PURCHASING POWER PARITIES 1999 BENCHMARK RESULTS

Paul Schreyer and Francette Koechlin, OECD Statistics Directorate

1.

Introduction

How does one compare economic data between countries that is expressed in units of national
currency? And in particular, how should measures of production and GDP be converted? One answer
to this question is to use market exchange rates. While straightforward, this turns out to be an
unsatisfactory solution for many purposes primarily because exchange rates reflect so many more
influences than the direct price comparisons that are required to make volume comparisons.
Purchasing Power Parities (PPPs) provide such a price comparison and this is the rationale for the
work of the OECD and other international organisations in this field (see Box 1). The OECD publishes
new sets of benchmark PPPs every three years, drawing on detailed international price comparisons.
Every time a new set of benchmark PPPs is released, this also gives rise to a new set of international
comparisons of levels of GDP and economic welfare.
The present article presents the set of benchmark results for 1999 which cover 43 countries. Second,
and importantly, it aims at shedding light on some of the uses and interpretation of PPPs. The
discussion includes the most frequent usage of PPPs that of generating comparable measures of GDP
per capita across countries but it will also point to other areas such as labour productivity, and the
comparisons of price levels. Furthermore, the paper takes a look at the evolution of PPPs and
associated measures over time thereby combining the spatial and the temporal aspect of PPPs.
2.

Spatial comparisons

2.1. 1999 GDP benchmark results: overview

Thirty OECD Member countries and 13 other countries are covered by the latest set of benchmark
results, based on purchasing power parities (PPPs) for the year 1999. These were drawn up in
collaboration with Eurostat (the Statistical Office of the European Commission) who provided data for
the 15 EU Member states, the 13 Candidate Countries to the European Union and Norway, Iceland and
Switzerland. The OECD is 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. Benchmark comparisons reflect 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. Along with the benchmark GDP results for 1999 (see Table 1) come new
estimates for OECD Member countries for the years 2000 and 2001 (Table 2) based on extrapolation
1

of the 1999 PPPs. For the first time, data for Korea came from a benchmark survey (earlier results
were based on OECD estimates).

Box 1. What are PPPs?

In their simplest form, PPPs are price relatives, which show the ratio of the prices in national
currencies of the same good or service in different countries. A well-known example of a one product
comparison is The Economists BigMacCurrency index, presented by the journal as burgernomics,
whereby the BigMac PPP is the exchange rate that would mean hamburgers cost the same in America
as abroad. The OECD-Eurostat PPPs, however, are not only calculated for individual products, they
are also calculated for product groups and for each of the various levels of aggregation up to and
including GDP. The purpose is similar: to obtain rates of currency conversion that eliminate the
differences in price levels between countries and so permit volume comparisons.
The calculation of PPPs is undertaken in three stages: first, at the product level, then, at the product
group level, where the price relatives are averaged to obtain unweighted PPPs for the group. Finally, at
the aggregation levels, the PPPs are weighted and averaged. The weights used in this last stage are the
expenditures on the product groups. All this is described in much more detail in the OECD publication
Purchasing Power Parities and Real Expenditures.
PPPs are price comparisons in space, and in many ways similar to the better-known price comparisons
over time. Comparisons over time often face the problem of changing products and consumption
patterns, especially when the years of comparison are remote. Somewhat similar, a great challenge
with spatial comparisons is that volumes or prices of sometimes very different economies have to be
compared. Goods and services and their prices that are characteristic in one country may be
uncharacteristic in another one, and yet common ground has to be found to make meaningful
comparisons. Regular benchmark surveys help keep product lists up to date so as to maximise
comparability.

Table 1 Purchasing Power Parities and International Comparisons of GDP

1999
Members
Non Members
PPPs
US$=1

Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
EURO 12
EU 15
OECD 30
Bulgaria
Croatia
Cyprus
Estonia
Israel
Latvia
Lithuania
Macedonia
Malta
Romania
Russian Federation
Slovenia
Ukraine

Exchange
rates
US$=1

Per capita GDP

OECD 30=100

Per cent of OECD 30

Indices
based on
PPPs

Indices
based on
exchange
rates

1.30
13.0
37.7
1.19
13.5
8.24
5.92
6.38
1.91
231
98.4
81.2
0.724
1554
162
755
39.7
5.63
1.97
1.43
9.25
1.77
127
13.6
125
9.64
1.89
197157
0.650
1.00
0.895
0.918

1.55
12.9
37.9
1.49
34.6
6.98
5.58
6.16
1.84
306
237.2
72.4
0.739
1817
114
1187
37.9
9.55
2.07
1.89
7.80
3.97
188
41.4
156
8.27
1.50
419688
0.618
1.00
0.938
..
..

109
113
109
117
60
124
103
102
109
70
51
122
114
106
110
60
190
37
117
83
128
40
75
49
84
104
127
26
103
149
102
102
100

92
115
109
94
24
147
111
106
114
53
21
138
113
91
158
39
201
22
112
64
153
18
51
16
68
122
161
12
109
150
98
101
100

0.443
3.80
0.383
6.21
3.64
0.246
1.52
16.7
0.294
4414
5.54
116
0.705

1.84
7.11
0.543
14.68
4.14
0.585
4.00
56.9
0.400
15339
24.62
182
4.130

28
37
86
38
88
29
34
27
56
24
26
70
16

7
20
61
16
78
12
13
8
42
7
6
45
3

Price Level Indices

GDP based
= PPPs/Exchange
GDP based
on
rates

84
102
100
81
39
119
107
104
105
76
42
113
99
86
143
64
106
59
96
76
119
45
68
33
80
118
127
47
106
101
96
99
100
24
54
71
43
89
42
38
30
74
29
23
64
17

Note: The cut-off date for all data used in the table was 31-December-2001.

Population

on PPPs

exchange
rates

1.85
0.82
1.00
3.19
0.55
0.59
0.48
5.48
7.98
0.66
0.46
0.03
0.38
5.45
12.50
2.53
0.07
3.22
1.65
0.28
0.51
1.38
0.67
0.24
2.98
0.82
0.81
1.55
5.48
36.38
27.64
34.53
100.0

1.56
0.83
1.00
2.57
0.22
0.70
0.51
5.73
8.38
0.50
0.19
0.03
0.38
4.70
17.92
1.62
0.08
1.91
1.59
0.22
0.61
0.62
0.46
0.08
2.40
0.97
1.03
0.73
5.81
36.66
26.55
34.02
100.0

1.70
0.72
0.92
2.73
0.92
0.48
0.46
5.39
7.35
0.94
0.90
0.02
0.34
5.16
11.35
4.20
0.04
8.73
1.42
0.34
0.40
3.46
0.89
0.48
3.55
0.79
0.64
5.90
5.33
24.45
27.18
33.78
100.0

0.20
0.15
0.05
0.05
0.48
0.06
0.11
0.05
0.02
0.48
3.45
0.12
0.73

0.05
0.08
0.04
0.02
0.43
0.03
0.04
0.01
0.01
0.14
0.78
0.08
0.13

0.74
0.40
0.06
0.13
0.55
0.21
0.33
0.18
0.03
2.01
13.10
0.18
4.49

Table 2 Purchasing Power Parities and International Comparisons of GDP

OECD Countries, 1999-2001*
PPPs
1999
Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
OECD TOTAL
EURO 15
EURO 12

Price Level Indices =

PPPs/Exchange rates

US$=1
2000* 2001**

1999

1.30
1.32
1.34
13.0
12.9
12.9
37.7
37.3
37.5
1.19
1.21
1.21
13.5
13.4
13.7
8.24
8.36
8.45
5.92
5.99
5.99
6.38
6.30
6.27
1.91
1.86
1.85
231
233
237
98.4 104.9 112.4
81.2
81.7
86.8
0.724 0.738 0.751
1554
1554
1567
162
156
150
755
727
726
39.7
40.2
39.7
5.63
6.10
6.33
1.97
1.99
2.06
1.43
1.44
1.48
9.25 10.51 10.74
1.77
1.85
1.91
127
128
132
13.6
14.1
14.6
125
126
128
9.64
9.52
9.48
1.89
1.87
1.86
197157 290424 439831
0.650 0.647 0.649
1.00
1.00
1.00

0.918
0.895

0.923
0.886

0.921
0.888

2000* 2001**

Per capita GDP

OECD 30=100
1999 2000* 2001**

84
102
100
81
39
119
107
104
105
76
42
113
99
86
143
64
106
59
96
77
119
45
68
33
80
118
127
47
106
101

81
91
90
86
36
109
98
93
93
67
39
110
91
78
153
68
97
68
88
69
126
45
62
32
74
110
117
49
103
105

76
92
91
86
39
112
99
94
93
68
43
98
94
80
136
62
97
74
92
68
131
51
65
33
76
101
121
43
103
110

109
113
109
117
60
124
103
102
109
70
51
122
114
106
110
60
190
37
117
83
128
40
75
49
84
104
127
26
103
149

108
113
109
117
60
123
106
101
108
70
52
122
122
105
108
63
195
38
116
83
126
40
75
49
84
104
126
27
102
149

108
113
110
117
61
124
105
102
108
73
54
122
127
106
107
63
199
38
117
83
127
40
76
50
85
105
127
24
103
148

100
99
96

100
90
86

100
91
87

100
102
102

100
102
102

100
103
102

*Extrapolated PPPs, based on 1999 benchmark. See text section 4 for explanation on extrapolation methodology.
**Results for 2001 were revised in April 2002.
Note: The cut-off date for all data used in the table was 31-December-2001.

2.2. GDP per capita: a measure of economic welfare

One of the most frequent uses of PPPs is in the computation of GDP and GDP per capita across
countries. Although GDP per capita has often been criticized as an incomplete statistic of economic
well being, it remains a cornerstone indicator of economic performance of individual countries. Policy
and analytical interest in this indicator goes a long way to explain the importance of PPPs as a
statistical tool.
It is to be noted in passing that market exchange rates are particularly ill-suited for comparisons of
living standards. This emerges from the fact that exchange rates tend to exhibit large swings over short
periods of time, implying rapid shifts of living standards between countries which cannot possibly
have occurred. Thus, OECD comparisons of GDP per capita here are based on PPPs and Table 1
shows the relevant results from the 1999 benchmark exercise.
To summarise results, four country groups have been identified on the basis of Table 1. Using groups
instead of a country-by-country ranking avoids possibly misleading interpretations (see Box) when
indices are clustered around a small range of results. Country indices are based on OECD 30 = 100,
which corresponds to an average of USD 21 500 per capita:
a high-income group (above 120): Denmark, Iceland, Luxembourg, Norway, Switzerland
and the United States;
a high-middle income group (between 100 and 119): Australia, Austria, Belgium,
Canada, Finland, France, Germany, Ireland, Italy, Japan, the Netherlands, Sweden and
the United Kingdom;
a low-middle income group (between 50 and 99): Cyprus, the Czech Republic, Greece,
Hungary, Korea, Israel, Malta, New Zealand, Portugal, the Slovak Republic, Slovenia,
Spain;
a low-income group (less than 50): Bulgaria, Croatia, Estonia, Latvia, Lithuania,
Macedonia, Mexico, Poland, Romania, the Russian Federation, Ukraine and Turkey.
For the first time, benchmark results are available for Korea. They show that Koreas per capita GDP
is 60 percent of the OECD average, a level similar to the Czech Republics and that Koreas economy
is somewhat smaller than the ninth largest OECD economy, Spain. However, when making these
comparisons, it should be kept in mind that PPP calculations are more difficult in countries that are
included for the first time than for those that have a history of participation going back several years.
Luxembourg occupies a remarkable position in these statistics. One of the reasons for Luxembourgs
high GDP per capita index is the large share of frontier workers in total employment (close to one
quarter). These persons contribute to GDP and employment but are not included in the total population
figure.
The 1999 results provide some telling examples for the differences between PPP-based and exchange
rate based comparisons. Consider per capita GDP for Japan, Norway or Switzerland relative to the
OECD average. When based on exchange rates, income per person would appear to exceed that of the
United States. However, when PPPs are used, these countries per capita GDP turns out to be lower
than that of the United States. This is because price levels are higher in these countries than in the

United States. When the price level effect is removed, the volume of goods and services purchased in
the United States is higher, on a per capita basis, than in any other country included in the comparison
except Luxembourg.
Generally, the gap between high-income and low-income countries narrows when PPPs are used
instead of exchange rates. Thus, the per capita indices based on PPPs of Mexico, Greece, Hungary,
Poland, Portugal, Turkey and the Russian Federation are closer to those of the United States than are
their per capita indices based on exchange rates. Again, this is because the price levels in these
countries are low compared to richer countries.
2.3. GDP levels: a measure of the size of economies
PPPs are also a tool to measure the relative size of economies. Table 1 shows each countrys GDP as a
percentage of total GDP of the 30 OECD countries. On this basis, the ten largest economies covered
by the comparison are the United States, Japan, Germany, France, Italy, United Kingdom, the Russian
Federation, Mexico, Canada and Spain. It is also confirmed that the 15 EU countries as a group are
virtually the same economic size as the United States.
Generally, there is a marked difference in the appreciation of the size of economies, depending on
whether exchange rates or PPPs are used to compare GDP data: the discrepancy is in particular present
in the group of low income countries. For example, on an exchange rate basis, the Russian Federation
corresponds to less than one per cent of total GDP in the OECD area. Corrected for differences in the
price level, this number rises to 3.5 per cent.

Box 2. Ranking may be misleading

When countries are clustered around a very narrow range of outcomes, it may be misleading to use the
per capita volume index based on PPPs to establish a strict order of ranking. As is often the case with
statistical information, there is a level of uncertainty associated with the data sources and procedures
on which PPP computations rely. Relatively minor differences in the measured per capita GDP can
result in a different country order that may not be statistically or economically significant.
As an example, consider the table below with GDP per capita indices for the EU15 countries in 1999,
and the corresponding ranking. Suppose there is a 1 per cent rise in the index for one country, say
Austria. With all other countries GDP per capita index unchanged, this would change Austrias
position from rank 5 to rank 4. Or, by the same argument, Germanys position would shift from 7 to 6.
What is important for the present purpose is the observation that as small a difference as 1 per cent in
results can yield quite different rankings when countries are as similar as for example the large
economies of the European Union. In fact, ranks change easily within the group of high-middle
income countries and it would, for example, be difficult to attribute economic significance to the
difference in the ranks of Sweden, Finland, the United Kingdom, Italy, or France. At the same time,
those EU countries that are part of the high-income group clearly keep their position, as do those EU
countries that are part of the low-middle income group.
GDP per capita 1999
EU 15=100
Index
Luxembourg
Denmark
Netherlands
Ireland
Austria
Belgium
Germany
Italy
Sweden
Finland
United Kingdom
France
Spain
Portugal
Greece
EU 15

Rank
186
121
114
112
111
107
106
103
101
101
101
100
82
74
68
100

1
2
3
4
5
6
7
8
9
9
9
12
13
14
15

2.4. Labour productivity: a measure of efficiency

Although GDP per capita comparisons command significant interest among analysts, they are not the
only pertinent statistic based on PPPs. One other useful indicator that also requires PPP-based volume
comparisons of output is the level of labour productivity, i.e., output per employed person. Estimates
of relative productivity levels provide insights into the possible scope for further gains in productivity
and competitiveness and also place a countrys growth experience in the perspective of its current
level of income and productivity. For inter-country comparisons of levels of labour productivity, PPPs
are indispensable: by definition, a measure of productivity has to put a volume measure of output in
proportion to a volume measure of inputs. Converting the value of output into a common currency by
applying exchange rates would seriously hamper the goal of capturing volume indicators of
production.
In Table 3, the 1999 benchmark PPPs and volume measures of GDP are used to compare relative
labour productivity measures across countries. When these productivity indices are put next to their
GDP per capita counterparts, considerable differences become apparent. They point to differences in
labour utilisation and demographic structures (and possibly statistical issues concerning the
measurement of labour input) and have been used extensively in analytical studies, such as the OECD
Growth Project1.

Box 3. Are PPPs suitable for industry-level comparisons?

PPPs at the GDP level are not normally adequate for productivity comparisons at the level of
individual industries. It is common practice to use PPPs for total GDP to convert output measures of
individual industries or sectors (such as total manufacturing) into a common currency. This may,
however, produce misleading statements about volume output and productivity of these industries as
economy-wide PPPs reflect relative prices of all goods and services, and not only the relative prices of
industry-specific output. There is no easy way out here: although researchers1 have developed
industry-specific PPPs, they only exist for a limited number of countries, industries and time periods.
1

See ICOP Project at the University of Groningen: http:/www.eco.rug.nl/GGDC/icop.html.

1.

See OECD (2001); The New Economy: Beyond the Hype; Paris.

Table 3 From GDP per Inhabitant to Labour Productivity*

Selected OECD Countries, 1999

Country

GDP per
capita

Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Portugal
Spain
Sweden
Switzerland
United Kingdom
United States

73
76
73
78
40
83
69
68
73
47
34
82
77
71
74
40
128
25
79
56
86
51
56
70
85
69
100

GDP per
person
employed
USA=100
78
81
96
81
43
81
76
87
81
62
45
73
89
97
72
46
111
32
80
60
84
52
79
75
78
74
100

*See Pilat and Schreyer (2002) Measuring Productivity; OECD Economic Studies No 33, 2001/2 for a discussion.
Differences in results with regard to Pilat and Schreyer are due to the use of new 1999 benchmark PPPs and
small revisions in the underlying GDP, population and employment data.
Sources: GDP and population data: OECD Annual National Accounts; GDP expressed at 1999 benchmark PPPs;
Employment from OECD Labour Force Statistics.

2.5. Comparative price levels: a measure of purchasing power disparities

Another key statistic derived from PPP measures is comparative price levels or the ratio between PPPs
and current exchange rates. If PPPs and exchange rates coincide, it can be concluded that, on average,
one unit of a currency buys as much in the country under consideration as it does in the reference
country. When PPPs exceed exchange rates, it can be concluded that one unit of the currency under
consideration buys less domestically than on other markets and vice versa. For example, Table 2
shows that a given volume of GDP costs on average 106 dollars in the United Kingdom, 68 dollars in
Portugal and 143 dollars in Japan. In other words, the general price level of Japan is higher than that of
the United Kingdom and of Portugal.
Box 4. Are PPPs obsolete in the Euro area?
Until 1999, in the OECD area, national territories and currency areas were identical in the vast
majority of cases. Since the introduction of the euro, this has changed markedly and 12 countries now
share a common currency. Does it still make sense to compute PPP measures for each national
territory? Or should there be a single PPP measure that converts the euro into other currencies?
From an economic viewpoint, there is no need to abandon country-specific PPPs for the Euro area.
One euro may very well have different purchasing power in different parts of the Euro area and so buy
more or less of the same bundle of goods. It is certainly interesting to find out whether a salary of
3 000, when paid in Finland, buys the same volume of goods and services as when paid in Germany
or Greece. In principle, the evolution of PPPs within the Euro area over time could be a tool to
monitor price convergence or absence thereof in the single currency area.
GDP measures of the Euro area as a whole are typically constructed by adding up GDP of the 12 Euro
area countries, each one expressed in euros. This is a perfectly acceptable measure of Euro area GDP
but it has to be kept in mind that this measure does not take into account the differences in purchasing
power of the euro within the currency area.
The price level effect is particularly visible in countries with low income per capita: there, exchange
rates often exceed PPPs rates by a substantive margin, indicating a comparatively low price level.
Partly, this is due to the economic importance of non-traded goods and services that are bought
relatively cheaply in low-income countries. It has long been noted that there is a positive correlation
between comparative price levels and GDP per capita: the richer a country, the higher its relative price
level tends to be, and vice versa. The 1999 results confirm this observation. In Figure 1, the line
depicting comparative price levels follows quite closely the sequence of bars that represent GDP per
capita indices. Overall, it can be concluded that an exchange rate based comparison will yield widely
understated measures for volume GDP and income in international comparison.
Just as the 1999 benchmark results gave rise to a grouping of countries according to relative income,
country groups according to relative price levels can be formed. These are:

A high price level group (110 and above): Denmark , Iceland , Japan , Norway,
Sweden and Switzerland ;

A medium-high price level group (between 90 and 109): Austria , Belgium, Finland,
France, Germany, Ireland, Luxembourg, the Netherlands, the United Kingdom and the
United States;

10

A medium-low price level group (between 60 and 89): Australia, Canada, Cyprus,
Greece, Israel, Italy, Korea, Malta, Mexico, New Zealand, Portugal, Slovenia and Spain;

A low price level group (less than 60): Bulgaria, Croatia, the Czech Republic, Estonia,
Hungary, Latvia, Lithuania, Macedonia, Poland, the Slovak Republic, Romania, the
Russian Federation, Turkey and Ukraine.

Box 5. PPPs: a tool to predict exchange rate movements?

It has been argued that in a world where all goods are traded and where markets are fully efficient, the
exchange rate should converge towards the PPP of a currency. Such convergence, proposed by the
PPP theory of exchange rates would imply that the same price levels should be observed across
countries. The PPP theory of exchange rates goes at least back to Gustav Cassels who applied it to
currency movements during World War I:
[] I propose to call this parity purchasing power parity. As long as anything like free movement of
merchandise and a somewhat comprehensive trade between [the] two countries takes place, the actual rate of
exchange cannot deviate very much from the purchasing power parity. (Cassels 1918)1.

On these grounds, comparative price levels are sometimes used to pass a judgement about the over- or
under-valuation of a particular currency. Empirically, however, many of the premises needed for the
PPP theory of exchange rates do not hold or do not hold any more in practice in particular, there is a
large number of non-traded goods, including construction and most services, and there are significant
influences of other factors (e.g., growth prospects, equity prices, capital flows) on exchange rates.
Thus, it is far from obvious to consider comparative price levels as reliable predictors for medium- or
even long-term currency movements. R. Dornbusch2 concludes that:
The theory remains controversial [] because strict versions are demonstrably wrong while soft versions
deprive it of any useful content. In between there is room for theory and empirical evidence to specify the
circumstances under which [] PPP provides a useful though not exact description of exchange rate behviour.
1

Cassels, G. (1918); Abnormal deviations in international exchanges; Economic Journal 28.

Dornbusch, R. (1987); Purchasing Power Parity; in Eatwell et al. (eds.); The New Palgrave: A Dictionary of
Economics.

11

200

180

160

140

120

40

100

40

20

80

20

60

0
Luxembourg
United States

High
income
group

Norway
Switzerland
Denmark

120

Iceland
Netherlands
Canada
Ireland
Austria
Japan

Italy
Sweden
Finland
United Kingdom

90

France
Israel
Cyprus

12

Spain
New Zealand

Greece
Korea

Comparative Price Levels

Low middle
income group

Portugal
Slovenia

Czech Republic
Malta

50

Hungary
Slovak Republic
Poland
Estonia
Mexico

Low income
group

Croatia
Lithuania
Latvia
Bulgaria
Macedonia
Russia
Turkey
Romania
Ukraine

Figure 1 Per capita GDP and comparative price levels, 1999

Belgium
Germany

Per capita volume indices

High middle
income group

Australia

200

180

160

140

120

100

80

60

3.

Inter-temporal comparisons: has the picture changed over past years?

So far, PPPs have been discussed as currency conversion rates for a given point in time they provide
a snapshot of relative prices and expenditure on GDP data converted by PPPs provide a snapshot of
relative volumes in that particular year. For many analytical purposes, it is of interest to observe the
evolution of volume GDP between countries and over time. There are at least two ways of setting up
such a comparison, each with its specific interpretation and use.
3.1. One option for year-to-year comparisons: a sequence of current PPPs
The first possibility of combining spatial and temporal observations is by using a sequence of current
or benchmark PPPs, i.e., a new set of price data compiled in Member countries, weighted and
aggregated to yield rates of currency conversion for total GDP and its expenditure components. This
means that prices and price structures are allowed to vary over time. Comparable volume levels of
GDP are obtained by applying these current PPPs to GDP measures at current national prices. Within
a given year, (spatial) comparisons between countries are straightforward volumes are measured
with the same price structure. Comparisons over time, however, incorporate several effects: relative
volume changes, changes in relative prices between countries and, possibly, changes in definitions and
methodologies. One can also say that by carrying out this calculation for every period, GDP
comparisons across countries are based on current international prices.
3.2. Another possibility: constant PPPs
A second approach to generate time series of PPPs is to fix a base year and to extrapolate PPPs for
other years. Extrapolation is done by applying the relative rates of inflation observed in different
countries to the base year PPPs. Consider a simple two-country example and suppose that PPPs in year
1 are equal to 1: one currency unit of country A buys, on average, the same amount of goods and
services as one currency unit in country B. Now assume that, between the two periods, the price level
of GDP in country A rises by 20% whereas, on average, prices in country B remain unchanged. PPPs
are then extrapolated by applying a ratio of 120/100=1,2 to the initial PPP. Thus, extrapolated PPPs
for year 2 are PPP of year 1*1,2=1,2.
GDP series in national currency and at current prices can now be converted with these PPPs to yield
volume measures that are comparable across countries. The resulting measures of GDP comparisons
are volume indices at constant prices and PPPs. The same result would have been achieved by
applying volume growth rates of GDP to the comparative GDP levels of the base year.
Whichever way they are calculated, these time series have a very convenient property: they replicate
exactly the relative movements of volume GDP growth of each country. While such a characteristic
facilitates the use and interpretation of PPPs over time, it shares an important drawback with other
indices that use a fixed base: the assumption that price structures do not change over time. Economic
reality has it, however, that relative prices do change over time and it is well known that ignoring
these shifts over longer periods can generate a biased picture of economic developments. Another
consequence of fixing price structures at a base year is the dependence of results on the choice of the
base year.

13

3.3. What are the conceptual and methodological differences?

The key conceptual difference between using current and constant PPPs is that the former capture
changes in volume as well as changes in relative prices, whereas the latter only capture volume
changes. Even if the volumes of goods and services remain identical over time, a GDP comparison
based on current PPPs may change over time if prices and price structures shift. This factor comes into
play when some countries are large producers and exporters of products with marked price changes, as
has been the case for Norway as an important exporter of oil.
Another source of differences between the two approaches is methodological changes between
successive rounds of price collection. For example, the introduction of the 1993 System of National
Accounts brought with it changes in product classification that affected PPP computations. While such
changes help to improve comparability across countries once they are put in place, they also reduce
comparability with observations before their introduction and a break in series occurs. Sometimes,
simple changes in price collection methodologies have similar effects and reduce inter-temporal
comparability. The OECD is currently analysing the impact of certain breaks in series on overall
results.
There may also be differences in the ways in which statistical offices construct implicit price indices
for their GDP series. Such differences will directly influence the extrapolated PPP measures and so
account for some of the observed differences between GDP based on current and constant PPPs.
Even when there is no change in methods, systematic methodological differences exist. For example,
price changes of exports and imports in the national accounts are based on import and export price
indices. For PPP evaluations, export and import price ratios between countries are simply measured as
bilateral exchange rates. Also, price comparisons in the PPP programme treat products as identical
even when they are sold at different locations or at different types of outlets. National accounts
principles foresee that products delivered in different locations should be treated as different qualities
even if they are otherwise physically identical. Lastly, simple measurement errors can enter the
picture. Such measurement errors may be due to small samples and due to the difficulty of comparing
like with like across countries.
3.4. Does it matter in practice? A comparison
To illustrate the differences between methods, results for calculations based on current PPPs are
compared with results based on constant PPPs. Consider Table 4 below which takes Portugal as an
example. When the table is read along a line, one obtains the GDP per capita index. Thus, the change
in Portugals GDP per capita relative to other OECD countries between the years 1990 and 1999 has
been 5 percentage points, whatever the base year.
When reading the table diagonally, one follows Portugals GDP per capita index at current PPPs.
Now, the change between 1990 and 1999 is 11 points (75 minus 64), or almost double. The differences
between the two results comprise shifts in price and volume structures but also changes in statistical
methodology, concerning both PPPs and national accounts. At present, it is not possible to disentangle
the purely methodological from the price and volume structure effects.
Finally, when reading the last column of the table, we measure the relative position of Portugals GDP
per capita according to different benchmark years. The range of estimates varies between 69 and 77 or
8 percentage points. Over a larger number of countries, the range of estimates is about 5 percentage
14

points, and this provides one justification why a 5 percentage point error margin is sometimes quoted
in conjunction with these results.
Table 5 provides a comparison between GDP per capita indices based on current and constant 1999
PPPs for a broader number of countries. Some of the differences are large and probably more
indicative of changes in methodologies than changes in price structures. On the whole, therefore, the
OECD recommends indices based on constant PPPs for the analysis of relative growth performance
between countries and the latest current (benchmark) PPPs for the latest snapshot comparisons of the
GDP and GDP per capita.
Box 6. Current and constant PPPs: what the OECD publishes1 and recommends
The OECD has followed a two-way strategy concerning the choice between benchmark and
extrapolation results for international comparisons.
First, in Annual National Accounts Vol. 1, the OECD publishes time series based on both current and
constant PPPs. The former enter under the heading GDP per capita at current prices and current
PPPs, the latter are labelled as GDP per capita at price levels and PPPs of 1995. The accompanying
text recommends GDP per capita data based on constant prices and constant PPPs for comparisons
over time. Thus, these data are best suited to answer a question such as: How has the relative position
of a countrys GDP per capita changed, given its measured growth performance?
Current (benchmark) PPPs are put forward as the appropriate tool to compare GDP levels for the latest
period available, as they reflect the most recent and most relevant price structure. These data are wellsuited to respond to the question: What is a countrys position in terms of GDP (per capita), given the
most recent set of international prices?
Second, the OECD has to deal with the fact that benchmark PPPs are available for the European
OECD countries on an annual basis provided by Eurostat - but on a three-yearly basis only for other
OECD countries. There is thus a choice of presenting benchmark PPPs only at three-year intervals and
thereby losing valuable information for many European countries or of producing estimates for
intermediate years for non-European countries. The OECD has followed the latter option. Estimates
on intermediate years are one-year extrapolations backwards or forward to benchmark years. For
example, with 1996 as a benchmark years, the 1995 PPP is extrapolated backwards from 1996, and the
1997 PPP is extrapolated forwards. Of course, these estimates are only necessary for those countries
that are not covered by Eurostats annual benchmark exercise. Currently, the PPP comparisons
published in the OECD Main Economic Indicators reflect this method.
Further, the OECD has to decide how to update the latest benchmark results. To date, the latest
available benchmark year for all countries is 1999. PPP series for 2000 and 2001 are obtained by
extrapolating on the basis of relative price indices for GDP with the results presented in Table 2 and,
more generally, in statistical publications that relate to the years 2000-2001. From a conceptual point
of view, extrapolating PPPs from 1999 onwards and applying them to GDP data is tantamount to
expressing this GDP data in 1999 constant prices and constant PPPs.
Lastly, it should be noted that there is full consistency of current PPPs between the data published by
Eurostat and by the OECD for European countries. Differences may occur when publication dates are
different or due to small differences in GDP or population data. Also, the OECD expresses its results
in dollars and typically related indices to OECD30=100.
1

All OECD books and statistical publications are available on line at www.SourceOECD.org.

15

Table 4 GDP per capita with current and constant PPPs for Portugal
OECD 30=100
Based on:

-constant PPP of
benchmark year:

1990
1991
1992
1993
1994
1995
1996
1997
1998
1999

1990
64
64
66
68
71
70
69
72
70
70

1991
67
67
68
71
74
73
72
75
73
73

1992
66
67
68
71
74
73
72
75
73
73

1993
65
65
66
69
72
71
70
73
71
71

Results for:
1994 1995
64
65
64
65
65
67
68
69
70
72
69
71
69
70
72
73
70
71
70
72

1996
66
66
68
70
73
72
71
74
72
73

1997
67
67
68
71
74
73
72
75
73
73

1998
68
68
70
72
75
74
73
77
74
75

Table 5 GDP per capita at current and constant PPPs

OECD 30=100
Current prices and
PPPs
Australia
Austria
Belgium
Canada
Czech Republic
Denmark
Finland
France
Germany
Greece
Hungary
Iceland
Ireland
Italy
Japan
Korea
Luxembourg
Mexico
Netherlands
New Zealand
Norway
Poland
Portugal
Slovak Republic
Spain
Sweden
Switzerland
Turkey
United Kingdom
United States
EU15
OECD TOTAL

1990 1993 1996 1999

106 104 108 109
109 112 112 113
109 114 111 109
120 114 117 117
..
..
65
60
110 113 121 124
106
92
97 103
111 107 101 102
106 109 108 109
60
64
67
70
..
..
46
51
113 108 117 122
76
83
93 114
106 102 104 106
118 121 122 110
..
..
..
60
157 171 169 190
..
..
36
37
107 106 109 117
88
85
88
83
113 123 127 128
..
..
37
40
64
69
71
75
..
..
46
49
79
80
79
84
114 100 102 104
139 136 126 127
30
32
30
26
104
99 100 103
148 146 145 149
101 100 100 102
100 100 100 100

1999 constant prices and

PPPs
1990 1993 1996
104 103 106
114 114 113
110 109 108
120 114 113
72
62
66
122 120 123
107
92
96
107 105 102
115 114 111
70
69
68
..
47
47
126 117 118
79
82
95
111 109 108
118 119 118
46
54
61
154 174 174
39
39
36
113 113 114
89
85
87
116 122 129
34
33
37
70
71
73
..
43
47
81
80
81
109 100 101
150 140 130
27
29
28
102 100 103
144 144 145
104 103 102
100 100 100

Note: The cut-off date for all data used in the table was 31-December-2001.

16

1999
109
113
109
117
60
124
103
102
109
70
51
122
114
106
110
60
190
37
117
83
128
40
75
49
84
104
127
26
103
149
102
100

1999
69
69
70
73
76
75
74
77
75
75