How Much do Trade and Financial Linkages Matter for Business Cycle Synchronization

1
How Much do Trade and Financial Linkages Matter
for Business Cycle Synchronization?

• Alicia García-Herrero and Juan Ruiz
• Bank of Spain

EcoMod Conference Istanbul, 29 June 2005
2
Outline

• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions

3
Introduction

• Increasing globalization in recent years
• Increase in trade
• Increase in financial flows
• Does that increase in trade and financial links
  tend to synchronize business cycles?
• Not clear theoretically
• Ultimately the data must answer this

4
Why does Synchronization matter?

• Transmission of shocks
• Design of currency areas
• Why does the source of synchronization matter?
• Effect of stabilization policies might be smaller
  if source is external
• If trade links determine synchronization, then
  exchange rate policy might not be able to boost
  demand and dampen economic fluctuations

5
This paper

• Examines relationship between
• Business cycle synchronization
• Trade links
• Financial links
• Similarity of economic structure
• Examines and separates explicitly direct and
  indirect effects of the last 3 variableson
  business cycle synchronization

6
This paper

• Tries to overcome usual problems of analyzing
  this issue only for big economies (usually US or
  a group of rich economies)
• Main problem availability of bilateral financial
  linkages.
• We use bilateral financial flows data for Spain
  1997-2003
• Quality of data Bilateral financial flows
• Small, open economy. No other channels of
  influence as in the case of the US.
• Sample of 101 countries.
• Includes many small, less developed economies,
  unlike previous empirical studies.

7
What do we find?

• Stronger trade and financial links do promote
  business cycle synchronization
• but only indirectly, by generating more similar
  productive structures
• There are no direct effects of trade or financial
  links beyond those accounted for by
• a similar productive structure
• similar macroeconomic policies
• This result is robust to a number of different
  model specifications and samples of countries

8
Outline

• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions

9
Related Literature
if there is more intra-industry trade and
shocks are industry-specific. or because of
aggregate demand shocks
Financial Linkages

• Frankel Rose (1998)
• Clark van Wincoop (2001)
• Imbs (2004a,b)

Trade Linkages
Output Synchronization
-

• Kose and Yi (2001)

if there is more inter-industry trade and
shocks are industry-specific.
More Similar Productive Structure (Less
Specialization)
Literature Summary
10
Related Literature
Financial Linkages

Trade Linkages
Output Synchronization
-
Sector-specific shocks or even aggregate shocks

More Similar Productive Structure (Less
Specialization)

• Frankel Rose (1998)
• Clark van Wincoop (2001)
• Imbs (2004a,b)
• Kraay and Ventura (2001)

Literature Summary
11
Related Literature
Financial Linkages
_

• Kalemli-Ozcan et al (2003)
• Helpman Razin (1978)

Trade Linkages
Output Synchronization
-
FDI might be concentrated on sectors where source
country has comparative advantage

More Similar Productive Structure (Less
Specialization)
Allows unhinging of production and consumption,
and thus specialization
Literature Summary
12
Related Literature
demand spillovers allows transfer of
resources across countries
Imbs (2004a,b) Kose, Prasad, Terrones
(2003b) - Heathcote Perry (2003b)
Financial Linkages
/-
_
_


Trade Linkages
Output Synchronization
-
Less correlation more financial flows (risk
diversification)

More Similar Productive Structure (Less
Specialization)
Literature Summary
13
Related Literature

Financial links allow reallocation of capital by
comparative advantage
Financial Linkages
/-
_

_


Trade Linkages
Output Synchronization
-
Trade links might promote FDI in export oriented
industries, or foster intl loans(Rose Spiegel
(2004))

More Similar Productive Structure (Less
Specialization)
Literature Summary
14
Related Literature

Financial Linkages
/-
_

_


Trade Linkages
Output Synchronization
-
/-

/-
More Similar Productive Structure (Less
Specialization)
Inter- vs intra-industry trade
Literature Summary
15
Related Literature Summary

Financial Linkages
/-
_

_


Trade Linkages
Output Synchronization
-
/-

/-
More Similar Productive Structure (Less
Specialization)
16
Related Literature Summary

• Many conflicting theoretical conflicting effects
• Ultimately, the effect of trade and financial
  linkages on output synchronization is an
  empirical question
• Quality of data and level of aggregation is an
  important issue

17
Outline

• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions

18
Estimation Issues

• Single equation estimations of determinants of
  output synchronization will suffer from
  endogeneity bias
• which in principle could be solved by IV
  estimation.
• Many direct and indirect effects from trade and
  financial integration into output synchronization
• which cannot be identified by IV estimation
• Sometimes conflicting indirect effects could give
  rise to not-significant net effects
• In theory there are effects that could go in
  opposite directions in some cases
• Case for a simultaneous equation estimation of
  these relationships (3-stage least-squares)

19
Data description
GDP synchronization
20
Data description
Trade linkages
21
Data description
FDI linkages
22
Estimation

• Taking into account direct and indirect effects
  we estimate the system
• (1) ri a0 a1 Ti a2 Si a3 Fi a.X(r) er
• (2)? Ti b0 b1 Si b2 Fi b.X(T ) eT
• (3) Fi d0 d1 ri d2 Ti d.X(F ) eF
• (4) Si g0 g1 Ti g2 Fi g.X(S ) eS
• where
• ri output correlation, Ti trade linkages,
• Fi financial linkages, Si similarity of
  productive structure, X(.) controls.

GDP Synchroniz.
Trade linkages
FDI linkages
Similarity of Econ. Structure
23
Outline

• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions

24
Results OLS and IV estimation. Dependent
variable GDP synchroniz. (r).
System of Equations
GDP Synchroniz.
Significant at 10, Significant at 5,
Significant at 1
25
Results 3SLS estimation of the whole system of 4
equations
Significant at 10, Significant at 5,
Significant at 1
26
Statistically significant effects

Financial Linkages (F )
/-
_

_


Trade Linkages (T )
Output Synchronization (r)
-
/

/-
More Similar Productive Structure (S ) (Less
Specialization)
Economically significant?
27
Main message

• Only similarity of productive structure (S )
  affects directly output synchronization (r).
• More similarity higher synchronization.
• Perhaps captures prevalence of industry-specific
  shocks in the 90s.
• Net effect of trade links (T )
• increases output synchronization but indirectly
• promotes less similar productive structures (S ),
  but
• increases financial links (this effect dominates)
• Financial links (F ) also increase output
  synchronization (r)
• But indirectly by promoting more similar
  productive structures (S )
• Also increases trade links (T ), but net effect
  on S is still positive

28
Robustness checks

• Controlling for common oil shocks
• Construct index of similar fuel dependency
• Index is not statistically significant to explain
  output correlations.
• Does not change other estimates (OLS, IV, 3SLS)
• Might account for low prevalence of oil shocks in
  the 90s.
• Alternative measures of similarity of productive
  structure
• Both the significance of regressors and the
  estimated total effect of trade and financial
  integration on output synchronization remain
  relatively unchanged.

Index description
29
Outline

• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions

30
Conclusions and Extensions

• Paper studies the role of trade and financial
  links in the synchronization of business cycles
• Allows for multiple, indirect links
• In theory, many direct and indirect links
• Some of them may offset each other
• The sign of some of them is not even clear a
  priori
• This paper shows that the ambiguity is just
  theoretical trade and financial integration do
  promote cycle synchronization
• but only by making the productive structure in
  both countries more similar
• A result that cannot be unveiled in reduced-form
  estimation

31
Conclusions (contd)

• Results seem robust to different specifications
  and samples
• External liberalization (trade or financial)
  without flexibility to reallocate resources might
  not provide expected foreign demand boosts

32
Future Work

• Further robustness checks with different measures
  of T, F, S, r.

33

• Thanks for your attention!

34
GDP Synchronization (r)

• Pearson correlation coefficient of log
  differences of annual GDP. 1990 - 2003

35
Trade Linkages (T )

• Measured as bilateral sum of imports and exports
  over sum of GDPs.
• Computed as time average.

36
Financial Linkages (F )

• Sum of inflows and outflows of FDI and portfolio
  flows between Spain and a given country.
• where Ii,j,t represents FDI and portfolio flows
  from country i to country j at time t

37
Similarity inproductive structure (S )

• Time average of discrepancies in economic
  structures.
• Measured using shares sn,i,t of value added for
  industrial sector n in country i at time t and
  constructing an indicator of distance (using data
  at 2-digit ISIC level).
• SESP,i can take values between -2 (disjoint
  structures) and 0 (same structure)

38
Index of similar oil dependency

• Constructed as the product of average oil
  dependency in both countries
• where Moili,t and Xoili,t are imports and exports
  of oil by country i in year t.

39
Gravity Variables used in Trade regression

• Distance to main city (in km)
• Common Language (dummy var.)
• Access to seacoast (dummy var.)
• Sum of land areas (km2)
• Product of populations
• Product of GDPs

40
Gravity Variables used in Financial links
regression

• Distance to main city
• Common language
• Time difference to main financial center
• Sum of per capita GDPs

41
Are these effects economically significant?

• Recall system of equations
• (1) ri a0 a1 Ti a2 Si a3 Fi a.X(r) er
• (2)? Ti b0 b1 Si b2 Fi b.X(T ) eT
• (3) Fi d0 d1 ri d2 Ti d.X(F ) eF
• (4) Si g0 g1 Ti g2 Fi g.X(S ) eS
• Total effect of T on r a2 (g1 d2 g2 )
  2911.32
• Total effect of F on r a2 (g2 b2 g1 ) 3.82 x
  10-6

Estimates
42
Are Trade and Financial effects on
synchronization economically significant?

• Total effect of T on r a2 (g1 d2 g2 )
  2911.32
• Total effect of F on r a2 (g2 b2 g1 ) 3.82 x
  10-6
• Increasing trade links by 1 standard deviation
  starting from its mean, increases bilateral cross
  country correlation of GDP from 0.710 to 0.717
• Increasing financial links by 1 standard
  deviation starting from its mean, increases
  bilateral cross country correlation of GDP from
  0.710 to 0.727
• NOT MUCH!!!

Statistically significant effects (chart)
43

• End of slides, return to
• Title page
• Introduction
• Related Literature
• Estimation
• Results
• Conclusions and Extensions