Our World in Data?
In recent years many charts from Max Roser and his colleagues at Our World in Data have been popping up on social media. Roser and OWID seek to build on the admirable data visualization legacy of Hans Rosling. Unfortunately they risk lulling us into a complacent sense that the job of defeating deep poverty is nearly done. Residual mop-up operations can be expected to soon be over, with little additional effort, since the graphical trajectory is already well established. Unfortunately, OWID’s headline charts, like the inaccurate stories they trigger, are generating a great deal of confusion along with the real insights they deliver.
I love the villages of the developing world, and have shared much of my life with the people who live in them. To help bring that perspective to OWID and its readers, I’ll use two examples: poverty and how it gets confused with cash, and literacy and how it gets confused with livelihoods. In these areas absolute progress is simply not being measured, and until we start measuring it we need to carefully manage the presentation of available datasets that are clearly inadequate proxies.
Poverty and Cash
Chart 1 below purports to depict the “World population living in extreme poverty, 1820–2015”. The subtitle: “world population living at a consumption (or income) level below 1.90 “international $” per day” may be quite accurate — at least for the modern era for which our data is modestly reliable. But the unhelpful conflation of ‘extreme poverty’ in the headline with ‘$1.90 a day’ threatens to mire a very serious challenge in complacency and indecision at a decisive moment in human evolution.
Chart 1: Trajectory of Poverty Since 1820 (OWID)
Looking at this chart, a reasonable person might conclude that the plummeting trend-line since 1950 has solved the problem. Chart 2 is a projection using OWID’s data. Will extreme poverty in fact evaporate from the face of the earth by 2025?
Chart 2: Projection of Extreme Poverty (based on Chart 1)
No, it won’t - and the conceptual framing of Chart 1 can tell us why.
In 1820 very few of the 1 billion or so people living on our planet earned more than $1.90 US a day (in 2017 dollars, adjusted for inflation). This did not mean they lived in poverty, however. With only a billion humans on a land mass that was as large then as it is now, and substantially less environmentally damaged, anyone with secure access to a modest amount of land could raise a family and avoid extreme poverty with no money at all — either coins or notes. Silver and gold coins, cowry shells and private paper money were mostly used for long-distance trade, not daily life. Monetization of daily life only really began when central banks began printing state-backed money in the early-mid 19th century.
OWID is quite open about its bias here.
The available long-run evidence shows that in the past, only a small elite enjoyed living conditions that would not be described as ‘extreme poverty’ today [emphasis mine]. But with the onset of industrialization and rising productivity, the share of people living in extreme poverty started to decrease. Accordingly, the share of people in extreme poverty has decreased continuously over the course of the last two centuries.
By this logic I could argue — adopting the perspective the year 2200, after universal artificial intelligence eliminates the need to work and the disease of old age has been cured — that we all live in ‘extreme poverty’ in 2017. Reader — are you in ‘extreme poverty’? I submit that only if you were an unreformed ‘2017’ human transported into 2200 without any of the resources — material or human — to access its benefits, would you begin to feel that your current material condition is ‘extreme poverty’.
As of 2017 the $1.90 US-a-day figure, supplied by the World Bank, is a somewhat useful proxy for extreme poverty. This International Poverty Line has been revised upwards several times since 1990, when it was first established at $1.00 US a day. It is adjusted for local purchasing power, and the periodic changes that governments make in their own national poverty lines. It excludes ownership of assets like land, housing and human capital.
Two centuries ago such a poverty line would have been of little use — but those centuries have wreaked havoc on the subsistence and in-kind economies that shaped much of our ancestral history. There is far less land per capita now. There are also far more people today than the technological endowments and human capabilities of 1820 could have supported. The monetization of the modern world has been one of its defining narratives, transforming money from a luxury to a basic survival need. In the process money has reshaped both the nature of poverty and our very nature as human beings.
But vestiges of the older world remain: some of the last survivors of total cyberwar might well be hundreds of millions of villagers who live where money remains — even today — a marginal factor in household decisions about food security, shelter and other basic needs. A monetary proxy like the IPL is designed for fully monetized environments, and can capture poverty well there. It is much less effective in exactly the partially-monetized places where most people who earn less than $1.90 a day live.
As we continue to trend towards global monetization, the case for the IPL (or something like it) is sound. But to understand poverty as it exists today we must also understand the effects of the world we have made. Half a century ago, Marshall McLuhan put it this way.
Environments work us over and remake us. It is man who is the content of and the message of the media, which are extensions of himself. Electronic man must know the effects of the world he has made above all things.
OWID could offer real insights by adding a separate indicator to the chart that shows how much money was needed to avoid extreme poverty from one generation to the next since 1820 (see Chart 3). Has the money supply actually commanded by people risen faster than this minimum?
Chart 3: Concept — Poverty and Cash, 1820–2015
In 1820 most people lived subsistence lives on the land, so for that year this indicator would have to be set very close to zero. As population growth has gradually forced more and more of the population to rely on money to meet the basic needs of daily life, the red dotted line shows an accelerating upward progression, with $1.90 a day being plausible in 2017 — at least for populations that still maintain some subsistence-based land access to supplement their cash incomes.
Reading, Writing and Livelihoods
Chart 4 shows OWID’s understanding of our progress against illiteracy since 1800. Again, any reasonable person might conclude from the graph that the plummeting trend-line, especially since 1950, has solved the problem.
This conclusion would also be incorrect, and again the framing of the chart can tell us why. In 1800 literacy may have been a valuable asset, but (like paper money) was not remotely essential to survival. In only a few leading nations was it common to find printing presses, newspapers or public signs containing writing. Elsewhere only a tiny minority could read or write — and most of their efforts were devoted to religious texts. Few livelihoods required literacy, and oral apprenticeships supplied the skills needed for many advanced trades: trades that rarely included reading or writing.
Chart 4: Trajectory of Literacy since 1800 (OWID)
Like the economy without money, the culture without text lingers in many places, in slow transition into modernity. Even villagers who still live a subsistence life receive ID cards as well as written information about their children’s education, their health, their property rights and their financial transactions. Failure to understand this information is common, and can have uncomfortable consequences for themselves and for their families.
The blue line in Chart 5 adjusts the OWID data for absolute numbers. There were 1.07 billion illiterate adults (aged 15 or more) in 2014 — down from a peak of 1.95 billion in 1980. However, progress has virtually stalled since 2000, when there were 1.10 billion. (This is muddied by OWID’s headline chart, which shows illiteracy continuing to drop — but only as a percent of the rapidly rising human population.) The red line in Chart 5 sketches a hypothetical but plausible trendline for the percentage of adults who have required the ability to read and write to avoid extreme poverty, by era. In 1800 this indicator would again be set very close to zero. This indictor shows an accelerating upward progression, with literacy being optional for a shrinking minority of humans in 2017.
Chart 5: Concept - Reading, Writing and Livelihoods
The stall-out since 2000 highlights two problems, one technical and the other substantive.
- Data quality: The data is collected by the world’s 200-odd national governments. Some collect it on census forms by asking people at their doorsteps if anyone in the home is illiterate. Others collect it by inference: functional literacy is decreed to be a curriculum goal by grade 4, ergo any grade 4 graduate must be literate. Both these methods tend to systematically underestimate adult illiteracy. Through direct nationally randomized testing, UNESCO is assembling a clearer picture, and finding hundreds of millions of additional illiterate adults. But there is a long way to go, and we still don’t know the real numbers.
- The legacy effect: Subsistence farmers often devalue literacy, because it doesn’t help them feed their families. To test for this effect, UNESCO has done direct cohort testing in several countries. For example, 21% of Ethiopian women 20–34 were literate in 2000, but only 16% of the same cohort could pass a literacy test a decade later. This pattern turns out to be very common, and raises questions about OWID’s assertion that “younger generations are progressively better educated than older generations. And … this intergenerational change is happening especially quickly in the least educated regions of our world”. As long as hundreds of millions of people depend on land and animals — not money or other products of literate culture — for their survival, illiteracy is likely to prove much more stubborn than Chart 4 suggests.
Remade by Our Environment
Roser argues that industrialization has been wiping out poverty. He narrates the story of Nathan Rothschild, who was “surely the richest man in the world when he died in 1836. But the cause of his death was an infection — a condition that can now be treated with antibiotics sold for less than a couple of cents.”
Technology is wealth — but there is an important corollary. Even the poorest people today must have capabilities that were beyond those of all but the most capable a few centuries ago. Antibiotics are only useful when people know why and how to use them. The web remains very daunting for those who can’t read (it is slowly improving). This human transformation is what McLuhan meant when he remarked that “[e]nvironments work us over and remake us. It is man who is the content of and the message of the media …” This is not an argument against monetization or literacy — let alone an idealization of a past era. It is a call for level-headed awareness of who we are, and where we are going, in the present.
An example of just how powerfully environments remake us is the Flynn Effect, documented in multiple studies since James Flynn’s groundbreaking research in the 1980s. The average American or European today has an IQ approximately 30 points higher than his ancestors in 1900, due to a process that has seen a secular rise of about 1 IQ point every three years in the industrialized West. Flynn explains:
Our ancestors in 1900 were not mentally retarded. Their intelligence was anchored in everyday reality. We differ from them in that we can use abstractions and logic and the hypothetical to attack the formal problems that arise when science liberates thought from concrete situations.
But the Flynn effect is not confined to advanced economies. In 1964, long before the Flynn effect was documented, McLuhan wrote:
… it is in our IQ testing that we have produced the greatest flood of misbegotten standards. Unaware of our typographic cultural bias, our testers assume that uniform and continuous habits are a sign of intelligence, thus eliminating the ear man and the tactile man.
The Flynn effect has been identified in studies in several developing nations, including among school children in a rapidly urbanizing Kenyan community (an increase of 11 IQ points over a 14-year period) and in the city of Belo Horizonte, Brazil (an increase of 17 IQ points over 72 years).
There is much debate about IQ these days, and the World Economic Forum published comparisons by country in 2013. In the 19th century the quality of a student’s hand-writing was considered a mark of intelligence and education. Today the fact that students graduate from high school scarcely able to write by hand elicits general indifference. Is literacy facing a similar trajectory? As voice soaks into every aspect of digital information processing from input to storage and from output to search, are we nearing a moment when literacy joins hand-writing in history’s scrapyard? This is not clear yet, but anyone persuaded that such a trend would automatically improve the lives of illiterate adults may be sadly misguided. How can smartphone ownership — by itself — overcome what Flynn calls the ‘pre-scientific’ mindset among adults whose survival depends on their crops and animals?
Industrialization has reduced extreme poverty and illiteracy over the centuries — but it has also profoundly reshaped them. Stranded just outside the gates of the modern industrial world, with its regular paychecks and its accelerating internet-of-everything, the oral world is getting trapped on the wrong side of an increasingly unbridgeable chasm.
There are solid grounds for an optimistic stance on the challenges we face: our progress against maternal and child mortality, life expectancy, and many tropical diseases, deserve all the favourable headlines we can lavish on them.
But let’s not succumb to the temptation to indulge in click-bait optimism that diverts attention from the serious challenges we must still confront. A fairer depiction would leave no one — not politicians, journalists or practitioners — confused about the magnitude of the task still ahead of us, or the scale of the risk resulting from a widening technological divide.
We like to think of our world as a ‘global village’. Let’s give the real villagers a chance to join us.
If you are interested in more discussion of this vital but sadly neglected topic, please leave a comment, and if you like this, don’t forget to clap :)
 Flynn, James R. What is Intelligence? Cambridge University Press, Cambridge, New York, 2009, pp. 10–11.
 McLuhan, Marshall. Understanding Media: The Extensions of Man. MIT Press, 1994, p. 17.
 Daley, Tamara C. et al. IQ on the rise: The Flynn Effect in rural Kenyan children. Psychological Science, 14 (3), American Psychological Association, 2003.
 Colom, Roberto, Carmen E. Flores-Mendoza and Francisco J. Abad. Generational changes on the ‘Draw-A-Man’ Test: A comparison of Brazilian urban and rural children tested in 1930, 2002 and 2004. Journal of Biosocial Science, Cambridge University Press, 2007.