Positioning and Internalization in GVC: The Case of Tuscan Firms

Positioning and Internalization in GVC:
The Case of Tuscan Firms
Giorgia Giovannetti
Enrico Marvasi 1
1 Università
2 European
1 2
di Firenze
University Institute
April 28, 2016
Introduction
Positioning and Governance
Data
Descriptive Statistics
Section 1
Introduction
Econometric Analysis
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Different approaches: theory
I
Firms are nowadays organizing production globally, exploiting
the CA of different locations in increasingly specific tasks
I
Heterogeneous firms models explain how different ex-ante productivity results in ex-post productivity sorting of firms
I
These (theoretical) models highlight precise mechanisms, but
rely on strong simplifying assumptions
I
Evidence from the business literature shows ample variety between firms along several dimensions
I
To explain recent firms’ behavior, we have two choices:
get clear mechanisms and testable hypotheses, but neglect
some important aspects of actual firms
2 provide a more comprehensive picture of reality, but be
somewhat less formal in the exact mechanisms
1
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Different GVC measurements: data
I
Traditional trade statistics suffer from double-counting and do
not to track flows across firms’ boundaries
I
GVC can be traced back to I-O tables, but they only provide
an aggregate picture
I
In firm-level data, inter and intra-firms relations at the international level are usually unobserved
I
Case studies exploit very detailed information, but they are
case-specific and can be hard to generalize
I
Researchers have two main choices:
Use economy-wide representative data, but be less accurate in
the measurement of GVC (aggregate trade or I-O tables data)
2 Have better GVC measures from firm-level data, but get hard
to generalize evidence (case studies and firm-level surveys)
1
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Two perspectives on GVC
I
Different approaches from two strands of the literature
I
They identify positioning and governance as the two main dimensions of GVC
I
The business literature focused on the complexity of power relations and the governance structure, generalizing the available
evidence from case studies
I
The trade literature developed theoretical models to explain
internalization in GVC, showing clear mechanisms, but ruling
out governance complexity
Best of both worlds
I Ideally, unifying these two perspectives would make it possible
to take into account the actual GVC complexity, while
building on sound theoretical models
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Our contribution
I
We relate these two strands of the literature
I
In particular, we link the work by Gereffi and coauthors with
the work by Antràs and coauthors
I
We construct a taxonomy of firms by positioning and
governance in DVC and GVC
I
We study about 16,000 firms from the 2011 Tuscan census
Preview of results (preliminary)
I
GVC firms outperform DVC firms, which in turn underperform
I
Suppliers are penalized, while buyers have a premium
I
Most firms are in market VC, but hierarchical VC dominate
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Section 2
Positioning and Governance
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Two main dimensions of GVC
I
Firms in GVC are identified by two main dimensions:
1
Positioning: where to produce along the chain, i.e. upstream
vs. downstream, buyer vs. supplier
2
Governance: how to organize and coordinate the different
activities; degree of coordination, power asymmetry
I
Wide use of the positioning-governance types elaborated by
Gereffi et al. (2005)
I
More recently, heterogeneous firms and property-rights theory
used to model the internalization choices of buyers and suppliers
(e.g. Antràs & Yeaple, 2015)
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
The complexity of the positioning-governance relation
Centralized
Decentralized
Integrated
Figure: Governance types from Gereffi et al. (2005)
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
The internalization choice and productivity
Figure: Profits and productivity from Antràs & Yeaple (2015)
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Insights from the two perspectives
I
Gereffi et al. (2005) identify 5 governance types, positioning
may include many buyer-supplier firms
I
Antràs & Yeaple (2015) model outsourcing vs. integration, and
firms are either buyers or suppliers (but not both)
I
In Gereffi et al. (2005), firms operate in a complex world, but
mechanisms are not clear-cut
I
In Antràs & Yeaple (2015), a formal model improves our understanding, but we risk over-simplification
I
Along these lines, we construct a 3x3 taxonomy of firms
I
Necessarily reliant on data availability
Introduction
Positioning and Governance
Data
Descriptive Statistics
Section 3
Data
Econometric Analysis
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Data
I
16,435 firms from the 2011 Tuscan census (Istat)
I
45% in manufacturing (Ateco classification)
I
About 1/3 of manufacturing firms produce intermediate goods
I
7.7% are pure suppliers, 19% only buyers and 34.5% both
I
22% sell to final consumers (households)
I
15% is owned by another firm, 84% by an individual or a family
I
4% is involved in offshore outsourcing (FDI or JV)
I
53% has 10 or more emp., 3% of these firms is foreign owned
I
owner decisional power (about 72%); 28% managers or others
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
GVC participation and positioning
I
VC as inter-firm exchange of intermediate goods (manuf.)
I
DVC if buyer and supplier in Italy, GVC if in different countries
1
Downstream producers or (pure) buyers
Buyers, produce final goods, sell to households (7.8%)
2
Midstream producers or mid-suppliers
Buyer-suppliers, produce int. goods, sell to other firms (72%)
3
Upstream producers or (pure) suppliers
Suppliers, produce interm. goods, sell to other firms (20%)
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Governance and decisional centralization
I
Independent from VC participation
I
Observed also for non-VC firms
1
Market (Centralized firms)
Independent, high decisional centralization, owner (50%)
2
Relational (Decentralized firms)
Independent, lower decisional centralization, managers (35%)
3
Hierarchical (Integrated firms)
Ownership structure, parent or affiliate of other firms (15%)
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Positioning-Governance taxonomy: all firms, VC vs. no-VC
VC 1,370 firms; no-VC 15,065 firms
Positioning
Governance
Market Relational Hierarchical Total
Downstream
4.23
2.55
1.02
7.81
Midstream
36.86
24.38
10.80 72.04
Upstream
11.31
6.50
2.34 20.15
Total VC
52.41
33.43
14.16
100
Total no VC
49.32
35.39
15.29
100
Figure: Firms’ distribution by positioning and governance, in percentage
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Positioning-Governance taxonomy: DVC vs. GVC
DVC 1,140 firms; GVC 230 firms
Market Relational Hierarchical Total
Downstream
Midstream
Upstream
Total DVC
4.47
37.63
12.02
54.12
DVC
2.63
25.18
6.58
34.39
0.96 8.07
7.89 70.70
2.63 21.23
11.49 100
Downstream
Midstream
Upstream
Total GVC
3.04
33.04
7.83
43.91
GVC
2.17
20.43
6.09
28.7
1.30 6.52
25.22 78.70
0.87 14.78
27.39 100
Figure: Firms’ distribution by positioning and governance, in percentage
Introduction
Positioning and Governance
Data
Descriptive Statistics
Section 4
Descriptive Statistics
Econometric Analysis
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
10
.8
12
Sales 2011 (ln)
14
16
18
Median of Sales 2011 (2008=100)
.9
1
20
1.1
Sales, employment and export
2008
DVC
GVC
NO VC
DVC
GVC
2009
2010
NO VC
2011
DVC
2012
GVC
5
0
Export 2011 (ln)
10
15
Employment 2011 (ln)
2
4
6
20
NO VC
NO VC
DVC
GVC
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
0
10
.2
Density
.4
Sales/Empl. 2011 (ln)
11
12
13
14
.6
Sales per employee by internationalization mode
10
9
9
Domestic
Exporter
Foreign owned
FDI
Domestic
11
12
Sales/Empl. 2011 (ln)
Exporters
13
Foreign owned
14
FDI
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
0
.2
Density
.4
Mean of Sales/Empl. 2011 (ln)
10
12
14
.6
Sales per employee, GVC and internalization
10
8
9
DVC
11
12
Sales/Empl. 2011 (ln)
NO VC
GVC
DVC
13
14
GVC
0
10
.2
Density
.4
Sales/Empl. 2011 (ln)
14
12
.6
16
NO VC
8
9
DVC non-hier.
DVC hier.
GVC non-hier.
GVC hier.
10
11
12
Sales/Empl. 2011 (ln)
DVC non-hierarchical
GVC non-hierarchical
13
DVC hierarchical
GVC hierarchical
14
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Sales per employee by Positioning and Governance
Downstream
Market
Downstream
Midstream
Market
Upstream
Midstream
Upstream
Downstream
Relational
Downstream
Midstream
Relational
Upstream
Midstream
Upstream
Downstream
Hierarchical
Midstream
Downstream
Upstream
Hierarchical
0
5
10
Median of Sales/Empl. 2011 (ln)
DVC
GVC
15
Midstream
Upstream
0
1
2
3
GVC-DVC medians ratio of Sales/Empl. 2011
Introduction
Positioning and Governance
Data
Descriptive Statistics
Section 5
Econometric Analysis
Econometric Analysis
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Firms’ productivity in GVC
How are Tuscan firms positioned?
Where are the productivity gains generated?
I
We focus on productivity (measured as sales per employee)
I
We study correlation with Positioning and Governance modes
I
Control variables: employment, internationalization modes, innovation, human capital and 2-digit sector fixed effects
I
Preliminary analysis based on OLS regressions for 2011
1
DVC and GVC vs. other firms
2
DVC and GVC by Positioning-Governance (pooled regressions)
3
DVC and GVC by Positioning-Governance (governance modes)
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
DVC and GVC vs. other firms
Dep. Sales per empl. (ln) (1)
DVC
All firms
(2)
(3)
(4)
Manufacturing firms
(5)
(6)
-0.142***
(0.034)
0.752***
(0.072)
-0.213***
(0.032)
0.602***
(0.065)
-0.246***
(0.032)
0.654***
(0.066)
-0.223***
(0.032)
0.597***
(0.064)
Sector f.e.
-0.098***
(0.030)
0.297***
(0.062)
0.147***
(0.008)
0.506***
(0.017)
0.106*
(0.059)
0.225***
(0.040)
0.386***
(0.084)
0.141***
(0.017)
0.122***
(0.026)
11.626*** 11.828*** 11.264***
(0.009)
(0.110)
(0.106)
No
Yes
Yes
R-squared
N
0.008
16090
0.249
16090
0.023
7266
0.085
7266
GVC
Employment (ln)
Exporter
For. Owned
JV
FDI
Innovation
Skilled empl.
Constant
0.328
16090
S.e. in parentheses; * p<0.1, ** p<0.05, *** p<0.01.
-0.084***
(0.030)
0.271***
(0.060)
0.148***
(0.012)
0.570***
(0.023)
0.159**
(0.081)
0.336***
(0.050)
0.246**
(0.104)
0.142***
(0.023)
0.148***
(0.036)
11.724*** 11.828*** 11.241***
(0.013)
(0.109)
(0.104)
No
Yes
Yes
0.231
7266
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
DVC and GVC by Positioning-Governance
Pooled regressions
Dep. Sales per empl. (ln) Market
Downstream
Regressions by governance modes
Relational Hierarchical
DVC vs. no-VC
-0.367*** 0.003
-0.184
(0.129)
(0.167)
(0.275)
Midstream
-0.068
(0.044)
Upstream
-0.401*** -0.308*** 0.100
(0.077)
(0.104)
(0.159)
Downstream
0.126
(0.326)
Midstream
0.282***
(0.101)
0.354***
(0.127)
Upstream
-0.119
(0.204)
-0.203
(0.232)
R2
N
0.028
(0.053)
0.185**
(0.094)
Dep. Sales per empl. (ln) Market
Downstream
Relational Hierarchical
DVC vs. no-VC
-0.306** -0.037
-0.439
(0.132)
(0.153)
(0.308)
Midstream
-0.033
(0.047)
Upstream
-0.341*** -0.329*** -0.175
(0.079)
(0.098)
(0.183)
Downstream
0.134
(0.331)
0.487***
(0.117)
Midstream
0.339***
(0.104)
0.319***
(0.118)
0.380***
(0.136)
0.158
(0.612)
Upstream
-0.063
(0.206)
-0.268
(0.220)
0.065
(0.669)
R2
N
0.22
3846
0.24
2407
0.21
1013
GVC vs. no-VC
-0.103
0.991
(0.386)
(0.611)
0.24
7266
Manufacturing only; 2-digit sector f.e. and controls included.
Control group: all no-VC firms.
S.e. in parentheses; * p<0.1, ** p<0.05, *** p<0.01.
0.025
(0.052)
0.007
(0.112)
GVC vs. no-VC
-0.137
0.881
(0.353)
(0.675)
Manufacturing only; 2-digit sector f.e. and controls included.
Control group: all no-VC firms by governance type.
S.e. in parentheses; * p<0.1, ** p<0.05, *** p<0.01.
Introduction
Positioning and Governance
Data
Descriptive Statistics
Section 6
Conclusion
Econometric Analysis
Conclusion
Introduction
Positioning and Governance
Data
Descriptive Statistics
Econometric Analysis
Conclusion
Conclusion
I
We document the existence of a clear productivity sorting among
among Tuscan manufacturing firms
I
In line with trade theory, internationalized firms perform better
I
The productivity sorting applies also for GVC vs. DVC
I
Suppliers tend to be penalized in value chains
I
Moreover, we document the existence of productivity heterogeneity by governance type
I
Hierarchically organized firms are a minority but tend to be
among the top performers
I
The highest productivity premia are found for midstream GVC
firms, across all governance types