I’ve just finished reading a book by the left-wing Peter Saunders, entitled The Poverty Wars. Much of it is a critique of the right-wing Peter Saunders, which as the author mentions at the outset has caused them both frustration.* He gets around this by referring to his opponent as "Peter Saunders" and to himself as "Saunders". In this post, I’ll take an even simpler approach – referring only to the left-wing Peter Saunders from now on.
There’s plenty to like in the book. Peter strikes a nice balance between the politics, statistics and anecdotes of poverty. I had some minor quibbles (eg. I think the connection between inequality and health is more tenuous than he acknowledges – though Imagining Australia also overstates the evidence, so I’m a glasshouse-dwelling stonethrower on this point).
But the big point is the "right" measure of poverty. Whereas Americans use a basket of goods approach** ("absolute poverty"), Australian and European poverty researchers prefer to measure poverty relative to average earnings ("relative poverty"). A commonly used standard in Australia is the fraction of people with size-adjusted family incomes that are below half the median (typical) family income.
So which is better? Critics of absolute poverty – like the left-wing Peter Saunders – charge that societal standards about what constitutes "deprivation" change over time. What Australians regarded as poverty in 1900 is clearly different to 2005. The only way of working out poverty, he argues, is to calculate it by reference to the income of the typical Australian.
On the other hand, critics of relative poverty charge that it’s really measuring inequality. If we tripled the income of those on and above the median, and doubled the income of those below the median, the poverty rate would rise dramatically. Somehow I don’t think this is what most Australians have in mind when they think about poverty. Indeed, Peter quotes from several surveys in which people seem to be rejecting the relative poverty notion in favour of an absolute standard. I’d reckon most Australians would be surprised to learn that regardless of how fast the incomes of the poor are rising, the poverty rate will rise if the median worker is doing better still.
Although Peter points out that relative poverty doesn’t measure inequality at the top of the distribution, it’s clearly highly correlated to inequality at the bottom of the distribution. And the two measures co-move closely. For example, when you take the Luxembourg Income Study data, the correlation between the gini and relative poverty is very high (0.91). I happen to think that inequality matters, so I’d rather see a serious debate about inequality, than having it dressed up as poverty.
At the end, Peter gives the example of the Blair government, whose "abolish child poverty by 2020" goals will take both absolute and relative poverty into account. This seems a sensible start. Better yet, they might consider jettisoning relative poverty, and taking about two concepts: absolute poverty and inequality. Absolute poverty lines might have to be rejigged every few decades, but they’re closer to what most people mean when they talk about poverty.
* I once met a 3rd Peter Saunders, who works in the Department of Finance.
** The US poverty line starts by tripling the food requirements for a family of three, and is then adjusted to account for family size.