Are measures of happiness prone to Goodhart's law - that they become useless if used for policy purposes? A new paper by Bruno Frey claims so. He contends that if happiness indices become a benchmark for judging policy success, they will become distorted.

One reason for this is that respondents will mis-report their well-being. Supporters of the government will overstate their happiness and opponents understate, and there's no reason to suppose the two will cancel out.

What's more, he says, governments can manipulate the results of happiness surveys more than they can orthodox economic indicators. For example, you can increase reported happiness by surveying people on Friday (pdf) rather than Monday, and by not preceding questions on happiness with questions on politics. And this is not to mention more other less professional methods: "here's a picture of a kitten: how happy are you?"

I would add another danger - that if governments focus more on happiness, they'll have even stronger incentives to promulgate ideologies that diminish people's expectations. If you can convince people that poverty is unavoidable, you might reduce discontent even if injustice increases.

However, I don't think this is an argument against happiness policy.

We know there are some things governments can do to improve well-being such as: reduce unemployment; give people more control over their lives; and - credit Ed Miliband here - do more to reduce mental illness. We don't need a happiness index to do any of this.I suspect the desire to use such an index is a form of statistical fetishism - a wish for the mere appearance of useful knowledge. But as Keynes said, it is sometimes better to be roughly right than precisely wrong.