This is the author’s post at Oxford: https://www.rsc.ox.ac.uk/news/global-poverty-trends-through-a-new-lens-olivier-sterck-article-for-voxdev

Has global poverty fallen since 1990? Depending on which poverty line you use, the answer ranges from “we’ve made huge progress” to “nothing has changed”.

Using the World Bank’s extreme poverty line of US$2.15/day (in 2017 PPP), the share of people in poverty fell from 38% of the world’s population in 1990 (about 2 billion people) to 8.5% in 2024 (690 million people) (Figure 1). This is often cited as a historic success.

But raise the line – say to $21.5/day, as suggested by Pritchett and Viarengo (2025), or $30/day, as argued by Roser (2024) – and the picture changes entirely. The poverty rate is then extremely high, above 75%, and has barely budged since 1990. In absolute terms, the poverty headcount has even increased, from over 4 billion poor people in 1990 to over 6 billion poor people in 2024. Based on these numbers, the fight against global poverty appears to have failed.

This divergence is not just a statistical quirk. All mainstream poverty measures share the same fundamental feature: they ignore everyone above the chosen line. With the extreme poverty line of the World Bank ($2.15/day), someone earning $2.16/day is treated as equally non-poor as someone earning $10, $100, or $1,000/day. Billions of low-income people – who most would agree still live in poverty – are therefore excluded from the statistics. And because there is no consensus on where to set the line, it is tempting to pick the one that tells the story you want.

In Sterck (2026), I propose to measure income poverty without a poverty line. The idea is to measure poverty across the entire income distribution, rather than classifying people as poor or non-poor based on an arbitrary threshold.

The measure’s key intuition is simple: if person A earns half as much as person B, then A is twice as poor. Poverty is therefore simply measured as the reciprocal of income, and its unit is simply inverted. If incomes are measured in dollars per day ($/day), poverty is measured in days per dollar (days/$).

Average poverty is simply the average time it takes to earn $1 in a given population.

In 2024, that value was equivalent to:

• 1 day in DR Congo, Madagascar, South Sudan, and Mozambique
• 12 hours in Haiti
• 2 hours in China
• 85 minutes in the US
• 25 minutes in Switzerland.