Measuring Poverty

The folks who run HILDA (Australia’s leading panel data survey) are doing some interesting thinking about consumption inequality, income inequality and wealth inequality, which could significantly change our estimates of poverty in Australia. Here’s a preview. (HT: Jeremy Lawson)

This entry was posted in Inequality. Bookmark the permalink.

1 Response to Measuring Poverty

  1. christine says:

    They’re actually asking people to make suggestions on how to improve the data? And telling people to call them about it? And asking anyone interested in using the data to ask for help if they want it, “day or night”???


    On substance (caveat – I’m not an expert on this): no surprise that measured consumption inequality is less than measured income inequality. We’d expect that on a permanent income basis, if nothing else. But given that compression, is defining poverty as 50% of the median, and comparing the poverty estimates across the two measures sensible? If you use any (completely ad hoc) simple Keynesian consumption function – that has some fixed level of consumption – to determine consumption from income, you’re going to get lower poverty estimates from the made-up consumption numbers than from the income numbers (according to simple calcs from a dummy data set). But that’s not really getting at permanent income/lack of poverty, just at the impossibility of consuming nothing, even if you have no income.

    Of course, the 50% (or 60%) rule is pretty ad hoc anyway. But still, I’m not sure what conclusions to draw from a lower poverty estimate using the same definition of poverty but attached to consumption instead of income.

    Wouldn’t what we want be both consumption lower than some limit (given family type) AND decline in net assets greater than expected given stage in the lifecycle? So: a students who is income poor, (relatively) consumption rich, has declining or stagnant net financial assets, but increasing human capital => net asset position increasing => probably not poor (policy solution: loans). Retiree with low income and relatively high consumption has net asset position decreasing, but would be expected at their stage of life => not poor. Retiree with low income, low consumption and rapidly diminishing net asset position => poor.

    Poverty estimates don’t often really control for stage of life, though, except by calculating estimates by individual demographic groups. Probably partly because things then get massively overcomplicated?

Comments are closed.