Inequality by Country

This new module introduced in UQICD 3.0 represents a major step forward in providing researchers and users with panels of real incomes as well as the distribution of incomes - two major components of economic welfare of a society. The real GDP series provides estimates of the size of the economies with real per capita GDP or income serving as in indicator of standard of living in these economies. It is well recognized (Sen, 1976; and Stiglitz-Sen- Fitoussi, 2009) that inequality in the distribution of income is a critical determinant of economic welfare.

Availability of income distribution data is largely determined by the frequency with which household income and expenditure surveys are conducted. These data play a major role in the estimation of incidence and severity of poverty at the national, regional and global level. Despite the importance attached to inequality data, availability is sparse. More importantly, detailed unit record data are usually not available for the users and often researchers rely on aggregated distribution data in the form of income shares for decile or ventile groups. While aggregated data are used in approximating inequality measures like the Gini coefficient and the Theil's measure, details of the underlying income distributions are seldom available for purposes of analysis. This UQICD module on inequality serves to fill this gap.

A series of research advances achieved over the last fifteen years have enabled the UQICD team to compile this panel of income distributions over the period 1970 to 2019. Data is available for 157 countries that are included in the computations for this module (see detailed information).

Charts depicting density functions, for each country, are available from this page providing users with a comparison of the fitted log-normal, pareto log-normal, GB2 and mixture log-normal for six selected years (1970, 1980, 1990, 2000, 2010, 2019).

The user is advised to consult the UQ International Comparisons Database: UQICD User Guide V3.0 for details of the income distributions included in the data base. Econometric methodology used in the estimation of income distributions is described in the User Guide, and further details can be found in various publications and working papers.

Inequality measures by region are also available.



Select Data Series to Download

View the notes on the countries.

Available countries:



Select by Region:

Select by Income Group:


Selected countries:


Available years:



Selected years:


These provide essential country and year identifiers associated with downloaded data - including country codes, regions, years and currency units (present and historical). Real GDP, real per capita GDP at current and constant prices as well as exchange rate and population data are available to users.

description: World Bank 3-letter code
details: World Bank 3-letter code
source: WDI

description: Year
details: Year
source:

description: Country name
details: country name
source:

description: IMF International Financial Statistics (IFS) Code
details: IMF International Financial Statistics (IFS) Code
source: IFS

description: The currency unit used in 2019
details: This is the currency unit used in 2019. All 'local currency' variables are expressed in 2019 currency unit.
source: Miscellaneous sources

description: The currency code (ISO 4217) by country used in computing the PPP
details: The currency code (ISO 4217) by country used in computing the PPP
source: currency-iso.org

description: Geographical regions
details: Geographical regions as defined by the World Bank
source: World bank

These provide information on: GDP in PPP terms and at current prices and total population.

description: Total population
details: Total population is based on the de facto definition of population, which counts all residents regardless of legal status or citizenship--except for refugees not permanently settled in the country of asylum, who are generally considered as a part of the population of their country of origin. The values shown are midyear estimates (Definition from WDI)
source: UN data base

description: Real GDP per capita in current dollars using UQICD PPP Series
details: Real GDP per capita in current dollars using UQICD PPP Series
source: UQICD estimates

Prameters associated with each of the distributions included in this module - log-normal; generalized beta-2 (GB-2); Pareto-lognormal; and mixture of lognormals - are estimated using procedures described in the appendix in UQICD V3.0 Use Guide.


Lognormal distribtion is characterised by two parameters MU and SIG. This means that log of income is distributed as normal with parameters MU and SIG.

description: Parameter MU of lognormal distribution
details: Lognormal distribtion is characterised by two parameters MU and SIG. This means that log of income is distributed as normal with parameters MU and SIG.
source: Our estimates

description: Standard error of estimate of MU
details: Standard error can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter sigma of lognormal distribution
details: Lognormal distribtion is characterised by two parameters MU and SIG. This means that log of income is distributed as normal with parameters MU and SIG.
source: Our estimates

description: Standard error of estimate of SIG
details: Standard error can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Estimate of mean income
details: Estimate of average income computed using the values of the estimated parameters of the distribution.
source: Our estimates

description: Estimate of median income
details: Estimate of median income, level of income that splits the population (ordered by income level) into two equal groups, computed using the values of the estimated parameters of the distribution.
source: Our estimates

The Pareto distribution is known to fit higher incomes well and in contrast the lognormal distribution fits lower income levels better. The Pareto-lognormal distribution combines virtues of both of these distributions. It is a modification of the lognormal distribution which exhibits Pareto behaviour in the right tail. It is determined by three parameters denoted by MU, SIGMA and ALPHA. The density and distribution functions and other properties are given in UQICD V3.0 User Guide.

description: Parameter MU of the lognormal in Mixed Pareto-lognormal distribution- MU
details: This is the first parameter of the lognormal distribution in the mixture of Pareto and lognormal distributions
source: Our estimates

description: Standard error of estimate of MU
details: This is a measure of reliability of the estimated MU. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter SIGMA of the lognormal in the Mixed Pareto-lognormal distribution- SIG
details: This is the second parameter of the lognormal distributions in the mixture.
source: Our estimates

description: Standard error of estimate of SIG
details: This is a measure of reliability of the estimated SIG. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter ALPHA of Pareto
details: This is the key parameter of the Pareto distribution in the mixed Pareto-lognormal distribution. It determines the mean as well as variance (and hence inequality) in the distribution.
source: Our estimates

description: Standard error of ALPHA
details:
source: Our estimates

description: Estimate of mean income
details: Estimate of average income computed using the values of the estimated parameters of the distribution.
source: Our estimates

description: Estimate of median income
details: Estimate of median income, level of income that splits the population (ordered by income level) into two equal groups, computed using the values of the estimated parameters of the distribution.
source: Our estimates

The GB2 distribution, its mean, variance, and other inequality measures depend on all the four parameters A, B, P and Q. The distribution has been re-parametrized before estimation. The new parameters are A, M, P and Q. The analytical expression relating m to the original parameters can be found in the UQICD User Guide.

description: GB2 - Parameter M
details: The GB2 distribution, its mean, variance, and other inequality measures depend on all the four parameters A, M, P and Q. Refer to UQICD User Guide V3.0 for analytical expressions.
source: Our estimates

description: Standard error of estimate of M of GB2
details: This is a measure of reliability of the estimated Parameter M. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: GB2 - Parameter A
details: The GB2 distribution, its mean, variance, and other inequality measures depend on all the four parameters A, M, P and Q. Refer to UQICD User Guide V3.0 for analytical expressions.
source: Our estimates

description: Standard error of estimate of A of GB2
details: This is a measure of reliability of the estimated Parameter B. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: GB2 - Parameter P
details: The GB2 distribution, its mean, variance, and other inequality measures depend on all the four parameters A, M, P and Q. Refer to UQICD User Guide V3.0 for analytical expressions.
source: Our estimates

description: Standard error of estimate of P of GB2
details: This is a measure of reliability of the estimated Parameter P. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: GB2 - Parameter Q
details: The GB2 distribution, its mean, variance, and other inequality measures depend on all the four parameters A, M, P and Q. Refer to UQICD User Guide V3.0 for analytical expressions.
source: Our estimates

description: Standard error of estimate of Q of GB2
details: This is a measure of reliability of the estimated Parameter Q. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Estimate of mean income
details: Estimate of average income computed using the values of the estimated parameters of the distribution.
source: Our estimates

description: Estimate of median income
details: Estimate of median income, level of income that splits the population into equal groups, computed using the values of the estimated parameters of the distribution.
source: Our estimates

Mixtures of lognormal distributions provide a flexible function form to model income distributions. In UQICD, mixtures of three lognormal distributions with the second and third distributions having same variance, denoted MLN32. This mixture fits data better than the other three distributions included in UQICD. Using mixtures of lognormal distribution has become a common practice. In UQICD we consider mixture of three log-normal distributions. Further technical details about the distribution can be found in UQICD V3.0 User Guide

description: Parameter MU of 1st lognormal component - MU1
details: This is the MU parameter of the first of the three lognormal distributions in the mixture. This distribution is accorded the biggest weight among the distributions.
source: Our estimates

description: Standard error of estimate of MU1
details: This is a measure of reliability of the estimated MU1. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter SIGMA of 1st lognormal component - SIG1
details: This is the SIGMA parameter of the first of the three lognormal distributions in the mixture. This distribution is accorded the biggest weight among the distributions.
source: Our estimates

description: Standard error of estimate of SIG1
details: This is a measure of reliability of the estimated SIG1. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter MU of the 2nd lognormal component - MU2
details: This is the MU parameter of the second of the three lognormal distributions in the mixture. This distribution is accorded the biggest weight among the distributions.
source: Our estimates

description: Standard error of estimate of MU2
details: This is a measure of reliability of the estimated MU2. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter SIGMA of 2nd (and 3rd when applicable) lognormal component - SIG2
details: This is the SIGMA parameter of the second of the three lognormal distributions in the mixture. Note that in the mixture of lognormal distributions used here, variances of the second and third lognormal are the same. This distribution is accorded the second biggest weight among the distributions.
source: Our estimates

description: Standard error of estimate of SIG2
details: This is a measure of reliability of the estimated SIG2. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Parameter MU of 3nd lognormal component - MU3
details: This is the MU parameter of the third of the three lognormal distributions in the mixture. This distribution is accorded the smallest weight among the distributions. Note that for some countries the mixture of lognormal comprises only two lognormal distributions. In that case, "NA" will appear for MU3
source: Our estimates

description: Standard error of estimate of MU3
details: This is a measure of reliability of the estimated MU3. It can be used in constructing confidence intervals and hypothesis testing. Note that for some countries the mixture of lognormal comprises only two lognormal distributions. In that case, "NA" will appear for SE of MU3
source: Our estimates

description: Weight for first lognormal - W1
details: This is the estimated weight accorded to the first lognormal in the mixture.
source: Our estimates

description: Standard error of estimate of W1
details: This is a measure of reliability of the estimated W1. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Weight for second lognormal - W2
details: This is the estimated weight accorded to the second lognormal in the mixture. Note that the weight for the third lognormal can be derived as the three weights add up to 1.
source: Our estimates

description: Standard error of estimate of W2
details: This is a measure of reliability of the estimated W2. It can be used in constructing confidence intervals and hypothesis testing.
source: Our estimates

description: Estimate of mean income
details: Estimate of average income computed using the values of the estimated parameters of the distribution.
source: Our estimates

description: Estimate of median income
details: Estimate of median income, level of income that splits the population into equal groups, computed using the values of the estimated parameters of the distribution.
source: Our estimates

Ginicoefficient is a popular measure of inequality in the distribution of income. It takes values in the range 0 to 1. Zero value indicates perfectly equal distribution whereas a value of 1 indicates extreme inequality. Gini coefficient is equal to one minus the area under the Lorenz curve (shows the share of income of the poorest x% of population for different values of x in the range 0 to 100). See UQICD V3.0 User guide for further details.

description: Gini coefficient for lognormal distribution
details: The Gini coefficient for lognormal is completely determined by parameter SIG. Details about lognormal distribution can be found in UQICD User Guide V3.0.
source: Our estimates

description: Gini coefficient for Pareto-lognormal distribution
details: Gini coefficient for a mixed Pareto-lognormal distributions Details about Pareto-lognormal distribution can be found in UQICD User Guide V3.0.
source: Our estimates

description: Gini coefficient for GB2 distribution
details: The Gini coefficient for generalized beta-2 distribution. Details about GB2 distribution can be found in UQICD User Guide V3.0.
source: Our estimates

description: Gini coefficient for mixture of lognormal distributions
details: The Gini coefficient for lognormal is completely determined by parameter SIG parameters of the components of the mixture of lognormal distributions. Details about mixture of lognormal distributions can be found in UQICD User Guide V3.0.
source: Our estimates

Theil's L index is a member of a class of generalized entropy measure taking values between 0 and infinity (zero implies perfectly equal distribution) determined by a parameter alpha. Low values of alpha means that the inequality measure is more sensitive to lower levels of income. Theil's L index corresponds to a value of alpha equal to zero. It is also known as the Mean Log Deviation. The index is additively decomposable among population sub-groups. See UQICD V3.0 User Guide for further details.

description: Theil's L measure of inequality lognormal distribution
details: This is an additively decomposible measure of inequality. Commonly used in decomposition of inequality into within and between sub-groups of population.
source: Our estimates

description: Theil's L measure of inequality
details: This is an additively decomposible measure of inequality. Commonly used in decomposition of inequality into within and between sub-groups of population.
source: Our estimates

description: Theil's L measure of inequality using GB2 distribution
details: This is an additively decomposible measure of inequality. Commonly used in decomposition of inequality into within and between sub-groups of population.
source: Our estimates

description: Theil's L measure of inequality using mixture of lognormal distributions
details: This is an additively decomposible measure of inequality. Commonly used in decomposition of inequality into within and between sub-groups of population.
source: Our estimates

This is the share of the poorest 10 percent percent of the population. This is computed using the functional form of the income distribution.

description: Share of poorest 10 percent population using lognormal distribution
details: Income share of the poorest 10 percent of the population. This share would be less than or equal to 10 percent.
source: Our estimates

description: Share of poorest 10 percent population using Pareto-lognormal distribution
details: Income share of the poorest 10 percent of the population. This share would be less than or equal to 10 percent.
source: Our estimates

description: Share of poorest 10 percent population using GB2 distribution
details: Income share of the poorest 10 percent of the population. This share would be less than or equal to 10 percent.
source: Our estimates

description: Share of poorest 10 percent population using mixture of lognormal distributions
details: Income share of the poorest 10 percent of the population. This share would be less than or equal to 10 percent.
source: Our estimates

This is the share of the poorest 30 percent percent of the population. This is computed using the functional form of the income distribution.

description: Share of poorest 30 percent population using lognormal distribution
details: Income share of the poorest 30 percent of the population. This share would be less than or equal to 30 percent.
source: Our estimates

description: Share of poorest 30 percent population using Pareto-lognormal distribution
details: Income share of the poorest 30 percent of the population. This share would be less than or equal to 30 percent.
source: Our estimates

description: Share of poorest 30 percent population using GB2 distribution
details: Income share of the poorest 30 percent of the population. This share would be less than or equal to 30 percent.
source: Our estimates

description: Share of poorest 30 percent population using mixture of lognormal distribution
details: Income share of the poorest 30 percent of the population. This share would be less than or equal to 30 percent.
source: Our estimates

This is the share of the richest 10 percent percent of the population. This is computed using the functional form of the income distribution.

description: Share of the richest 1 percent population using lognormal distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 1 percent population using Pareto-lognormal distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 1 percent population using GB2 distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 1 percent population using mixture of lognormal distributions
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

This is the share of the richest 1 percent percent of the population. This is computed using the functional form of the income distribution.

description: Share of the richest 10 percent population using lognormal distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 10 percent population using Pareto-lognormal distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 10 percent population using GB2 distribution
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

description: Share of the richest 10 percent population using mixture of lognormal distributions
details: Income share of the richest or top 10 percent of the population. This share would be generally greater 10 percent.
source: Our estimates

Charts

Cite UQICD

Cite UQICD:

Rao, D.S. Prasada, A. N. Rambaldi, G. Hajargasht, D. Chotikapanich and W.E. Griffiths, UQ International Comparisons Database: UQICD V3.0, School of Economics, The University of Queensland, St Lucia, QLD 4072, Australia. 2022.