Poverty Isn’t A Thing

Here’s a little secret development nerds know that muggles struggle with: poverty isn’t a thing.

Those newspaper stories you see about rises and falls in “the poverty rate”? They boil the question down to a single summary statistic. A rate. At most two rates (“poverty” and “extreme poverty”).

That gives the illusion that poverty is like the temperature outside or GDP or the murder rate, things that can be captured completely in one number.

But a second’s engagement is enough to realize that that doesn’t make sense.

Poverty isn’t one thing: poverty is the outcome of a huge big amorphous mass of interconnected things. The standard bit of jargon is “multi-dimensional”. Poverty is the interstice of a series of deprivations that build on one another and tend to reinforce one another: in health, in money, in education, in housing, in safety, in employment conditions, in cultural capital and access to credit and access to decision-makers and dignity and a sense of power over your own life.

This, Rodrigo, is why I can’t endorse your denunciation of the Unmet Basic Needs methodology for measuring poverty. You argue that Unmet Basic Needs is a deeply flawed measure of poverty. And you’re right, because any summary number that pretends to stand in for a concept as slippery and evanescent as poverty will be deeply flawed.

You argue the components of the UBN approach are badly conceived. But any proxy that you could imaginably construct for poverty would be flawed because poverty isn’t a thing.

You could, of course, try to build better, more detailed proxies for each of the dimensions. You could try to measure education quality rather than just years in school, or housing adequacy rather than merely rooms-per-family-member. But you very quickly realize that there’s an inverse relationship between the availability of data and its level of detail.

Of course, there’s a reason why very detailed social data is not generally available: it’s murderously expensive to gather. At some point, social scientists with limited budgets are forced to make a judgment about the marginal value of the next datum. Returns are diminishing to detail. That stands to reason.

But to stress this aspect would be a mistake. The reason it’s hard to catch smoke using a wire mesh isn’t fundamentally that it’s expensive to buy enough wire to make a fine enough mesh.

Plenty of researchers have been tempted to just roll their eyes at all this multidimensionality mumbo-jumbo and just proxy poverty via money. That seems sensible…until you start to prod it.

First, you need to be clear if by money you mean “wealth” or “income”.  Make wealth your key proxy and you quickly find yourself arguing absurdities. It’s Felix Salmon who brilliantly pointed out that on a pure wealth basis, Jerome Kerviel would be the world’s poorest person: a suave, well-educated Parisian who knows his claret from his beaujolais but happens to be 6.3 billion dollars in debt. Try telling a starving Darfuri refugee that some stock trader in Paris is less wealthy than he is.

OK, so not wealth then: income. Except that’s yet another conceptual dog’s breakfast. How? Let us count the ways.

There is an enormous academic literature on the major (arguably, insurmountable) problems with cross-country comparisons of income-based poverty rates, and with comparisons over distinct periods of time.

Newspaper editorialists tend to gloss over these difficulties, but they’re real and, tellingly, the closer researchers look into these questions, the more they tend to become convinced that such comparisons are meaningless: Angus Deaton, for one, has more or less thrown in the towel, and that guy invented purchasing power parity.

But it gets worse. There isn’t any real consensus on how you build a “poverty line” and some people argue the concept really tells us nothing at all. There isn’t real consensus on whether defining poverty as some percentage of the median wage is a sensible thing to do, though some big organizations do just that. So even in first-world, data rich settings, trying to wiggle out of poverty’s non-thingness by saying “it’s all about income” doesn’t really solve your problems: in some ways, it makes them worse.

And of course Venezuela is anything but first world and data-rich: when basic consumption basket items are unavailable, a measure that compares their notional cost to incomes quickly becomes meaningless. In the presence of the mass of contradictions our Distortioland fairground-mirror-hall economy produces, money as a synthetic measure of wellbeing is just another mirage. And where the government can manipulate monetary variables for electoral gains, you see wild swings in income-based poverty rates (like its precipitous collapse in Venezuela in 2012 and its just-as-precipitous rise in 2013) that really don’t tell us very much about the deeper social dynamics at play.

The implication in your piece, Rodrigo, was that taking a money-approach to poverty measurements would sidestep the conceptual pitfalls of counting a house as an anti-poverty win regardless of how well (or badly) the house is built. And you’re right, income-measures do sidestep that conceptual frying pan, and drop you directly into the fire.

Poverty is like that. We’re desperate to corral it into thingness, but it bucks and thrashes all the way. The more you push and prod at the concept the blobbier it gets. The more we get exasperated and demand researchers cut to the chase and give us a number that tells me how many people are poor the more violence we do to the truth.

Personally, I’ve come around to Duncan Green’s formulation: the opposite of poverty isn’t wealth, it’s power. Power both in the big political sense of having a government that responds to your needs but also, and more relevantly, in the little personal sense of having meaningful choices over your life, having the power to command quality medical care when you’re sick, having the power to ensure your children are properly educated, having the power to go out for a stroll at night without fear of violence.

In this reading, the real problem with chavismo is the way it’s hijacked the rhetoric of empowerment while providing a cheap, knock-off version of the real thing that doesn’t really expand people’s choices meaningfully. In the Venezuelan context, it’s hard to even describe the idea of poverty-as-powerlessness without coming across like a deranged chavistoid ñángara. That’s how far up the creek we are.

But I digress.

The point is, poverty can’t be measured. It can only be proxied. Every proxy will have serious problems, and the more you closely you look into them, the more you realize that those problems aren’t really brushable-offable. They aren’t the sort of things you could fix with more and better data. They cut to the conceptual bone.

Which brings us back to Unmet Basic Needs, Rodrigo. It’s bad, of course. That’s not the question. The question is whether it, (and its even cruder cousin in the measurement of multi-dimensionality, the Human Development Index,) are better than the alternatives.

To me, the answer is clear: the UBN approach is the worst possible solution to the problem of measuring poverty…except for all the others.