Published on Data Blog

Why have the 2011 PPPs been revised and what does it mean for estimates of poverty?

This page in:

In May 2020, the International Comparison Program (ICP) revised its 2011 purchasing power parities (PPPs), which were originally published in 2014, together with publishing its most recent 2017 PPPs. This was the first time that previously published ICP PPPs were revised (further details on the ICP Revision Policy).

The different input data used by the World Bank for its global poverty estimation, such as consumer price indices (CPIs), gross domestic product (GDP) expenditures or population data, are periodically revised to reflect the most up-to-date and accurate information available. In a similar fashion, the revised 2011 PPPs were adopted for the global poverty measures in the PovcalNet update published in September 2020 (also included in the Poverty and Shared Prosperity report 2020). The use of the revised 2011 PPPs led to slight changes in estimates of the global and regional extreme poverty rate, defined as the share of the population living on less than $1.90 a day in 2011 PPP terms. Extreme poverty in the world in 2017 increased by 0.24 percentage points, equivalent to an additional 17.7 million poor people. The largest change in the extreme poverty rate in 2017 was observed in Sub-Saharan Africa (an increase of 1.1 percentage points, representing 11.4 million more poor people). This paper and associated blog provide more details on the impact of the revisions in the 2011 PPPs on estimates of global and regional poverty. The main aim of this blog is to shed light on the factors underlying the revisions to the 2011 PPPs at the country level. While the changes at the global and regional level might be small, the country-level changes can be more significant, as discussed below.


Impact of PPP revisions on country poverty rates at different poverty lines

The figure below compares the latest poverty headcount rates for countries estimated using both the original and revised 2011 PPPs. Estimates are provided at the three poverty lines used by the World Bank in its global poverty measurement—the international poverty line of $1.90 a day, as well as $3.20 and $5.50 a day, which are more typical of the national poverty lines found in lower- and upper-middle income countries, respectively. Note that for each country, the figure plots the latest survey year, and hence the poverty rates are difficult to compare across countries. Poverty estimates are based on either income or consumption expenditure, which also creates comparability issues. The chart view can be filtered by both poverty line and region to allow easy exploration of the data. The size of the bubble for each country is proportional to the (absolute) change in the number of poor as a result of the revision in the 2011 PPPs. Countries close to the 45-degree line saw the smallest changes in their poverty rate as a result of the revision.

Figure 1: Poverty rates with original and revised 2011 PPPs

Revisions to the ICP 2011 PPPs

We have grouped the drivers behind the revisions to the 2011 PPPs into four main factors, which are illustrated with country examples. It is important to note that the revision in a particular country may be due to several factors that are happening at the same time.

First, the revisions to the 2011 PPPs are primarily driven by revisions to national accounts expenditures.  National accounts expenditures are used as weights when basic heading PPPs are aggregated to estimate PPPs at higher levels of ICP expenditure classification. Examples of the ICP expenditure classification headings are rice (a basic heading), bread and cereals (a class), food (a group), food and alcoholic beverages (a category) and individual consumption by households (a main aggregate), for details on the ICP classification see Rissanen and Song (2020)Expenditure data revisions are a common practice as countries routinely revise their national accounts data to incorporate new or revised information. In general, revisions are a common practice in official statistics.  Larger revisions are sometimes required because of methodological changes due to, for example, changes to GDP base year or the adoption of a new United Nations System of National Accounts (SNA) version. The 1993 SNA was used as the conceptual framework to classify expenditures for the original 2011 ICP cycle, while the 2008 SNA (the latest version of the SNA) was used in the revised 2011 PPPs. Revisions to national accounts expenditures are the primary reasons for many of the changes in the Sub-Saharan African countries, such as Angola, Cameroon, Comoros, Liberia, Nigeria, São Tomé and Príncipe, Seychelles, and Sierra Leone.

Second, some 2011 PPPs were revised due to changes in the underlying price data. For instance, the Eurostat-OECD PPP Programme follows a rolling survey approach, in which prices are collected over a three-year period, resulting in revisions to price data, or other data, used to calculate PPPs for a particular reference year. The changes in PPPs for Albania and the Russian Federation are partly explained by this factor. Furthermore, additional quality assurance has been conducted on the original price data collected for the ICP 2011, based on comparisons with price data collected in subsequent years. This applies to selected ICP regions, such as Asia and the Pacific and Western Asia, and has resulted in revisions in the underlying price data in Fiji, Iraq, the Maldives and Myanmar, for example.

Third, the model that is used to impute PPPs for a selection of countries that do not participate in the ICP has been updated with new input data. The same regression model used in the original ICP 2011 has been used for the revision, but has been re-estimated with revised input data such as GDP per capita, import and export shares, and the age dependency ratio. More details on the model used for nonparticipating countries are available in the section PPPs for nonparticipating economies of Chapter 5 of the ICP 2017 Report. This explains the revisions for Kosovo, South Sudan, and Turkmenistan. In addition, the revised 2011 PPPs for Argentina and Guyana are based on extrapolations from 2017, instead of imputations, as the two countries joined the ICP in 2017. 

Lastly, because PPPs are multilateral price indexes, revisions for one country (e.g. as a result of changes in expenditure shares and/or changes in the price data) will impact other countries’ PPPs, although this effect is typically small.  Furthermore, PPPs are first estimated at a regional level, while the regional PPPs are subsequently linked together to form a global set of PPPs, typically, based on prices collected in all regions (for more details see the section PPP calculation and estimation, Chapter 5, ICP 2017 Report). Special linking approaches for selected regions, including the Commonwealth of Independent States (CIS) explains the revisions in Armenia, Kazakhstan, Kyrgyz Republic, and Tajikistan. These countries are linked to the global comparison through the Russian Federation, whose PPP estimates have been revised because of revisions to the structure of national accounts expenditures or other input data.

The World Bank has adopted the revised 2011 PPPs into its global poverty measures published in PovcalNet (September 2020 update) and included in the Poverty and Shared Prosperity report 2020. The 2017 PPPs, which were published together with the revised 2011 PPPs in May 2020, are not currently used in the poverty measures. The 2017 PPPs are being analyzed and their possible use in the World Bank’s poverty measurement is being assessed.


We gratefully acknowledge financial support from the UK government through the Data and Evidence for Tackling Extreme Poverty (DEEP) Research Programme. This blog has been edited by Edie Purdie of the ICP team. The ICP team encourages users to share their data applications and findings using ICP data via


Samuel Kofi Tetteh Baah

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

Aziz Atamanov

Senior Economist, Poverty Global Practice, World Bank

Christoph Lakner

Program Manager, Development Data Group, World Bank

Daniel Gerszon Mahler

Senior Economist, Development Data Group, World Bank

Marko Rissanen

Program Manager, Development Data Group, World Bank

William Vigil-Oliver

International Comparison Program (ICP), The World Bank

Mizuki Yamanaka

Senior Statistician, Development Data Group

Join the Conversation

The content of this field is kept private and will not be shown publicly
Remaining characters: 1000