Transcript The correlation of Demand and Supply shocks – Evidence

Academy of Economic Studies
Doctoral School of Finance and Banking
Demand and supply shocks synchronization – Evidence
from Romania in the context of European Integration
MSc Student: Nora Rusu
Supervisor: Professor PhD. Moisă Altăr
Topics of the paper

Optimal currency area and business cycle
correlation approach
 Brief literature review
 Objectives of the paper
 Theoretical considerations and shock
identification
 Data analysis
 Empirical estimation (Structural VAR)
 Results
 Conclusions
The importance of studying business cycle correlation in
the context of optimal currency area theory

Optimal currency area theory (since 1961 Mundell)


Formation of the European Union



“If nations are dissimilar to regions, fixed exchange rates may do as
well as flexible exchange rates”
Is at least the “old” Europe an optimum currency area?
Will it be costly for the economies to adopt a single currency
Formation of the Eurozone


Can it receive new members?
How can one determine if a country is ready or not for adoption
 Business cycle correlation → Shock correlation (one
approach)
 Benefits: reduction in transactions costs and stronger integration
of markets
The importance of studying business cycle correlation in
the context of optimal currency area theory

•
•
Costs:
 Giving up the flexible exchange rate
 Giving up the independence of monetary policy
If asymmetric shocks occur
If responses to shocks are different


Sole instrument: fiscal policy (limited by Maastricht criteria)
CEECs once they join EU they have to join EMU


Are they ready for EMU adherence?
Is Romania ready for EMU adherence?
Brief literature review – Optimal currency area theory
Mundell (1961) - the argument for flexible exchange rates rests on the
closeness with which countries correspond to regions. If a nation is an
economic region with internal factor mobility and external factor immobility,
the argument for flexible exchange rates holds.
Bayoumi and Eichengreen (1992) when they used data from 11 European
Union member countries to extract information on underlying aggregate
supply and demand disturbances using VAR decomposition
CEECs: - topical subject; they are expected to join EMU
Fidrmuc and Korhonen (2001), Horvath (2000), Frenkel and Nickel (2002),
Babetski, Boone and Maurel (2003), Horvath and Ratfai (2004), Fridmuc
(2001), Frankel and Rose (1998)
 The correlation in shocks has a high degree of dispersion and differ from
correlations in EMU; still some strong correlations are shown by some
countries (Hungary
Objectives of the paper






Identify aggregate supply and demand shocks for the countries
included in the study
Study the response of the economy (real GDP and GDP deflator)
to a supply or demand shocks
Study the correlations between responses
Study the correlations in shocks between the considered
countries
Time varying correlations
Shock importance (Error forecast variance decomposition)
Shock identification
Methodology: Blanchard and Quah (1989), Bayoumi (1991) and Bayoumi
and Eichengreen (1992)
Blanchard and Quah (1989)
They interpret fluctuations in real GNP and unemployment as due to two
types of disturbances: disturbances that have a permanent effect on
output and disturbances that do not. The first is interpreted as supply
disturbances and the second as demand disturbance.
Bayoumi and Eichengreen (1991)
They examine time series behavior of real GDP and the price level. To
identify the structural shocks they impose the restriction that aggregate
demand disturbances have only a temporary effect on output but a
permanent impact on prices while aggregate supply disturbances
permanently affect both output and prices.

Bivariate SVAR (real GDP growth and variation in prices)
Theoretical considerations
The Aggregate Demand and Supply Model (The Model)
Theoretical considerations
AD → AD’ + SRAS
•
Equilibrium E → D’
•
Temporary increase in Output (Y’)
•
Increase in Prices (P’)
Supply curve becomes vertical LRAS
•
Equilibrium D’ → D’’
•
Output returns to its initial level (Y)
•
Permanent increase in Prices (P’’)
Positive demand shock:
 Temporary
positive effect on
Output; Long run zero effect
 Permanent
positive effect on
Prices
Theoretical considerations
Technology shock raises long run
potential level of output → both SRAS
and LRAS move rightwards to SRAS’
and LRAS’
Short-run equilibrium S’
•
Increase in Output (Y’)
•
Decrease in Prices (P’)
Supply curve becomes vertical LRAS’
•
Equilibrium S’ → S’’
•
Output increases further (Y”)
•
Prices decline further (P’’)
Positive supply shock:
 Permanent positive effect on Output
 Permanent decline in Prices
Shock identification
The two variables that compose the VAR:
BX t  0  1 X t 1  ...  p X t  p   t
 y t 
Xt   
pt 
εt is a vector of the two structural (demand and supply) errors.
Assuming that B is invertible, that is (1  b12 b21  0)
X t  B 10  B 11 X t 1  ...  B 1p X t  p  B 1 t
X t  A( L) LX t  et (1)
The bivariate moving average representation of VAR:

b
 y t 
i  11i

L
p   b
 t  i 0  21i
b12 i   dt 
b22 i   st 
(2)
Shock identification
Using (1) we can say that e1t is the one-step forecast error of Δyt. From the
BMA representation in (2) we can further obtain that:

e1t  b11( 0 )  dt  b12 ( 0 )  st


e2t  b21( 0 )  dt  b22 ( 0 )  st
or
 e1t  b11( 0 )
e   b
 2t   21( 0 )
b12 ( 0 )   dt 
b22 ( 0 )   st 
If the b coefficients were known, it would be possible to recover and
from the residuals e1 and e2. We need four additional restrictions. We
can use the residuals e1 and e2 to construct the covariance matrix so we
would know var(e1), var(e2) and cov(e1,e2) .
Shock identification - Restrictions
Restriction 1:
var(e1t )  var(b11(0) dt  b12(0) st )
Knowing that E(εdt, εst) = 0 since the two disturbances are uncorrelated and
assuming at the same time that the two disturbances have unit variance, we
obtain restriction no 1:
(3)
var(e1 )  b112 (0)  b122 (0)
Restriction 2:
In the same manner we obtain restriction no 2:
var(e2 )  b212 (0)  b222 (0)
(4)
Shock identification - Restrictions
Restriction 3
e1t e2t  [b11(0) dt  b12(0) st ][b21(0) dt  b22(0) st ]
Assuming once more that the structural disturbances are not correlated
and that they have unit variance we obtain restriction no 3:
Ee1t e2t  b11(0) b21(0)  b12(0) b22(0)
(5)
Restriction 4
For all possible realizations of the sequence, demand shocks will have
only temporary effects on the sequence if:




1   a22 (k )b11( 0)   a12 (k )b 21( 0)  0
k 0
 k 0

(6)
Data analysis
Countries included in the analysis:
 Romania
 Core economies of the Euro Area: Germany, France and Italy
 Slovakia
 Poland
 Hungary
Variables:
Nominal GDP (SA and NSA) and Real GDP (Eurostat, IFS (IMF))
Inflation: GDP Deflator = (Nominal GDP) / (Real GDP) * 100
Sample:
1998Q1 : 2008Q1
Initial data managing:
Eliminating the seasonal effects using Demetra (TRAMO SEATS) –
alternative: seasonal dummies – lost in degrees of freedom
Data analysis
Growth
Country
Mean
Standard
Deviation
Prices
Mean
Standard
Deviation
EA
0.006001
0.004968
0.004638
0.001430
Germany
0.003855
0.005262
0.002017
0.002060
France
0.005541
0.003789
0.003818
0.003459
Italy
0.003283
0.004395
0.005888
0.001365
Romania
0.010964
0.007787
0.010964
0.007787
Hungary
0.009352
0.003578
0.015097
0.012894
Poland
0.010266
0.007400
0.009495
0.030153
Slovakia
0.012120
0.008560
0.014607
0.022207
Data analysis – Initial correlation
0.8
Germany
Correlations of Growth
0.7
Italy
0.6
0.5
France
0.4
0.3
0.2
Poland
0.1
0.0 Hungary
-0.1
-0.2
-0.1
Slovakia
0.0
0.1
Romania
0.2
0.3
0.4
Correlations of Prices
0.5
0.6
0.7
0.8
Empirical estimation SVAR
Estimate 8 bivariate SVAR
realGDP (SA) and GDPdeflator (SA)
1) Testing for Unit Root – all variables are I(1) → first differences in
realGDP and GDPdeflator
d(realGDP) – real growth
d(GDPdeflator) – inflation
2) Optimal number of lags – Sequential LR, Akaike, Schwartz,
Hannan-Quinn
Optimal LAG length
Country\Criteria
Sequential LR
AIC
SC
HQ
Chosen
Euro Area
1
1
1
1
1
Germany
1
1
1
1
1
France
2
2
2
2
2
Italy
3
3
1
1
3
Romania
4
4
5
5
4
Slovakia
3
4
3
4
3
Poland
2
3
1
3
3
Hungary
1
2
1
1
2
Empirical estimation SVAR
3) VAR stability condition – the absolute values of the eigenvalues of
the matrix lie inside the unit circle
4) Residual tests:



Autocorrelation (LM Autocorrelation test)
Normality (Jarque-Berra test)
White Heteroskedasticity test
5) Granger Causality test
Impose the STRUCTURAL restriction that the aggregate demand
shock does not have a permanent effect on output → Structural
aggreagate demand and supply shocks
Results 1- BQ restriction and overidentifying restrictions
Response of Output to Demand Shock
Accumulated Response of DL_GDP_EA to Structural
One S.D. Shock2
.0030
Accumulated Response of DL_GDP_GE1 to Structural
One S.D. Shock2
.0020
.0028
.0016
.0026
.0024
.0012
.0022
.0008
.0020
.0018
.0004
.0016
.0014
.0000
1
2
3
4
5
6
7
8
9
10
1
Accumulated Response of DL_GDP_PL to Structural
One S.D. Shock2
2
3
4
5
6
7
8
9
10
Accumulated Response of D(LOG(GDP_RO)) to Structural
One S.D. Shock2
.0020
.0024
.0016
.0020
.0016
.0012
.0012
.0008
.0008
.0004
.0004
.0000
.0000
-.0004
-.0004
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Results 1- BQ restriction and overidentifying restrictions
Response of Output to Supply Shock
Accumulated Response of DL_GDP_EA to Structural
One S.D. Shock1
.006
Accumulated Response of DL_GDP_GE1 to Structural
One S.D. Shock1
.009
.008
.005
.007
.004
.006
.005
.003
.004
.002
.003
1
2
3
4
5
6
7
8
9
10
1
Accumulated Response of DL_GDP_PL to Structural
One S.D. Shock1
.013
2
3
4
5
6
7
8
9
10
Accumulated Response of D(LOG(GDP_RO)) to Structural
One S.D. Shock1
.014
.012
.012
.011
.010
.010
.009
.008
.008
.006
.007
.004
.006
.005
.002
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Results 1- BQ restriction and overidentifying restrictions
Response of Prices to Demand Shock
Accum ulated Res pons e of DL_DEFL_EA to Structural
One S.D. Shock2
.0020
Accum ulated Res pons e of DL_DEFL_GE1 to Structural
One S.D. Shock2
.0036
.0034
.0016
.0032
.0030
.0012
.0028
.0008
.0026
.0024
.0004
.0022
.0000
.0020
1
2
3
4
5
6
7
8
9
10
Accum ulated Res pons e of DL_DEFL_PL to Structural
One S.D. Shock2
.07
1
2
3
4
5
6
7
8
9
10
Accum ulated Res pons e of D(LOG(DEFL_RO)) to Structural
One S.D. Shock2
.014
.013
.06
.012
.05
.011
.04
.010
.009
.03
.008
.02
.007
.01
.006
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Results 1- BQ restriction and overidentifying restrictions
Response of Prices to Supply Shock
Accumulated Response of DL_DEFL_EA to Structural
One S.D. Shock1
.0000
Accumulated Response of DL_DEFL_GE1 to Structural
One S.D. Shock1
-.0018
-.0020
-.0004
-.0022
-.0008
-.0024
-.0012
-.0026
-.0016
-.0028
-.0020
-.0030
1
2
3
4
5
6
7
8
9
10
Accumulated Response of DL_DEFL_PL to Structural
One S.D. Shock1
1
3
4
5
6
7
8
9
10
Accumulated Response of D(LOG(DEFL_RO)) to Structural
One S.D. Shock1
.000
.00
-.002
-.01
-.004
2
-.02
-.006
-.03
-.008
-.04
-.010
-.05
-.012
-.014
-.06
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Result 2
Demand shock:
 Temporary and positive effect on output (BQ restriction)
 Permanent and positive effect on prices
Supply shock:
 Permanent and positive effect on output
 Permanent and negative effect on prices
In most of the cases (as in Frenkel, Nickel and Schmidt (1999)):
 The supply shocks seem to be more important then the demand
shocks for output response even in the short run
Size: The response of output to a supply shock in EA is almost half of
the magnitude of the similar reaction at the same type of shock for
the Romanian economy.
Result 3
Response of output to a positive demand shock
Speed of adjustment:

Demand shock to output: very
quick absorption in EA (1-3
quarters)
 Demand shock to prices
stabilizes in 4-5 quarters (EA)
 Supply shock to output: 7-8
quarters in EA, 5-6 in CEECs
 Supply shock to prices
stabilizes quicker in only 3-4
quarters (EA)
In general: in CEEC’s it takes
longer to absorb the shocks
and the effect is volatile
Response of DL_GDP_EA to Structural
One S.D. Shock2
Response of DL_GDP_GE1 to Structural
One S.D. Shock2
.0024
.0020
.0020
.0015
.0016
.0010
.0012
.0005
.0008
.0000
.0004
-.0005
.0000
-.0004
-.0010
1
2
3
4
5
6
7
8
9
10
1
Response of D(LOG(GDP_FR)) to Structural
One S.D. Shock2
2
3
4
5
6
7
8
9
10
Response of D(LOG(GDP_RO)) to Structural
One S.D. Shock2
.0030
.0024
.0020
.0025
.0016
.0020
.0012
.0015
.0008
.0010
.0004
.0000
.0005
-.0004
.0000
-.0008
-.0005
-.0012
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Result 4 Correlation of aggregate supply shocks
EA
EA
GE
FR
IT
RO
PL
SK
HU
1
Germany
0.71
France
0.53
0.36
Italy
0.47
0.28
0.25
Romania
0.15
0
0.27
0.08
Poland
0.19
0.27
0.15
0.14
0.13
1
Slovakia
0.13
0.07
-0.10
0.13
-0.06
-0.05 1
Hungary
0.30
0.12
0.14
0.31
0.17
0.19
1
1
1
1
-0.30
1
Result 4 Correlation of aggregate demand shocks
EA
GE
FR
IT
RO
PL
SK
EA
1
Germany
0.14
1
France
0.02
0.07
1
Italy
0.58
0.34
0.16
1
Romania
0.00
-0.06
-0.14
0.01
1
Poland
0.05
-0.05
0.19
-0.11
-0.01
1
Slovakia
0.08
0.41
0.00
0.28
-0.08
0.25
1
Hungary
0.02
0.00
0.19
-0.04
-0.05
-0.26
-0.39
HU
1
Result 4 Correlation of shocks
The Euro Area countries
 “EMU will reduce the incidence of country specific shocks”
(European Commission 1990)
 Strong correlation of supply shocks; Weak correlation in demand
shocks → not a homogenous zone
For CEECs countries:
 The correlation of demand shocks is much weaker and confusing
in terms if signs than the correlation in supply shocks (Firmuc and
Korhonen (2003))
 Differences in demand shocks mostly emanate from different
economic policies (e.g. Fiscal policy in developing countries) and
differences and changes in exchange rate regimes
Result 4 The correlation of supply shocks is more important
0.6
Germany
France
0.4
Italy
Hungary
0.2
Poland
Romania
Slovakia
0.0
Correlations of Supply Shocks
0.8
for assessing the degree of business cycle integration
-0.1
0.0
0.1
0.2
0.3
0.4
Correlations of Demand Shocks
0.5
0.6
0.7
Result 5 Correlation in shocks between EA and Romania

Contemporaneous correlation: positive supply shock correlation (0.15)
and null demand shock correlation
 Time evolution of correlation in shocks:
2004
2006
2008
Supply
Shocks
0.54
0.52
0.15
Demand
Shocks
-0.37
-0.37
0.00
 Supply shocks correlation positive and rather strong (Caution – small
sample reduced the significance (2/√n))
 Supply shock correlation decline in 2007 (floods) – different reaction
 Demand shock correlation negative: different policies, change in
exchange rate regime, complete liberalization of the capital account in 2005
Result 6 Correlation coefficients of Impulse Response
Functions to Supply Shocks
Impulse response of Output
EA
EA
Germany
France
Italy
1
Impulse response of Prices
EA
Germany
France
Italy
1
Germany
0.99
1
France
0.95
0.92
1
Italy
0.97
0.96
0.92
Romania
0.80
0.85
Poland
0.96
Slovakia
Hungary
0.88
1
0.72
0.81
1
1
0.26
-0.16
0.06
1
0.66
0.70
0.26
0.31
0.50
0.33
0.92
1.00
0.91
-0.26
-0.34
-0.02
0.47
0.43
0.36
0.63
0.35
-0.50
-0.74
-0.84
0.22
0.97
0.96
0.91
0.94
-0.76
-0.88
-0.96
0.02
Result 6 Correlation coefficients of Impulse Response
Functions to Supply Shocks



Strong correlation in response of the economy to shocks for EA
countries; Weaker results for correlation of IRF to Demand
Shocks
Lower speed of adjustment in case of some of the CEECs
countries: Romania and Slovakia
Strong correlation in response of output to a supply shock for
Poland and Hungary
Result 7 Forecast error Variance Decomposition

Variation in real GDP growth is explained (after one
quarter):




80-95% by supply shocks in EA countries Germany, Italy and
France (curiously enough in EA as a whole the percentage is
50%)
70% in Romania
Technology shocks not only dominate variations of real
GDP growth in the long run but they are also important
for short-term output movements
Variation in GDP deflator inflation is explained almost
equally by the two shocks both in EA countries and in
CEECs
Conclusions






The most important issue: Correlation of supply shocks
Strong and significant correlations between EA countries; still, the
correlation among EA countries is not perfect
Significantly weaker shock correlation with CEECs; weaker
demand correlation (different policies)
Correlated responses but still differences with CEECs
For Romania, the correlation in supply shocks has been positive
and the correlation of demand shocks negative (different policies
and exchange rate regimes)
Acceptance of new countries would not affect that much the EA
countries. It’s rather a problems of the acceding economies to be
correlated with the EA and not to have too high costs.
Selective Bibliography 1
Alesina, A., R. J. Barro, and S. Tenreyro (2002), "Optimal Currency Areas", National Bureau of Economic
Research Working Papers, 9072
Babetski, J., L. Boone and M. Maurel (2003), “Exchange Rate Regimes and Supply Shocks Asymmetry: the
Case of the Accession Countries”, Discussion Paper No. 3408, CEPR, London
Bayoumi, T. and B. Eichengreen (1992), “Shocking Aspects of Monetary Unification,” NBER Working Paper
No.39499
Bayoumi, T. and B. Eichengreen (1994), “One money or many? Analyzing the prospects for monetary
unification in the various parts of the world”, Princeton Studies in International Finance, No. 76
Blanchard, O. and D. Quah (1989), “The dynamic effects of aggregated demand and supply disturbances”,
American Economic Review, Vol. 79, No. 4, pp. 655-673
Darvas, Z. and G. Vadas, (2004), “Univariate Detrending and Business Cycle Similarity Between the Euroarea and New Members of the EU”, Magyar Nemzeti Bank Working Papers
Enders, W. and S. Hurn, , (2007), “Identifying aggregate demand and supply shocks in a small open
economy”, Oxford Economic Papers, No. 59, pp. 411-429
Enders, W. (2004), “Applied Econometric Time Series” 2nd ed. Wiley
Fidrmuc, J. (2002), “Migration and Regional Adjustment to Asymmetric Shocks in Transition Economies”,
CEPR Di
scussion Paper
Fidrmuc, J., Korhonen, I. (2001), “Similarity of Supply and Demand Shocks between the Euro Area and the
CEECs,” BOFIT Discussion Paper 13, Bank of Finland, Institute for Economies in Transition, Helsinki
Fidrmuc, J., Korhonen, I. (2003), “The euro goes East. Implications of the 2000-20002 economic slowdown
for synchronization of business cycles between the euro area and CEECs” BOFIT Discussion Paper 6,
Bank of Finland, Institute for Economies in Transition, Helsinki
Frankel, J. A., Rose A.K. (1996), “The Endogeneity of the Optimum Currency Area Criteria”, CEPR
Discussion Paper No. 1473
Selective Bibliography 2
Frenkel, M. and Nickel, C. (2002), “How symmetric are the shocks and the shock adjustment dynamics
between the Euro Area and Central Eastern European Countries?”, International Monetary Fund,
Working Paper, No. 222
Frenkel, M., C. Nickel and G. Schmidt (1999), “Some shocking aspects of EMU enlargement”, Research
note No. 99-4, Deutsche Bank, Frankfurt am Main
Horvath, J. (2000), “Supply and Demand Shocks in Europe: Large-4 EU Members, Visegrad-5 and Baltic-3
Countries”, Central European University, mimeo
Horvath, J. (2003), “Optimum Currency Area Theory: A Selective Review”, Bank of Finland, Institute for
Economies in Transition Discussion Paper Supply and Demand Shocks in Europe: Large-4 EU
Members, Visegrad-5 and Baltic-3 Countries”, Central European University, mimeo
Horvath, J. and A. Ratfai (2004), "Supply and demand shocks in accession countries to the European
Monetary Union," Journal of Comparative Economics, Vol 32, No.2, 202-211
Korhonen, I. (2001), “Some empirical tests on the integration of economic activity between the Euro area
and the accession countries”, BOFIT Discussion Paper 9, Bank of Finland, Institute for Economies in
Transition, Helsinki
Kouparitsas, M., (1999), "Is the EMU a viable common currency area? A VAR analysis of regional business
cycles" Economic Perspectives, Federal Reserve Bank of Chicago, issue Q IV, pages 2-20
Mongelli F., (2002) “’New’ views of the optimum currency area theory: What is EMU telling us?”, European
Central Bank, Working Paper No. 138
Mundell, R. A. (1961), "A Theory of Optimum Currency Areas," American Economic Review 51