Transcript Slide 1
Searching for Coherence between Spatial
and Temporal Measures of International
Price Change
Getachew Tessema
and
Peter Rossiter
Outline
1. Background
2. Research objective
3. Methodology: Rhoades' ideas
4. Data sources and issues
5. Empirical results
6. Conclusions
1. Purchasing power parities (PPPs)
Often economists are interested in comparisons
between countries such as:
- economic activity/GDP/GDP per capita
- level of poverty
- productivity
These comparisons require
the conversions of national
PPPs serve this purpose
currencies into comparable
units
1.1 Construction of PPPs
SNA data in
national
currencies
Prices of goods
and services
Construction of PPPs
involves
A reference
country or
countries
Aggregation
method
1.2 Construction of PPPs
SNA data in
national
currencies
Prices of goods
and services
collected
infrequently
(every 3 years
by the ECP))
Construction of PPPs
involves
A reference
country or
countries
Provided by
member
countries
Index formulae
Usually
EU15
The EKS
method
1.3 Some points to note about PPPs
PPPs constructed for the years where prices are
directly collected are called Benchmark PPPs.
PPPs for subsequent years are extrapolated using price
movements from the time-series based national accounts.
1.4 The problem
When extrapolated PPPs are compared with
latter benchmarks there is almost always
inconsistency between them.
1.5 Potential sources of inconsistency
Theoretical
(index formulas)
Benchmark PPPs
versus
extrapolated PPPs
Measurement
(e.g., prices)
Revisions to SNA
(e.g., expenditure)
2. Objective
To get a better understanding of the sources of
inconsistency
To measure their relative contributions
To make suggestions to improve future PPP
projections
3. Methodology: Rhoades' ideas
Decomposes inconsistency into its sources
Provides an integrated framework to analyse
PPP-implied and time-series based estimates
systematically
Based on progressive substitution of data items from
PPP system to the Time-series system (or vice versa)
3.1 Methodology: analytical framework
GDPtS= i p Sit q Sit ,
S(R)
PPPt
w S(R)
it
i
GDPtR= i p Rit q Rit
S 1 2
p it
R
p it i
S 1 2
R p it
w it R
p it
= Bilateral Fisher Index
+ e S(R)
t
+ adjustment
Then convert S to R's currency as follows:
S(R)
GDPt =
GDPtS
PPPtS(R)
S(R)
GDPt+
k=
S
GDPt+ k
S(R)
PPPt+ k
3.1 Methodology: analytical framework
Consider the spatial relatives at time t and t+k:
S(R)
GDPt+ k
R
GDPt+ k
S(R)
GDPt
R
GDPt
PPP-implied
Real Growth in S
R R
p
i it q it+ k
R R
p
i it q it
S S
p
i it q it+ k
S S
p
i it q it
GDPS(R)
t+ k
R
GDP
t+
k
GDPtS(R)
R
GDPt
IPDSt+ k
R
IPD
t+ k
PPPS(R)
t
S(R)
PPP
t+ k
3.1 Methodology: PPP-implied price change
PPP-implied
price change in S
R
R
p
q
i it+ k it+ k
Time-series based
Price deflator (IPD)
R R
p
i it qit+ k
PPPS(R)
t+ k
R
PPP
t+ k
PPPtS(R)
R
PPPt
PPP-b ase d
t io s
ch an ge s in PPP r a
3.1 Decomposing inconsistency
Time series system
Step 1.
PPP-system
Time series aggregate price
change
PPP-implied aggregate price change
Expenditure
1
Re-calculate PPP-implied aggregate
price change
+
Relative prices
+
Fisher index
2
3
Re-calculate PPP-implied aggregate
price change
Re-calculate PPP-implied aggregate
price change
Compare
Step 2. Substitute data items from PPP-system to Time series system
3.1 Decomposing inconsistency
Time series system
Step 1.
PPP-system
Time series aggregate price
change
PPP-implied aggregate price change
Expenditure
1
Re-calculate PPP-implied aggregate
price change
+
Relative prices
+
Fisher index
2
3
Re-calculate PPP-implied aggregate
price change
Re-calculate PPP-implied aggregate
price change
Compare
Step 2. Substitute data items from PPP-system to Time series system
4. Data used
Data for OECD countries provided by Eurostat
Three countries are selected for case study (Germany, Ireland and the
United Kingdom)
Three-level of disaggregation of Domestic Final Demand (DFD)
Actual Individual Consumption (AIC)
Collective Consumption (CC)
Gross Fixed Capital Formation (GFCF)
The study covers the years 1999-2002
5. Discussion of Results
% points
5.1 Aggregate price change – Domestic final demand, 1999–2002
SN
IR UK
A Germany
SN
DE UK
A Ireland
SN
DE IR
A United Kingdom
% points
5.2 Real growth – Domestic final demand, 1999–2002
UK
SN
IR
A
Germany
SN
A
DE
Ireland
UK
SN
DE IR
A
United Kingdom
% points
5.3 Net effects on aggregate price changes,
1999 - 2002
* indicates reference country
% points
% points
5.4 Net effects on aggregate price changes, 1999 - 2002
% points
% points
5.5 Net effects on aggregate price changes, 1999 - 2002
% points
% points
5.6 Net effects on aggregate price changes, 1999 - 2002
% points
5.7 Net effects on real growth in DFD, 1999 - 2002
% points
% points
5.8 Net effects on real growth in DFD, 1999-2002
% points
% points
5.9 Net effects on real growth in DFD, 1999-2002
% points
5.10 Sub-aggregate level effects on aggregate price
change, 1999-2002
5.11 Sub-aggregate level expenditure effects on aggregate
price, 1999-2002
6. Conclusions and recommendations
Inconsistency is predominantly driven by price differences
(based on relative prices that depend on the reference country)
Index effects can be large depending on the reference country
The issue of a reference country is important
We believe that the framework provides important
insights for post-hoc analysis
There appears to be some potential to minimise
inconsistency of future PPP projections
For example:
1. Extrapolate component-level PPPs using the time-series price
movements,
2. Then use Fisher Index to combine these extrapolated PPPs
(depending on the availability of disaggregated data)
Thank you