The New York Times reports on a conference in Nova Scotia: A New Measure of Well-Being From a Happy Little Kingdom* (registration req’d). [Thanks to the New Economist, who in turn thanks other two bloggers.]

Everyone agrees that GDP is not the same thing as social welfare: economists learn this point early on in our principles courses. A more useful tool for policy analysis would be what we call a ‘Social Welfare Function’ (SWF): an index that increases when society as a whole is better off. Things like General Progress Indicators (GPI) – such as the one generated by GPI Atlantic – are attempts to construct SWFs.

Those of us who remember our social choice theory will recall Arrow’s Impossibility Theorem, which demonstrates that there is no way of generating a democratic, coherent social ordering. It also implies that there is no democratic, coherent SWF, either.

From the NYT article:

Ronald Colman, a political scientist and the research director for Canada’s well-being index, said one challenge was to decide how much weight to give different indicators.

In fact, that is the challenge, and it’s insurmountable. I don’t think that anyone would dispute the notion that (for example) the 22 indicators listed here are of interest. But after 30 minutes of rootling around that site, I still can’t find any description of how much of an increase in a given health indicator would be required to compensate for an increase in inequality, or how they would be balanced against a reduction in employment, let alone any justification for why that tradeoff is reasonable – or even why that tradeoff should be constant.

Later on, it is suggested that

…Canada’s index could eventually take the form of a report card rather than a single G.D.P.-like number.

This is a much more sensible exercise. As an econometrician, I welcome new data sets, and especially ones that can be used to evaluate policy. But I’m very suspicious of any attempt to aggregate them into a single index.

*Note: the kingdom in question is Bhutan, not Canada.