.

This is the problem with much of what passes for evidence-based policy. It bypasses the question of why by focusing on the question of how. The fundamental question of whether the state should be doing it in the first place is never seriously scrutinised.

Take the issue of minimum unit pricing, for example. This is a classic example of reformers insisting that they have strong—nay, 'overwhelming'—evidence that makes it imperative for the state to act. As it happens, the Sheffield model which forms the evidence base in this instance is a mass of dubious conjecture and optimistic guesses, but the soundness of the data is not the point here. It is not unreasonable to assume, all other things being equal, that higher prices will reduce consumption somewhat so let's forget about the unintended consequences for a moment and suppose that a 50p unit price will reduce alcohol consumption by two per cent.

If that is the case, then minimum pricing is an evidence-based way of reducing alcohol consumption. But so what? People like drinking alcohol and it has many benefits. Why, then, not set a maximum price of alcohol and increase alcohol consumption by five per cent?

The answer, I suppose, is that the government has decided that alcohol is a Bad Thing and that it would be better if consumption fell. Fine. That is their view. So why not make the minimum price 60p or 70p or, come to that, £5 or £10? What is the evidence base for setting this minimum price at 50p?

There is none, of course. They believe that £5 a unit would be an excessive infringement on people's wallets and liberties, just as those who campaign for 20 mph speed limits consider a 5 mph speed limit to be an excessive burden on motorists, despite the fact that the same 'evidence' they cite for 20 mph would show that a 5 mph limit would 'save' even more lives.

By taking the discussion to the logical extremes of a £5 minimum price and a 5 mph speed limit, we see that policy is not really based on objective evidence, but comes down to a subjective assessment of costs and benefits. If I say that I find a 50p minimum price and a 20 mph speed limit to be an excessive burden on drinkers and motorists respectively, what evidence can be produced to show that I am wrong? There is none because—like my opponents—I am expressing a personal view.

The belief that alcohol consumption should be lower is nothing more than a subjective opinion. Evidence doesn't come into it. If there was any real science behind the policy, the government would state that there is an optimal level of alcohol consumption of, say, seven litres per capita per annum. If any politician made such a claim, he would be asked to provide empirical evidence for why that particular figure had been chosen and he would have to admit that there isn't any. How could there be?

Likewise, the 50p a unit minimum is entirely arbitrary. There is no evidence to show that the minimum price shouldn't be 75p or 95p or 15p. There is no evidence showing that there should be a minimum price at all. Again, how could there be?

The 50p figure was picked because it was politically feasible, not because the evidence showed—let alone demanded—that 50p was either optimal or necessary. No amount of evidence can tell us what the optimum number of alcohol-related deaths is (even the staunchest temperance advocate accepts that zero is unrealistic), nor can it tell us whether the nation is drinking too much or too little.

When people say that minimum pricing is an 'evidence-based policy', what they mean is that they think alcohol consumption should be lower and that higher prices are likely to reduce consumption. That is not an 'evidence-based policy'. It is personal opinion combined with the law of demand. In this instance, as in many others, the fundamental assumptions behind the policy are not—and can never be—evidence-based.

4 comments:

Chris Price said...

Christopher, I started to post a comment on your very nice analysis - but found it had become an article...

Sorry about that :)

http://www.ecigarette-politics.com/the-meaning-of-evidence-based-policy.html

Ivan D said...

Agreed 100%. Never in the field of public policy has so much been misunderstood by so many to the political benefit of so few.

themself said...

Fundamentally policy is driven by elected politicians who base their decisions on what they think gets them re-elected. Civil servants, who are the people that want to do evidence based policy, are confined to advice on options on how to meet Ministerial objectives. So that is where the evidence base is used, rather than on whether or not the Ministerial objectives are supported by evidence.

themself said...

Fundamentally policy is driven by elected politicians who base their decisions on what they think gets them re-elected. Civil servants, who are the people that want to do evidence based policy, are confined to advice on options on how to meet Ministerial objectives. So that is where the evidence base is used, rather than on whether or not the Ministerial objectives are supported by evidence.