Transcript Slide 1
Offshoring and Productivity:
A Micro-data Analysis
Jianmin Tang and Henrique do Livramento
Presentation to
The 2008 World Congress on National Accounts and Economic
Performance Measures for Nations
May 12–17, 2008
Outline
•
Introduction
•
Objective
•
Data
•
Offshoring and its Associated Factors
•
Offshoring and Productivity
•
Conclusion
2
What is offshoring?
Within countries
Between Firms
Domestic
Outsourcing
Between countries
International
Outsourcing
OFFSHORING
Within Firms
Domestic Supply
International
Insourcing
This paper covers
•
Material offshoring (both international outsourcing and insourcing)
•
R&D service offshoring (only international outsourcing)
3
Introduction
Firms are heterogeneous even in the same industry
•
Not all firms will participate in offshoring
•
Participants may pursue different offshoring models
•
They offshore to different geographical locations
•
They may obtain different productivity dividends from offshoring
•
Geographical locations of offshoring may matter for productivity
Better to use micro data with information on geographical
location
4
Introduction (continued)
Offshoring may depend on many factors
•
Transaction costs
– relationship-specific investment,
– contractual incompleteness,
– search efforts,
– transportation costs, and
– coordination and communication costs
•
Market conditions of foreign suppliers
– trade liberalization,
– reduction in FDI restrictions,
– the availability of cheap skilled labour, and
– the ability to produce and supply high quality products and services
•
Firm-specific factors
– productivity level,
– the skill level of workers and managers, and
– size
Offshoring is not for every firm and every foreign location
5
Introduction (continued)
Different offshoring business models
•
Simply replacing expensive domestic suppliers by cheap foreign suppliers
•
Offshoring low productive and low value added components,
•
Employing foreign expertise (e.g. R&D services) and using high-tech components
abroad for designing quality products, or
•
Offshoring due to adopting foreign technologies (e.g., investing in foreign M&E)
Offshoring is associated with different business models or
organization, which leads to different geographical locations
6
Introduction (continued)
Possible productivity dividends from offshoring
•
The composition effect
- Moving up the value chain
Focusing on high value added component
Introducing high quality products
- Specializing and obtaining economy of scale
Focusing on competency
•
The innovation effect
- Facing intense international competition
- Exposing to the world technology frontier and best management practices
•
The effects may depend on offshoring business models and geographical locations
7
Geographical location of offshoring may matter for productivity dividend
Objectives
•
What factors are associated with material or R&D offshoring?
•
Is material or R&D offshoring associated with plants’ productivity?
Does the geographical location of offshoring matter for the
association?
8
Data
The Statistics Canada’s 2005 Survey of Innovation (SI)
•
6,143 plants in logging and manufacturing industries with at least 20 employees and
$250,000 in revenue
•
The response rate is 72%
•
One-time cross-sectional, covering the year of 2004 (for the purpose of this study)
•
Outward-oriented business activities with geographical locations
Percentage of total expenditures on raw materials and components from overseas
(material offshoring)
Percentage of R&D services contracted out to independent foreign firms
(R&D offshoring)
Other plants and operations in the firm (multinationals)
Percentage of investment on foreign M&E
Percentage of revenue from exporting
•
Other important variables
Percentage of workers with university education
Plant size
9
Data (continued)
The SI was linked to 2002 and 2004 Annual Survey of
Manufacturers (ASM)
• 6,109 in-sample manufacturing plants, representing a
subpopulation of17,367 manufacturing plants.
• Production data
10
Offshoring and Its Associated Factors
11
Hypothesis 1
Offshoring is associated with:
•
outward-oriented business activities
-
•
being multinationals (establishing foreign affiliates through FDI)
adoption of foreign technology (investment in foreign M&E)
exporting
plant-specific factors
-
productivity level
skill level of workers and managers
plant size
12
outward-oriented business strategies
Being a multinational
•
•
intra-firm trade accounted for significant part of imports (47% in U.S. in 2005).
most of the intra-firm transaction is associated with intermediate inputs
Adopting foreign technologies
•
•
requiring specific materials or accessories from the foreign manufacturers
requiring the manufacturers’ expertise (R&D services)
Exporting (indirectly influencing)
•
•
•
forcing the firm to improve its cost-competitiveness by exposing a firm to
international competition
better understanding local markets
lower offshoring transaction costs due to established networks
13
Plant-specific factors
Productivity level
•
•
offshoring is often considered to be endogenous to productivity (Amiti and Wei (2006).
high-productivity firms more likely engage in offshoring activities than low-productivity firms
(Antràs and Helpman 2004)
Skill level
•
•
knowledge and skills are required to coordinate the complexity involved in offshoring
skilled workers are required to specify R&D projects for offshoring and to develop absorptive
capacity to benefit from R&D offshoring.
Plant size
•
Large firms are perceived to be more likely to engage in offshoring than small firms.
the cost of financing offshoring projects is lower
larger benefit from economies of scale
lower risk of offshoring due to its large scope
14
Regression model 1
Oi , 04 0 1 M i 2 Ti , 04 3 Ei , 04 4 Pi , 02 5 U i , 04 6 S i , 02
3
6
20
j 1
k 1
m 1
6 j Di , j 9 k Li ,k 15 m I i ,m i ,
where Oi , 04 is the percentage of materials or R&D services in 2004 that are offshored;
M i is a dummy variable for plant i to be part of a multinational that has operation in
foreign location;
Ti , 04 is the percentage of plant i’s total expenditures on new M&E in 2004 that is
supplied from overseas;
Ei ,04 is the percentage of plant i’s total revenue in 2004 that come from abroad;
Pi ,02 is defined as value-added per worker in 2002;
Qi , 04 is a variable for skills, indicated by the percentage of workers with a university
education in 2004;
S i , 02 is a firm size dummy based on employment in 2002, taking the value one for
large firms and zero otherwise;
Di , j is a binary offshoring location dummy, taking the value one if plant i is
offshoring to a foreign country/region j and zero otherwise;
Li ,k is a binary operating location dummy, taking the value one if plant i is located k
in Canada and zero otherwise;
I i ,m is a binary industry dummy, taking the value one if plant i belongs to industry m
and zero otherwise; and
j
i is the error term that is associated with geographical location j.
15
Estimation results: material offshoring and the associated factors
Variable
Multinationals
Percentage of investment in foreign M&E in 2004
Share of revenue from exports in 2004
Productivity in 2002
Share of uni. educated workers
Dummy: large-sized plant
(1)
5.873***
(5.7)
0.168***
(15.6)
0.138***
(10.9)
-1.449
(-0.6)
0.112***
(3.4)
0.891
(0.6)
Dummy: material offshoring to U.S.
Dummy: material offshoring to Europe
Dummy: material offshoring to Asia Pacific
Dummy: plant located in Quebec
Dummy: plant located in Ontario
Dummy: plant located in Manitoba
Dummy: plant located in Saskatchewan
Dummy: plant located in Alberta
Dummy: plant located in British Columbia
Industry fixed effects
Adjusted R-squares
Number of observations
Yes
0.22
5073
(2)
4.126***
(4.6)
0.103***
(10.9)
0.071***
(6.3)
-2.289
(-1.1)
0.036
(1.2)
-0.593
(-0.5)
22.044***
(26.9)
11.886***
(12.5)
15.974***
(16.8)
1.023
(0.7)
2.485
(1.6)
1.025
(0.5)
-0.462
(-0.2)
-2.568
(-1.4)
1.711
(1.0)
Yes
0.41
5073
Note: t-statistics are in parenthesis. “*”, “**”, and “***” denote significance at 10%, 5% and 1%, respectively.
16
Estimation results: R&D offshoring and the associated factors
Variable
Multinationals
Percentage of investment in foreign M&E in 2004
Share of revenue from exports in 2004
Productivity in 2002
Share of uni. educated workers
Dummy: large-sized plant
(1)
-0.085
(-0.3)
0.014***
(4.7)
0.018***
(5.1)
0.382
(0.6)
0.067***
(7.3)
1.057**
(2.5)
Dummy: R&D offshoring to U.S.
Dummy: R&D offshoring to Europe
Dummy: R&D offshoring to Asia Pacific
Dummy: plant located in Quebec
Dummy: plant located in Ontario
Dummy: plant located in Manitoba
Dummy: plant located in Saskatchewan
Dummy: plant located in Alberta
Dummy: plant located in British Columbia
Industry fixed effects
Adjusted R-squares
Number of observations
Yes
0.04
5073
(2)
-0.104
(-0.6)
0.003*
(1.8)
0.002
(0.9)
0.141
(0.3)
-0.013**
(-2.2)
0.133
(0.5)
32.159***
(58.0)
24.675***
(29.3)
30.864***
(28.9)
-0.536*
(-1.7)
-0.828***
(-2.7)
-0.067
(-0.1)
-0.815
(-1.5)
-0.787**
(-2.1)
-0.722**
(-2.1)
Yes
0.62
5073
Note: t-statistics are in parenthesis. “*”, “**”, and “***” denote significance at 10%, 5% and 1%, respectively.
17
Productivity and Offshoring
18
Hypothesis 2
•
Productivity (MFP) is associated with offshoring and the
association differs across geographical locations
-
The composition effect
Moving-up the value chain
Specializing and obtaining economy of scale
-
The innovation effect
Increased competition
Exposure to advanced technologies and business practices
-
These effects may differ across geographical locations
19
Regression model 2
EU
AP
OT
ln(Pi ,04 ) 0 1 ln(Fi ,04 ) 2 OiUS
, 04 3 Oi , 04 4 Oi , 04 5 Oi , 04 6 Ri , 04
6
20
k 1
m 1
7 M i 8 Qi ,04 9 S i ,02 9 k Li ,k 15 m I i ,m i ,
where ln(Pi ,04 ) is defined as value-added per worker in 2004;
ln(Fi ,04 ) is fuel and power consumption per worker in 2004, a proxy for capital intensity;
OiUS
, 04 is the percentage of total expenditure on materials that imported from the U.S. in
2004;
OiEU
, 04 is the percentage of total expenditure on materials that imported from Europe in
2004;
AP
Oi ,04 is the percentage of total expenditure on materials that were supplied from Asia
Pacific in 2004;
OT
Oi ,04 is the percentage of total expenditure on materials that were supplied from the rest
of countries including Mexico in 2004;
Ri,04 is the percentage of total expenditure on R&D services that were supplied from
overseas in 2004;
M i is a dummy variable for being a plant of a multinational, taking the value one if the
plant is part of a multinational and zero otherwise;
Qi ,04 is a variable for skills, indicated by the percentage of workers with a university
education in 2004;
Si,04 is a plant size dummy based on employment in 2002, taking the value one for large
firms and zero otherwise (from ASM);
Li ,k is a binary operating location dummy, taking the value one if plant i is located k in
Canada and zero otherwise;
I i ,m is a binary industry dummy, taking the value one if firm i belongs to industry m and
zero otherwise; and
i is the error term.
20
Control variables
Fuel and power consumption per worker, as a proxy for capital intensity
•
•
working capital stock is highly correlated with fuel and power consumption
industry differences in energy intensity are accounted for by industry dummies.
Being multinationals
•
multinationals are more productive than non-multinationals (Baldwin and Gellatly,
2007)
scale, scope, diversified markets,
unique technology, and superior business organizations
Skills
•
•
important for technology adoption and innovation
Forming and managing business organizations
Size, operating location and industry dummies
•
introduced to capture specific effects from differences in local business
environment, financial and technological opportunities across different size
groups, operating locations and industries.
21
Estimation results: offshoring and productivity
Variable
Regression Regression Regression Regression Regression
(1)
(2)
(3)
(4)
(5)
Fuel and power consumption per 0.285***
0.286***
0.275***
0.270***
0.269***
worker
(40.0)
(40.0)
(38.2)
(37.5)
(37.4)
Material offshoring
0.155***
0.080***
(7.1)
(3.7)
to U.S.
0.151***
0.051*
0.039
(4.2)
(1.9)
(1.5)
to Europe
0.243***
0.153**
0.169***
(3.8)
(2.4)
(2.7)
to Asia Pacific
0.219***
0.153***
0.132**
(4.1)
(2.9)
(2.5)
to other countries
0.204***
0.195***
0.157**
(2.8)
(2.7)
(2.2)
R&D offshoring
0.159**
0.157**
0.031
0.053
0.054
(2.1)
(2.1)
(0.4)
(0.7)
(0.7)
Multinationals
0.158***
0.150***
0.149***
(9.5)
(9.1)
(9.1)
Share of university educated
0.484***
0.467***
0.470***
workers
(9.9)
(9.6)
(9.7)
Dummy: large-sized firms
0.058**
0.066***
0.066***
(2.3)
(2.7)
(2.6)
Dummy: plant located in Quebec
0.224***
0.225***
(7.6)
(7.6)
Dummy: plant located in Ontario
0.277***
0.275***
(9.7)
(9.6)
Dummy: plant located in Manitoba
0.174***
0.171***
(4.1)
(4.1)
Dummy: plant located in
0.226***
0.221***
Saskatchewan
(4.5)
(4.4)
Dummy: plant located in Alberta
0.302***
0.302***
(8.7)
(8.7)
Dummy: plant located in British
0.314***
0.315***
Columbia
(9.7)
(9.7)
Industry-fixed effects
Yes
Yes
Yes
Yes
Yes
Adjusted R-squares
0.36
0.37
0.39
0.40
0.40
Number of observations
5653
5653
5653
5653
5653
22
Concluding remarks
•
Material offshoring was highly associated with firms’ outward-oriented business
activities including foreign operation, investing in foreign M&E, and exporting,
after controlling for offshoring and operating locations advantages, and industryspecific effects.
•
For R&D offshoring (international outsourcing only), it is found that it was mainly
associated with investment in foreign M&E.
•
Material offshoring is positively associated with productivity and the association is
significantly larger for material offshoring to non-U.S. countries than for material
offshoring to the U.S. after controlling for other effects including being
multinationals, the education level of workers, and plant size.
23
Concluding remarks (continued)
•
The results should be interpreted with the understanding:
The results are only about association not about causal relationship
No delayed effects
The result that material offshoring to non-U.S. countries tends to be associated
with higher productivity than material offshoring to the U.S may be justified by
higher transaction cost associated with material offshoring to non-U.S. countries.
24