90259 International Comparison Program Proposed ICP Revision Policy Contents 1. BACKGROUND ............................................................................................................................................... 3 2. TRIGGERS FOR REVISING ICP INDICATORS ..................................................................................................... 3 2.1 REVISIONS IN INPUT DATA ...................................................................................................................................3 2.2 NEW METHODOLOGY .........................................................................................................................................4 3. GUIDELINES FOR REVISING ICP INDICATORS .................................................................................................. 4 3.1 HISTORICAL REVISIONS .......................................................................................................................................4 3.2 GEOGRAPHICAL SCOPE .......................................................................................................................................4 3.3 FIXITY .............................................................................................................................................................5 3.4 CLASSIFICATION LEVEL ........................................................................................................................................5 3.5 CATEGORIES OF INDICATORS ................................................................................................................................5 3.6 QUALITY REVIEW ...............................................................................................................................................5 3.7 PUBLICATION OF REVISED RESULTS ........................................................................................................................5 3.8 CONSISTENCY BETWEEN PUBLISHED AN UNPUBLISHED DATASETS.................................................................................5 4. TIMING AND COMMUNICATION OF REVISIONS ............................................................................................. 6 2 Proposed ICP Revision Policy 1. Background The motivation for proposing an ICP revision policy stems from user interest to compare ICP results from subsequent benchmark rounds. ICP benchmark rounds publish indicators that include: (i) real and per-capita real expenditures, (ii) Purchasing Power Parities (PPPs) and Price Level Indices (PLIs), and (iii) nominal and per-capita nominal expenditures for all participating countries. Indicators are published at the analytical level. The analytical level is the level of detail (i.e. GDP, major components and categories) for which the Global Office, OECD, Eurostat and Regional Coordinating Agencies agree to publish data in the ICP benchmark rounds. Results at various levels of detail below the published level - down to the basic heading level - for all countries are available to researchers through a data access process, as stipulated in the ICP Data Access and Archive Policy. Comparing results from ICP benchmarks is complicated by three significant factors. First, ICP benchmark rounds are designed to provide a one-time “snapshot” of ICP indicators. Second, the collection and estimation methodologies for some components are improved between rounds. Third, additional countries are included plus others may drop out or shift from one region to another. This affects comparisons between countries because the multilateral results will differ depending on the countries included and their ICP grouping. In addition, National Accounts in virtually all countries are revised over time as additional data become available. For instance, the 2005 estimates of expenditure on GDP, and its major components, have been revised in most countries since the 2005 ICP results were released in December 2007; in the case of 22 countries, these revisions have been in excess of 10%. Additionally, analogous problems arise with revisions to population estimates. While ICP benchmarks produce a one-time “snapshot” of ICP indicators, the World Bank’s World Development Indicators produce a time series of ICP indicators on an annual basis for non-benchmark years starting from 1980. These interim estimates are computed based on an extrapolation method. This paper proposes a revision policy that defines how ICP indicators will be revised. It describes the triggers and guidelines for revising ICP indicators, as well as the timing of revisions and the steps to be taken to communicate these revisions to users. 2. Triggers for Revising ICP Indicators The proposed triggers for revisions are as follows: 2.1 Revisions in input data - Revisions in aggregate GDP estimates: trigger a revision of real and per-capita real expenditures and nominal and per-capita nominal expenditures. - Revisions in major components of GDP: trigger a revision of real and per-capita real expenditures and nominal and per-capita nominal expenditures. It may also trigger a revision of PPPs and PLIs. - Revisions in population figures: trigger a revision of per-capita real expenditures and per-capita nominal expenditures. 3 - Changes in countries’ currency units: trigger a revision of real and per-capita real expenditures, PPPs, PLIs, and nominal and per-capita nominal expenditures. - Release of new results from regional non-benchmark exercises: may trigger a revision of global level results related to real and per-capita real expenditures, PPPs and PLIs for non-benchmark years. - Correction of errors in source data or results: may trigger a revision of real and per-capita real expenditures, PPPs and PLIs, and nominal and per-capita nominal expenditures. 2.2 New methodology - Materially improved PPP computation and aggregation methods: trigger a revision of real and per-capita real expenditures, PPPs and PLIs. - Materially improved global linking approach: triggers a revision of real and per-capita real expenditures, PPPs and PLIs. - Materially improved extrapolation method: triggers a revision of real and per-capita real expenditures, PPPs and PLIs for non-benchmark years. 3. Guidelines for Revising ICP Indicators The following guidelines are proposed for revising ICP indicators: 3.1 Historical revisions - The World Bank assumes that historical estimates of country prices and exchange rates used in the ICP benchmarks will remain unchanged. - Revisions triggered by changes in a country’s National Accounts estimates may require a revision to the time series of ICP indicators -including benchmark data- going back historically as far as necessary to incorporate the changes in the country’s National Accounts time series. For example, revisions in a country’s GDP from 2004 to 2011 would trigger a revision in the real and per-capita real expenditures of the 2005 ICP benchmark. However, if the revisions spanned 2006 through 2011, the 2005 ICP benchmark data would not be revised. - Revisions triggered by a new benchmark ICP methodology should not go beyond the last benchmark. For example, ICP 2011 uses an improved global linking method so revisions could be done for the last benchmark indicators (2005) but not for the previous benchmark indicators (1993-1996). 3.2 Geographical scope - It is desirable to introduce revisions first to regional benchmark indicators then to global benchmark indicators, in order to preserve the consistency between regional and global datasets. Revised within-region PPPs will be the input to the estimation of between-region linking factors needed to determine global level PPPs. - The World Bank is responsible for revising benchmark indicators at the global level (i.e. denominated in the global numeraire). - If a particular region does not revise its regional benchmark indicators, the World Bank will revise and publish global benchmark indicators to take into account any revision in other regions. - Results of regional interim exercises may be incorporated in the time series of ICP indicators if the results are deemed of sufficient quality by the World Development Indicators team. 4 3.3 Fixity - Revised global benchmark results should respect, to the extent possible, regional price fixity - the convention whereby the price relativities established between countries in a regional comparison are retained when the countries are included in the global comparison. 3.4 Classification level - PPPs could be revised at various classification levels (GDP, major components, categories, groups, or classes) depending on the level of detail of the National Accounts expenditure revisions. For example, revised expenditures at the class level would result in revisions to higher-level PPP aggregates (GDP, major components, categories, and groups) through the aggregation process. 3.5 Categories of indicators - Real and per-capita real expenditures will be revised when National Accounts expenditure data and/or population data are revised. - PPPs and PLIs may be revised depending on the level of detail of the National Accounts expenditure revisions. When countries revise their expenditure data for major components, categories, groups, classes, or basic headings, then PPPs (and resulting PLIs) may be revised at levels above the lowest level for which expenditures were revised. - Nominal and per-capita nominal expenditures will be revised when National Accounts expenditure data and/or population data are revised. 3.6 Quality review - Revised ICP indicators should go through an expert review before they are published, to ensure data quality. - The World Bank will maintain a database containing the various vintages of the data, for quality purposes. 3.7 Publication of revised results - Revised ICP indicators should be published once they have been compiled and undergone the quality review process. - Revised ICP indicators should be released at the analytical level of the benchmark results, as established in the ICP Data Access and Archive Policy. 3.8 Consistency between published an unpublished datasets - The World Bank provides researchers with a detailed dataset from benchmark ICP exercises, as stipulated in the ICP Data Access and Archive Policy. This dataset includes real and per-capita real expenditures, PPPs and PLIs, and nominal and per-capita nominal expenditures at all levels - down to the basic heading level. When revising ICP indicators, this unpublished detailed dataset may need to be revised to maintain the consistency between the published and the unpublished datasets. For example, if a revision of nominal and per-capita nominal expenditures is triggered, it may be desirable to reflect it at all levels - down to the basic heading level - depending on the World Bank’s assessment of the need to do so. 5 4. Timing and Communication of Revisions - The time series of ICP indicators will be revised by the World Bank on an annual basis, as part of the World Development Indicators April update. - The schedule of revisions should be announced to stakeholders and users well in advance. - When methodology is improved, the new methods need to be communicated to users well in advance. - Results should be made publicly available on the Internet, ICP portal, and all other relevant sites. 6