ľ¹ÏÓ°Ôº

Marko Rissanen and Robert Inklaar

September 2023

The concept of purchasing power parities (PPPs) is well-known and widely used within economic analyses. PPPs control for the differences in price levels between economies and equalize the purchasing power of currencies. In this way, PPPs show the relative price of a given basket of goods and services in each of the compared economies, with reference to a base economy. The primary use of PPPs is to convert volume and per capita measures of gross domestic product (GDP) and its expenditure components into a common currency, while also controlling for differences in price levels between economies. PPPs and PPP-based indicators are key tools in data-led policy making and are considered a global public good. ľ¹ÏÓ°Ôº publication Purchasing Power Parities for Policy Making: A Visual Guide to Using Data from the International Comparison Program provides a comprehensive account of the myriad applications of PPPs and elaborates on their limitations. They are used in administrative applications by public and private organizations, including the European Union, the International Monetary Fund (IMF), and the World Bank, among others.

Estimates of PPPs are available from several different sources. This knowledge brief identifies the better-known sources of PPP datasets and discusses their commonalities and differences. We compare the International Comparison Program (ICP) PPPs with those published by (i) the World Bank¡¯s (WDI); (ii) the (PWT); (iii) the IMF¡¯s (WEO), and (iv) the Central Intelligence Agency¡¯s (CIA) . The ICP is chosen as the basis for the comparison given its status as a statistical program under the United Nations and also as other datasets base their PPP estimates on the ICP results or its underlying data.

[ Back to top ]

Overview of global PPP datasets

The table below summarizes the available global PPP datasets and their main features.

Summary of global PPP datasets published by ICP, PWT, WDI, WEO, and the CIA World Factbook

1/ ICP Global Office has released benchmark estimates for 2005, 2011 and 2017, and PPP time series estimates for years in between 2011-2017. The 2021 ICP results are expected to be released in early 2024.

2/ Forecasts are included in the dataset.

[ Back to top ]

The ICP produces and disseminates global [1] PPPs for 44 detailed GDP expenditure categories, including GDP, actual individual consumption, household consumption, food and non-alcoholic beverages, clothing and footwear, transport, etc., for specific ICP benchmark years. In addition, the ICP produces global PPPs for a more limited number of categories for years between the benchmark years.

The WDI publishes PPPs at the level of and , and also extrapolates PPPs beyond the benchmark years. PWT, in turn, estimates PPPs for GDP and its main components, and provides the longest PPP time-series dating back to 1950. The IMF¡¯s WEO publishes PPPs at the GDP level and extrapolates and forecasts PPPs. The CIA World Factbook publishes PPPs at the GDP level, however, many estimates are provided for countries that are otherwise absent from other databases.

As noted above, all listed providers of PPPs utilize ICP data as the main source of information. Columns [4] to [6] in the table assess the extent to which each providers uses (i) price data collected by the ICP, (ii) GDP expenditure data compiled by the ICP, and (iii) regional and global PPPs published by the ICP. All datasets, other than the ICP, use the latter, that is published ICP PPPs, and make no direct use of the ICP¡¯s price and expenditure data. [2]

In terms of frequency of publication, columns [8] and [9] in the table show that WDI, WEO, and the CIA World Factbook publish PPPs at a higher frequency than ICP and PWT. The calculation of PPPs for the latter two involves more methodological steps and requires more details than the former three, and thus takes longer to produce, resulting in a less frequent publication timetable.

[ Back to top ]

Global PPP datasets in more detail

This section reviews each PPP dataset in more detail. In doing so, it compares global PPP datasets along different dimensions, such as (i) the publishing institutions¡¯ motivation or goal for producing PPPs; (ii) aspects related to methodology and/or input data; and (iii) the publishing institutions¡¯ publication timeline and frequency, including revision policies or lack thereof.

1.     International Comparison Program (ICP)

The ICP is the only PPP data source that collects and compiles national input data for PPP estimation. These data are provided by ICP national implementing agencies working within a well-established governance framework and structure. The program is coordinated by the World Bank, under the auspices of the United Nations Statistical Commission (UNSC), and implemented at the regional level through the (AfDB), the (ADB), the (CIS-STAT),  the (UN-ECLAC), the (UN-ESCWA), the (Eurostat), and the (OECD).

The ICP¡¯s main purpose is to produce global PPPs, price level indices, and PPP-based expenditures, in order to measure the global economy in both aggregate and per capita form. ICP PPPs are calculated based on item-level price data for a common basket of goods and services collected through surveys, and detailed expenditure weights on groups of items in each of the participating economies. All data collected and compiled is benchmarked to a reference year for each comparison cycle, which is currently with an interval of 3 to 4 years. In addition to releasing PPPs for a new benchmark year in each cycle, the ICP revises previous benchmark year results based on the latest available data. The World Bank and the regional implementing agencies, linked above, publish ICP results and PPPs on their respective websites and data portals.  

As a widely used global public good, ICP follows an established statistical methodology, with methodological consistency between benchmark years a priority of the ICP. The sets out the policy, triggers, and procedures for revising previously published and unpublished results from the ICP, to ensure the program¡¯s outputs reflect the latest available information and methodologies, are of the highest quality, and remain relevant to users.

[ Back to top ]

2.     World Development Indicators (WDI)

(WDI) is the World Bank¡¯s flagship database of cross-country comparable development data. It features global PPPs and PPP-based indicators from 1990 onwards. WDI extrapolates ICP PPPs, currently provided for 2011-2017, for years not covered by the ICP. For years before 2011 and after 2017, PPPs provided by the ICP are extrapolated to create a time series from 1990 to the latest year available.

In order to , WDI applies inflation observed in an economy relative to that of the base economy, the United States, over the same period, onto the benchmark PPP estimates. GDP deflators are used to extrapolate , while consumer price indexes (CPIs) are used to extrapolate household consumption, or , -level PPPs. As such, the WDI PPPs are the result of extrapolating global ICP PPPs to cover years other than those covered by the ICP.

GDP-level PPPs in the WDI database are updated twice a year in line with updates of national accounts data. Private consumption PPPs are updated a few times a year, upon release of new or revised CPI data. In addition, the most recent results of the annual - PPP Programme are regularly incorporated into WDI data on a frequent basis.

[ Back to top ]

3.     Penn World Table (PWT)

The was originally developed at the University of Pennsylvania in the 1960s [3] with the main purpose of extending coverage of ICP PPPs to additional countries and years. Data from later ICP comparisons and longer time series were subsequently integrated and enabled academic research on the cross-country sources of growth and development.

Currently, PWT is managed by the University of Groningen and the University of California, Davis. The institutional background and intended audience of the PWT differs from that of the ICP and leads to differences in methodological choices. While the ICP operates as an official statistical program under a well-established governance framework, the PWT is an independent academic research project that undertakes experimental developments. These approaches are complementary, as the PWT depends on the ICP institutional framework to collect reliable data, while the ICP benefits from the PWT experimentation to improve its own methodology. The main methodological differences between the ICP and the PWT lie in PPP aggregation methods, treatment of regional data, treatment of trade balance, utilization of additional data sources, and treatment of historical ICP data, among others.

As of 2023, PWT is one of the largest databases on relative levels of income, output, input, and productivity. The latest version, , released in 2023, covers up to 183 economies from 1950 to 2019 and incorporates historical ICP data from 1970 to 2017. Regarding revisions, PWT does not follow a strict revision schedule, however, it incorporates a new ICP benchmark in its major revisions.

[ Back to top ]

4.     World Economic Outlook (WEO)

The publication (WEO) presents the IMF¡¯s analyses of global economic developments in the near and medium term, including projections. It is released twice a year at the IMF¡¯s Spring and Fall meetings. The report includes a containing a wide set of different macroeconomic indicators covering several years, including three series that use global ICP PPPs: PPP-adjusted GDP per capita in current and constant prices, and PPP-adjusted GDP in current prices.

The WEO extrapolates ICP PPPs backwards from 2011 and forwards from 2017 by employing the same extrapolation technique as the WDI, that is, using the growth in relative GDP deflators between each economy and the United States.[4] The source of the underlying GDP deflators is the IMF¡¯s own data and estimates, and hence the resulting extrapolated PPPs may differ from those used in the WDI. The WEO also publishes PPP forecasts based on its own .

[ Back to top ]

5.     The Central Intelligence Agency (CIA) World Factbook

The Central Intelligence Agency (CIA) produces and its for use by United States government officials based on information collected from ¨C and coordinated with ¨C a wide variety of United States Government agencies and from hundreds of published sources. It provides a wide range of information for economies and territories. The economy section of the World Factbook presents PPP-based GDP figures for economies around the world.

The CIA mainly extrapolates PPPs from ICP benchmark results, however, based on the , the CIA does not exclusively use ICP PPPs. Further details on the sources for the economic data used are not publicly provided, as noted by the CIA: ¡°The Factbook staff uses many different sources to publish what we judge are the most reliable and consistent data for any particular category.¡±

[ Back to top ]

Concluding notes

This knowledge brief outlines the commonalities and differences between the PPPs disseminated by different sources (ICP, WDI, PWT, IMF, and CIA) with the purpose of supporting users on their use and application of PPPs. Although all these sources are based on ICP PPPs, there are important differences that users should be aware of and the choice for which source to use ultimately depends on the intended use and the user¡¯s discretion.

Looking forward, data from the ongoing ICP cycle, benchmarked to the year 2021, will be released in 2024. The knowledge brief will be kept updated upon this and future releases of the ICP, as well as to reflect any changes in PPP data sources.

[ Back to top ]

[1] Note that ¡°Global PPPs¡± refers to PPP datasets covering the seven World Bank geographical regions. Furthermore, this knowledge brief focuses only on the global PPPs provided by each data source. It will not examine differences resulting from the source or vintage of national accounts and/or population data used by each dataset. Users of each dataset will often access PPP-based national accounts data, and not PPPs, which are not always directly published. These PPP-based national accounts data, such as PPP-based GDP per capita, may differ from one source to another not only because of the PPPs used, but also due to differences in national accounts and/or population data.

[2] It is noted, however, that the PWT computations start from the lowest level of aggregation, the basic heading level, utilizing global PPPs and expenditure data (i.e., below published level), and then aggregating basic heading level data up to GDP and its main aggregates level.

[3]  provides a comprehensive account of the history of the ICP.

[4] See Section 4 of the .