Transcript Integration among G7 Equity Market : Evidence from iShares

Balassa-Samuelson: Still haven’t found what we’re
looking for?
Trond-Arne Borgersen and Roswitha M. King
Østfold University College, Halden, Norway
FIW – FORSCHUNGSKONFERENZ ‘INTERNATIONAL ECONOMICS‘
WIRTSCHAFTSUNIVERSITÄT WIEN 11. 12. 2009
1
Balassa-Samuelson (B-S) effect: empirically weak

Mihaljek and Klau (2008,2004)

Also small effects reported by:


B-S explains around 16% of overall CPI
inflation in selected east European
countries.
B-S explains on average between 0.2%
and 2% of annual inflation differentials
between east European countries and the
euro area.
Egert (2005), Egert (2002a), Egert (2002b),
Egert et al (2003)
Policy Relevance





Breuss (2003)
Cipriano (2001)
Coricelli & Jazbec (2001)
Flek et al (2002)
Kovacs (2002)
2
Official stance of the European Commission and ECB

“…notwithstanding its importance, the Balassa-Samuelson effect
should not be overstated when explaining current inflation rates in
the accession countries…. The Maastricht inflation criterion…. will
not be revised to take into account any possible Balassa-Samuelson
effect…”

(Christian Noyer, 2001)
3
Why have we not found what we are looking for?

Is there nothing to find?
or

Have we been looking in the wrong places?
4
Three limiting (theory-based) features of B-S

First, mixing of supply-side and demand-side channels ( structural
inflation relates to the supply side so analysis should be anchored
there)

Second, B-S treats as independent what may be considered strongly
related elements: the productivity growth differential between
sectors and the sector’s relative magnitude.

Third, structural changes evolve over the longer term. However, by
construction, B-S has relatively short-term orientation.
5
First, mixing of supply-side and demand-side channels (B-S is about
structural inflation and that relates to the supply side)

Typically B-S computes sectoral shares in CPI rather than in GDP.

Mihaljek and Klau (2008): market based non-tradables account for
only 20%-30% in the Baltic States’ CPI basket, while accounting for
more than 60% of value added in GDP.
using GDP based measures increases B-S
and anchors it firmly on the supply side –
where structural shifts reside
6
Second, B-S treats as independent what in reality are strongly related
elements: the productivity growth differential between sectors and the
sector’s relative magnitude.

B-S uses two key elements
Relative sector size
 Productivity growth differences between sectors
But does not exploit relations between them.

Our alternative approach:
Defining a functional relation between
them.
7
Endogenizing the sector size:
function of productivity growth difference
'

 t   qT  where  (qT )  0
8
How productivity growth affects structural inflation:
Productivity Growth Pass-Through (sector size exogenous/endogenous)
PSTR  (1   )qT  q NT 
dPSTR
 (1   )  0
dqT

dPSTR 
T  (qT ) ' (qT )q
T 
 (1   )q

 



dqT
PE
SE


9
Third, structural changes evolve over the longer term.
However, by construction, B-S has relatively short-term
orientation.

Long-run supply side transition dynamics come in two layers,
which in the following is referred to as reallocation and
restructuring .

Claim: Conventional B-S effect captures the former, but not
the latter.
10
The relationship between the size of the tradable sector
and its productivity measures - reallocation
qT
qT
0
0
0
0
qT
0
0
0
0
t
11
Long-term transition, the size of the tradable sector and tradable
sector productivity - restructuring
qT
qT
0
0
0
0
T
q
0
0
0
0
t
12
The Scandinavian Model of Inflation
Expressions for Inflation and Structural Inflation
p

W  e  (1   )qT  q NT   p
W  e  P
P
STR
PSTR  (1   )qT  q NT 
13
From the literature: 2 variants of defining structural inflation
When following the B-S effect over a number of years the literature provides two variants:
Constant relative size of the non-tradable sector and time-indexed productivity growth
differentials
T
(1−α)
 q
T
 q NT t
t 1
alternatively, time-indexed productivity growth differentials and time-indexed relative sizes
of the non-tradable sector
 1    qT  q NT t
T
t 1
t
Both expressions are interpreted as structural inflation over a period of time
14
Our definition of (cumulative) structural inflation
T
T
 1     q
t 1
t
t 1
T
 q NT
t
15
q T  q NT
T
T
 1     q
t 1
(1-α)t( q  q )t
T
NT
(1-α)t( ) t+1
t
t 1
T
(1-α)t( ) t+2
(1- α)t+1( )t (1-α)t+1( )t+1 (1-α)t+1( )t+2
(1- α)t+2( )t (1-α)t+2( )t+1 (1-α)t+2( )t+2
(1- α)t+3( )t (1-α)t+3( )t+1 (1-α)t+3( )t+2
.
 NT
q
t
.
.
(1-α)t( ) T-1
.
.
.
.
(1-α)t+1( )T-1 (1-α)t+1( )T
(1-α)t+2( )T-1 (1-α)t+2( )T
(1-α)t+3( )T-1 (1-α)t+3( )T
.
.
(1-α)t( ) T
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
(1- α)T-1( )t (1-α)T-1( )t+1
(1- α)T( )t (1-α)T( )t+1
(1-α)T-1( )t+2
(1-α)T( )t+2
(1-α)T-1( )T-1
(1-α)T( )T-1
(1-α)T-1( )T
(1-α)T( )T
16
Sector size elasticity with respect to productivity growth. Conditions for when increased
productivity differential translates into structural inflation depends on the relation between sector size elasticity
and the prevailing distribution of value added between secors.

dP
STR
0
dq T
  

El qT  

1 
  
El qT  

1 
  ' (qT )

El qT  
qT 
 

qT
 0
qT
 0
17
Each of the curves represents a given relationship between value
added in the tradable and the non-tradable sector.

*
i
 i
 
1 i



18
When the cumulative productivity growth differential makes structural
inflation exceed a threshold (T), the distribution of value added
changes...........the curve shifts.
 *
 t 1

*
t  
ˆ *
t


iff
iff

dPSTR
T
C
dq T


dPSTR


T
C

dq T

19
Condition for long-term transition. There is a critical value for the
growth rate of structural inflation that can bring about transition.

dP
STR
T
C
dqT
  

El qT  
 T
1  
20
Interpretation of Threshold T


T   1   t  qT  q NT t    (1   ) t (qT  q NT ) t
t 1
 t 1
 t 1
T
T
T
21
Our approach departs from standard B-S analysis in the following ways:

We endogenize the relative sector sizes by defining them as a function of the
productivity growth differentials.

We introduce leads and lags into the relation between productivity growth
differentials and relative sector sizes, thereby capturing not only instantaneous
effects, but also the effects of larger longer-term shifts in the structure of an
economy.

We distinguish between short-term dynamics motivated by profit seeking
economic agents under a given interval of sectoral composition, and long-term
dynamics representing shifts in the prevailing interval of sectoral composition. We
link these two dynamics.

We define transition as shifts of relative sector sizes in favor of the non-tradable
sector
22