The starting point for the analysis is household survey-based estimates of regional income distributions from the World Bank’s PovcalNet database. The global regions are as defined by the World Bank. In this report, “advanced economies” refers to the group of countries listed as “other high income” by the World Bank.
These data were used to estimate a benchmark distribution of the population in each of seven major global regions across five income tiers. The income tiers are defined by the daily per capita income or consumption of people in a region as follows: poor ($2 or less daily), low income ($2.01-$10), middle income ($10.01-$20), upper-middle income ($20.01-$50) and high income (more than $50). All dollar figures are expressed in 2011 prices and purchasing power parity dollars.
The year for which the benchmark estimates were drawn was the latest one in which the household surveys covered at least 75% of the region’s population. This varied across regions as follows: 2018 in advanced economies, Latin America and the Caribbean, East Asia and the Pacific, and Europe and Central Asia; 2017 in Sub-Saharan Africa; 2015 in the Middle East and North Africa; and 2014 for South Asia.
Extrapolating the benchmark estimates to 2020 required the following three steps:
1. Extrapolate benchmark-year incomes to 2020
2. Estimate income distributions in 2020 based on extrapolated incomes
3. Determine the population in each income tier in 2020
Extrapolating incomes from benchmark years to 2020
Household incomes are extrapolated to 2020 using estimates of the growth in real gross domestic product (GDP) per capita in a region. This, in turn, required several sets of data.
The first is the level of GDP per capita in the benchmark year, say 2017 for Sub-Saharan Africa, and in 2019, the latest year for which data are available in the World Bank’s World Development Indicators database. In this database, GDP per capita is expressed in constant 2017 international dollars. The result is an estimate of the percentage change in per capita income in a region from the benchmark year to 2019.
Next, the change in income from 2019 to 2020 is approximated by two measures of growth in real GDP per capita (at market prices) issued by the World Bank in its reports on global economic prospects. The first measure of growth in 2020 is a forecast released in January 2020, prior to the emergence of the pandemic. The second measure is an estimate released in January 2021, with the economic effects of the pandemic more apparent. The result is two estimates of how much per capita income changed in each region from the benchmark year to 2020, one in a counterfactual “pre-pandemic world” and the other in a “post-pandemic” world, the period since the emergence of the coronavirus.
In a slight variation to this procedure, the change in per capita income in 2020 for advanced economies is estimated as the difference in the growth in GDP and the growth in population. This is because estimates of the change in GDP per capita in 2020 for this group of countries is not directly reported in the source data.
Estimating income distributions in 2020 based on extrapolated incomes
The estimated change in income from the benchmark year to 2020 is used to reset the boundaries of the income tiers. For example, based on the pre-pandemic forecast, it was estimated that per capita income in East Asia and the Pacific would have increased by 10.5% from 2018 to 2020. As a result, people living on between $1.82 to $2 a day in East Asia and the Pacific in 2018 would have seen their income increase to more than $2 a day in 2020. Thus, they would no longer be poor in 2020; only people with incomes of $1.81 or less in 2018 would still be poor in 2020. Thus, $1.81 is used as the poverty boundary to derive an estimate of the share who were poor in East Asia and the Pacific in 2020 in the hypothetical pandemic-free world.
With the emergence of the pandemic, the growth in per capita income in East Asia and the Pacific in 2020 is estimated to have been 5.6%. In this circumstance, only people with incomes from $1.90 to $2 a day in 2018 would have emerged from poverty in 2020. Thus, $1.89 is used as the poverty boundary to estimate the state of poverty in East Asia and the Pacific in 2020 in the post-pandemic scenario.
A similar procedure is used to reset the boundaries of all income tiers in all regions. The estimation of income distributions for 2020 was greatly facilitated by the “povcalnet” command in Stata developed by World Bank researchers. (Castaneda Aguilar, R.A., C. Lakner, E.B. Prydz, J.S. Lopez, R. Wu and Q. Zhao, Estimating Global Poverty in Stata: The Povcalnet Command, September 2019).
Determining the population in each income tier in 2020
The estimates derived from PovcalNet yield the percentages of people in a region who are poor, low income and so on. These percentages are applied to the estimated population of each region to determine the number who fall into each income tier.
The overall population in a region in 2020 is derived from two sources: World Bank data on population through 2019, as listed on PovcalNet, and United Nations forecasts of the change in population from 2019 to 2020. This exercise is conducted at the country level, some 218 in all, and the results are summed to obtain an estimate of the population in each region in 2020.
Key assumptions and potential uncertainties
As is the case with any projection, the estimates reported in this analysis are subject to a degree of uncertainty. One issue is that final end-of-year estimates of real GDP growth are not yet available for most countries, if any. Thus, the January 2021 estimates from the World Bank that are used in the analysis are likely to be revised in the coming months. In its projections of the increase in extreme poverty ($1.90 or less a day) in 2020, the World Bank assumes a global contraction of either 5% or 8%. The resulting change in extreme poverty in 2020 is estimated to range from 88 million to 115 million.
The methodology in this analysis assumes that incomes change at the same rate for all people, whether poor, low income, middle income, upper-middle income or high income. In other words, it is assumed that there is no change in inequality from the benchmark year to 2020. If the COVID-19 recession has worsened inequality, the increase in the number of poor is likely greater than estimated in this analysis, and the decrease in the number who are high income is likely less than estimated. The middle class may have shrunk by more than projected.
Although World Bank researchers have simulated the relationship between inequality and the poverty rate, more precise measures of changes in inequality in the COVID-19 era will not be forthcoming until newer household survey data become available. Also, the extent to which social spending by governments, especially in advanced economies, may have ameliorated the effects of the COVID-19 downturn is not yet clear.
Another assumption in this analysis is that personal incomes change at the same rate as GDP per capita. Empirically, it is not unusual to observe that personal income or consumption increases at a slower rate than GDP per capita. However, the relationship between the two measures varies widely across countries and it is not clear what the relative rates of growth might be at the regional or global level. The World Bank often assumes a “pass-through” rate of 0.85, i.e., only 85% of national output growth is passed through to household income or consumption. If a pass-through rate of 0.85 is assumed for this analysis the estimated change in the number of people in each income tier is as follows: Poor (+117 million), low income (+11 million), middle income (-45 million), upper-middle income (-30 million) and high income (-53 million).
From the point of view of measuring global poverty, a key issue is that the latest available benchmark survey for India was conducted in 2011. As the second most populous country in the world, India accounts for a large share of the global population in poverty. In 2011, 969 million people lived in extreme poverty globally, on $1.90 or less daily, and 285 million of that number were from India. The World Bank has applied a variety of estimation techniques to update the poverty counts for India, but uncertainty lingers in these updates for India and, by implication, in current measures of global poverty.