As that title is broad, let me say in preface this is written in a Red Teaming spirit[1]. Outside this essay, my conviction in market capitalism is stronger. The biases I wish to interrogate are typically more aligned with its shorthand criteria and vernacular proxies. That proviso should also suitably prime interpretation of the empirical evidence I am about to share.
I am trying to get you to entertain two propositions:
There is no such thing as a developed country
Net global progress is impossible to measure without misleading shortcuts
In explaining why I raise those, I’ll focus on what data we have access to and how we interpret it. ‘Models are not reality’ is far from a new message. But alongside the main models, there are supplementary models around with more evasive data. If you have only a minute, skip the text and glance at the graphics below.
Development as Progress
The biases peek right from the introductory pages at EA, in axiomatic reading of economic models spun from ‘that which has already been measured because it was easy’. Typically, incomes and output.
Until a few days ago, as an example of EA in practice, the introduction page (archival link) presented the common graph of historical economic output through the Industrial Revolution[2]. It was transformative. AI is likely to be. Hence we should ensure that AI only does ‘good’. But did the Industrial Revolution only do good? Researchers in the subject regularly attempt to first operationally define the good and then proceed to estimate how much good it did. The assumption in omitting any mention of downsides seems to be that there were none. It is far rarer to see anyone even listing the categories of bads that came from the Industrial Revolution and hence hardly any attempts to estimate the global total such that we may compare. In the absence of that estimate or an acknowledgement of it, invoking the Industrial Revolution as an example of why AI needs attention, as that page still does (August 22), indicates gaps in thought.
As I lay the argument that ‘net progress’ is an extremely difficult question, I will mainly steer attention away from Europe and North-America, but even if EA only looked closer to home in space and time to begin with, the Industrial Revolution would reveal its mixed wages.
In the EA Handbook, Max Roser of Our World in Data begins to address problems with the quantification of progress but is not nearly nuanced enough. A different post at Our World in Data better sums up apparent EA articles of faith: an exhaustive one on extreme poverty.
One graph on that page, ‘Share of the World Population living in Absolute Poverty, 1820-2015’, shows the drop in the last two hundred years. Though the ‘$2 a day’ you see there is loosely adjusted for inflation, it imposes 2002’s North-Atlantic elite ideas of what people need (that they can only get at a market) on everyone who lived in 1820 anywhere. The painstaking work in the paper at the source sure has its worth. (In fact, I’ll lean on it in a bit.) But to use the estimates[3] as the definitive guide to assessing our economic systems is a failure to truly put oneself in the milieu we wish to understand. The authors even acknowledge some of the difficulty, but the commentary accompanying another graph on that same page only underscores how deep biases sit. Here’s the graph:
I see a negative correlation, but a very poor one. But the authors believe:
It shows that… people living in poorer countries tend to be less satisfied with their living standards. This suggests that economic prosperity is not a vain, unimportant goal but rather a means for a better life.
‘Economic prosperity’ as in output, on that graph. That is nearly the opposite of my reading. Bangladesh, Bolivia, Egypt and Tajikistan report much the same levels of satisfaction as USA, Finland, France and Japan. India, where I live, also appears in the vicinity.
The disparity in interpretation is revealing[4]. Beside that two-centuries graph, the authors say:
Nathan Rothschild 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. Today, only the very poorest people in the world would die in the way that the richest man of the 19th century died.
Yes, 19th century aristocrats and tycoons died from diseases that can today be cured with antibiotics available to billions over the counter. But still not available to everyone. How many people today do not know of or cannot access these drugs? As the otherwise comprehensive page does not mention it, I suspect we do not have a good estimate. Also, in absolute terms, was the number of people susceptible to that disease larger in 1820s or is it today? There is a rational, mathematical reason to ask: some of the world’s unequal progress has relied on unequal population growth. (More on this below.)
Imagine a teenaged British aristocrat from your favorite period TV drama whom you manage to befriend and introduce to one of the people on those data charts who’s been lifted just above extreme poverty in recent years. You offer them to trade places with a factory worker in 2018 with a guaranteed longer life. Do you think they’d jump at the chance? ‘But ah’, you counter, ‘the comparison should be to the closest job to factory work that existed in 1820’. Right, but if it is easy to imagine that an aristocrat would be giving something up that they value in 1820, how sure are we that a ‘peasant’ from 1820 might not lose something they value in the same time travel too? Should we dismiss the question because we cannot directly ask? Or should we look around the world today to inform a better guess?
More to the economic point, when average life spans were in the 40s, what proportion of people had to spend 50 hours a week or more year-round on hard work plus a commute just to make ends meet? What proportion had to spend at least 12 years fulltime in education to even have a shot at joining that daily grind? What proportion were hired as a draught animal or minded like a machine? How much of every year was life and how much work? How stressed did everyone constantly feel?
We cannot have incontrovertible evidence to settle any of this. Why then do we try to reassure ourselves that industrialization was only good for humans in the balance? Why aren’t any downsides at least footnoted in paeans to the Industrial Revolution even amid growing reports of an anti-work movement and quarter-life crises?
The somewhat vacuous ‘living standards’ comparison across centuries is something my fellow economists came up with to answer the quibbling critique that capitalism seems to both require and perpetuate inequality visibly in our lifetimes.
To help make some of this feel relatable, let me pitch from the shoulders of someone influential. In June 2021, Ezra Klein wrote a column titled ‘What the Rich Don’t Want to Admit About the Poor’. Here’s the thrust of it:
The American economy runs on poverty, or at least the constant threat of it. Americans like their goods cheap and their services plentiful and the two of them, together, require a sprawling labor force willing to work tough jobs at crummy wages.
The constant threat of it. In the 21st century, just as we glimpsed some of the earliest galaxies being born. You might think this has always been in plain view and a subject of scholarship for two centuries at least. Anyone who has lived in India or another low-middle income country would also recognize the economic system described there. Even for those who didn’t before, the 2020 pandemic laid it bare.
But there is a wider aspect we don’t broadly and universally acknowledge. What is true within the US economy and within the Indian economy is also true of the global economy. In fact, if the phenomenon concerns you, the only rational way to analyze it is to examine the whole. And the global picture is startling.
Network Dependent Structures of Economic Progress
That ‘poverty since 1820’ paper is Bourguignon and Morrison (2002). (I urge you to read it alongside this section.) After deriving their estimates, the authors inform you that if all the growth in income since 1820 had been more equitable, the world in 2002 would have 1/4th the number of people living in poverty and 1/8th living in extreme poverty. By switching to the proportion of poor in the total population, Our World in Data uses the same estimates to leave you with the impression that industrialized growth has been good for reducing poverty. It is not merely a change in the denominator. Volumes of information on the gears of our global economic system lie across the two tellings. Newer papers marshal the data to visualize it.
How does economic output grow?
Structural Transformation is a story of how economies grow more prosperous. Over time, the economic output of a country shifts from being predominantly agricultural to manufacturing and then into services. Relative differences in productivity between sectors ‘naturally’ drive this transition. After all, all the countries that are wealthy today went through this transition in the last century or so.
There is very good evidence to support that theory and there is little to dispute that the account is valid for wealthy countries. There is less consensus on whether the account offers any iron-clad direction to low-income countries today. One of the complications is that the structural transformation story leaves something out.
Work in rich nations didn’t move from agriculture to manufacturing exclusively because spectacular productivity gains meant that fewer agriculture workers could entirely meet their own country’s need for agricultural produce. Instead, with rapidly expanding supply chain networks, they did not need to. They could offload some work to agricultural labor in poorer nations. For the next transition away from manufacturing, rich nations again had a population of poorer factory workers (migrating away from shrinking farms) across the ocean waiting to take over.
But of course, the same process cannot cascade down endlessly. New parts of the world can only get (slightly) richer while there is someone in the queue waiting to take up less lucrative sectors of the economy, and they in turn must have someone even poorer able to supply primary goods to them and so forth. For countries that were poor near the end of the last century, there weren’t any large and poorer populations far away eager to take over. Dani Rodrik has found evidence that the wave of structural transformation stopped propagating down global income levels early in this century.
That dynamic alone should give us pause. But there are other seldom-aired parts to the Structural Transformation story.
“Can we go to the park today?”
Humanity has made great progress by some measures. But measuring progress in the averages on a select list of criteria obscures the wealth of evidence for the fact that improvement in the quality of life at the high end of the global spectrum came partly by directly worsening quality of life on several indicators at the lower end.
When they hear, ‘Environmental Justice’, EA members might think back to Flint, Michigan and how historically, polluting industries have been located close to marginalized groups in the US (or vice versa). Just as with the Ezra Klein article though, the US picture is a small, milder portion of the global collage.
In the same period that North-Atlantic Structural Transformation was underway, US and EU citizens have enjoyed cleaner air and water not primarily because their lifestyles got cleaner, but because they have been increasingly buying things that emit the same pollutants all over China, India and Mexico instead. The graph below shows the shift in pollutant emissions from 1970 to 2008. Blue is where the pollution used to occur. Red and yellow show where it occurs today. This graph only shows pollution from what US citizens consume.
Map CC-BY from: Daniel Moran and Keiichiro Kanemoto 2016 Environmental Research Letters ,11, 094017. The paper includes similar graphs for SO2 and NOx.
What good is a borrowed healthier environment if you cannot also borrow some free time to enjoy it? In a very real and measurable way, North-Atlantic gains in leisure and comfort come from siphoning leisure and comfort from the Global South. The graph below shows the labor embodied in goods consumed, from high-income countries, through upper-middle, China, lower-middle, India and the low-income group.
Permission to use this graphic obtained from the lead author of Dorninger, C., Hornborg, A., Abson, D. J., von Wehrden, H., Schaffartzik, A., Giljum, S., Engler, J.-O., Feller, R. L., Hubacek, K., & Wieland, H. (2021). Global patterns of ecologically unequal exchange: Implications for sustainability in the 21st century. Ecological Economics, 179, 106824.
If you or I save up to buy a hypothetical product or service with its entire supply chain contained within high-income countries, those labor flows still help. As long as there are goods in our monthly consumption basket that are only cheap because numerous people in their supply chains were paid much lower than we’d want to be paid for the same work. If they weren’t, our lives wouldn’t be as good. Prices are lower in part because more people were born locked into poverty somewhere.
To be clear, in technical terms the shifts and flows seen in those graphs are also partly because productivity in poorer parts of the world remains lower.
But equally, it’s clear the pace of North-Atlantic progress was predicated on inequality as it stood 200 years ago. Just as how that inequality came to be is an unanswerable question about the world’s most complex system, so is the counter-factual of whether any part of the world would have experienced the same pace of growth in those two centuries if they had all started at comparable wage levels and geopolitical clout.
Trade would still certainly bring gains to everyone in that hypothetical world, but would the quality of life of the top 10% be the same as today? Where would the averages stand? How would the size of the average donation to EA causes and average time spent on this forum change?
In the world that we do have now, a poor country can still hope to grow and get somewhat richer by improving productivity and institutions, but both of those need financial capital and human capital. And net flows of both kinds of capital tend towards places that already have higher productivity and better institutions, in a century-old success-breeds-success dynamic.
There are many global channels that amount to a Matthew Effect. Just one topical example: As inflation spikes, high-income economies with a broad import portfolio have a nice cushion to help relieve “demand pressure on the domestic front” – cut tariffs on imports from poorer nations. Even as domestic inflation breaks historical records and topples Prime Ministers, you can say you’re helping “eradicate poverty” across African nations.
It is important to clarify that those graphs don’t show the work of mysterious malevolent forces. It is more important to accommodate them in our mental models of how progress happens and what appraising net progress would mean.
Looking for keys near the lamp
Largely because the global economy is so complex and we cannot conceive 8 billion lives, many of which are nothing like ours, we understand the world through summary data. Most of the data we have on far-flung places is that which is easiest to get. A worldview that values output as the primary good casts a person holding pills to treat their allergies and newly christened mental health disorders and another who has never encountered either variety of ailment respectively as haves and have-nots.
Economic output is easily measured. It is harder to tally what we lose with every innovation and step-up in production. Not least because the benefits start with a small group in one place and in the period it takes for those benefits to reach a bit wider, the losses as they accumulate are dispersed away from where our eager focus remains.
Antibiotics are definitely good, but we don’t anticipate and measure the losses from not teaching judicious use to those who enjoyed them first. Where will the first superbug pandemic cause most damage, around its origin or the reaches of the world where antibiotics have been scarce[5] for decades? That’s yet another ‘global’ problem that matches a familiar template: the early focus on what economics can easily measure first bestows skewed gains to one corner of the world and when the unmeasured risk mounts, the eventual damage hurts the rest of the world most.
And by losses associated with economic activity I don’t mean intangibles like ‘community’ or the spreading epidemic of loneliness. I mean material things that some people had easier access to outside the market. Like indigenous fruit (a story below). Or the real cost of things they once didn’t need. Like purifying river water. Medicine for curing something that came from not purifying it. Or climate change mitigation.
Development as Distinction
Paradoxically, despite being hooked into a system that both thrives on and generates inequality, the nations of the world are more alike that we intuitively feel.
‘Developed’ is the past participle form of the verb. It suggests all systemic problems solved. Dust off your hands, everybody! It’s noble as an aspiration. What evidence had we to begin using it to divide the world in two? Category labels have a well-documented cognitive effect – once labels are created, all information is filtered through them, molded to fit them, in lay and expert minds alike.
The USA, an economically peerless country, has been dealing with persistent problems in food insecurity, functional illiteracy, child marriage, child labor, maternal mortality and homelessness, to list only the ones that don’t make it into regular news headlines outside the USA (and to omit some that I frankly just don’t understand well). Many poor countries deal with those same problems at larger scales – a difference of degrees, not kind.
And yet, understanding the world as split into the developed and the developing makes most voters and policy makers on both sides of the stipulated divide imagine that many of these problems are caused primarily by poverty. The failure to see lives and what they value on a spectrum makes us overlook the pockets of jurisdictions that may have crucial insight into long-term solutions.
In fact, when we see solutions in the wrong place, we may not recognize them. Worse, we may name and describe a solution someone ‘developing’ comes up with as a problem for everyone ‘developed’, just a few short years before everyone on either side sheepishly adopts it. (I’m working on another essay laying out a prominent recent case and will link here if any outlet publishes it.)
‘Liberal democracy’ is another premature phrase that distances the losses from global economic change farther away from the minds that arbitrate the authoritative verdict on the goods of the global economy. Recent years have induced more US and EU observers to allow that liberalism and democracy are not coterminous. It is my hunch that some of the crises that have made us all more aware that countries of the world are not all that dissimilar were partly due to the filtering or obfuscation of any contrary evidence by the premature classification.
For another angle on how similar parts of the world are, the last graph is one I saw early in this year. I had to get off my chair, still staring at it. It is from the World Inequality Report 2022.
Graphic under CC 4.0 from: Chancel, L., Piketty, T., Saez, E., Zucman, G. et al. World Inequality Report 2022, World Inequality Lab.
After 4000 years of elaborate markets and institutions, after 300 years of divergence between Asia and Europe, and after multiple Industrial Revolutions, the distribution of wealth on every continent is virtually identical. That picture begs to know: what have we been developing all this time? How much further potential progress really is rationed to the majority?
I’ll just leave it at that because there are two Piketty books and a report where that came from. And I’ll add that the main motivation for posting this was to juxtapose those three graphs. Kindling any level of guilt among wider membership emphatically was not.
All the Mangoes
Near the very top of the essay I added a parenthesis to what people in 1820 might need — ‘that they can only get at a market’. Lest a reader in a hurry get the impression this essay calls for rolling back the clock and returning to hunting-gathering, let me insert an anecdotal case. Also to complement those graphs and rub in the point about what empirical evidence cannot show.
In the land of Alphonso mangoes nestled beside one of the world’s eight biodiversity hotspots on the Konkan coast, is a village so small its name still doesn’t appear on Google Maps. I knew an old woman who lived alone in an adobe house there for decades. She owned four buffaloes. She spent her days in routines that might be familiar to farmers all over the world. A chunk of her day was spent bathing the buffaloes as a parent would. Then she cleaned the house and the barn where the buffaloes lived. (The floor in the human and buffalo parts of the house were always spotless and indistinguishable.) She then milked the buffaloes. Another chunk of her day was spent turning that milk first into butter and then ghee. The next part I think confounds economic theory. She went around the village giving away the ghee to anyone who wanted some. Then she returned home and did it all over again. In the right season, a neighbor offered he a fruit called ‘kokum’ which she turned into a dried flavoring popular in the region’s cuisine. Another regularly brought coconuts. Before you think ‘barter economy’, no one kept an account of how much of whatever produce a household shared with another was worth relative to another. The only thing you never saw exchange hands were mangoes. Everyone had far too many of those of their own to go looking elsewhere. The world’s most expensive fruit, littered beneath 30-meter trees, a fragrance to rival flowers, even as some of it rots through the June monsoon. A hundred years ago, the village was not devoid of children who lived in relative poverty. But the children had easy access to fresh fruits. It was a narrower variety of fruit than available in Indian supermarkets today, but they had some fruits not seen in supermarkets today and the fruits were all free.
That was grandma; I was born in that adobe house. Was she an altruist? Several industries effectively snuffed out her way of life in her last decade. Is eighty-something Kamalaathaal of Coimbatore, who ‘sells’ breakfast to poor workers and seems to reject the notion of the profit motive? Can you be sure some of the workers along the supply chain of the clothes you’re wearing now don’t only get by thanks to some such defiant extra-market holdouts?
Grandma lived half her life under British rule and half in what we now call India. She saw enormous economic change sweep through the country. We often argued about the worth of electricity. If she were alive today and asked me whether the changeover to the British economic mold brought me anything I truly value personally, I’d probably say Wikipedia and a few online publications (before paywalls) and that’s about it. (No, I’ve not needed pharmaceutical drugs or any ‘alternative’ medicine in 22 years. Yes, I have visited a dentist and an optometrist in that time.)
To round off my answer to her, I’d have to also concede that for all the knowledge freely available on the internet, I have little power to use everything I have learned. In the world that Thomas Piketty envisages, this value the internet brings me would actually amount to something for the world outside my head. I’d have the financial security to make things and share, the way my grandma did. The way many Americans on the internet could afford to for my benefit[6].
If you’re stunned by some of the policy proposals in the book at that last link, this is part of the imbalance Piketty is trying to correct. It is not that industry and markets did not or cannot bring progress. But to qualify as ‘human’ progress, it has to feel less zero-sum for groups of humans.
Farmers in every part of the world would relate to some part of my grandma’s lifestyle, her values and, to varying degrees, with her sense of loss. And the agrarian life is not the only kind displaced by expanding industry. Humans have tried various forms of economic organization. Most others are now threatened. Elevating industrialization above them all means endorsing a system that causes the lifestyles of one group to make another group rapidly give up theirs and then, at best, wait. Such an endorsement must meet that high bar of showing true ‘net’ progress, which, in a rational world, needs an exhaustive global measurement exercise for all the downsides, wherever they may be manifest.
Again, given my training, I’m inclined to believe that if we did have the data, the networked global economy or an improved version could probably clear that bar too. But before we resolve to stay on this trajectory collectively, every single citizen of the world, I’d like to see it in numbers please.
What EA Should Do
Impeach all received economics[7]. More openly and open-mindedly. Be clearer about assumptions and measurement shortcuts in economics papers. What many of the economists aligned with the movement profess isn’t wrong in any substantive way. It is simply incomplete. Even as they acknowledge the ‘if it doesn’t get counted, it doesn’t count’ nature of economics, you’ll find prominent economists giving confident answers when the evidence is even more mixed than is common in economics. For instance, when Gregory Mankiw advises that the USA should never try and emulate West European welfare states. Or, more commonly, when they inevitably bring up South Korea or Japan as models for everyone else to emulate, as Tyler Cowen is wont to.
Mankiw’s unoriginal reasoning is that taxes discourage innovation. Even before you look up the evidence, ask yourself, wouldn’t the medium and long-term effects on innovation depend on what you do with the taxes? Then, when you read actual papers, check how ‘innovation’ is defined. Does it include new product lines, which can mean a ‘new’ flavor of potato chips? That’s weighed against a world[8] where losing a job or falling sick doesn’t derail your life. Where college is a genuine choice and vocational work has equal social status.
So also with the East Asian ‘miracles’, long a Rorschach case. Economists rarely follow that talk with a mention of the extensive post-war US engagement with aid, training and comingled trade ties, besides the happy circumstance of having a giant rapidly diversifying economy right next door to smoothly buffer your transition[9].
Constant appraisal and renewal of a neutral stance matters. In the gap between earnest do-gooders encountering Bourguignon and Morrison (2002) in a graphic at Our World in Data and reading its concluding paragraphs are suspended a few billion lives in the present and countless more in the future.
I read an earlier post pointing out the emphasis on analysis over practice at EA, which in turn linked to another concerned that EA runs the risk of ending up a group “dedicated to a particular set of conclusions about the world”. I’m a bit unsure if the analysis and the conclusions are all distilled somewhere in a compact form other than as implicit in the introductory pages I referred to above.
I should say I’m an economist who used to be an engineer. I read the philosophical discussion here with great interest, but as an outsider. When I come across phrases like ‘moral patients’, ‘latent variables’, ‘welfarist approach’, they raise certain questions (or expectations) around information that is organized systematically.
On goals: Subjective wellbeing is hard to capture. Conventional proxies miss the mark[10]. Where to look beyond income or consumption? Could you (fund research to) account for other measurables such as leisure time available per day? How about the choice of pursuits available in such leisure? When we measure consumption, how broadly have we asked what needs are? What is getting commoditized that locals prefer were not?
On process: what is the consensus on such questions and how is it arrived at? Since there is much talk of expanding the moral circle, it is a given all humans are moral patients. What groups of humans are represented in the EA community directly or indirectly? Which assumptions of members with an elite North-Atlantic education are least representative of humans alive today?
Some of the people with influence in this community might spend some of their time publicly examining assumptions around what a good life is, why the current economic system deprives so many humans of what their parents defined as a good life.
In simplified summation, what is the ideal allocation of resources between:
Saving a life now (with no further measure to ensure one’s own lifestyle doesn’t directly lower the quality of that extended life)
Preventing the systemic root cause that endangered that life to begin with, to save many others in the future, however near or far
Reducing the gap between the quality of the lives in need of saving and that of those who have acquired the resources to spare to save them
Hope I have managed to make you wonder if the central question in this essay isn’t just as pivotal to the very long term as to the near term.
All lasting ‘good’ emerges from or is contingent on systems. Above are some less salient features of the one we designed, the scaffolding of institutions, rules and expectations in which all our endeavors are embedded. Our economic system generates both, some of the ‘externalities’ that charities seek to address and the surplus funds at the source of their donations. Parts of the system work reasonably well, but we are far from done perfecting the system. It is too soon for past participles. More smart people should search the tacit view that ‘if some could attain the developed status in this system, surely the system’s fine and others soon can’. Then more of them might spend professional hours thinking of how we might improve the global economic system. That would do a great good.
Also in the same spirit, because I’m wary of how text on the internet is read, I want to explicitly add: I have learned a lot from every person I cite by name, outside the specific point I’m critiquing in each case. The same goes for countries. For this post if I list a fault in Country A and commend one good idea from Country B, it has no bearing on how much I admire Country A overall or oppose other policies of Country B.
A January 2020 prize-winning essay on this forum by Hauke Hillebrandt and John G. Halstead actually called that Industrial Revolution graph the “story of human welfare”. The essay suggested then that EA pays “too little attention” to growth theory, albeit that was in relation to development interventions informed by Randomized Control Trials.
The authors caution that since some of their assumptions particularly for low-income countries are “arbitrary”, “it would be unwise” to take their estimates “at face value”.
If you’re unsure what to make of it, the holistic picture stacking up all the other graphs with self-reported data on Our World in Data’s more comprehensive page on Happiness and Life Satisfaction is even more mixed (August 2022). Something I wouldn’t be able to point to if they had not gone to the trouble.
That’s not to say a superbug could not arise in the Global South. It is a mark of the extent of within-country inequality and structural barriers to growth that South Asia might well become the source of the first superbug pandemic.
It never escapes me that most of what I learned on the internet especially in the early days came from people in the US and EU, in their personal free time, just leaving bits of information online for me to look up. As on Wikipedia articles or all the free source code from which I learned JavaScript. EtymOnline.com is just one guy from Pennsylvania who insists it isn’t altruism! Well, it certainly is very ‘effective’.
Signs apart from the thrust of this essay: curious gaps in the topic labels available to tag posts with, created by members with sufficient karma. Or the premise in the answer to the ‘I already pay taxes’ question in FAQs. Taxes are not tithe, not even partially. They are club membership fees to belong and benefit. People who pay higher taxes tend to consume or befoul public goods more. (If you buy more, you dispose more waste. You can afford gasoline, you drive more miles, wear off more roads, pollute more air, disturb more wildlife...). There’s more, but that’s a different essay. Barack Obama said a part of it at GWU in 2011.
Going only by the op-ed, I’m inclined to wager Mankiw has not spent years living in a flat society. My own economics persuasions dramatically changed over my time in first Sweden and then Netherlands, observing everyday what flatter income distributions meant for human dignity.
Which is of course not to deny indigenous factors like management innovation all credit. Again, just a complex system that doesn’t recognize political right or left.
The Wages of North-Atlantic Bias
As that title is broad, let me say in preface this is written in a Red Teaming spirit[1]. Outside this essay, my conviction in market capitalism is stronger. The biases I wish to interrogate are typically more aligned with its shorthand criteria and vernacular proxies. That proviso should also suitably prime interpretation of the empirical evidence I am about to share.
I am trying to get you to entertain two propositions:
There is no such thing as a developed country
Net global progress is impossible to measure without misleading shortcuts
In explaining why I raise those, I’ll focus on what data we have access to and how we interpret it. ‘Models are not reality’ is far from a new message. But alongside the main models, there are supplementary models around with more evasive data. If you have only a minute, skip the text and glance at the graphics below.
Development as Progress
The biases peek right from the introductory pages at EA, in axiomatic reading of economic models spun from ‘that which has already been measured because it was easy’. Typically, incomes and output.
Until a few days ago, as an example of EA in practice, the introduction page (archival link) presented the common graph of historical economic output through the Industrial Revolution[2]. It was transformative. AI is likely to be. Hence we should ensure that AI only does ‘good’. But did the Industrial Revolution only do good? Researchers in the subject regularly attempt to first operationally define the good and then proceed to estimate how much good it did. The assumption in omitting any mention of downsides seems to be that there were none. It is far rarer to see anyone even listing the categories of bads that came from the Industrial Revolution and hence hardly any attempts to estimate the global total such that we may compare. In the absence of that estimate or an acknowledgement of it, invoking the Industrial Revolution as an example of why AI needs attention, as that page still does (August 22), indicates gaps in thought.
As I lay the argument that ‘net progress’ is an extremely difficult question, I will mainly steer attention away from Europe and North-America, but even if EA only looked closer to home in space and time to begin with, the Industrial Revolution would reveal its mixed wages.
In the EA Handbook, Max Roser of Our World in Data begins to address problems with the quantification of progress but is not nearly nuanced enough. A different post at Our World in Data better sums up apparent EA articles of faith: an exhaustive one on extreme poverty.
One graph on that page, ‘Share of the World Population living in Absolute Poverty, 1820-2015’, shows the drop in the last two hundred years. Though the ‘$2 a day’ you see there is loosely adjusted for inflation, it imposes 2002’s North-Atlantic elite ideas of what people need (that they can only get at a market) on everyone who lived in 1820 anywhere. The painstaking work in the paper at the source sure has its worth. (In fact, I’ll lean on it in a bit.) But to use the estimates[3] as the definitive guide to assessing our economic systems is a failure to truly put oneself in the milieu we wish to understand. The authors even acknowledge some of the difficulty, but the commentary accompanying another graph on that same page only underscores how deep biases sit. Here’s the graph:
I see a negative correlation, but a very poor one. But the authors believe:
‘Economic prosperity’ as in output, on that graph. That is nearly the opposite of my reading. Bangladesh, Bolivia, Egypt and Tajikistan report much the same levels of satisfaction as USA, Finland, France and Japan. India, where I live, also appears in the vicinity.
The disparity in interpretation is revealing[4]. Beside that two-centuries graph, the authors say:
Yes, 19th century aristocrats and tycoons died from diseases that can today be cured with antibiotics available to billions over the counter. But still not available to everyone. How many people today do not know of or cannot access these drugs? As the otherwise comprehensive page does not mention it, I suspect we do not have a good estimate. Also, in absolute terms, was the number of people susceptible to that disease larger in 1820s or is it today? There is a rational, mathematical reason to ask: some of the world’s unequal progress has relied on unequal population growth. (More on this below.)
Imagine a teenaged British aristocrat from your favorite period TV drama whom you manage to befriend and introduce to one of the people on those data charts who’s been lifted just above extreme poverty in recent years. You offer them to trade places with a factory worker in 2018 with a guaranteed longer life. Do you think they’d jump at the chance? ‘But ah’, you counter, ‘the comparison should be to the closest job to factory work that existed in 1820’. Right, but if it is easy to imagine that an aristocrat would be giving something up that they value in 1820, how sure are we that a ‘peasant’ from 1820 might not lose something they value in the same time travel too? Should we dismiss the question because we cannot directly ask? Or should we look around the world today to inform a better guess?
More to the economic point, when average life spans were in the 40s, what proportion of people had to spend 50 hours a week or more year-round on hard work plus a commute just to make ends meet? What proportion had to spend at least 12 years fulltime in education to even have a shot at joining that daily grind? What proportion were hired as a draught animal or minded like a machine? How much of every year was life and how much work? How stressed did everyone constantly feel?
We cannot have incontrovertible evidence to settle any of this. Why then do we try to reassure ourselves that industrialization was only good for humans in the balance? Why aren’t any downsides at least footnoted in paeans to the Industrial Revolution even amid growing reports of an anti-work movement and quarter-life crises?
The somewhat vacuous ‘living standards’ comparison across centuries is something my fellow economists came up with to answer the quibbling critique that capitalism seems to both require and perpetuate inequality visibly in our lifetimes.
To help make some of this feel relatable, let me pitch from the shoulders of someone influential. In June 2021, Ezra Klein wrote a column titled ‘What the Rich Don’t Want to Admit About the Poor’. Here’s the thrust of it:
The constant threat of it. In the 21st century, just as we glimpsed some of the earliest galaxies being born. You might think this has always been in plain view and a subject of scholarship for two centuries at least. Anyone who has lived in India or another low-middle income country would also recognize the economic system described there. Even for those who didn’t before, the 2020 pandemic laid it bare.
But there is a wider aspect we don’t broadly and universally acknowledge. What is true within the US economy and within the Indian economy is also true of the global economy. In fact, if the phenomenon concerns you, the only rational way to analyze it is to examine the whole. And the global picture is startling.
Network Dependent Structures of Economic Progress
That ‘poverty since 1820’ paper is Bourguignon and Morrison (2002). (I urge you to read it alongside this section.) After deriving their estimates, the authors inform you that if all the growth in income since 1820 had been more equitable, the world in 2002 would have 1/4th the number of people living in poverty and 1/8th living in extreme poverty. By switching to the proportion of poor in the total population, Our World in Data uses the same estimates to leave you with the impression that industrialized growth has been good for reducing poverty. It is not merely a change in the denominator. Volumes of information on the gears of our global economic system lie across the two tellings. Newer papers marshal the data to visualize it.
How does economic output grow?
Structural Transformation is a story of how economies grow more prosperous. Over time, the economic output of a country shifts from being predominantly agricultural to manufacturing and then into services. Relative differences in productivity between sectors ‘naturally’ drive this transition. After all, all the countries that are wealthy today went through this transition in the last century or so.
There is very good evidence to support that theory and there is little to dispute that the account is valid for wealthy countries. There is less consensus on whether the account offers any iron-clad direction to low-income countries today. One of the complications is that the structural transformation story leaves something out.
Work in rich nations didn’t move from agriculture to manufacturing exclusively because spectacular productivity gains meant that fewer agriculture workers could entirely meet their own country’s need for agricultural produce. Instead, with rapidly expanding supply chain networks, they did not need to. They could offload some work to agricultural labor in poorer nations. For the next transition away from manufacturing, rich nations again had a population of poorer factory workers (migrating away from shrinking farms) across the ocean waiting to take over.
But of course, the same process cannot cascade down endlessly. New parts of the world can only get (slightly) richer while there is someone in the queue waiting to take up less lucrative sectors of the economy, and they in turn must have someone even poorer able to supply primary goods to them and so forth. For countries that were poor near the end of the last century, there weren’t any large and poorer populations far away eager to take over. Dani Rodrik has found evidence that the wave of structural transformation stopped propagating down global income levels early in this century.
That dynamic alone should give us pause. But there are other seldom-aired parts to the Structural Transformation story.
“Can we go to the park today?”
Humanity has made great progress by some measures. But measuring progress in the averages on a select list of criteria obscures the wealth of evidence for the fact that improvement in the quality of life at the high end of the global spectrum came partly by directly worsening quality of life on several indicators at the lower end.
When they hear, ‘Environmental Justice’, EA members might think back to Flint, Michigan and how historically, polluting industries have been located close to marginalized groups in the US (or vice versa). Just as with the Ezra Klein article though, the US picture is a small, milder portion of the global collage.
In the same period that North-Atlantic Structural Transformation was underway, US and EU citizens have enjoyed cleaner air and water not primarily because their lifestyles got cleaner, but because they have been increasingly buying things that emit the same pollutants all over China, India and Mexico instead. The graph below shows the shift in pollutant emissions from 1970 to 2008. Blue is where the pollution used to occur. Red and yellow show where it occurs today. This graph only shows pollution from what US citizens consume.
What good is a borrowed healthier environment if you cannot also borrow some free time to enjoy it? In a very real and measurable way, North-Atlantic gains in leisure and comfort come from siphoning leisure and comfort from the Global South. The graph below shows the labor embodied in goods consumed, from high-income countries, through upper-middle, China, lower-middle, India and the low-income group.
If you or I save up to buy a hypothetical product or service with its entire supply chain contained within high-income countries, those labor flows still help. As long as there are goods in our monthly consumption basket that are only cheap because numerous people in their supply chains were paid much lower than we’d want to be paid for the same work. If they weren’t, our lives wouldn’t be as good. Prices are lower in part because more people were born locked into poverty somewhere.
To be clear, in technical terms the shifts and flows seen in those graphs are also partly because productivity in poorer parts of the world remains lower.
But equally, it’s clear the pace of North-Atlantic progress was predicated on inequality as it stood 200 years ago. Just as how that inequality came to be is an unanswerable question about the world’s most complex system, so is the counter-factual of whether any part of the world would have experienced the same pace of growth in those two centuries if they had all started at comparable wage levels and geopolitical clout.
Trade would still certainly bring gains to everyone in that hypothetical world, but would the quality of life of the top 10% be the same as today? Where would the averages stand? How would the size of the average donation to EA causes and average time spent on this forum change?
In the world that we do have now, a poor country can still hope to grow and get somewhat richer by improving productivity and institutions, but both of those need financial capital and human capital. And net flows of both kinds of capital tend towards places that already have higher productivity and better institutions, in a century-old success-breeds-success dynamic.
There are many global channels that amount to a Matthew Effect. Just one topical example: As inflation spikes, high-income economies with a broad import portfolio have a nice cushion to help relieve “demand pressure on the domestic front” – cut tariffs on imports from poorer nations. Even as domestic inflation breaks historical records and topples Prime Ministers, you can say you’re helping “eradicate poverty” across African nations.
It is important to clarify that those graphs don’t show the work of mysterious malevolent forces. It is more important to accommodate them in our mental models of how progress happens and what appraising net progress would mean.
Looking for keys near the lamp
Largely because the global economy is so complex and we cannot conceive 8 billion lives, many of which are nothing like ours, we understand the world through summary data. Most of the data we have on far-flung places is that which is easiest to get. A worldview that values output as the primary good casts a person holding pills to treat their allergies and newly christened mental health disorders and another who has never encountered either variety of ailment respectively as haves and have-nots.
Economic output is easily measured. It is harder to tally what we lose with every innovation and step-up in production. Not least because the benefits start with a small group in one place and in the period it takes for those benefits to reach a bit wider, the losses as they accumulate are dispersed away from where our eager focus remains.
Antibiotics are definitely good, but we don’t anticipate and measure the losses from not teaching judicious use to those who enjoyed them first. Where will the first superbug pandemic cause most damage, around its origin or the reaches of the world where antibiotics have been scarce[5] for decades? That’s yet another ‘global’ problem that matches a familiar template: the early focus on what economics can easily measure first bestows skewed gains to one corner of the world and when the unmeasured risk mounts, the eventual damage hurts the rest of the world most.
And by losses associated with economic activity I don’t mean intangibles like ‘community’ or the spreading epidemic of loneliness. I mean material things that some people had easier access to outside the market. Like indigenous fruit (a story below). Or the real cost of things they once didn’t need. Like purifying river water. Medicine for curing something that came from not purifying it. Or climate change mitigation.
Development as Distinction
Paradoxically, despite being hooked into a system that both thrives on and generates inequality, the nations of the world are more alike that we intuitively feel.
‘Developed’ is the past participle form of the verb. It suggests all systemic problems solved. Dust off your hands, everybody! It’s noble as an aspiration. What evidence had we to begin using it to divide the world in two? Category labels have a well-documented cognitive effect – once labels are created, all information is filtered through them, molded to fit them, in lay and expert minds alike.
The USA, an economically peerless country, has been dealing with persistent problems in food insecurity, functional illiteracy, child marriage, child labor, maternal mortality and homelessness, to list only the ones that don’t make it into regular news headlines outside the USA (and to omit some that I frankly just don’t understand well). Many poor countries deal with those same problems at larger scales – a difference of degrees, not kind.
And yet, understanding the world as split into the developed and the developing makes most voters and policy makers on both sides of the stipulated divide imagine that many of these problems are caused primarily by poverty. The failure to see lives and what they value on a spectrum makes us overlook the pockets of jurisdictions that may have crucial insight into long-term solutions.
In fact, when we see solutions in the wrong place, we may not recognize them. Worse, we may name and describe a solution someone ‘developing’ comes up with as a problem for everyone ‘developed’, just a few short years before everyone on either side sheepishly adopts it. (I’m working on another essay laying out a prominent recent case and will link here if any outlet publishes it.)
‘Liberal democracy’ is another premature phrase that distances the losses from global economic change farther away from the minds that arbitrate the authoritative verdict on the goods of the global economy. Recent years have induced more US and EU observers to allow that liberalism and democracy are not coterminous. It is my hunch that some of the crises that have made us all more aware that countries of the world are not all that dissimilar were partly due to the filtering or obfuscation of any contrary evidence by the premature classification.
For another angle on how similar parts of the world are, the last graph is one I saw early in this year. I had to get off my chair, still staring at it. It is from the World Inequality Report 2022.
After 4000 years of elaborate markets and institutions, after 300 years of divergence between Asia and Europe, and after multiple Industrial Revolutions, the distribution of wealth on every continent is virtually identical. That picture begs to know: what have we been developing all this time? How much further potential progress really is rationed to the majority?
I’ll just leave it at that because there are two Piketty books and a report where that came from. And I’ll add that the main motivation for posting this was to juxtapose those three graphs. Kindling any level of guilt among wider membership emphatically was not.
All the Mangoes
Near the very top of the essay I added a parenthesis to what people in 1820 might need — ‘that they can only get at a market’. Lest a reader in a hurry get the impression this essay calls for rolling back the clock and returning to hunting-gathering, let me insert an anecdotal case. Also to complement those graphs and rub in the point about what empirical evidence cannot show.
In the land of Alphonso mangoes nestled beside one of the world’s eight biodiversity hotspots on the Konkan coast, is a village so small its name still doesn’t appear on Google Maps. I knew an old woman who lived alone in an adobe house there for decades. She owned four buffaloes. She spent her days in routines that might be familiar to farmers all over the world. A chunk of her day was spent bathing the buffaloes as a parent would. Then she cleaned the house and the barn where the buffaloes lived. (The floor in the human and buffalo parts of the house were always spotless and indistinguishable.) She then milked the buffaloes. Another chunk of her day was spent turning that milk first into butter and then ghee. The next part I think confounds economic theory. She went around the village giving away the ghee to anyone who wanted some. Then she returned home and did it all over again. In the right season, a neighbor offered he a fruit called ‘kokum’ which she turned into a dried flavoring popular in the region’s cuisine. Another regularly brought coconuts. Before you think ‘barter economy’, no one kept an account of how much of whatever produce a household shared with another was worth relative to another. The only thing you never saw exchange hands were mangoes. Everyone had far too many of those of their own to go looking elsewhere. The world’s most expensive fruit, littered beneath 30-meter trees, a fragrance to rival flowers, even as some of it rots through the June monsoon. A hundred years ago, the village was not devoid of children who lived in relative poverty. But the children had easy access to fresh fruits. It was a narrower variety of fruit than available in Indian supermarkets today, but they had some fruits not seen in supermarkets today and the fruits were all free.
That was grandma; I was born in that adobe house. Was she an altruist? Several industries effectively snuffed out her way of life in her last decade. Is eighty-something Kamalaathaal of Coimbatore, who ‘sells’ breakfast to poor workers and seems to reject the notion of the profit motive? Can you be sure some of the workers along the supply chain of the clothes you’re wearing now don’t only get by thanks to some such defiant extra-market holdouts?
Grandma lived half her life under British rule and half in what we now call India. She saw enormous economic change sweep through the country. We often argued about the worth of electricity. If she were alive today and asked me whether the changeover to the British economic mold brought me anything I truly value personally, I’d probably say Wikipedia and a few online publications (before paywalls) and that’s about it. (No, I’ve not needed pharmaceutical drugs or any ‘alternative’ medicine in 22 years. Yes, I have visited a dentist and an optometrist in that time.)
To round off my answer to her, I’d have to also concede that for all the knowledge freely available on the internet, I have little power to use everything I have learned. In the world that Thomas Piketty envisages, this value the internet brings me would actually amount to something for the world outside my head. I’d have the financial security to make things and share, the way my grandma did. The way many Americans on the internet could afford to for my benefit[6].
If you’re stunned by some of the policy proposals in the book at that last link, this is part of the imbalance Piketty is trying to correct. It is not that industry and markets did not or cannot bring progress. But to qualify as ‘human’ progress, it has to feel less zero-sum for groups of humans.
Farmers in every part of the world would relate to some part of my grandma’s lifestyle, her values and, to varying degrees, with her sense of loss. And the agrarian life is not the only kind displaced by expanding industry. Humans have tried various forms of economic organization. Most others are now threatened. Elevating industrialization above them all means endorsing a system that causes the lifestyles of one group to make another group rapidly give up theirs and then, at best, wait. Such an endorsement must meet that high bar of showing true ‘net’ progress, which, in a rational world, needs an exhaustive global measurement exercise for all the downsides, wherever they may be manifest.
Again, given my training, I’m inclined to believe that if we did have the data, the networked global economy or an improved version could probably clear that bar too. But before we resolve to stay on this trajectory collectively, every single citizen of the world, I’d like to see it in numbers please.
What EA Should Do
Impeach all received economics[7]. More openly and open-mindedly. Be clearer about assumptions and measurement shortcuts in economics papers. What many of the economists aligned with the movement profess isn’t wrong in any substantive way. It is simply incomplete. Even as they acknowledge the ‘if it doesn’t get counted, it doesn’t count’ nature of economics, you’ll find prominent economists giving confident answers when the evidence is even more mixed than is common in economics. For instance, when Gregory Mankiw advises that the USA should never try and emulate West European welfare states. Or, more commonly, when they inevitably bring up South Korea or Japan as models for everyone else to emulate, as Tyler Cowen is wont to.
Mankiw’s unoriginal reasoning is that taxes discourage innovation. Even before you look up the evidence, ask yourself, wouldn’t the medium and long-term effects on innovation depend on what you do with the taxes? Then, when you read actual papers, check how ‘innovation’ is defined. Does it include new product lines, which can mean a ‘new’ flavor of potato chips? That’s weighed against a world[8] where losing a job or falling sick doesn’t derail your life. Where college is a genuine choice and vocational work has equal social status.
So also with the East Asian ‘miracles’, long a Rorschach case. Economists rarely follow that talk with a mention of the extensive post-war US engagement with aid, training and comingled trade ties, besides the happy circumstance of having a giant rapidly diversifying economy right next door to smoothly buffer your transition[9].
Constant appraisal and renewal of a neutral stance matters. In the gap between earnest do-gooders encountering Bourguignon and Morrison (2002) in a graphic at Our World in Data and reading its concluding paragraphs are suspended a few billion lives in the present and countless more in the future.
I read an earlier post pointing out the emphasis on analysis over practice at EA, which in turn linked to another concerned that EA runs the risk of ending up a group “dedicated to a particular set of conclusions about the world”. I’m a bit unsure if the analysis and the conclusions are all distilled somewhere in a compact form other than as implicit in the introductory pages I referred to above.
I should say I’m an economist who used to be an engineer. I read the philosophical discussion here with great interest, but as an outsider. When I come across phrases like ‘moral patients’, ‘latent variables’, ‘welfarist approach’, they raise certain questions (or expectations) around information that is organized systematically.
On goals: Subjective wellbeing is hard to capture. Conventional proxies miss the mark[10]. Where to look beyond income or consumption? Could you (fund research to) account for other measurables such as leisure time available per day? How about the choice of pursuits available in such leisure? When we measure consumption, how broadly have we asked what needs are? What is getting commoditized that locals prefer were not?
On process: what is the consensus on such questions and how is it arrived at? Since there is much talk of expanding the moral circle, it is a given all humans are moral patients. What groups of humans are represented in the EA community directly or indirectly? Which assumptions of members with an elite North-Atlantic education are least representative of humans alive today?
Some of the people with influence in this community might spend some of their time publicly examining assumptions around what a good life is, why the current economic system deprives so many humans of what their parents defined as a good life.
In simplified summation, what is the ideal allocation of resources between:
Saving a life now (with no further measure to ensure one’s own lifestyle doesn’t directly lower the quality of that extended life)
Preventing the systemic root cause that endangered that life to begin with, to save many others in the future, however near or far
Reducing the gap between the quality of the lives in need of saving and that of those who have acquired the resources to spare to save them
Hope I have managed to make you wonder if the central question in this essay isn’t just as pivotal to the very long term as to the near term.
All lasting ‘good’ emerges from or is contingent on systems. Above are some less salient features of the one we designed, the scaffolding of institutions, rules and expectations in which all our endeavors are embedded. Our economic system generates both, some of the ‘externalities’ that charities seek to address and the surplus funds at the source of their donations. Parts of the system work reasonably well, but we are far from done perfecting the system. It is too soon for past participles. More smart people should search the tacit view that ‘if some could attain the developed status in this system, surely the system’s fine and others soon can’. Then more of them might spend professional hours thinking of how we might improve the global economic system. That would do a great good.
Also in the same spirit, because I’m wary of how text on the internet is read, I want to explicitly add: I have learned a lot from every person I cite by name, outside the specific point I’m critiquing in each case. The same goes for countries. For this post if I list a fault in Country A and commend one good idea from Country B, it has no bearing on how much I admire Country A overall or oppose other policies of Country B.
A January 2020 prize-winning essay on this forum by Hauke Hillebrandt and John G. Halstead actually called that Industrial Revolution graph the “story of human welfare”. The essay suggested then that EA pays “too little attention” to growth theory, albeit that was in relation to development interventions informed by Randomized Control Trials.
The authors caution that since some of their assumptions particularly for low-income countries are “arbitrary”, “it would be unwise” to take their estimates “at face value”.
If you’re unsure what to make of it, the holistic picture stacking up all the other graphs with self-reported data on Our World in Data’s more comprehensive page on Happiness and Life Satisfaction is even more mixed (August 2022). Something I wouldn’t be able to point to if they had not gone to the trouble.
That’s not to say a superbug could not arise in the Global South. It is a mark of the extent of within-country inequality and structural barriers to growth that South Asia might well become the source of the first superbug pandemic.
It never escapes me that most of what I learned on the internet especially in the early days came from people in the US and EU, in their personal free time, just leaving bits of information online for me to look up. As on Wikipedia articles or all the free source code from which I learned JavaScript. EtymOnline.com is just one guy from Pennsylvania who insists it isn’t altruism! Well, it certainly is very ‘effective’.
Signs apart from the thrust of this essay: curious gaps in the topic labels available to tag posts with, created by members with sufficient karma. Or the premise in the answer to the ‘I already pay taxes’ question in FAQs. Taxes are not tithe, not even partially. They are club membership fees to belong and benefit. People who pay higher taxes tend to consume or befoul public goods more. (If you buy more, you dispose more waste. You can afford gasoline, you drive more miles, wear off more roads, pollute more air, disturb more wildlife...). There’s more, but that’s a different essay. Barack Obama said a part of it at GWU in 2011.
Going only by the op-ed, I’m inclined to wager Mankiw has not spent years living in a flat society. My own economics persuasions dramatically changed over my time in first Sweden and then Netherlands, observing everyday what flatter income distributions meant for human dignity.
Which is of course not to deny indigenous factors like management innovation all credit. Again, just a complex system that doesn’t recognize political right or left.
If you only click one external link, make it that Economist article.