Digitized by the Internet Archive
in
2011 with funding from
Boston Library Consortium
Member
Libraries
http://www.archive.org/details/productpricesoecOOkraa
DEWEY
HB31
.M415
working paper
department
of
economics
PRODUCT PRICES AND THE OECD CYCLE
Aart Kraay
Jaume Ventura
No. 99-01
January, 1999
massachusetts
institute of
techinology
50 memorial drive
Cambridge, mass. 02139
WORKING PAPER
DEPARTMENT
OF ECONOMICS
PRODUCT PRICES AND THE OECD CYCLE
Aart Kraay
Jaume Ventura
No.
99-01
January, 1999
MASSACHUSEHS
INSTITUTE OF
TECHNOLOGY
50 MEMORIAL DRIVE
CAMBRIDGE, MASS. 02142
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
Product Prices and the
OECD
Cycle
Aart Kraay
The World Bank
Jaume Ventura
Massachusetts
Institute of
Technology
January 1999
Abstract
:
It
is
well
known
a
the same group
that business cycles in industrial countries exhibit
of
remarkable degree of synchronization. Much less known is that In
peak of the cycle is associated with high prices of labourintensive products and low prices of capital-intensive ones. We document this cyclical
behavior of product prices and argue that it offers an important clue as to why business
cycles are so synchronized. Positive shocks In one or more countries raise the prices
of labour-intensive products and, as a result, the demand for labour throughout the
industrialized world. This generates increases in wages, employment and output in all
industrial countries. Through this channel, shocks are positively transmitted across
countries, creating a force towards the synchronization of business cycles.
industrial countries, the
We thank Daron Acemoglu for helpful comments. Some of the material
under the
title
Trade and
Fluctuations".
necessarily reflect those of
In this
The views expressed here are the
The Worid Bank.
paper was circulated
and do not
authors'
earlier
It
hardly
seems necessary
among
synchronized
document
to
industrial countries.
Table
that business cycles are strongly
reports the time-series correlations
1
OECD average growth excluding that
country, over the period 1960-1996, for all OECD countries. These correlations are
substantial, averaging 61% in the G7 and 47% in the full OECD sample. One might
of annual real per capita
GDP
growth with
think that these correlations are high
increases
in
most
of the
even
if
we
the price of
the
oil in
OECD economies.
respectively.
Since
These
much
a strong
The
of
of
Table
54% and 46%
1
percent
in
the
G7 and
figures certainly justify the notion of
of the fluctuations in output
correlation
2 countries
with
shows
that
for
is
it
among
an
full
can be traced back
to fluctuations in
not surprising that labour market indicators also
industrial countries.
Table 2 documents
a
this for
which internationally-comparable indicators are available.
OECD average per capita GDP growth
high, averaging
41%, 43%, and 24%,
respectively.
demand
of correlations is that shifts in the
employment and
real
excluding that country are quite
A
natural interpretation of this set
for labour tend to
occur at the
same
time
in
countries.
Why are shifts in the demand for labour highly correlated
this paper,
we
argue that the
clue towards the
we
answer
cyclical
refer to the observation that
products. This can be
AInpirt
-Xi,)-
year
t
relative to the
and decreases
In
cyclical
behaviour of product prices,
in
increases
in
the
the price of capital-intensive
estimating the following regression model:
AinY, +(Xi
where Alnpid denotes the growth
in
By the
OECD booms are associated with
shown by
=(p, +P2
across countries?
behavior of product prices offers an important
to this question.
price of labour-intensive products
c
1
for
OECD samples,
OECD cycle.
the
correlations of annual growth rates of manufacturing hours,
wages
all
970s which constituted a large adverse shock
However, the second column
employment and hours worked,
sample
large
exclude the 1973-1981 period from our sample, the cross-country growth
correlations average
exhibit
1
because the sample period includes the
+X2 •X|,)AIny^
+fi,
+u
ict
rate of the price index of output in industry
consumer
price index in country c in year
1
t;
Xic Is
i
in
country
the average
over time of labour's share
and denote
real
respectively;
per capita
and
fic
and
in
value-added
GDP
industry
in
growth rates
in
the
and country
i
OECD and
in
c; AInYct
country c
and
in
Ainyct
year
t,
are a country- and industry-specific fixed effect and a well-
Uict
We estimate the model using data on 28 manufacturing
sectors in a sample of 20 OECD economies over the period 1963-1995.
behaved
error, respectively.
The
results reported in
specifications, the estimate of
we
is
interpret
Table 3 point to three conclusions.
(32 is
as evidence that the
positive
and
significantly different
sensitivity of product price
higher for labour-intensive products. For instance,
if
we
across
First,
changes
all
from zero, which
to
OECD growth
take two products with
labour shares of 10 and 70 percent (roughly corresponding to the bottom and top
deciles
our sample), the
in
point increase
in
first
OECD growth
column
of
Table 3 suggests that a one percentage
leads to a change
percent, respectively. Second, the cyclical pattern
in
product prices
by OECD-wide shocks and not by domestic shocks. This
and
fourth
columns
of
Table
3,
where we
prices of -1 .2
in their
is
driven primarily
may be seen
relax the restriction X,i=X^=0
and 0.6
the second
In
and
find that
these coefficients are not significantly different from zero. Third, the cyclical behavior
of product prices
is
not driven by the
oil
shocks of the
1
970s.
The
Table 3 indicate that our results are qualitatively unaffected but
absolute value
when we drop
When viewed
last
two columns of
slightly smaller In
the oil-shock years from the sample.
as a whole,
this
evidence
is
consistent with the view that there
a channel of transmission of business cycles that works through changes
product prices.
In particular, individual
positively
employment and output
commodity trade plays a
between product prices
In
commodities, raising wages and
home. Through
this
channel, shocks are
propagated across countries, contributing to the creation of the
International
link
at
in relative
countries find that positive shocks abroad lead
to increases in the prices of labour-intensive
stimulating
is
crucial role in our
in different
argument, since
OECD cycle.
it
creates a
countries.
the remainder of this short paper,
we
develop a stylized model that
formalizes this channel of transmission through relative product prices, and examine
some
of
its limitations.'
interaction
among
highlight this channel, the
countries:
endowments. As a
models
To
result,
we
commodity trade based on differences
a discussion
commodity
trade, factor
reviews of
how shocks
are transmitted across countries
in
of
effective factor
in
movements and/or financial
of these alternative transmission channels,
Mussa (1979) and Backus, Kehoe and Kydland
reader to
in
abstract from other transmission channels that arise
that feature other sorts of
linkages. For
model features a single source
(1995),
who
we
refer the
provide useful
the Mundell-Fleming and
the stochastic growth models, respectively.
A Stylized Model
of Trade
and Fluctuations
We consider a world (named the "OECD") consisting of a continuum of
countries with
and the
the
mass one; two
K-industry;
same
and two
industries producing perishable products, the L-industry
factors of production, labour
measured by an index
allow countries to trade goods, but
assume
factors are immobile. For simplicity,
goods are perishable,
Each country
we
that there
this implies that countries
is
of factor productivity, k.
is
no trade
in financial
have
the
in
We
assets
also rule out capital accumulation. Since
do not save
populated by a continuum of consumers, each endowed with a
unit of capital.
Consumers decide how much
to
to work. Let 9
be the
between
ratio of
capital. All countries
technology, preferences and factor endowments. Countries only differ
state of their business cycle, as
and
and
elasticity of substitution
spending on L- and K-products
-^-^
is
consume and save, and whether or
;
L-
and K-products, so
not
that the
where Pl and Pk are the prices
of L-
IPkJ
and K-products. The parameter
determines how much prices need to change to
convince consumers to vary their relative consumption of L- and K-products. To
produce a
unit of the K-product, e"" units of capital are required.
L-products, e*" (e>0) workers are needed.
associated with productivity fluctuations.
'
See Kraay and Ventura (1998)
for
a
If
To produce one
The parameter e determines
unit of
the factor bias
e<1 the productivity of labour relative to
,
related model.
,
capital is low
when n
is
high and high
when n
is
low.
Consumers
differ in their
wage
opportunity cost of work or reservation wage. Let 1/z denote the reservation
consumer with index
z,
The parameter
{[^>^).
works
with index z
employment
Since the
w = Pl
•
is 1
wage
if
where ze [1 ,~) follows a Pareto
and only
If
w>1/z, where
w Is the wage
and the wage-elasticity
w''"^
e^" the productions of L- and K-products are
productivity shocks.
component,
we
p[J"^
The index n
7i-n.
Each
is
the
sum
of these
of
a and 1-a
respectively, with
index of domestic productivity,
is
A consumer
Therefore, aggregate
e''^ "
is
drift
is
.^
|i-1
i.e.
and e"
i.e.
refer to
a global component,
components
changes
IT,
in
n as
and a country-
an independent Brownian
and instantaneous variances
8>0 and O<0<1. Let the country-specific
components be independent and uniformly
cross-sectional distribution of jr-n
We
a
= 1 - z^~^
of the labour supply is
define y as the income of the country,
motion reflected on the interval [-5, 5] with zero
It
rate.
•
,
productivity.
F(z)
nothing but the value of the L-products a worker can produce,
is
Business cycles arise as n fluctuates randomly.
equal to
i.e.
determines the dispersion of reservation wages.
\i
- F(w"^ ) =
respectively. Finally,
specific
distribution,
of
distributed
on [-6,6] This means
.
time-invariant.^ While
that the
% has been defined as an
n serves as an index of OECD-wide average
straightforward to
show that the instantaneous
domestic shocks, 6n, and OECD-wide shocks, dn,
regulates the extent to which the variation
in
country-specific components,
it
i.e.
whether
is
yfa
."
The parameter a therefore
domestic productivity
comes from dn
between
correlation
is
due
to global or
or d(7r-n). Figure
1
shows
possible sample paths of n under three alternative assumptions regarding a.
^
When the wage
reaches one, the entire population
We assume that all countries are always operating
is
in
working and the labour supply
is vertical.
the elastic section of their labour supply.
This can be ensured through some alternative (and innocuous) parameter restrictions.
^ See Harrison
(1990), Chapter 5.
*
This
is
true except
when
either
rare events since the dates at
n or
n are
These are
the time line.
reflected at their respective boundaries.
which they occur constitute a set
of
measure zero
in
—
be the
Let p
consumer
relative price of L-products in
price index as the numeraire,
we can
_i_
i.e.
Y=
5
d(7t
y—
j
-5
to the
(1)
dlnY-E[dlnY]=
J
^
'
is
shows
in
average
'LJlI}^
productivity,
that increases in
if
and only
proportional to increases
in
that innovations or
.dn
^n
dn. Not
in
surprisingly. Equation (1)
to higher than
average growth
average productivity lead
1
if
shows
and
OECD average
that
Equation
rates.
to increases in the relative price
>ep.. Increases in the production of the K-industry are
the productivity of capital,
i.e.
proportional to n; while
the production of the L-industry are proportional to increases
productivity of workers
in
show
p take this form:^
average productivity lead
of the L-industry
increases
manipulations
OECD share of labour in income. Equations (1) and (2) characterize the
i.e. how the OECD growth rates of output and relative prices, dInY and
dinp, vary with
(2)
—^. A few
2"^
letYbeOECD
the
OECD cycle,
increases
and K-
e+(|x-i)(i-S)
dlnp-Erdlnpl =
where S
respectively. Also,
^-^-^-^-^^Q^^-^)-(^-^)
^
(2)
+ p^"®)«-S
- n)
^^
Y and
growth rates of
shocks
write the prices of the L_!_
products as Pl =(l + p®~^)e-i and p^ =(l
average income,
terms of K-products. Using the ideal
their
number;
i.e.
proportional to e|An.
If
in
'\>e\i,
both the
increases
productivity require that the relative price of L-products increase to
maintain market equilibrium.
From now
on,
we
shall
proceed under the empirically
relevant assumption of 1>e^.
^
To see
this,
equate the
ratios of
OECD expenditure and production of L and
K-products,
i.e.
pi-e
-8
^
'
2-8
This condition implicitly defines p as a function of n, i.e. p=p(n). To obtain Equations (1) and (2)
apply Ito's lemma to Y(n) and p(n), and use the implicit function theorem.
Transmission Through Relative Product Prices
Innovations or shocks to the growth rate of income
in
a country with index n
take this form:
dlny-E[dlny] = [l-s(1-£ ^)]d7i + [s(^-1)(1-S) + s-S]-
(3)
-dn
^^J_^'^^_
where s
is
the country's share of labour
rate reacts to
Equation
shocks
(3) is that,
home and
at
in
income. Equation
abroad. For our purposes, the key result
economy
if
and only
if
s
S
>
.
1
effects of
through changes
two channels
in relative
price of L-products
e
in
is
of transmission of shocks.
product prices. Since
^- dn
.
Both channels work
we have assumed
procyclical. Consequently, positive
—
In addition, positive
This condition
+ (H-1)(1-S)
that 1>eM., the
shocks abroad raise wages
home, stimulating employment and production. This wage
————
shows how the growth
holding constant domestic growth, shocks abroad are positively
transmitted to the domestic
combines the
(3)
effect is
at
measured by
shocks abroad constitute
+ (^-1).(1-S)
favourable (unfavourable)
(K-products),
i.e.
movements
in
s>S
countries for which
the terms of trade for exporters of L-products
(s<S). This terms-of-trade effect
is
measured
(s-S)-(1-e-H)^^e
e+(^-i)(i-s)
How much do these
transmission channels contribute to the creation of the
OECD business cycle? To answer this question, we calculate the correlation
coefficient
that the
®
between domestic and the
OECD average growth
rates,
i.e.
diny and dInY,
model predicts under reasonable parameter values. This calculation requires
A shock at home
has two
industries. This effect
is
effects. First, increased productivity raises production of both
measured by
(s
•
e
+ 1 - s)
djt
and, not surprisingly, coincides with the
Solow residual, i.e. that fraction of the GDP growth rate that cannot be accounted for by
changes in factor inputs. Second, increased productivity raises wages at home and, therefore,
employment and production in the L-industry. This effect is measured by s (ji - 1) e dii Both
•
effects are expansionary.
6
.
us to make assumptions regarding S,
in
the
OECD
0.53 and
is
three values for
countries
in
s,
Table
its
e
standard deviation
s={0.49,0.53,0.57}.
and found
1
s, a, n,
and
9.
Since the average labour share
0.04,
is
we
set
S=0.53 and consider
We have calculated Solow residuals for all the
that the cross-country correlation of
0.32. Since the theory predicts this correlation to
be equal toVo
,
Solow residuals
we
is
set a=0.1024.
Using these assumptions, Table 4 shows the correlation coefficients that arise for
values of
different
0.32, the
e
and
)x.
Given the observed correlation
model can explain the observed correlation
this requires
(e=0);
6,
and
us to assume a strong factor bias
either
a less than
in
of productivity
of growth rates of 0.47.
seem
to
unit elasticity of substitution
generate cross-country correlations
those observed
in
among commodities
in
for
all
of the
it
is
to believe that this particular
observed correlations
To understand
cycle,
no reason
in
one adopts the
income and
it
is
the role that commodity trade plays
easy
to
relative prices in
see
benchmark case
that innovations or
(5)
dlnp-E[dlnp] =
^^
—/
e
^'^
of autarky.
to the
If
OECD
countries are not
growth rates of
this form:
dn
+ (^i-1).(1-s)
The reader will immediately
(2),
channel should account
the creation of the
a country with index n would take
= ^:^:ii:i±^^:tiLLlHIzl).dK
dlny-E[dlny]
^J
^
^
e+(^-i).(i-s)
^
in
shocks
(4)
^
in relative
outputs.
useful to briefly discuss the
allowed to trade,
(9=0.5)
emphasized
extreme position that business cycles are on/y transmitted through changes
is
However,
growth rates that are smaller than
the data. But this should not be surprising. Unless
product prices, there
of
the nature of productivity shocks
or a high elasticity of the labour supply (|i-1=1). Otherwise, the effects
here
shocks
notice that these are nothing but Equations
which now apply to every country and not only to
OECD averages.
In
(1 )
and
a world of
autarky, relative product prices are determined by the domestic relative production of
L-
and K-products and therefore depend only on domestic shocks. Also,
in
such a
world there are no channels of transmission, and the cross-country correlation of
income growth rates
exactly equal to those of shocks,
is
contrast with a world of
commodity trade
OECD-wide shocks and
appear
to
be
at
look at the data
odds with the
based on commodity
in
product prices
is
while cross-country
trade.
.
This stands
prices
in relative
in
depend on
GDP
shows
(3), respectively).
that the predictions of the autarky
facts, further motivating
model
our analysis of transmission
We have already seen in Table 3 that the cyclical pattern
driven by
residual correlations
which changes
^fa
the cross-country correlations of income growth rates exceed
those of shocks (See Equations (2) and
A simple
in
i.e.
OECD-wide shocks and
not domestic ones. Moreover,
growth correlations average 0.47, cross-country Solow
average only 0.32. Figure
2,
which plots the correlations of
GDP
growth (on the Y-axis) against those of Solow residuals (on the X-axis), indicates that
this
ranking of correlations
is
not an artifact of averaging across countries.
results are quite striking for their clarity:
45%
line,
with
GDP
of
in
24 countries are located above the
interpreting this evidence
positive transmission of shocks. First,
of productivity
shocks
it
is
is
well
as a hard "proof that there
known
that interpreting
Solow residuals
is
Solow residuals
quite problematic. Second, there are surely
other shocks that drive the business cycle and these could be
countries than
the
growth correlations exceeding those of Solow residuals.^ Clearly,
one should be cautious
as a measure
22 out
In fact,
are. Nevertheless,
we
find this
more
correlated across
evidence suggestive and
consistent with the arguments presented here.
To sum
up, our
argument goes as
OECD countries lead to
increases
in
follows: Favourable
shocks
in
one or more
the relative prices of labour-intensive products
and, therefore, raise wages, employment and output
in
the rest of the
presented evidence showing that periods of high growth
in
the
OECD.
First,
we
OECD are associated
with increases in the relative prices of labour-in'.ensive products
in all
countries.
by no means a new observation. See Backus, Kehoe and Kydland (1995) who argue
GDP growth rates are more correlated across countries than Solow
residuals is one of the main puzzles that arise when one attempts to interpret the data from the
perspective of the stochastic growth model.
^
This
is
that the finding that
8
we developed a
Second,
stylized
model
that illustrates
how
this cyclical
behavior of
product prices creates a channel of positive transmission of shocks that might be one
of the (possibly
many)
factors behind the
Going beyond the
details of the
should be apparent that the key ingredients of our argument are:
model,
it
booms
labour-intensive products
for labour increases in
increases
would
OECD cycle.
in
the
all
demand
become
countries; (2)
for labour.
relatively
scarce and therefore the
Wages and employment respond
Any model
that
(1)
During
demand
positively to
combines these two features
rationalize our claim that the cyclical behavior of product prices provides
important clue on
how shocks
are transmitted across
an
OECD countries.
References
Backus, David, Patrick Kehoe and Finn Kydland (1995), "International Business
in T.F.CooIey (ed.) Frontiers of Business Cycle
Research, Princeton University Press.
Cycles: Theory and Evidence"
Driscoll,
Spatially
John and Aart Kraay (1998), "Consistent Covariance Matrix Estimation with
Dependent Data," The Review of Economic and Statistics, Vol LXXX, No. 4.
pp. 549-560.
Harrison, Michael M. (1990), Brownian Motion
Kraay, Aart and
Jaume Ventura
section of Business Cycles,"
and Stochastic Flow Systems,
(1998), "Comparative
Krieger.
Advantage and the Cross-
MIT Working Paper.
Mussa, Michael (1979), "Macroeconomic Interdependence and the Exchange Rate
Regime" in R.Dornbusch and J.A. Frenkel (ed.) International Economic Policy, MIT
Press.
Nehru, Vikram and Ashok Dhareshwar (1993). "A New Database on Physical Capital
Stock: Sources, Methodology and Results". Revista de Analisis Economico. Vol. 8,
No. 1,pp. 37-59.
9
The OECD
Table 1
Cycle: Output Fluctuations
Correlation of Real Per Capita
With
OECD Average
Full
Sample
GDP Growth
Excluding that Country
Non-Oil Shock Sample
(1960-72, 1982-96)
(1960-96)
G7
Canada
0.69
France
0.74
Germany
0.66
Italy
0.54
Japan
0.50
United Kingdom
United States
Rest of
0.61
0.50
***
***
***
***
***
***
***
0.79
0.67
0.53
0.54
0.47
0.39
0.36
***
***
***
**
**
*
OECD
Australia
0.52
Austria
0.53
Belgium
0.62
Denmark
0.58
Finland
0.41
Greece
0.58
Iceland
0.15
Ireland
0.31
Luxembourg
0.29
Netherlands
0.67
New Zealand
0.29
Nonvay
0.31
Portugal
0.39
0.54
Sweden
***
***
***
***
**
***
0.50
0.56
0.54
0.66
0.42
0.55
0.57
*
***
*
**
***
***
***
0.66
***
***
0.27
0.37
**
0.18
0.61
0.73
Turkey
-0.06
0.02
0.61
0.54
0.42
0.41
0.47
0.46
OECD Average
**
***
0.07
0.54
OECD Average
•**
0.25
Switzerland
G7 Average
***
***
0.20
*
*
Spain
Rest of
***
0.66
***
•**
***
This table reports the correlation of annual real per capita GDP growth with OECD average real per capita GDP
growth excluding that country, for the indicated periods. Annual data on real per capita GDP at PPP and population
are taken from Penn World Table, Version 5.6 (codes RGDPCH and POP), and are extended through 1996 using
World Bank data on real per capita GDP at PPP and population (codes NYGDPCAPPPKD87 and SPPOPTOTL). *
(**) (***) indicate significance at the 10% (5%) (1%) level respectively.
10
The OECD
Table 2
Cycle: Labour Market Fluctuations
Correlations with
OECD Average
of
Real Per Capita
Growth
Manufacturing
Manufacturing
Employment
Hours
Full
GDP Growth
in:
Manufacturing
Real
Wages
Sample (1960-96)
G7
Canada
0.61
France
0.44
Germany
0.40
***
0.52
"
0.46
*
Italy
0.31
Japan
0.39
United Kingdom
"*
0.58
"*
0.49
United States
0.61
*"
0.40
"
0.53
0.61
0.42
***
***
***
*•
***
***
***
0.16
0.07
0.41
*
0.14
0.33
*
-0.11
0.31
*
Other
0.49
Denmark
*"
"*
0.57
Netherlands
0.33
0.24
0.44
Nonway
0.17
0.04
0.27
Sweden
0.37
0.24
0.26
G7 Average
G12 Average
0.46
0.51
0.19
0.43
0.41
0.24
Belgium
0.44
0.39
"
"
***
0.32*
*•
0.23
*
*
Non-Oil Shock Sample (1960-1972, 1982-1996)
07
*"
Canada
0.71
France
0.44
Germany
0.32*
Italy
0.41
0.30*
0.63
United States
0.38
0.49
***
0.38*
***
0.24
0.33*
**
Japan
United Kingdom
0.66
"
0.39
0.40
***
0.57
**
**
**
***
0.46
*
0.26
0.41
*
0.09
0.38*
0.24
Other
Belgium
0.49
Denmark
0.41
***
0.38
**
**
**
0.59*
0.36
0.24
0.53*
0.45*
Netherlands
0.41
Nonway
0.26
Sweden
0.59
07 Average
0.46
0.44
0.32
012 Average
0.45
0.37
0.40
0.16
***
*
0.25
0.37
**
0.61
*
This table reports the correlation of annual growth rates in each variable with OECD average real per capita GDP
growth, excluding that country. Annual indices of hours worked, employment and wages in manufacturing are from
the United States Bureau of Labour Statistics, International Labour Statistics (codes 0003, 0006, 0022, and 0024,
available at ftp://ftp.bls.gov/pub/time.series/in/) Real wage growth is obtained by deflating by the consumer price
index as reported by the IMF International Financial Statistics (code 64...zf). * (**) (***) indicate significance at the
10% (5%) (1%)
level respectively.
11
.
Table 3
Labour Intensity and Price Cyclicality
(Dependent Variable is Growth in Price Index of
Industry in Country c at Time t)
i
Full
Sample
(1960-96)
OECD Growth
OECD Growth
Labour
x
Intensity
-1.462
-2.009
-0.752
(0.621)**
(0.743)***
(0.440)*
(0.775)"
2.965
3.120
(0.833)***
(0.937)***
1.916
(0.670)"*
(1.003)"
Domestic Growth
Domestic Growth x
Labour Intensity
Number of Observations
Non-Oil Shock Sample
(1960-72, 1982-96)
16418
-1
.363
2.128
0.478
0.501
(0.446)
(0.588)
0.207
0.143
(0.617)
(0.727)
16418
11494
11494
Standard errors are corrected for heteroskedasticlty and serial and cross-sectional dependence using the
procedure suggested by Driscoll and Kraay (1998). * (") (***) indicates significance at the 10% (5%) (1%)
level. Country-industry fixed effects are removed by taking deviations from country-industry means. The
price and labour-intensity data cover 28 three-digit ISIC manufacturing sectors in 20 OECD countries for
all available years over the period 1963-1995 (Iceland, Luxembourg, Switzerland and Turkey are excluded
due to inadequate data coverage), and are drawn from the United Nations Industrial Development
Yearbook. The dependent variable is the growth rate of the implicit gross output deflator (constructed as
the growth in nominal local currency gross output less growth in the industrial production index) relative to
growth in the consumer price index. Labour intensity is measured as the share of employee
compensation in value added. OECD and domestic growth are constnjcted as in Table 1
12
Table 4
Theoretical Growth Correlations
s=0.49
9=1.0
0=0.5
9=0.1
£=0.25
6=0.5
S=0.53
0=0.1024
s=0.53 3=0.57
s=0.49
E=0
S=0.53 S=0.57
S=0.49
s=0.53 s=0.57
[1=1.0
0.31
0.32
0.33
0.31
0.32
0.33
0.30
0.32
0.35
H=1.5
H=2.0
0.32
0.33
0.33
0.34
0.35
0.36
0.35
0.38
0.41
0.32
0.32
0.32
0.34
0.35
0.36
0..39
0.42
0.45
H=1.0
0.30
0.32
0.34
0.29
0.32
0.35
0.27
0.32
0.37
[1=1.5
0.33
0.33
0.34
0.35
0.36
0.38
0.38
0.42
0.46
H=2.0
0.32
0.32
0.32
0.36
0.37
0.38
0.42
0.46
0.51
[1=1.0
0.24
0.32
0.40
0.17
0.32
0.46
0.07
0.32
0.55
[1=1.5
0.34
0.35
0.36
0.38
0.41
0.45
0.44
0.52
0.60
[1=2.0
0.32
0.32
0.32
0.38
0.40
0.42
0.49
0.55
0.61
This table reports the theoretical correlation coefficient between domestic and
OECD average growth,
1
dlnY, which Isp (r-\
is;
"
1
a
-s
•
(1
-
e •^i)
1-s (1-e[i) + [s-(S ^.•(i-s))-s].^^J_
parameter values.
13
e-[i
1)-(1-S)J
2
diny and
Figure 1
Sample Paths of the Productivity Index
Country-Specific Variation Only
(o=Q)
n+s
Time
n-5
Global Variation Only
(a =
1)
n+5
Time
n-5
Both Country-Specific and Global Variation
(0<a<1)
Time
14
.
.
Figure 2
GDP Growth and Solow Residual
Correlations
(1960-1996)
1.00-
0.80
X
-
>/
FRA
CAN
1o
O
Q.
Q
0.60
GBR
-
v''^
y^
BEL
GRC
DNK
45°
y^
.USy^
'^E
0.40
CD
(0
*•>
q.
(0
0.20-
o
ISL
V^
Q.
1
n r\r\
u.uu
>
yo. 30
-0.20
-0.20
0.40
'eC20
0.60
0.80
1.00
-
Solow Residuals
This figure plots the correlation of annual real per capita GDP growth with OECD average growth excluding that
country (on the vertical axis) against the correlation of annual Solow residuals with OECD average Solow residuals
excluding that country (on the horizontal axis). Real per capita GDP growth correlations are drawn from Table 1
Solow residuals are constructed as the growth in total real GDP, less the economy-wide share of wages in GDP
in total civilian employment, less one minus the share of wages in GDP times growth in the total
capital stock. The share of wages in GDP is constnjcted using current price local currency compensation of
employees and GDP by expenditure components taken from the OECD National Accounts (codes MOCOM and
MOGDPE). Total civilian employment is drawn from the OECD Labour Force Statistics. The total capital stock is in
1987 constant local currency units and is drawn from Nehru and Dhare^hwar (1993) and is updated to 1995 using
Worid Bank data. The OECD average Solow residual is constructed as a population-weighted average of country
Solow residuals. Due to missing data on employment, Solow residual correlations are computed over the period
1960-1995. 22 out of 24 countries lie above the 45-degree line, and the null hypothesis that the growth rate
correlations are equal to the Solow residual correlations is easily rejected. The p-value associated with a simple
times growth
sign test
is
less than 0.01
15
707
Date Dub
Lib-S6-67
^lJ
MIT LIBRARIES
3 9080 01972 0736
.
«
fe
**
*
pi
il^
^-
5f
%
<§i
W ^ ^
-^
m W
^'f'
P'^S'
"?*
f7'ir^"'
;!«
*, p'
^t:
f P
^f^'
WWff9.w..P....f^^:^J::.
3:
f'W
?^
f W H ^
f ?%
yf|-
-
.
-:
^'*
p" %'
jl'
:.*'•;».**'
t
'.'
.
S^'-'#
'f.'*^^^
'^If "vf
%' %'
'
'^?-P"
'$"'-^:"''y- s|
^' *"!?'"
ii''^^!''
:l
s|.| 5':#..Vye," j~'
##
'"-f|'' ^?|.
'':^^
!«"
f^
'"f
?;H
^.
4s 41
'%"**.
o^'^'
r*
1/;!,
t
If
i^
"^''
'.«'
I"
i'
l'i.
'fl'\|i'>'>'\rif>'^^fi'^~if
I
w
^-
i'^''
tiii'"
if
i;r
m'"
'
•.. ;;?
f:^/'-'^
ti''>
OREGOMRULECO.
^
-
11'
p
U
'
V
ft
fr
'
^;ii;;-#"'
r|.""pi
?
/
>
:;,;
#1
-..lij
%f"^;" -
tW^'l"'
:j,,
>'
..,„
^f?«">'
&
H" ^ S:'- !^
f/ ft"' $/' ly,"
|j
'
k--"
If'
i^'
:-"^ji''
im'4*'
^f
f|-
vif''
t*^'
I!f'"
>
p' if lg"%
%;
^:
if p' p'
'!%'
«
i;
p"
%"
;:^."'
r
;%'
>i,;
£?"'^i
*.-'
W-
»
¥
"ir
y^
IV
i'f