Frequently Asked Questions

The Human Development Index (HDI) was created to emphasize that expanding human choices should be the ultimate criteria for assessing development results. Economic growth is a means to that process but is not an end in itself. The HDI can also be used to question national policy choices, asking how two countries with the same level of Gross National Income (GNI) per capita can end up with different human development outcomes.

For example, Kuwait has GNI per capita more than two times higher than Croatia, but Croatia’s life expectancy at birth is three years longer, expecting years of schooling and mean years of schooling are longer one year and four years longer than in Kuwait, respectively, resulting in Croatia having a higher HDI value than Kuwait and being ranked 21 ranks higher. These contrasts can stimulate debate about government policy priorities.

The 2019 HDI covers 189 countries. The wide coverage is the result of efforts by the Human Development Report Office (HDRO) to work with UN agencies and the World Bank, which provide internationally standardized data, and with national statistical agencies to obtain required development indicators for the HDI. For a full explanation of the results and methodology of the 2019 HDI and other composite indices, please see Technical Notes 1-6 at

The Human Development Report Office strives to include as many UN Member States as possible in the HDI. To include a country in the HDI we need recent, reliable and comparable data for all three dimensions of the Index. For a country to be included, statistics should be available from the national statistical authority through mandated relevant international data agencies.

For countries for which only one HDI indicator is missing, the HDRO estimates the missing value using an alternative source or a cross-country regression model. For example, mean years of schooling (MYS) for Liechtenstein is based on MYS of neighbouring Austria. For 10 countries— Comoros, Djibouti, Eritrea, Grenada, Lebanon, Madagascar, Micronesia (Federated States), Saint Kitts and Nevis, South Sudan and Syria — mean years of schooling was estimated by a cross-country regression model. Expected years of schooling was estimated by cross-country regression for nine countries—Bahamas, Congo, Equatorial Guinea, Fiji, Gabon, Haiti, Liberia, Libya, and Vanuatu.

In general, the rankings change a little between two successive years because of the nature of the HDI component indicators. With the exception of gross national income per capita, other indicators change very slowly year to year.

Based on the consistent data series that were available on the cut-off date for downloading data for the computation of composite indices for the 2019 HDR, there are several countries with ranks that changed between 2018 and 2019. The HDI values for 2018 and 2019 are given in Table 2 of the statistical annex. Table 2 also provides the change in ranks between 2014 and 2019.

The consistent data are based on the latest updates and data revisions and are obtained using the same methodology. The effect of change in achievements (improvement or decline) in human development indicators in terms of health, education and living standards is captured by comparing the HDIs obtained from such a consistent data series.

The difference between HDI values (and ranks) published in different editions of HDR represents a combined effect of data revision, change in methodology, and the real change in achievements in indicators. We advise users of the HDI not to compare the estimates from different editions of Reports, but to always use the consistent data given in Table 2 of the latest report or to use the data series available in the Internet database

The major revision was made by the World Bank of GNI and GDP data in PPP terms (World Bank, May 2020.) Data collected in the 2017 International Comparison Program were used for computation of the PPP conversion factors with the new base year set at 2017. Also, the new population data from ‘The World Population Prospect, 2019 Revision’ (United Nations Population Division, June 2019) were used as denominator for computation of indicators expressed per capita and as averages, thus affected GNI per capita and education indicators.

Although the HDI is calculated with a larger number of decimals, we report only the HDI rounded to three decimals. Often there are ties in the HDI three-decimal values of countries, which is also reflected in ties in their ranks. The HDI values, by the very nature of the estimated components, are not significant beyond three decimal places.

Life expectancy at birth is provided by the UN Population Division in the UN Department of Economic and Social Affairs (UNDESA); mean years of schooling (MYS) is based on UNESCO Institute for Statistics (UIS) educational attainment data, for countries for which UIS data are not available, Barro and Lee (2018) estimates and projections were used; expected years of schooling (EYS) is provided by UIS; and GNI per capita (in 2017 $PPP) by the World Bank and the International Monetary Fund. For several countries, mean years of schooling and expected years of schooling are estimated from nationally representative household surveys, and for some countries GNI was obtained from the UN Statistical Division’s database – National Accounts Main Aggregates Database.

Differences between national and international values of indicators exist for some countries. HDRO actively advocates for the improvement of the quality of human development data at all levels – national and international - and for an efficient communication and collaboration between national statistical authorities and the UN statistical entities. The Human Development Report Office does not take data directly from national statistical sources.

The HDI attempts to make an assessment of 189 diverse countries and territories, with very different price levels. To compare economic statistics across countries, the data must first be converted into a common currency. Unlike market exchange rates, PPP rates of exchange allow this conversion to take account of price differences between countries. In that way GNI per capita (PPP $) reflects people's living standards comparably across countries. In theory, 1 PPP dollar (or international dollar) has the same purchasing power in the domestic economy of a country as $1 (USD) has in the US economy.
The current PPP conversion rates have been introduced in May 2020. They were based on the 2017 International Comparison Programme (ICP) Surveys, which covered more than 176 economies from all geographical regions and from the OECD.

No. Income is a means to human development, not its end. GNI per capita only reflects average national income. It does not reveal how that income is spent, nor whether it translates to better health, education, and other human development outcomes. In fact, comparing the GNI per capita rankings and the HDI rankings of countries can reveal much about the results of national policy choices. Equatorial Guinea with the GNI per capita of $13,944 (PPP$) has a GNI rank of 88, but an HDI rank 145 – almost the same HDI as Zambia whose GNI per capita is only $3,326 (PPP$).

No. The concept of human development is much broader than what can be captured by the HDI, or by any other composite index in the Human Development Report (Inequality-adjusted HDI, Gender development index, Gender Inequality Index or Multidimensional Poverty Index). The composite indices are focused measures of human development, zooming in on a few selected areas. A comprehensive assessment of human development requires analysis of other human development indicators and information presented in the statistical annex of the report (see the Reader’s guide to the Report).

Yes, the HDI indicators can be adapted to country-specific indicators provided they meet other aspects of statistical quality. For example, some countries have used under-5 mortality rates at sub-national levels instead of life expectancies, and some have used average disposable income per capita instead of GNI per capita. The HDI can also be disaggregated at sub-national level to compare levels and disparities among different subpopulations within a country, provided that appropriate data at the level of disaggregation are available or can be estimated using sound statistical methodology. The highlighting of internal disparities using HDI methodology has prompted constructive policy debates in many countries.

In 2010, the geometric mean was introduced to compute the HDI. Poor performance in any dimension is directly reflected in the geometric mean. In other words, a low achievement in one dimension is not linearly compensated for by a higher achievement in another dimension. The geometric mean reduces the level of substitutability between dimensions and at the same time ensures that a 1 percent decline in the index of, say, life expectancy has the same impact on the HDI as a 1 percent decline in the education or income index. Thus, as a basis for comparisons of achievements, this method is also more respectful of the intrinsic differences across the dimensions than a simple average.

Income is instrumental to human development, but the contribution diminishes as incomes rise. Also, a high income without being translated into other human development outcomes is of less relevance for human development. Fixing the maximum at $75,000 means that for countries with GNI per capita greater than $75,000, only the first $75,000 contributes to human development. In this way the higher income is prevented from dominating the HDI value. Currently we have only three countries with GNI per capita above the cap – Liechtenstein, Qatar, and Singapore.

In addition to capping, income enters the HDI as a logarithmically transformed variable. The idea is to emphasize the diminishing marginal utility of transforming income into human capabilities. This means that the concave logarithmic transformation makes clearer the notion that an increase of GNI per capita by $100 in a country where the average income is only $500 has a much greater impact on the standard of living than the same $100 increase in a country where the average income is $5,000 or $50,000.

There are arguments for and against transforming the health and education variables to account for diminishing returns. It is true that health and education are not only of intrinsic value; they, like income, are instrumental to other dimensions of human development not included in the HDI (Sen, 1999). Thus, their ability to be converted into other ends may likewise incur diminishing returns. The approach is to value each year of age or education equally, and therefore the principle has been applied only to the income indicator.

Generally, the minimum values are set to the levels that a society needs to survive over time. For life expectancy – 20 years is based on historical evidence (Maddison, 2010, and Riley, 2005), which indicates 20 years as the minimum. If a society or a subgroup of society has a life expectancy below the typical age of reproduction, that society would die out. Lower values have occurred during some crises, such as the Rwandan genocide, but these were exceptional cases that were not sustainable.

Maddison, A. 2010. Historical Statistics of World Economy: 1-2008 AD. Paris: Organization for Economic Cooperation and Development.

Riley, J.C. 2005. Poverty and Life Expectancy. Cambridge, UK: Cambridge University Press. Noorkbakhsh (1998). The Human Development Index: Some Technical Issues and Alternative Indices. Journal of International Development 10, 589-605.

For both education indicators, the minimum is set to 0 since societies can subsist without formal education. For income, it is set at $100 per capita GNI, which is lower than the lowest value attained by any country in recent history (Liberia, 1995). Should any country’s per capita GNI fall close to or below $100, the minimum will be changed accordingly.

The HDI assigns the same weight to all three dimension indices; the two education sub-indices are also weighted equally. The choice of weights is based on the normative assumption that all human beings value the three dimensions equally. The right choice of minima and maxima for the transformation of component indicators into indices gives more equal ranges of variation of dimension indices. Research papers that provide a statistical justification for this approach include:
Noorkbakhsh (1998). The Human Development Index: Some Technical Issues and Alternative Indices. Journal of International Development 10, 589-605.

Decancq, K. and Lugo, A. (2013). Weights in multidimensional indices of wellbeing: An overview. Econometric Reviews, 2013 - Taylor & Francis

As a simple summary index, the HDI is designed to reflect average achievements in three basic aspects of human development – leading a long and healthy life, being knowledgeable and enjoying a decent standard of living. Instead of bringing additional dimensions and indicators into the HDI, other composite indices were introduced – Inequality-adjusted HDI, Gender inequality index, and gender development index. Participation and other aspects of well-being are measured using a range of objective and subjective indicators and are regularly discussed in the Reports. Measurement issues related to these aspects of human development demonstrate the conceptual and methodological challenges that need to be further addressed.

The aggregation of the HDI across countries in a group (HD category, developing region, etc.) is done by applying the HDI formula to the weighted group-averages of component indicators. The weights used to obtain such averages of component indicators are – total population for life expectancy and gross national income per capita, population (ages 5 to 24) for expected years of schooling and population (ages 25 and above) for mean years of schooling.

The HDI represents a national average of human development achievements in the three basic dimensions making up the HDI: health, education and income. Like all averages, it conceals disparities in human development across the population within the same country. Two countries with different distributions of achievements can still have the same average HDI value.

The IHDI takes into account not only the average achievements of a country in health, education, and income, but also how those achievements are distributed among its population by “discounting” each dimension’s average value according to its level of inequality.

The approach is based on a distribution-sensitive class of composite indices proposed by Foster, Lopez-Calva, and Szekely (2005), which draws on the Atkinson (1970) family of inequality measures. It is computed as the geometric mean of dimension indices adjusted for inequality. The inequality in each dimension is estimated by the Atkinson inequality measure, which is based on the assumption that a society has a certain level of aversion to inequality. For details see Technical note 2.

The IHDI has been calculated for years 2010-2019. By analyzing the trend in the IHDI one can assess the direction of the change. Dashboard 5 on socioeconomic sustainability includes the annual average change in the overall loss due to inequality in HDI distribution across population.

The IHDI relies on data on income/consumption and years of schooling from major publicly available databases, which contain micro data from national household surveys harmonized to common international standards: Eurostat’s European Union Statistics on Income and Living Conditions, the Luxembourg Income Study, the World Bank’s International Income Distribution Database, the United Nations Children’s Fund’s Multiple Indicators Cluster Survey, ICF Macro’s Demographic and Health Survey, the Socio-economic database for Latin America and the Caribbean (SEDLAC), and the United Nations University’s World Income Inequality Database. Inequality in the health dimension is computed from the abridged life tables from the United Nations Population Division. A list of surveys used for the 2019 IHDI estimation is given at .

The IHDI refers to 2019. It uses the HDI indicators that refer to 2019 and measures of inequality that are based on the most recent household surveys available from 2008 to 2020 and life tables that refer to the 2015-2020 period. The logic is to use the year to which the HDI indicators refer.

While the HDI can be viewed as an index of average achievements in human development dimensions, the IHDI is the level of human development when the distribution of achievements across people in the society is accounted for. The IHDI will be equal to the HDI when there is no inequality but falls below the HDI as inequality rises. The difference between the HDI and IHDI, expressed as a percentage of the HDI, indicates the loss in human development due to inequality.

The Coefficient of Human Inequality, introduced in the 2014 HDR as an experimental measure, is a simple average of inequalities in health, education, and income. The average is calculated by an unweighted arithmetic mean of estimated inequalities in these dimensions. When all inequalities are of a similar magnitude, the coefficient of human inequality and the overall loss in HDI differ negligibly; when inequalities differ in magnitude, the loss in HDI tends to be higher than the coefficient of human inequality.

The IHDI captures the inequality in distribution of the HDI dimensions. However, it is not association sensitive, meaning that it does not account for overlapping inequalities—whether the same people are at the lower end of each distribution. Also, individual values of education and income can be zero or even negative (for income), so they have been adjusted to non-negative non-zero values uniformly across countries. This adjustment can reshape the distributions a bit, so it is fair to say that the Atkinson inequality measure provides an approximation of the magnitude of inequality.

The IHDI allows a direct link to inequalities in dimensions of the HDI and the resulting loss in human development. Thus, it can help inform policies towards inequality reduction and to evaluate the impact of various policy options aimed at inequality reduction.

The IHDI and its components can be useful as a guide to help governments better understand the inequalities across population and their contribution to the overall loss in the level of human development due to inequality.

The IHDI in its current form is inspired by a similar index produced by Mexico’s national HDR. The IHDI can be adapted to compare the inequalities in different subpopulations within a country, provided that the appropriate data are available. National teams can use proxy distributions for indicators, which may make more sense in the national context.

The IHDI was initiated as an experimental index in 2010, alongside the Gender Inequality Index and the Multidimensional Poverty Index. It has been critically examined and discussed at conferences on measuring human progress held in 2012, 2013, 2014 and 2018. It became a regularly computed composite index and it may evolve over time like all other human development indices.

One of the key properties of the approach is that it is ‘subgroup consistent’. This means that if inequality declines in one subgroup and remains unchanged in the rest of population, then the overall inequality declines. The second important property is that the IHDI can be obtained by first computing inequality for each dimension and then across dimensions, which further implies that it can be computed by combining data from different sources, thus it is not necessary that micro data on all components come from the same survey.

The Gini coefficient is commonly used as a measure of inequality of income, consumption, or wealth. There was an attempt to apply the Gini index to measure inequality in the HDI distribution (Hicks, 1998).
The choice of the Atkinson inequality measure is guided by three factors: (i) subgroup consistency, (ii) sensitivity to the inequality in the lower end of distribution, and (iii) simplicity of computation and mathematical elegance of the resulting composite Inequality-adjusted Human Development Index.

Subgroup consistency means that if inequality declines in one subgroup (region, ethnic group, etc.) and remains unchanged in the rest of population, then the overall inequality declines. The Gini coefficient does not have this property. By its construction, the Gini coefficient puts equal weights to the entire distribution, while the Atkinson inequality measure puts more weight to the lower end, thus it accounts better for child mortality, illiteracy, and income poverty.

Finally, the geometric form of the HDI in combination with the Atkinson index provides a simple and elegant composite IHDI, obtained by first computing inequality for each dimension and then across dimensions, which further implies that it can be computed by combining data from different sources (life tables and different surveys for education and income).

No. Due to data limitations and the construction, the IHDI does not capture overlapping inequalities—whether the same person is at the lower end of distributions in all three dimensions. Also, at this time we are not able to estimate inequality in distribution of expected years of schooling, so the inequality in education dimension is assessed only from distribution of years of education for the adult population.

By their very nature, income and consumption yield different levels of inequalities, with income inequality being higher than inequality in consumption. Income seems to correspond more naturally to the notion of “command over resources.” Consumption data are arguably more accurate in developing countries, less skewed by high values, and directly reflect the conversion of resources. Income data also pose technical challenges because of the greater presence of zero and negative values. In an ideal world, one would be consistent in the use of either income or consumption data to estimate inequality. However, to obtain sufficient country coverage, it is necessary to use both. The final estimates are modestly influenced by whether inequality refers to income or consumption distribution.

Life expectancy is calculated as the average life span of a hypothetical cohort of 100,000 people born at the same time who progress through successive ages, with the cohort shrinking from one age to the next according to a set of age-specific death rates, until all people in the cohort die. Such an average is the life span that people born in the same year can be expected to live under the constant-mortality assumption, i.e., age-specific mortality is maintained constant throughout the life of the cohort at the level estimated for the reference year of birth or the reference period.

As the cohort shrinks from one age to the next, it implies that people from the cohort are dying at different ages – thus there is an inequality in life spans. We capture this inequality and use it as the inequality measure of the health dimension - in the same way as we use the average of these life spans (i.e., life expectancy) to estimate the average achievement in health dimension.

Inequality in the education dimension is approximated only by inequality in years of schooling of the adult population drawn from nationally representative household surveys. For some countries this inequality is computed from the education attainment tables of the UNESCO Institute for statistics.

Expected years of schooling is an aggregate measure and inequality in its distribution is reflected in current school enrolment ratios. Certainly, there is a difference in inequalities in the two distributions – years of schooling for the adult population and expected years of education for the school age population, with the inequality in distribution of expected years of schooling across the school-age population being lower. Thus, one can speculate that overall inequality in the HDI distribution would be reduced if expected years of schooling were used.

Years of schooling of adults is mostly derived from the highest level of schooling achieved. Using UNESCO’s country information on the duration of schooling needed for each level, the highest level of schooling is converted into years. While the duration of primary, secondary and most of post-secondary education is more or less standardized, the very high levels of education—masters and doctoral studies—vary in length across countries. However, the Atkinson measure of inequality, which is used to assess inequality in the education components, is less sensitive to differences at the upper end of the distribution.

The GDI measures differences in male and female achievements in three basic dimensions of human development: health, education and command over economic resources. Gender disaggregated data is used in each dimension. The health dimension is captured by female and male life expectancy at birth. Education is measured using two indicators—female and male expected years of schooling for children and female and male mean years of schooling for adults ages 25 and older. Command over economic resources is measured by female and male estimated earned income.

The GDI is the ratio of female HDI to male HDI. To calculate it, the HDI is first calculated separately for females and for males. The same goalposts as in the HDI are used for transforming the indicators into a scale lying between zero and one. The only exception is life expectancy at birth where the goalposts are adjusted, to reflect the empirical finding that on average, women have a biological advantage over men, and live about 5 years longer.

The income component, female and male estimated earned income, is calculated based on female and male shares of the population, female and male shares of economically active population, ratio of female to male wages in all sectors, and GNI per capita in PPP$ (2017 constant prices).

The income component of the GDI is a proxy to command over economic resources. This component captures income gaps in a way similar to the focus on gender gaps in other HDI components.

The global average female to male wage ratio across all sectors is about 0.8 since 2018. This global average is used to estimate the wage ratio for countries with missing sex-disaggregated wage data. We recognize the limitations in assuming that the global average applies to all countries with missing wage data. The International Labour Organization (ILO) is working to improve availability of sex-disaggregated wage statistics.

Estimating the female and male HDIs for all countries relies on many approximations, such as assuming wage ratios of 0.8 for many countries. Because of this the estimated HDIs need to be interpreted with caution. We prefer not to rank the countries based on these approximated HDIs. Instead, we group countries into five GDI groups by absolute deviation from gender parity in HDI values.

Group 1 countries have high equality in HDI achievements between women and men: absolute deviation less than 2.5 percent; group 2 has medium-high equality in HDI achievements between women and men: absolute deviation between 2.5 percent and 5 percent; group 3 has medium equality in HDI achievements between women and men: absolute deviation between 5 percent and 7.5 percent; group 4 has medium-low equality in HDI achievements between women and men: absolute deviation between 7.5 percent and 10 percent; and group 5 has low equality in HDI achievements between women and men: absolute deviation from gender parity greater than 10 percent.

The GDI helps in better understanding of the gender gap in human development achievements. It provides insights into gender disparities in achievements in three basic capabilities: health, education and command over economic resources, and is useful for designing and monitoring policies to close the gaps.

The GII is an inequality index. It shows the loss in potential human development due to disparity between female and male achievements in three dimensions: reproductive health, empowerment and the labour market. Overall, the GII reflects how women are disadvantaged in these dimensions.

The GII ranges between 0 and 1. Higher GII values indicate higher inequalities between women and men and thus higher loss to human development. There is no country with perfect gender equality. All countries suffer some loss in achievements in key aspects of human development when gender inequality is taken into account. The GII is similar in method to the Inequality-adjusted Human Development Index (IHDI)—see Technical Note 4 for details. It can be interpreted as a combined loss to achievements in reproductive health, empowerment and labour market participation due to gender inequalities. Since the GII includes different dimensions than the HDI, it cannot be interpreted as a loss in HDI itself. For more details, please refer to the discussions in this paper and Technical Note 4 for the current methodology.

The GII includes reproductive health and measures empowerment beyond the female literacy and primary education. It also reveals gender disparities in labour market participation and avoids using flawed sex-disaggregated income measures.

Like all composite measures, the GII has some limitations. First, it does not capture the length and breadth of gender inequality. For example, it captures national parliamentary representation but excludes participation at the local government level and elsewhere in community and public life. The labour market dimension lacks information on employment and the quality of jobs. The index misses other important dimensions, such as unpaid work, and the fact that many women carry an unfair burden of caregiving and housekeeping. Asset ownership, childcare support, gender-based violence and participation in community decision-making are also not captured in the GII, mainly due to limited data availability.

The GII relies on data from major publicly available international databases, including the maternal mortality ratio from World Health Organization (WHO), United Nations Children’s Fund (UNICEF), United Nations Population Fund (UNFPA), World Bank Group and United Nations Population Division; adolescent birth rates from the UN Department of Economic and Social Affair’s World Population Prospects; educational attainment statistics from the UNESCO Institute for Statistics educational attainment tables and the Barro-Lee data sets; parliamentary representation from the International Parliamentary Union (IPU); and labour market participation from the International Labour Organization (ILO).

It is true that reproductive health indicators used in the GII do not have equivalent indicators for males. In this dimension, the reproductive health of girls and women is compared to what should be the societal goals—no maternal death and no adolescent pregnancy. The rationale is that safe motherhood reflects the importance society attaches to women’s reproductive role. Early childbearing, as measured by the adolescent birth rate, is associated with greater health risks for mothers and infants; also, adolescent mothers often are forced out of school and into low-skilled jobs.

Only 1 out of 162 countries included in the GII has female shares of parliamentary seats equal to zero. Because the functional form is multiplicative, we replaced the zero value with 0.1 percent to make the computation possible. The rationale is that while women may not be represented in parliament, they do have some political influence. The relative rank of the country is sensitive to the choice of the replacement value. The lowest observed non-zero female shares of parliamentary representation was 1.0% for Yemen.

The GII provides insights into gender disparities in reproductive health, empowerment and labour market participation in 162 countries. It can help governments and others understand the extent of gender inequalities in empowerment. The component indicators highlight areas in need of critical policy intervention. The GII, like any other global composite index, is constrained by the need for international comparability. But it could be readily adapted for use at the national or local level.

The Global Multidimensional Poverty Index (MPI) identifies multiple deprivations at the household and individual level in health, education and standard of living. It uses micro data from household surveys, and—unlike the Inequality-adjusted Human Development Index—all the indicators needed to construct the measure must come from the same survey. Each person in a given household is classified as poor or non-poor depending on the weighted number of deprivations his or her household, and thus, he or she experiences. These data are then aggregated into the national measure of poverty. The MPI reflects both the incidence of multidimensional deprivation (a headcount of those in multidimensional poverty) and its intensity (the average deprivation score experienced by poor people). It can be used to create a comprehensive picture of people living in poverty, and permits comparisons both across countries, regions and the world and within countries by ethnic group, urban or rural location, as well as other key household and community characteristics. The MPI offers a valuable complement to income-based poverty measures.

The 2020 release of the MPI presents estimates for 107 developing countries with a combined population of 5.9 billion (77% of the world total). About 1.3 billion people in the countries covered—22% of their entire population—lived in multidimensional poverty between 2008 and 2019. We could not include other developing countries due to data constraints as comparable data on each of the indicators were not available. There was also a decision not to use data from surveys conducted earlier than 2008.

The MPI identifies overlapping deprivations that people experience across the same three dimensions as the Human Development Index (health, education and standard of living) and shows the proportion of people that are poor and the average number of deprivations each poor person experiences at the same time. For more details see Technical note 5.

Using the United Nations Development Programme (UNDP) classification for developing regions, the MPI 2020 covers 11 (out of 20) countries in the Arab States region; 12 (out of 24) in East Asia and the Pacific; 13 (out of 17) countries in Europe and Central Asia; 21 (out of 33) countries in Latin America and the Caribbean; 8 (out of 9) countries in South Asia; and 42 (out of 46) countries in Sub-Saharan Africa.

The MPI relies on two main databases that are publicly available and comparable for most developing countries: the ICF Macro Demographic and Health Survey (DHS) and the UNICEF’s Multiple Indicators Cluster Survey (MICS). For several countries, national household surveys with the same or similar content and questionnaires are used: Botswana, 2015-2016 Multi-Topic Household Survey; Brazil, 2015 Pesquisa Nacional por Amostra de Domicílios (PNAD); China, 2014 China Family Panel Studies; Cuba, 2017 Encuesta Nacional de Ocupacion; Ecuador, 2013-2014 Encuesta de Condiciones de Vida (ECV); Jamaica, 2014 Jamaica Survey of Living Conditions (JSLC); Libya, 2014 Pan Arab Population and Family Health Survey (PAPFAM); Mexico, 2016 Encuesta Nacional de Salud y Nutricion (ENSANUT); Morocco, 2011 Pan Arab Population and Family Health Survey (PAPFAM); Peru, 2018 Encuesta Demográfica y de Salud Familiar; Seychelles, 2019 Quarterly Labor Force Survey, Sri Lanka, 2016 Demographic and Health Survey; and Syrian Arab Republic, 2009 Pan Arab Population and Family Health Survey (PAPFAM).

Table 1 of the 2020 release of the MPI indicates for each country if data come from the DHS, MICS or from a national survey.

Since the 2014 HDR, the statistical programmes used to calculate the MPI have been available at the HDRO’s website. The statistical programs associated with the MPI 2020 are available at the HDRO’s website for a large selection of countries (

This year, for the first time, we are releasing programs that calculate the MPI in R, a free software available at For now, these programs are available for 4 selected countries (Benin, Republic of Congo, India and Iraq). This is still an experimental phase and we plan to expand the availability of MPI programs in R; in the meantime, we welcome feedback from the users.

Also, DHS and MICS data are publicly available online. Therefore, national governments, civil society groups, and research communities can replicate the MPI results as well as adapt the programmes to their own country-specific poverty needs.

The MPI was first developed in 2010 by OPHI and UNDP for the UNDP’s flagship Human Development Reports. In February 2012 and March 2013 a critical review of the family of human development indices including the MPI was conducted during the two conferences on measuring human progress organized by UNDP. As an outcome of these critical reviews, a number of adjustments were made to the MPI. They were justified on the grounds of being more in line with the MDGs. At that time, OPHI continued publishing their own estimates using the original 2010 specifications. In 2018, OPHI and UNDP revised and unified the MPI methodology and the joint work is available at both OPHI’s and UNDP’s websites. The MPI numbers and analysis are expected to be updated at least once per year to include newly released data.

The new round of adjustments to the MPI introduced in 2018 were justified on the grounds of being more in line with the 2030 Agenda. Since 2018 the MPI is therefore a contribution to the implementation and monitoring of Sustainable Development Goal 1 which aims to end poverty in all its forms everywhere, and to the achievement of the Agenda’s ambition and fundamental principle of “Leaving No One Behind”.

The difference between the previous approaches stands in the definition of deprivations for some of the indicators.

  • Health dimension:
    - Nutrition: a household is deprived if there is a stunted or an underweight child (instead of only stunted). Because, if a child is stunted, the damage is mostly irreversible. As Anthony Lake of UNICEF described it: “That child will never learn, nor earn, as much as he or she could have if properly nourished in early life.” Adults above 20 years are considered malnourished if their BMI is below 18.5 kg/m2; for individuals aged 15-19 years age-specific BMI z scores, defined by WHO, are calculated.

    - Child mortality: a household is deprived if any child has died in the family in the five-year period preceding the survey. This captures recent improvements in child mortality. When the survey lacks information about the date of the child death, mortality that occurred anytime is used. (one additional revision was introduced in 2019, see next question)

  • Education dimension:
    - School attainment: a household is deprived if no member aged 10 years or older has completed six years of schooling. Six years is the duration of primary education in most countries, so this change reinforces SDG 4 “Quality education.” (one additional revision was introduced in 2020, see next question)

    - School attendance: a household is deprived if any school-aged child is not attending school up to the age at which he/she would complete class 8.

  • Standard of living:
    - Housing: a household is deprived if it has inadequate housing in at least one of the three: the floor is of natural materials or the roof or the walls are of rudimentary materials.

    - Assets ownership: a household is deprived if it does not own more than one of these assets: radio, TV, telephone, computer, animal cart, bicycle, motorbike, or refrigerator, and does not own a car or truck.
    - Electricity, improved sanitation, improved drinking water and cooking fuel remain the same.

Further details about the original MPI methodology from 2010 can be found in Alkire and Santos (2010)

Further details about the UNDP/HDRO revised methodology from 2014 can be found in Kovacevic and Calderon (2014)’s-multidimensional-poverty-index-methodology-paper-2014

Further details about the jointly revised methodology can be found in: Alkire and Jahan (2018)

Yes. This year, we adjusted the years of schooling indicator by introducing a country-specific age cutoff: No household member of ages 'school entrance age + six' or older has completed six years of schooling.

Previously, the age cutoff was 10 years, but this did not recognize the fact that by age 10 children do not normally complete 6 years of schooling. For example, if the official school entrance age is 6, a child would normally complete 6 years of schooling by age 12; therefore, children of ages 10 or 11 should be considered ineligible for years of schooling. If, as an exception, children of ages 10 or 11 completed 6 years of schooling, this is counted as an achievement. Note that this adjustment has been implemented only to the countries with an updated survey in 2020. Although this adjustment is conceptually better, the effect on the empirical estimates is minimal.

Also, in 2019, the only methodological adjustment was applied to child mortality. Following the Convention on the Rights of the Child, a child is defined as a person below the age of 18. With this new adjustment, the death of an offspring reported by a mother does not count as a deprivation if the death occurred when the son or daughter was above the age of 18. Previously, there was no age limit and child was understood as a son or daughter of any age.

The indicators for the 2010 MPI were drawn from the Millennium Development Goals (MDGs). These original MPI indicators related, to MDG indicators: nutrition (MDG 1), child mortality (MDG 4), access to drinking water (MDG 7), access to sanitation facility (MDG 7) and use of an improved source of cooking fuel (MDG 9).

With the subsequent adjustments of the methodology, the 2018 revision of the MPI is a contribution to the implementation and monitoring of Sustainable Development Goal 1 which aims to end poverty in all its forms everywhere, and to the achievement of the Agenda’s ambition and fundamental principle of “Leaving No One Behind”. In addition, national measures of Multidimensional Poverty may directly serve the purpose of monitoring SDG Indicator 1.2.2 (proportion of men, women and children of all ages living in poverty in all its dimensions according to national definitions).

The global MPI shows the simultaneous deprivations of people sharing the same household across ten indicators that relate to SDGs 1, 2, 3, 4, 6, 7, and 11. In doing so, it provides a policy tool that can inform policies that seek to address interlinked SDGs and their targets.

Yes. We are continuously working on improving the MPI methodology to better measure the multidimensional aspect of human’s deprivations, to reflect the needs of policy makers and to incorporate new indicators or components when surveys update the information collected.

The MPI constitutes a set of poverty measures. These measures are explained as follows. Headcount or incidence of multidimensional poverty: the proportion of people who are poor according to the MPI (those who are deprived in at least one third of the weighted indicators). Intensity of multidimensional poverty: the weighted average number of deprivations poor people experience at the same time. The MPI value summarizes information on multiple deprivations into a single number. It is calculated by multiplying the poverty headcount by the intensity of poverty. These measures can be unpacked to show the composition of poverty both across countries, regions and the world and within countries by ethnic group, urban/rural location, as well as other key household and community characteristics.

One deprivation alone may not represent poverty. The MPI requires a person to be deprived in multiple indicators at the same time. A household, and so all its members, is multidimensionally poor if it is deprived in one third or more of the weighted indicators. We also count persons who are vulnerable to multidimensional poverty as those who are deprived in one fifth or more but less than one third of the weighted indicators. Those who are deprived in one half or more are considered living in severe multidimensional poverty.

The number of people living in multidimensional poverty have been computed using the United Nations Department of Economic and Social Affairs/Population Division’s total population data for the year 2018 for all countries. This approach assumes a constant headcount ratio (obtained by applying the MPI methodology to the survey) which implies an assumption of no change in poverty rates between the year of the survey and 2018.

In addition, the 2020 release of the MPI provides the number of people living in multidimensional poverty during the year of the survey. In countries with surveys conducted in a two-year period, i.e. 2015/2016, the population of the second year is used.

The MPI relies on the most recent and reliable data available since 2008. The difference in dates limits direct cross-country comparisons, as circumstances may have improved, or deteriorated, in the intervening years. This is the reason why we do not rank countries based on the MPI value.

Data for the 2020 MPI are from 2008–2019, though 5.6 billion of the 5.9 billion people covered and 1.2 billion of the 1.3 billion multidimensionally poor people identified are captured by surveys from 2014 or later (83 of the 107 countries covered).

The MPI methodology shows aspects in which the poor are deprived and helps to reveal inter-connections among those deprivations. This enables policymakers to target resources and design policies more effectively. This is especially useful where the MPI reveals areas or groups characterized by severe deprivations.

The multidimensional poverty approach can be adapted using indicators and weights that are more relevant to national context at the country level to create tailored national poverty measures. The MPI can be useful as a guide to help governments tailor a poverty measure that reflects local indicators and data. In 2009, Mexico became the first country to adopt a measure of multidimensional poverty as an official national statistic.

The MPI methodology can be, and often is, modified to generate national measures of Multidimensional Poverty that reflect local cultural, economic, climatic and other factors. Such national measures of Multidimensional Poverty may directly serve the purpose of monitoring SDG Indicator 1.2.2 (proportion of men, women and children of all ages living in poverty in all its dimensions according to national definitions). The MPI was devised as an analytical tool to compare acute poverty across nations.

The MPI has some drawbacks, due mainly to data constraints.

  • The indicators may not reflect capabilities but instead reflect outputs (such as years of schooling) or inputs (such as cooking fuel).
  • The health dimension indicators are not consistently collected across countries and overlook some groups’ deprivations, especially for nutrition, though the patterns that emerge are plausible and familiar.
  • Child deaths that occurred at any time (instead of in the last 5 years previous to the survey) are used in 10 countries because the survey did not collect the date of child deaths. We are currently working on finding a better alternative to these cases.
  • In some cases, careful judgments were needed to address data that was not collected. For example, data on nutrition was not collected for children older than 5 years in both DHS and MICS, data on adult nutrition was not collected in MICS, data on adult nutrition was collected only for women of ages 15 and 49 in DHS (though in a few small number of countries it was collected for men), and child mortality was not collected in households without women of reproductive age. In these situations, households without eligible members to be measured for nutrition and without eligible members to report child mortality are assumed to be not deprived in these indicators. This is a strong assumption that may biases the results and we are conducting research to find a better alternative. However, to be considered multidimensionally poor, households must be deprived in at least six standard of living indicators or in three standard of living indicators and one health or education indicator, or in two health or education indicators. This requirement makes the MPI less sensitive to such assumption.
  • Intra-household inequalities may be severe, but these could not be captured.
  • While the MPI goes well beyond a headcount ratio to include the intensity of poverty, it does not measure inequality among the poor, although a separate measure of inequality has been published since 2019 (variance of individual deprivation scores of poor people). Also, decompositions by groups can be used to reveal group-based inequalities.
  • The estimates presented are mostly based on publicly available data and cover various years between 2008 and 2019, which limits direct cross-country comparability.

The MPI can be used to study the changes in poverty patterns over time providing that the data at different points in time and MPI estimates are strictly harmonized. If estimates are not harmonized, readers are advised to carefully interpret the changes over time for a particular country because different indicators could be missing from the surveys.

This year, we are releasing changes over time estimates for 75 countries with suitable surveys. These estimates are presented in Table 2 of the 2020 MPI publication. This table presents harmonized estimates and it allows for a strict comparison across time. The harmonization process aligns the indicator definitions, meaning that, where necessary, this process re-creates the indicators in the global MPI so that they are using exactly the same information and deprivation cutoffs in both years. For instance, if a survey in one year collects nutrition information from children under 5 years and for women aged 15–49 years and in the other year it collects nutrition information from children under 5 years only, the harmonized computations will only use information from children under 5 in both years, the common denominator, so that any change in the nutrition indicator is due to a change in the nutritional status of children rather than the inclusion of a group of women who may tend to be more (or less) undernourished.

Users are advised to distinguish between the unharmonized estimates in table 1 and the harmonized estimates in table 2. The first ones represent the best possible MPI estimate for a particular year and survey and include all the information that is available, while the second ones allow for a strict comparison across time by removing the components that were not collected in the previous point in time. Unharmonized and harmonized estimates may be identical in some cases, but they may differ in many others.

The effects of shocks are difficult to capture in any poverty measure. Because the standard survey data used to estimate the global measure are collected in differing intervals, the ability to detect changes depends upon the frequency of available data. The MPI will reflect the impacts of shocks as far as these cause children to leave primary education or to become malnourished, or lose adequate housing, for example. If more frequent data are available at the country or local level, this can be used to seek to capture the effects of larger scale economic and other shocks.

The MPI will be a valuable tool to measure the effects of COVID-19 and how the crisis has affected many of the indicators such as nutrition, school attendance, years of schooling and housing.

Yes. The global MPI estimates are constrained by need for comparability. National teams should use the indicators and weights that are more relevant to the country context. At the country level, the multidimensional poverty approach to assessing household deprivations can be tailored using country-specific data and indicators to provide a richer and more representative picture of poverty.

Incomplete data on the health dimension were collected in the Brazil’s Pesquisa Nacional por Amostra de Domicílios (PNAD). While PNAD collects rich data on social, economic and demographic characteristics of households and their members, some data that are needed for the MPI computation, such as basic nutrition characteristics (body weight and height of children and adults) are not collected. Although the survey collects household information on deaths of children (of any age), it does not collect the date of the death, so it is not possible to distinguish recent deaths from those that happened more than 5 years prior to the survey. Thus, the child mortality indicator has incomplete information on deaths of children. Given the lack on nutrition, this implies that the imperfect child mortality indicator would receive the full dimension weight of 1/3.

In such a case, the MPI methodology was modified to compensate for this limitation in the health dimension and two adjustments were performed: (i) only child mortality reported by mothers of ages between 15 and 49 was considered, and (ii) a household that is deprived only in child mortality cannot be declared MPI poor. The first adjustment increases the likelihood that the reported deaths happened more recently and that they refer to a child under the age of 18 year, while the second removes the possibility that a household is declared MPI poor based only on the death of a child that happened any time before the survey. This second rule means that an additional deprivation in at least one of the remaining indicators is needed for such a household to be declared MPI poor.

Brazil is the only country that lacks information about nutrition and that lacks information about date of death for the child mortality indicator. There are 9 countries that lack data on deprivations in nutrition but with complete data on mortality. Likewise, there are another 10 countries without date of death, but with data on deprivations in nutrition, therefore, the weight attached to child mortality is only 1/6 making the MPI estimates less dependent on the treatment of child mortality.

The MPI reflects the multiple deprivations that people experience at the same time. We have described the MPI as a measure of “acute” poverty because it reflects overlapping deprivation in basic needs and also to avoid confusion with the World Bank’s measure of “extreme” poverty that captures those living on less than $1.90 (in 2011 $PPP) a day.

The MPI complements income poverty measures. It measures various deprivations directly. In practice, although there is a clear overall relationship between the MPI and the $1.90 per day measure, the estimates do differ for many countries. This is a topic for further research, but some factors can include the provision of public services, as well as different abilities to convert income into positive outcomes such as good nutrition.

The MPI, like the $1.90 per day line, is a globally comparable measure of poverty. It measures acute multidimensional poverty, and only includes indicators that are available for many countries. National poverty line measures are typically monetary measures, and, thus, capture something different. The fact that there are differences does not mean that the national poverty number or the MPI headcount are wrong—these simply measure different concepts of poverty. At the same time, just as national poverty measures are designed to reflect the domestic situation more accurately and often differ in very useful ways from the $1.90 measure, countries are encouraged and assisted by UNDP to build a national multidimensional poverty index that is tailored to their context and many times based on the national surveys, to complement this global MPI.

Including income as an indicator might imply double counting people’s deprivations. The standard of living dimension of the MPI acts as a proxy for economic wellbeing. The MPI, which also includes other dimensions of wellbeing, should be seen as a complementary measure of income poverty that goes beyond the monetary aspect of people’s lives.

An interesting analysis would be to explore the overlap between income poverty and multidimensional poverty, i.e. to distinguish those who are income poor and MPI poor versus those who are poor according to one concept only or those who are not poor at all. However, income data come from different surveys, and these surveys often do not have information on the health dimension. For most countries, we are not able to identify whether the same people are income poor and also deprived in the MPI indicators.

We could not include empowerment due to data constraints. The Demographic and Health Surveys (DHS) collect data on women’s empowerment for some countries, but not every DHS survey includes empowerment, and the other surveys do not have these data. Data on men’s empowerment or political freedom are missing.

The dashboard approach has become popular for monitoring development outcomes. The 2020 Human Development Report features five colour-coded tables also termed dashboards covering five topics: quality of human development, life-course gender gap, women’s empowerment, environmental sustainability and socioeconomic sustainability. The colour-coded tables evaluate progress of human development by highlighting levels and changes of various indicators.

The dashboards visualize grouping of countries by each indicator, thus partially, contrary to a complete grouping by a composite measure, which combines all listed indicators after making them commensurable. A good example of a complete grouping is the grouping of countries into four human development groups by the Human Development Index (HDI). The complete grouping by a composite index depends on the way the component indicators are combined into the index. On the other hand, the partial grouping does not require any assumption on normalization, weighting or the functional form of the composite index. A partial grouping may depend on the predefined values considered as thresholds needed for grouping, such as what is considered a good performance or as a target to be achieved.

The decision was to group countries to a small number of groups, say three, according to the values of an indicator achieved by countries. Countries are divided into three groups of approximately equal sizes (terciles): the top third, the middle third and the bottom third. A distinct shade of a selected colour is attached to a group of countries with a similar level of performance. The colour-coding scale graduates from darkest to lightest.

The darker shade represents the top third group; the moderate shade represents the middle third; and the lightest shade represents the bottom third of countries. Partial grouping of countries applies to all indicators listed in five dashboards. Sex ratio at birth of Dashboard 2 is an exception—countries are divided into two groups: the natural group (countries with a value between 1.04-1.07, inclusive), which uses darker shading, and the gender-biased group (all other countries), which uses lighter shading. See Technical note 6 at for details.

When grouping countries into tercile groups according to each indicator, the intention is not to suggest thresholds or target values for any indicator, but rather to allow a crude assessment of country’s performance relative to others. A country that is in the top group performs better than at least two thirds of countries; a country that is in the middle group performs better than at least one third but worse than at least one third; and a country that is in the bottom third performs worse than at least two thirds of countries. The observed ranges of values that define tercile groups for all indicators in dashboards 1-5 are given in Technical note 6 at

Countries with values of ratios around 1 form the group with the top achievements in that indicator. Large gaps in favor of men are treated equally as large gaps in favor of women.

In dashboard 4 on environmental sustainability, countries are not grouped by percentage of total land area under forest, but rather by the change in forest area since 1990. The reason is to respect the fact that the forest area is in a way determined by environmental and climate conditions, while the recent change in forest area is caused by men’s activities. Similarly, in dashboard 5 on socioeconomic sustainability, indicator military expenditure (% of GDP) was not used for grouping of countries, instead the ratio of education and health expenditure to military expenditure was used for grouping and coloring. The reason is that military expenditure was not considered as an indicator in this table, but rather as an auxiliary indicator.

Group aggregates were not used to define the tercile groups. However, based on the value of an aggregate, it was placed in a tercile group and coloured accordingly.