Can economic growth reduce life satisfaction?

Last week Andrew Leigh linked to this paper by Angus Deaton which, using poll data from 132 countries, came up with findings contrary to previous subjective well-being research:

Average happiness is strongly related to per capita national income; each doubling of income is associated with a near one point increase in life satisfaction on a scale from 0 to 10. Unlike most previous findings, the effect holds across the range of international incomes; if anything, it is slightly stronger among rich countries. Conditional on national income, recent economic growth makes people unhappier… (emphasis added)

I’m not as convinced as Andrew that this is a good paper. The ‘life satisfaction’ question doesn’t actually directly ask whether people are satisfied with their lives, but instead asks them to imagine a ladder where the bottom represents the ‘worst possible life for you’ and the top the ‘best possible life for you’. The respondents are then asked where on the ladder they feel they stand at the present time. It would be quite possible to be dissatisfied with your life and to think that it won’t get any better, so low on a conventional life satisfaction rating but high on the imaginary ladder (personally, I’d put myself higher on the ladder than I would rate myself for life satisfaction). This could possibly explain why rich countries do well in the survey, since when things are already going well there is less scope for a better life.

And perhaps someone can explain the statistics to me, but I can’t see that Deaton has the data to show that ‘recent economic growth’ makes people unhappier. He has only one year of life satisfaction (sic) data, 2006. To know whether recent economic growth makes people unhappier we need to know life satisfaction in earlier years for that country, not life satisfaction in other countries, which is the data he has. Otherwise, how do we know we are not just measuring the well-known persistent differences between countries in self-reported happiness?

Another problem is that the ‘recent’ economic growth is not very recent, but 2000 to 2003, three to six years before the life satisfaction survey. Conventional subjective well-being theories of adaptation would suggest that by 2006 people would have lost much of the well-being impact of any increase in their standard of living between 2000 and 2003. One of the many papers using German panel study data found that most of the gains to an increase in income were lost in four years.

There is also a theory, advanced by Richard Easterlin and others, that aspirations rise with income, and neutralise the well-being gains from more material resources. The question in Deaton’s survey lends itself to this kind of result. In a period of economic growth, it perhaps easier to imagine yourself moving up the ladder. Therefore now is further away from the ‘best possible life’, even if it is better than any previous life.

Generally, Easterlin’s thesis helps explain why life satisfaction levels fluctuate without trend over time. But it is possible for aspirations to cause dips in life satisfaction. If a period of economic growth is followed by a recession – though I don’t think this happened in most countries between 2003 and 2006 – aspirations may not be met, leading to disappointment or worse. People who think that their standard of living is going down are prone to unhappiness. For example, in the 1999 Coping with Economic and Social Change survey, which was used as the basis for a book by tax-and-spend Peter Saunders, nearly a third of people who thought their standard of living was lower than five years ago classed themselves as ‘unhappy’, compared to 7% of those who thought their standard of living was the same and 4% of those who thought their standard of living was higher.

But this is a very particular set of circumstances, not a social science generalisation that can support Deaton’s hypothesis that recent economic growth causes a loss of happiness. Long recession-free periods are important in minimising the number of people going down as well as maximising the number of people going up. On the loss aversion hypothesis, supported by the Saunders survey, people are hurt by losses more than they benefit from gains. Australia’s long run of growth has seen retrenchment rates drop, sparing more people one of the proven causes of misery, unemployment.

In his paper, Deaton describes the result on ‘recent economic growth’ as ‘puzzling and surprising’. Indeed. If it it isn’t just wrong, as I suspect, it is likely to be due to idiosyncratic factors.