Purchasing Power Parities

An alternative approach to convert measures of income from national currencies into a common currency is by using conversion factors that reflect the purchasing power of currencies -- Purchasing Power Parities (PPP).

PPP eliminates the inconsistencies inherent in exchange rate conversions, which are sometimes volatile and fail to reflect properly the differences in price levels between countries - particularly with respect to non-traded items. (Devaluation of a country’s currency will reduce its GDP in US$ over night, but it does not make the citizens less well off unless they buy imported goods. The exchange rate is the price on foreign currency, and is relevant for actual transfers across the border, but it is not too relevant for the part of GDP that does not enter international trade.)

The PPP rate is defined as the number of units of a country’s currency that is required to buy the same amount of goods and services in the country as one US$ would buy in the US. PPP as a rate of conversion ensures that money exchanged for a dollar buys the same volume of goods and services in every country. By equalizing prices, PPP rates deliver a measure of relative GDP which is based on what constitutes "real" income, the volume of goods and services embodied in GDP. The method of using PPP is analogous to measuring GDP in different years at fixed base year prices.

As to actual PPP data, there are concerns related to coverage, continuity and timeliness of surveys, quality of results and aggregation procedures.

The World Bank does not use PPP converted data for administrative purposes. For setting the terms for lending at the World Bank, the atlas method is used to convert income from local currencies to a common currency (US$). However, the World Bank uses available PPP-based numbers for analytical and poverty reduction policy purposes, as demonstrated in recent editions of the World Development Report and the World Development Indicators.

PPP rates can be derived using several methods, each yielding different estimates. PPP rates are estimated on the basis of data from special price surveys. Price ratios of comparable items between countries are computed, and aggregated using corresponding weights based on GDP expenditure data. Several methods of aggregation exist, and there is no universal agreement as to which is superior - it depends on the purpose.

PPPs can be used either for binary comparisons, or for comparison of a group of countries. Binary comparisons between pairs of countries are obtained by computing the “ideal” index, the Fisher index. However, the Fisher index is not transitive, thus, other methods are (should be) used for multilateral comparisons. (Transitivity means that comparing country A with C directly should give the same result as comparing country A with B and C with B -- making the comparison of A and C indirectly.)

The two most commonly used methods of aggregation in multilateral comparisons are (i) the Geary-Khamis and (ii) the Elteto, Koves and Szulc, which both produce transitive and base-country invariant results.

(i) The Geary-Khamis method involves using observed price and expenditure data to obtain implicit quantity estimates, and evaluate these quantities at a single set of average “international prices” denominated in a common currency, like the “international dollar”.
(An “international dollar” has the same purchasing power as an US$ for total GDP in the US, but the purchasing power of the components are determined by the average international price structure, not the US price relatives.)

(ii) The Elteto, Koves and Szulc, involves a two-step process. First step is to get a set of binary Fisher indexes for all pairs of countries, and step two is to make these comparisons transitive by computing geometric means of all the direct and indirect indexes.

The results using the Geary-Khamis method will generally differ both in ranking and level compared to the results using the Elteto, Koves and Szulc method. The Geary-Khamis method has one advantage over the Elteto, Koves and Szulc method: it is additive. This means that components can be added to reach a total, making it possible to add expenditure at “international prices” to reach GDP at international prices. Thus, the use of the Geary-Khamis method makes it possible to put up an internally consistent set of national accounts data at “international prices”. However, the Geary-Khamis method tends to result in inflated quantity estimates for poorer countries.

International Comparison Programme

The United Nations International Comparison Programme (ICP) was launched in 1968 as a worldwide effort to compare country income levels on a purchasing power adjusted basis. The initiative has been developed under the guidance of a group from the University of Pennsylvania in cooperation with international agencies, including the United Nations Statistical Office, the Statistical Office of the European Union, the Statistical Office of the Organization of Economic Cooperation and Development and the World Bank. The work of the ICP has been carried out in several faces over a long time, with compilation of data for various benchmark years.

Furthermore, effort has been made to develop shortcut and reduced information methods to extend real income comparisons to countries which may not be able to participate in the benchmark work and to obtain annual estimates for inter-benchmark years.

The ICP-methods

In ICP, GDP by expenditure is divided into 150 (or more) groups/basic headings, in which all items selected for pricing are classified. Out of a list of specifications containing approximately 2000 items, each country collects the prices on 3 or more items per basic heading. An average price for each group / basic heading is estimated, in order to derive price ratios for each of these groups (un-weighted parities). These price ratios are then aggregated, a procedure which involves weighting and summing up, to arrive at PPPs and "real" expenditures for each category of expenditure up to the level of GDP. For the purpose of aggregation, two alternative methods of aggregation can be / are used in the ICP programme, the Geary-Khamis and the Elteto, Koves and Szulc. As mentioned earlier, both methods produce multilateral, rather than binary comparisons, and both methods lead to transitive and base-country invariant results.

Special price surveys are conducted at about five year intervals, and have so far covered about 90 countries at one point of time. However, while surveys have been regular and more or less complete in the industrialized countries, surveys for developing countries and countries in transition have been less regular and of more uncertain quality. Since not all countries have participated in all years, it has been necessary to extrapolate to construct an array for a given year for all the 90 countries. Furthermore, a recourse has been taken to shortcut procedures for the estimation of numbers for countries that have not participated in the ICP.

As PPP numbers are available every fifth year, PPP converted data like GDP and private consumption expenditure are extrapolated using real growth of the respective countries adjusted for US inflation. However, using this method, the estimates never match data based on a new benchmark year (new price survey), and there is no reason they should. This method allows prices only to change, while PPP-estimates include the effect of changes in quantities as well. This extrapolation method is also used whenever figures for a country is not available for a new benchmark year. Using this method means that data are less reliable the further away from the last benchmark year.

In addition to extrapolation over time, it has also been necessary to come up with ICP or PPP numbers for countries not participating in the ICP programme. The usual method has been to use regression techniques to establish a structural relationship between ICP-type estimates on the one hand and a set of easily observable explanatory variables on the other. The World Bank has in World Development Indicators, World Development Report and World Bank Atlas published PPP-estimates derived on the basis of two independent variables; Atlas estimates of per capita GDP and secondary school enrollment. Such short-cut estimates are useful for analytical purposes, even though the estimates can have large residual errors.

Despite limitations, the ICP-database constitutes the largest single source of data on international comparisons of price and expenditure patterns.