petersuberMar 7

New from Oxford economist #OlivierSterck: "Measuring poverty on a spectrum instead of an arbitrary line conveys a more accurate picture of inequality."
https://theconversation.com/measuring-poverty-on-a-spectrum-instead-of-an-arbitrary-line-conveys-a-more-accurate-picture-of-inequality-271912

PS: I find this take on poverty very illuminating. Instead of measuring poverty by incomes (dollars per year), we should measure it by the time it takes to earn $1 (years per dollar). The classical measure allows a small number of very rich people to pull up the average income, and makes widespread poverty seem to be a minor problem. Sterck's new measure allows a large number of struggling people to pull up the average time needed to earn money, and makes a necessary correction to our understanding. The classical measure gives more voice or weight to the rich, when they're very rich, while Sterck's gives more to the poor, when they're very numerous. This helps us see two things otherwise invisible: widespread poverty hidden by average wealth, and the role of income inequality in hiding that poverty.

While the average income is higher or better in the US than in Europe, the average time needed to earn $1 is significantly longer or worse in the US. This kind of poverty exists alongside that high average income. We need to bring in severe income inequality to explain this, and of course income inequality is significantly higher in the US than Europe.

#Economics #Income #IncomeInequality #Poverty

Measuring poverty on a spectrum instead of an arbitrary line conveys a more accurate picture of inequality

An economist proposes a new method of estimating the scope of poverty in different countries.

The Conversation
Show threadJanne MorenMar 7

@petersuber
The problem is using averages. When it's not gaussian (a bell curve) it will give you the wrong answers.

We should use the median (and related measures like percentiles). It removes the distortion from a few outliers. It's also easier to understand, and we already use it when we refer to the top 1% earners and the like.

Show threadDan Neuman 🇨🇦
@jannem @petersuber Using an inverse causes the few rich outliers to have very little impact on the average. 1/$Billion = almost zero. I'm not sure using a median would have a significant impact here.
Mar 7 at 11:20pmWeb
Show threadJanne MorenMar 7

@dan613 @petersuber
With medians you don't need to use an inverse in the first place.

Also, the average still gives you the wrong results; the distribution is still not gaussian. The median is the better measure just about always (with gaussian distributions the median and average converges).

Show threadNaga (he/him)Mar 8

@jannem @dan613 @petersuber I'm not trying to contradict the broad point here by the article or Peter, but just weighing in on median:

As a quantitative social scientist, income and wealth are precisely the standard examples of when you should use median and never mean/average. The point is general about non-Gaussian distributions, but in the cases of income and wealth, it's especially about outliers. The moment you add Musk's wealth into an otherwise Gaussian distribution, the mean breaks down badly.

Show threadNaga (he/him)Mar 8
@jannem @dan613 @petersuber Non-linear transformations like inverse are patches on the problem. Helpful for some situations, depending on the next steps, but don't necessarily get you where you want, and have other implications.