Global Extreme Poverty

OWID presents work from many different people and organizations. When citing this entry, please also cite the original data source. This entry can be cited as:

Max Roser and Esteban Ortiz-Ospina (2017) – ‘Global Extreme Poverty’. Published online at Retrieved from: [Online Resource]

This entry is concerned with extreme poverty. The World Bank is the main source for global information on extreme poverty today and sets the International Poverty Line. This poverty line was revised in 2015 – since then a person is considered to live in extreme poverty if he or she is living on less than 1.90 international-$ per day. The poverty measurement is based on the monetary value of a person’s consumption, but since consumption measures are unfortunately not available for all countries, the World Bank has to rely on income measures for some countries.

A key difficulty for measuring global poverty is that price levels are very different in different countries. For that reason it is not sufficient to convert the consumption levels of people in different countries simply by the market exchange rate; it is additionally necessary to adjust for differences in price levels between different countries. This is done through Purchasing Power Parity adjustments (explained below), which allow consumption and incomes to be expressed in so-called ‘international dollars’.

It is important to notice that the International Poverty Line is extremely low. Indeed, ‘extreme poverty’ is the adequate name for those living under this low threshold. Focusing on extreme poverty is important precisely because it captures those most in need. However, it is also important to point out that living conditions well above the International Poverty Line can still be characterized by poverty and hardship. Accordingly, in this entry we will also discuss the global distribution of people below poverty lines that are higher than 1.90 international-$.

In the past only a small elite enjoyed living conditions that we would not describe as a life in extreme poverty today. With the onset of industrialization and rising productivity the share of people living in extreme poverty started to decrease. The share of people in extreme poverty has decreased continuously over the course of the last 2 centuries.

Closely linked to this improvement in material living conditions is the improvement of global health and the expansion of global education that we have seen over these last 2 centuries. We also discuss the link between education, health, and poverty in this entry.

During the first half of the last century, the growth of the world population caused the absolute number of poor people in the world to increase, even though the share of people in poverty was going down. After around 1970, the decrease in poverty rates became so steep that the absolute number of people living in extreme poverty started falling as well. This trend of decreasing poverty – both in absolute numbers and as a share of the world population – has been a constant during the last three decades.

# Empirical View

# The importance of material living conditions

# What do poor people think about poverty?

Extreme poverty has been going down in the last two centuries. But why should we care? Is it not the case that poor people might only have less income but enjoy their lives just as much – or even more – than rich people?

One way to find out is to simply ask. This is what the Gallup Organization did. In the World Poll, Gallup asked people around the world what they thought about their standard of living – not only about their income. The economist Angus Deaton1 compared the answers that people in different countries gave with the average income in these countries. This research is summarized in the following graph; it shows that people living in poorer countries tend to be less satisfied with their living standards in a broader sense.

This suggests that economic prosperity is not a vain, unimportant goal but a means for a better life. The correlation between rising incomes and higher self-reported life satisfaction is shown in the entry on happiness.

We are pointing this out at the beginning of this entry because there is often a romantic take on what life in poverty is like. There is just no empirical evidence that would suggest that poverty is romantic for people who live in poverty.

It is somewhat understandable to romanticise poverty for anyone who has not experienced extreme poverty personally, since it is very difficult to imagine what it is like to live with very little material means. Even economists who think a lot about income and poverty find it difficult to understand what it means to live on a given income level. It is just hard to get an image of what life is like when all you know is a dollar per day figure. To address this Anna Rosling Rönnlund put together a very nice project at in which she portrays the living conditions of people living at different income levels. At Dollar Street you find portraits of families and you can see how they cook, what they eat, how they sleep, what toilets they have available, what their children’s toys look like and much more.

Fractions of people reporting that they are dissatisfied with their standard of living – Deaton2



# Poverty lines

The simplest, most straightforward way to measure poverty is to set a poverty line and to count the number of people living below that poverty line, and express it as a fraction of the total number of people in that society. Fittingly this is called the poverty headcount ratio.

The difficulty here is that it is not easy to draw a line between people that one should consider poor and the rest of the society. If you look at the previous graph, it is easy to understand why: Drawing the line would be easy if there was a clear delineation between people who were satisfied with their standard of living and people who aren’t. But this is not the case. We see, on the contrary, that having more income improves the self-perceived living standards of people. This is not only true for self-reported life satisfaction, but also for many welfare indicators; and it is never straightforward to define who should be considered poor and who should not.

The fact that higher income improves the standards of living means that defining poverty by drawing a line is always rightly subject to debate. It is necessarily controversial to propose a line that separates those that are poor from those that ‘merely’ have a low income.

The World Bank, which gathers data on income from people around the world, defined extreme poverty as living on less than $1.90 per day and you find more information on this poverty line in the section dedicated to measurement below. Global poverty is measured in international dollars (in prices of 2011) that are adjusted for the fact that people in different countries face different price levels (PPP adjustment). It is also expressed in real terms to adjust for price changes over time. (How these adjustments are made is explained in the entry on GDP data.)

# The historical perspective on extreme poverty

# 200 years of lifting the world out of poverty

The World Bank publishes data on absolute poverty from 1981 onwards, but researchers have tried to reconstruct information about the living standards of the more distant past. The seminal paper was written by Bourguignon and Morrison in 2002.3 In this paper, the two authors reconstructed measures of poverty as far back as 1820. The poverty line of 1.90 Int. Dollars per day was introduced only in 2015, so the 2002 paper used the measure of ‘one dollar per day’. This difference in the definition of poverty should be kept in mind when considering the following graph.

In 1820, the vast majority of people lived in extreme poverty and only a tiny elite enjoyed higher standards of living. Economic growth over the last 200 years completely transformed our world, and poverty fell continuously over the last two centuries. This is even more remarkable when we consider that the population increased 7-fold over the same time. In a world without economic growth, an increase in the population would result in less and less income for everyone. A 7-fold increase in the world population would be potentially enough to drive everyone into extreme poverty. Yet, the exact opposite happened. In a time of unprecedented population growth we managed to lift more and more people out of poverty.

It is very difficult to compare income or consumption levels over long periods because goods and services that are available change so much, and even completely new goods and services become available. This is so important that it would not be wrong to say that every person in the world was extremely poor in the 19th century. Nathan Rothschild was surely the richest man in the world when he died in 1836. But the cause of his death was an infection; this is a condition that can be treated with antibiotics, available for purchase today for less than a couple of cents.

This means that the level of poverty, especially for the distant past, is hard to judge. Indeed, this is one of the reasons why the ‘dollar a day’ poverty line was updated. What is however clear is the trend over time. As more and more countries industrialised and increased the productivity of work, the economies started to grow and poverty started to decline. According to Bourguignon and Morrison a little more than a quarter of the world population was not living in poverty by 1950.

From 1981 onward we have better empirical data on global extreme poverty. The Bourguignon and Morrison estimates for the past are based on national accounts and additional information on the level of inequality within countries. The data from 1981 onwards comes from the World Bank, which bases their estimates on household surveys.

According to these household surveys, 44% of the world population lived in absolute poverty in 1981. Since then, the share of poor people has declined very fast – in fact faster than ever before in world history. In 32 years the share of people living in extreme poverty was divided by 4, reaching levels below 11% in 2013. Although the World Bank estimates for 2015 are not available, the projections suggest that the incidence of extreme poverty has fallen below 10%.

There is also an interactive version of the same visualization.


# The world population in extreme poverty in absolute numbers

We have seen that the chance of being born into poverty has declined dramatically over the last 200 years. But what about the absolute number of people living in extreme poverty?

The visualisation below combines the information on the share of extreme poverty shown in the last chart, with the number of people living in the world. Prior to 1970 we use here the estimates of people ‘living in poverty’ from Bourguignon and Morrison (2002); from 1981 we use the World Bank estimates.

As we can see, in 1820 there were just under 1.1 billion people in the world, of which more than 1 billion lived in extreme poverty. Over the next 150 years the decline of poverty was not fast enough to offset the very rapid rise of the world population, so the number of non-poor and poor people increased. Since around 1970 we are living in a world in which the number of non-poor people is rising, while the number of poor people is falling. According to the estimates shown below there were 2.2 billion people living in extreme poverty in 1970, and there were 705 million people living in extreme poverty in 2015. The number of extreme poor people in the world is 3 times lower than in 1970.

In 1990 there were 2 billion people living in extreme poverty. With a reduction to 705 million in 2015 this means that – on average – on every day in these 25 days 137,000 fewer people were living in extreme poverty.4

On every day in the last 25 years there could have been a newspaper headline saying “The number of people in extreme poverty fell by 137,000 since yesterday”. Unfortunately the slow developments that transform our world entirely never make the news, and this is very reason why we are working on this online publication.

# The history of extreme poverty in today’s rich countries

In the thousands of years before the beginning of the industrial era, the vast majority of the world population lived in conditions that we would call extreme poverty today. Productivity levels were low and food was scarce – material living standards were generally very low.

The first countries in which people improved their living conditions were those that industrialised first. The chart below shows the decline of extreme poverty in these countries. It is based on estimates published by Martin Ravallion5 and it should be noted that extreme poverty here is measured against the older poverty line of 1.25 international $ in 2005 prices. The difference between this and the newer poverty line that is used in the rest of this entry is not large, so the estimates here are roughly comparable.

It shows that the living conditions of large parts of today’s rich countries were characterised by extreme poverty. The reduction of extreme poverty in these countries was only achieved very recently.

Ravallion writes that “today there is virtually no extreme poverty left in today’s rich world, when judged by the standards of poor countries today.” An exception to this is the US – a country with exceptionally high inequality among rich countries – where still a small but sizeable fraction of the population is living in extreme poverty (more about this below).

In other rich countries extreme poverty has ended and we should not forget what a very recent achievement this is. We definitely can end extreme poverty, we have done it before.

# Share of population below the International Poverty Line

Extreme poverty estimates, measuring the share of population living below the International Poverty Line, are available for many countries over the last 3 decades. These estimates are shown in the world map below.

By default the map shows the latest data, but with the slider below the map you can explore the change over time.

You can also switch to the ‘chart’ tab to see the change over time for individual countries or world regions. Or you simply click on a country to see how the poverty headcount has changed.

# Share of population living with less than 3.10$

Living on $1.90 per day is very difficult – the term extreme poverty is appropriate. As argued before it would be wrong to think that a person living on a little more than 1.90 international dollars is not poor.

It is therefore interesting and important to also measure poverty with higher poverty lines. The second poverty line for which the World Bank presents such estimates is 3.10 international dollars per day. The following chart shows the share of a country’s population that is living with less than this.

In this visualisation you can compare the share living with less than 1.90 international-$ and 3.10 international-$ for the same country.

# Poverty headcount across world regions

The following visualization shows the number of people living below the International Poverty Line in different world regions.

As we can see, in 1990 very few people lived in poverty in the early-industrialized parts of the world; yet other regions experienced substantial deprivation. Regional inequalities were very large. In the subsequent years, as emerging economies started catching up in the process of industrialization, some of these inequalities were reduced. Poverty started declining in other parts of the world. This can be appreciated in the trends for East Asia and the Pacific, where rapid economic growth meant that poverty decreased particularly rapidly.

Globally the number of people living in extreme poverty fell by more than 1 billion during this period; from 1.85 billion in 1990 to 0.76 billion in 2013. On average the number of people living in extreme poverty declined by 47 million every year since 1990 (or 130,000 every single day).

# Poverty incidence across world regions

In the following graph we show the share of the population, by world region, living below the International Poverty Line. This is a measure of the incidence of extreme poverty in different parts of the world. It is usually referred to as the ‘poverty headcount ratio’.

As we can see, an outstanding achievement of the last decades was the very rapid reduction of poverty in East Asia and Pacific; from 60% in 1990 to 3.5% in 2013. This reduction is also especially important as it is a very populous part of the world.

As we can see, the decline of global poverty is not only due to the rapid development in East Asia. As the visualisation shows this is part of a larger global development. The share of people living in extreme poverty has been falling in all world regions.

# The population in rich countries is largely unaware of the decline of global extreme poverty

The chart below shows the perceptions that survey-respondents in the UK have regarding global achievements in poverty reductions. While the share of extremely poor people has fallen faster than ever before in history over the last 30 years, the majority of people in the UK thinks that the opposite has happened, and that poverty has increased!

The chart below presents evidence from a survey in the UK, but ignorance of global development is even greater in other countries that were also surveyed. The extent of ignorance in the UK is particularly bad if we take into account that the shown result correspond to population with a university degree. See the Gapminder Ignorance Project for more evidence.

A more recent survey commissioned by Oxfam and others6 asked similar questions in poorer countries. They find that there are considerable difference in the answers provided in rich and poor countries: in Germany and the US only 8% of the survey respondents know that extreme poverty has declined, while in India and China the corresponding figures are 27% and 50% respectively.

Survey response in the UK to the question how global poverty has changed7


# The poverty gap is falling globally

So far we have been focusing on ‘headcount’ measures of poverty. These measures are easy to understand, but they have disadvantages, since they do not tell us anything about the intensity of poverty: a person that earns only 0.5 international-$ and a person that earns 1.89 international-$ per day are both called extremely poor. Alternative poverty measures, such as the ‘poverty gap’ account for the fact that poverty is more severe if incomes fall drastically below the poverty line.

The poverty gap is the mean shortfall of the total population from the poverty line (counting the non-poor as having zero shortfall), expressed as a percentage of the poverty line. It is explained in more detail here at the web page of the UN.

The visualisation below shows the poverty gap for the international poverty line of 1.90 international-$ per day.

The thick line shows the declining global poverty gap; from 19% in 1981 to below 4% in the latest estimates. The case of Madagascar shows that not all countries are making progress towards ending extreme poverty and some countries experience rising poverty levels instead. You can add other countries to the chart or switch to the map tab for a broader overview.

# The distribution of poverty in the world today

The incidence of extreme poverty has gone down from almost 100% in the 19th century, to 10.7% in 2013. While this is a great achievement, there is absolutely no reason to be complacent: a headcount ratio of 10.7% means that 746 million people are still living in extreme poverty. Where do they live?

The visualisation below shows the number of people in extreme poverty by continent and country.8

Above we discuss regional trends. In 1990 the world region with the largest number of poor people was Asia (505 million in South Asia, plus 966 million in East Asia and the Pacific). With the rapid economic growth in Asia over the following two decades, the number of people in extreme poverty fell rapidly. As can be seen from the visualisation below, the number was down to 327 million in 2013 – by that time Africa became the continent with the largest number of people living in extreme poverty.

The breakdown by continent is as follows:

  • 383 Million in Africa
  • 327 Million in Asia
  • 19 Million in South America
  • 13 Million in North America
  • 2.5 Million in Oceania
  • 0.7 Million in Europe

India is the country with the largest number of people living in extreme poverty (218 million people). Nigeria and the Congo (DRC) follow with 86 and 55 million people.


# The future of extreme poverty

The declared goal of the World Bank is to end extreme poverty by 2030. How realistic is this goal?

An interesting scenario is to assume that the current growth rates remain constant. This scenario, based on each countries’ growth rate over the last 10 years, is presented in the following visualisation. World poverty will further decrease, but it will remain quite a bit higher than the stated goal of 0%.


# Beyond income – poverty in other dimensions

So far we have only looked at poverty in terms of consumption and income. There is good reason to regard these measures as important; but income is a means to an end and not an end in itself. Other measures of deprivation allow us to take this into account.

One approach is to study deprivation across several independent dimensions of well-being – such as education, health, human rights, etc. One can even consider work as one more dimension, and attempt to capture other work-related aspects of well-being, such as safety and quality of labor. This is the approach that Our World in Data follows and you find the empirical research and data on all these topics right here.

A second approach consists in aggregating various dimensions of deprivation into one single metric of ‘multidimensional poverty’. Under this approach the aggregation of information regarding deprivation is made at the individual level, rather than at the social or national levels.

An example of this second approach is the Multidimensional Poverty Index published by the Oxford Poverty & Human Development Initiative. It is possible to visualize their research results on a world map.

# Correlates, Determinants & Consequences

# Poverty reduction and economic growth

By 1820 only a few places in the world had achieved economic growth – and only to a rather small extent. The reduction of poverty from the last 200 years was largely possible because economic growth brought higher incomes to more and more people in the world.

The following graph shows the correlation between average income and the share of the population living in extreme poverty. As we can see, the incidence of poverty tends to be lower in richer countries; but the correlation is of course not perfect, and there are also countries that have low poverty incidence at low average income levels.

The scatter plot above shows cross-country evidence of the link between economic prosperity and poverty. In the following graph we look at this relationship over time for Brazil. This country, with a population of almost 200 million people, achieved an average annual growth rate of 1.7% over 20 years. Through economic growth and redistributive policies, Brazil halved the share of people living in poverty over this short period of time.

Similarly to other industrializing countries, this reduction in poverty went together with a reduction in income inequality.

# Economic growth, decreasing inequality and decreasing poverty in Brazil, 1990-2011 – The Economist9

Economic Growth, Decreasing Inequality and Decreasing Poverty in Brazil (1990-2011) - The Economist

# Measurement and Data Quality

# Key Definitions

# What is the ‘poverty headcount ratio’?

The poverty headcount ratio is defined as the share of people with incomes (or consumption) below a certain poverty line. This measure is easy to understand, but it has disadvantages, since it does not tell us anything about the intensity of poverty: a person that earns only 0.5 international-$ and a person that earns 1.89 international-$ per day are both called extremely poor. Alternative poverty measures account for how far below the poverty line incomes are – this is what the poverty gap does.

# What is the ‘poverty gap’?

In contrast to the poverty headcount, the poverty gap takes the intensity of poverty into account. It takes into account that poverty is more severe if incomes fall drastically below the poverty line.

The poverty gap is the mean shortfall of the total population from the poverty line (counting the non-poor as having zero shortfall), expressed as a percentage of the poverty line. It is explained in more detail here at the web page of the UN. The issue of course with the poverty gap measure is that it is not as straightforward to explain as the poverty headcount and for public discussions it is therefore rarely used (but it should!).

# What is the difference between absolute and relative poverty?

The word ‘poverty’ is used in two distinct ways in social research. Not distinguishing between these causes confusion when comparing poverty across time and across countries.

Absolute poverty is measured relative to a fixed standard of living; that is, an income threshold that is constant across time and countries. This notion of poverty allows counting the number of people who are destitute by a common global standard.

Relative Poverty on the other hand is measured relative to living standards in a particular society, and varies both across time and between societies. This notion of poverty measures deprivation relative to the incomes and opportunities of other individuals in a given society at a given point in time – so it is closely related to inequality.

It is important to bear in mind that whether a given poverty measure is ‘relative’ or ‘absolute’ depends on the nature of the poverty line. Hence, it is possible to estimate the poverty headcount and the poverty gap both in absolute and relative terms – it just depends on whether the poverty line is defined relative to a fixed standard of living, or relative to the changing standard of living of other individuals.

# Where do absolute poverty lines come from?

The pioneering work to count the number of people in poverty by a common global standard was published by Montek Ahluwalia, Nicholas Carter, and Hollis Chenery in 197910. The three authors based their estimates of global poverty figures on the Indian poverty line at the time.

A global ‘dollar-a-day’ poverty line was introduced in the World Development Report 1990, and was subsequently used for the formulation of the Millennium Development Goals, through the commitment to “halve, between 1990 and 2015, the proportion of people whose income is less than $1 a day.” Since then the global poverty line has been revised, and today it is based on the national poverty lines of a number of low-income countries. It remains, however, a global poverty line for absolute measurement of deprivation– so it is not recurrently adjusted as low-income countries grow richer.

Another possibility to set a poverty line is to ask people about the level of income that is adequate in their society. This tradition of self-reports by poor people is based on the work of Bernard van Praag (1968)11. Above we show that there is a close relationship between the self-assessment of living conditions and the mean income in the society. Stevenson and Wolfers (2008)12 showed that this relationship not only holds between but also within countries – more information can be found in the entry about happiness and subjective well-being. National poverty lines are in fact often informed by the self-assessment of citizens of these countries.

# Data quality and limitations

# Should we care about data quality?

Any estimate of poverty — of either its level or change over time — is surrounded by a margin of error. In any consideration of the extent or the history of poverty it is crucial to keep this in mind.

Assessing the quality of data in this context is important for many reasons, not least because poverty estimates are used to make important decisions on the allocation of international aid. Indeed, reviewing the many problems of poverty measurement Deaton (2010)13 concludes that “probably the most urgent area for the poverty counts is not the [International Comparison Program], but the improvement in the consistency and timeliness of household surveys, and the upgrading of national accounts.”

Below we discuss how extreme poverty is measured and what the limitations of available estimates are. But before moving on, we want to stress that even if poverty measurement is imperfect, the estimates do tell us something important about how material living conditions are distributed, and how they change over time. Estimates of extreme poverty are imperfect but not meaningless. They should be used as one more source of information to assess living standards; such as health, education, nutrition, etc.

# What is the underlying source of data for World Bank estimates?

The World Bank is the most important institution measuring the extent of global poverty. This poverty data is published via Povcal Net and also in the World Development Indicators.

Although the World Bank is by far the most important source producing poverty estimates, the coverage of the underlying sources is not as good as one would hope. The following map shows the availability of surveys used to construct the World Bank’s poverty estimates.

For all countries that are shown in grey in this map there is not a single survey available to the World Bank in the last 3 decades. Many of these countries are rich countries in which extreme poverty is very low. But there is also missing data for some poorer countries, in which surely a considerable share of the population is living in extreme poverty.

As we can also see from this map, there are some countries with very few observations. This is the case for many African countries, where there is only one survey available in the last decade. This is extremely low, even in comparison to Latin America and Central Asia, where many countries have almost annual surveys.

# What are the main limitations of the World Bank estimates?

Ferreira et al (2015)14 describe in detail the method used for estimating the global poverty figures published in Povcal Net. The poverty estimates are primarily based on household survey data. For the global poverty estimates published in October 2015, the World Bank used income and consumption distribution data “from more than one thousand surveys covering 131 developing countries. The estimates are based on survey interviews of more than 2 million households, representative of 95 percent of the population in the developing world”.

These surveys are not carried out by the World Bank, but by the different National Statistical Offices and are therefore designed to be useful for the varying goals of these national offices. The challenge is therefore to harmonize this empirical data that was not designed to be internationally comparable. Here we summarize some of the most important difficulties.

Level of measurement

The first key issue for the World Bank is that the Povcal data is published in per capita terms, but the national surveys are mostly carried out on the household level.

Mix of consumption and income data: The second challenge for harmonizing the data is that the underlying household survey are a mix of income and consumption data. In poor countries, where many people are self-employed in agriculture, consumption is typically much easier to measure in surveys than is income. The table shows that approximately 75 percent of the countries in the PovcalNet database have data on per capita consumption and only a quarter of the countries – mostly in Latin America and the Caribbean – report income per capita data. The fact that the Povcal data is based on these two very different welfare measures is a key weakness of the data and one that is according to Ferreira et al (2015) impossible to correct for; the authors write “indeed, in part because there are few or no fixed, observable patterns in the relationship between income and consumption, no adjustments are made to income or consumption distributions in PovcalNet to make them more comparable.” Comparisons of income and consumption figures has shown that the inequality and poverty level of consumption estimates is lower than of that same economy’s income inequality and income poverty.15

Income and consumption distributions in PovcalNet for global poverty estimates16
 All of PovcalNet (as of October 2015)  2012 Poverty estimates  
Micro dataGrouped dataTotalMicro dataGrouped dataTotal

Survey nonresponse

There is also theoretical and empirical evidence presented by Korinek et al. (2005)17 showing that as people become richer, they are less likely to respond to surveys. When richer individuals or households are less likely to answer surveys than poor people it has the consequence that survey-based estimates of consumption and income will understate the mean level of prosperity and overstate the share of people in poverty.

Survey design

Questionnaires are designed differently in different countries and these differences matter significantly for the level of measured poverty. For example the recall period for food consumption matters for the assessment of food consumption in a population – research indicates that the reported food consumption is lower if the recall period is longer (for an example on India see Deaton and Kozel (2005)18 and for a study on how the questionnaire design matters see Beegle et al. (2012)19). Jolliffe (2001)20 finds that questionnaires with more food items listed report higher food consumption when compared with a questionnaire with fewer items. However,  higher consumption totals, as reported in the surveys with a shorter recall period, are not necessarily more accurate as the Indian NSSO Expert Group on Sampling Errors (2003)21 found.

Non-available microdata

For six of the 131 countries covered the World Bank relied (in October 2015) on grouped data. According to Ferreira et al (2015) this is because the World Bank was not authorized to share the microdata publicly and it could only rely on the grouped data that refers to quintile or decile shares.

Urban bias

The urban bias in pricing referring to the worry that in price comparisons more expensive goods in urban markets are measured and the often lower prices in rural areas are neglected.

Revisions limiting comparability

The entry on GDP data explains how adjustments are made for price changes over time (inflation) and between countries (PPP adjustments). For poverty measurements, the revision of PPPs have substantial consequences for the regional distribution but not for the trend over time as Deaton (2010)22 explains:

“The revision from 1985 to 1993 had relatively little effect on the global count but made sub-Saharan Africa appear to be much poorer, and Latin America appear to be much less poor. Indeed, it is the 1993 revision that “established” sub-Saharan Africa as the region with the highest headcount ratio, a fact that has dominated subsequent discussions of world poverty; prior to revision, the measured prevalence of poverty in South Asia was substantially higher than in sub-Saharan Africa. At the time of that revision, I commented that “changes of this size risk swamping real changes, and it seems impossible to make statements about changes in world poverty when the ground underneath one’s feet is changing in this way,” Deaton (2001). The 2005 revision shifts the ground even more. The global count for 1993 jumps by almost half a billion people, the headcount ratio for East Asia doubles, there are substantial increases in Africa and South Asia, and a large reduction in Latin America. The Table shows these numbers for 1993, for which there are estimates using three sets of PPPs, […]. Let me emphasize that the general trends […] reappear whichever PPPs and poverty lines are used. The shifting of the ground refers only to the level of global poverty, and to its regional distribution.”

Regional aggregates and timing of surveys

For individual countries the World Bank publishes poverty estimates only for years in which household survey data is available. For regional and global aggregates however it is necessary to estimate the poverty figures for all countries in a specific year. Ferreira et al (2015)23 explain that in cases in which survey data before and after that specific year is available the World Bank reports an average weighted by the relative distance from these two surveys. To arrive at data for all countries in a specific year the consumption or income data from the latest survey is extrapolated by using growth rates from the national accounts. In this extrapolation the World Bank assumes distribution-neutral growth (no change in inequality since the last survey).

There are some countries for which no poverty estimates are available. Not including these countries in the regional or global aggregates would understate the level of global poverty. To avoid this the average poverty estimate of the world region in which this country is located is then assigned to that country for which no data is available.

High-income countries – according to the definition of the World Bank – are assumed to have no people living in extreme poverty in the global poverty figures published by the World Bank. Ferreira et al (2015) explain that “although some people in rich countries report household per capita incomes that are below the international poverty line, per capita consumption is above this threshold for nearly everyone.” Chandy and Smith (2014) for example find that 1% to 4% of the US population have incomes of less than $2 a day, but when measured by consumption this share of Americans living with less than 2$ per day falls to less than 0.1%.

The timing of surveys can be problematic for several reasons. Most countries do not have regular annual surveys and to arrive at regional or global estimates for a particular year the World Bank uses data from the National Accounts to fill in the gaps. This makes it particularly problematic to link changes in poverty estimates to current events – such as the financial crisis. It is preferable to study the change of poverty over longer time periods for which the data is based on household surveys. Additionally the timing of survey within a given year can matter for the measurement as for example the food provision varies throughout the year.

# National accounts offer an alternative for measuring aggregate deprivation

The poverty figures published by the World Bank are based on household surveys as explained above. A second possibility to measure poverty is to start from the National Accounts. This is done by using National Accounts data (from the Penn World Tables) combined with inequality measures. This technique was first used by Ahluwalia et al. (1979)24. More recent papers that followed this approach include Bourguignon and Morrisson (2002) whose data is shown above.

While in theory national account measures of poverty should match those from household surveys, in reality there are often large discrepancies. Deaton (2005)25 reviews the reasons for these discrepancies.

# How much weight should we give to poverty estimates from national accounts?

Economic activity around the world has been shown to correlate with satellite-recorded data on nighttime lights from the surface of the Earth, which are visible from space. This is intuitive, since nighttime lights are mainly generated by human activity – typically lights in buildings and cars.

The following visualization shows how economic activity, as measured by nighttime lights, changed in South Asia between the years 1994 and 2010. We can see that night lights in 2010 cover areas that were unlit in 1994; and there is also a substantial increase in the intensity of lights in major cities over the same period. This is clearly indicative of the underlying growth in economic activity that South Asia witnessed during this period.

Based on this correlation between nighttime lights and economic activity, Pinkovskiv and Sala-i-Martin (2016)26 have evaluated the relative quality of aggregate income data generated by household surveys and national accounts.

The intuition for their approach is that, as long as the measurement error in nighttime lights is unrelated to the measurement errors in either national accounts or survey means, one can use nightlights as a tool to see how much weight to give national accounts income estimates (relative to household-survey income estimates) in measuring true income.

The authors find that nighttime lights are much better correlated with GDP per capita than with survey means – which suggests, under their assumptions, that the optimal ‘weights’ to calculate aggregate income should be very large for national accounts and very modest for survey means.

This is important since, as we have pointed out, there are large discrepancies between estimates generated by household surveys and national accounts.


# Data Sources

# Long-term development of global poverty

# Bourguignon and Morrisson (2002)

Economists Xavier Sala-i-Martin and Maxim Pinkovskiy estimated the share of the world population living in absolute poverty.27

An important recent paper on absolute poverty is Chen and Ravallion (2010) – The Developing World is Poorer than We Thought, But No Less Successful in the Fight Against Poverty. In The Quarterly Journal of Economics, 125, 4, 1577–1625.

# Data on global poverty in recent decades

# World Bank
  • Data: Several measures of absolute poverty.
  • Geographical coverage: Global – by country and world region.
  • Time span: Since 1980
  • Available at: World Bank’s PovcalNet – an interactive tool which visualizes absolute poverty and makes the data available for download.
  • There is a collection of World Bank articles about declining poverty.
  • The World Bank data on extreme poverty (% of people below 1.25$ a day) is also available on Gapminder where the relationship with other measures of wellbeing can be visualised.

Data on the sub-national level (with huge coverage!) is available from the World Bank. These are data on the poverty headcount – at national poverty line, urban poverty line, and the rural poverty line.