research report of poverty

March 2024 global poverty update from the World Bank: first estimates of global poverty until 2022 from survey data

R. andres castaneda aguilar, carolina diaz-bonilla, christoph lakner, minh cong nguyen, martha viveros, samuel kofi tetteh baah.

Global poverty estimates were updated today on the Poverty and Inequality Platform (PIP).   As explained in more detail in the What’s New document, more than 100 new surveys were added to the PIP database, bringing the total number of surveys to more than 2,300. With more recent survey data, this March 2024 PIP update is the first to report a global poverty number for 2020-2022, the period of the COVID-19 pandemic. We estimate that COVID-19 increased extreme poverty in the world, as measured by the international poverty line of $2.15, from 8.9 percent in 2019 to 9.7 percent in 2020 (see Figure 1). This is the first increase in global poverty in decades. It is in line with earlier estimates of the COVID-19 impact which used limited survey data and GDP growth projections.  The global increase in extreme poverty in 2020 is driven by South Asia, where extreme poverty increased by 2.4 percentage points to 13 percent between 2019 and 2020. In Latin America and the Caribbean, however, extreme poverty continued to decline in 2020, which is driven by Brazil. This can be explained by the role of fiscal policy in mitigating the economic impacts of the COVID-19 shock . At a higher poverty line of $3.65 ( the poverty line more relevant for assessing poverty in lower-middle-income countries ), poverty also fell in Latin America and the Caribbean and in East Asia and the Pacific even in 2020. At $6.85 ( the poverty line more relevant for assessing poverty in upper-middle-income countries ), poverty also declined in 2020 in Europe and Central Asia and in advanced countries (“Other High Income”). Unfortunately, survey coverage during the post-2019 period is still limited in Sub-Saharan Africa and the Middle East and North Africa, so we cannot report poverty estimates beyond 2019 for these regions. 

Figure 1: Global and regional poverty estimates, 1990 - 2022

Following the widespread recession in 2020, economies around the world started to recover in 2021 and extreme poverty levels were lower than pre-pandemic levels in the more prosperous regions of the world by 2022 (East Asia and the Pacific, Latin America and the Caribbean, advanced countries, Europe and Central Asia, and South Asia). For the world, however, global poverty was still marginally above pre-pandemic levels by 2022, though on a declining trend.  The new estimates of extreme poverty in the world in the period 2020-2022 are quite similar to earlier projections.   An estimated 23 million more people were living in extreme poverty in 2022, compared to 2019. That extreme poverty levels were lower in 2022 relative to 2019 for more prosperous regions, but not for the world, suggests that the economic recovery from the pandemic was uneven and slower for Sub-Saharan Africa where more than half of the extreme poor live. The year 2022 also came with another global shock – Russia’s invasion of Ukraine, which contributed to rising inflation in low-, middle- and high-income countries. At the $3.65 and $6.85 poverty lines, the global poverty rate in 2022 are lower the levels recorded in 2019. This result is consistent with the recovery being faster in more prosperous regions, considering that Sub-Saharan Africa accounts for a smaller share of the global poor at these higher lines compared to the extreme poverty line. This March 2024 global poverty update from the World Bank incorporates updated CPI, national accounts and population data, and revises previously published global and regional estimates from 1981 to 2022. The methodology used for lining up regional and global poverty has also been revised, which leads to small changes. For more details, see the What’s New document. Figure 1 shows global and regional poverty trends at all three global absolute poverty lines of the World Bank (see the poverty series using 2011 PPPs here. ) Table 1 summarizes the revisions to the regional and global poverty estimates between the September 2023 data vintage and the March 2024 data vintage for the 2019 reference year at all three poverty lines.

Table 1: Poverty estimates for reference year 2019, changes between the September 2023 and March 2024 PIP vintages

Given all the new data points and revisions to PIP data and methodology in this update, global extreme poverty in 2019 has been revised down marginally by 0.1 percentage points to 8.9 percent, resulting in a downward revision in the number of poor people from 701 to 689 million. The global reduction in the millions of extreme poor occurs despite an upward revision in Sub-Saharan Africa (14 million). The reduction is driven by Europe and Central Asia and the Middle East and North Africa, where new survey data have recently become available to replace extrapolations of very old surveys (the regional estimate for Middle East and North Africa cannot be shown since it does not meet the 50% population cut-off.) For example, new survey data for 2022 have been added for Syria and Uzbekistan, for which the latest surveys were 2003 in the previous vintage of the data. At $3.65 and $6.85, poverty rates have been revised down by 0.7 and 0.6 percentage points, representing a reduction in global poverty counts by 52 and 44 million, respectively. These downward revisions in global poverty estimates at these higher poverty thresholds are driven by Europe and Central Asia and South Asia. For more details on the March 2024 PIP update from the World Bank, see the What’s New document.

Would you like to be updated with the latest news on PIP? Register to our newsletter  here .

The authors gratefully acknowledge financial support from the UK Government through the Data and Evidence for Tackling Extreme Poverty (DEEP) Research Program.

• Development Economics

Get updates from Data Blog

Thank you for choosing to be part of the Data Blog community!

Your subscription is now active. The latest blog posts and blog-related announcements will be delivered directly to your email inbox. You may unsubscribe at any time.

Economist, Development Data Group, World Bank

Senior Economist, Poverty and Equity Global Practice, World Bank

Data Scientist

Program Manager, Development Data Group, World Bank

Senior Data Scientist, Poverty and Equity Global Practice, World Bank

Consultant, Development Data Group, World Bank

Economist, Global Poverty and Inequality Data (GPID), Development Data Group, World Bank

Join the Conversation

• Share on mail
• comments added

By: Joe Hasell , Max Roser , Esteban Ortiz-Ospina and Pablo Arriagada

Global poverty is one of the most pressing problems that the world faces today. The poorest in the world are often undernourished , without access to basic services such as electricity and safe drinking water ; they have less access to education , and suffer from much poorer health .

In order to make progress against such poverty in the future, we need to understand poverty around the world today and how it has changed.

On this page you can find all our data, visualizations and writing relating to poverty. This work aims to help you understand the scale of the problem today; where progress has been achieved and where it has not; what can be done to make progress against poverty in the future; and the methods behind the data on which this knowledge is based.

Key Insights on Poverty

Measuring global poverty in an unequal world.

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 $24.55 per day, in Ethiopia the poverty line is set more than 10 times lower – at $2.04 per day. You can read more about how these comparable national poverty lines are calculated in this footnote. 1

To measure poverty globally, however, we need to apply a poverty line that is consistent across countries.

This is the goal of the International Poverty Line of $2.15 per day – shown in red in the chart – which is set by the World Bank and used by the UN to monitor extreme poverty around the world.

We see that, in global terms, this is an extremely low threshold indeed – set to reflect the poverty lines adopted nationally 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 .

From $1.90 to $2.15 a day: the updated International Poverty Line

What you should know about this data.

• Global poverty data relies on national household surveys that have differences affecting their comparability across countries or over time. Here the data for the US relates to incomes and the data for other countries relates to consumption expenditure. 2
• The poverty lines here are an approximation of national definitions of poverty, made in order to allow comparisons across the countries. 1
• Non-market sources of income, including food grown by subsistence farmers for their own consumption, are taken into account. 3
• Data is measured in 2017 international-$, which means that inflation and differences in the cost of living across countries are taken into account. 4

Global extreme poverty declined substantially over the last generation

Over the past generation extreme poverty declined hugely. This is one of the most important ways our world has changed over this time.

There are more than a billion fewer people living below the International Poverty Line of $2.15 per day today than in 1990. On average, the number declined by 47 million every year, or 130,000 people each day. 5

The scale of global poverty today, however, remains vast. The latest global estimates of extreme poverty are for 2019. In that year the World Bank estimates that around 650 million people – roughly one in twelve – were living on less than $2.15 a day.

Extreme poverty: how far have we come, how far do we still have to go?

• Extreme poverty here is defined according to the UN’s definition of living on less than $2.15 a day – an extremely low threshold needed to monitor and draw attention to the living conditions of the poorest around the world. Read more in our article, From $1.90 to $2.15 a day: the updated International Poverty Line .
• Global poverty data relies on national household surveys that have differences affecting their comparability across countries or over time. 2
• Surveys are less frequently available in poorer countries and for earlier decades. To produce regional and global poverty estimates, the World Bank collates the closest survey for each country and projects the data forward or backwards to the year being estimated. 6
• Data is measured in 2017 international-$, which means that inflation and differences in the cost of living across countries are taken into account . 4

The pandemic pushed millions into extreme poverty

Official estimates for global poverty over the course of the Coronavirus pandemic are not yet available.

But it is clear that the global recession it brought about has had a terrible impact on the world’s poorest.

Preliminary estimates produced by researchers at the World Bank suggest that the number of people in extreme poverty rose by around 70 million in 2020 – the first substantial rise in a generation – and remains around 70-90 million higher than would have been expected in the pandemic’s absence. On these preliminary estimates, the global extreme poverty rate rose to around 9% in 2020. 7

• Figures for 2020-2022 are preliminary estimates and projections by World Bank researchers, based on economic growth forecasts. The pre-pandemic projection is based on growth forecasts prior to the pandemic. You can read more about this data and the methods behind it in the World Bank’s Poverty and Shared Prosperity 2022 report. 8

Hundreds of millions will remain in extreme poverty on current trends

Extreme poverty declined during the last generation because the majority of the poorest people on the planet lived in countries with strong economic growth – primarily in Asia.

The majority of the poorest now live in Sub-Saharan Africa, where weaker economic growth and high population growth in many countries has led to a rising number of people living in extreme poverty.

The chart here shows projections of global extreme poverty produced by World Bank researchers based on economic growth forecasts. 9

A very bleak future is ahead of us should such weak economic growth in the world’s poorest countries continue – a future in which extreme poverty is the reality for hundreds of millions for many years to come.

• The extreme poverty estimates and projections shown here relate to a previous release of the World Bank’s poverty and inequality data in which incomes are expressed in 2011 international-$. The World Bank has since updated its methods, and now measures incomes in 2017 international-$. As part of this change, the International Poverty Line used to measure extreme poverty has also been updated: from $1.90 (in 2011 prices) to $2.15 (in 2017 prices). This has had little effect on our overall understanding of poverty and inequality around the world. You can read more about this change and how it affected the World Bank estimates of poverty in our article From $1.90 to $2.15 a day: the updated International Poverty Line .
• Figures for 2018 and beyond are preliminary estimates and projections by Lakner et al. (2022), based on economic growth forecasts. You can read more about this data and the methods behind it in the related blog post. 10
• Data is measured in 2011 international-$, which means that inflation and differences in the cost of living across countries are taken into account. 4

The rapid progress seen in many countries shows an end to poverty is possible

Each of the countries shown in the chart achieved large declines in extreme poverty over the last generation. 11

The fact that rapid progress against poverty has been achieved in many places is one of the most important lessons we can learn from the available data on extreme poverty.

For those who are not aware of such progress – which is the majority of people – it would be easy to make the mistake of believing that poverty is inevitable and that action to tackle poverty is hence doomed to fail.

The huge progress seen in so many places shows that this view is incorrect.

Interactive visualization requires JavaScript.

After 200 years of progress the fight against global poverty is just beginning

Over the past two centuries the world made good progress against extreme poverty. But only very recently has poverty fallen at higher poverty lines.

Global poverty rates at these higher lines remain very high:

• 25% of the world lives on less than $3.65 per day – a poverty line broadly reflective of the lines adopted in lower-middle income countries.
• 47% of the world lives on less than $6.85 per day – a poverty line broadly reflective of the lines adopted in upper-middle income countries.
• 84% live on less than $30 per day – a poverty line broadly reflective of the lines adopted in high income countries. 12

Economic growth over the past two centuries has allowed the majority of the world to leave extreme poverty behind. But by the standards of today’s rich countries, the world remains very poor. If this should change, the world needs to achieve very substantial economic growth further still.

The history of the end of poverty has just begun

How much economic growth is necessary to reduce global poverty substantially?

• The data from 1981 onwards is based on household surveys collated by the World Bank. Earlier figures are from Moatsos (2021), who extends the series backwards based on historical reconstructions of GDP per capita and inequality data. 13
• All data is measured in international-$ which means that inflation and differences in purchasing power across countries are taken into account. 4
• The World Bank data for the higher poverty lines is measured in 2017 international-$. Recently, the World Bank updated its methodology having previously used 2011 international-$ to measure incomes and set poverty lines. The Moatsos (2021) historical series is based on the previously-used World Bank definition of extreme poverty – living on less than $1.90 a day when measured in 2011 international-$. This is broadly equivalent to the current World Bank definition of extreme poverty – living on less than $2.15 a day when measured in 2017 international-$. You can read more about this update to the World Bank’s methodology and how it has affected its estimates of poverty in our article From $1.90 to $2.15 a day: the updated International Poverty Line .
• The global poverty data shown from 1981 onwards relies on national household surveys that have differences affecting their comparability across countries or over time. 2
• Such surveys are less frequently available in poorer countries and for earlier decades. To produce regional and global poverty estimates, the World Bank collates the closest survey for each country and projects the data forward or backwards to the year being estimated. 6
• Non-market sources of income, including food grown by subsistence farmers for their own consumption, are taken into account. This is also true of the historical data – in producing historical estimates of GDP per capita on which these long-run estimates are based, economic historians take into account such non-market sources of income, as we discuss further in our article How do we know the history of extreme poverty?

Explore Data on Poverty

About this data.

All the data included in this explorer is available to download in GitHub , alongside a range of other poverty and inequality metrics.

Where is this data sourced from?

This data explorer is collated and adapted from the World Bank’s Poverty and Inequality Platform (PIP).

The World Bank’s PIP data is a large collection of household surveys where steps have been taken by the World Bank to harmonize definitions and methods across countries and over time.

About the comparability of household surveys

There is no global survey of incomes. To understand how incomes across the world compare, researchers need to rely on available national surveys.

Such surveys are partly designed with cross-country comparability in mind, but because the surveys reflect the circumstances and priorities of individual countries at the time of the survey, there are some important differences.

Income vs expenditure surveys

One important issue is that the survey data included within the PIP database tends to measure people’s income in high-income countries, and people’s consumption expenditure in poorer countries.

The two concepts are closely related: the income of a household equals their consumption plus any saving, or minus any borrowing or spending out of savings.

One important difference is that, while zero consumption is not a feasible value – people with zero consumption would starve – a zero income is a feasible value. This means that, at the bottom end of the distribution, income and consumption can give quite different pictures about a person’s welfare. For instance, a person dissaving in retirement may have a very low, or even zero, income, but have a high level of consumption nevertheless.

The gap between income and consumption is higher at the top of this distribution too, richer households tend to save more, meaning that the gap between income and consumption is higher at the top of this distribution too. Taken together, one implication is that inequality measured in terms of consumption is generally somewhat lower than the inequality measured in terms of income.

In our Data Explorer of this data there is the option to view only income survey data or only consumption survey data, or instead to pool the data available from both types of survey – which yields greater coverage.

Other comparability issues

There are a number of other ways in which comparability across surveys can be limited. 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.

In collating this survey data the World Bank takes a range of steps to harmonize it where possible, but comparability issues remain. These affect comparisons both across countries and within individual countries over time.

To help communicate the latter, the World Bank produces a variable that groups surveys within each individual country into more comparable ‘spells’. Our Data Explorer provides the option of viewing the data with these breaks in comparability indicated, and these spells are also indicated in our data download .

Global and regional poverty estimates

Along with data for individual countries, the World Bank also provides global and regional poverty estimates which aggregate over the available country data.

Surveys are not conducted annually in every country however – coverage is generally poorer the further back in time you look, and remains particularly patchy within Sub-Saharan Africa. You can see that visualized in our chart of the number of surveys included in the World Bank data by decade.

In order to produce global and regional aggregate estimates for a given year, the World Bank takes the surveys falling closest to that year for each country and ‘lines-up’ the data to the year being estimated by projecting it forwards or backwards.

This lining-up is generally done on the assumption that household incomes or expenditure grow in line with the growth rates observed in national accounts data. You can read more about the interpolation methods used by the World Bank in Chapter 5 of the Poverty and Inequality Platform Methodology Handbook.

How does the data account for inflation and for differences in the cost of living across countries?

To account for inflation and price differences across countries, the World Bank’s data is measured in international dollars. This is a hypothetical currency that results from price adjustments across time and place. It is defined as having the same purchasing power as one US-$ would in the United States in a given base year. One int.-$ buys the same quantity of goods and services no matter where or when it is spent.

There are many challenges to making such adjustments and they are far from perfect. Angus Deaton ( Deaton, 2010 ) provides a good discussion of the difficulties involved in price adjustments and how this relates to global poverty measurement.

But in a world where price differences across countries and over time are large it is important to attempt to account for these differences as well as possible, and this is what these adjustments do.

In September 2022, the World Bank updated its methodology, and now uses international-$ expressed in 2017 prices – updated from 2011 prices. This has had little effect on our overall understanding of poverty and inequality around the world. But poverty estimates for particular countries vary somewhat between the old and updated methodology. You can read more about this update in our article From $1.90 to $2.15 a day: the updated International Poverty Line .

To allow for comparisons with the official data now expressed in 2017 international-$ data, the World Bank continues to release its poverty and inequality data expressed in 2011 international-$ as well. We have built a Data Explorer to allow you to compare these, and we make all figures available in terms of both sets of prices in our data download .

Absolute vs relative poverty lines

This dataset provides poverty estimates for a range of absolute and relative poverty lines.

An absolute poverty line represents a fixed standard of living; a threshold that is held constant across time. Within the World Bank’s poverty data, absolute poverty lines also aim to represent a standard of living that is fixed across countries (by converting local currencies to international-$). The International Poverty Line of $2.15 per day (in 2017 international-$) is the best known absolute poverty line and is used by the World Bank and the UN to measure extreme poverty around the world.

The value of relative poverty lines instead rises and falls as average incomes change within a given country. In most cases they are set at a certain fraction of the median income. Because of this, relative poverty can be considered a metric of inequality – it measures how spread out the bottom half of the income distribution is.

The idea behind measuring poverty in relative terms is that a person’s well-being depends not on their own absolute standard of living but on how that standard compares with some reference group, or whether it enables them to participate in the norms and customs of their society. For instance, joining a friend’s birthday celebration without shame might require more resources in a rich society if the norm is to go for an expensive meal out, or give costly presents.

Our dataset includes three commonly-used relative poverty lines: 40%, 50%, and 60% of the median.

Such lines are most commonly used in rich countries, and are the main way poverty is measured by the OECD and the European Union .More recently, relative poverty measures have come to be applied in a global context. The share of people living below 50 per cent of median income is, for instance, one of the UN’s Sustainable Development Goal indicators . And the World Bank now produces estimates of global poverty using a Societal Poverty Line that combines absolute and relative components.

When comparing relative poverty rates around the world, however, it is important to keep in mind that – since average incomes are so far apart – such relative poverty lines relate to very different standards of living in rich and poor countries.

Does the data account for non-market income, such as food grown by subsistence farmers?

Many poor people today, as in the past, rely on subsistence farming rather than a monetary income gained from selling goods or their labor on the market. To take this into account and make a fair comparison of their living standards, the statisticians that produce these figures estimate the monetary value of their home production and add it to their income/expenditure.

Research & Writing

Despite making immense progress against extreme poverty, it is still the reality for every tenth person in the world.

$2.15 a day: the updated International Poverty Line

What does the World Bank’s updated methods mean for our understanding of global poverty?

Global poverty over the long-run

How do we know the history of extreme poverty?

Joe Hasell and Max Roser

Breaking out of the Malthusian trap: How pandemics allow us to understand why our ancestors were stuck in poverty

The short history of global living conditions and why it matters that we know it

Poverty & economic growth.

The economies that are home to the poorest billions of people need to grow if we want global poverty to decline substantially

Global poverty in an unequal world: Who is considered poor in a rich country? And what does this mean for our understanding of global poverty?

What do poor people think about poverty?

More articles on poverty.

Three billion people cannot afford a healthy diet

Hannah Ritchie

Homelessness and poverty in rich countries

Esteban Ortiz-Ospina

OWID Data Collection: Inequality and Poverty

Joe Hasell and Pablo Arriagada

Interactive Charts on Poverty

Official definitions of poverty in different countries are often not directly comparable due to the different ways poverty is measured. For example, countries account for the size of households in different ways in their poverty measures.

The poverty lines shown here are an approximation of national definitions, harmonized to allow for comparisons across countries. For all countries apart from the US, we take the harmonized poverty line calculated by Jolliffe et al. (2022). These lines are calculated as the international dollar figure which, in the World Bank’s Poverty and Inequality Platform (PIP) data, yields the same poverty rate as the officially reported rate using national definitions in a particular year (around 2017).

For the US, Jolliffe et al. (2022) use the OECD’s published poverty rate – which is measured against a relative poverty line of 50% of the median income. This yields a poverty line of $34.79 (measured using 2017 survey data). This however is not the official definition of poverty adopted in the US. We calculated an alternative harmonized figure for the US national poverty using the same method as Jolliffe et al. (2022), but based instead on the official 2019 poverty rate – as reported by the U.S. Census Bureau.

You can see in detail how we calculated this poverty line in this Google Colabs notebook .

Jolliffe, Dean Mitchell, Daniel Gerszon Mahler, Christoph Lakner, Aziz Atamanov, and Samuel Kofi Tetteh Baah. 2022. Assessing the Impact of the 2017 PPPs on the International Poverty Line and Global Poverty. The World Bank. Available to read at the World Bank here .

Because there is no global survey of incomes, researchers need to rely on available national surveys. Such surveys are designed with cross-country comparability in mind, but because the surveys reflect the circumstances and priorities of individual countries at the time of the survey, there are some important differences. In collating this survey data the World Bank takes steps to harmonize it where possible, but comparability issues remain.

One important issue is that, whilst in most high-income countries the surveys capture people’s incomes, in poorer countries these surveys tend to capture people’s consumption. The two concepts are closely related: the income of a household equals their consumption plus any saving, or minus any borrowing or spending out of savings.

To help communicate the latter, the World Bank produces a variable that groups surveys within each individual country into more comparable ‘spells’ (which we include in our data download ). Our Data Explorer provides the option of viewing the data with these breaks in comparability indicated.

The international-$ is a hypothetical currency that results from price adjustments across time and place. It is defined as having the same purchasing power as one US-$ in a given base year – in this case 2017. One int.-$ buys the same quantity of goods and services no matter where or when it is spent. There are many challenges to making such adjustments and they are far from perfect. But in a world where price differences across countries and over time are large it is important to attempt to account for these differences as well as possible, and this is what these adjustments do. Read more in our article From $1.90 to $2.15 a day: the updated International Poverty Line .

​​According to World Bank data, in 1990 there were 2.00 billion people living in poverty, and in 2019 that had fallen to 0.648 billion. The average fall over the 29 years in between is: (2.00 billion – 0.648 billion)/29 = 46.6 million. Dividing by the number of days (29 x 365) gives the average daily fall: (2.00 billion – 0.648 billion)/(29 x 365) = 128,000. (All figures rounded to 3 significant figures).

The projections are generally made on the assumption that incomes or expenditure grow in line with the growth rates observed in national accounts data. You can read more about the interpolation methods used by the World Bank in Chapter 5 of the Poverty and Inequality Platform Methodology Handbook.

We use the figures presented in the World Bank’s Poverty and Shared Prosperity 2022 report. Earlier estimates were also published in Lakner, C., Mahler, D.G., Negre, M. et al. How much does reducing inequality matter for global poverty?. J Econ Inequal (2022). https://doi.org/10.1007/s10888-021-09510-w . Available online here .

Earlier estimates were also published in Lakner, C., Mahler, D.G., Negre, M. et al. How much does reducing inequality matter for global poverty?. J Econ Inequal (2022). https://doi.org/10.1007/s10888-021-09510-w . Available online here .

The figures are taken from a World Bank blog post by Nishant Yonzan, Christoph Lakner and Daniel Gerszon Mahler. The post builds on and updates the estimates published by Lakner et al. (2022).In September 2022, the World Bank changed from using 2011 international-$ to 2017 international-$ in the measurement of global poverty. The International Poverty Line used by the World Bank and the UN to define extreme poverty was accordingly updated from $1.90 a day (in 2011 prices) to $2.15 (in 2017 prices).In order to match up to the projected figures, the extreme poverty estimates shown here relate to a previous release of the World Bank’s data using data expressed in 2011 prices, which vary slightly from the latest data in 2017 prices. You can read more about this change and how it affected the World Bank estimates of poverty in our article From $1.90 to $2.15 a day: the updated International Poverty Line .Lakner, C., Mahler, D.G., Negre, M. et al. How much does reducing inequality matter for global poverty?. J Econ Inequal (2022). https://doi.org/10.1007/s10888-021-09510-w . Available online here .

We use the figures provided in the blog post, which extend the methods presented in Lakner et al. (2022).Lakner, C., Mahler, D.G., Negre, M. et al. How much does reducing inequality matter for global poverty?. J Econ Inequal (2022). https://doi.org/10.1007/s10888-021-09510-w . Available online here .

Shown are those countries with a decline of more than 30 percentage points over a period of 15 years or more.There are a number of ways in which comparability across the different household surveys on which this data is based can be limited. These affect comparisons both across countries and within individual countries over time.The World Bank’s Poverty and Inequality Platform Methodology Handbook provides a good summary of the comparability and data quality issues affecting this data and how it tries to address them. In collating this survey data the World Bank takes a range of steps to harmonize it where possible, but comparability issues remain. To help communicate the latter, the World Bank produces a variable that groups surveys within each individual country into more comparable ‘spells’. Our Data Explorer provides the option of viewing the data with these breaks in comparability indicated.

You can read more about how the World Bank sets these higher poverty lines, as well as the International Poverty Line against which it measures extreme poverty, in our article From $1.90 to $2.15 a day: the updated International Poverty Line . To the three poverty lines adopted officially by the World Bank – $2.15, $3.65 and $6.85 – we add a higher line broadly consistent with definitions of poverty in high income countries. See our article Global poverty in an unequal world: Who is considered poor in a rich country? And what does this mean for our understanding of global poverty?

For details of the methods used to produce the long-run poverty data see, Moatsos, M. (2021). Global extreme poverty: Present and past since 1820. In van Zanden, Rijpma, Malinowski and Mira d’Ercole (eds.) How Was Life? Volume II: New Perspectives on Well-Being and Global Inequality since 1820. Available from the OECD here .

Cite this work

Our articles and data visualizations rely on work from many different people and organizations. When citing this topic page, please also cite the underlying data sources. This topic page can be cited as:

BibTeX citation

Reuse this work freely

All visualizations, data, and code produced by Our World in Data are completely open access under the Creative Commons BY license . You have the permission to use, distribute, and reproduce these in any medium, provided the source and authors are credited.

The data produced by third parties and made available by Our World in Data is subject to the license terms from the original third-party authors. We will always indicate the original source of the data in our documentation, so you should always check the license of any such third-party data before use and redistribution.

All of our charts can be embedded in any site.

Our World in Data is free and accessible for everyone.

Help us do this work by making a donation.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings

Preview improvements coming to the PMC website in October 2024. Learn More or Try it out now .

• Advanced Search
• Journal List

The Social Consequences of Poverty: An Empirical Test on Longitudinal Data

Carina mood.

Institute for Futures Studies, Box 591, 101 31 Stockholm, Sweden

Swedish Institute for Social Research (SOFI), Stockholm University, Stockholm, Sweden

Jan O. Jonsson

Nuffield College, OX1 1NF Oxford, England, UK

Poverty is commonly defined as a lack of economic resources that has negative social consequences, but surprisingly little is known about the importance of economic hardship for social outcomes. This article offers an empirical investigation into this issue. We apply panel data methods on longitudinal data from the Swedish Level-of-Living Survey 2000 and 2010 (n = 3089) to study whether poverty affects four social outcomes—close social relations (social support), other social relations (friends and relatives), political participation, and activity in organizations. We also compare these effects across five different poverty indicators. Our main conclusion is that poverty in general has negative effects on social life. It has more harmful effects for relations with friends and relatives than for social support; and more for political participation than organizational activity. The poverty indicator that shows the greatest impact is material deprivation (lack of cash margin), while the most prevalent poverty indicators—absolute income poverty, and especially relative income poverty—appear to have the least effect on social outcomes.

Introduction

According to the most influential definitions, poverty is seen as a lack of economic resources that have negative social consequences—this is in fact a view that dominates current theories of poverty (Townsend 1979 ; Sen 1983 ; UN 1995 ), and also has a long heritage (Smith 1776 /1976). The idea is that even when people have food, clothes, and shelter, economic problems lead to a deterioration of social relations and participation. Being poor is about not being able to partake in society on equal terms with others, and therefore in the long run being excluded by fellow citizens or withdrawing from social and civic life because of a lack of economic resources, typically in combination with the concomitant shame of not being able to live a life like them (e.g., Sen 1983 ). Economic hardship affects the standard of life, consumption patterns, and leisure time activities, and this is directly or indirectly related to the possibility of making or maintaining friends or acquaintances: poverty is revealed by not having appropriate clothes, or a car; by not being able to afford vacation trips, visits to the restaurant, or hosting dinner parties (e.g., Mack and Lansley 1985 ; Callan et al. 1993 )—in short, low incomes prevent the poor from living a life in “decency” (Galbraith 1958 ).

The relational nature of poverty is also central to the social exclusion literature, which puts poverty in a larger perspective of multiple disadvantages and their interrelationships (Hills et al. 2002 , Rodgers et al. 1995 ; Room 1995 ). While there are different definitions of the social exclusion concept, the literature is characterized by a move from distributional to relational concerns (Gore 1995 ) and by an emphasis on the importance of social integration and active participation in public life. The inability of living a decent or “ordinary” social life may in this perspective erode social networks, social relations, and social participation, potentially setting off a downward spiral of misfortune (Paugam 1995 ) reinforcing disadvantages in several domains of life. This perspective on poverty and social exclusion is essentially sociological: the playing field of the private economy is social. It is ultimately about individuals’ relations with other people—not only primary social relations, with kin and friends, but extending to secondary relations reflected by participation in the wider community, such as in organizations and in political life (UN 1995 ).

Despite the fact that the social consequences of limited economic resources are central to modern perspectives on poverty and marginalization, this relation is surprisingly seldom studied empirically. Qualitative research on the poor give interesting examples on how the negative effects of poverty works, and portray the way that economic problems are transformed into social ones (Ridge and Millar 2011 ; Attree 2006 ). Such studies, however, have too small sample sizes to generalize to the population, and they cannot tell us much about the range of the problem. The (relatively few) studies that have addressed the association between poverty and social outcomes on larger scale tend to verify that the poor have worse social relations (Böhnke 2008 ; Jonsson and Östberg 2004 ; Levitas 2006 ), but Barnes et al. ( 2002 ) did not find any noteworthy association between poverty (measured as relative income poverty, using the 60 %-limit) and social relations or social isolation. Dahl et al. ( 2008 ) found no relation between poverty and friendships, but report less participation in civic organizations among the poor. All these studies have however been limited to cross-sectional data or hampered by methodological shortcomings, and therefore have not been able to address the separation of selection effects from potentially causal ones.

Our aim in this study is to make good these omissions. We use longitudinal data from the Swedish Level of Living Surveys (LNU) 2000 and 2010 to study how falling into poverty, or rising from it, is associated with outcomes in terms of primary and secondary social relations, including participation in civil society. These panel data make it possible to generalize the results to the Swedish adult population (19–65 in 2000; 29–75 in 2010), to address the issue of causality, and to estimate how strong the relation between economic vulnerability and social outcomes is. Because the data provide us with the possibility of measuring poverty in several ways, we are also able to address the question using different—alternative or complementary—indicators. Poverty is measured as economic deprivation (lack of cash margin, self-reported economic problems), income poverty (absolute and relative), and long-term poverty, respectively. The primary, or core, social outcomes are indicated by having social support if needed, and by social relations with friends and relatives. We expand our analysis to secondary, or fringe, social outcomes in terms of participation in social life at large, such as in civil society: our indicators here include the participation in organizations and in political life.

Different Dimensions/Definitions of Poverty

In modern welfare states, the normal take on the issue of poverty is to regard it as the relative lack of economic resources, that is, to define the poor in relation to their fellow citizens in the same country at the same time. Three approaches dominate the scholarly literature today. The first takes as a point of departure the income deemed necessary for living a life on par with others, or that makes possible an “acceptable” living standard—defined as the goods and services judged necessary, often on the basis of consumer or household budget studies. This usage of a poverty threshold is often (somewhat confusingly) called absolute income poverty , and is most common in North America (cf. Corak 2006 for a review), although most countries have poverty lines defined for different kinds of social benefits. In Europe and in the OECD, the convention is instead to use versions of relative income poverty , defining as poor those whose incomes fall well behind the median income in the country in question (European Union using 60 % and OECD 50 % of the median as the threshold). As an alternative to using purchasing power (as in the “absolute” measure), this relative measure defines poverty by income inequality in the bottom half of the income distribution (Atkinson et al. 2002 ; OECD 2008 ).

The third approach argues that income measures are too indirect; poverty should instead be indicated directly by the lack of consumer products and services that are necessary for an acceptable living standard (Mack and Lansley 1985 ; Ringen 1988 ; Townsend 1979 ). This approach often involves listing a number of possessions and conditions, such as having a car, washing machine, modern kitchen; and being able to dine out sometimes, to have the home adequately heated and mended, to have sufficient insurances, and so on. An elaborate version includes information on what people in general see as necessities, what is often termed “consensual” poverty (e.g., Mack and Lansley 1985 ; Gordon et al. 2000 ; Halleröd 1995 ; van den Bosch 2001 ). Other direct indicators include the ability to cover unforeseen costs (cash margin) and subjective definitions of poverty (e.g., van den Bosch 2001 ). The direct approach to poverty has gained in popularity and measures of economic/material deprivation and consensual poverty are used in several recent and contemporary comparative surveys such as ECHP (Whelan et al. 2003 ) and EU-SILC (e.g., UNICEF 2012 ; Nolan and Whelan 2011 ).

It is often pointed out that, due to the often quite volatile income careers of households, the majority of poverty episodes are short term and the group that is identified as poor in the cross-section therefore tends to be rather diluted (Bane and Ellwood 1986 ; Duncan et al. 1993 ). Those who suffer most from the downsides of poverty are, it could be argued, instead the long-term, persistent, or chronically poor, and there is empirical evidence that those who experience more years in poverty also are more deprived of a “common lifestyle” (Whelan et al. 2003 ). Poverty persistence has been defined in several ways, such as having spent a given number of years below a poverty threshold, or having an average income over a number of years that falls under the poverty line (e.g., Duncan and Rodgers 1991 ; Rodgers and Rodgers 1993 ). The persistently poor can only be detected with any precision in longitudinal studies, and typically on the basis of low incomes, as data covering repeated measures of material deprivation are uncommon.

For the purposes of this study, it is not essential to nominate the best or most appropriate poverty measure. The measures outlined above, while each having some disadvantage, all provide plausible theoretical grounds for predicting negative social outcomes. Low incomes, either in “absolute” or relative terms, may inhibit social activities and participation because these are costly (e.g., having decent housing, needing a car, paying membership fees, entrance tickets, or new clothes). Economic deprivation, often indicated by items or habits that are directly relevant to social life, is also a valid representation of a lack of resources. Lastly, to be in long-term poverty is no doubt a worse condition than being in shorter-term poverty.

It is worth underlining that we see different measures of poverty as relevant indicators despite the fact that the overlap between them often is surprisingly small (Bradshaw and Finch 2003 ). The lack of overlap is not necessarily a problem, as different people may have different configurations of economic problems but share in common many of the experiences of poverty—experiences, we argue, that are (in theory at least) all likely to lead to adverse social outcomes. Whether this is the case or not is one of the questions that we address, but if previous studies on child poverty are of any guidance, different definitions of poverty may show surprisingly similar associations with a number of outcomes (Jonsson and Östberg 2004 ).

What are the Likely Social Consequences of Poverty?

We have concluded that poverty is, according to most influential poverty definitions, manifested in the social sphere. This connects with the idea of Veblen ( 1899 ) of the relation between consumption and social status. What you buy and consume—clothes, furniture, vacation trips—in part define who you are, which group you aspire to belong to, and what view others will have of you. Inclusion into and exclusion from status groups and social circles are, in this view, dependent on economic resources as reflected in consumption patterns. While Veblen was mostly concerned about the rich and their conspicuous consumption, it is not difficult to transfer these ideas to the less fortunate: the poor are under risk of exclusion, of losing their social status and identity, and perhaps also, therefore, their friends. It is however likely that this is a process that differs according to outcome, with an unknown time-lag.

If, as outlined above, we can speak of primary and secondary social consequences, the former should include socializing with friends, but also more intimate relations. Our conjecture is that the closer the relation, the less affected is it by poverty, simply because intimate social bonds are characterized by more unconditional personal relations, typically not requiring costs to uphold.

When it comes to the secondary social consequences, we move outside the realm of closer interpersonal relations to acquaintances and the wider social network, and to the (sometimes relatively anonymous) participation in civil or political life. This dimension of poverty lies at the heart of the social exclusion perspective, which strongly emphasizes the broader issues of societal participation and civic engagement, vital to democratic societies. It is also reflected in the United Nation’s definition, following the Copenhagen summit in 1995, where “overall poverty” in addition to lack of economic resources is said to be “…characterized by lack of participation in decision-making and in civil, social, and cultural life” (UN 1995 , p. 57). Poverty may bring about secondary social consequences because such participation is costly—as in the examples of travel, need for special equipment, or membership fees—but also because of psychological mechanisms, such as lowered self-esteem triggering disbelief in civic and political activities, and a general passivity leading to decreased organizational and social activities overall. If processes like these exist there is a risk of a “downward spiral of social exclusion” where unemployment leads to poverty and social isolation, which in turn reduce the chances of re-gaining a footing in the labour market (Paugam 1995 ).

What theories of poverty and social exclusion postulate is, in conclusion, that both what we have called primary and secondary social relations will be negatively affected by economic hardship—the latter supposedly more than the former. Our strategy in the following is to test this basic hypothesis by applying multivariate panel-data analyses on longitudinal data. In this way, we believe that we can come further than previous studies towards estimating causal effects, although, as is the case in social sciences, the causal relation must remain preliminary due to the nature of observational data.

Data and Definitions

We use the two most recent waves of the Swedish Level-of-living Survey, conducted in 2000 and 2010 on random (1/1000) samples of adult Swedes, aged 18–75. 1 The attrition rate is low, with 84 % of panel respondents remaining from 2000 to 2010. This is one of the few data sets from which we can get over-time measures of both poverty and social outcomes for a panel that is representative of the adult population (at the first time point, t 0 )—in addition, there is annual income information from register data between the waves. The panel feature obviously restricts the age-groups slightly (ages 19–65 in 2000; 29–75 in 2010), the final number of analyzed cases being between 2995 and 3144, depending on the number of missing cases on the respective poverty measure and social outcome variable. For ease of interpretation and comparison of effect sizes, we have constructed all social outcome variables and poverty variables to be dichotomous (0/1). 2

In constructing poverty variables, we must balance theoretical validity with the need to have group sizes large enough for statistical analysis. For example, we expand the absolute poverty measure to include those who received social assistance any time during the year. As social assistance recipients receive this benefit based on having an income below a poverty line that is similar to the one we use, this seems justifiable. In other cases, however, group sizes are small but we find no theoretically reasonable way of making the variables more inclusive, meaning that some analyses cannot be carried out in full detail.

Our income poverty measures are based on register data and are thus free from recall error or misreporting, but—as the proponents of deprivation measures point out—income poverty measures are indirect measures of hardship. The deprivation measure is more direct, but self-reporting always carries a risk of subjectivity in the assessment. To the extent that changes in one’s judgment of the economic situation depend on changes in non-economic factors that are also related to social relations, the deprivation measure will give upwardly biased estimates. 3 As there is no general agreement about whether income or deprivation definitions are superior, our use of several definitions is a strength because the results will give an overall picture that is not sensitive to potential limitations in any one measure. In addition, we are able to see whether results vary systematically across commonly used definitions.

Poverty Measures

• Cash margin whether the respondent can raise a given sum of money in a week, if necessary (in 2000, the sum was 12,000 SEK; in 2010, 14,000 SEK, the latter sum corresponding to approximately 1600 Euro, 2200 USD, or 1400 GBP in 2013 currency rates). For those who answer in the affirmative, there is a follow-up question of how this can be done: by (a) own/household resources, (b) borrowing.
• Economic crisis Those who claim that they have had problems meeting costs for rent, food, bills, etc. during the last 12 months (responded “yes” to a yes/no alternative).
• Absolute poverty is defined as either (a) having a disposable family income below a poverty threshold or (b) receiving social assistance, both assessed in 1999 (for the survey 2000) or 2009 (for the survey 2010). The poverty line varies by family type/composition according to a commonly used calculation of household necessities (Jansson 2000 ). This “basket” of goods and services is intended to define an acceptable living standard, and was originally constructed for calculating an income threshold for social assistance, with addition of estimated costs for housing and transport. The threshold is adjusted for changes in the Consumer Price Index, using 2010 as the base year. In order to get analyzable group sizes, we classify anyone with an income below 1.25 times this threshold as poor. Self-employed are excluded because their nominal incomes are often a poor indicator of their economic standard.
• Deprived and income poor A combination of the indicator of economic deprivation and the indicator of absolute poverty. The poor are defined as those who are economically deprived and in addition are either absolute income-poor or have had social assistance some time during the last calendar year.
• Long - term poor are defined as those interviewed in 2010 (2000) who had an equivalized disposable income that fell below the 1.25 absolute poverty threshold (excluding self-employed) or who received social assistance in 2009 (1999), and who were in this situation for at least two of the years 2000–2008 (1990–1998). The long-term poor (coded 1) are contrasted to the non-poor (coded 0), excluding the short-term poor (coded missing) in order to distinguish whether long-term poverty is particularly detrimental (as compared to absolute poverty in general).
• Relative poverty is defined, according to the EU standard, as having a disposable equivalized income that is lower than 60 % of the median income in Sweden the year in question (EU 2005). 4 As for absolute poverty, this variable is based on incomes the year prior to the survey year. Self-employed are excluded.

Social and Participation Outcomes

Primary (core) social relations.

• Social support The value 1 (has support) is given to those who have answered in the positive to three questions about whether one has a close friend who can help if one (a) gets sick, (b) needs someone to talk to about troubles, or (c) needs company. Those who lack support in at least one of these respects are coded 0 (lack of support).
• Frequent social relations This variable is based on four questions about how often one meets (a) relatives and (b) friends, either (i) at ones’ home or (ii) at the home of those one meets, with the response set being “yes, often”, “sometimes”, and “no, never”. Respondents are defined as having frequent relations (1) if they have at least one “often” of the four possible and no “never”, 5 and 0 otherwise.

Secondary (fringe) Social Relations/Participation

• Political participation : Coded 1 (yes) if one during the last 12 months actively participated (held an elected position or was at a meeting) in a trade union or a political party, and 0 (no) otherwise. 6
• Organizational activity : Coded 1 (yes) if one is a member of an organization and actively participate in its activities at least once in a year, and 0 (no) otherwise.

Control Variables

• Age (in years)
• Educational qualifications in 2010 (five levels according to a standard schema used by Statistics Sweden (1985), entered as dummy variables)
• Civil status distinguishes between single and cohabiting/married persons, and is used as a time-varying covariate (TVC) where we register any changes from couple to single and vice versa.
• Immigrant origin is coded 1 if both parents were born in any country outside Sweden, 0 otherwise.
• Labour market status is also used as a TVC, with four values indicating labour market participation (yes/no) in 2000 and 2010, respectively.
• Global self - rated health in 2000, with three response alternatives: Good, bad, or in between. 7

Table  1 shows descriptive statistics for the 2 years we study, 2000 and 2010 (percentages in the upper panel; averages, standard deviations, max and min values in the lower panel). Recall that the sample is longitudinal with the same respondents appearing in both years. This means, naturally, that the sample ages 10 years between the waves, the upper age limit being pushed up from 65 to 75. Both the change over years and the ageing of the sample have repercussions for their conditions: somewhat more have poor health, for example, fewer lack social support but more lack frequent social relations, and more are single in 2010 (where widows are a growing category). The group has however improved their economic conditions, with a sizeable reduction in poverty rates. Most of the changes are in fact period effects, and it is particularly obvious for the change in poverty—in 2000 people still suffered from the deep recession in Sweden that begun in 1991 and started to turn in 1996/97 (Jonsson et al. 2010 ), while the most recent international recession (starting in 2008/09) did not affect Sweden that much.

Table 1

Descriptive statistics of dependent and independent variables in the LNU panel

N for variables used as change variables pertains to non-missing observations in both 2000 and 2010

The overall decrease in poverty masks changes that our respondents experienced between 2000 and 2010: Table  2 reveals these for the measure of economic deprivation, showing the outflow (row) percentages and the total percentages (and the number of respondents in parentheses). It is evident that there was quite a lot of mobility out of poverty between the years (61 % left), but also a very strong relative risk of being found in poverty in 2010 among those who were poor in 2000 (39 vs. 5 % of those who were non-poor in 2000). Of all our respondents, the most common situation was to be non-poor both years (81 %), while few were poor on both occasions (6 %). Table  2 also demonstrates some small cell numbers: 13.3 % of the panel (9.4 % + 3.9 %), or a good 400 cases, changed poverty status, and these cases are crucial for identifying our models. As in many panel studies based on survey data, this will inevitably lead to some problems with large standard errors and difficulties in arriving at statistically significant and precise estimates; but to preview the findings, our results are surprisingly consistent all the same.

Table 2

Mobility in poverty (measured as economic deprivation) in Sweden between 2000 and 2010

Outflow percentage (row %), total percentage, and number of cases (in parentheses). LNU panel 2000–2010

We begin with showing descriptive results of how poverty is associated with our outcome variables, using the economic deprivation measure of poverty. 8 Figure  1 confirms that those who are poor have worse social relationships and participate less in political life and in organizations. Poverty is thus connected with both primary and secondary social relations.

The relation between poverty (measured as economic deprivation) and social relations/participation in Sweden, LNU 2010. N = 5271

The descriptive picture in Fig.  1 does not tell us anything about the causal nature of the relation between poverty and social outcomes, only that such a relation exists, and that it is in the predicted direction: poor people have weaker social relations, less support, and lower levels of political and civic participation. Our task now is to apply more stringent statistical models to test whether the relation we have uncovered is likely to be of a causal nature. This means that we must try to rid the association of both the risk for reverse causality—that, for example, a weaker social network leads to poverty—and the risk that there is a common underlying cause of both poverty and social outcomes, such as poor health or singlehood.

The Change Model

First, as we have panel data, we can study the difference in change across two time-points T (called t 0 and t 1 , respectively) in an outcome variable (e.g., social relations), between groups (i.e. those who changed poverty status versus those who did not). The respondents are assigned to either of these groups on the grounds of entering or leaving poverty; in the first case, one group is non-poor at t 0 but experiences poverty at t 1 , and the change in this group is compared to the group consisting of those who are non-poor both at t 0 and t 1 . The question in focus then is: Do social relations in the group entering poverty worsen in relation to the corresponding change in social relations in the group who remains non-poor? Because we have symmetric hypotheses of the effect of poverty on social outcomes—assuming leaving poverty has positive consequences similar to the negative consequences of entering poverty—we also study whether those who exit poverty improve their social outcomes as compared to those remaining poor. We ask, that is, not only what damage falling into poverty might have for social outcomes, but also what “social gains” could be expected for someone who climbs out of poverty.

Thus, in our analyses we use two different “change groups”, poverty leavers and poverty entrants , and two “comparison groups”, constantly poor and never poor , respectively. 9 The setup comparing the change in social outcomes for those who change poverty status and those who do not is analogous to a so-called difference-in-difference design, but as the allocation of respondents to comparison groups and change groups in our data cannot be assumed to be random (as with control groups and treatment groups in experimental designs), we take further measures to approach causal interpretations.

Accounting for the Starting Value of the Dependent Variable

An important indication of the non-randomness of the allocation to the change and comparison groups is that their average values of the social outcomes (i.e. the dependent variable) at t 0 differ systematically: Those who become poor between 2000 and 2010 have on average worse social outcomes already in 2000 than those who stay out of poverty. Similarly, those who stay in poverty both years have on average worse social outcomes than those who have exited poverty in 2010. In order to further reduce the impact of unobserved variables, we therefore make all comparisons of changes in social outcomes between t 0 and t 1 for fixed t 0 values of both social outcome and poverty status.

As we use dichotomous outcome variables, we get eight combinations of poverty and outcome states (2 × 2 × 2 = 8), and four direct strategic comparisons:

• Poverty leavers versus constantly poor, positive social outcome in 2000 , showing if those who exit poverty have a higher chance of maintaining the positive social outcome than those who stay in poverty
• Poverty leavers versus constantly poor, negative social outcome in 2000 , showing if those who exit poverty have a higher chance of improvement in the social outcome than those who stay in poverty
• Poverty entrants versus never poor, positive social outcome in 2000 , showing if those who enter poverty have a higher risk of deterioration in the social outcome than those who stay out of poverty, and
• Poverty entrants versus never poor, negative social outcome in 2000 , showing if those who enter poverty have a lower chance of improvement in the social outcome.

Thus, we hold the initial social situation and poverty status fixed, letting only the poverty in 2010 vary. 10 The analytical strategy is set out in Table  3 , showing estimates of the probability to have frequent social relations in 2010, for poverty defined (as in Table  2 and Fig.  1 above) as economic deprivation.

Table 3

Per cent with frequent social relations in “comparison” and “change” groups in 2000 and 2010, according to initial value on social relations in 2000 and poverty (measured as economic deprivation) in 2000 and 2010

LNU panel 2000–2010. N = 3083

The figures in Table  3 should be read like this: 0.59 in the upper left cell means that among those who were poor neither in 2000 nor in 2010 (“never poor”, or 0–0), and who had non-frequent social relations to begin with, 59 % had frequent social relations in 2010. Among those never poor who instead started out with more frequent social relations, 90 per cent had frequent social relations in 2010. This difference (59 vs. 90) tells us either that the initial conditions were important (weak social relations can be inherently difficult to improve) or that there is heterogeneity within the group of never poor people, such as some having (to us perhaps unobserved) characteristics that support relation building while others have not.

Because our strategy is to condition on the initial situation in order to minimize the impact of initial conditions and unobserved heterogeneity, we focus on the comparisons across columns. If we follow each column downwards, that is, for a given initial social outcome (weak or not weak social relations, respectively) it is apparent that the outcome is worse for the “poverty entrants” in comparison with the “never poor” (upper three lines). Comparing the change group [those who became poor (0–1)] with the comparison group [never poor (0–0)] for those who started out with weak social relations (left column), the estimated probability of frequent social relations in 2010 is 7 % points lower for those who became poor. Among those who started out with frequent relations, those who became poor have a 17 % points lower probability of frequent relations in 2010 than those who stayed out of poverty.

If we move down Table  3 , to the three bottom lines, the change and comparison groups are now different. The comparison group is the “constantly poor” (1–1), and the change group are “poverty leavers” (1–0). Again following the columns downwards, we can see that the change group improved their social relations in comparison with the constantly poor; and this is true whether they started out with weak social relations or not. In fact, the chance of improvement for those who started off with non-frequent social relations is the most noteworthy, being 33 % units higher for those who escaped poverty than for those who did not. In sum, Table  3 suggests that becoming poor appears to be bad for social relations whereas escaping poverty is beneficial.

Expanding the Model

The model exemplified in Table  3 is a panel model that studies change across time within the same individuals, conditioning on their initial state. It does away with time-constant effects of observed and unobserved respondent characteristics, and although this is far superior to a cross-sectional model (such as the one underlying Fig.  1 ) there are still threats to causal interpretations. It is possible (if probably unusual) that permanent characteristics may trigger a change over time in both the dependent and independent variables; or, put in another way, whether a person stays in or exits poverty may be partly caused by a variable that also predicts change in the outcome (what is sometimes referred to as a violation of the “common trend assumption”). In our case, we can for example imagine that health problems in 2000 can affect who becomes poor in 2010, at t 1 , and that the same health problems can lead to a deterioration of social relations between 2000 and 2010, so even conditioning on the social relations at t 0 will not be enough. This we handle by adding control variables, attempting to condition the comparison of poor and non-poor also on sex, age, highest level of education (in 2010), immigrant status, and health (in 2000). 11

Given the set-up of our data—with 10 years between the two data-points and with no information on the precise time ordering of poverty and social outcomes at t 1 , the model can be further improved by including change in some of the control variables. It is possible, for example, that a non-poor and married respondent in 2000 divorced before 2010, triggering both poverty and reduced social relations at the time of the interview in 2010. 12 There are two major events that in this way may bias our results, divorce/separation and unemployment (because each can lead to poverty, and possibly also affect social outcomes). We handle this by controlling for variables combining civil status and unemployment in 2000 as well as in 2010. To the extent that these factors are a consequence of becoming poor, there is a risk of biasing our estimates downwards (e.g., if becoming poor increases the risk of divorce). However, as there is no way to distinguish empirically whether control variables (divorce, unemployment) or poverty changed first we prefer to report conservative estimates. 13

Throughout, we use logistic regression to estimate our models (one model for each social outcome and poverty definition). We create a dummy variable for each of the combinations of poverty in 2000, poverty in 2010 and the social outcome in 2000, and alternate the reference category in order to get the four strategic comparisons described above. Coefficients do thus express the distance between the relevant change and comparison groups. The coefficients reported are average marginal effects (AME) for a one-unit change in the respective poverty variable (i.e. going from non-poor to poor and vice versa), which are straightforwardly interpretable as percentage unit differences and (unlike odds ratios or log odds ratios) comparable across models and outcomes (Mood 2010 ).

Regression Results

As detailed above, we use changes over time in poverty and social outcomes to estimate the effects of interest. The effect of poverty is allowed to be heterogeneous, and is assessed through four comparisons of the social outcome in 2010 (Y 1 ):

• Those entering poverty relative to those in constant non-poverty (P 01  = 0,1 vs. P 01  = 0,0) when both have favourable social outcomes at t 0 (Y 0  = 1)
• Those exiting poverty relative to those in constant poverty (P 01  = 1,0 vs. P 01  = 1,1) when both have favourable social outcomes at t 0 (Y 0  = 1)
• Those entering poverty relative to those in constant non-poverty (P 01  = 0,1 vs. P 01  = 0,0) when both have non-favourable social outcomes at t 0 (Y 0  = 0)
• Those exiting poverty relative to those in constant poverty (P 01  = 1,0 vs. P 01  = 1,1) when both have non-favourable social outcomes at t 0 (Y 0  = 0)

Poverty is a rare outcome, and as noted above it is particularly uncommon to enter poverty between 2000 and 2010 because of the improving macro-economic situation. Some of the social outcomes were also rare in 2000. This unfortunately means that in some comparisons we have cell frequencies that are prohibitively small, and we have chosen to exclude all comparisons involving cells where N < 20.

The regression results are displayed in Table  4 . To understand how the estimates come to be, consider the four in the upper left part of the Table (0.330, 0.138, −0.175 and −0.065), reflecting the effect of poverty, measured as economic deprivation, on the probability of having frequent social relations. Because these estimates are all derived from a regression without any controls, they are identical (apart from using three decimal places) to the percentage comparisons in Table  3 (0.33, 0.14, −0.17, −0.07), and can be straightforwardly interpreted as average differences in the probability of the outcome in question. From Table  4 it is clear that the three first differences are all statistically significant, whereas the estimate −0.07 is not (primarily because those who entered poverty in 2010 and had infrequent social relations in 2000 is a small group, N = 25).

Table 4

Average marginal effects (from logistic regression) of five types of poverty (1–5) on four social outcomes (A-D) comparing those with different poverty statuses in 2000 and 2010 and conditioning on the starting value of the social outcome (in 2000)

Right columns control for sex, education, age, immigrant status, health in 2000, civil status change between 2000 and 2010, and unemployment change between 2000 and 2010. P values in parentheses. Excluded estimates involve variable categories with N < 20. Shaded cells are in hypothesized direction, bold estimates are statistically significant ( P  < 0.05). N in regressions: 1A: 3075; 1B: 3073; 1C: 3075; 1D: 3069; 2A: 3144; 2B: 3137; 2C: 3144; 2D: 3130; 3A: 3074, 3B: 3072; 3C: 3074; 3D: 3068; 4A: 2995; 4B: 2988; 4C: 2995; 4D: 2981; 5A: 3128; 5B: 3121; 5C: 3128; 5D: 3114

In the column to the right, we can see what difference the controls make: the estimates are reduced, but not substantially so, and the three first differences are still statistically significant.

The estimates for each social outcome, reflecting the four comparisons described above, support the hypothesis of poverty affecting social relations negatively (note that the signs of the estimates should differ in order to do so, the upper two being positive as they reflect an effect of the exit from poverty, and the lower two being negative as they reflect an effect of entering poverty). We have indicated support for the hypothesis in Table  4 by shading the estimates and standard errors for estimates that go in the predicted direction.

Following the first two columns down, we can see that there is mostly support for the hypothesis of a negative effect of poverty, but when controlling for other variables, the effects on social support are not impressive. In fact, if we concentrate on each social outcome (i.e., row-wise), one conclusion is that, when controlling for confounders, there are rather small effects of poverty on the probability of having access to social support. The opposite is true for political participation, where the consistency in the estimated effects of poverty is striking.

If we instead follow the columns, we ask whether any of the definitions of poverty is a better predictor of social outcomes than the others. The measure of economic deprivation appears to be the most stable one, followed by absolute poverty and the combined deprivation/absolute poverty variable. 14 The relative poverty measure is less able to predict social outcomes: in many instances it even has the non-expected sign. Interestingly, long-term poverty (as measured here) does not appear to have more severe negative consequences than absolute poverty in general.

Because some of our comparison groups are small, it is difficult to get high precision in the estimates, efficiency being a concern particularly in view of the set of control variables in Table  4 . Only 14 out of 62 estimates in models with controls are significant and in the right direction. Nonetheless, with 52 out of 62 estimates in these models having the expected sign, we believe that the hypothesis of a negative effect of poverty on social outcomes receives quite strong support.

Although control variables are not shown in the table, one thing should be noted about them: The reduction of coefficients when including control variables is almost exclusively driven by changes in civil status. 15 The time constant characteristics that are included are cross-sectionally related to both poverty and social outcomes, but they have only minor impacts on the estimated effects of poverty. This suggests that the conditioning on prior values of the dependent and independent variables eliminates much time invariant heterogeneity, which increases the credibility of estimates.

Conclusions

We set out to test a fundamental, but rarely questioned assumption in dominating definitions of poverty: whether shortage of economic resources has negative consequences for social relations and participation. By using longitudinal data from the Swedish Level-of-living Surveys 2000 and 2010, including repeated measures of poverty (according to several commonly used definitions) and four social outcome variables, we are able to come further than previous studies in estimating the relation between poverty and social outcomes: Our main conclusion is that there appears to be a causal relation between them.

Panel models suggest that falling into poverty increases the risk of weakening social relations and decreasing (civic and political) participation. Climbing out of poverty tends to have the opposite effects, a result that strengthens the interpretation of causality. The sample is too small to estimate the effect sizes with any precision, yet they appear to be substantial, with statistically significant estimates ranging between 5 and 21 % units.

While these findings are disquieting insofar as poverty goes, our results also suggest two more positive results. First, the negative effects of poverty appear to be reversible: once the private economy recovers, social outcomes improve. Secondly, the negative consequences are less for the closest social relations, whether there is someone there in cases of need (sickness, personal problems, etc.). This is in line with an interpretation of such close relations being unconditional: our nearest and dearest tend to hang on to us also in times of financial troubles, which may bolster risks for social isolation and psychological ill-being,

Our finding of negative effects of poverty on civic and political participation relates to the fears of a “downward spiral of social exclusion”, as there is a risk that the loss of less intimate social relations shrinks social networks and decreases the available social capital in terms of contacts that can be important for outcomes such as finding a job (e.g., Lin 2001 ; Granovetter 1974 ). However, Gallie et al. ( 2003 ) found no evidence for any strong impact of social isolation on unemployment, suggesting that the negative effects on social outcomes that we observe are unlikely to lead to self-reinforcement of poverty. Nevertheless, social relations are of course important outcomes in their own right, so if they are negatively affected by poverty it matters regardless of whether social relations in turn are important for other outcomes. Effects on political and civic participation are also relevant in themselves beyond individuals’ wellbeing, as they suggest a potentially democratic problem where poor have less of a voice and less influence on society than others.

Our results show the merits of our approach, to study the relation between poverty and social outcomes longitudinally. The fact that the poor have worse social relations and lower participation is partly because of selection. This may be because the socially isolated, or those with a weaker social network, more easily fall into poverty; or it can be because of a common denominator, such as poor health or social problems. But once we have stripped the analysis of such selection effects, we also find what is likely to be a causal relation between poverty and social relations. However, this effect of poverty on social outcomes, in turn, varies between different definitions of poverty. Here it appears that economic deprivation, primarily indicated by the ability of raising money with short notice, is the strongest predictor of social outcomes. Income poverty, whether in absolute or (particularly) relative terms, are weaker predictors of social outcomes, which is interesting as they are the two most common indicators of poverty in existing research.

Even if we are fortunate to have panel data at our disposal, there are limitations in our analyses that render our conclusions tentative. One is that we do not have a random allocation to the comparison groups at t 0 ; another that there is a 10-year span between the waves that we analyze, and both poverty and social outcomes may vary across this time-span. We have been able to address these problems by conditioning on the outcome at t 0 and by controlling for confounders, but in order to perform more rigorous tests future research would benefit from data with a more detailed temporal structure, and preferably with an experimental or at least quasi-experimental design.

Finally, our analyses concern Sweden, and given the position as an active welfare state with a low degree of inequality and low poverty rates, one can ask whether the results are valid also for other comparable countries. While both the level of poverty and the pattern of social relations differ between countries (for policy or cultural reasons), we believe that the mechanisms linking poverty and social outcomes are of a quite general kind, especially as the “costs for social participation” can be expected to be relative to the general wealth of a country—however, until comparative longitudinal data become available, this must remain a hypothesis for future research.

1 http://www.sofi.su.se/english/2.17851/research/three-research-departments/lnu-level-of-living .

2 We have tested various alternative codings and the overall pattern of results in terms of e.g., direction of effects and differences across poverty definitions are similar, but more difficult to present in an accessible way.

3 Our deprivation questions are however designed to reduce the impact of subjectivity by asking, e.g., about getting a specified sum within a specified time (see below).

4 In the equivalence scale, the first adult gets a weight of one, the second of 0.6, and each child gets a weight of 0.5.

5 We have also tried using single indicators (either a/b or i/ii) without detecting any meaningful difference between them. One would perhaps have assumed that poverty would be more consequential for having others over to one’s own place, but the absence of support for this can perhaps be understood in light of the strong social norm of reciprocity in social relations.

6 We have refrained from using information on voting and membership in trade unions and political parties, because these indicators do not capture the active, social nature of civic engagement to the same extent as participation in meetings and the holding of positions.

7 We have also estimated models with a more extensive health variable, a s ymptom index , which sums responses to 47 questions about self-reported health symptoms. However, this variable has virtually zero effects once global self-rated health is controlled, and does not lead to any substantive differences in other estimates. Adding the global health measure and the symptom index as TVC had no effect either.

8 Using the other indicators of poverty yields very similar results, although for some of those the difference between poor and non-poor is smaller.

9 We call these comparison groups ”never poor” and ”constantly poor” for expository purposes, although their poverty status pertains only to the years 2000 and 2010, i.e., without information on the years in between.

10 With this design we allow different effects of poverty on improvement versus deterioration of the social outcome. We have also estimated models with a lagged dependent variable, which constrains the effects of poverty changes to be of the same size for deterioration as for improvement of the social outcome. Conclusions from that analysis are roughly a weighted average of the estimates for deterioration and improvement that we report. As our analyses suggest that effects of poverty differ in size depending on the value of the lagged dependent variable (the social outcome) our current specification gives a more adequate representation of the process.

11 We have also tested models with a wider range of controls for, e.g., economic and social background (i.e. characteristics of the respondent’s parents), geography, detailed family type and a more detailed health variable, but none of these had any impact on the estimated poverty effects.

12 It is also possible that we register reverse causality, namely if worsening social outcomes that occur after t 0 lead to poverty at t 1 . This situation is almost inevitable when using panel data with no clear temporal ordering of events occurring between waves. However, reverse causality strikes us, in this case, as theoretically implausible.

13 We have also estimated models controlling for changes in health, which did not change the results.

14 If respondents’ judgments of the deprivation questions (access to cash margin and ability to pay rent, food, bills etc.) change due to non-economic factors that are related to changes in social relations, the better predictive capacity of the deprivation measure may be caused by a larger bias in this measure than in the (register-based) income measures.

15 As mentioned above, this variable may to some extent be endogenous (i.e., a mediator of the poverty effect rather than a confounder), in which case we get a downward bias of estimates.

Contributor Information

Carina Mood, Phone: +44-8-402 12 22, Email: [email protected] .

Jan O. Jonsson, Phone: +44 1865 278513, Email: [email protected] .

• Atkinson AB, Cantillon B, Marlier E, Nolan B. Social indicators: The EU and social inclusion. Oxford: Oxford University Press; 2002. [ Google Scholar ]
• Attree P. The social costs of child poverty: A systematic review of the qualitative evidence. Children and Society. 2006; 20 :54–66. [ Google Scholar ]
• Bane MJ, Ellwood DT. Slipping into and out of Poverty: The Dynamics of Spells. Journal of Human Resources. 1986; 21 :1–23. doi: 10.2307/145955. [ CrossRef ] [ Google Scholar ]
• Barnes M, Heady C, Middleton S, Millar J, Papadopoulos F, Room G, Tsakloglou P. Poverty and social exclusion in Europe. Cheltenham: Edward Elgar; 2002. [ Google Scholar ]
• Böhnke P. Are the poor socially integrated? The link between poverty and social support in different welfare regimes. Journal of European Social Policy. 2008; 18 :133–150. doi: 10.1177/0958928707087590. [ CrossRef ] [ Google Scholar ]
• Bradshaw J, Finch N. Overlaps in dimensions of poverty. Journal of Social Policy. 2003; 32 :513–525. doi: 10.1017/S004727940300713X. [ CrossRef ] [ Google Scholar ]
• Callan T, Nolan B, Whelan CT. Resources, deprivation, and the measurement of poverty. Journal of Social Policy. 1993; 22 :141–172. doi: 10.1017/S0047279400019280. [ CrossRef ] [ Google Scholar ]
• Corak M. Principles and practicalities for measuring child poverty in the rich countries. International Social Security Review. 2006; 59 :3–36. doi: 10.1111/j.1468-246X.2006.00237.x. [ CrossRef ] [ Google Scholar ]
• Dahl E, Flotten T, Lorentzen T. Poverty dynamics and social exclusion: An analysis of Norwegian panel data. Journal of Social Policy. 2008; 37 :231–249. doi: 10.1017/S0047279407001729. [ CrossRef ] [ Google Scholar ]
• Duncan GJ, Gustafsson B, Hauser R, Schmauss G, Messinger H, Muffels R, Nolan B, Ray J-C. Poverty dynamics in eight countries. Journal of Population Economics. 1993; 6 :215–234. doi: 10.1007/BF00163068. [ CrossRef ] [ Google Scholar ]
• Duncan GJ, Rodgers W. Has children’s poverty become more persistent? American Sociological Review. 1991; 56 :538–550. doi: 10.2307/2096273. [ CrossRef ] [ Google Scholar ]
• Galbraith J. The affluent society. Boston: Houghton-Mifflin; 1958. [ Google Scholar ]
• Gallie D, Paugam S, Jacobs S. Unemployment, poverty and social isolation: Is there a vicious cycle of social exclusion? European Societies. 2003; 5 :1–32. doi: 10.1080/1461669032000057668. [ CrossRef ] [ Google Scholar ]
• Gordon D, Adelman L, Ashworth K, Bradshaw J, Levitas R, Middleton S, Pantazis C, Patsios D, Payne S, Townsend P, Williams J. Poverty and social exclusion in Britain. York: Joseph Rowntree Foundation; 2000. [ Google Scholar ]
• Gore C. Introduction: Markets, citizenship and social exclusion. In: Rodgers G, Gore C, Figueiredo JB, editors. Social exclusion: Rhetoric, reality, responses. Geneva: International Labour Organization; 1995. [ Google Scholar ]
• Granovetter, M. S. (1974). Getting a job. A study of contacts and careers . Cambridge: Harvard University Press.
• Halleröd B. The truly poor: Direct and indirect measurement of consensual poverty in Sweden. Journal of European Social Policy. 1995; 5 :111–129. doi: 10.1177/095892879500500203. [ CrossRef ] [ Google Scholar ]
• Hills J, Le Grand J, Piachaud D. Understanding social exclusion. Oxford: OUP; 2002. [ Google Scholar ]
• Jansson, K. (2000). Inkomstfördelningen under 1990-talet. In Välfärd och försörjning 2000, pp. 15–60. SOU 2000:40.
• Jonsson, J. O., Mood, C., & Bihagen, E. (2010). Fattigdomens förändring, utbredning och dynamik, Chapter 3. In Social Rapport 2010 . Stockholm: Socialstyrelsen.
• Jonsson, J. O., & Östberg, V. (2004). Resurser och levnadsförhållanden bland ekonomiskt utsatta 10-18-åringar: Analys av Barn-LNU och Barn-ULF. pp. 203–55 in Ekonomiskt utsatta barn, Socialdepartementet, Ds. 2004:41. Stockholm: Fritzes.
• Levitas R. The concept and measurement of social exclusion. In: Pantazis C, Gordon D, Levitas R, editors. Poverty and social exclusion in Britain. Bristol: Policy Press; 2006. pp. 123–162. [ Google Scholar ]
• Lin, N. (2001). Social capital. A theory of social structure and action . Cambridge: Cambridge University Press.  
• Mack J, Lansley S. Poor Britain. London: Allen & Unwin Ltd; 1985. [ Google Scholar ]
• Mood C. Logistic regression: Why we cannot do what we think we can do and what we can do about it. European Sociological Review. 2010; 26 :67–82. doi: 10.1093/esr/jcp006. [ CrossRef ] [ Google Scholar ]
• Nolan B, Whelan CT. Poverty and deprivation in Europe. New York: Oxford University Press; 2011. [ Google Scholar ]
• OECD . Growing unequal? Income distribution and poverty in OECD countries. Paris: OECD Publishing; 2008. [ Google Scholar ]
• Paugam S. The spiral of precariousness: A multidimensional approach to the process of social disqualification in France. In: Room G, editor. Beyond the threshold: The measurement and analysis of social exclusion. Bristol: Policy Press; 1995. pp. 47–79. [ Google Scholar ]
• Ridge T, Millar J. Following families: Working lone-mother families and their children. Social Policy & Administration. 2011; 45 :85–97. doi: 10.1111/j.1467-9515.2010.00755.x. [ CrossRef ] [ Google Scholar ]
• Ringen S. Direct and indirect measures of poverty. Journal of Social Policy. 1988; 17 :351–365. doi: 10.1017/S0047279400016858. [ CrossRef ] [ Google Scholar ]
• Rodgers G, Gore C, Figueiredo JB, editors. Social exclusion: Rhetoric, reality, responses. Geneva: International Labour Organization; 1995. [ Google Scholar ]
• Rodgers JR, Rodgers JL. Chronic poverty in the United States. Journal of Human Resources. 1993; 28 :25–54. doi: 10.2307/146087. [ CrossRef ] [ Google Scholar ]
• Room G, editor. Beyond the threshold: The measurement and analysis of social exclusion. Bristol: Policy Press; 1995. [ Google Scholar ]
• Sen A. Poor, relatively speaking. Oxford Economic Papers. 1983; 35 :153–169. [ Google Scholar ]
• Smith, A. (1776). An inquiry into the nature and causes of the wealth of nations (republished by R. H. Campbell and A. S. Skinner (Eds.). Oxford: Clarendon Press, 1976).
• Townsend P. Poverty in the United Kingdom. Harmondsworth: Penguin; 1979. [ Google Scholar ]
• van den Bosch K. Identifying the poor: Using subjective and consensual measures. Aldershot: Ashgate; 2001. [ Google Scholar ]
• United Nations. (1995). United nations world summit (Copenhagen) for social development. programme of action , Chapter 2. New York: United Nations.
• UNICEF. (2012). Measuring child poverty. New league tables of child poverty in the world’s rich countries. In Innocenti Report Card 10 . Florence: UNICEF Innocenti Research Centre.
• Veblen T. The theory of the leisure class. New York: McMillan; 1899. [ Google Scholar ]
• Whelan CT, Layte R, Maitre B. Persistent income poverty and deprivation in the European Union: An analysis of the first three waves of the European community household panel. Journal of Social Policy. 2003; 32 :1–18. doi: 10.1017/S0047279402006864. [ CrossRef ] [ Google Scholar ]

inequality.com

The stanford center on poverty and inequality, search form.

• like us on Facebook
• follow us on Twitter
• See us on Youtube

Custom Search 1

• Stanford Basic Income Lab
• Social Mobility Lab
• California Lab
• Social Networks Lab
• Noxious Contracts Lab
• Tax Policy Lab
• Housing & Homelessness Lab
• Early Childhood Lab
• Undergraduate and Graduate Research Fellowships
• Minor in Poverty, Inequality, and Policy
• Certificate in Poverty and Inequality
• America: Unequal (SOC 3)
• Inequality Workshop for Doctoral Students (SOC 341W)
• Postdoctoral Scholars & Research Grants
• Research Partnerships & Technical Assistance
• Conferences
• Pathways Magazine
• Policy Blueprints
• California Poverty Measure Reports
• American Voices Project Research Briefs
• Other Reports and Briefs
• State of the Union Reports
• Multimedia Archive
• Recent Affiliate Publications
• Latest News
• Talks & Events
• California Poverty Measure Data
• American Voices Project Data
• California Data Dashboard
• About the Center
• History & Acknowledgments
• Center Personnel
• Stanford University Affiliates
• National & International Affiliates
• Employment & Internship Opportunities
• Graduate & Undergraduate Programs
• Postdoctoral Scholars & Research Grants
• Research Partnerships & Technical Assistance
• Talks & Events
• History & Acknowledgments
• National & International Affiliates
• Get Involved

Recent Statistics

Featured Research Areas

Numbers, Facts and Trends Shaping Your World

Read our research on:

Full Topic List

Regions & Countries

Publications

• Our Methods
• Short Reads
• Tools & Resources

Read Our Research On:

Income, Wealth & Poverty

Key facts about asian americans living in poverty.

Burmese (19%) and Hmong Americans (17%) were among the Asian origin groups with the highest poverty rates in 2022.

Black Americans’ Views on Success in the U.S.

Key facts about the wealth of immigrant households during the covid-19 pandemic, the varied landscape of minimum wages and tip credits in the u.s., sign up for our weekly newsletter.

Fresh data delivery Saturday mornings

Income inequality is greater among Chinese Americans than any other Asian origin group in the U.S.

Among all Asian origin groups in the U.S., Chinese American households had the highest income inequality in 2022.

The State of the Asian American Middle Class

The share of Asian Americans in the U.S. middle class has held steady since 2010, while the share in the upper-income tier has grown.

The State of the American Middle Class

As the financial divide has grown, a smaller share of Americans now live in middle-class households. Here are key facts about this group.

Is College Worth It?

Americans have mixed views on the importance of having a degree. 47% say the cost is worth it only if someone doesn’t have to take out loans.

7 facts about Americans and taxes

A majority of U.S. adults say they’re bothered a lot by the feeling that some corporations (61%) and some wealthy people (60%) don’t pay their fair share.

1 in 10: Redefining the Asian American Dream (Short Film)

Of the 24 million Asians living in the United States, about 2.3 million live in poverty. This short film explores their diverse stories and experiences.

The Hardships and Dreams of Asian Americans Living in Poverty

About one-in-ten Asian Americans live in poverty. Pew Research Center conducted 18 focus groups in 12 languages to explore their stories and experiences.

A booming U.S. stock market doesn’t benefit all racial and ethnic groups equally

Nearly two-thirds of White families (66%) owned stocks directly or indirectly, compared with 39% of Black families and 28% of Hispanic families.

Who Are You? The Art and Science of Measuring Identity

As a shop that studies human behavior through surveys and other social scientific techniques, we have a good line of sight into the contradictory nature of human preferences. Here’s a look at how we categorize our survey participants in ways that enhance our understanding of how people think and behave.

REFINE YOUR SELECTION

Research teams.

1615 L St. NW, Suite 800 Washington, DC 20036 USA (+1) 202-419-4300 | Main (+1) 202-857-8562 | Fax (+1) 202-419-4372 |  Media Inquiries

Research Topics

• Email Newsletters

ABOUT PEW RESEARCH CENTER  Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes and trends shaping the world. It conducts public opinion polling, demographic research, media content analysis and other empirical social science research. Pew Research Center does not take policy positions. It is a subsidiary of  The Pew Charitable Trusts .

© 2024 Pew Research Center

Mental health effects of poverty, hunger, and homelessness on children and teens

Rising inflation and an uncertain economy are deeply affecting the lives of millions of Americans, particularly those living in low-income communities. It may seem impossible for a family of four to survive on just over $27,000 per year or a single person on just over $15,000, but that’s what millions of people do everyday in the United States. Approximately 37.9 million Americans, or just under 12%, now live in poverty, according to the U.S. Census Bureau .

Additional data from the Bureau show that children are more likely to experience poverty than people over the age of 18. Approximately one in six kids, 16% of all children, live in families with incomes below the official poverty line.

Those who are poor face challenges beyond a lack of resources. They also experience mental and physical issues at a much higher rate than those living above the poverty line. Read on for a summary of the myriad effects of poverty, homelessness, and hunger on children and youth. And for more information on APA’s work on issues surrounding socioeconomic status, please see the Office of Socioeconomic Status .

Who is most affected?

Poverty rates are disproportionately higher among most non-White populations. Compared to 8.2% of White Americans living in poverty, 26.8% of American Indian and Alaska Natives, 19.5% of Blacks, 17% of Hispanics and 8.1% of Asians are currently living in poverty.

Similarly, Black, Hispanic, and Indigenous children are overrepresented among children living below the poverty line. More specifically, 35.5% of Black people living in poverty in the U.S. are below the age of 18. In addition, 40.7% of Hispanic people living below the poverty line in the U.S. are younger than age 18, and 29.1% of American Indian and Native American children lived in poverty in 2018. In contrast, approximately 21% of White people living in poverty in the U.S. are less than 18 years old.

Furthermore, families with a female head of household are more than twice as likely to live in poverty compared to families with a male head of household. Twenty-three percent of female-headed households live in poverty compared to 11.4% of male-headed households, according to the U.S. Census Bureau .

What are the effects of poverty on children and teens?

The impact of poverty on young children is significant and long lasting. Poverty is associated with substandard housing, hunger, homelessness, inadequate childcare, unsafe neighborhoods, and under-resourced schools. In addition, low-income children are at greater risk than higher-income children for a range of cognitive, emotional, and health-related problems, including detrimental effects on executive functioning, below average academic achievement, poor social emotional functioning, developmental delays, behavioral problems, asthma, inadequate nutrition, low birth weight, and higher rates of pneumonia.

Psychological research also shows that living in poverty is associated with differences in structural and functional brain development in children and adolescents in areas related to cognitive processes that are critical for learning, communication, and academic achievement, including social emotional processing, memory, language, and executive functioning.

Children and families living in poverty often attend under-resourced, overcrowded schools that lack educational opportunities, books, supplies, and appropriate technology due to local funding policies. In addition, families living below the poverty line often live in school districts without adequate equal learning experiences for both gifted and special needs students with learning differences and where high school dropout rates are high .

What are the effects of hunger on children and teens?

One in eight U.S. households with children, approximately 12.5%, could not buy enough food for their families in 2021 , considerably higher than the rate for households without children (9.4%). Black (19.8%) and Latinx (16.25%) households are disproportionately impacted by food insecurity, with food insecurity rates in 2021 triple and double the rate of White households (7%), respectively.

Research has found that hunger and undernutrition can have a host of negative effects on child development. For example, maternal undernutrition during pregnancy increases the risk of negative birth outcomes, including premature birth, low birth weight, smaller head size, and lower brain weight. In addition, children experiencing hunger are at least twice as likely to report being in fair or poor health and at least 1.4 times more likely to have asthma, compared to food-secure children.

The first three years of a child’s life are a period of rapid brain development. Too little energy, protein and nutrients during this sensitive period can lead to lasting deficits in cognitive, social and emotional development . School-age children who experience severe hunger are at increased risk for poor mental health and lower academic performance , and often lag behind their peers in social and emotional skills .

What are the effects of homelessness on children and teens?

Approximately 1.2 million public school students experienced homelessness during the 2019-2020 school year, according to the National Center for Homeless Education (PDF, 1.4MB) . The report also found that students of color experienced homelessness at higher proportions than expected based on the overall number of students. Hispanic and Latino students accounted for 28% of the overall student body but 38% of students experiencing homelessness, while Black students accounted for 15% of the overall student body but 27% of students experiencing homelessness. While White students accounted for 46% of all students enrolled in public schools, they represented 26% of students experiencing homelessness.

Homelessness can have a tremendous impact on children, from their education, physical and mental health, sense of safety, and overall development. Children experiencing homelessness frequently need to worry about where they will live, their pets, their belongings, and other family members. In addition, homeless children are less likely to have adequate access to medical and dental care, and may be affected by a variety of health challenges due to inadequate nutrition and access to food, education interruptions, trauma, and disruption in family dynamics.

In terms of academic achievement, students experiencing homelessness are more than twice as likely to be chronically absent than non-homeless students , with greater rates among Black and Native American or Alaska Native students. They are also more likely to change schools multiple times and to be suspended—especially students of color.

Further, research shows that students reporting homelessness have higher rates of victimization, including increased odds of being sexually and physically victimized, and bullied. Student homelessness correlates with other problems, even when controlling for other risks. They experienced significantly greater odds of suicidality, substance abuse, alcohol abuse, risky sexual behavior, and poor grades in school.

What can you do to help children and families experiencing poverty, hunger, and homelessness?

There are many ways that you can help fight poverty in America. You can:

• Volunteer your time with charities and organizations that provide assistance to low-income and homeless children and families.
• Donate money, food, and clothing to homeless shelters and other charities in your community.
• Donate school supplies and books to underresourced schools in your area.
• Improve access to physical, mental, and behavioral health care for low-income Americans by eliminating barriers such as limitations in health care coverage.
• Create a “safety net” for children and families that provides real protection against the harmful effects of economic insecurity.
• Increase the minimum wage, affordable housing and job skills training for low-income and homeless Americans.
• Intervene in early childhood to support the health and educational development of low-income children.
• Provide support for low-income and food insecure children such as Head Start , the National School Lunch Program , and Temporary Assistance for Needy Families (TANF) .
• Increase resources for public education and access to higher education.
• Support research on poverty and its relationship to health, education, and well-being.
• Resolution on Poverty and SES
• Pathways for addressing deep poverty
• APA Deep Poverty Initiative

UNDERSTANDING POVERTY

We face big challenges to help the world’s poorest people and ensure that everyone sees benefits from economic growth. Data and research help us understand these challenges and set priorities, share knowledge of what works, and measure progress.

Free and open access to global development data

• Nowcast of extreme poverty, 2015-2022
• Nowcast of the Global Poverty Rate at the $1.90 Line, 2015–21
• The number of extremely poor people continues to rise in Sub-Saharan Africa
• By Indicator

After several decades of continuous global poverty reduction, a period of significant crises and shocks resulted in around three years of lost progress between 2020-2022. Low-income countries, which saw poverty increase during this period, have not yet recovered and are not closing the gap.

For almost 25 years, extreme poverty was steadily declining. Now, for the first time in a generation, the quest to end poverty has suffered its worst setback. This setback is largely due to major challenges — COVID 19, conflict, and climate change — facing all countries, but in particular those with large poor populations. The increase in extreme poverty from 2019 to 2020 is projected to be larger than any time since the World Bank started tracking poverty globally in a consistent manner. While COVID-19 is a new obstacle, conflicts and climate change have been increasing extreme poverty for years in parts of the world.  Explore Data

Globally, extreme poverty has rapidly declined. New poverty estimates by the World Bank suggest that the number of extremely poor people — those who live on $1.90 a day or less — has fallen from 1.9 billion in 1990 to about 736 million in 2015.  Read More .

Research and Publications

Analysis and advice for developing countries..

Access more than 200,000 publicly available documents and reports.

Essential Reading

Poverty and Shared Prosperity 2022

World Development Report 2019: The Changing Nature of Work

Fair Progress? Economic Mobility across Generations Around the World

Accelerating Poverty Reduction in Africa

Climate Change

Without urgent action, climate change impacts could push an additional 100 million people into poverty by 2030.

• View More Data
• View More Research

Access to energy is essential to reduce poverty. Globally, 1.1 billion people still do not have access to electricity.

Fragility, Conflict and Violence

The challenge of fragility, conflict, and violence is widespread, and not confined to low-income countries.

Financial Inclusion

Around 2 billion people don’t use formal financial services, including more than 50% of adults in the poorest households. Financial inclusion is a key to reducing poverty and boosting prosperity.

This site uses cookies to optimize functionality and give you the best possible experience. If you continue to navigate this website beyond this page, cookies will be placed on your browser. To learn more about cookies, click here .

Watch CBS News

"Persistent poverty" exists across much of the U.S.: "The ultimate left-behind places"

By Aimee Picchi

Edited By Alain Sherter

July 6, 2023 / 5:00 AM EDT / MoneyWatch

Although the U.S. has periodically sought to reduce poverty around the country since the 1960s, roughly 35 million Americans — or almost 1 in 10 — live in communities suffering from "persistent poverty," a recent analysis shows.

That troubling number is 72% higher than previously thought, according to the Economic Innovation Group , which focused on areas where the poverty rate has remained above 20% for more than three decades. To arrive at their figures, the public policy group examined poverty by Census tract — smaller geographic divisions of a county — rather than at the current county level, which can mask pockets of impoverishment.

For instance, by some measures there are no counties in Maine, New Hampshire or Vermont that rank as persistently poor. But each of these states encompasses smaller Census tracts that meet the definition, with most of them home to thousands of deeply poor residents, according to the analysis. 

"Left-behind places"

The findings shed light on overlooked parts of the U.S. that have largely failed to benefit from the significant economic growth the country has enjoyed over the last 30 years. Although many of the persistently poor regions sit in areas long known for their high level of economic deprivation, such as Appalachia and the rural South, EIG found pockets of enduring poverty in every U.S. state. 

"These are the parts of the country that need the most help," EIG Director of Research Kenan Fikri told CBS MoneyWatch, describing them as "the ultimate left-behind places."

"They have been impervious to multiple cycles of economic growth," he added.

"If large tracts of the country are full of people not reaching their full potential, then the country as a whole isn't reaching its full potential," Fikri noted.

To be sure, some Americans who live within persistently impoverished communities aren't poor. Regardless of their income, however, people in such areas may struggle with issues such as access to quality schools, health care and infrastructure. 

Nationally, almost 12% of Americans, or about 38 million people, fall below the poverty line,  according to Census data. Single adults who earn less than $14,580 a year are considered poor, while a family of four earning less than $30,000 is poor, according to federal guidelines.

"Economic or demographic shock"

The regions that suffer from persistent poverty typically experienced "some sort of economic or demographic shock that set them on this path of high poverty, and there hasn't been a countervailing intervention," noted August Benzow, research lead at EIG. 

Those forces can vary, such as Appalachia's dependence on the declining coal mining industry, while many impoverished urban neighborhoods have long faced issues such as racial segregation and lack of access to capital. Yet despite such differences, these communities tend to share a common trait: Once they fall into persistent poverty, it is very difficult to climb out.

Only 7% of counties that experienced poverty rates above 20% in 1990 fell "comfortably below" that level by 2019 while also experiencing population growth, the analysis found. Most of these counties were able to escape persistent poverty because of exurban sprawl or growth in regional industries.

"Once it takes root, it can be very difficult to turn the tide," Fikri said.

Persistently poor communities tend to remain deprived due to their disconnection from regional growth, poor infrastructure, "anemic" small business development and a small tax base that is vulnerable to local economic distress, EIG found. 

"Once places become high poverty, financial institutions and investors tend not to invest in these places, and this creates a calcification or a lack of opportunity to where it's much more difficult to start a business or to purchase a home," Benzow told CBS MoneyWatch.

How to uproot poverty 

Tackling the problem of persistently poor neighborhoods may require multiple initiatives, according to EIG. 

"There is no single silver bullet to fix the issue," Fikri said.

That includes investing in infrastructure and broadband as well as workforce development and education. Communities could be aided by grants to support those efforts, such as to support childcare for parents to re-enter or remain in the workforce. The federal government could also help foster private-sector investment in these areas to attract private capital, EIG said.

"There need to be more investment but it needs to be smarter," Benzow said. Federal investment "needs to be more experimental and innovative."

Aimee Picchi is the associate managing editor for CBS MoneyWatch, where she covers business and personal finance. She previously worked at Bloomberg News and has written for national news outlets including USA Today and Consumer Reports.

More from CBS News

How much can you save by shopping for a home equity loan?

Maps show dengue fever risk areas as CDC warns of global case surge

How ratings for first presidential debate of 2024 compare with past

Boeing sanctioned by NTSB for releasing details of Alaska Airlines probe

Advancing social justice, promoting decent work ILO is a specialized agency of the United Nations

Main Figures on Forced Labour

27.6 million

people are in forced labour.

US$ 236 billion

generated in illegal profits every year.

3.9 million

of them are in State-imposed form of forced labour.

of them are women and girls (4.9 million in forced commercial sexual exploitation, and 6 million in other economic sectors).

of them are children (3.3 million). More than half of these children are in commercial sexual exploitation.

3 times more

risk of forced labour for migrant workers.

• Victims of forced labour include 17.3 million exploited in the private sector; 6.3 million in forced commercial sexual exploitation, and 3.9 million in forced labour imposed by State.
• The Asia and the Pacific region has the highest number of people in forced labour (15.1 million) and the Arab States the highest prevalence (5.3 per thousand people).
• Addressing decent work deficits in the informal economy, as part of broader efforts towards economic formalization, is a priority for progress against forced labour.

Source: 2022 Global Estimates

Forced Labour Observatory

The Forced Labour Observatory (FLO) is a database that provides comprehensive global and country information on forced labour, including on international and national legal and institutional frameworks; enforcement, prevention and protection measures, as well as information related to access to justice; remedies, and cooperation.

Global Reports

2021 Global Estimates of Modern Slavery: Forced Labour and Forced Marriage

The latest estimates show that forced labour and forced marriage have increased significantly in the last five years, according to the International Labour Organization, Walk Free and the International Organisation for Migration.

• Full Report
• Executive Summary
• Press Release
• Third estimates: Modern Slavery (2017)
• Second estimates: Forced Labour (2012)
• First estimates: Forced Labour (2005)

Profits and Poverty: The Economics of Forced Labour (2024)

The study investigates the underlying factors that drive forced labour, of which a major one is illegal profits.

• Press release
• First edition of the report and second estimates (2014)
• First estimates of illegal profits from forced labour (2005)

Main Statistical Tools on Forced Labour

• Hard to See, Harder to Count: Guidelines for Forced Labour Surveys
• Ethical Guidelines for Research on Forced Labour
• Evidence Gap Map on Forced Labour
• Global Research Agenda (child labour, forced labour and human trafficking)

ICLS and forced labour

The International Conference of Labour Statisticians (ICLS) is the authoritative body to set global standards in labour statistics. During its 20th meeting, in October 2018, the ICLS adopted the "Guidelines concerning the measurement of forced labour". The intent of the guidelines is to facilitate the process of testing the measurement of forced labour in different national circumstances and/or measurement objectives.

• Guidelines concerning the measurement of forced labour (ICLS 2018)
• Measurement of forced labour: stocktaking and way forward (ICLS 2023 Room document 22)
• All ICLS documents