Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
On the Measurement of Upstreamness and Downstreamness in Global Value Chains Pol Antr` as Harvard & NBER
Davin Chor NUS
Dec 2017
Antr` as, Chor
On the Measurement of GVC Positioning
1 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Overview I
Backdrop: Production processes are now global in nature, with intermediate inputs obtained from multiple countries and industries along global value chains (GVCs).
I
Spurred an increased interest in understanding: (i) Where are countries and industries located along GVCs? (ii) What are the underlying determinants of country positioning along GVCs?
Antr` as, Chor
On the Measurement of GVC Positioning
2 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Overview I
Backdrop: Production processes are now global in nature, with intermediate inputs obtained from multiple countries and industries along global value chains (GVCs).
I
Spurred an increased interest in understanding: (i) Where are countries and industries located along GVCs? (ii) What are the underlying determinants of country positioning along GVCs?
I
On (i): Recent work employing techniques and concepts from input-output analysis to provide descriptive answers. (Fally 2012, Antr` as et al. 2012, Antr` as and Chor 2013, Alfaro et al. 2017, Miller and Temurshoev 2017, Wang et al. 2017)
Antr` as, Chor
On the Measurement of GVC Positioning
2 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Overview I
Backdrop: Production processes are now global in nature, with intermediate inputs obtained from multiple countries and industries along global value chains (GVCs).
I
Spurred an increased interest in understanding: (i) Where are countries and industries located along GVCs? (ii) What are the underlying determinants of country positioning along GVCs?
I
On (ii): An emerging body of general equilibrium models, though not easy to structurally discipline these models. (Yi 2003, Kohler 2004, Yi 2010, Harms et al. 2012, Baldwin and Venables 2013, Costinot et al. 2013, Antr` as and Chor 2013, Fally and Hillberry 2014, Kikuchi et al. 2017, Tyazhelnikov 2017, Alexander 2017, Antr` as and de Gortari 2017, de Gortari 2017)
Antr` as, Chor
On the Measurement of GVC Positioning
2 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
What we do This paper: Attempts to provide a bridge between these two strands of work. 1. Review four measures that have been used to capture countries’ and industries’ GVC positioning (that are based on patterns of intermediate input and final-use purchases)
2. Document how these measures have evolved over time in the 1995-2011 World Input-Output Database (WIOD). Will report several salient patterns and puzzling correlations 3. Develop an extension of Caliendo and Parro (2015) that can match all entries of a WIOD (i.e., the full pattern of intermediate input purchases and final-use expenditures)
4. Perform model-based counterfactuals to explore the role of: (i) trade cost movements; and (ii) changes in final consumption shares, in explaining the evolution of GVC positioning.
Antr` as, Chor
On the Measurement of GVC Positioning
3 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Roadmap for this talk
1. Motivation and Introduction 2. Review of GVC Measures 3. The Evolution of GVC Positioning from 1995-2011 4. Proximate Explanations 5. Structural Model 6. Counterfactuals 7. Conclusion
Antr` as, Chor
On the Measurement of GVC Positioning
4 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Measures of GVC Positioning
Antr` as, Chor
On the Measurement of GVC Positioning
5 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
World Input-Output Table (WIOT) GVC measures are defined with a WIOT in mind: I
Unit of analysis: Industry r in country i
I
J countries and S industries/sectors
I
JS by JS block of intermediate-use values
I
JS by J block of final-use values
Intermediate
Country 1
inputs
···
supplied
Country J Value added Gross output
Industry 1 ··· Industry S ··· Industry 1 ··· Industry S
Input use Country 1 Industry 1 · · · Industry S 11 1S Z11 ··· Z11 rs ··· Z11 ··· S1 SS Z11 ··· Z11 ··· ··· ··· 11 1S ZJ1 ··· ZJ1 rs ··· ZJ1 ··· S1 SS ZJ1 ··· ZJ1 V A11 ··· V AS1 Y11 ··· Y1S
Antr` as, Chor
& value added ··· Country J · · · Industry 1 · · · Industry S 11 1S ··· Z1J ··· Z1J rs ··· ··· Z1J ··· S1 SS ··· Z1J ··· Z1J Zijrs ··· ··· ··· 11 1S ··· ZJJ ··· ZJJ rs ··· ··· ZJJ ··· S1 SS ··· ZJJ ··· ZJJ V Asj V A1J ··· V ASJ Yjs YJ1 ··· YJS
Final use Country 1 · · · Country J 1 F11 ··· S F11 ··· 1 FJ1 ··· S FJ1
··· ··· ··· Fijr ··· ··· ···
1 F1J ··· S F1J ··· 1 FJJ ··· S FJJ
Total use
Y11 ··· Y1S Yir YJ1 ··· YJS
···
On the Measurement of GVC Positioning
6 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
World Input-Output Table (WIOT) GVC measures are defined with a WIOT in mind: I
Rows: Uses of output from industry r in country i
I
Columns: Value-add and input-use in the production of industry s in country j
Intermediate
Country 1
inputs
···
supplied
Country J Value added Gross output
Industry 1 ··· Industry S ··· Industry 1 ··· Industry S
Input use Country 1 Industry 1 · · · Industry S 11 1S Z11 ··· Z11 rs ··· Z11 ··· S1 SS Z11 ··· Z11 ··· ··· ··· 11 1S ZJ1 ··· ZJ1 rs ··· ZJ1 ··· S1 SS ZJ1 ··· ZJ1 V A11 ··· V AS1 Y11 ··· Y1S
Antr` as, Chor
& value added ··· Country J · · · Industry 1 · · · Industry S 11 1S ··· Z1J ··· Z1J rs ··· ··· Z1J ··· S1 SS ··· Z1J ··· Z1J Zijrs ··· ··· ··· 11 1S ··· ZJJ ··· ZJJ rs ··· ··· ZJJ ··· S1 SS ··· ZJJ ··· ZJJ V Asj V A1J ··· V ASJ Yjs YJ1 ··· YJS
Final use Country 1 · · · Country J 1 F11 ··· S F11 ··· 1 FJ1 ··· S FJ1
··· ··· ··· Fijr ··· ··· ···
1 F1J ··· S F1J ··· 1 FJJ ··· S FJJ
Total use
Y11 ··· Y1S Yir YJ1 ··· YJS
···
On the Measurement of GVC Positioning
6 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Upstreamness from Final Use Define the direct requirements coefficient: aijrs = Zijrs /Yjs . I
Start from the output accounting identity: Yir =
S X J X
aijrs Yjs + Fir .
(1)
s=1 j=1 I
Iterate to obtain: Yir = Fir +
S X J X
aijrs Fjs +
s=1 j=1
Antr` as, Chor
S X J X S X J X
st t aijrs ajk Fk + . . .
(2)
s=1 j=1 t=1 k=1
On the Measurement of GVC Positioning
7 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Upstreamness from Final Use Define the direct requirements coefficient: aijrs = Zijrs /Yjs . I
Start from the output accounting identity: Yir =
S X J X
aijrs Yjs + Fir .
(1)
s=1 j=1 I
Iterate to obtain: Yir = Fir +
S X J X
aijrs Fjs +
s=1 j=1
S X J X S X J X
st t aijrs ajk Fk + . . .
(2)
s=1 j=1 t=1 k=1
Simple measure: Final use share in gross output F /GO = Fir /Yir
Antr` as, Chor
On the Measurement of GVC Positioning
(3)
7 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Upstreamness from Final Use Richer measure: Taking into account production-staging distance from final use S P J P
Uir
=1×
Fir Yir
+2×
S P J P S P J P
aijrs Fjs
s=1 j=1
Yi r
+3×
st t aijrs ajk Fk
s=1 j=1 t=1 k=1 Yir
+ ...
(4)
Remarks: I
Uir ≥ 1
I
Uir larger if use occurs on average more stages upstream from final demand
I
Computation: I
Numerator = [I − A]−2 F
I
Denominator = [I − A]−1 F
where A is the JS by JS matrix of the aijrs ’s (direct requirements matrix); and F is the JS by 1 vector of the Fir ’s.
Antr` as, Chor
On the Measurement of GVC Positioning
8 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Upstreamness from Final Use
I
An alternative recursive formulation: ˜ ir = 1 + U
S X J X
˜ js bijrs U
(5)
s=1 j=1
so industries are upstream if they sell to industries that are themselves relatively upstream. I
Equivalence result from Antr` as et al. (2012): ˜ ir = Uir U
Antr` as, Chor
On the Measurement of GVC Positioning
9 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Downstreamness from Primary Factors Define: bijrs = Zijrs /Yir . I
Output accounting identity from the perspective of sources of value-added: Yjs =
S X J X
bijrs Yir + VAsj .
r =1 i=1 I
Iterate to obtain: Yjs = VAsj +
S X J X
bijrs VAri +
r =1 i=1
Antr` as, Chor
S X J X S X J X
bkitr bijrs VAtk + . . .
r =1 i=1 t=1 k=1
On the Measurement of GVC Positioning
10 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Downstreamness from Primary Factors Define: bijrs = Zijrs /Yir . I
Output accounting identity from the perspective of sources of value-added: Yjs =
S X J X
bijrs Yir + VAsj .
r =1 i=1 I
Iterate to obtain: Yjs = VAsj +
S X J X
bijrs VAri +
r =1 i=1
S X J X S X J X
bkitr bijrs VAtk + . . .
r =1 i=1 t=1 k=1
Simple measure: Value-added share in gross output VA/GO = VAsj /Yjs
Antr` as, Chor
On the Measurement of GVC Positioning
(6)
10 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Downstreamness from Primary Factors Richer measure:
Djs
=1×
VAsj Yjs
S P J P
+2×
S P J P S P J P
bijrs VAri
r =1 i=1
Yjs
+3×
bkitr bijrs VAtk
r =1 i=1 t=1 k=1 Yjs
+ ...
(7)
Remarks: I
Djs ≥ 1
I
Djs larger if use occurs on average more stages downstream from primary factors
I
Computation: I
Numerator = [I − B]−2 V
I
Denominator = [I − B]−1 V
where B is the JS by JS matrix of the bijrs ’s (allocation matrix); and V is the JS by 1 vector of the VAsj ’s. Antr` as, Chor
On the Measurement of GVC Positioning
11 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Downstreamness from Primary Factors
I
Alternative recursive formulation: ˜ js = 1 + D
J X S X
˜ ir . aijrs D
(8)
i=1 r =1
so industries are downstream if they purchase inputs from industries that are themselves relatively downstream. Result from Fally (2012) and Miller and Temurshoev (2017): ˜ js = Djs D
Antr` as, Chor
On the Measurement of GVC Positioning
12 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Aggregation Four GVC measures at the country-industry level: F /GO, U, VA/GO, D. Two approaches to aggregate to the country level: (i) Take a GO-weighted average of the country-industry GVC measures (ii) Collapse the WIOT to a country-by-country I-O table, and compute the GVC measures I
Both approaches clearly equivalent for F /GO and VA/GO.
I
Not equivalent, but very highly correlated for U and D.
Antr` as, Chor
On the Measurement of GVC Positioning
13 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Aggregation Four GVC measures at the country-industry level: F /GO, U, VA/GO, D. Two approaches can be applied to aggregate to the world level as well: I
Both approaches clearly equivalent for F /GO and VA/GO still.
I
But at the world level, aggregate finale expenditures equal aggregate payments to primary factors, so: F¯ /GO = VA/GO
I
Far less obvious, but GO-weighted U and D at the world level are also equal (Miller and Temurshoev 2017):
¯ =D ¯ U I
Thus: At the world-level, view these more as measures of production complexity, rather than positioning.
Antr` as, Chor
On the Measurement of GVC Positioning
13 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
GVC Positioning from 1995-2011 (from the World Input-Output Database)
Antr` as, Chor
On the Measurement of GVC Positioning
14 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
To the WIOD Data. . . Use 2013 edition of the World Input-Output Database, c.f. Timmer et al. (2015) I
J = 41 countries
I
S = 35 industries/sectors
I
16 years: 1995-2011
I
A lot of data points! Zijrs matrix in any year: (35 × 41)2 = 2, 059, 225 Fijr matrix in any year: 35 × 412 = 58, 835
I
Computational detail: Apply a net inventories correction Antr` as et al. (2012)
Antr` as, Chor
On the Measurement of GVC Positioning
15 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
For the World as a whole. . . U and D on the rise
I
Upshot: GVCs appear to be “lengthening”
.46 .48
.5
.52 .54 .56
F /GO and VA/GO on the decline
I
F/GO over time (World)
I
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2004
2006
2008
2010
2012
2.1 2 1.9 1.8
U over time (World)
2.2
Year
1994
1996
1998
2000
2002 Year
Antr` as, Chor
On the Measurement of GVC Positioning
16 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Country-level GVC Measures over Time
VA/GO over time (country percentiles)
Year
2.1 2 1.8
1.9
D over time (country percentiles)
2.1 2 1.9
U over time (country percentiles)
.42.44.46.48 .5 .52.54.56.58 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Year 2.2
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
2.2
F/GO over time (country percentiles)
.42.44.46.48 .5 .52.54.56.58
Similar patterns present across different percentiles of the country-level distribution of the respective GVC measures
1.8
I
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
1994 1996 1998 2000 2002 2004 2006 2008 2010 2012
Year
Year
Antr` as, Chor
On the Measurement of GVC Positioning
17 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Country-level GVC Measures over Time (cont.) I
Striking stability and persistence in rank order Table 1 Country-Level GVC Position by Rank (Top and Bottom Five) F/GO (1995) China 0.384 Luxembourg 0.388 Slovakia 0.394 Czech Rep. 0.408 Russia 0.444
F/GO (2011) Luxembourg 0.296 China 0.340 Korea 0.377 Taiwan 0.396 Czech Rep. 0.401
37. 38. 39. 40. 41.
Denmark Brazil Turkey Greece Cyprus
0.558 0.572 0.605 0.625 0.709
Brazil USA Mexico Cyprus Greece
Rank: 1. 2. 3. 4. 5.
Cyprus Greece Turkey Brazil Denmark
1.451 1.611 1.666 1.755 1.810
Russia Luxembourg Czech Rep. Slovakia China
2.185 2.242 2.331 2.389 2.535
Rank: 1. 2. 3. 4. 5.
37. 38. 39. 40. 41.
Rank: 1. 2. 3. 4. 5.
VA/GO (1995) China 0.373 Czech Rep. 0.403 Slovakia 0.416 Estonia 0.430 Romania 0.454
VA/GO (2011) China 0.325 Luxembourg 0.362 Korea 0.372 Czech Rep. 0.383 Bulgaria 0.401
0.563 0.575 0.575 0.576 0.625
Brazil USA Mexico Cyprus Greece
0.557 0.569 0.586 0.637 0.668
37. 38. 39. 40. 41.
Austria Turkey Brazil Greece Cyprus
Greece Cyprus Mexico USA Brazil
1.546 1.617 1.737 1.786 1.824
Rank: 1. 2. 3. 4. 5.
Cyprus Brazil Turkey Greece Austria
1.662 1.748 1.758 1.759 1.800
Greece Cyprus Mexico Brazil USA
1.657 1.763 1.779 1.806 1.808
Czech Rep. Taiwan Korea Luxembourg China
2.358 2.463 2.544 2.581 2.819
37. 38. 39. 40. 41.
Romania Estonia Slovakia Czech Rep. China
2.155 2.209 2.306 2.344 2.591
Luxembourg Bulgaria Czech Rep. Korea China
2.348 2.370 2.444 2.534 2.900
U (1995)
U (2011)
D (1995)
0.561 0.562 0.581 0.586 0.628 D (2011)
Notes: Rank order based on the respective GVC measures computed at the country-level, i.e., based on the WIOD aggregated to a country-by-country panel of Input-Output tables. The top and bottom five countries in the rank order are reported, for both 1995 and 2011.
Antr` as, Chor
On the Measurement of GVC Positioning
18 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Country-industry GVC Measures over Time I
Focusing on the “pure” within-component of the variation: Still find F /GO and VA/GO on the decline; U and D on the rise Table 2 GVC s = β1 Year + FEjs + sj,t . over Time Evolution of GVCj,tMeasures withint Country-Industries Dependent variable:
Year (Dum: Year=1996) (Dum: Year=1997) (Dum: Year=1998) (Dum: Year=1999) (Dum: Year=2000) (Dum: Year=2001) (Dum: Year=2002) (Dum: Year=2003) (Dum: Year=2004) (Dum: Year=2005) (Dum: Year=2006) (Dum: Year=2007)
(9)
F/GOj,ts
F/GOj,ts
VA/GOj,ts
VA/GOj,ts
(U)j,ts
(U)j,ts
(D)j,ts
(D)j,ts
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
-0.0009* [0.0004]
-0.0017*** [0.0005] -0.0002 [0.0025] -0.0015 [0.0020] 0.0026** [0.0010] 0.0029*** [0.0004] -0.0015 [0.0014] -0.0022 [0.0020] -0.0010 [0.0024] -0.0033 [0.0027] -0.0052 [0.0030] -0.0061* [0.0031] -0.0084** [0.0033] Antr` as, Chor -0.0119***
0.0064*** [0.0015] -0.0012 [0.0026] -0.0024 [0.0020] 0.0002 [0.0015] -0.0005 [0.0005] -0.0094*** [0.0016] -0.0122*** [0.0021] -0.0091*** [0.0022] -0.0102*** [0.0022] -0.0135*** [0.0025] -0.0153*** [0.0032] -0.0208*** [0.0036] On the Measurement -0.0237***
0.0084*** [0.0017] -0.0060 [0.0083] 0.0026 [0.0068] -0.0129*** [0.0032] -0.0086*** [0.0010] 0.0140*** [0.0045] 0.0182** [0.0065] 0.0069 [0.0069] 0.0204** [0.0082] 0.0346*** [0.0100] 0.0421*** [0.0099] 0.0598*** [0.0117] of GVC Positioning 0.0797***
0.0019 [0.0079] 0.0061 [0.0062] -0.0085* [0.0043] -0.0073*** [0.0025] 0.0311*** [0.0044] 0.0394*** [0.0053] 0.0218*** [0.0054] 0.0334*** [0.0059] 0.0490*** [0.0079] 0.0657*** [0.0101] 0.0919*** [0.0115] 0.1103***
19 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Correlations in GVC Measures across Countries
I
In an autarkic world, aggregate F and VA would be equal as a national accounting identity.
⇒ Would expect a perfect positive correlation in F /GO and VA/GO across countries I
Conversely, would expect that as trade costs fall and production becomes more fragmented, this tight correlation between F /GO and VA/GO would weaken.
I
Logic should carry over to correlation between U and D as well, since F /GO and U are negatively correlated (as are VA/GO and D)
Antr` as, Chor
On the Measurement of GVC Positioning
20 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Puzzling Correlations
.7
Correlation between F /GO and VA/GO (as well as between U and D) shows no signs of weakening! .7
CYP
.6 .5
F/GO (2011)
MEX USA BRA LTU FRA IND PRT TUR JPN ROU ESP GBR ITA DNK CAN DEU RoW IDN SVN POL NLD LVA AUT AUS MLT SWE SVK HUN FINEST BEL BGR IRL RUS CZE TWN KOR
.4
.6 .5 .4
F/GO (1995)
GRC CYP
GRC TUR BRA DNK AUT RoWMEX MLT USA DEU PRT FRA CAN JPN ESP GBR LTU ITA IND SVN LVA NLD IRL IDN SWE HUN BGR POL TWN ROU AUS FIN BEL ESTKOR RUS CZE LUX CHN SVK
.3
.3
CHN
.3
.4
.5
.6
.7
LUX .3
.4
.6
.7
3 CYP 1.5
1.75
2
2.25
2.25 2.5 2.75
U (2011)
1.5 1.75
SVK CZE LUX RUS EST KOR AUS FIN BEL ROU TWN POL HUN IDN BGR SWE NLD IRL LVA SVN IND LTU ITA GBR ESP JPN CAN FRA PRT MLT DEU USA AUT MEXRoW DNK BRA TUR GRC
CHN LUX KOR TWN CZE RUS BGR FIN IRL BEL HUN SVK AUS EST MLT SWE AUT LVA NLD POL IDN RoW SVN DEU DNK CAN GBR ITA ESP JPN ROU TUR PRT FRA IND LTU BRA USA MEX
2
2.25 2.5 2.75
CHN
2
U (1995)
.5
VA/GO (2011)
3
VA/GO (1995)
1.5 1.75
I
2.5
2.75
D (1995)
3
CYP GRC 1.5
1.75
2
2.25
2.5
2.75
3
D (2011)
Antr` as, Chor
On the Measurement of GVC Positioning
21 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Puzzling Correlations
1 .95 .9 .85 .8
Correlation: F/GO on VA/GO
Correlation between F /GO and VA/GO (as well as between U and D) shows no signs of weakening!
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2004
2006
2008
2010
2012
.7
.8
.9
1 1.1 1.2 1.3
Year Slope coefficient: F/GO on VA/GO
I
1994
1996
1998
2000
2002 Year
Antr` as, Chor
On the Measurement of GVC Positioning
21 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Puzzling Correlations
.95 .9 .85 .8
Correlation: U on D
1
Correlation between F /GO and VA/GO (as well as between U and D) shows no signs of weakening!
1994
1996
1998
2000
2002
2004
2006
2008
2010
2012
2004
2006
2008
2010
2012
1.1 1.2 1 .9 .8
Slope coefficient: U on D
Year
.7
I
1994
1996
1998
2000
2002 Year
Antr` as, Chor
On the Measurement of GVC Positioning
21 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Puzzling Correlations (at the country-industry level) I
Positive slope coefficient even in the country-industry GVC measures s s s s Table j,t 3 + FEj + FE + j,t , F /GOj,t = β1 VA/GO Correlation between Country-Industry GVC Measures (1995 and 2011)
(10)
F/GOj,ts
F/GOj,ts
F/GOj,ts
F/GOj,ts
F/GOj,ts
F/GOj,ts
1995
1995
1995
2011
2011
2011
(1)
(2)
(3)
(4)
(5)
(6)
0.5438*** [0.1815]
0.5196** [0.1924]
0.0775 [0.0543]
0.6543*** [0.1647]
0.6373*** [0.1740]
0.2647*** [0.0527]
Country FE? Industry FE?
N N
Y N
Y Y
N N
Y N
Y Y
Observations R2
1,417 0.1285
1,417 0.1488
1,417 0.8392
1,414 0.1927
1,414 0.2033
1,414 0.8479
Dependent variable:
VA/GOj,ts
Uj,ts
Uj,ts
Uj,ts
Uj,ts
Uj,ts
Uj,ts
1995 (7)
1995 (8)
1995 (9)
2011 (10)
2011 (11)
2011 (12)
0.5308***
0.4820**
0.2413***
0.6213***
0.5707***
0.3772***
[0.1640]
[0.1902]
[0.0604]
[0.1454]
[0.1698]
[0.0617]
Country FE? Industry FE?
N N
Y N
Y Y
N N
Y N
Y Y
Observations R2
1,435 0.1350
1,435 0.1742
1,435 0.8264
1,435 0.1946
1,435 0.2232
1,435 0.8325
Dependent variable:
Dj,ts
Notes: The sample comprises all countries (41), industries (35), and years (17) in the WIOD. Standard errors are multi-way clustered by Antr` Measurement of GVC Positioning country and industry; ***, **, and * denote significance atas, theChor 1%, 5%, and On 10%the levels respectively. The dependent variables are GVC
22 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
Two Candidate Explanations
Antr` as, Chor
On the Measurement of GVC Positioning
23 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
1. Trade Costs
I
Could be that trade costs remain high in absolute levels
I
Use the Head-Reis index to get an empirical handle on this: τijrs
=
τijrF
=
!− 1 2θ Zijrs Zjirs , and Ziirs Zjjrs !− 1 2θ Fijr Fjir . r r Fii Fjj
(11)
(12)
(Assume either: (i) θ = 5; or (ii) use Caliendo-Parro (2015) sectoral-level estimates.)
Antr` as, Chor
On the Measurement of GVC Positioning
24 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
1. Trade Costs Average τ remains high at the end of the period. . .
I
However: Clear overall downward trend particularly in the earlier half of the period
3
Iceberg τ 3.5 4
4.5
5
I
Assume: θ=5 2.5
Input-use mean τ Final-use mean τ 1995
1997
1999
2001
Antr` as, Chor
2003 Year
2005
2007
2009
2011
On the Measurement of GVC Positioning
25 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
1. Trade Costs I
Decline in trade costs borne out robustly in regressions Table 4 rs rs Head-Reis Trade for Intermediate Time ln τij,t = Costs β0 Year + rs , and t + FEij Inputs ij,tover Dependent variable: Industries: Year Dum: Year=1996 Dum: Year=1997 Dum: Year=1998 Dum: Year=1999 Dum: Year=2000 Dum: Year=2001 Dum: Year=2002 Dum: Year=2003 Dum: Year=2004 Dum: Year=2005 Dum: Year=2006
(13)
log Trade Costs for Intermediate Inputs (1)
(2)
(3)
(4)
(3)
(4)
All
All
Goods
Goods
Services
Services
-0.0164*** [0.0022] -0.0052 [0.0093] -0.0782*** [0.0048] -0.1108*** [0.0039] -0.1129*** [0.0048] -0.1562*** [0.0056] -0.1653*** [0.0067] -0.1594*** [0.0064] -0.1778*** [0.0100] -0.2019*** [0.0097] -0.2239*** [0.0109] -0.2491*** [0.0107] Antr` as, Chor
-0.0181*** [0.0024]
-0.0150*** [0.0026]
-0.0211* [0.0108] -0.0982*** [0.0075] -0.1503*** [0.0077] -0.1598*** [0.0074] -0.1850*** [0.0079] -0.2021*** [0.0093] -0.1936*** [0.0079] -0.2141*** [0.0123] -0.2219*** [0.0115] -0.2558*** [0.0147] -0.2781*** [0.0154] On the Measurement of GVC Positioning
0.0085 [0.0142] -0.0609*** [0.0061] -0.0768*** [0.0055] -0.0725*** [0.0063] -0.1313*** [0.0086] -0.1336*** [0.0099] -0.1299*** [0.0107] -0.1465*** [0.0154] -0.1846*** [0.0155] -0.1965*** [0.0159] -0.2241*** [0.0152]
26 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
1. Trade Costs I
Decline in trade costs borne out robustly in regressions Table 5 rF r ln τij,t = Costs β0 Year + FEijrover + Time Head-Reis Trade for tFinal-Use ij,t . Dependent variable: Industries: Year Dum: Year=1996 Dum: Year=1997 Dum: Year=1998 Dum: Year=1999 Dum: Year=2000 Dum: Year=2001 Dum: Year=2002 Dum: Year=2003 Dum: Year=2004 Dum: Year=2005 Dum: Year=2006
(14)
log Trade Costs for Final Goods/Services (1)
(2)
(3)
(4)
(3)
(4)
All
All
Goods
Goods
Services
Services
-0.0212*** [0.0039] -0.0217 [0.0332] -0.1123*** [0.0104] -0.1588*** [0.0099] -0.1479*** [0.0193] -0.2001*** [0.0257] -0.2370*** [0.0233] -0.2295*** [0.0179] -0.2500*** [0.0251] -0.2814*** [0.0304] -0.3024*** [0.0323] -0.3260*** [0.0307] Antr` as, Chor
-0.0231*** [0.0041]
-0.0196*** [0.0051]
-0.0510* [0.0278] -0.1306*** [0.0146] -0.2095*** [0.0116] -0.1910*** [0.0090] -0.2445*** [0.0162] -0.2708*** [0.0245] -0.2589*** [0.0238] -0.2825*** [0.0251] -0.2951*** [0.0303] -0.3417*** [0.0372] -0.3652*** [0.0372] On the Measurement of GVC Positioning
0.0029 [0.0550] -0.0968*** [0.0115] -0.1161*** [0.0168] -0.1115** [0.0441] -0.1628*** [0.0456] -0.2084*** [0.0360] -0.2047*** [0.0359] -0.2225*** [0.0423] -0.2699*** [0.0435] -0.2693*** [0.0435] -0.2930*** [0.0425]
26 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition
I
Key observation: Services tend to feature a higher share of output going straight to final demand, as well as a higher share of use of primary factors In other words: Services are in “short chains”, while goods are in “long Details chains”
I
If some countries are more specialized in goods and others in industries, this can account for the positive cross-country correlation between F /GO and VA/GO (as well as between U and D)
I
Could be consistent with a decline in trade costs, if this decline reinforces pre-existing patterns of specialization
Antr` as, Chor
On the Measurement of GVC Positioning
27 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition I
A secular rise over time in service share of gross output. . .
I
BUT: Some signs that patterns of specialization in services have become more concentrated over time CHN
CHN IND IDN ROU BGR TUR KOR TWN MEX IRL RoW SVK POL HUN RUS LTU SVN FIN CZE MLT
KOR TWN IDN HUN IND RoW CZE MEX TUR RUS POL BGR IRL ROU DEU SVK BRA CAN FIN EST ITA BRA PRT CAN LVA ESP DEU JPN SWE BEL NLD GBR AUT GRC DNK FRA USA AUS CYP
LTU JPN SVN SWE ITA AUT NLD BEL ESP FRA EST PRT AUS DNK USA MLT GRC LVA GBR CYP LUX
-.3
-.2
-.1
0
.1
.2
LUX
.3
Services share of Gross Output, De-meaned (1995)
Antr` as, Chor
-.3
-.2
-.1
0
.1
.2
.3
Services share of Gross Output, De-meaned (2011)
On the Measurement of GVC Positioning
28 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition
1
SS
SS SS SS SSS SSSS SSSSS SSS S SSSSSS SSS SS S S S SSS S S SSS S SS SSS S SS G SGS SSSSS S SS S SS S SS SSSS SS S SSS SS S SS SS SSS G SGGS SS S S S S G G S S SS SS SS S GS S S S SGGS S S S G S SS S S GSSS S G G SSS S GS S S SSS SS S S GSG G SG SS S G G SS G G GS G SS S S S S G G G S GG S GSG SS SS G SS SS G SS G S S G SGS GG SSS GG GGS GSS S SSS SS GGG S S G S SS S SS S S SS S SG GS G GG S SSS G G G S S G S S G S SG S G GGG SS S S SG S GG GG G GSG G GGGG S SS GG GG GG GS GG SS GG GG SGSSS S S S GSGG G S S S S G G G G GGS S G GG S G S S GSGSS G S SS G G G G SG G S S SS SSSG G GG S GG G G S S G G S G G G S G S G S S S GG GG G SG G SSSSS SSS S S S GGG S SS GGGG S SSS S S S GSG SG SG GS GG G SG G S SSSS SS G G S G SS GGS GG SS G GG G GS SS S GG S GGSS G GG G SS SSG G S SS GSG GG G GGG G G GG S G GSSS SS GG G GG GGSSGSS GSSSSSG S SSSS SS G S S S GG G SSSS GG G G GS G SS GS SGS S SSSSS G S G GG G G G G G G GG S S G G S G S S GG SGG G SS SS G GG SS SSS SSSS G S GSG GG SSSS S S SS GG GGSGSG G SSS G S S SS SSS G S GSS S G S S S S G S S S SSS SS G S GS S SS G GG G GS S SG S S S SS S S S GGSSG GGS SSS S SS GS G SSS S G SS S G SGS G SG SS S GG GSS G GGGG S SS SS G G S G GG SSS G G SG SGGS GG G G SSS G G G G G G S G G G S G S S S S S S G G S S GG GG S S S G G SGGG GG G SS S G GS S S G S SG S GG SGG S S G S S S G G S S G GG G G SS SSS G S SSS GG G SG G G SGGG S SGSSSGSGS S SG GS S G G S S GSSS S GGSSSG SSGS S S GG GS SS SSSG G GG G SG G SG G S S SS S GSS SSS GG S SSGG S GSSS S SG G G G SSG S SS S G G GG GG SSS SS GG G G G SG S G S G G G S S GG SSS S SGS GSS S S S SG S G G SG G G G G G G G GGG SGSGSSS S SGSG SG S S S S GSS SSSS SG G G GS S S GGSS G S GG SS G SS G G G GSSS GG G G S G G GG S S G S G G G S GS SSS SGS S GG SG SG G GG G G G S SSG S G GGG GGSG SGS G SSG G G S G GG G S G GSG G G G GS G G G SG G GG GG GG G G S S G S G G G G G G G G G S GSSS S G S G GG G SG G G GG GG GSGGS G SG S G SG G GGG G GG G GGGG GG G G GG GGG G G SG G G G G G G G G G GG GG GGGG G G GG G G G GG G G G G G G GG G G GG GGG G G GG S
.9
S S
.7
.8
S
.5 .4
F/GO
.6
S
.3
SS
.2
.9 .8 .7 .6 .5
F/GO
.4 .3 .2
S
.1
1
SS SS SS SSS SSSS SSSSS SSS S SS SSS SSS SS S SS S S S S SS S S SSS SS SSS SSS S SSS SSSSS SS S SS S SS S SSS SS SS SS S S S SSS SSSSS SS S S SSSS SS SS SS S S S S SSS S S SS S S S S S SSS S S SSS S SS SS S S S SS S S SSS S SS S S S S S S SSSS SS S SS S SSSSSS S S S SS S SSSSS SS S S SS S S S S S SSSS S S S S S S S S S S S S SS SS S S SS S S S S SS SSS S S S SS S S S S S S SS S SSSSS S SS SSS S S S S S S S S S SSSSS SSS S S S SS S SS S S S SSS SSSS SSS S SS S SS SS SS S S SS S SS S S SSS SS S S S S SS S S S SSSSSS S S S SS S S SS SSSSSS S SS S S SSSSSS S S S S S S S S S S SS SSS S S S SSSSSSSSS S SSS S SS SSS SS S SSSSSS S S S S SSS S S SSS SS S SS S SS S S S S S SS S S S SS S SS SSS S SS SSS SS S SSSSSS SS S S SS SS S SSS S S S SSSSS S S S S SSS S S S SS S SSS S SSS S SS S S S S S S SS SSS S SSS SS S S S S S S SSS S S S SSSS S SS SS S S SSSS SSS S S S S SSS S S SS SS SSS SSS S SS S S SSSSSS S S SSSS S S SSSSSS SS SS SS S S S S SSS SS SSS S S S S SS S SS SS SSS SSSS S SS S S S S S S S S S SS SS SS S SS S SS S SSS S SS S S SS S S SS SSSS S SS S S S
.1
S
0
G G G G G GG G GG GG GG G G GG G G G G GG G G G G GG GG GGG G GG G G GG G GGG G G G GG G G GG GG G G G GG G G G GG GG G G G GGGG GGGGG GG G GG G G G GG G GG G G G GG G G G GG G GG G G G G G G G G G GG G G G G G G G G G GG G G G GGGGGG GGGG G G G GG G G GGGGG G G GGG G G GG G GG GG GG G G GG GG G GGG G G G GG GGG G G GGG G G G G G G GG GGG G GG G G G GG G GG GG G GG GG GG G G G G GG G GG G GG G GG GGG G G G GG G G G G G GGGG GG GGGG G G G G G G G G G G G G GGG G G GG G G GG GG G G G G G G G GGG G G GG GG G G GGG G G G GGG G GGG GGG G G G GG G GG G G GG G GGG GG G GG GG G GG G G GG GGG G G G G G G G G GG G G GG GG GG G G GGGG G G GG G G G G GGGG GG G G G GG GGGGG GGGG G G GG GGG G G G G G GGG G GG G G GG GG G GG G G GG G G G G G GG GGG G GG G G GG G G G G G G GGG G G G G G GG G GGG G G G G G G G GG GG GG GGG G G G G G G G G GG G G G G GG G G G GG GGG G G GGG G G GGG G GG G GGGG GG G G GG GGGG G G G G G G G G G G G GG GG GGGG G G GG G G G GG G G G G G G GG G G GG GGG G G GG
0
.5 .4 .3 .2 .1
F/GO
.6
.7
.8
.9
1
At the country-industry level: Positive slope coefficient between GVC measures is driven by services
0
I
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
VA/GO
VA/GO
VA/GO
Goods only (2011)
Services only (2011)
All industries (2011)
Antr` as, Chor
SS S S
S
S
S
0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1
On the Measurement of GVC Positioning
29 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition At the country-industry level: Positive slope coefficient between GVC measures is driven by services S
G
G
G S S
G
1.5
2
2.5
3
3.5
4
1.5
2
2.5
3
3.5
G G G
G
G G
G
3.5 3
U 2.5 2 1.5
U 2.5
3.5 3
U 2.5 2 1.5
G
S S S S S S S SSS S S SS S S S SSS S SSS S SS S S S SSS S S SSS SSSSSS S S S S SSS S SS SS S SS S SSSS SS SS S SS S S S SSS S SSSSS S SS S SSSSSS SSS SSS S S S SS S S SS SSSS SS SS SS SSS S S SS SSSSS S SS SS S S SS SSSSSS S SSS SSS SS S SS S SS SS SS S S S S SSSS S SS SS SSSSS S S SSS SS SS S SS SS SS S SSSSS SSS S S SS SSSSSS SS SSS S S S S S SS S S S SSS S SSSS S SSSSS S S S S S S S S S S SSS SS SSSSS SS S SS SSSSS SSS SS SSS SS S S SSS SS SSSSSSS S S S SS SS SS SSSS SSSS S S SSS S SSSSSSS S S S SS SSS SSS SS S S S SSSSSSSS SSS S S S SSS S SSS S S S S SSSSS S S S SSSS SS S SSSS S SSS S S SS SS S SS S S SS SS S S SSSSS SSS SSS S S S S SS S S S SSSS S S SS S SSS SSS S SSSS S SS S SS S SSS SS S SS SS S S S S S S S S SS S SSSSSS S S S SS S SS S S SS S SSS SS S S S SS S SS SSS SS SS SSS SSSSSS S S SSSSS S SS SSSSS SSSS S SSS SS S SS S SSSSS S S S SSS SS SSS S S S S S S S S S S S S S S S S S SS SSSS S SSSSSS S SS S SSSSS SS S SS SS SS SS SS S S S S SS S
1
G
G G SG S G GG GG G SG G G G GS G G S G GS G GG G GG G G GG S S S GG G G S G GG SG GS GG S S SS G S GS G SGG GSGS G GSS GGG G S GGGG G SG GG G GG S SSSS G GG GG G GGG G G GS S G SGSSS G G G G G G SG G G G S S SSSGS G GGG SG S SG G G G G S G S SGSS G S G SGS G SG GG S GS G GS GS G GS SG GS SG G SS GG GS S SSSS G S S S G G SS GG G G GG S S GG G S SSS SS G SG G SGG S S SS G S S G G S G SS GG G G G GG G S SS GG G GS G G G G S G G GG G G S G GG G G S G G G S S G S S SGS G G G G G S S S SSSS S G S SS SS G G G G G G SG G GG G S S S S G S G SG G G SS S G G S G G SS S G SG GSS G S S G G SG SS SG G G G S S SGG G G SSSSS G SGGS S GG G G G S S G G S G G S G S S S G S SS SS SSSG S GSGGS G S SS SG S GG SSS GG G SS SS GSGG G S G G S SS SSSG G SSS G SGG S G G SG G SG G G G G G GG S G S SS GS G G G GG SSG S S S G SS SSSSSS S GG S S S S SS S SGGG G G G GGS S GS G G G S S S G G SG SS G G G SSS GGG G SG SG G SS G SS SS G SS G G S SSS SSS S GGG S G SG SS SSS SSS SSS SSS S G G S GS G GGS G G G GG G S S S SS GS S SS SS S G G S S GG GS G S G G G G G S GG S SSG G S G SS G G G S S S S G S S G S G S S G S GGG G GGG SG GG GG SS S SS SGSG S GS G GG S SS G S SSS S SSS SG G SGGG G S S SG S S G G SG SS S S G G GS SS SS G S G S G GG G G S GG G G S SG S SSS G S G G S S G S G S G S G S S S G S G S S G G G G S SS GS G S G S G SG SG SSGS G G SS G GGGG G G G S G G G SS G S SSSS S G S S S G S S G G G S SS SG GG SG S GSSG SSSS S SSS G GGS SGG G G GGG G SG GS G SSSSS S GG G S GG S S SSS G SG GG S G G G G SGG G G SSS GS G GGGGG GG G SSSS S G G G GG G GG G SG G S SS SS S G SSS S SSS SGS GG G G S G G G SG S GGGG SS S SSS SS G G SG G S SS SSG S S S GG GG S SS S S GG G G S SS SS G G SSSS SG G S S GS G G S SS S G G G G GS G GGG S S SGSG SS GGG S G S S S SSSS G SSS S SS S SS G S G SG SSSSSSGS SS GG G SSS S S S S S S S G G S G G SS S S SSSSSSSGSGS SS G G S SS SSSSS SS GS S S SS SSSG SG S S SSSSS S S SSS S S G S S S SS G S S SS SSS G S S S S S SS S SS SS S SSSSS SS SS SS S S S S S GSS
S S
3
G
1
G
3.5
G
GG
G G G
4
G G
G GG G G G G G G G G G G G G GG G GG G GGGG G G G GG G G GG GGGG G GGG G G G G GGGG G G G GG G GG GG G GG G G GGG G G GG G G G GGGG G GG G GG G G GGG G G G G G G G G GGGG GG GGG GG G G G G GG GGG G G G G G GG G GG GG GG G G G G GG GG GG GG G G G GG G G G G GG G G G GG G GG G G G G G G G G G GG G G G G G GGG G G G G G G G G G G G GG GGG GG GG G G G GG G G GG GGG GG G G GGG G G G G GGG GG GG GG G G GG G GG G G G G G GGG G G G GG GGGG G GGG G G GG G G G GGG G G GGG GG GG G GG G G G G GG G G G GG GG G G GGG G GGG G GG G GG G G G GG G GG GG G GGGGG G G GG G G G G G GGGG GGGGG GGG G G GG GGG GG GG G GG G G GGGG GG GG G GG G G G G G G G GGGG G GG G G G G G G G G G G G G G G G G GGGG G G G G GG G G GG G G G G G G GG G GG GG G GG GGG G GG GGG G GG G GG G G G G GGG G G GGGGG GG G G G GG G GG G G G GG G G GG G G G G G GGGG GG G GG G GG G GG GGG G G GG G G G GGG GG GGG GGGGGG G GGG GGGG G GG GGG G GG G G G G G G
1
G
2
G
G G
1.5
G
G
1
4
G G G
4
S G
1
I
4
1
1.5
2
2.5
3
3.5
D
D
D
Goods only (2011)
Services only (2011)
All industries (2011)
Antr` as, Chor
On the Measurement of GVC Positioning
4
29 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition I
Share of services in final expenditures has been on the rise, while that for goods has fallen Table 6
s s Final-Use ln αj,t =Expenditure β0 Yeart Shares + FEjover + Time sj,t . Dependent variable: Industries: Year Dum: Year=1996 Dum: Year=1997 Dum: Year=1998 Dum: Year=1999 Dum: Year=2000 Dum: Year=2001 Dum: Year=2002 Dum: Year=2003 Dum: Year=2004 Dum: Year=2005 Dum: Year=2006 Dum: Year=2007
log Expenditure Shares,
(15) js
(1)
(2)
(3)
(4)
(5)
(6)
All
All
Goods
Goods
Services
Services
-0.0038 [0.0025] 0.0016 [0.0148] -0.0078 [0.0117] -0.0139 [0.0081] -0.0070*** [0.0022] 0.0028 [0.0060] -0.0013 [0.0099] -0.0123 [0.0144] -0.0192 [0.0159] -0.0165 [0.0185] -0.0136 [0.0216] -0.0207 [0.0234] Antr` as, Chor -0.0226
-0.0127*** [0.0037]
0.0037 [0.0025]
-0.0087 [0.0183] -0.0230* [0.0127] -0.0493*** [0.0158] -0.0633*** [0.0077] -0.0497*** [0.0130] -0.0718*** [0.0119] -0.0991*** [0.0104] -0.1171*** [0.0134] -0.1130*** [0.0250] -0.1123*** [0.0344] -0.1164** [0.0396] On the Measurement of GVC Positioning -0.1188**
0.0104 [0.0200] 0.0050 [0.0137] 0.0160** [0.0059] 0.0405*** [0.0036] 0.0470*** [0.0086] 0.0582*** [0.0155] 0.0609** [0.0231] 0.0633** [0.0222] 0.0648*** [0.0212] 0.0697*** [0.0232] 0.0601** [0.0245] 0.0584**
30 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Definitions and Concepts Stylized Facts and Puzzling Correlations Proximate Explanations
2. Sectoral Composition I
Share of services in input purchases has been on the rise, while that for goods has fallen Table 7
rs rs rs Input-Use Shares ln γj,t = β0 Year FETime t +over j + j,t , Dependent variable: Industries: Year Dum: Year=1996 Dum: Year=1997 Dum: Year=1998 Dum: Year=1999 Dum: Year=2000 Dum: Year=2001 Dum: Year=2002 Dum: Year=2003 Dum: Year=2004 Dum: Year=2005 Dum: Year=2006 Dum: Year=2007
log Input-Use Shares,
(16) ɣjrs
(1)
(2)
(3)
(4)
(5)
(6)
All
All
Goods
Goods
Services
Services
0.0000 [0.0031] 0.0098 [0.0160] 0.0093 [0.0134] -0.0031 [0.0110] 0.0013 [0.0043] 0.0204*** [0.0032] 0.0450*** [0.0119] 0.0335** [0.0154] 0.0234 [0.0183] 0.0353 [0.0230] 0.0353 [0.0260] 0.0300 [0.0269] Antr` as, Chor 0.0309
-0.0113** [0.0043]
0.0097*** [0.0031]
-0.0050 [0.0186] -0.0209 [0.0167] -0.0580*** [0.0186] -0.0728*** [0.0087] -0.0544*** [0.0091] -0.0660*** [0.0103] -0.0894*** [0.0116] -0.1068*** [0.0166] -0.1028*** [0.0276] -0.1032** [0.0372] -0.1072** [0.0415] On the Measurement of GVC Positioning -0.1154**
0.0227 [0.0142] 0.0351** [0.0131] 0.0437*** [0.0130] 0.0643*** [0.0087] 0.0840*** [0.0062] 0.1391*** [0.0157] 0.1377*** [0.0193] 0.1337*** [0.0209] 0.1524*** [0.0242] 0.1526*** [0.0250] 0.1462*** [0.0235] 0.1549***
31 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Model
Antr` as, Chor
On the Measurement of GVC Positioning
32 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Recap: Caliendo-Parro (2015) General equilibrium model of cross-country production and trade with inter-sectoral linkages, that builds on the Eaton-Kortum machinery Setup: I
i and j denote countries; ij subscript indicates a shipment from i to j
I
r and s denote industries; rs superscript indicates a shipment from r to s
Preferences: Cobb-Douglas u(Cj ) =
S Y
Cjs
αsj
(17)
s=1
where Cjs is a sector-s composite over a unit measure of varieties (see below)
Antr` as, Chor
On the Measurement of GVC Positioning
33 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Recap: Caliendo-Parro (2015) General equilibrium model of cross-country production and trade with inter-sectoral linkages, that builds on the Eaton-Kortum machinery Setup: I
i and j denote countries; ij subscript indicates a shipment from i to j
I
r and s denote industries; rs superscript indicates a shipment from r to s
Production: Cobb-Douglas over labor and intermediates from all sectors S rs 1−PSr=1 γjrs Y s γj Mrs yjs (ω s ) = zjs (ω s ) ljs (ω s ) j (ω )
(18)
r =1 s
where zjs (ω s ) are iid draws from a Fr´echet distribution: exp{−Tjs z −θ }.
Antr` as, Chor
On the Measurement of GVC Positioning
33 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Recap: Caliendo-Parro (2015) General equilibrium model of cross-country production and trade with inter-sectoral linkages, that builds on the Eaton-Kortum machinery Setup: I
i and j denote countries; ij subscript indicates a shipment from i to j
I
r and s denote industries; rs superscript indicates a shipment from r to s
CES aggregator for Cjs and Msr j composites: Qjs =
Z
s
qjs (ω s )1−1/σ dω s
σs /(σs −1) (19)
Iceberg trade costs: τijr ≥ 1.
Antr` as, Chor
On the Measurement of GVC Positioning
33 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Caliendo-Parro (2015): Equilibrium System Tis (cis τijs )−θ
πijs
= PJ
k=1
s
Tks (cks τkjs )−θs
P 1− Sr=1 γjrs
cjs = Υsj wj
S Y
Pjr
γjrs
(20)
(21)
r =1
" Pjr
r
=A
J X
Tir
−θ cir τijr
r
#−1/θr (22)
i=1
Xjs =
S X r =1
γjsr
J X
s=1
Xjs =
(23)
i=1
| S X
Xir πjir + αjs (wj Lj + Dj )
S X J X
{z
Yjr
}
Xjs πijs =
s=1 i=1
Antr` as, Chor
S X J X
Xis πjis + Dj
(24)
s=1 i=1
On the Measurement of GVC Positioning
34 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Limitation I
Trade shares πijs do not differ by the identity of the purchaser, i.e., whether this is used to meet final demand or as an intermediate input.
I
When mapping to the data: F˜ijr πijr = PJ
k=1
F˜kjr
˜ijrs Z = PJ k=1
˜ rs Z kj
for j = 1, . . . , J.
(25)
I
Would not be satisfied for a generic WIOT
I
Why this matters: As GVC measures are computed from final-use and intermediate-use shares, desirable to have a model that can match these shares exactly
Antr` as, Chor
On the Measurement of GVC Positioning
35 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
A More Flexible Model Consider trade costs from country i to j in industry r : I
Now allow these to differ by identity of the purchasing entity.
I
If used as an input by another industry s: τijrs
I
If used to meet final demand: τijrF
I
Implies that trade share expressions (and hence price indices) will depend on the identity of the purchaser.
New equilibrium system: πijrs
Tir (cir τijrs )−θ
= PJ
k=1
πijrF
r
Tkr (ckr τkjrs )−θr
Tir (cir τijrF )−θ
= PJ
k=1
Antr` as, Chor
(26)
r
Tkr (ckr τkjrF )−θr
On the Measurement of GVC Positioning
(27)
36 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
A More Flexible Model P 1− Sr=1 γjrs
cjs = Υsj wj
S Y
Pjrs
γjrs
(28)
r =1
" Pjrs
r
=A
J X
−θ cir τijrs
Tir
r
#−1/θr (29)
i=1
" PjrF
r
=A
J X
Tir
cir τijrF
−θr
#−1/θr (30)
i=1
PjF =
S Y
PjsF /αjs
αsj
(31)
s=1
Yjs =
J X
πjksF αks (wk Lk + Dk ) +
i=1 r =1 s=1
πijsr γjsr Yjr + wj Lj =
πjksr γksr Ykr
(32)
r =1 k=1
k=1 J X S X S X
S X J X
J X S X S X i=1 r =1 s=1
πjisr γisr Yir +
S X J X
πjisF αis (wi Li + Di )
s=1 i=1
(33) Antr` as, Chor
On the Measurement of GVC Positioning
37 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Mapping to the WIOD I
Recover final expenditure share and input use share parameters in a standard way: PJ ˜ rs i=1 Zij γjrs = ˜s Y j PJ ˜ s Fij αjs = PS i=1 r fj + D ˜j VA
(34)
(35)
r =1
Antr` as, Chor
On the Measurement of GVC Positioning
38 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Mapping to the WIOD I
Recover final expenditure share and input use share parameters in a standard way: PJ ˜ rs i=1 Zij γjrs = ˜s Y j PJ ˜ s Fij αjs = PS i=1 r fj + D ˜j VA
(34)
(35)
r =1
I
Theoretical result: Suppose that these and all other underlying model parameters – other than the trade costs – are given. Then, there exist a unique set of values of τijrs and τijrF that will exactly match the observed ˜ijrs ’s and F˜ijs . Z
I
Upshot: The more flexible model can now exactly match all entries of a WIOT.
Antr` as, Chor
On the Measurement of GVC Positioning
38 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Counterfactual Changes via the “Hat Algebra” I
Caveat: While the set of τijrs ’s and τijrF ’s exist that fully match a WIOT, these are computationally not easy to back out.
I
Instead: Turn to “hat-algebra” techniques. Re-express the equilibrium system of equations in changes, following Dekle et al. (2008) and Caliendo and Parro (2015).
I
Denote change in variable X by X 0 ; and percentage changes by ˆ = X 0 /X . X
I
To evaluate counterfactual changes, need only: (i) the initial trade shares, πijrs and πijrF ; (ii) the demand and technological Cobb-Douglas parameters γjrs and αsj ; (iii) a vector of θr ’s.
Antr` as, Chor
On the Measurement of GVC Positioning
39 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Counterfactual Changes via the “Hat Algebra”
π ˆijrs
!−θr
cˆir τˆijrs ˆ rs P
=
(36)
j
π ˆijrF
cˆir τˆijrF ˆ rF P
=
!−θr (37)
j
cˆjs = (w ˆ j )1−
PS
r =1
γjrs
S Y
ˆ jrs P
γjrs
(38)
r =1
" ˆ jrs P
=
J X
πijrs
−θ cˆir τˆijrs
r
#−1/θr (39)
i=1
" ˆ jrF P
=
J X
πijrF
cˆir τˆijrF
−θr
#−1/θr (40)
i=1
Antr` as, Chor
On the Measurement of GVC Positioning
40 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Counterfactual Changes via the “Hat Algebra”
Yjs
0
=
J X
sF πjk
0
(αsk )0 (w ˆ k wk Lk + Dk ) +
J S X X
0
γksr (Ykr )0
(41)
J X S X S 0 X πjisr γisr (Yir )0
(42)
sr πjk
r =1 k=1
k=1 J X S X S 0 0 X πijsr γjsr Yjr + w ˆ j wj L j i=1 r =1 s=1
=
i=1 r =1 s=1
+
S X J X
πjisF
0
(αsi )0 (w ˆ i wi Li + Di )
s=1 i=1
Antr` as, Chor
On the Measurement of GVC Positioning
41 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Counterfactuals
Antr` as, Chor
On the Measurement of GVC Positioning
42 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
Two exercises 1. Initialize to 1995 and – holding deficits constant – evaluate whether changes in the τ ’s and/or changes in the α’s can explain observed evolution of GVC measures 2. Initialize to 2011 – holding deficits constant – explore how changes in the τ ’s and/or changes in the α’s are projected to affect the future movements in country GVC positioning
Note: Not meant to be a definitive decomposition. I
Put aside changes in the T ’s, since harder to discipline this empirically
I
Also: “hat algebra” in the current Cobb-Douglas framework not equipped to handle changes in γ’s.
Antr` as, Chor
On the Measurement of GVC Positioning
43 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
1. Changes from 1995 to 2011 I
Consider τˆijrs ’s and τˆijrF ’s from 1995-2011 (from Head-Reis, for goods vs services).
I
Consider actual changes in the final-use shares for the α’s.
I
Set: θ = 5.
Table 8 Evaluating the Role of Changes in Trade Costs and Expenditure Shares
A: Country-level GVC measures
Mean F/GO
Mean VA/GO
Correlation F/GO, VA/GO
Mean U
Mean D
Correlation U, D
Real wage change (Min, Mean, Max)
1995 baseline (from data)
0.507
0.503
0.825
1.976
1.987
0.868
---
2011 baseline (from data)
0.484
0.487
0.925
2.085
2.070
0.912
---
1995 to 2011 Shifts Change trade costs
0.518
0.502
0.612
1.940
1.984
0.666
(1.003, 1.104, 1.512)
Change expenditure shares
0.516
0.513
0.857
1.945
1.953
0.889
(0.993, 1.001, 1.017)
Both changes
0.525
0.511
0.660
1.917
1.952
0.705
(1.002, 1.093, 1.434)
Regress F/GOj,ts on VA/GOj,ts
Regress Uj,ts on Dj,ts
(Coefficient on VA/GOj,ts)
(Coefficient on Dj,ts)
B: Country-industry GVC measures 1995 baseline (from data)
0.5434***
0.5184**
0.0851
0.5337***
0.4839**
0.2564***
---
2011 baseline (from data)
0.6543***
0.6373***
0.2647***
0.6286***
0.5785***
0.4156***
---
1995 to 2011 Shifts Change trade costs
0.5534***
0.5321***
0.1101*
0.5270***
0.4844**
0.2474***
---
Change expenditure shares
0.5942***
0.5760***
0.1029*
0.5930***
0.5540***
0.2753***
---
Both changes
0.6009***
0.5854***
0.1193**
0.5856***
0.5512***
0.2609***
---
Country FE?
N
Y
Y
N
Y
Y
---
Industry FE?
N
N
Y
N
N
Y
---
Notes: Quantitative evaluations based on the multi-country, multi-industry general equilibrium model described in Section 5.2. Panel A reports moments and correlations for the country-level GVC measures, as well as real wage changes. Panel B reports the partial correlation between the country-industry level GVC measures based on the
Antr` as, Chor
On the Measurement of GVC Positioning
44 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
1. Changes from 1995 to 2011 I
Observed trade costs tend to weaken the key cross-country correlation between F /GO and VA/GO (as well as that between U and D)
I
Conversely, rise in importance of Table services (relative to goods) moves these 8 Evaluating the Role of Changes in Trade Costs and Expenditure Shares correlations in the opposite direction
A: Country-level GVC measures
Mean F/GO
Mean VA/GO
Correlation F/GO, VA/GO
Mean U
Mean D
Correlation U, D
Real wage change (Min, Mean, Max)
1995 baseline (from data)
0.507
0.503
0.825
1.976
1.987
0.868
---
2011 baseline (from data)
0.484
0.487
0.925
2.085
2.070
0.912
---
1995 to 2011 Shifts Change trade costs
0.518
0.502
0.612
1.940
1.984
0.666
(1.003, 1.104, 1.512)
Change expenditure shares
0.516
0.513
0.857
1.945
1.953
0.889
(0.993, 1.001, 1.017)
Both changes
0.525
0.511
0.660
1.917
1.952
0.705
(1.002, 1.093, 1.434)
Regress F/GOj,ts on VA/GOj,ts
Regress Uj,ts on Dj,ts
(Coefficient on VA/GOj,ts)
(Coefficient on Dj,ts)
B: Country-industry GVC measures 1995 baseline (from data)
0.5434***
0.5184**
0.0851
0.5337***
0.4839**
0.2564***
---
2011 baseline (from data)
0.6543***
0.6373***
0.2647***
0.6286***
0.5785***
0.4156***
---
1995 to 2011 Shifts Change trade costs
0.5534***
0.5321***
0.1101*
0.5270***
0.4844**
0.2474***
---
Change expenditure shares
0.5942***
0.5760***
0.1029*
0.5930***
0.5540***
0.2753***
---
Both changes
0.6009***
0.5854***
0.1193**
0.5856***
0.5512***
0.2609***
---
Country FE?
N
Y
Y
N
Y
Y
---
Industry FE?
N
N
Y
N
N
Y
---
Notes: Quantitative evaluations based on the multi-country, multi-industry general equilibrium model described in Section 5.2. Panel A reports moments and correlations for the country-level GVC measures, as well as real wage changes. Panel B reports the partial correlation between the country-industry level GVC measures based on the
Antr` as, Chor
On the Measurement of GVC Positioning
44 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
1. Changes from 1995 to 2011 I
Lower panel: These two forces go some way towards accounting for the rise in the slope-coefficient between the country-industry GVC measures Table 8 Evaluating the Role of Changes in Trade Costs and Expenditure Shares
A: Country-level GVC measures
Mean F/GO
Mean VA/GO
Correlation F/GO, VA/GO
Mean U
Mean D
Correlation U, D
Real wage change (Min, Mean, Max)
1995 baseline (from data)
0.507
0.503
0.825
1.976
1.987
0.868
---
2011 baseline (from data)
0.484
0.487
0.925
2.085
2.070
0.912
---
1995 to 2011 Shifts Change trade costs
0.518
0.502
0.612
1.940
1.984
0.666
(1.003, 1.104, 1.512)
Change expenditure shares
0.516
0.513
0.857
1.945
1.953
0.889
(0.993, 1.001, 1.017)
Both changes
0.525
0.511
0.660
1.917
1.952
0.705
(1.002, 1.093, 1.434)
Regress F/GOj,ts on VA/GOj,ts
Regress Uj,ts on Dj,ts
(Coefficient on VA/GOj,ts)
(Coefficient on Dj,ts)
B: Country-industry GVC measures 1995 baseline (from data)
0.5434***
0.5184**
0.0851
0.5337***
0.4839**
0.2564***
---
2011 baseline (from data)
0.6543***
0.6373***
0.2647***
0.6286***
0.5785***
0.4156***
---
1995 to 2011 Shifts Change trade costs
0.5534***
0.5321***
0.1101*
0.5270***
0.4844**
0.2474***
---
Change expenditure shares
0.5942***
0.5760***
0.1029*
0.5930***
0.5540***
0.2753***
---
Both changes
0.6009***
0.5854***
0.1193**
0.5856***
0.5512***
0.2609***
---
Country FE?
N
Y
Y
N
Y
Y
---
Industry FE?
N
N
Y
N
N
Y
---
Notes: Quantitative evaluations based on the multi-country, multi-industry general equilibrium model described in Section 5.2. Panel A reports moments and correlations for the country-level GVC measures, as well as real wage changes. Panel B reports the partial correlation between the country-industry level GVC measures based on the
Antr` as, Chor
On the Measurement of GVC Positioning
44 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
2. Forward Projections I
Consider a further decline in trade costs and/or the α’s for another 16 years, based on per annum rate of change from earlier regression estimates
I
Interestingly: a decline in trade costs that is biased towards services appears to induce greater specialization in services for countries with pre-existing comparative advantage. . . and this actually strengthens the correlation between F /GO and VA/GO Table 9 (and between U and D) Counterfactual Projections
Country-level measures
Mean F/GO
Mean VA/GO
Correlation F/GO, VA/GO
Mean U
Mean D
Correlation U, D
Real wage change (Min, Mean, Max)
2011 baseline (from data)
0.484
0.487
0.925
2.085
2.070
0.912
---
1995 to 2011 Shifts Change trade costs
0.482
0.476
0.840
2.095
2.101
0.815
(1.070, 1.207, 1.485)
Change trade costs (Goods only)
0.483
0.480
0.836
2.089
2.091
0.811
(1.058, 1.151, 1.269)
Change trade costs (Services only)
0.486
0.485
0.914
2.081
2.073
0.908
(1.010, 1.048, 1.286)
Change expenditure shares
0.492
0.494
0.934
2.054
2.042
0.923
(0.997, 1.000, 1.006)
Change trade costs (goods & services) and expenditure shares
0.489
0.483
0.867
2.066
2.072
0.849
(1.064, 1.189, 1.456)
Notes: Quantitative evaluations based on the multi-country, multi-industry general equilibrium model described in Section 5.2. Moments and correlations for the country-level GVC measures, as well as real wage changes, are reported. The "2011 baseline" row reports summary statistics calculated directly from the 2011 WIOT data. Under "Change trade costs", this simulates the effects of a decrease commencing in 2011 for sixteen more years, in which trade costs for intermediate goods decline at a rate of 1.81% per year, trade costs for intermediate service inputs decline at a rate of 1.50% per year, trade costs for final goods decline at a rate of 2.31% per year, and trade costs for final-use services decline at a rate of 1.96% per year, these being the rates of change estimated from Tables 4 and 5. The subsequent two rows simulate the effects of this trade cost decrease, but applying the decrease to Goods (respectively, Services) industries only. Under "Change expenditure shares", this simulates the effects of a decrease
Antr` as, Chor
On the Measurement of GVC Positioning
45 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Extending Caliendo and Parro (2015) Mapping the Model to Data Counterfactuals
2. Forward Projections Changes in individual countries’ GVC positioning (U and D) from a further trade cost decline: CHN
CHN LUX IRL
RUS BGR CZE BGR IND BEL NLD DNK HUN SVK ESP POL KOR
IRL ROU AUS IND BRA CAN KOR GRC USA CZE TUR GBR JPN RoW PRT ITA LVA POL ESP CYP IDN FIN SWE FRA SVN DEU HUN LTU EST MEX MLT AUT SVK BEL TWN NLD DNK LUX
SVN ITA DEU MEX SWE FRA CAN CYP AUT PRT GBR MLT AUS JPN USA LVA EST ROU GRC TUR FIN RoW RUS IDN BRA LTU TWN
-.25
-.2
-.15
-.1
-.05
0
.05
.1
.15
.2
.25
Counterfactual Projected Change (U)
Antr` as, Chor
-.25
-.2
-.15
-.1
-.05
0
.05
.1
.15
.2
.25
Counterfactual Projected Change (D)
On the Measurement of GVC Positioning
46 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Concluding Remarks
Antr` as, Chor
On the Measurement of GVC Positioning
47 / 48
Introduction and Overview Measure of GVC Positioning and Empirical Patterns A Quantitative Model of the WIOT
Conclusion I
Documented the evolution of GVC positioning of countries and industries, in the 1995-2011 WIOD
I
Uncovered several salient facts and puzzling correlations: I
Countries (and country-industries) that are far removed from final demand also have a high production-staging distance from primary factors
I
Explored the possible role of two forces: (i) declines in trade costs; and (ii) the rising importance of services in final consumption shares.
I
Done through the lens of a model (extending Caliendo-Parro), that fully rationalizes all the entries of a WIOT, and thus provides a more flexible basis for counterfactual exercise on countries’ GVC positioning.
Antr` as, Chor
On the Measurement of GVC Positioning
48 / 48
Supplementary Slides
Antr` as, Chor
On the Measurement of GVC Positioning
1/2
Summary Statistics
Table A.1 Back Summary Statistics: Country-Industry GVC Measures 10th
Median
90th
Mean
Std Dev
N
All industries
0.125
0.444
0.901
0.473
0.271
24,076
Goods industries only
0.076
0.373
0.700
0.379
0.240
11,105
Service industries only
0.216
0.496
0.956
0.553
0.270
12,971
All industries
0.279
0.456
0.738
0.489
0.186
24,395
Goods industries only
0.247
0.360
0.499
0.371
0.118
11,152
Service industries only
0.378
0.575
0.812
0.589
0.175
13,243
All industries
1.153
2.126
2.914
2.098
0.649
24,395
Goods industries only
1.523
2.298
3.048
2.291
0.605
11,152
Service industries only
1.055
1.982
2.771
1.936
0.640
13,243
All industries
1.502
2.141
2.624
2.092
0.450
24,395
Goods industries only
2.033
2.381
2.728
2.376
0.316
11,152
Service industries only
1.356
1.846
2.363
1.852
0.404
13,243
F/GO
VA/GO
U
D
Notes: Based on the country-industry GVC measures calculated from the WIOD for 1995-2011; the sample comprises all 41 countries, 35 industries, and 17 years. Goods industries are defined as primary and manufacturing industries, namely industries 1-16 in the WIODOn classification. The service industries are industries Antr` as, Chor the Measurement of GVC Positioning
2/2
