There is no single definition of poverty. Our understanding of the extent of poverty and how it is changing depends on which definition we have in mind.

In particular, richer and poorer countries set very different poverty lines in order to measure poverty in a way that is informative and relevant to the level of incomes of their citizens.

For instance, while in the United States a person is counted as being in poverty if they live on less than roughly $27.10 per day, in Ethiopia the poverty line is set more than 10 times lower, at $2.59 per day. You can read more about how these comparable national poverty lines are calculated in this footnote.

Because the definition of poverty varies so widely, national poverty lines can’t be used to compare poverty across countries. To measure poverty globally, we need to use a poverty line that is applied consistently across all countries.

This is the goal of the International Poverty Line of $3 per day, which is set by the World Bank and used by the UN to monitor extreme poverty around the world.

In global terms, this is an extremely low threshold — set to represent the typical poverty lines adopted in the world’s poorest countries. It marks an incredibly low standard of living — a level of income much lower than just the cost of a healthy diet.

How does the World Bank set the International Poverty Line?

The exact method used by the World Bank to set the International Poverty Line has changed over time. But each time, the goal has remained broadly the same: to find a “typical standard by which the poorest countries of the world judge their citizens to be impoverished.”

The method used in the latest update to arrive at a figure of $3 (expressed in 2021 international-$) can be summarized as follows.

The World Bank begins by collecting a large set of national poverty lines — the lines used by individual countries to estimate official poverty rates among their populations. The World Bank has developed an approach to make these comparable.

The IPL aims to reflect the typical definition of poverty adopted among “low-income” countries, classified using the World Bank’s income classification system. For this, the World Bank uses the median value of these poverty lines.

Although the International Poverty Line is by far the most prominent international line, the same method is also used by the World Bank to set two higher poverty lines that reflect the national definitions adopted in lower-middle and upper-middle income countries. The median poverty line among these two groups of countries are $4.20 and $8.30, respectively.

You can read more in our article: The $3 a day International Poverty Line.

Much of the economic data we use to understand the world, such as the incomes people receive or the goods and services firms produce and people buy, is recorded in the local currencies of each country. That means the numbers start out in rupees, US dollars, yuan, and many others, and without adjusting for inflation over time. This is known as being in “current prices” or “nominal” terms.

Before these figures can be meaningfully compared, they need to be converted into common units. International dollars (int.-$) are a hypothetical currency that is used for this.

The idea is simple: one international dollar should buy the same quantity and quality of goods and services, no matter where or when it is spent. To achieve this, international dollars adjust for two things. First, they account for inflation within each country, so that values from different years can be compared (showing “constant” prices). Second, they account for differences in living costs across countries. This second adjustment uses purchasing power parity (PPP) rates, which reflect how much local currency is needed to buy what one US dollar would buy in the United States.

The United States is the benchmark, so that one 2021 int.-$ is defined as the value of goods and services that one US dollar would buy in the US in 2021. One 2011 int.-$ is defined in the same way, but for prices in 2011.

You can read more in our article, What are international dollars?

There is no single global survey of incomes. What we have instead are national surveys, each designed by a different statistical agency, using different methodologies. The World Bank collects and harmonizes this data, but important comparability issues remain.

One key issue is that high-income countries typically measure people's incomes, while lower-income countries more often measure consumption expenditure — what households spend on goods and services. Pooling both types of survey is unavoidable if we want a global picture of inequality, but it means that somewhat different things are being measured depending on the country or year.

The two concepts are closely related: the income of a household equals its consumption plus savings.

At the bottom end of the income distribution, people’s consumption may be somewhat higher than their income. While zero consumption is not a feasible value — people must consume something to survive — a zero income is a feasible value. A common example is retired people drawing down their savings: they may have a very low, or even zero, income, but still have a high level of consumption.

At the top end of the distribution, consumption is typically lower than income. The gap rises with income, with households generally saving a higher share of their income the richer they are.

For both reasons, the distribution of consumption is generally more equal than the distribution of income. This means that inequality estimates tend to be somewhat lower when based on consumption surveys.

There are other comparability issues too — differences in survey design, coverage, and methodology. The PIP Methodology Handbook provides a good summary of the comparability and data quality issues affecting this data and how it tries to address them.

To help readers see where comparisons may be less reliable, the World Bank groups data points within each country into "spells" — periods where the underlying surveys are considered more comparable. Where available, you can reveal these breaks in our charts using the "breaks in data" option.