Factory Asia: The Determinants of
Multinational Activity in the Context of
Global Value Chains
Natalia Ramondo
UCSD and NBER
June 2016
Motivation: GVC and MNCs
• Increasing complexity of—and countries’ engagement in—the GVC
• MNCs: Main coordinators of the GVC
I 80% of global trade, UNCTAD 13
• GVC Foreign Direct Investment (FDI) = MNC trade Activities
I particularly important for ”factory Asia”
This Paper: Determinants of GVC FDI in Asia
• Unique plant-level data from D&B (large business registry)
I crucial: plant exposure to international trade
I plus: industries of parent & affiliates, revenues, location
I plus: foreign and domestic plants
• Detailed and comprehensive analysis of MNCs activities
I till now, analysis only possible for a handful of countries
• Host market- vs export-oriented activities of MNCs
I traditional factors
– comparative advantage, trade costs, economies of scale
I two twists
– country’s engagement in the GVC: domestic content of exports,
upstreamness of exports, ...
– degree of production fragmentation:I-O links within the MNC
Why do We Care?
• Level—and type—of attachment to the GVC seems to matter for
growth (UNCTAD, 13)
• (Rigid) GVCs seem to be the main channel through which shocks
get transmitted internationally (Boehm et al, 15)
• GVC changes the incentives for trade policy and lobbying, even
within an industry (Blanchard et al, 16)
Firm, location, and industry determinants of MNCs’ trade
activities are key for understanding the factors influencing the
participation of countries in the GVC and the attraction of MNCs
MNC Data: WorldBase
• Cross section of establishments for 2015
• Information on establishment and global ultimate HQ (GUH)I 4-digit SIC code of primary industry of operation
– establishments in manufacturing, mining, and business services
– ... belonging to GUH in any sector
I location of operation / origin
– establishments operating in Asian countries
– ... belonging to GUH in ”any” country
I Revenues and/or employment of establishment
– use both variables to impute missing revenues
I 3 indicators: establishment only exports, only imports, or both
I foreign MNC affiliate: country of GUH 6= country of establishment
• 230,130 establishments belonging to 176,054 GUH
MNC Data: Descriptive Statistics, by sector
1 Exporters are the minority of plants
2 Foreign plants are even a smaller set
3 ... but are more export-oriented than domestic plants
Exporters as % of Foreign as % of
all plants foreign plants all plants
All sectors 19 63 9
Manufacturing 24 74 10
Mining 12 38 7
Services 7 28 7
MNC Data: Descriptive Statistics, by country
• Countries with data on foreign and domestic plants
Host country Foreign as % of Exporters as % of
all plants all plants foreign plants
Malaysia 63 67 71
Indonesia 50 44 52
Singapore 41 42 52
Hong Kong 41 48 48
Thailand 31 52 72
China 23 48 82
Taiwan 22 81 77
Korea 14 45 44
Australia 13 14 23
New Zealand 11 18 0.0
India 2 16 47
Japan 1 3.4 20
MNC Data: Descriptive Statistics, by country
• Countries with data on foreign plants only
Host country Foreign exporters, as % of foreign plants
Viet Nam 76
Philippines 66
Georgia 42
Cambodia 41
Sri Lanka 36
Uzbekistan 29
Brunei 17
Bangladesh 10
Zero MNC exporters: Afghanistan; Armenia; Azerbaijan; Kyrgyzstan; Nepal; and Pak-
istan. Kazakhstan has only one observation which is an exporter.
MNC Data: Outward FDI, Selected Countries
Australia 4% Cambodia
0%
China 48%
Hong Kong 1%
India 2%
Indonesia 7%
Korea 3%
Malaysia 4%
New Zealand 0%
Pakistan 0%
Philippines 4%
Singapore 5%
Taiwan 5%
Thailand 7%
Viet Nam 7%
Japan
Australia 20%
Cambodia 0%
China 26%
Hong Kong 3%
India 13%
Indonesia 2%
Japan 10%
Korea 4%
Malaysia 2%
Pakistan 0%
Philippines 3% Singapore
12%
Sri Lanka 0%
Taiwan 2%
Thailand 2% Viet Nam
1% United States
Australia 28%
Cambodia 1%
Hong Kong 29%
India 4% Indonesia
2%
Japan 7%
Korea 1%
Malaysia 2%
Pakistan 0%
Philippines 2%
Singapore 14%
Taiwan 1%
Thailand 1%
Viet Nam 8%
China
Quality Check: ”Stylized” Facts Hold in WorldBase
1 Exporters are larger than non-exporters
2 Foreign plants are larger than domestic plants
3 Export-oriented are larger than host-market oriented foreign plants
Where do we observe GVC FDI? Country Pairs
Country of plant Country of parent Number of plants w/ imports & exports
15. Singapore Japan 164
14. Philippines Japan 171
13. Malaysia Japan 175
12. China France 177
11. Taiwan Japan 212
10. Indonesia Japan 214
9. Thailand Japan 258
8. Viet Nam Japan 306
7. China Singapore 337
6. China Korea 358
5. China Taiwan 401
4. China Germany 625
3. China United States 646
2. China Hong Kong 1314
1. China Japan 2260
Where do we observe GVC FDI? Industries
Affiliate Industry Number of plants
w/imports & exports
15. Computer Systems Design Services 391
14. Other Engine Equipment 395
13. All Other Miscellaneous General Purpose Machinery 433
12. All Other Petroleum and Coal Products 434
11. Plastics Material and Resin 465
10. Ethyl Alcohol 477
9. Farm Machinery and Equipment 490
8. Telemarketing Bureaus 532
7. Custom Computer Programming Services 542
6. Semiconductor and Related Device 694
5. Paint and Coating 710
4. Pharmaceutical Preparation 859
3. Plastics Pipe and Pipe Fitting 980
2. Other Electronic Component 1,358
1. Motor Vehicle Brake System 1,925
Data on GVC Trade and Other Data
• Domestic value-added of exports, as a share of gross exports
I Wang, Wei, and Zhu, 2014
• Index of exports upstreamness of a country
I Antras, Chor, Fally, and Hillberry, 2014
• Direct requirement coefficients, U.S. input-ouput matrix, 2002
• Other country controls: usual sources (PWT, WB, ...)
More GVC Trade, More GVC FDI
• GVC FDI = fraction of exporter plants in country n belonging to parents in i
• GVC Trade = (L) DVA of exports from country n to i ; (R) ratio of upstreamness
of exports, country n to i
0.2
.4.6
.81
fract
ion
of fo
reig
n af
filia
tes
that
exp
ort,
coun
try-p
air l
evel
.5 .6 .7 .8 .9 1DVA share, country-pair level
0.2
.4.6
.81
fract
ion
of fo
reig
n af
filia
tes
that
exp
ort,
coun
try-p
air l
evel
-.4 -.2 0 .2 .4Export upstreamness (host relative to source country), country-pair level
OLS coefficients: (L) -1.23*** (s.e. 0.18); (R) -0.16* (s.e. 0.08).
Larger Difference in Income, More GVC FDI
0.2
.4.6
.81
fract
ion
of fo
reig
n af
filia
tes
that
exp
ort,
coun
try-p
air l
evel
-4 -2 0 2 4log rgdpl (host relative to source country), country-pair level
OLS coefficient: -0.025** (s.e. 0.11).
Stronger Engagement in GVC, More Growth
Low High Observations
GVC-Trade 2.1% 3.3% 8
GVC-FDI 2.8% 3.6% 27
FDI Intensity 2.5% 3.4% 12
Annual average growth rate of real GDP per capita, for 2000-2010, median among countries in each
group. Low/High FDI intensity refers to countries with sales of foreign plants, as a share of total
sales, below/above the median share across countries. Low/High GVC-FDI refers to countries with
a fraction of foreign plants that export below/above the median share across countries. Low/high
GVC-trade refers to countries with DVA shares above/below the median across countries.
Country Determinants of GVC FDI — OLS
Dep var D(exports>0)
log of affiliate sales 0.044** 0.044** 0.044**
log rgdp -0.108** -0.067** -0.088**
× D(foreign>0) 0.072** 0.060** 0.022*
log rgdpl 1.015** 1.292** 1.342**
× D(foreign>0) -0.163+ -0.494** -0.462**
log K/L -0.688** -1.225** -1.360**
× D(foreign>0) 0.011 0.393* 0.403**
log yr. sch. -0.008 0.031* 0.047**
× D(foreign>0) 0.000 -0.042** -0.016*
rule of law -2.167**
× D(foreign>0) 0.787**
entry costs 0.006
× D(foreign>0) -0.005**
Export upstreamness -0.343**
× D(foreign>0) -0.101+
Observations 159,390 157,323 159,390
R-squared 0.480 0.465 0.480
Parent-affiliate industry and origin country fixed effects included. S.E. clustered at the parent level.
Levels of significance: ** p<0.01, * p<0.05, + p<0.1.
Industry Determinants of GVC FDI — OLS
Dep Var D(export > 0)
log affiliate sales 0.057**
(0.001)
R&D 0.521**
(0.01)
× D(foreign > 0) 0.003*
(0.235)
log kl 0.002
(0.003)
× D(foreign > 0) -0.034**
(0.008)
log sl 0.025**
(0.007)
× D(foreign > 0) -0.089**
(0.02)
Observations 96,554
R-squared 0.465
Parent industry and origin-destination country fixed effects included. S.E. clustered at the parent
level. Levels of significance: ** p<0.01, * p<0.05, + p<0.1.
Vertical Links within the MNC
• Direct requirement coefficient between the industries of parent and
affiliate
I drap: affiliate is upstream; drpa: affiliate is downstream
I strong I-O link = drap > 0 or/and drpa > 0
Where do We Observe Strong I-O Links?
Ctry of Affiliate Ctry of Parent Ind of Affiliate Ind of parent
15. China Japan Other Electrical eq ... Semiconductors ...
14. China Japan Motor Vehicle parts Plastic Products
13. China Japan Motor Vehicle parts Semiconductors ...
12. China Japan Semiconductors ... Other Electrical eq ...
11. China Taiwan Communications Eq Semiconductors ...
10. China Taiwan Computer and peripheral eq Semiconductors ...
9. China Korea Semiconductors... Communications eq
8. China Japan Semiconductors ... Motor Vehicle parts
7. China Japan Semiconductors ... Communications eq
6. China Japan Metal Working Machinery Plastic Products
5. China Japan Oth. Fabricated Metal Products Motor Vehicle parts
4. China Germany Electrical eq Navigational, Measuring, ...
3. China Taiwan Semiconductors ... Computer and peripheral eq
2. China USA Basic Chemical Oth. Gral. Purpose Machinery
1. China Japan Semiconductors ... Electrical Eq
Industry pairs in manufacturing only, at four-digit NAICS.
Facts on Vertical Links and GVC FDI
1 Plant-level
I Plants with stronger I-O links with the parent are larger
2 Country-industry of affiliates × country-industry of parent -level
I More exporter are associated with stronger I-O links
I More MNC activity is associated with stronger I-O links
– Alfaro & Charlton (09); Atalay et al. (14); Ramondo et al (15)
In The Paper
• Gravity analysis of bilateral activity of MNC affiliates
• Analysis of determinants of GVC trade, country and bilateral level
• Analysis for the U.S. using detailed aggregates extracted from the
BEA confidential data
Conclusions
• GVC FDI is done by large plants with strong I-O links with their
parents
• GVC FDI has stronger presence in countries that
I are easier to access
I comparative advantage motives are stronger
I are more engaged in GVC trade at later stages of the production
process
• GVC FDI appears to be correlated with more growth