1

OECD PPP Meeting WellingtonFebruary 2004

The PPP Programme and the Uses of PPPs

Paul Schreyer OECD

2

Upfront…

PPPs are not a concept that is immediately and easily grasped by users (everything is relative, unclear links with national accounts and national price indices). This has generated a fair bit of mis-understandings

PPPs are used for purposes for which they are not suited

PPPs are not used even when they are the right concept

3

Contents

1.What are PPPs?

2.The PPP programme

3.Uses of PPPs

4

1. What are PPPs?

PPPs are price relatives calculated in three stages:

first for individual products (tomatoes) then for product groups (food) finally for aggregates (PHC, GDP)

the weights for aggregation are provided by the National Accounts

PPPs are spatial price indices that allow “real” volume comparisons across countries

5

2. The organisation of the PPP programme

• Australia• Canada• Japan• Korea• Mexico• New Zealand• United States• Israel

• Finland• Denmark• Estonia• Iceland• Ireland• Latvia• Lithuania• Norway• Sweden• UK

• Austria• Belgium• Czech Rep.• Germany• Hungary• Luxembourg• Netherlands• Poland• Slovak Rep.• Slovenia• Switzerland

• Italy• Bulgaria• Cyprus• France• Greece• Malta• Portugal• Romania• Spain• Turkey

• Croatia• Macedonia• Russian Fed.• Ukraine

Eurostat OECD

Northern group Central group Southern group Pacific countries Non-OECD members

6

Basic requirements from countries

A set of national annual prices for a selection of products chosen from a common basket of goods and services

the basket consists of about 3000 goods and services these products should be both comparable and representative

A breakdown of national expenditures

7

Selection of products for the basket

The basket of goods and services is a sample of all goods and services covered by the GDP

for individual consumption of households : surveys on consumer goods and services (e.g., Food, beverages and tobacco; Transport services; Rents)

for gross fixed capital formation : surveys on construction projects, equipment goods...

for government consumption : surveys on salaries

8

How are products chosen?

Product lists Products have to be :

representative of the consumption patterns

comparable between countries Lists established in co-operation

between Eurostat groups with ‘Pacific-rim’ countries

9

How are PPPs calculated?

PPPs are price relatives calculated in three stages

I: prices relatives for individual products II: price relatives for product groups by averaging

the price relatives for individual products

III: price relatives for aggregates by weighting and averaging the price relatives for the product groups

10

3. Uses of PPPs: a traffic light approach

1. Areas where PPPs are well placed

2. Areas where PPPs may be used but with limitations

3. Areas where PPPs should not be used

11

Main indicators:

1. Volume comparisons of GDP

- GDP per capita

- GDP per hour worked

- Size of economies

2. Comparisons of relative price levels

Areas where PPPs are well placed

12

Generally: compare with care…

• PPPs are a useful instrument but with limited precision

• Small differences between countries should not be over-interpreted

• Results may change due revisions in GDP and population data

• Example: GDP per capita across OECD countries

GDP per capita

13

1999 ‘Benchmark’ Results: GDP/head

0

10

20

30

40

50

60

70

80

90

100

110

120

130

140

150

160

170

180

190

200

High income group

High middle income group

Low middle income group

Low income group

120 90 50

14

0

20

40

60

80

100

120

Icel

and

Canad

a

Nether

lands

irela

nd

Austria

Japan

Belgiu

m

Ger

man

y

Austra

lia

Swed

en Italy

United

Kingdom

Finla

nd

France NZL

Spain

Portugal

Gre

ece

Korea

Czech

Repu

blic

Hungar

y

1999 Results: Middle income group(Information set: December 2003)

450 $

15

1999 Results: Middle income group(Information set: December 2003)

0

20

40

60

80

100

120

Icel

and

Canad

a

Nether

lands

irela

nd

Austria

Japan

Belgiu

m

Ger

man

y

Austra

lia

Swed

en Italy

United

Kingdom

Finla

nd

France NZL

Spain

Portugal

Gre

ece

Korea

Czech

Repu

blic

Hungar

y

5158 $

16

1999 Results: High middle income groupunder two information sets

40

50

60

70

80

90

100

110

120

130

Nethe

rland

s

Canad

a

Irelan

d

Austri

a

Japa

n

Austra

lia

Belgiu

m

Ger

man

yIta

ly

Sweden

Finlan

d

Unite

d Kin

gdom

Franc

eIs

rael

Spain

New Z

eala

nd

Portu

gal

Gre

ece

Korea

Czech

Rep

ublic

Hunga

ry

Jan-02 Jul-03

17

GDP per capita

• A measure of output, and with some reservations, a measure of economic well-being (NDP per capita and PPPs for NDP?)

• EU: allocation of structural funds

• OECD: no use for membership fees but

• Analysis: from GDP per capita to GDP per hour worked: a useful analysis

• Statistics: zone totals

18

(7) (8)(1) - (6) (1) - (2) - (3) - (4)

Canada 85 3 2 -2 -2 84 82Mexico 26 -2 -5 1 1 31 32United States 100 0 0 0 0 100 100

Australia 76 0 0 0 0 76 77Japan 74 1 2 0 -1 72 72Korea 48 3 -6 1 12 37 49New Zealand 61 -1 1 0 0 61 61

Sweden 74 -2 2 0 -11 85 74Total OECD 75 0 -3 0 -2 81 78

Un-employ-

ment

Working hours

Effect of (%)

Working-age

population (15-64

years) to

Labour force to

working-agepopulation

GDP per capita(US = 100)

GDP per hour

worked(US = 100)

GDP per person

employed(US = 100)

(1) (2) (3) (4) (5)

19

Mexico

Canada

Spain

Korea

Australia

Turkey

Sweden

Switzerland

Norway

New Zealand

Slovak Republic

Based on PPPs Based on exchange rates

34 2 1 0 1 2 3 4

Size of economies

20

CPL is intuitively familiar to any international traveller: comparison between the relative prices of the same product at home and abroad and the market exchange rate.

Example:

• NZL traveller pays hotel room in Chicago: 100 USD

• Traveller’s calculation whether this is “expensive” or not:

Convert 100 USD into NZD with market exchange rate, say 1.5 NZD/USD: 100 USD*1.5 NZD/USD = 150 NZD

• Take price of an equivalent hotel room in Wellington.

Comparative price levels (1)

21

• If the equivalent hotel room there costs 100 NZD, the traveller will perceive the price level for hotels in the United States as high compared to Wellington.

• Thus, travelling in the USA is ‘expensive’ if the PPP for hotel rooms (100 USD/100 NZD=1 USD/NZD) exceeds the exchange rate (1 USD/1.5 NZD=0.67 USD/NZD) and vice versa.

• Comparative price level = PPP/Exchange rate

• Thus, CPL varies with exchange rates

• This has to be kept in mind whether observing CPLs over time

Comparative price levels (2)

22

A look at New Zealand

New Zealand’s and Australia’s price levels compared to the United States…

Comparative price levels (3)

1999 2000 2001 2002NZL NZD 76 65 52 68Australia AUD 83 76 68 73United States USD 100 100 100 100

23

• Cost of living index across countries

• Time series analysis of relative GDP per capita

• Analysis of price convergence

Areas where PPPs may be used but with limitations

24

• Relative price level of actual final consumption

• For average consumption patterns of resident population Not for expatriates

• Some important aspects of prices of private consumption are inadequately covered, in particular housing (volume index)

Cost of living index yes, but…

25

• Take NZL’s per capita GDP relative to that of the OECD as a whole in 1996

• Multiply that measure by the rate of GDP growth for NZL relative to that of the OECD as a whole between 1996 and 1999 and divide by the relative population growth

• Surely, the resulting GDP per capita index for NZL for 1999 corresponds to the one that comes directly out of the 1999 survey?

Time series analysis of relative GDP or price levels (1)

26

• Unfortunately, no. And this is a source of much discussion

• Even more so as this is what policy makers want to know: where does my country stand relative to others and how has its position evolved over time?

• Why are there differences?

Time series analysis of relative GDP or price levels (2)

27

• The first calculation used fixed PPPs (fixed 1996 international prices)

• The second result comes from the 1999 survey and thus reflects current 1999 international prices

• One source of differences are thus changes in price structures

• But there are other sources. Here is an important one, the underlying samples

Time series analysis of relative GDP or price levels (3)

28

Differences in samples: price indices underlying national accounts and PPPs

CPI sample inperiod t

CPI sample inperiod t+1

PPP sample in period t

PPP sample in period t+1

t+1/t

t+1/t

29

• PPP samples tend to be small and change over time: this makes the temporal comparison of price levels difficult. But PPP samples are conceived to maximise comparability of items across countries.

• Samples for temporal price indices tend to be larger and conceived to maximise comparability over time within a country. But they are not set up to deliver international comparability of items.

• Thus, when PPP samples change over time, it makes little sense to compare prices over time.

Time series analysis of relative GDP or price levels (4)

30

• OECD recommends constant PPPs for time-series analysis (but note trade-off between advantage of national accounts consistency and disadvantage of imposing constant international price structure)

• How about NZL, then?

• Despite some differences between series based on constant and current PPPs, the discernable trends are very similar change in NZL’s relative position of GDP/head is not a PPP issue

Time series analysis of relative GDP or price levels (5)

31

75

80

85

90

95

100

105

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

Constant PPPs

Current PPPs

32

Time series analysis of relative GDP or price levels (6): Example Norway and

oil prices

75

85

95

105

115

125

135

145

155

165

1980

1982

1984

1986

1988

1990

1992

1994

1996

1998

2000

Constant PPPs

Current PPPs

33

• In principle, the spread of prices over time can be used to make a statement about price convergence, in particular in the Euro area

• But: does not make sense for all products (e.g. non-tradeables) – special basket may be needed

• Also: small sample size in PPP surveys, such convergence cannot be measured at product level

• Also: which prices are expected to converge? How to account for taxes and subsidies?

Analysis of price convergence

34

1. As an indicator for the over- or under-valuation of a currency

2. As a precision tool to establish rankings between

3. As a measure to generate output and productivity comparisons by industry (unless there are industry-specific PPPs)

4. As a way of constructing national growth rates

5. As a ‘volume’ measure of exports or imports across countries

PPPs are not recommended…