Recently, I posted about how global poverty increased in 2020, and how that’s bad. I mostly talked about people who are above or below a 1.90 dollar cutoff, and only mentioned other, higher poverty lines as a sidenote. The questions here are: how do we choose one of those? And how much does the definition of poverty actually matter?
Poverty Measurement 101
My second post here, all the way back in April, was about this exact question: how many people are poor in Argentina?. How to measure poverty is a hotly debated question in economics, for a variety of reasons.
The first distinction to be made is between absolute and relative poverty. Absolute poverty is focused on whether or not someone has the minimal necessary things for life: food, shelter, water, etc. This is fairly alright as a universal measuring stick, but for many places it doesn’t make that much sense: how does “not having food” work to measure poverty in, say, California, a place that is generally decent at fulfilling those needs but with a very high cost of living? That’s where relative poverty comes in. Each place can define its own poverty threshold relative to the living conditions there.
You can measure both types of poverty in a variety of ways, such as defining a poverty line by determining how much fulfilling those basic needs cost, and then estimating how many people earn more/less than that amount. Other places use a different metric, such as a percentage of median income, which might sound arbitrary but it’s not more or less arbitrary than any particular poverty line.
Normally, most “standard” definitions prefer you use consumption, rather than income, to measure poverty. The reason is that many things that make or break someone’s standard of living aren’t related to how much they earn, but rather to what they consume. For instance, public education or housing isn’t a thing that can actually be easily deduced from your income, but having access to it should count towards not being in poverty. The main reason income is frequently used is because consumption data is much harder to find and create, and many places don’t actually have it, or have it but only fairly recently.
There’s also the notion of multidimensional poverty, which aims to include other indicators related to quality of living but that aren’t easily identified from either consumption or income: does someone have clean running water? have they had access to education? do they have electricity?. These surveys are generally more accurate at defining quality of life, but have a lot of issues related to sensitivity, i.e. how many questions are asked and how many "positives” are needed to count as poor.
Obviously each step of this is an enormous can of worms with a huge number of questions: which needs are basic? How do you settle on which prices to use? How can you extrapolate consumption from income data? How do you estimate the value of non-market items? How do you adjust for quality? And the list
Bill Gates’s Povertygate
Two years ago, Bill Gates tweeted out a chart showing how much progress had been made in certain metrics: extreme poverty, democratization, vaccinations, literacy, and child mortality. He linked to an Our World In Data post about these charts, which mostly explained how living conditions had been improving worldwide by many metrics (the ones cited above, in fact).
Gates’ tweet was followed by a The Guardian article by economist and anthropologist Jason Hickel, who I’ve spoken about in my post about degrowth, an ideology he’s a leading proponent of. Hickel’s article goes straight for the jugular, starting with the headline “Bill Gates says poverty is decreasing. He couldn’t be more wrong” - to be fair, he didn’t choose it himself and doesn’t like it, but it was picked by an editor. His core thesis is:
The trend that the graph depicts is based on a poverty line of $1.90 (£1.44) per day, which is the equivalent of what $1.90 could buy in the US in 2011. It’s obscenely low by any standard, and we now have piles of evidence that people living just above this line have terrible levels of malnutrition and mortality. Earning $2 per day doesn’t mean that you’re somehow suddenly free of extreme poverty. Not by a long shot.
Scholars have been calling for a more reasonable poverty line for many years. Most agree that people need a minimum of about $7.40 per day to achieve basic nutrition and normal human life expectancy, plus a half-decent chance of seeing their kids survive their fifth birthday. And many scholars, including Harvard economist Lant Pritchett, insist that the poverty line should be set even higher, at $10 to $15 per day.
This is fairly reasonable, and I’ll circle back because it’s a core part of this debate. He also makes two other main points, which are complete nonsense. The first is that everyone who brings up these figures is an austerity-loving neoliberal who’s ready to shower praise on the free market for reducing global poverty - which is far from the case, and is something that only a handful of people actually claim. The second relates to colonization: people who are counted as poor by the aforementioned figures before their countries were colonized weren’t actually poor, they just didn’t earn any money but had their needs otherwise satisfied. This isn’t actually true: life in premodern times was generally awful, since people had very short life expectancies because living and health conditions were appalling. I’ve written a bit on the “Malthusian phase” of human life, and it wasn’t pretty - the kind of lives Thomas Hobbes called “nasty, brutish, and short”.
Stephen Pinker, one of the people quoted in the Hickel piece, responded to a blogger’s email about the subject. He raises a number of objections to Hickel’s point: that the decline in global poverty isn’t some neoliberal artifact conjured up to defend free trade, but a generally neutral claim based purely on facts; that the precise poverty line used isn’t relevant, because the entire distribution of income shifted to the right (i.e. everyone is generally better off); that the decline in global poverty is a global trend, not a China and India specific topic; that his claims about idyllic primitive people are baloney; that other measures of wellbeing (not just poverty) corroborate this story. He then makes claims about a “far left agenda” that is humiliated by the fact that markets and globalization improved lives, but that’s neither here nor there.
Hickel then responded to Pinker, mostly by doubling down on most of his points, from the inadequacy of the poverty lines to the evils of colonialism (which, to be fair, he’s right about, except for the “and pre-modern life was good actually” part) and then makes a couple of pretty interesting points on how exactly global poverty is measured. Developing countries don’t have a lot of data: for Asia and Latin America, only three countries in each have data older than 1900; for Africa, no data that old exists, and only 3 countries have data older than 1950. And the data generally used is the Maddison Project’s GDP per capita estimates which have the same problem, plus the problem of imperfect translation and not so good treatment of non-monetary resources (he cites rivers and forests). And lastly, he mentions a very weird thing: the number of people living at higher poverty lines increased, which is mainly because of the number of people on the planet increasing. And the “poverty headcounts are what really matters” line of thinking has had horrific consequences in the fact when taken to its logical conclusion, as I’ve talked about on my post on overpopulation.
Then, Our World In Data decided to rebuff Hickel’s criticisms of their methodology by explaining it in detail in a post of their own. They rebuff the criticisms that estimates of GDP per capita don’t take into account non-market income (we have fairly comprehensive data on, say, medieval Britain) and that they’re not very accurate - they’re mostly based on work that experts on each place actually estimate on their own, and are then verified and put together by the Maddison Project. Poverty estimates flowing from those were done fairly carefully by works like “The Economics of Poverty” (2016) by Martin Ravallion, “More Relatively-Poor People in a Less Absolutely-Poor World” (2012) by Chen & Ravallion, “The Changing Shape of Global Inequality 1820–2000; Exploring a New Dataset” (2014) by Van Zanden, Baten, Foldvari, & van Leeuwen, or the oldest source generally used, “Inequality Among World Citizens: 1820-1992” (2002) by Bourguignon & Morrison. They also talk about colonialism being really bad for economic growth, which isn’t something anyone who’s looked at data for more than five minutes disagrees with.
That post also brings up something that should be really obvious: people were really badly off in pre-modern civilizations: no medicine, no running water, war, disease, etc. And even if, for some reason, the people in the past were richer than we expect, we can only really say so much about its impact on growth: very few people lived in places that we have very little data on: Africa particularly only accounted for 8% of the global population, so that’s the most it can reduce it by (and we know, factually, that a big chunk of that was extremely poor, so it’s less than that at most. Saying “it’s only India and China” (for which we have fairly long and comprehensive data) is far less of an own when a third of the global population lives there, and that fraction used to be much higher.
Hickel responded, mostly by focusing on the inadequacy of the poverty line, the inadequacy of estimates for before 1981 (the year the World Bank started collecting extreme poverty directly), and a few other methodological quibbles.
Finally, someone who’s a huge deal in poverty economics showed up: Branko Milanovic. Milanovic is one of the biggest experts on global poverty and inequality in the world. He starts by owning Hickel real bad, because he thinks that the Maddison Project GDP per capita data is great, and that Martin Ravallion and Max Roser are great dudes who’ve done a lot of great work. But he also agrees with Hickel on a few points: that non-monetized activities aren’t very well handled, that pre-colonial lifestyles might have been better because colonialism was that bad, and that the data isn’t very high quality but it tends to check out (except for some of the Bourgnignon and Morrison data, which he doesn’t think is very good).
What we agree on…
The points of agreement across mainstream voices (and also Jason Hickel) are plenty: data from after 1981 is acceptable, and it shows a very clear story: extreme poverty is declining, regardless of measurement (1.90 a day, 3.20 a day, 5.50 a day, 7.40 a day) though at different speeds and degrees. Hickel and Charles Kenny of the Centre for Global Development even wrote a pretty concise summary of what everyone agrees on, that I’m just going to summarize:
There is no single simple narrative of progress; much depends on the metrics we use. (…)
There are many problems with the global income and consumption data—the underlying sources are imperfect and the aggregates rely on numerous methodological choices…
The $1.90/day (2011 PPP) line is not an adequate or in any way satisfactory level of consumption; it is explicitly an extreme measure (…)
Aggregate numbers can obscure diverse experiences.
The proportion of people living under $1.90 per day has declined significantly, but poverty as measured by $7.40/day has declined more slowly…
The absolute number of people living under $1.90/day has declined significantly, while the number of people living under $7.40/day has risen
The average consumption of people below both the $1.90 and $7.40 poverty lines and above those lines has increased. The “poverty gap” (the average distance below the poverty line) has been shrinking.
Between 1981 and 2002 most of the gains against global poverty at $7.40/day came from East Asia and the Pacific…
Since 2002, every developing region has seen a decline in the proportion of people living under both $1.90 and $7.40…
Income and consumption does not tell us the whole story about poverty…
The present rate of poverty reduction is too slow for us to end $1.90/day poverty by 2030, or $7.40/day poverty in our lifetimes…
Ultimately, the more morally relevant metric is not proportions or absolute numbers, but rather the extent of poverty vis-a-vis our capacity to end it
I’d also add “colonialism is bad” to the list, for what it’s worth.
… and what we disagree on
Anyhow, what do people disagree on? The significance of any precise poverty line. When Hickel says 1.90 dollars a day is inadequate, he isn’t talking out of his ass. Going back to 2003, this has been a hot topic of debate.
Development economics heavyweight Lant Pritchett argued, all the way back in 2003, that the World Bank should use a different poverty line, one that’s the average of its member countries weighted by economic might - which is about 15 dollars per day PPP (Hickel also had some thoughts on the PPP methodology, but that’s not worth delving into). Besides from methodological concerns, Pritchett makes a pretty interesting argument focused on unfairness: why do we accept that someone born poor in a rich country counts as poor when they’d be at least upper middle class in a poor country? Unless we define a poverty line that works for everyone, we’re just accepting gross (global) inequalities.
(Future) Nobel Laureate Angus Deaton also had some opinions on the matter. He claims that, because consumption isn’t measured from the same sources as GDP, then it’s measured differently - particularly, it’s measured at a lower level, which might also be because there are fewer rich participants in these surveys. Because of inequality in consumption hanging around and making everything a pain, we’re understating poverty reductions across the world, and overstating income growth.
Philosopher Thomas Pogge and economist Sanjay Reddy also developed an interesting position. Besides from a pretty well laid out case against the methodology (those come a dime a dozen it seems), particularly how international comparisons of prices are done, how the estimates for the amount are estimated, lack of representativeness, plus the fact that poor people (and poor countries) face different prices, because of the more limited set of choices they possess.
Economic historian Robert Allen also had a take on this: the World Bank’s approach is fine, but an approach based on how much food costs, and how many times of what each country spends on food expenses it spends on other non-food expenses. This would solve the representativeness issue, would make comparing across time easier (at least as long as food prices are available), doesn’t have problems with prices, and doesn’t have information collection issues.
Also, Max Roser of Our World In Data has endorsed a higher poverty line in a paper of his own, and it seems Jason Hickel came up with his threshold for poverty out of thin air - for someone intent on not shifting the goalposts and not defining poverty around an arbitrary line, he sure seems to be doing that.
Conclusion
In the end, any individual poverty line is meaningless - even if any specific level has fewer problems, it doesn’t tell you that much about actual living conditions. Regardless, we should take each individual threshold as more of a general indicator - fewer people living in objectively horrific living conditions, and instead “graduating” to just terrible conditions. Choosing the poverty line of a developed country might work, but only as a final stage in a project of building a more equal world.
The fact is, people in rich countries are many times more richer than people in poor ones - global inequality is basically as high as it was 100 years ago, with a Gini coefficient higher than any country in the world’s. Adjusting for the higher quality of life, the higher acess to public services, and the higher quality of those services, the world’s poor don’t hold a candle to everyone else.
Sources
Measurements
Atkinson (1987), “On the Measurement of Poverty”
The controversy
Jason Hickel, “Bill Gates says poverty is decreasing. He couldn’t be more wrong”, The Guardian, 2019
Jerry Coyne, “Is the world really getting poorer? A response by Steve Pinker”, 2019 (Pinker’s email is at the bottom of this piece)
Hasell & Roser, “How do we know the history of extreme poverty?”, OWID, 2019
Jason Hickel, “A Response to Max Roser: How Not to Measure Global Poverty”, 2019
Branko Milanovic, “Global poverty over the long-term: legitimate issues”, 2019
What we agree on
Hickel & Kenny, “12 Things We Can Agree On about Global Poverty, CGDev, 2018
What we don’t agree on
Pritchett (2003), “Who is not Poor? Proposing a Higher International Standard for Poverty”
Deaton (2005), “Measuring Poverty in a Growing World (Or Measuring Growth in a Poor World)”
Pogge & Reddy (2005), “How Not to Count the Poor”
Allen (2017), “Absolute Poverty: When Necessity Displaces Desire”
Conclusion
Charles Kenny, “Really, Global Poverty *Is* Falling. Honest.”, CGDev, 2017
Milanovic (2011), “A short history of global inequality: The past two centuries”