Open seminar, Eesti Pank 03.09.2015

1. The trade and demand nexus:
Do global value chains matter?
Julia Wörz
Foreign Research Division
Oesterreichische Nationalbank
Joint work with Alexander Al-Haschimi, Frauke Skudelny and
Elena Vaccarino (ECB)
Eesti Pank Seminar, 3 September 2015
www.oenb.at
WORK IN PROGRESS…

2. Rubric
www.oenb.at oenb.info@oenb.at2
Overview
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results

3. Rubric
www.oenb.at oenb.info@oenb.at3
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview

4. Rubric
www.oenb.at oenb.info@oenb.at4
Puzzle 1: Trade grew consistently faster than GDP.
Puzzle 2: The trade-GDP ratio declined.
Ratio of global import growth to GDP growth
(in PPP)
Source: ECB staff calculations.
Motivation

5. Rubric
www.oenb.at oenb.info@oenb.at5
Puzzle 1: Trade grew consistently faster than GDP.
Puzzle 2: The trade-GDP ratio declined since the mid-90s.
Ratio of global import growth to GDP growth
(in PPP) (in USD)
Source: ECB staff calculations.
Motivation

6. Rubric
www.oenb.at oenb.info@oenb.at6
• Standard trade models failed to explain changes in trade-GDP
growth ratio
• Reasons for these changes are unknown, cyclical or
structural?
• This paper analyses the role of GVCs in this context
(= possible structural reason).
• We include indicators for the participation in GVCs in a
standard import demand equation.
• Alternatively, one could analyse differences between gross
and value added trade.
Motivation

7. Rubric
www.oenb.at oenb.info@oenb.at9
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview

8. Rubric
www.oenb.at oenb.info@oenb.at10
• Yi (2001):
 Falling trade barriers too weak and too early to explain high trade
growth, but vertical specialisation can explain the puzzle
• Eaton, Kortum & Romalis (2011):
 During the crisis, spending shifted away from durable goods 
composition effect hypothesis
• Alessandria, Kaboski & Midrigan (2010); Altomonte, di Mauro,
Ottaviano, Rungi & Vicard (2012); Bems, Johnson & Yi (2012):
 Disproportionally large inventories in GVC trade, higher sensitivity of
trade to foreign income shocks, bullwhip effect  supply chain effect
hypothesis
• Constantinescu, Mattoo & Ruta (2015):
 decline in trade-GDP ratio started long before the crisis, thus reflecting
longer term structural reasons
• Ollivaud & Schwellnus (2015):
 No decline in the ratio with correct GDP measurement and treatment of
intra-EU flows, thus weak global demand is responsible
Literature

9. Rubric
www.oenb.at oenb.info@oenb.at11
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview

10. Rubric
www.oenb.at oenb.info@oenb.at
Import demand function …
… augmented by an index of GVC participation:
 Most G20 and EU countries
 Estimations for 1980-2012 and 1995-2012
 2 alternative GVC integration measures:
 Vertical specialisation index (only backward integration)
 GVC participation (based on decomposition of gross exports)
Empirical set-up
12
𝒍𝒏 𝑴𝒊𝒋𝒕 = 𝜶𝒊𝒋 + 𝜶 𝟏 𝒍𝒏 𝑻𝑭𝑬𝒊𝒕 + 𝜶 𝟐 𝒍𝒏
𝑷𝒋𝒕
𝑷𝒊𝒕
+ 𝜶 𝟑 𝒍𝒏 𝑬𝑹𝒊𝒋𝒕 + 𝜺𝒊𝒋𝒕
𝒍𝒏 𝑴𝒊𝒋𝒕 = 𝒊𝒋 +  𝟏 𝒍𝒏 𝑻𝑭𝑬𝒊𝒕 +  𝟐 𝒍𝒏
𝑷𝒋𝒕
𝑷𝒊𝒕
+  𝟑 𝒍𝒏 𝑬𝑹𝒊𝒋𝒕
+ 𝟒 𝒍𝒏 𝑮𝑽𝑪_𝒑𝒂𝒓𝒕𝒊𝒕 + 𝜺𝒊𝒋𝒕

11. Rubric
www.oenb.at oenb.info@oenb.at13
Data sources
IMF DOTS
Value of imports between each country and its
trading partners
IMF WEO + IFS
GDP, deflators, prices, exchanges rates
WIOD
Available across 40 countries over the period
1995-2011
CEPII – CHELEM Database
Bilateral trade values, 79 countries, 1967-2012
GVC
participation
Index à la
Koopman et
al. (2014)
Vertical
Specialisation
Index (VSI) à
la Amador and
Cabral (2009)
Trade
elasticities

12. Rubric
www.oenb.at oenb.info@oenb.at
1st GVC-integration measure: VSI
14
Vertical Specialisation Index (VSI)
2010=100
Source: update of Amador & Cabral, 2009.
0
20
40
60
80
100
120
140
160
180
200
1967 1972 1977 1982 1987 1992 1997 2002 2007 2012
China
India
Brazil
0
20
40
60
80
100
120
140
160
180
200
1967 1972 1977 1982 1987 1992 1997 2002 2007 2012
United States
Germany
United Kingdom
We observe increasing vertical specialisation, especially by EMEs

13. Rubric
www.oenb.at oenb.info@oenb.at
2nd GVC-integration measure: GVC-participation
15
GVC participation index
2010=100
Source: authors’ calculations based on WIOD.
Global Crisis induced a reversal in GVC participation intensity
𝑮𝑽𝑪 𝑷𝒂𝒓𝒕𝒊𝒄𝒊𝒑𝒂𝒕𝒊𝒐𝒏𝒊𝒕 = 𝑽𝑺𝟏𝒊𝒕 + 𝑭𝑽𝒊𝒋𝒕

14. Rubric
www.oenb.at oenb.info@oenb.at16
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview

15. Rubric
www.oenb.at oenb.info@oenb.at
Results
17
Regression results advanced and emerging economies, 1980-2012
Pooled mean group estimator, dependent variable: import volumes
*** p<0.01, ** p<0.05, * p<0.1
Source: authors’ calculations.
(1) (2) (3)
long-run
ln(TFE) 1.51*** 0.69*** 0.79***
ln(rel.M-prices) -0.96*** -0.97*** -0.97***
ln(er) 0.019** 0.022** 0.021**
ln(VSI) 0.33***
ln(TFE) x ln(VSI) 0.050***
EC 0.18*** 0.17*** 0.17***
short-run
d_ln(TFE) 3.21*** 3.02*** 3.38***
d_ln(rel.M-prices) -0.83*** -0.82*** -0.82***
d_ln(er) -0.031*** -0.027** -0.029**
d_ln(VSI) 0.095***
d_[ln(TFE) x ln(VSI)] 0.36
const -0.35*** -0.76*** -0.75***
ADVANCED ECONOMIES
1980-2012
(4) (5) (6)
1.36*** 0.60 0.25
-0.13** -0.13** -0.12**
0.033 0.045 0.048
0.29**
0.081***
0.13*** 0.14*** 0.14***
3.85*** 2.85*** 3.83***
-0.50*** -0.48*** -0.49***
-0.14*** -0.14*** -0.15***
0.30***
2.37***
0.084 -0.19 -0.38
EMERGING ECONOMIES
1980-2012

16. Rubric
www.oenb.at oenb.info@oenb.at
Results
18
Regression results advanced and emerging economies, 1995-2012
Pooled mean group estimator, dependent variable: import volumes
*** p<0.01, ** p<0.05, * p<0.1
Source: authors’ calculations.
(7) (8) (9)
long-run
ln(TFE) 1.55*** 1.00*** 0.96***
ln(rel.M-prices) -0.78*** -0.79*** -0.79***
ln(er) -0.086* -0.044 -0.047
ln(VSI) 0.23***
ln(TFE) x ln(VSI) 0.037***
EC 0.28*** 0.28*** 0.28***
short-run
d_ln(TFE) 3.00*** 2.87*** 3.17***
d_ln(rel.M-prices) -0.88*** -0.87*** -0.87***
d_ln(er) -0.071** -0.084*** -0.084***
d_ln(VSI) 0.067***
d_[ln(TFE) x ln(VSI)] -0.14
const -0.42** -0.80*** -0.92***
ADVANCED ECONOMIES
1995-2012
(10) (11) (12)
1.57*** 0.72** 0.43
-0.27 -0.24 -0.27*
-0.17 -0.089 -0.12
0.35***
0.085***
0.21*** 0.22*** 0.22***
3.82*** 2.62*** 3.72***
-0.68*** -0.69*** -0.64***
-0.20*** -0.25*** -0.24***
0.32***
1.42**
0.058 -0.32 -0.66*
EMERGING ECONOMIES
1995-2012

17. Rubric
www.oenb.at oenb.info@oenb.at
Results
19
1995-2011: alternative GVC-indicator (and alternative estimation methods)
Advanced economies Emerging economies
Fixed effects
with AR(1) dist. Dynamic panel
Fixed effects with
AR(1) dist. Dynamic panel
Lagged dep. 0.72*** 0.72*** 0.69*** 0.69***
TFE 1.51*** 1.41*** 0.35*** 0.25*** 1.42*** 1.40*** 0.48*** 0.36***
Long-term coef. 1.27 0.90 1.55 1.16
Relative prices -0.02 -0.05* 0.21*** 0.13*** -0.023 -0.06 0.15*** 0.07*
Long-term coef. 0.76 0.46 0.47 0.24
ER 0.43*** 0.3*** 0.13*** 0.06** 0.40*** 0.38*** 0.11*** -0.01
Long-term coef. 0.46 0.21 0.37 -0.03
TFE*GVC_part 0.05*** 0.03*** 0.03*** 0.065***
Long-term coef. 0.11 0.21

18. Rubric
www.oenb.at oenb.info@oenb.at
Results
20
Recursive estimates, including TFE*GVC-part interaction term
Fixed effects with AR, advanced economies
recursive
lntfe_d
0.00
0.50
1.00
1.50
2.00
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
with interaction term
recursive
lntfe_d
0.00
0.50
1.00
1.50
2.00
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
lntfe_d_part
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
with interaction term
recursive
lntfe_d
0.00
0.50
1.00
1.50
2.00
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
lntfe_d_part
-0.04
-0.02
0.00
0.02
0.04
0.06
0.08
0.10
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Fixed effects with AR, emerging economies
recursive
lntfe_d
0.00
0.50
1.00
1.50
2.00
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Results

19. Rubric
www.oenb.at oenb.info@oenb.at
Results
21
Recursive estimates, including TFE*VSI interaction
Advanced economies
lnyer_usd_d
0.00
0.50
1.00
1.50
2.00
2.50
3.00
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
Emerging economies
lnyer_usd_d
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
lnyer_usd_d_part
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
lnyer_usd_d_part
0.00
0.05
0.10
0.15
0.20
0.25
0.30
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
lnyer_usd_d
0.00
0.50
1.00
1.50
2.00
2.50
3.00
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
lnyer_usd_d
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012

20. Rubric
www.oenb.at oenb.info@oenb.at22
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview

21. Rubric
www.oenb.at oenb.info@oenb.at
Conclusions and way forward
23
• Countries with stronger involvement in GVCs tend to import more,
beyond the effect of the demand variable
• Omitting these variables leads to higher estimates of income elasticities
 larger difference for EME and for longer sample
 Confirming our hypothesis that GVCs explain part of the high trade to
GDP growth ratio
• Income elasticity higher and GVC elasticity lower in advanced than
emerging countries
• Sensitivity of imports to GVC-integration rising for EMEs, but declining for
AEs
• On the agenda: more (and proper) robustness checks, allow for time-
varying effect of GVC-participation, differentiate demand components

22. Rubric
www.oenb.at oenb.info@oenb.at
Thank you !