Daniel sent us this one — he wants us to dig into the Human Development Index, the H. , as a way of measuring economic development and what we might call modernity. Where it came from, what it actually measures, its strengths, its well-documented weaknesses, and the family of related indices it spawned. And the deeper question: can any single composite number meaningfully rank development across radically different societies, or is the H. really just a conversation-starter dressed up as a leaderboard? There's a lot to unpack here.
There really is. And I should mention — DeepSeek V four Pro is writing our script today, so let's see how it handles development economics. But to the topic: the H. launched in nineteen ninety, and the intellectual backstory is genuinely fascinating. You had Mahbub ul Haq, a Pakistani economist, working at the U. , and he brought in Amartya Sen, who'd later win the Nobel Prize, to help shape the thinking. Haq's explicit goal was to shift development economics away from national income accounting and toward people-centered policies. He wanted a number that could compete with G. per capita in public discourse.
Sen was apparently reluctant about the whole thing.
Sen had developed this entire capabilities approach — the idea that development should be measured by what people are actually able to be and do, not just what they can buy. Being well-fed, being sheltered, being healthy, being educated, participating in community life. It's a rich, nuanced framework. And Haq came to him and said, essentially, we need to boil this down to one number. Sen thought it was reductive. But Haq's argument was brutally pragmatic: if you want to change the conversation, you need a simple measure. Politicians and journalists aren't going to read a philosophical treatise.
There's something almost tragic about that trade-off. You take this elegant capabilities framework, and you compress it into a single scalar between zero and one. Lossy compression, as Daniel would probably put it.
That's exactly what it is. But Haq was right about the political communication problem. is seductive because it's one number. If you want to dethrone it, you need another one number. So let me walk through what the H. actually measures, because the methodology matters. There are three dimensions. Health, measured by life expectancy at birth. Education, measured by mean years of schooling for adults and expected years of schooling for children, combined into a single education index. And income, measured by G. per capita at purchasing power parity.
Each of these gets normalized to a zero-to-one scale, right? With fixed goalposts.
Life expectancy runs from a minimum of twenty years to a maximum of eighty-five. Mean years of schooling from zero to fifteen, expected years from zero to eighteen. per capita from a hundred dollars to seventy-five thousand dollars, with a log transform applied to the income component. And then the three sub-indices are combined using a geometric mean rather than an arithmetic mean. That switch happened in twenty ten, and it's mechanically important.
Explain why the geometric mean matters.
Because it prevents perfect substitutability. Under the old arithmetic mean, you could have abysmal health outcomes but enormous income, and your H. would still look respectable. The geometric mean punishes imbalance. If any one dimension is very low, the whole score gets dragged down substantially. You can't buy your way out of a health crisis or an education deficit. Mathematically, it's the cube root of the product of the three sub-indices.
It encodes a normative assumption that balanced development across all three dimensions is what we're aiming for. Which seems reasonable on its face, but it's still an assumption.
And that's actually one of the through-lines of H. criticism — every methodological choice encodes a normative judgment. The log transform on income is another example. By logging G. , you're saying that an extra thousand dollars matters a lot more to a poor country than to a rich one. Diminishing returns to income for human development. That's not a factual claim about the world, it's a value judgment about what counts as development.
Which I've always found interesting, because it means the H. is essentially saying that beyond a certain point, more money doesn't make you more developed. And yet we never stop hearing about G. growth targets for rich countries.
The ceiling effect is a real issue. The income component maxes out at seventy-five thousand dollars P. , and life expectancy plateaus too. So wealthy nations all cluster near the top with very little differentiation. is least informative precisely for the countries that care most about their rankings. Switzerland, Norway, Australia — they're all bunched up within a few thousandths of a point. You get reshuffling year to year that's basically statistical noise.
Which means the index is most useful for low and middle-income countries, where variation is actually meaningful. But those are also the countries where data quality is worst.
Life expectancy figures depend on vital registration systems that may be incomplete. School enrollment data can be inflated for political reasons. calculations for informal economies are inherently fuzzy. So you've got the most measurement error where the index is supposed to be most informative.
Let's talk about what the H. gets right, though. Because I don't want to just pile on.
Its genuine strengths are substantial. First, it really did change the conversation. Before nineteen ninety, development was overwhelmingly discussed in terms of G. forced ministries of finance and international institutions to pay attention to health and education outcomes in a way they hadn't before. now publishes these annual Human Development Reports that get genuine media coverage and political attention. Countries have shuffled budget priorities because they didn't want to look bad on the H.
There are interesting things you see when you compare H. rankings to G. Countries that punch well above their economic weight.
Sri Lanka is the classic example. per capita around twenty-five hundred to four thousand dollars, but an H. of about zero point seven eight. That puts it well above Equatorial Guinea, which has oil-driven wealth and an H. of only zero point six seven. Sri Lanka invested heavily in universal healthcare and education for decades, and the H. Former Soviet republics are another interesting case — Tajikistan retained high literacy and decent health outcomes even after the economic collapse of the nineties. shows that human capital doesn't evaporate the way financial capital does.
On the flip side, you get countries with enormous resource wealth that look terrible on the H. Equatorial Guinea, as you mentioned. Some of the Gulf states actually underperform relative to their income levels once you strip out the G. component and look at what their health and education indices are doing.
The United States is a surprisingly interesting case here. It ranks seventeenth in the world on H. despite having one of the highest G. per capita figures. Countries like Australia, Canada, Ireland, and several European nations all outrank it. gets dragged down by relatively lower life expectancy — which has actually been declining or stagnating in recent years — and by education metrics that don't fully reflect the quality issues in parts of the system. For the richest large country on earth, seventeenth is not where you'd expect to be.
Which tells you something real. But it also highlights one of the weaknesses — the H. has nothing about inequality in the headline number. has enormous internal variation. I saw a study that found if you took the poorest income quintile in the United States and ranked them separately, they'd land around fifty-fifth in the global H. But the overall U. score masks that entirely.
That's from a twenty ten study, and it's a devastating finding. The standard H. is an average. It tells you nothing about distribution. A country can have a respectable H. while leaving a substantial portion of its population in conditions that look like a much poorer country. That's why they introduced the I. , the Inequality-adjusted H. , in twenty ten. It applies Atkinson inequality measures to discount each dimension based on how evenly distributed the outcomes are.
The gap between the H. and the I. can be substantial.
calls it the loss in potential human development due to inequality. For countries with very unequal outcomes, the I. can be fifteen or twenty percent lower than the headline H. But here's the thing — the I. gets far less media attention than the main index. The headline number is what drives political conversation.
Of course it does. The whole point of the H. was to have a simple number that competes with G. If you start adding asterisks and alternative versions, you lose the communicative punch. So the thing that makes it politically useful is also what makes it analytically limited.
That's the central tension. And it extends to other dimensions the H. simply doesn't capture at all. There's no environmental dimension in the standard index. A country can have a stellar H. while being an ecological disaster. No measure of carbon emissions, resource depletion, biodiversity loss. Ambuj Sagar wrote a widely cited critique of this back in nineteen ninety-eight, and the criticism still holds.
No political freedom or governance measures either.
is completely silent on democracy, civil liberties, press freedom, rule of law, corruption. A country can score well on health, education, and income while being an authoritarian state. This is where the legitimacy laundering concern comes in — if an authoritarian regime delivers material wellbeing, the H. gives it a gold star and never asks about political rights.
Which brings us to the family of derivative indices. Because the U. clearly recognized these gaps and kept bolting things on.
So you've got the I. for inequality, which we mentioned. The Gender Inequality Index, or G. , which measures disparities in reproductive health, empowerment, and labor market participation. The Multidimensional Poverty Index, the M. , which looks at household-level deprivations across health, education, and living standards — about one point seven billion people were estimated to be in M. poverty across a hundred and four countries. And then the Planetary-adjusted H. , which adjusts the score downward based on carbon emissions and material footprint.
is the one that really reshuffles the rankings, isn't it? Wealthy high-consumption countries drop substantially.
When you penalize for planetary pressures, the countries that look best on the standard H. take a significant hit. Norway, which often tops the H. rankings, doesn't look nearly as impressive. essentially says: your human development achievements are less impressive if they're built on unsustainable resource consumption that undermines future generations' prospects.
Which raises the question of whether the headline H. should incorporate environmental sustainability directly, or whether that's mixing two fundamentally different normative goals. Human flourishing today versus planetary boundaries for tomorrow.
There's no technical answer to that question. It's a values choice. has chosen to keep the standard H. focused on the original three dimensions and offer the P. as a separate metric. But that means the index that actually drives headlines and policy conversations is the one without environmental constraints.
Let me push on something that's been bothering me throughout this discussion. The education component.
measures years of schooling. Not what anyone actually learned. Not whether the education was any good. Just seat time. And it implicitly assumes that formal Western-style schooling is the only path to being an educated, knowledgeable person. What about societies with strong apprenticeship traditions? Indigenous knowledge systems? Informal education structures? doesn't see any of that. It only sees years spent in a classroom.
This is one of the deepest critiques, and it's under-discussed. There's a World Bank working paper by Martin Ravallion from twenty twelve that found the H. 's valuations of gains from extra schooling seem unreasonably high — many times greater than the actual economic returns to schooling. So the index is overweighting education relative to what schooling actually delivers in terms of capabilities or productivity.
It encodes a very specific developmental path. The assumption is that all societies should converge toward long life expectancy, many years of formal schooling, and high income. That's the modernity package. But who decided that? The whole framework was designed by development economists trained in Western institutions, working at a U. agency headquartered in New York.
I think this is where Sen's original reluctance becomes really poignant. His capabilities approach was actually quite open-ended about what capabilities matter and how different societies might prioritize them differently. operationalizes capabilities in a very specific, arguably culturally specific way. Haq's pragmatism won out, but something was lost.
Yet, I have to admit, when I look at the alternatives, I'm not sure there's a better simple metric available. per capita is clearly worse. It tells you nothing about how long people live or whether they can read. , for all its flaws, at least gestures toward human wellbeing.
The rank correlation between H. per capita is about zero point nine three to zero point nine six. So in practice, the H. mostly tracks income anyway. Gapminder did an analysis showing that if you plot H. per capita, most countries hug the curve tightly. There are only about twelve notable outliers.
Which means for most countries, the H. isn't telling you much that G. per capita doesn't already tell you. The outliers are where it gets interesting — Sri Lanka, Equatorial Guinea, former Soviet states, the U. But for the bulk of the distribution, it's basically a transformation of income data.
A twenty twenty-one paper even demonstrated that the H. is mathematically equivalent to a paternalistic social welfare function. The analyst's preferences drive the interpretation, not some objective measurement of development.
Where does that leave us? is better than G. alone, but it's still a lossy compression of something irreducibly multidimensional. It captures health and education, which is valuable, but it ignores inequality, environment, governance, and culture. It's most informative for the countries with the worst data, and least informative for the countries that obsess over rankings. And for most countries, it doesn't diverge much from income metrics anyway.
I think the fair verdict is that the H. 's real value isn't as a leaderboard. It's as a conversation-starter. When Sri Lanka outranks Equatorial Guinea on the H. despite having a fraction of the G. , that forces a question: what is Sri Lanka doing right, and what is Equatorial Guinea doing wrong? When the United States ranks seventeenth, it prompts a discussion about why the richest country can't translate wealth into health and education outcomes more efficiently. Those are valuable conversations.
The rankings create a kind of productive discomfort. Countries don't like looking bad, and the H. makes some countries look bad in ways that G. per capita doesn't. That's probably its greatest legacy — not the precision of the scores, but the policy debates the scores generate.
seems to understand this. The Human Development Reports have become progressively richer over time, adding thematic chapters, country notes, and the derivative indices we discussed. is the hook, but the actual substance is in the surrounding analysis. I suspect the U. will keep patching the framework rather than building something entirely new, partly because of institutional inertia and partly because any replacement would face the same fundamental trade-offs.
The fundamental problem is that development is irreducibly multidimensional. Any single number is going to be a compromise. The question is whether the compromise is worth making. Haq thought yes, and he was probably right in nineteen ninety, when G. utterly dominated the discourse and there was no counterweight at all. Whether the same calculus holds today, when we have dozens of composite indices and much richer data, is less clear.
I'd argue it still holds, but barely. remains the most recognized alternative to G. , and that brand recognition matters. If you're a finance minister in a developing country and your H. is stagnating while G. is growing, that's a conversation you have to have with your cabinet. That's not nothing.
But I come back to the schooling critique. If the H. is driving policy attention toward years of schooling rather than learning outcomes, it could actually be counterproductive. Countries might focus on building schools and boosting enrollment — which are visible and score well on the index — while neglecting whether anyone is actually learning anything inside those schools.
That's a real risk. And it connects to a broader point about metric-driven governance. Once you attach policy incentives to a metric, people optimize for the metric, not necessarily for the underlying reality the metric is supposed to capture. is vulnerable to Goodhart's Law like any other indicator.
If you were advising a policymaker, what would you tell them to actually use?
I'd say look at the H. as a starting point, not an endpoint. Check the I. to understand inequality. Look at the M. if poverty is your concern. Check the G. for gender disparities. If environmental sustainability matters to you, consult the P. And then go beyond the U. family entirely — look at governance indicators, press freedom rankings, environmental performance indices. No single number is adequate. is one lens among many.
Which is basically admitting that the whole project of replacing G. with a single alternative number has failed. What we've ended up with is a dashboard, not a replacement.
Dashboards are harder to communicate than single numbers. That's the trade-off Haq identified, and it hasn't gone away. But I think the dashboard approach is intellectually honest in a way that the single-number approach isn't. Development is complicated. Pretending otherwise doesn't make it simpler.
And now: Hilbert's daily fun fact.
The shortest war in recorded history was between Britain and Zanzibar on August twenty-seventh, eighteen ninety-six. It lasted between thirty-eight and forty-five minutes.
Where does this leave listeners who want to use the H. First, understand what it actually measures — and what it doesn't. When you see an H. ranking, ask about the distribution underneath. Check the I. Look at the sub-components separately rather than just the composite score. A country with a high H. driven entirely by income looks very different from one driven by health and education.
Second, pay attention to the outliers. The countries where H. diverge are where the interesting stories are. Sri Lanka, Cuba, Vietnam on the positive side — they've achieved decent human development without high income. Equatorial Guinea, Angola, some Gulf states on the negative side — they've got wealth that isn't translating into broad-based wellbeing. Those cases tell you more about development than the countries that hug the regression line.
Third, if you're comparing rich countries, recognize that the H. has limited discriminatory power. The difference between rank five and rank fifteen may not be meaningful. Small changes in methodology or data can reshuffle the top of the table. Don't overinterpret small ranking shifts.
Fourth, if you care about environmental sustainability, consult the P. or other environmental metrics alongside the standard H. The standard index rewards consumption patterns that may be ecologically unsustainable, and that's an increasingly important dimension of development that the headline number simply ignores.
was a genuine intellectual achievement in nineteen ninety. It shifted the discourse from national income to human capabilities, and that was a real contribution. Thirty-plus years later, it remains useful as a rough starting point and a conversation-generator. But treating it as a definitive ranking of national development is a category error. It's a heuristic, not a truth machine.
That's probably how Haq and Sen would want us to use it. Sen never intended his capabilities framework to be reduced to a single number. was a tactical compromise — a way to get capabilities thinking into policy conversations that were otherwise dominated entirely by G. The question is whether the compromise has outlived its usefulness, or whether we still need that simple counterweight to income metrics.
Given that G. growth remains the dominant policy objective in most countries, I suspect we still need it. But we should use it with clear eyes about what it is and isn't telling us.
Thanks to Hilbert Flumingtop for producing. This has been My Weird Prompts. You can find every episode at myweirdprompts dot com.
If you've got thoughts on the H. or a prompt you'd like us to tackle, send it our way. We'll be back soon.
