Embed Size (px)
Citation preview
The trade and demand nexus:Do global value chains matter?
Julia WörzForeign Research DivisionOesterreichische Nationalbank
Joint work with Alexander Al-Haschimi, Frauke Skudelny andElena Vaccarino (ECB)
Eesti Pank Seminar, 3 September 2015
www.oenb.at
WORK IN PROGRESS…
Rubric
www.oenb.at [email protected]
Overview
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Rubric
www.oenb.at [email protected]
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview
Rubric
www.oenb.at [email protected]
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
Rubric
www.oenb.at [email protected]
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
Rubric
www.oenb.at [email protected]
• 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
Rubric
www.oenb.at [email protected]
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview
Rubric
www.oenb.at [email protected]
• 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
Rubric
www.oenb.at [email protected]
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview
Rubric
www.oenb.at [email protected]
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 (𝑴 𝒊 𝒋 𝒕 )=𝜶𝒊𝒋+𝜶𝟏 𝒍𝒏 (𝑻𝑭𝑬𝒊𝒕 )+𝜶𝟐 𝒍𝒏(𝑷 𝒋𝒕
𝑷𝒊𝒕 )+𝜶𝟑𝒍𝒏 (𝑬𝑹𝒊𝒋𝒕 )+𝜺𝒊𝒋𝒕
Rubric
www.oenb.at [email protected]
Data sources
IMF DOTSValue of imports between each country and its
trading partners
IMF WEO + IFSGDP, deflators, prices, exchanges rates
WIODAvailable across 40 countries over the period
1995-2011
CEPII – CHELEM DatabaseBilateral 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
Rubric
www.oenb.at [email protected]
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
Rubric
www.oenb.at [email protected]
2nd GVC-integration measure: GVC-participation
15
GVC participation index2010=100
Source: authors’ calculations based on WIOD.
Global Crisis induced a reversal in GVC participation intensity
𝑮𝑽𝑪𝑷𝒂𝒓𝒕𝒊𝒄𝒊𝒑𝒂𝒕𝒊𝒐𝒏𝒊𝒕=𝑽𝑺𝟏𝒊𝒕+𝑭 𝑽 𝒊𝒋𝒕
Rubric
www.oenb.at [email protected]
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview
Rubric
www.oenb.at [email protected]
Results
17
Regression results advanced and emerging economies, 1980-2012Pooled mean group estimator, dependent variable: import volumes
*** p<0.01, ** p<0.05, * p<0.1Source: authors’ calculations.
(1) (2) (3)long-runln(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 ECONOMIES1980-2012
(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 ECONOMIES1980-2012
Rubric
www.oenb.at [email protected]
Results
18
Regression results advanced and emerging economies, 1995-2012Pooled mean group estimator, dependent variable: import volumes
*** p<0.01, ** p<0.05, * p<0.1Source: authors’ calculations.
(7) (8) (9)long-runln(TFE) 1.55*** 1.00*** 0.96***ln(rel.M-prices) -0.78*** -0.79*** -0.79***ln(er) -0.086* -0.044 -0.047ln(VSI) 0.23***ln(TFE) x ln(VSI) 0.037***EC 0.28*** 0.28*** 0.28***short-rund_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)] -0.14const -0.42** -0.80*** -0.92***
ADVANCED ECONOMIES1995-2012
(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 ECONOMIES1995-2012
Rubric
www.oenb.at [email protected]
Results
19
1995-2011: alternative GVC-indicator (and alternative estimation methods)
Advanced economies Emerging economies
Fixed effects with
AR(1) dist. Dynamic panelFixed effects with
AR(1) dist. Dynamic panelLagged 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 1.16Relative 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 0.24ER 0.43*** 0.3*** 0.13*** 0.06** 0.40*** 0.38*** 0.11*** -0.01Long-term coef. 0.46 0.21 0.37 -0.03TFE*GVC_part 0.05*** 0.03*** 0.03*** 0.065***Long-term coef. 0.11 0.21
Rubric
www.oenb.at [email protected]
Results
20
Recursive estimates, including TFE*GVC-part interaction term Fixed effects with AR, advanced economiesrecursivelntfe_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 termrecursivelntfe_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 termrecursivelntfe_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 economiesrecursivelntfe_d
0.00
0.50
1.00
1.50
2.00
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Results
Rubric
www.oenb.at [email protected]
Results
21
Recursive estimates, including TFE*VSI interactionAdvanced economieslnyer_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 economieslnyer_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
Rubric
www.oenb.at [email protected]
1
2
3
Literature
Empirical set-up
Motivation
4
5 Concluding remarks
Results
Overview
Rubric
www.oenb.at [email protected]
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
