Before we can ask why some people are poor, we need to know who is poor and how we know. Measuring poverty sounds simple, but every measure involves choices, and each one shows some things and hides others.

The international poverty line

The World Bank counts people living in extreme poverty using an international poverty line. It began in 1990 as roughly a dollar a day, based on the poverty lines of some of the world's poorest countries. Prices are converted using purchasing power parity (PPP), which adjusts for the fact that a dollar buys more in some countries than others. The line is revised periodically as new price data arrive. It has been raised several times, to 2.15 dollars in 2022 and to about 3 dollars a day in 2025. Each revision changes the count somewhat, so figures from different years must be compared with care.

The line is deliberately very low. It marks a level at which survival itself is a struggle.

The long decline, stated carefully

By historians' rough estimates, most of humanity lived in what we would now call extreme poverty two centuries ago. Around 1990, on the World Bank's measures, more than a third of the world's population was below the extreme line. By the late 2010s the share had fallen to roughly a tenth, even as the population grew. This is one of the largest improvements in human welfare on record. Several caveats belong beside it:

• Much of the fall happened in East Asia, especially China. Progress elsewhere was slower, and most people in extreme poverty now live in sub-Saharan Africa.
• Progress stalled around 2020, when the COVID-19 pandemic pushed the count up for the first time in decades.
• The numbers rest on household surveys that are infrequent in some countries and missing in others, so estimates carry real uncertainty.
• Escaping the line is not escaping poverty. Hundreds of millions remain extremely poor, and far more live on modest amounts above it.

GDP per capita

The most common measure of a country's economic development is gross domestic product (GDP) per capita: the value of goods and services produced in a year, divided by the population. It is useful because it is widely available and tracks average material living standards. But it is an average, so it says nothing about how income is shared, and it ignores unpaid work, leisure, environmental damage and much of what makes life go well.

The Human Development Index

In 1990 the United Nations Development Programme published the first Human Development Index (HDI), led by the Pakistani economist Mahbub ul Haq. It combines three dimensions: a long and healthy life, education, and a decent standard of living, measured by life expectancy, years of schooling and income per person. Countries with similar incomes can rank very differently on the HDI, which reveals choices about health and schooling that GDP hides. Critics note that its weighting of the three dimensions is somewhat arbitrary.

Sen's capability approach

Behind the HDI lies the thinking of the Indian economist and philosopher Amartya Sen, who received the Nobel memorial prize in economics in 1998. In paraphrase, Sen argued that development should be judged not by income or goods alone but by people's capabilities: their real freedom to do and be the things they have reason to value, such as being well nourished, reading, taking part in community life and having a say in politics. Income matters, but as a means. Two people with the same income may have very different capabilities if one is disabled, or lives where there are no schools or clinics. Sen summed up this view in his book Development as Freedom (1999).

His approach has inspired other measures, such as a multidimensional poverty index that counts deprivations in health, education and living conditions directly.

Worked example: two invented villages

Consider two fictional villages, each of 1,000 people, with the same average income: about 4 dollars per person per day, adjusted for prices.

In Village A, income is spread fairly evenly. Almost everyone is above the extreme poverty line. But there is no clinic within a day's walk, the school has one teacher for 90 children, and girls usually leave after three years. Many adults cannot read.

In Village B, a few landowners earn a great deal and many laborers earn very little, so several hundred people fall below the line. Yet there is a health post with a nurse, and the school has enough teachers and keeps most girls through primary school.

Now apply the measures. GDP per capita says the villages are identical. The poverty headcount says Village B is much worse. An index like the HDI, which counts health and education, would favor Village B on those dimensions while noting its lower incomes for many. Sen's capability approach asks a more searching question: what can people in each village actually do and be? In Village A, a girl has enough to eat but little chance to learn to read or to see a nurse when sick. In Village B, a laborer's child may go hungry some weeks but can learn and get care.

No single number captures both villages' situations. That is the point. Each measure answers a different question, and a good analyst uses several, asks what each leaves out, and remembers that behind every figure are particular people whose lives are not averages.