Inequality kills

Inequality causes suicide. That's the message of this paper (pdf) from the Federal Reserve Bank of San Francisco:

Inequality in the upper tail [the ratio of 90th to 50th percentile incomes] raises suicide rates and inequality in the lower tail [the 50th-10th percentile ratio] lowers suicide rates...A doubling of the 90/50 ratio is associated with an increase in the suicide rate of 2.5 per 100,000 in 1990 and 1.7 per 100,000 in 2000.

One interpretation of this is that Richard Layard was right. Inequality, in some forms, makes us miserable. Someone around median income (where most suicides occur) becomes unhappy if people earn a lot more than him, but happier if some people earn a lot less. At the margin, these changes in happiness affect suicide rates.
This could be a little embarrassing for those (including me) who have been critical of Layard. So, what replies are possible? 
One is that broad-brush data contradict the notion of clear link between inequality and suicide. Suicide rates among blacks, who suffer from inequality, are lower than those among whites. And in the UK,  the suicide rate recently hit a record low, even though inequality is higher now than in the 1970s or early 80s. 
Another reply is that these effects, whilst statistically significant, are small. In the same data, a one percentage point rise in a county's unemployment rate raises the male suicide rate by 26 per 100,000. If inequality is the price to pay for fuller employment, therefore, it's one worth paying, measured in terms of suicide-reduction.
Thirdly - as the authors recognize - it might not be that people are comparing their incomes to others directly, as Layard thinks.
Instead, inequality might matter for suicide because it's a measure of income risk. Someone around median income might want a low 90-50 ratio, because this indicates a high probability of rising to a high income point,  but they might also want a high 50-10 ratio as a sign that there's a low probability of their income falling a long way. 
If this is right, the case for Shiller-style income insurance is stronger, isn't it?

Comments

[dearieme]

"raises suicide rates": not "is associated with higher suicide rates"? Correlation = cause, eh? Perhaps suicides cause inequality, then? Maybe I've become too fond of this, but 'In social science, data is the plural of bollock.'

[Alex]

errrrr....

Correlation only implies causation if you can propose a plausible mechanism for such causation. Layard's theory is arguably such a mechanism.

What was your point again?

[chris]

"Data is the plural of bollock" - except when they corroborate our prejudices.

[dearieme]

"Correlation only implies causation if you can propose a plausible mechanism": you'll have to do better than that, or accept the argument of the wag who said that global warming was caused by demography - all those female baby-boomers having hot flushes.

[dearieme]

Got it. The inequality means that rich business types in late middle age are able to land young trophy wives, thus causing younger, poorer men to have a lower standard of happiness - both from spouse-deprivation and from envy - and thus be more prone to top themselves. Any evidence that suicide is commoner among men, and particularly among men a couple of years older than the average bimbo? The degree of inequality in the top tail is crucial because it's wealth that allows convex older guys to attract concave younger women.

[Tim Worstall]

Surely desirability in women depends not on covexity or concavity but on the location of such?

[Chris Williams]

Suicide rates are related to opportunity. Someone worked this out by doing a regional breakdown during the period when the UK was converting from coal gas - the suicidal housewife's best mate - to North Sea gas. This probably also has some effect on the higher sucide rates among farmers and medical workers: it's easier for them.

So the lowering of the UK suicide rate since the 1970s (although I'm rather sceptical of headline rates like this) may well be accounted for in part by changes in opportunity.

[Alex]

"Correlation only implies causation if you can propose a plausible mechanism": you'll have to do better than that, or accept the argument of the wag who said that global warming was caused by demography - all those female baby-boomers having hot flushes.

Posted by: dearieme | May 02, 2006 at 04:08 PM

Is that the real dearieme or a parody troll? Still, I'll bite. That's not a plausible mechanism. 2-0.

[NewDearieme]

Hot flushes are a more plausible mechanism than CO2: absorption phenomena are logarithmic not exponential.