Firm Selection and Organizational Choice

Firm Selection and Organizational Choice:
Complex Patterns of Global Sourcing
Valérie Smeets∗
Frédéric Warzynski†
October 27, 2015
Abstract
In this paper, we use detailed survey data about offshoring decisions for a large
subset of Danish firms over the last decade. We investigate whether the patterns of
global sourcing that we observe in the data are in line with recent models suggested
in the literature. We show that the most productive firms self select in complex
offshoring patterns that involve bi-sourcing and location in various geographical
areas both in developed and developing nations. We discuss how these results can
be related to recent theoretical developments.
1
Introduction
Over the last fifteen years, companies have taken advantage of new offshoring possibilities,
facilitated by the information and communication technology (ICT) revolution and access
to large pools of cheap labor thanks to trade liberalization. They have created complex,
global value chains, where the production process is fragmented in different tasks, and
these various tasks are performed in several countries. As a consequence, fewer activities
are conducted in the home country of large multinational firms.
In this paper, we take advantage of two large scale surveys of Danish firms run by
Statistics Denmark (more than 4,000 companies were surveyed in two waves) about their
offshoring decisions from 2000 to 2011 to study which activities were offshored, either
within the boundaries the firm or outside, and where. Our aim is to better understand
how firms decide which activities should be outsourced and which activities should be
retained within the boundaries of the firm, and what are the consequences for the future
prospects of manufacturing activities.
In doing so, we test recent theories of global sourcing (our focus is on Antràs and
Helpman, 2004) that emphasize how organizational choices are made endogenously based
∗
†
Department of Economics and Business, Aarhus University, email: [email protected]
Department of Economics and Business, Aarhus University, email: [email protected]
1
on two key dimensions of firm heterogeneity: productivity and headquarter intensity. By
integrating the make or buy decision in a model of heterogeneous firms, these new theories
have produced new sets of testable predictions regarding the organizational choices made
by firms.
When choosing their global sourcing strategy, firms face two decisions: (1) where to
locate production and (2) the extent of control to exert over production processes. The
level of control depends on headquarter intensity, or how crucial it is for a firm to be
involved in its production process. Headquarter intensity can reflect a firm’s capabilities,
tangibles like special equipment or machinery, or intangibles like R&D knowledge, know
how or workers’ skills. When headquarter intensity is high, the firm’s involvement into
production is important and activities should take place inside the firm’s boundaries.
When headquarter intensity is low, the firm’s contribution to production is marginal
and firms should delegate those activities outside their boundaries. Firms can decide to
locate their production in the North or in the South and the location decision depends
on the fixed and variable costs of each location. Four organizational choices are therefore
possible: to make at home (produce in the North), to outsource at home (rely on suppliers
in the North), to make abroad (do FDI in the South) and to outsource abroad (rely on
suppliers in the South).
The theory predicts a clear sorting of firms into organizational choices depending on
headquarter intensity and productivity: when headquarter intensity is low, control is not
important and outsourcing is the only efficient strategy. High productivity firms outsource to the South, low productivity firms outsource to the North and the least productive firms exit. When headquarter intensity is high, both control and location decisions
matter. Most productive firms produce in the South (FDI), somewhat less productive
firms outsource in the South, firms with even lower productivity acquire components in
the North, among them the more productive integrate and the less productive outsource.
The least productive firms exit.
We start by presenting some stylized facts about firms’ organizational choices. We
show that around 10% of the firms in our sample offshored their core activities, either
within the group or outside the group, and sometimes both within and outside (bisourcing). We also show evidence that the most productive firms offshored their activities
both to the North and to the South, suggesting complex patterns of global sourcing. We
document a trend towards more offshoring within the groups and more to the South.
We then look at the characteristics of firms that choose specific forms of organization.
As expected, more productive firms are more likely to offshore within the group than
outside the group. Firms with lower headquarter intensity are more likely to outsource.
2
However, firms offshoring both within and outside the group are the most productive
in our sample. Headquarter intensity also plays an important role, as firms with higher
capital intensity are less likely to outsource their core activity, either in the North or in
the South. These results are in line with Du et al. (2009) who provide an extension of the
Antràs and Helpman framework introducing the possibility of bi-sourcing, and also with
Yeaple (2003), Grossman, Helpman and Szeidl (2006) and Antràs, Fort and Tintelnot
(2014) who show that selection and strong complementarities between sourcing origins
might explain the observed patterns of geographic dispersion.
Several recent papers have provided empirical tests of these theories (see e.g. Tomiura,
2007; Kohler and Smolka, 2014; see also the discussion in Antràs, 2014). Our aim is to
contribute to this growing literature and to provide additional evidence thanks to rich
survey data at our disposal.
Section 2 describes the rich datasets that we use for our analysis and provides summary statistics. Section 3 discusses some stylized facts . Section 4 introduces our simple
methodology and section 5 shows the results. Section 6 concludes and derives some implications for global sourcing theory.
2
Data
2.1
Description
To provide a proper statistical cover of the importance of international sourcing in the
EU, Eurostat launched a first survey in 2007 coordinated among member states, asking
medium and large companies about various aspects of their surcing activities. The survey
covered their activities between 2001 and 2006. The Danish survey was run by Statistics
Denmark and obtained a very high response rate.1
2.1.1
2007 wave
An important dimension of the survey is that firms are asked about the type of business
function that was sourced. Nine functions were identified: core function, distribution and
logistics, marketing, sales and after sales services (including help desk and call center); ICT
services; administrative and management functions; engineering work and other technical
services; R&D; facility management (cleaning, security, food, etc...); other functions.
While we will mostly be concerned with the core activity, we will also consider some of
the other functions in our analysis.
1
See Statistics Denmark, 2008 for more details.
3
The most important question that we exploit is whether the firm outsourced a specific activity domestically or abroad; additionally, the survey asks whether international
sourcing by business function was established through FDI (defined as insourcing by Eurostat) or through international outsourcing outside the enterprise group. Based on these
answers, we create several categorical variables and divide firms according to the various
categories considered for each function. Our first variable (InOut) only considers the
choice between within or outside the group (we also allow for the possibility that firms
might choose both):
• InOut = 0: no outsourcing
• InOut = 1: only domestic outsourcing
• InOut = 2: international sourcing outside the group
• InOut = 3: international sourcing within the group
• InOut = 4: international sourcing both outside and within the group
An additional question asked about the destination of the sourcing activity, when international sourcing was indicated, for each business function: ”old” EU countries (EU15);
”new” EU countries (EU12); other European countries; China; India; other Asian countries and Oceania; US and Canada; Central America; and Africa.
Based on the answers, we define two areas to proxy for the theoretical notions of North
and South used in the literature: one containing countries of equal level of development
(old EU countries, other European countries and North America) and countries at a
relatively lower level of developement and with lower wages (the other destinations).
We then create an alternative measure of organizational choice (InOutDest = 0) that
integrates the location of the activity in line with the model:
• InOutDest = 0: no outsourcing
• InOutDest = 1: only domestic outsourcing
• InOutDest = 2: international sourcing outside the group in the North
• InOutDest = 3: international sourcing outside the group in the South
• InOutDest = 4: international sourcing within the group in the North
• InOutDest = 5: international sourcing within the group in the South
4
• InOutDest = 6: international sourcing both outside and within the group in the
North
• InOutDest = 7: international sourcing both outside and within the group in the
South
2.1.2
2012 wave
The 2012 wave replicates most of the questions but also clarifies and deepens the concept
of offshoring activity and the implication of the firm in the manufacturing process. One
important difference is that the number of functions was reduced from nine to seven.
”R&D activities” and ”engineering and other technical services” were merged in one
category; and the function facility management was abandoned (as a much larger fraction
of firms reported having outsourced this function domestically compared to the other
functions).
Some of the new questions that were asked are the following:
• which business functions are undertaken by one of your subsidiaries and where?
• for each business function, in which countries are your suppliers located?
• how would you define your company’s activities?
The first two questions were trying to clarify the difference between the flow of activities
to new destinations, and the actual activities performed or delegated by firms in various
locations. The third question was intended to provide additional light on the variety of
tasks performed by firms, with a special interest on subcontracting and factoryless good
producers (FGP’s), i.e. firms involved in the design and development of goods, but not
in the production process itself.
Firms were asked to choose between the following options: manufacturing firm; subcontractor; factoryless good producer (FGP); construction; trade; regular service company; service company with manufacturing production activities; other service company;
other type of company. We can therefore see whether the firm considers its activity in
line with the industry in which it is supposed to operate.
2.2
2.2.1
Summary statistics
2007 wave
All functions
Out of 4,161 firms having answered the survey:
5
• 1,748 firms (42.01%) are not involved in any type of outsourcing (neither domestic
nor foreign)
• 2,158 firms are involved in some domestic outsourcing (51.86%)
• 732 firms are involved in some international outsourcing (17.59%)
• 477 firms are involved in some of both
Core function
Things look a bit different when we consider only the core function (table 1a). International sourcing is relatively more prevalent: 109 firms source only domestically, 338 firms
source internationally and 42 firms source in both. Of those sourcing internationally, an
equal number is sourcing within the group and outside the group (168 firms), while 44
source in both ways. In terms of destination areas, 97 outsourced in the North, 267 in
the South (China being one of the most popular destinations with 111 firms reporting
outsourcing there) and 92 did both.
2.2.2
2012 wave
All functions
Out of 4,461 respondents, 2,851 reported neither domestic nor foreign outsourcing (63.91%),
855 reported only domestic outsourcing (19.17%), 497 only foreign sourcing (11.14%) and
258 reported both (5.78%).
Core function
When looking only at the core function (table 1b), out of 4,454 respondents2 , 380 firms
were sourcing internationally only (8.53%), 157 domestically only (3.52%) and 76 were
doing both (1.71%), while 3,841 firms reported not sourcing their core activity (86.24%).
More firms (222) were reporting sourcing within the group (FDI) than outside the group
(183), while 51 firms reported sourcing both from within and outside. More sourcing was
also taking place in the South than in the North. We can thus observe a clear trend with
more firms offshoring to the South and within the group.3
2
3
Seven firms reported no core activities during that period.
The distribution of InOut is provided in the Appendix (table A1).
6
Main activity
Table 2 shows the distribution of answers according to the industry where the firm is officially classified. We observe obvious differences between sectors, even in manufacturing
only. In the food sector, most firms consider themselves as performing in regular manufacturing activities. In fabricated metals, the largest share of firms consider themselves
as subcontractors. In machinery and equipment, a large share (above 5%) defines their
main activity as FGP’s. Considering the wholesale sector, while the largest portion would
define their activities as trade, some firms consider themselves as manufacturers, others as
FGP’s or involved in design and R&D. The combined share of these three latest categories
is above 20%.
These summary statistics suggest that the borders between manufacturing and services have become blurred. These new facts complement previous work by Bernard and
Fort (2014) - for a small European economy - and also can enrich the previous work
on deindustrialization by Bernard, Smeets and Warzynski (2014). We also observe that
firms are sourcing more of their core economic activities inside the firm (i.e. through
subsidiaries, using FDI) and more to the ”South” (mostly in China and Eastern Europe).
Subsidiaries and suppliers
In our sample, 919 firms report having a subsidiary abroad (20.60%) while 1,277 firms
report having a foreign supplier (28.63%). Firms were also asked for which function they
used these suppliers or subsidiaries, and in which geographic area they were located.
Based on their answers, we can look at the distribution of the preferred sourcing strategy
by type of function and by type of region (table 3). As we can observe, having subsidiaries
in the North is the most common choice among firms sourcing internationally for their
core activities, while the choice of subsidiaries and suppliers in the South is relatively
more important for the other functions.
3
Stylized Facts (preliminary)
In this section we present three stylized about the complex patterns of global sourcing
strategies adopted by firms that we obse. First, a significant share of firms have subsidiaries and suppliers in more than one country; offshored to more than one country for
their core activity. Firms can locate their core activities in several locations, as we can
see in table 4. Around a quarter of all surveyed firms report having at least one foreign
supplier for their core activity. Among those, slightly less than a half have suppliers in
more than one geographic area. In the case of subsidiaries, out of 668 firms having sub7
sidiaries abroad, almost 40% report having subsidiaries in more than one area. For the
offshoring decision, the number of destinations is lower, as the period during which the
move is reported is relatively short. Most firms offshored their core activity to one or two
regions at most. Table 5 shows the most popular areas to offshore, establish subsidiaries
or supplier relationships. While most of the recent moves involved countries in the new
EU member states or Asia, we observe that the most suppliers or subsdidiaries are located
in old member states.
Second, more productive firms are more likely to offshore and to more geographic areas.
Table 6 shows a simple regression between the value added per worker and the number
of geographic areas where the firm is involved. Not surprisingly, more productive firms
are more likely to be have suppliers and subsidiaries in more geographic areas. Regarding
offshoring, it looks like more productive firms offshored to less destiantions.
Thirs, the more distant the country, the less firms offshore only to that country. Table
7 shows that firms are less likely to offshore only to distant areas such as India, China or
other Asian countries, compared to neighboring areas in Europe. Instead, when involved
to these distant areas, they are likely to choose a complex bundle of destinations.
4
Empirical Methodology
We first consider the core activity only. We run a multinomial logit with our two measures
of organizational choice as left hand side variable. The probability of firm i choosing
organizational form j is expressed as:
exp(αj + βj Xi(t−1) )
pijt = P r(yit = j) = P
, k = 1, ..., 8, t = 2006, 2011
exp(αk + βk Xi(t−1) )
(1)
k
where Xi(t−1) is a vector of lagged firm-level variables. In line with the theorey discussed in section 2, we consider a measure of productivity (log of value added per worker),
log of firm size and log of capital intensity as a proxy for headquarter intensity. (Note
that this last measure is firm-specific and not industry-specific)
In a next stage, we consider all functions
exp(αj + βj Xi(t−1) + δf )
pif jt = P r(yif t = j) = P
, k = 1, ..., 8, t = 2006, 2011
exp(αk + βk Xi(t−1) + δf )
k
where δf is a function dummy.
8
(2)
5
5.1
Results
Core activity
We start by discussing the results of our multinomial using InOut as our left hand side
variable and focusing on the core function only (table 8a). We find that both value added
per worker and firm size are more strongly associated to organizational choice where
firms are sourcing within the group. Firms sourcing outside the group are also larger
and more productive than non sourcers. However, the most productive firms are the
bi-sourcers. Capital intensity appears to be negatively related to the choice of offshoring
outside the group. Manufacturing are also much more likely to be involved in some form
of offshoring. These findings appear to be in line with the predictions of the AH model
and its extension to bi-sourcing by Du et al. (2008). We also observe that there is more
domestic outsourcing and more sourcing within the group in the 2nd wave compared to
the 1st one.
In table 8b, we look at our alternative measure of organizational choice: InOutDest
(note: we use the same definitions as used by Eurostat - see table A2 in Appendix). If
we compare firms choosing the same organizational firms but involved in FDI in different
regions, we observe that firms with activities in the South are more productive and larger.
We also find that sourcing within the group in the South has increased over time.
5.2
All functions
In table 9, we include all the other functions in the analysis. The main results described
with the core function extend to the more general setting. We also observe that almost
all functions (except R&D) are more likely to be domestically outsourced compared to
the core function, while the opposite is true for international sourcing. We also see more
international sourcing within the group of ICT and R&D.
5.3
Stock vs. flow
In table 10, we use the answer to the question about the presence of suppliers and subsidiaries for the various functions and to the different destinations rather than the offshoring question.
Panel A shows the results for the core function only. We find again that productivity
is more strongly associated to the choice of a subsidiary compared to supplier; and to
choosing location in the South compared to the North. More productive firms choose
bi-sourcing in the South. More capital intensive and larger firms appear to choose bisourcing, and capital intensity is negatively related to having foreign subsidiaries in the
9
North. Finally, manufacturing firms are more likely to be involved in all sourcing forms,
especially in the case of subsidiaries.
In panel B, we look at all functions. We observe that all functions are less likely to be
allocated to any type of organization form compared to the core activity.
6
Conclusion
In this paper, we analyzed offshoring decisions of Danish firms over the last decade and
related our findings to the offshoring literature. Our results suggest complex offshoring
patterns involving both insourcing and outsourcing, both in the North and in the South.
These decisions appear to be mostly driven by two measures of firm heterogeneity: productivity and capital intensity.
Our analysis at this stage is mostly descriptive. In future work, we would like to
understand better why firms adopt bi-sourcing strategies. Du et al. (2008) suggest an
improved bargaining position with suppliers. However, one can think of alternative explanations. Previous papers have shown that firms might decide to outsource some activities
outside the firm and keep others according to its ability to learn new knowledge (Azoulay,
2004). Another important dimension is the need to keep control over quality, so that
firms might decide to outsource low quality products but keep high quality ones in house.
More research is needed to better understand firms’ complex sourcing decisions.
References
[1] Antràs, P. (2003), ”Firms, Contracts, and Trade Structure”, Quarterly Journal of
Economics 118, 1375-1418.
[2] Antràs, P. (2014), ”Grossman-Hart (1986) Goes Global: Incomplete Contracts, Property Rights, and the International Organization of Production”, Journal of Law,
Economics, and Organization 30 (suppl 1), i118-i175.
[3] Antràs, P., Fort, T. and Tintelnot, F. (2014), ”The Margins of Global Sourcing:
Theory and Evidence from U.S. Firms”, work in progress.
[4] Antràs, P. and Helpman H. (2004), ”Global Sourcing”, Journal of Political Economy
112, 552-80.
[5] Antràs, P. and Yeaple, S. (2015), ”Multinational Firms and the Structure of International Trade”, in Helpman, E., Rogoff, K. and Gopinath, G., Handbook of
International Economics, Vol. IV, pp. 55-130
10
[6] Azoulay, P. (2004), ”Capturing Knowledge within and across Firm Boundaries: Evidence from Clinical Development” American Economic Review 94, 1591-1612.
[7] Bernard, A. and Fort T. (2013), ”Factoryless Goods Producers in the US”, NBER
Working Paper # 19396.
[8] Bernard, A., Smeets, V. and Warzynski, F. (2014), ”Rethinking Deindustrialization”,
working paper, Aarhus University.
[9] Du, J., Lu, Y. and Tao, Z. (2009), ”Bi-sourcing in the global economy”, Journal of
International Economics 77, 215-222.
[10] Grosman S. and Hart O. (1986), ”The Costs and Benefits of Ownership: A Theory
of Vertical and Lateral Integration”, Journal of Political Economy 94, 691-719.
[11] Grossman, G., Helpman, E. and Szeidl, A: (2006), ”Optimal integration strategies
for the multinational firm”, Journal of International Economics 70, 216-238.
[12] Kohler, W. and Smolka, M. (2014), ”Global Sourcing and Firm Selection”, Economics
Letters 124, 411-415.
[13] Statistics Denmark (2008), ”International Sourcing: Moving Business Functions
Abroad”.
[14] Tomiura, E. (2007), ”Foreign outsourcing, exporting, and FDI: A productivity comparison at the fim level”, Journal of International Economics 72, 113-127.
[15] Williamson, Oliver (1975), ”Markets and Hierarchies: Analysis and Antitrust Implications”, New York, Free Press.
[16] Yeaple, S. (2003), ”The complex integration strategies of multinationals and cross
country dependencies in the structure of foreign direct investment”, Journal of International Economics 60, 293-314.
11
Table 1: Summary statistics - core function
2007
2012
Domestic sourcing
109
2.62% 157
3.52%
International sourcing
338
8.12% 380
8.53%
Both
42
1.01% 76
1.71%
Not sourced
3,672 88.25% 3,841 86.24%
Total
4,161
4,454
International sourcing
- to enterprises within the group
- to enterprises outside the group
- both
International sourcing
- in the North
- in the South
- both
12
168
168
44
380
44.21% 222
44.21% 183
11.58% 51
456
48.68%
40.13%
11.18%
99
195
86
380
26.05% 97
51.32% 267
22.63% 92
456
21.27%
58.55%
20.17%
13
71
62
46
25
28
10
294
6.59%
Subcontr.
=2
7
3.23%
26
7.62%
129
42.43%
Trade
Regular
=5
service=6
9
4.15%
5
1.47%
4
1.32%
254
62.72%
51
20.56%
67
34.89%
123
555
2.76% 12.44%
FGP
=3
4
1.84%
18
5.28%
9
2.96%
30
7.41%
15
4.40%
7
2.30%
21
5.18%
36
14.52%
15
7.81%
177
3.97%
Design, R&D,...
service=7
71: also includes technical testing and analysis
Regular
manu=1
Food
184
84.79%
Machinery and
254
equipment
74.49%
Fabricated metals
127
41.78%
Wholesale
33
8.15%
Computer programming, 10
consulting,...
4.03%
Architectural and
6
engineering activities; ... 3.12%
All firms
1,373
30.78%
Table 2: How do firms define their main activity?
Other
Total
services=8
217
94.01%
341
93.26%
304
90.79%
405
83.46%
116
248
46.77%
85.89%
49
192
25.52%
71.3%
1,030
4,461
23.09%
79.62%
14
Table 3: Location and choice between subsidiaries and suppliers by function
15
Table 4: Number of geographic locations for sourcing, core function
16
Table 5: Location decisions for sourcing, core function
- # of firms reporting having offshored their core activity by geographic area
2007
2012
EU-15
109
127
EU-12
205
210
Other European countries
46
70
China
126
124
India
30
63
Other Asian countries and Oceania 60
62
US and Canada
25
19
# firms offshoring
380
456
9.13% 10.24%
- # firms with subsidiaries/suppliers by geographic area (2012 only)
Subsidiaries Suppliers
EU-15
405
684
EU-12
189
396
Other European countries
168
245
China
119
321
India
45
133
Other Asian countries and Oceania 107
176
US and Canada
102
170
# firms reporting subsidiaries
668
1,080
or suppliers abroad
14.97%
24.21%
17
Table 6: Productivity and number of geographic areas
18
Table 7a: Distance and sourcing
19
Table 7b: Distance and sourcing (ctd.)
Table 8a: The Determinants of Sourcing Decision. Multinomial logit analysis
20
21
Table 8b: The Determinants of Sourcing Decision. Multinomial logit analysis
Table 9a: The Determinants of Sourcing Decision by Function. Multinomial logit analysis
22
23
Table 9b: The Determinants of Sourcing Decision by Function. Multinomial logit analysis
24
Table 10: Choice between subsidiaries and suppliers by function
Table A1: Distribution of InOut
Table A2: Eurostat definition of InOut
25