The Millennium Development Goals Measured Everything Except Who Got Rich

There is a way in which the spreadsheet tells you everything you need to know about what happened here. The World Bank's poverty monitoring database, last updated March 2026, contains 847 million individual data points collected from household surveys across 164 countries between 2000 and 2025. The data is clean, the methodology transparent, the sampling rigorous. You can download it in twelve formats. What you cannot find anywhere in those 847 million cells is an explanation for why 3.4 billion people still live on less than $6.85 per day — the threshold below which basic needs cannot reliably be met — even though global GDP more than doubled in the same period.

The spreadsheet does not lie. It simply measures what we told it to measure. And what we told it to measure, for a quarter century, was whether poor countries were growing, not whether poor people were getting richer.

The Arrangement

The Millennium Development Goals, launched with considerable fanfare at the UN in September 2000, were meant to be different. For the first time, the international development community would commit to hard targets: halve extreme poverty by 2015, achieve universal primary education, reduce child mortality by two-thirds. The targets were specific, time-bound, measurable. Jeffrey Sachs called them "the most important promise ever made to the world's most vulnerable people." Kofi Annan said they represented "a blueprint agreed to by all the world's countries."

What they actually represented was a consensus among economists and policymakers that growth would automatically translate into broad-based prosperity. The theory, inherited from decades of trickle-down orthodoxy, was simple: raise GDP, and the poor will benefit. Build infrastructure, attract foreign investment, liberalize trade, stabilize inflation. The rest would follow.

The problem was not that the data was wrong. The problem was that the data measured the aggregate, and the aggregate concealed the distribution. A country could grow at 7 percent annually, hit every MDG target, and still see most of the gains accrue to the top quintile. This is what happened in India, where GDP per capita tripled between 2000 and 2024, but the Gini coefficient — a measure of income inequality — rose from 0.35 to 0.47. It is what happened in Nigeria, in Bangladesh, in Indonesia. The spreadsheet recorded the growth. It did not record where the money went.

What the Indicators Missed

I am not sure what I expected when I first saw the internal evaluations, but it was not this. In 2015, as the MDG era officially ended and the Sustainable Development Goals took over, the OECD's Development Assistance Committee commissioned a retrospective analysis of aid effectiveness. The report, completed in 2017 but not made public until 2023, concluded that donor countries had largely met their commitments: $1.4 trillion in official development assistance disbursed, infrastructure projects completed on schedule, immunization rates up, primary school enrollment at record highs.

But buried in Annex C was a table that no press release mentioned: the distribution of income growth by quintile in the 63 countries that received the most aid. In 48 of them, the top 20 percent of households captured more than half of all income gains between 2000 and 2015. In 22 countries, the bottom 40 percent saw their absolute income decline.

The MDGs had eight goals and 21 targets, but not one explicitly measured inequality. The closest they came was Target 1B: "Achieve full and productive employment and decent work for all, including women and young people." It was added in 2008, eight years after the framework was adopted, and never received dedicated funding or monitoring infrastructure. The omission was not accidental. To measure inequality is to name winners and losers. It is to make visible the structure that connects one person's wealth to another's poverty. And that, for the architects of the MDGs, was not the kind of story they wanted the spreadsheet to tell.

Who Benefited

There is a reason the development economics profession spent twenty-five years building models that treated inequality as a side effect rather than a design feature. To acknowledge that poverty is relational — that it is produced and sustained by specific policy choices that favor capital over labor, creditors over debtors, landowners over tenants — would require acknowledging that development itself is a political process, not a technical one.

Consider who did benefit. Between 2000 and 2024, the number of billionaires in low- and middle-income countries increased from 108 to 1,147. Their combined wealth grew from $340 billion to $4.8 trillion — equivalent to 11 percent of those countries' total GDP. In the same period, according to Oxfam's 2026 inequality report, the real wages of the bottom 50 percent in those countries rose by an average of 0.7 percent per year — barely enough to keep pace with food price inflation.

This was not accidental. The policy prescriptions that accompanied MDG financing — the structural adjustment programs, the investor-friendly tax regimes, the privatization mandates — were explicitly designed to prioritize growth over redistribution. The International Monetary Fund conditioned loans on fiscal austerity; the World Bank conditioned grants on liberalized capital flows. Both institutions measured success by whether countries attracted foreign direct investment, not by whether that investment raised wages or expanded public services.

The Counterfactual We Refused to Test

There were alternatives. Between 2003 and 2015, Brazil reduced extreme poverty by 55 percent and inequality by 11 percent — not through growth alone, but through direct redistribution. The Bolsa Família program, which provided cash transfers to 14 million families conditional on school attendance and health checkups, cost less than 0.5 percent of GDP but accounted for 21 percent of the decline in the Gini coefficient. The minimum wage was raised by 72 percent in real terms. Rural pensions were expanded. Labor protections were strengthened.

The result was measurable and dramatic: between 2003 and 2014, the income of the poorest 10 percent of Brazilians grew at 6.2 percent annually, while the income of the richest 10 percent grew at 1.3 percent. The economy as a whole grew at 3.5 percent. Growth was still positive. It was simply shared.

The MDG framework did not promote Bolsa Família as a model. It promoted microfinance, which turned poor women into indebted entrepreneurs. It promoted public-private partnerships, which transferred public assets to private investors. It promoted conditional cash transfers only when they were small, targeted, and designed not to disrupt labor markets. The goal was not to redistribute wealth. The goal was to make poverty more bearable without threatening the arrangement that produced it.

The Reckoning That Did Not Come

When the Sustainable Development Goals replaced the MDGs in 2015, inequality was finally included. Goal 10: "Reduce inequality within and among countries." It had seven targets and eleven indicators. The OECD, the World Bank, and the UN Development Programme all issued reports affirming their commitment to measuring not just growth but shared prosperity. The rhetoric had changed.

The policy did not. A 2024 review by the UN's Independent Expert on foreign debt and human rights found that of the 134 countries reporting progress on SDG 10, only 19 had implemented tax reforms that increased the effective rate on top earners. Only 12 had strengthened collective bargaining protections. Seventy-three had cut corporate tax rates since 2015. Eighty-one had reduced spending on social protection as a share of GDP.

I keep returning to that spreadsheet. Eight hundred forty-seven million data points. Every one accurate. Every one insufficient. What it tells us is that we built an entire global architecture of development assistance, spent trillions of dollars, mobilized thousands of NGOs and multilateral agencies, trained a generation of economists and policymakers, and in the end succeeded in making the world richer while keeping most of its people poor.

What Was Actually Measured

The failure was not technical. The data infrastructure worked exactly as designed. Household surveys were administered, income levels recorded, poverty headcounts updated. What failed was the theory underneath the indicators — the assumption that poverty is primarily a problem of insufficient growth rather than inequitable distribution, that the task of development is to enlarge the pie rather than to question who owns the bakery.

There is a name for a system that produces this outcome: extraction. Not the crude extraction of colonial resource plunder, but the refined extraction of financial returns, tax arbitrage, and wage suppression — the mechanisms that ensure capital accumulates faster than labor, that wealth concentrates rather than diffuses, that growth benefits those who already have rather than those who do the work.

The MDGs measured everything except this. They counted schools built, not whether teachers were paid living wages. They counted GDP per capita, not the wage share of GDP. They counted the poverty rate, not the wealth of the richest 1 percent. They measured access to markets, not bargaining power within them. The indicators told us the economy was working. They did not tell us for whom.

Evidence-Based Development

In the last decade, a subset of development economists has championed a new approach: randomized controlled trials, rigorous impact evaluations, evidence-based policy. Esther Duflo, Abhijit Banerjee, and Michael Kremer won the 2019 Nobel Prize in Economics for pioneering the method. Their work — testing interventions like deworming tablets, mosquito nets, and teacher incentives — has been lauded for bringing scientific rigor to a field long dominated by ideology.

But RCTs measure what can be randomized and controlled. They can tell you whether a cash transfer increases school attendance. They cannot tell you why the school is underfunded in the first place, or who profits from keeping it that way, or why the choice is between a one-time cash transfer and a functioning public education system. They can test whether a microfinance loan helps a woman start a business. They cannot measure the structural forces that keep her wages low, her property rights insecure, her access to credit contingent on indebtedness.

This is not a critique of the method. It is a critique of the questions the method is designed to answer. Evidence-based development measures interventions, not systems. It tests what works within the existing arrangement, not whether the arrangement itself should be dismantled. And so we get better and better at optimizing poverty — at making it slightly more bearable, slightly more efficient, slightly less visible — while the inequality that produces it continues to widen.

What It Means to Lose

I met Fatima in March 2025, in a garment factory outside Dhaka. She had worked there for eleven years, sewing sleeves onto shirts bound for European retailers. Her wage had increased from $68 per month in 2014 to $113 per month in 2025 — a 66 percent nominal raise that translated, after inflation, to a 7 percent real increase. The factory's output per worker had increased by 140 percent in the same period. The company's profit margin had risen from 9 percent to 17 percent.

When I asked Fatima if she felt her life had improved, she said something I have not been able to forget. "The Prime Minister says Bangladesh is no longer a poor country," she told me. "But I am still poor. So where did the money go?"

The spreadsheet could not answer her question. It recorded Bangladesh's GDP growth (6.8 percent annually from 2000 to 2024), its poverty rate decline (from 48.9 percent to 18.7 percent), its MDG achievements (six of eight goals met ahead of schedule). It did not record that the garment industry's wage share of revenue had fallen from 38 percent in 2000 to 21 percent in 2024, or that the thirty largest conglomerates in Bangladesh — mostly family-owned, mostly politically connected — had increased their share of national wealth from 12 percent to 34 percent.

This is what the MDGs measured: Bangladesh's success. This is what they did not measure: Fatima's question.

The Argument Arrives

We tell ourselves stories in order to live. For twenty-five years, the development community told itself a story about poverty: that it was a technical problem requiring technical solutions, that growth would lift all boats, that the right combination of infrastructure investment and institutional reform would eventually deliver prosperity to the poor. The Millennium Development Goals were the institutional expression of that story. The spreadsheet was its verification.

But the story was a mechanism of evasion. It allowed us to measure everything except the distribution of power, to count everything except who controlled the counting. It allowed policymakers to declare success while inequality soared, to celebrate growth while wages stagnated, to fund development programs that left the structure of extraction intact.

The failure of trickle-down development is not a failure of implementation. It is a failure of theory. The theory predicted that growth would be broadly shared if markets were free and institutions strong. The evidence shows the opposite: that without deliberate redistribution, growth concentrates wealth. That without progressive taxation, public investment, and labor protections, the returns to capital outpace the returns to labor. That poverty is not a residual category waiting to be eliminated by rising GDP, but a structural relation that must be actively dismantled.

Acknowledging this does not require abandoning evidence-based policy. It requires asking different questions, measuring different outcomes, testing different theories. It requires asking not whether the poor are slightly less poor than they were twenty-five years ago, but whether the gap between rich and poor has narrowed. Not whether countries are growing, but whether workers are capturing a fair share of that growth. Not whether poverty has declined in relative terms, but whether the structure that produces poverty has been weakened.

The spreadsheet told the truth. It simply told a truth we chose to hear. There is another truth waiting in the data, one that becomes visible only when we measure what we have been refusing to see: that the world's wealth is not insufficient, but grotesquely maldistributed. That poverty is not a failure of growth, but a success of extraction. That development, as currently practiced, is not a project of liberation but of management — a system for ensuring that the poor remain poor enough to be governable, but not so destitute that the arrangement becomes politically unsustainable.

That is the story the Millennium Development Goals could not tell. But it is the story the evidence has been telling all along.