1. Regular annual PPP revisions
PPPs are only established on an annual basis, therefore only annual revisions apply. According to the Eurostat regular publication calendar PPPs are released as follows: an earlier estimate of PPP a so-called "now-cast" at a high aggregation levell, published 5 months after the end of the reference year. The now-cast is replaced by preliminary PPPs 12 months after the end of the reference year and revised to final data 24 months after the end of the reference year. This regular PPP revision / release calendar is in line with the data delivery timetable for National Accounts data according to the ESA95 regulation1.
In addition, regular National Accounts revisions of the level of GDP and other aggregated expenditure data are taken into account with regard to volume indicators, as the PPPs are always applied to the latest available National Accounts data.
But the weights, which underlie the PPP calculations, are not always fully consistent with the latest aggregated GDP data to which the PPPs are applied. This is a built-in problem of the National Accounts compilation and revision process, as more aggregated data may be revised more often than the detailed expenditure weights, which usually are only revised once a year.
2. Major PPP revisions
The revision of PPPs from 1995 to 2000 falls in the category of the so-called major revisions. In fact, such a major revision has been attempted for the first time. This revision of PPPs incorporates changes in methods, concepts and classifications consistently into the underlying basic data and computation methods for all years in question. The occasion has also been used to correct real errors in reported data sets and data processing. Such major revisions of PPP will remain rather infrequent events and will be repeated only when new major changes in the underlying PPP and/or National Accounts methodology will occur.
2.1. Reasons for revision
The overall PPP calculations depend on three forms of input: the price survey results, GDP weights and other input data, as e.g. rents and salaries in General Government. Of these, only price survey results have snapshot character: once collected, prices cannot be re-collected or revised.
GDP in national currency, GDP weights and other National Accounts related input data are, however, subject to regular revisions in line with the national revision timetables for the National Accounts. These national revision timetables are not (yet) harmonised across the EU. In addition there was a major National Accounts revision concerning the switch from ESA79 to ESA95. This major change in National Accounts methodology in fact was one of the main reasons for the first major revision of PPPs.
The introduction of ESA95 did not occur at once in all its parts and not in all countries at the same point in time. This step by step introduction of ESA95 in the Member States led to a multitude of breaks in the time series and affected the comparability of the PPP based indicators over time or even between countries within one given year. The initially established PPPs were not anymore consistent with the data to which they were applied to and from which they were partially derived.
The situation was even more critical for the EU-Candidate Countries, which had to manage the introduction of ESA95 partly coming from older National Accounts systems such as MPS (Material Product System), SNA68 (System of National Accounts) or national systems. In addition the Candidate Countries did not all fully participate in all price surveys during the period 1995 to 1998. Respective extrapolation, which had to be undertaken, had to be reviewed as well.
The year 2000 was taken as a starting point for the revision because for the majority of the participating countries, the reference year 2000 was the first year where the National Accounts data was fully ESA95 based. Moreover, since the reference year 1999, all Candidate Countries have fully participated in the price surveys according to the survey cycle. Eurostat, therefore, believed that it was the right time to revise the years 1995 to 2000 in light of ESA95 and bring them in line with the data for 2000. This would help to facilitate the comparison over space and time of volume indicators, such as GDP per head in PPS, which plays an important role in economic analysis.
2.2. The scope of the revision
For the EU Member states and the three EFTA countries the revision was made at the same level of detail as the regular PPP calculation for the years 1995 to 2000.
In the case of the EU Candidate Countries, a detailed revision of PPP was limited to the years 1999 and 2000 because of the absence of detailed input data; for the moment their ESA95 base NA series are relatively short and, even more importantly, they have fully participated in the PPP price surveys of Eurostat only from the reference year 1999. For the EU Candidate Countries PPPs from 1995 to 1998 were retrapolated using 1999 PPPs and GDP implicit price deflator. The aggregation was based on the new consistent GDP expenditure weights, which were provided for most Candidate Countries for the period 1995 to 2000.
2.2.1. GDP and expenditure weights
One of the main reasons for undertaking the revision was to produce PPP data, which have underlying NA weights that are for all years in line with ESA95. In addition, there are two more requirements: a) the weights have to be for each country consistent within each year; b) the weights also have to be complete, no gaps should exist, except for those categories were in reality no expenditure exist in a given country, e.g. on railway services in Iceland in the absence of a railway system.
All 31 countries have provided revised GDP expenditure weights for all years except Poland (only 1999, 2000) and Malta (only 2000). In cases where countries could not supply complete structures, extensive work was undertaken by Eurostat to estimate missing values. Eurostat also put emphasis on checking if the development over time of the revised expenditure weights showed a reasonable pattern.
As important as the structure of GDP is the total level of GDP. Often the PPP deflated level of GDP is the focus of international price and volume comparisons. In the EU, also the structural funds allocations are based on level criteria, the GDP per head in PPP terms.
The data provided by the countries revealed for some countries also considerable changes to the level of GDP.
For the EU Accession and Candidate countries, however, the revised GDP volume shown for the years 1999 and 2000 may undergo further changes the extend of which currently is not foreseeable. The main reason is that all Eurostat National accounts projects for the full implementation of ESA95 in these countries, scheduled with a final deadline at mid-2004, will have an impact on the level of GDP.
2.2.2. Consumer price survey data and consumer price indices (CPI)
Collected consumer prices cannot be revised and they are assumed to be the most reliable part of the PPP exercise. However, different classifications were used during the period 1995 to 2000. During 1995 to 1998 the 217 basic headings for consumer goods and services were classified according to the Classification of households goods and services (CHGS) in line with ESA79, and from 1999 for the 199 basic headings the Classification of individual consumption by purpose (COICOP) in line with ESA95 was applied. So all respective basic headings have been re-classified according to the latest COICOP classification.
CPI plays an important role in the PPP calculations as they are used for extrapolating basic heading PPP between survey years. Normally, the CPI data itself is very seldom revised. However, as different basic heading classifications were used in different years, Eurostat asked the participating countries to provide a consistent set of CPI for all years according to the COICOP classification.
2.2.3. Other input data subject to revision
Housing services
During the revision various aspects of rents' data have been analysed: rents' data (prices and quantity indicators); conformity of the rents reported in PPP with the rents underlying NA; coverage of rents and the change in the basic heading classification. The countries undertook enormous efforts to provide improved rents data and weights, which show a much smoother development over time.
Equipment goods and construction prices
Over time analyses of Price Level Indices (PLIs) at basic heading level have shown remarkable volatility. Therefore, retrapolation of PPPs from 2000 or 2001 backward has been considered as a possible alternative solution. The implicit deflator from the National Accounts for Gross Fixed Capital Formation or for its sub-aggregates was used for all countries as suitable retrapolation factor.
Salaries in General Government
When comparing salaries, the working time they relate to should be an equal standard working time in all countries. During the revision adjustments for equal standard working time were made for years 1995 and 1996.
Weights in the form of the share of number of employees for the selected occupations in the total number of employees were revised as well. Countries were asked to revise the weights of at least 1-year, which were also applied to the other years. This was done under the assumption that structures of employment usually do not change very much from year to year.
Special attention was also paid to the consistent interpretation of the Eurostat survey guidelines.
Population
Population figures in most countries are based on rather infrequent population censuses and extrapolations, e.g. based on administrative information. During the period of the revision, in several countries censuses have been carried out. For the per capita calculations in the framework of the revision, the latest population figures have been used.
Remarkable revisions to the population figures have been reported for Greece, Portugal, UK, Cyprus, and Latvia, and too lesser extend for Luxembourg and Lithuania.
Eurostat also ensured that the population figures were those according to the residency concept used in the National Accounts.
3. Main results of the revision
In summary, two main observations can be made concerning the revision and resulting series. Firstly, the level of per capita GDP in PPS has been revised consistently for the years under revision, reflecting better now, in individual years, countries' relative position to the EU average and to other countries than in the results before (e.g. Denmark, Germany, France, Sweden, Iceland and Switzerland). Secondly, revised data show much more logical development over time, even if over time comparison is not the main purpose for which PPPs have been developed. Formerly observed "jumps", have been removed successfully (e.g. Belgium, Denmark, France and United Kingdom).
It can be concluded that the revision has well addressed the criticism raised with regard to the reliability of PPPs and related economic indicators, particularly when it comes to the development over time in comparison with the National Accounts.
1. ESA95; European System of Accounts 1995, Commission Regulation (EC) 2223/1996 of 25 June 1996
Published on 23.02.2004
