Learning Center
Plans & pricing Sign in
Sign Out

the oecd eurostat programme on purchasing power parities


									                                       For Official Use                                                           STD/HLG(2002)7
                                       Organisation de Coopération et de Développement Economiques
                                       Organisation for Economic Co-operation and Development                          27-May-2002
                                       _____________                                                            English - Or. English
                                       STATISTICS DIRECTORATE
For Official Use

                                       Meeting of the High Level Group on Statistics


                                       Agenda Item 3

                                       Château de la Muette
                                       13 June 2002
                                       Beginning at 9.30
               English - Or. English


                                       Document complet disponible sur OLIS dans son format d'origine
                                       Complete document available on OLIS in its original format



1.       The PPP programme was established in the early 1980s to provide internationally comparable
price and volume measures of GDP and its component expenditures for the Member countries of the
European Union and the OECD, as well as several other countries. The 1999 Round is the sixth round of
the Programme and the number of countries has grown from 18 in 1980 to 43 in 1999.

2.        The PPP programme underwent significant scrutiny a few years ago, with an outside assessment
by Ian Castles1 (then ABS), and another one on the broader International Comparison Programme (ICP) by
Jacob Ryten (then Statistics Canada). These reports led to several changes in the PPP programme, some of
which will be mentioned below. In 2002, and with the completion of the ‘1999 round’, ABS suggested that
the newly-founded high-level meeting of OECD Statisticians discuss some of the lessons learned from this
latest round and advise on possible adjustments in the methodology, institutional set-up or data gathering
process of the PPP programme. The present paper provides the background for this discussion. The outline
of the paper was presented to the OECD Statistical Advisory Group in March 2002, and met its approval.

Key features of the 1999 Round

3.        Under the Joint OECD-Eurostat PPP Programme, the OECD and Eurostat share the responsibility
for calculating PPPs. For the 1999 Round, Eurostat (the Statistical Office of the European Commission)
provided data for the 15 EU Member states, the 13 Candidate Countries to the European Union and
Norway, Iceland and Switzerland. The OECD was responsible for the programme with non-European
Member countries: Australia, Canada, Korea, Japan, Mexico, New Zealand, the United States as well as for
the Russian Federation, Ukraine, Israel, Croatia and Macedonia. The 1999 Benchmark comparison reflects
series of price quotations for a basket of about 3000 comparable and representative goods and services.
Price quotations are obtained from statistical surveys that follow a common framework elaborated by the
OECD and Eurostat. For the first time, data for Korea came from a benchmark survey (earlier results were
based on OECD estimates).

4.        In many respects, the 1999 Round can be considered as a transition Round with three key

           Significant organisational changes in the collection of price data, largely due to the reform
            of the European Comparison Programme (ECP);

1.   Ian Castles (1997); ‘Review of the OECD-Eurostat PPP Programme’ [OECD document STD/PPP(97)5].


           Extension of the number of participating countries, in particular Candidate countries to the
            European Union that participated in the Programme for the first time;

           Methodological changes, in particular introduction of SNA 93.

           Increased commitment of non-European countries;

5.        The first three features need to be kept in mind in any assessment of the 1999 Round: they were
sizeable, one-off changes that, in all likeliness are beneficial for the working of the programme in the
coming Rounds, but they have also meant certain costs and investments during the 1999 Round that did not
help, for example, the timeliness of publication of the 1999 results. The following paragraphs spell out
what the OECD Secretariat considers the most important institutional and substantive issues in the 1999

Institutional aspects

6.         Country groupings and ECP reform. The 1999 Round brought about an important
organisational change in data collection procedures among the countries for which Eurostat bears
responsibility. In the course of the reform of the European Comparison Programme (the ECP Reform was
announced in November 1997 and came into effect in January 1999), three groups of countries were
formed according to geographical proximity (Southern, Northern and Central Europe). Each group has a
leader country (Finland, Italy, Austria) that is responsible for the compilation of the survey product lists for
its group, the co-ordination of the price surveys within the countries in its group, the editing of the price
data for its group and the submission of the national average prices to Eurostat. So far, the ECP reform has
only concerned consumer goods and services.

7.       From an OECD perspective (who participates regularly in these meetings), this organisational
change has been a useful one, and will be conducive to improving international comparability. However,
during the 1999 Round, the ECP reform turned out to be time-consuming and required a fair amount of
meetings. Another element of transaction costs was the changeover of responsibilities for EU Candidate
countries (Czech Republic, Hungary, Slovak Republic, Slovenia, Turkey) from OECD to Eurostat. In
conclusion, the organisational reform has been the right move but is likely to produce net benefits mainly
from the present 2002 Round onwards.

8.         Creation of the OECD group of ‘Pacific Rim’ countries. One of the direct consequences of the
call in the Castles Report for greater transparency and visibility of the PPP programme and for countries to
be involved at all stages of the process, was the organisation of several meetings of ‘Pacific Rim’ countries
(United States, Canada, Mexico, Japan, Korea, Australia, New Zealand) plus Israel. The meetings of this
group (latest one in Paris, November 2001) turned out to be useful occasions for countries to discuss
common problems and to articulate a non-European perspective in the OECD/Eurostat programme.
Discussion of results by the group have improved data editing procedures and will continue to do so in the
2002 Round.

9.         Bilateral working relations with Pacific Rim countries. OECD has also started to intensify
direct relations with its counterparts in statistical offices of the Pacific Rim countries. This has led to visits
to Australia, New Zealand and Korea in 2002 with a view to tackling data collection and comparability
issues from a country-specific perspective.

10.       Extension of number of participating countries. On the Eurostat side, the inclusion of candidate
countries into the regular PPP work marked an important event. Concerning OECD, Korea participated for


the first time in the 1999 round of the PPP programme. With Korea, all 30 OECD member countries are
now covered. Until then, Korean PPPs had been estimated by the Secretariat using regression techniques.
The integration of a new country into the programme and the associated multiple contacts between the
Secretariat and the Korean authorities showed again how little of a purely mechanical exercise the PPP
programme is. Korea has been a particularly interesting case in point, given its consumption patterns,
brands and product characteristics that are often very different from those in other OECD countries. This
presents a demanding situation for national statisticians who constantly have to arbitrate between
comparability and representativeness: comparable products are often highly priced luxury goods or
services that tend to be unrepresentative and whose inclusion into the price comparison may generate an
upward bias. On the other hand, there is often no comparable counterpart for those products that are
representative of the Korean market. Considerations of this trade-off require case-by-case decisions that
are sometimes difficult when there is little experience with international price comparisons. Regular
bilateral contacts are therefore important and the above-mentioned visit to Korea in 2002 would have been
even more useful had it take place during the early phase of the 1999 Round.

11.       Timeliness of results. Timeliness is a major concern and every delay even by a single country
can have consequences on the quality of overall results. The reason is that editing procedures have to be
run on the prices of all countries at the same time if they are to be fully effective. Regular meetings such as
the one for the Pacific Rim countries help to synchronise provision of timely data.

12.        A negative short-term impact of the efforts to re-organise the PPP programme was that the 1999
benchmark results for all 43 countries were only finalised in December 2001 - significantly later than
initially planned. OECD basic data were final beginning of the year 2001 and preliminary results were
already shown in Ottawa in October 2001 but Eurostat data became only available in autumn 2002.
Eurostat has since put in place a new schedule and planning procedure to obtain data from countries and to
treat them so that in all probability deadlines will be met in the future.

13.       Overall, data supply by non-European countries was timely and timeliness tended to improve
during the Round. For example, Australia only provided results from the first surveys only during in early
2000 but caught up with the schedule by the end of the Round. Calculation procedures at the OECD
Secretariat are well in place and require comparatively little time. However, verification processes are time
consuming, in particular when the Secretariat goes back to one or several countries because comparative
results for a product (group) turn out to be counter-intuitive.

14.       Eurostat draft regulation on PPPs. Another important institutional/legal development among is
the forthcoming European regulation on PPPs. Eurostat has prepared a draft of this regulation which is
presently under review in the relevant bodies of the European Union. The draft regulation will make
delivery of price data for PPP comparisons by EU countries compulsory, and the regulation defines
standards for quality and timeliness of the data. A methodological manual to be developed alongside with
the regulation should also further the comparability and representativeness of the compiled data.

15.        OECD welcomes the draft regulation because it puts the PPP programme in the EU countries on
a strong institutional and legal footing. OECD has been consulted in the course of the elaboration of the
draft regulation and Eurostat took up a number of these suggestions. As a general point, OECD hopes that
the specifications in the Regulation remain general enough so as not to inhibit adjustments in methodology
if they are needed in the future.


Substantive issues

16.       In addition to the institutional settings, there are a number of substantive issues – some of them
unique to the 1999 Round (e.g. transition to the SNA 93), some of them of a recurring and long-standing
nature (e.g., difficulties in producing product lists that are as representative for non-European countries as
they are for European ones).

17.       Switch to SNA 1993. During the 1999 Round, most of the OECD countries switched from the
1968 System of National Accounts (SNA) to the SNA 93. There are two types of implications. First, some
changes in national accounts definitions affect the data used in PPP calculations. For example, the SNA
1993 identifies software purchases as part of investment. Second, the SNA 1993 brought along a change in
the classification of final consumption expenditure. More specifically, this change concerns government
consumption expenditure on goods and services that are provided free-of-charge or at nominal fees or with
reimbursement to households. The SNA 1993 adopted the “who pays” principle in their recording and this
change is also reflected in the PPP expenditure classification in the 1999 Round. While the change in
classification did not raise any controversial issues, it implies that the backward comparability of the 1999
PPPs at the level of broad expenditure components has been reduced.

18.        Number of products priced and number of basic headings. In the course of the 1999 Round, a
good deal of discussion took place with Eurostat about the possibility to improve cost-effectiveness of data
collection by way of reducing the number of products collected and by way of reducing the number of
basic headings, i.e., the lowest level of aggregation in the product classification used for price collection.
The OECD maintained that there is room for reduction of products priced and reduction of number of basic
headings but these should be considered as different aims (they are not complementary). Reductions of the
number of basic headings should be considered on a case-by-case basis, and survey by survey because an
‘automatic’ reduction can lead to biased results. When the number of products priced is reduced too
strongly, an implicit and strong emphasis is put on comparability of results over their representativeness.
Also, given that the number of prices per product supplied by non-European countries is much smaller than
the number of observations coming from European countries, any further reduction would be detrimental
to the reliability of the data for non-European countries.

19.       Representativeness of European product list for other countries. This is a specific and recurring
OECD concern: are the product lists defined by European countries representative of non-European
markets? An OECD-wide comparison based on products that are not equally representative of all countries
will result in biased price relatives. Relative price levels for countries having a smaller share of
representative products in all products that have been priced will be overestimated, while relative price
levels for countries with a larger number of representative products will be underestimated.

20.       One way to tackle this problem is the use of very detailed product descriptions and specifications
(‘generic’ specifications) rather than brands and models to identify comparable products. Between different
countries, there is a greater chance of finding a local product with similar specifications than a specific
brand model. Note, however, the loss of cost-effectiveness when moving from an identification via
brands/models towards an identification via detailed specifications: the latter requires much longer and
more elaborate descriptions which may be costly and imply more work for the price surveyors

21.       Another possibility is to add some products that are typical for non-European markets to the
European list. Of course, such an initiative runs counter to the above-mentioned move towards minimising
the number of product prices collected and is only possible if the non-European products actually exist on
European markets. When OECD participates in meetings of European countries, it asks for the inclusion of
certain products into the European lists whenever this is reasonable. For example, the European list now
contains prices of cars with engines of more than 2000 cm3.


22.       A third possible route for certain products is to make use of hedonic price estimators that exist in
some countries, in particular the United States and Canada. Given detailed specifications of certain
products – for example personal computers – a country may use its hedonic function for this type of
product and introduce the model characteristics into the hedonic function. This generates a price that
indicates what the model at hand would cost if it were delivered at a competitive price in the domestic
market. Where the possibility to use hedonic functions exists, it constitutes a cost-effective and
theoretically satisfactory way to obtain price estimates. However, the method is applicable only in a limited
number of cases, as hedonic techniques are only used for a minority of products and/or because
characteristics themselves (and not only their quantity) may be entirely different across countries.

23.       ‘Comparison-resistant’ goods and services. Some parts of the price comparison are specifically
difficult and had already been identified by the Castles and the Ryten report. They include housing
services, construction, capital goods, communication services and items of collective consumption. During
the 1999 Round, progress has been modest in these areas of PPP programme. Meanwhile, however,
Eurostat has started to review its collection of construction prices and OECD will benefit from any
progress made in this field. Generally, these difficult areas need tackling on a case-by-case basis and often
require additional resources to undertake research in conjunction with Member countries.

24.       Use of PPPs over time and consistency with national accounts. An important issue that arises
with the publication of every new set of benchmark results is the consistency between PPP and national
accounts time series. More precisely, it is asked why GDP volume results in consecutive benchmark
comparisons are not in line with GDP growth rates in that period. Although it is not too difficult to explain
why, from a theoretical viewpoint, the two series can be different (changes in the relative price structure
across countries, treatment of foreign trade etc.), it is much more difficult to assert users that all of the
differences are due to changes in the relative price structure. OECD undertook some analysis of the issue
to separate explainable differences from unexplainable ones. Though difficult to quantify exactly, the
conclusion from this work is that unexplainable differences do play a sizeable part, and this has
implications for the use and interpretation of PPPs. In early 2002, Eurostat initiated a project to review
historical PPPs. One of the fall-outs of this work will be more light shed on the above-mentioned
unexplained differences.

25.        Communication. When the results for 1999 were released at the beginning of this year, a
particular effort was undertaken to communicate them to the public from a user’s perspective. Two
initiatives were taken. First, a note was drafted to accompany the press release and to provide indications
for interpretation and use of PPPs. This document2 was well received by the press, by users and by member
countries. Along similar lines, but in a more concise form and with improved presentation, an issue of the
OECD Statistics Brief was devoted to PPPs. Again, this document found a favourable reception. Overall,
the experience showed that there is need to communicate results with a user’s perspective. This raises the
visibility of the programme and helps avoiding mis-interpretation of some results.

Lessons for the future

26.        The PPP programme generates a public good: quality will only be high if every country is
committed to the programme. The nature of the PPP programme is such that responsibilities have to be
shared between countries and international organisations. There is a public goods aspect to the production
of PPPs: the quality and intensity of each country’s effort by itself cannot guarantee high-quality results for
this particular country, but a collective effort will generate reliable results for everyone. Commitment to



the PPP programme in statistical offices is an issue of internal priority setting and resource commitment 3.
The forthcoming Regulation as well as the organisational changes put in place during the 1999 Round will
certainly strengthen the involvement of many European agencies.

27.        Involvement of national accountants is important. Too often, the PPP exercise is regarded as
work of price statisticians only. However, the aim is to make volume comparisons of GDP and its main
aggregates across countries. The quality of PPP-based volume comparisons is just as dependent on the
quality and comparability of detailed national accounts expenditure data as it is on price ratios. Also, prices
collected should be consistent in valuation and classification with the underlying expenditure data. From an
institutional perspective, therefore, the co-operation of price statisticians and national accountants in the
compilation and in the assessment of PPP results is key.

28.       Representativeness of product lists for ‘Pacific Rim’ countries needs improvement. Further
efforts will have to be undertaken to increase the number of comparable and representative products across
OECD countries, in particular between European and ‘Pacific Rim’ countries. This may entail pricing
more products in all countries, using generic and detailed specifications, and exploiting the potential of
hedonic functions where they exist. Bilateral contacts at the working level are also a useful tool to improve
comparability and visits should be organised on a regular basis.

29.       Research efforts on selected aspects have to continue. Many difficult areas remain in the PPP
programme. They cannot all be addressed at the same time and need focusing. Given the use of PPPs for
GDP comparisons, a useful strategy is to undertake methodological efforts in the PPP programme in the
same areas where specific efforts are made concerning price and volume measures in the national accounts.
A candidate for such an area is housing services, and work has started in the field of construction prices.
Possibilities to exploit alternative statistical sources and to involve expert knowledge from other parts of
the OECD work programme should equally be explored. This relates, for example, to prices of
telecommunication services.

30.        Communication counts. A sometimes-neglected part of the PPP programme is the effort to
communicate results, to enhance users’ understanding of PPPs and to ensure transparency of methods.
GDP per capita, PPP converted, remains the headline figure associated with the PPP programme and
attracts significant amount of attention. The significant coverage in the French press over the past year of
the evolution of France’s GDP per capita in relation to its European neighbours was a good example of the
visibility that PPP results can generate but also of some frequent mis-interpretations that may arise.

Conclusions for statistical offices

31.        To sum up, two consequences can be drawn for statistical offices. First, in a spirit of shared
responsibility between international organisations and national statistical offices for the published results, it
is important to maintain and where necessary to increase the commitment to the programme by allocating
the necessary resources and by raising awareness of the usage of PPP data. The first aspect applies in
particular to ‘Pacific Rim’ countries that do no benefit from international financing of the programme as is
the case for EU and Candidate countries. Second, within national statistical offices, the inter-action
between price statisticians and national accountants should be sought to bring in expertise from all
statistical areas concerned.

3.   At present, Eurostat devotes about 2.5 Million USD per year to the PPP programme. This amount includes the
     financing of data collection and treatment in countries. OECD does not have this possibility and depends on the
     voluntary participation of ‘Pacific Rim’ countries.



To top