G20 consumer price index
From Statistics Explained
This article presents the backgrounds and methodology of the G20 CPI, a consumer price index (CPI) intended to be a timely measure of inflation for the G20 countries. It provides some in-depth technical information on the compilation of the G20 CPI all-items and highlights comparability issues of CPIs across G20 economies.
The G20 CPI is part of the G20 Data Gaps Initiative (DGI) aimed at compiling zone aggregates for the G20 as a whole. It is produced by the OECD and released as part of OECD's regular monthly 'News Release on Consumer Price Index' at around one month after the reference period.
The first results of the G20 consumer price index were released through a Eurostat Press release on 14 October 2013, reporting on figures for August 2013.
In the aftermath of the 2008 global financial crisis, several data gaps have been recognised. In response, twenty recommendations to address the data gaps and improve statistical information were agreed upon by the G20 Finance Ministers and Central Bank Governors: the so-called G20 Data Gaps Initiative (DGI). Progress on the recommendations is coordinated and monitored by the Inter-Agency Group on Economic and Financial Statistics (IAG). The group consists of the Bank for International Settlements (BIS), the European Central Bank (ECB), Eurostat, the International Monetary Fund (IMF, chair), the Organisation for Economic Co-operation and Development (OECD), the United Nations and the World Bank. Annual progress reports are provided by the Financial Stability Board (FSB) Secretariat and the IMF staff to meetings of the G20 Finance Ministers and Central Bank Governors. Economic statistics for the G20 economies (and 10 non-G20 economies with systemically important financial sectors) are published on a dedicated Principal Global Indicators website, established by the IAG.
One of the goals of the G20 DGI is the compilation of zone-aggregates for the G20 as a total. As a first step, quarterly economic growth for the G20 aggregate has been compiled by the OECD following a methodology agreed by the IAG and is part of the regular OECD quarterly News Release on G20 GDP growth at around 70 days after the reference quarter.
Data sources and availability
The OECD will publish the CPI G20 aggregate each month as part of the OECD CPI News Release following the release of national CPI and HICP data by OECD and G20 economies (expected at the end of each month following the end of the reference period or the early beginning of the following month).
Each month, the CPI all-items for the G20 aggregate and the G20 economies will be published in two ways:
- as part of the OECD CPI Press Release of the current month, and
- on the following websites:
Compilation of the CPI all-items G20 aggregate
The G20 consists of the following economies: Argentina, Australia, Brazil, Canada, People’s Republic of China (China), France, Germany, India, Indonesia, Italy, Japan, Republic of Korea (Korea), Mexico, the Russian Federation, Saudi Arabia, South Africa, Turkey, the United Kingdom, the United States, and the European Union (EU).
The G20 CPI aggregate is calculated as a weighted average of the CPIs of the fifteen countries above that are not part of the European Union (that is, the fifteen individual country members of the G20 other than France, Germany, Italy and the United Kingdom) and the harmonised index of consumer prices (HICP) for the European Union.
For the fourteen countries that are not part of the European Union (except Turkey) only a national CPI is available while for the European Union, its Member States and for Turkey Harmonised Indices of Consumer Prices (HICPs) are available (note that for the European countries - France, Germany, Italy, and the United Kingdom - both national CPIs and HICPs are available).
Given their comparability, HICPs are used for the compilation of the G20 CPI for Turkey as well as for the European Union (including France, Germany, Italy and the United Kingdom), while the national CPIs are used for the other countries.
Table 1 shows the official sources of national CPIs and HICPs: National CPIs for the G20 economies are mainly sourced from National Statistics Offices (NSOs), while HICPs for Turkey and the European Union are directly sourced from Eurostat, which in turn receives the data from the NSOs.
The national CPI data are published according to national schedules (the IMF provides the advance CPI release calendar for countries subscribing to the Special Data Dissemination Standard-SDDS). The HICP for the European Union is in general released by the middle of the month following the reference period.
Revisions policy and reliability
National CPI and HICP data of G20 economies are final at the date of the publication and are usually not revised. Still, HICP data, including back data, are revisable under the terms of an explicit revision policy.
However, national CPI and HICP data may be exceptionally revised when major changes in methodology or underlying structural data are implemented. In this respect, it can be noted that the weight reference periods of the national indices are subject to change – often referred to as a rebasing – every five years on average in G20 economies. National providers also typically use this rebasing to introduce methodological changes and/or a new basket of products, and these exceptionally lead to revisions of the respective CPI series (HICPs should not be revised when new baskets are introduced).
The OECD will publish the CPI G20 aggregate each month as part of the OECD CPI News Release following the release of national CPI and HICP data by OECD and G20 economies (expected at the end of each month following the end of the reference period or the early beginning of the following month). However, some countries’ CPI data may not be available at the time of the release. India, for example, publishes its CPI on the last day of the month following the reporting period, and Saudi Arabia publishes its CPI with a time lag of two months in comparison to one month for other G20 economies. The current period CPI for countries whose CPI is unavailable will be imputed using the weighted average of the percentage CPI change of the previous period covering all the members whose CPI are available. Usually only the CPI data for Saudi Arabia are missing and the coverage is approximately 99.4 percent of G20 household consumption expenditure. The individual imputed values used for calculation of the G20 aggregate are not published.
When it comes to reliability, the primary goal of compiling and publishing G20 aggregates is the provision of indicators which sufficiently portray the reality of the overall economic development in the G20 area, in line with the principles of the quality frameworks subscribed by the international and supranational organisations constituting the IAG. This is necessary to ensure that G20 aggregates are fit for policy use and allow for a comparison with the average for the area under consideration as well as comparing developments across countries. From these goals, it follows that the IAG would want to arrive at the best possible estimates for the G20 area, using internationally-comparable data from the countries constituting the relevant area.
In this context, the IMF has issued a declaration of censure in connection with Argentina’s breach of obligation to the Fund under the Articles of Agreement with regard to the inaccurate reporting of its official data (please see the IMF press release dated February 1, 2013). Alternative estimates of CPI inflation point to considerably higher inflation rates than the official data since 2007. However, given the absence of a single validated alternative measure and the small weight of Argentina in the G20 CPI, the compilation of the G20 aggregate for the CPI uses Argentina’s officially reported data (each deviation in CPI inflation of 10 percentage points, for example, will affect the G20 aggregate inflation by about 0.1 percentage point). The impact of using the official data on the CPI G20 aggregate is communicated with the following footnote: The IMF has issued a declaration of censure and called on Argentina to adopt remedial measures to address the quality of the official CPI data. However, given the absence of a single validated alternative measure and the small weight of Argentina in the G20, the official data of Argentina have been included in the G20 aggregate (each deviation in Argentina’s CPI inflation of 10 percentage points, for example, will affect G20 aggregate inflation only by about 0.1 percentage point).
Regarding national CPI for Argentina, it has been agreed to publish the relevant official data with the following footnote: The data for Argentina are officially reported data. The IMF has, however, issued a declaration of censure and called on Argentina to adopt remedial measures to address the quality of the official CPI-GBA data. Alternative data sources point to considerably higher inflation rates than the official data since 2007. In this context, the Fund is also using alternative estimates of CPI inflation for the surveillance of macroeconomic developments in Argentina. A similar footnote is added to the relevant data series in the country page on the Principal Global Indicators (PGI) website.
Calculation of the CPI all-items for the G20 aggregate
The CPI for the G20 aggregate has been calculated for the headline indicator only (CPI/HICP all-items). For reasons of comparability and availability of data, calculation of a G20 aggregate for CPI Food, CPI Energy, CPI all-items less food less energy is currently not feasible (available CPI data for food for Brazil, India, China, and Russian Federation are not defined according to COICOP classification and are not fully comparable with COICOP 01. CPI data for energy are not currently available for Argentina, Brazil, India, China, Russian Federation and Saudi Arabia).
The CPI all-items for the G20 aggregate is calculated by aggregating the national CPI/HICP in each period using Household Final Consumption Expenditure (HFCE) and Final Consumption Expenditure of Non-Profits Institutions serving Households (NPISHs) as weights. The relevant country data are converted into a common currency (US dollars) using purchasing power parities (PPP) which are estimates of the rates of currency conversion that eliminate the differences in price levels between countries. The PPPs used in the zone estimates relate to HFCE specifically and are not the same as the PPPs for GDP, which are more commonly available.
The weights used to compile the G20 estimate are updated each year, usually in November or December. As a result of changes in the underlying national accounts data, the G20 estimate is subject to revision even though national CPIs/HICPs are rarely revised themselves. The main reason for updating the weights annually as opposed to aggregating with weights from a particular fixed base year is to allow for the changing importance of a country over time. Table 2 shows the weights of each G20 economy for the years 1995, 2000, 2005 and 2011 (the last year available at the time of writing for all G20 economies).
The G20 CPI aggregate is compiled from G-20 national CPIs and HICPs as a chain-linked Laspeyres-type index. Each monthly index during one particular calendar year of the index, say 2012, is re-referenced so that December 2011 = 100 and then aggregated using the weights of the previous year, i.e. 2011 in this example, price-updated to December 2011. This process is repeated each year to form links for each calendar year referenced to December of the previous year. Next, these links are chained together by successive multiplication to create a time series which is then referenced to a common year, currently 2005=100. Since weights are usually available with some time lag (of around one year), G20 estimates for the latest period continue to use the most recently available weights until new data become available.
Preliminary results and data presentation
As of the data of August 2013, in the news release and in the relevant websites, CPI data for the G20 aggregate and the G20 economies will be made available in the three following ways:
- Index with 2005=100 as index reference year;
- Annual change: percentage change on the same period of the previous year (%)
- Previous period change: percentage change on the previous period (%)
Comparability issues of CPI across G20 economies
The scope of this section is on issues affecting the international comparability of the CPI headline indicators across G20 economies. Subsequently, the following topics will be discussed:
All G20 economies compile a national CPI/HICP that gives a measure of inflation by measuring price changes of a representative basket of goods and services typically purchased by households.
National CPI versus HICP
As mentioned above, data used for the compilation of CPI all-items for the G20 aggregate refer to HICPs for the European Union (including France, Germany, Italy, and the United Kingdom) as well as for Turkey, and to the national CPI for the other 14 countries.
The preference for the use of the HICP, when available, rather than the national CPI is founded on the fact that Eurostat has developed a harmonised approach and a single set of definitions in order to arrive at comparable measures of inflation in the EU. Given the higher comparability of the HICPs, the use of HICPs when available is considered appropriate for the compilation of the CPI G20 aggregate. The availability among G20 economies is as follows:
- HICPs for the European Union covering all current 28 Member States: data from January 1996 (including HICPs for France, Germany, Italy, and the United Kingdom);
- HICP for Turkey: data from January 1996;
- Experimental HICP for the United States that differs from the US National CPI in two major aspects. The US HICP (i) includes the rural population, and (ii) excludes owner occupied housing. However, for reasons of prominence, main uses and timeliness (the US HICP is published with a delay of one or two months with historical data starting in January 1998 only) the use of the US national CPI is preferred in the calculation of the G20 aggregate.
France, Germany, Italy, Turkey and the United Kingdom publish both HICPs and national CPIs. The methodology used for the HICPs differs from the methodology used for the compilation of the national CPIs. A comparison of percentage changes in national CPIs and HICPs is shown in Table 3.
In general terms, the main methodological differences between the HICP and the national CPIs are as follows:
- Coverage: Excluded from HICPs are: Narcotics (COICOP/HICP 02.3); Imputed rentals for housing, including rentals of owner-occupiers (04.2); Games of chance (gambling) (09.4.3); FISIM (financial intermediation services indirectly measured) (12.5.2); and part of personal care services n.e.c. (12.1.3 ) (example: prostitution);
- Treatment of subsidized healthcare and education: the HICP includes the net price paid by consumers (after reimbursements);
- Formula used for elementary aggregates: the HICP use is ratios of arithmetic mean prices and/or ratios of geometric means;
- Geographical and population coverage: the HICP covers all expenditures on the territory of the country concerned, whether made by residents or non-residents (e.g. tourists).
Table 4 highlights the main methodological differences between the HICP and the national CPI for individual countries.
More generally, a number of dimensions relating to the reference population, and the types of expenditure and transactions covered need to be considered when comparing headline CPIs and HICPs.
As shown in table 5, the CPI covers only urban areas in two third of the G20 economies. Restricting collection of prices to urban areas may not be an issue if their movements can be considered representative of price movements in rural areas. In these cases national weights can be applied and the resulting index can be considered as a national CPI covering the whole country. If, however, price movements in urban and rural areas are sufficiently different and price collection is restricted to urban areas owing to resource constraints, then urban weights should be applied and the resulting index must be considered as merely an urban CPI, not a national one.
For CPI purposes, the definition of a household is essentially the same as in the System of National Accounts (SNA) 2008, which includes both private households, including those consisting of only one person, and institutional households such as religious orders, residential hospitals, prisons, etc. For many G20 economies, however, the institutional population and those living in non-private households are excluded from the scope of the CPI, often due to problems related to obtaining data on their consumption patterns. In these economies the indices are restricted to the private household population or a subset of the private household population. On the other hand, the HICP coverage of households includes institutional households.
Ideally, the CPI should be designed to be representative for households and not to a specific socio-economic group. However in some G20 economies such as Brazil, India and Indonesia, households with a low income level are excluded. Japan and Korea exclude households mainly engaged in agriculture, forestry and fishing, and all one-person households. Table 6 provides an overview of the socio-economic coverage.
The COICOP classification of consumer expenditures and prices on goods and services by purpose is only strictly applied by Korea and South Africa, while the COICOP/HICP classification (which is the COICOP classification adapted to the needs of the HICP) is used for the European Union and by Turkey. The other G20 economies use national classifications which in some countries, such as Australia and Japan are consistent with COICOP at the highest level. Table 7 gives an overview of the systems of classification used and highlights the specific exclusions or inclusions by country.
Treatment of housing
As shown in table 8, the cost of housing is a major component of household expenditure. However, the measurement of housing costs is far from straightforward and not consistent. Different approaches to the treatment of owner-occupied housing (OOH), e.g. the (net) acquisition approach, the imputed rents approach or the user cost approach, make the comparability across countries difficult.
Table 9 presents how housing is treated in the G20 economies. All G20 economies include a measure of rentals for housing. However, decisions on the coverage, the adjustment for quality, the treatment of regulated rents and the design of price surveys in cases where the rental market is small or unregulated, may affect comparability across countries.
Eleven countries do not include a measure of OOH. All HICPs currently exclude OOH. Among the 6 countries that include OOH in the CPI, practices on how to deal with OOH vary between countries:
- Australia uses an acquisition approach;
- Canada uses a user cost approach; and
- Japan, Mexico, South Africa, and United States use a rental equivalent approach.
The compilation of a CPI consists of two main stages: 1) the calculation of unweighted price indices (mostly price ratios) for elementary aggregates, and 2) the weighted aggregation of these elementary price indices for higher level aggregates.
Three main aggregation methods are used to calculate elementary aggregate indices:
- The ratio of arithmetic mean of prices or Dutot index. This is often referred to as the ratio of averages (RA);
- The arithmetic mean of price relatives or Carli index. This is often referred to as the average of relatives (AR);
- The geometric mean of the price relative, which is identical to the ratio of the geometric means of prices (GM) or Jevons index.
The second stage in the compilation process is the aggregation of elementary price indices to higher level indices. This aggregation is typically done using a Laspeyres-type approach. Laspeyres-type indexes include Lowe, Young and (retrospective) Laspeyres indexes .
Typical frequencies of updating the weights are either every five years or once a year. In most G20 economies weights are updated every five years. Either of the above formulas can be chained annually of less frequently reweighted.
Table 10 shows information on the methods used for index calculation both for elementary aggregates and higher level aggregation. Methods used for combining prices to give elementary aggregates vary widely among the individual countries of the G20, the geometric mean (Jevons index) is the method most commonly used (10 countries plus France, Italy and the United Kingdom). All G20 economies use a Laspeyres-type formula for their published CPIs in one of the aforementioned forms.
Further Eurostat information
- First release of G20 consumer price index G20 - annual inflation slows to 3.0% in August 2013 (Eurostat Press release, 14 October 2013]
- OECD.StatExtracts - G20 – CPI All items (OECD online database)
- Principal Global Indicators - The Inter-Agency Group on Economic and Financial Statistics