Masters Dissertation
Evaluating the effects of agglomerations on
location selection of MNEs based on different
equity-based entry modes
Student: Lars Schumann
Student no: NCL: B8057137 GRO: S3903362
Supervisors: Newcastle University: Prof. Tom McGovern
University of Groningen: Prof. Gary Ge
Programme: NCL: MSc Advanced International Business Management
GRO: MSc International Business & Management
Date of Submission: December 2nd, 2019
1
Abstract
This paper identified a gap in existing research as there have been no comparative studies
on the agglomeration levels effect on location decisions for various FDI. For now, there is
research on agglomerations impact on each JV, M&A, and Greenfield location decisions with
smaller firms being more affected than larger ones. However, there is no such research
including all these equity-based FDI methods in order to see which one is affected most. There
is a need for both public sectors, whose policies enable agglomerations, and business
understanding to find out just that: How does agglomeration affect FDI location choice
differently for various equity-based entry modes (Joint ventures, Mergers & Acquisitions,
Greenfields) of MNEs? Initially, the agglomeration levels of each state are being calculated for
both industry and country of origin agglomeration. Thereafter a compelling dataset of FDI from
Japan into the USA was obtained including various independent and control variables –
agglomeration level, knowledge intensity, deal type, subsidiary network and experience of
firms – that might also have an impact on strategic decision-making regarding location choices
of firms. Data is going to be drawn from the Bureau van Dijk resources Zephyr and Orbis. The
findings of the binary regression performed led to conclude that industrial agglomeration has a
bigger effect than country of origin agglomeration and that M&As are most significantly
affected by agglomeration, followed by Greenfields. JVs location choices are not found to be
influenced by industry agglomeration but strongest affected by country of origin agglomeration.
Moreover, this study motivates further researches in the field of agglomerations. For instance,
strategic relocation does not necessarily come with FDI but might as well be carried out within
one country.
Keywords: FDI, Joint Venture, Mergers and Acquisitions, Greenfield, agglomeration, location
2
Acknowledgements
Firstly, I would like to express my appreciation for my supervisors Gary Ge and Tom
McGovern. Their invaluable constructive feedback and guidance enabled a challenging but
rewarding experience in writing my thesis. Further, I would like to thank all my friends and
colleagues who I met as part of the AIBM course, both in Newcastle University, and University
of Groningen, for making the last 18 months an amazing experience. Lastly, I must thank my
mum and my girlfriend to offering me support wherever it was needed along my journey during
3
Table of Content
Abstract ... 1 List of Figures ... 4 List of Tables ... 4 List of Abbreviations ... 4 1 Introduction ... 5 2 Literature Review ... 112.1 Push and Pull Factors of Agglomerations ... 11
2.2 Industry and Country of Origin Agglomeration ... 14
2.3 Hypotheses ... 17 3 Research Methodology ... 23 3.1 Research Philosophy ... 23 3.2 Data ... 24 3.3 Sample ... 25 3.4 Variables ... 26 3.4.1 Dependent Variable... 26 3.4.2 Independent Variables... 27 3.4.3 Control Variables ... 29 4 Results ... 33 4.1 Preliminary tests ... 33 4.2 Descriptive Statistics ... 34 4.3 Correlations ... 38 4.4 Regression results ... 38 4.5 Robustness checks ... 40 5 Discussion ... 43 6 Conclusion ... 48 6.1 Theoretical implications ... 48 6.2 Practical implications ... 50
6.3 Limitations & Future Research ... 51
References ... 53
Appendices ... 62
Appendix 1: List of states and agglomeration levels ... 62
Appendix 2: Preliminary tests ... 63
4
List of Figures
Figure 1 - Conceptual model……….….………17
Figure 2 - Distribution of Greenfield investments in the USA……….……….……….28
Figure 3 - Distribution of M&As across the USA……….……….…28
Figure 4 - Distribution of JVs across the USA………..28
List of Tables Table 1 - Industry and Country of Origin Agglomeration………..……10
Table 2 - Variable description………..………..24
Table 3 - Investments into agglomerations and not into agglomerations………27
Table 4 - Descriptive Statistics and Correlations………...28
Table 5 - Regression - Location choice of subsidiaries……….30
Table 6 - Outlier test………..32
List of Abbreviations
COO Country of Origin Agglomeration
FDI Foreign Direct Investment
IT Internet Technology
JV Joint Venture
KBV Knowledge-based view
5
1 Introduction
There is a vast literature trying to link inward foreign direct investment (FDI) to
subsequent agglomeration (Wheeler & Mody, 1992; Head et al., 1995, 1999; He, 2002; Campos
& Kinoshita, 2003; Crozet et al., 2004; Du et al., 2008; Alcacer & Chung, 2014; etc.), however
there is only a little evidence that this creates a sustainable agglomeration as the dependence on
the MNE is too significant. The decision-making power lies solely with the MNE that the
technology transfer between foreign and domestic industries is limited (De Propris, 2001).
However, the attractiveness of agglomerations as an FDI destination is not affected by that lack
of sustainability. Many studies in the past found a strong agglomeration effect for FDI (Wheeler
& Mody, 1992; Head et al., 1995, 1999; He, 2002; Campos & Kinoshita, 2003; Crozet et al.,
2004; Du et al., 2008), however, Guimaraes and Figueiredo’s research (2000) actually found
contrary results and an insignificant negative correlation that indicated a dispersion of FDI. The
paper at hand makes the case that the different results could be up to the entry mode choice that
was looked at during these studies. As there has never been a comparative study amongst
different entry modes and the extends of effects varied between the studies mentioned
previously. Firms entering a new market choose their preferred method of entering based on
the objectives they have. These objectives are different for different entry modes and different
objectives cause different patterns in decision making that can be linked also to location
decisions. While, other predictors such as firm size, experience and knowledge intensity of the
firm have been assessed and controlled for in research before (e.g. Head et al., 1995; Li &
Bathelt, 2014), entry mode was not. Therefore, in a bid to clarify the different extents of the
effects of agglomeration on the locational choice of MNEs this paper addresses the question:
6
Relevant entry modes to assess include Joint Ventures (JVs), Mergers and Acquisitions
(M&As), as well as Greenfield investments as all of those, represent equity-based entry modes.
Leading to the entering firm to be exposed to the foreign market directly. JVs and M&As
emerge as possible strategies (Franco et al., 2010) to gain resources, knowledge and customers
amongst other things in the relative environment and then have the potential to use the learnings
elsewhere. The two governance forms mentioned, JVs and M&As, have their distinctions that
cause different effects in knowledge transfer: In the wake of an acquisition one company
purchases majority stake in another company to assume control; a JV is a legally distinct
business that is owned by two or more parent firms (Beamish & Lupton, 2009). While both
forms of investment ensure exposure to the local market, the extent of intertwining between
organisations differ. During a JV both sides own a (often somewhat equal) share of a newly
found enterprise that is there to enhance best practice and technology exchange. However, both
companies have the interest to gain a maximum of new knowledge while not giving away too
much of their expertise to weaken their own position in the competition. Given the differences
outlined before, both forms of FDI have in common that they both are equity-based entry modes
into a foreign market. Further to these investment options, Greenfield investments are gaining
relevance with the softening of regulations amongst more regulated markets including China
(OECD, 2019a). Equity-based FDI creates a knowledge and ownership advantage for firms that
cannot be generated by non-equity based FDI such as licensing which is therefore excluded
from this analysis.
FDI location choices by companies are shaped by two contrary forces: entrenchment
benefits and opportunity costs (Alcacer et al., 2013). Entrenchment benefits occur when a firm
is an early mover into a market. It gets the benefit of local learnings before competitors have
the opportunity to gain those (Alcacer et al., 2013). On the other hand, opportunity costs are a
concept of scarcity of capital: a firm has multiple markets to choose from to enter into but only
7
might have to postpone an investment into another. Giving competitors in those geographic
markets valuable lead time to improve their position for competition (Alcacer et al., 2013).
In order to capture entrenchment benefits, often in the form of spillovers, the subsidiaries
proximity to hotspots of innovation and production is necessary. That is where agglomeration
comes into the picture. Agglomeration of industries or companies can be defined as follows:
Agglomerations are “[…] geographic concentrations of industries related by knowledge, skills,
inputs and/or other linkages” (Delgado & Porter, 2015). These agglomerations have different
extent and therefore appeal to different motives behind FDI. Moreover, there are two different
kinds of agglomerations that have to be distinguished for this research. First, industry
agglomerations as they are defined by Delgado & Porter (2015). Second, country of origin
agglomerations suggesting foreign investors often locate near other FDI firms sharing the same
country of origin (Chang & Park, 2005; Chung & Alcacer, 2002;Head & Ries, 1996).
Firms in an agglomeration are characterized by high specialization and complementarity
which creates a dynamic knowledge creation (learning and innovation) and transfer (diffusion
and synergies). Collective learning processes are creating innovation and competitiveness as
well as improving productivity not only for the high-tech sector but also for other industries
(De Propris & Driffield, 2006). For instance, the ceramic tile industrial district in Sassuolo
(Italy) accounts for one-third of the sectors world export (Menghinello, 2003). As the
agglomeration grows, the extent of vertical and horizontal product differentiation increases.
Consequently, the agglomeration becomes a centre of collected competences across various
related industries, and stages of production. These localised centres of knowledge can be very
attractive to outside firms, and thus attractive to inward FDI decision-makers (De Propris &
Driffield, 2006). The extent of agglomerations attractiveness to FDI is measured in multiple
different studies with different focusses around the subject. These studies will be looked at in
8
Evidently, agglomerations have effects on companies’ strategic behaviour and decision
making. That includes the location decisions they make when investing abroad. Here the
benefits and risks of agglomerations play a central role in evaluating to what extent the
knowledge and expertise, as well as the efficient infrastructure, outweighs potential
appropriation risks. Multiple literatures all tested agglomeration effects on investment decisions
for M&As (e.g. Basile, 2004; Böckermann & Letho, 2006), JVs (e.g. Tuan & Ng, 2003; Zhao
& Zhu, 1998) and Greenfield investments (e.g. Head et al., 1995; Shaver & Flyer, 2000; Alcacer
& Chung, 2007) showed similar results. That will be assessed in more detail during the literature
review. Agglomeration was significant in attracting firms’ investment to a region with small
companies being relatively more affected by it than larger firms. In contrary Guimares and
Figueiredos (2000) found an insignificant negative correlation that indicated a dispersion of
FDI. However, with all the information provided by these researches, there is, to my best
knowledge, no work that evaluates and compares the strength of the effect for each of different
entry modes which could explain the differing results as well as effect strengths. This is where
the paper at hand is going to contribute to existing research in comparing the attractivity of
agglomerations for various types of FDI. Namely, JV, M&A and Greenfields that have been
used in previous research but were never analysed and compared in one study.
In order to find out how the entry modes can be differently affected in their location
choice, it is important to understand varying objectives behind the choice of entry. JVs are often
driven by the access to resources that otherwise would not be available for the companies
agreeing on cooperation. Namely, intellectual property and exploring new capabilities without
giving away knowledge that was not agreed on with the respective partner (Deloitte, 2016).
Moreover, risk mitigation in terms of a separation between the JV and both partners as well as
regulatory approval, such as FDI restrictions, played a role in the motivations of setting up a JV
(Deloitte, 2016). Many scholars consider that JVs enable easier access to the firms’
9
scale, develop new products faster and more cheaply than could be done by either of the
parenting firms acting alone or through acquisition (Beamish & Lupton, 2009; Choi &
Contractor, 2016). Complementary resources and technologies between the two partners play a
major role here as JVs are often contractually limited to a certain time in order to achieve a
certain goal that is in many cases related to technology transfer (Johnson & Turner, 2003). This
indicates a focus on the partners rather than the environment in terms of agglomerations.
However, in setting up a JV the staff is build up from both internal sources of both partners as
well as external skilled personal that helps to achieve said objectives. Therefore, the often
skilled and educated workers within agglomerated areas might be more relevant to gaining
technological advantage – and thereby influencing location choices – than first expected.
In contrary, for M&As the local labour pools are not as relevant. As the company, that is
either acquired, already exists and holds enough labour force as well as distribution channels,
equipment, supplier networks and customer base to operate at near full capacity (Johnson &
Turner, 2003). Therefore, at first, M&As do not require to develop new capacity and are not
dependent on outside benefits like JVs might be. However, given the fact that more companies
than not located in agglomerated areas – agglomeration, in the end, being caused by just that, a
high number of companies in one field – the location choice made by acquiring firms is
predetermined by the location of potential targets for a takeover. Indicating a potential tendency
to co-locate with other companies. While the choice of co-location for JVs might be free and
evaluated after taking cost and benefits into account. The decision of location choice in M&As
has to be made, as one factor of analysis in the target firm. Next to other reasons like strategic
fit, cultural fit, size of investment, economic benefit, consumer base and many others.
Lastly, for Greenfield investments, the location decision is the least restricted for the firm
pursuing FDI. The location choice can be made only based on environmental attributes such as
the extent of competition, supplier networks, availability of skilled labour et cetera. In
10
and involvement in a local environment (Johnson & Turner, 2003). In order to adapt to the new
environment spillovers from local knowledge, proximity to suppliers and customers and other
supporting industry is an advantage especially for firms with little to no experience in
international business either in general or in that specific region.
In summary, while the effects of agglomeration on M&As and Greenfield investments
are more explicit, JVs are rather dormant and secondary in strategic decision making. That does
not mean that the effects of agglomeration on location choice as it can be measured is
recognising the difference in drivers for co-location. In this research location decisions of
Japanese firms in cross-border FDI into the United States of America (USA) are being analysed.
The data consists of various investments in the three relevant equity-based entry modes of JVs,
M&As and Greenfield investments and is collected using the databases Orbis (Greenfield
investments) and Zephyr (M&A and JV), both provided by Bureau van Dijk (2019), to create a
compelling dataset of 326 observations between 2010 and 2019. It has been found that M&As,
as well as Greenfield investments locational choice in terms of a state, is related to the industry
agglomeration level of the state. While JVs are not attracted by higher industry agglomeration
of states, country of origin agglomeration proved to be a centripetal force. These results show
that the different entry modes have an influence on the effect strengths of agglomeration on the
location choice. Indicating that discrepancies in prior researches could be down to the different
entry modes assessed. Future research can gain insight for what differences entry modes have
on effect strengths of agglomeration in location choice by building on prior research and the
findings of this paper.
The following section is reviewing relevant literature in order to come up with hypotheses
that will be tested. Subsequently, the methods necessary to test the hypotheses will be
11
2 Literature Review
2.1 Push and Pull Factors of Agglomerations
Generally, firms select their branch location according to the expected profits, with
locations that promise high returns clearly in favour (Basile, 2004). Relevant determinants
include transportation cost, local market size, supplier availability amongst incentives like tax
breaks. There are multiple motivations for companies to establish subsidiaries within
agglomerated areas such as agglomerations. Following a cumulative causation approach, it is
suggested that firms tend to locate in areas where other firms are already established (Basile,
2004). Overall, there are two forces that must be considered when analysing the attractiveness
of agglomerations. On the one hand, there are centripetal forces that motivate MNEs to invest
in areas that already have other MNEs. This is down to the fact that foreign firms generate
spillovers in a host country environment that carry potential value for subsequent FDI; namely
specialized supplies and reduced cost of information (Shaver et al., 1997). Moreover, there are
multiple reasons highlighted in various literature: a significant market for local products and
services; reduced costs for transport and the supply chain; opportunities to reach more
customers through the high market concentration; more feasible access to resources,
opportunities for new companies to extend in this environment; increased motivation through
a competitive environment, greater potential cooperation between members of the
agglomeration, high degree of specialization in both services and products; the proximity of
firms of the same or similar industries allow for an exchange of knowledge and ideas through
direct contact and free movement of labour – so called spillovers; the number of competitive
firms in one area will result in a large pool of skilled workers; the direct contact with people in
the same field reduce risks and durations of the innovation process because of information
transfer between partners, companies, clients and research institutions (Malmberg et al., 1996),
12
& Swann, 1998), (Sölvell et al., 2003), (Krugman, 1991), (Malmberg & Maskell, 2001), (Boja,
2011). Further to these reasons, it is important to note that firms making their location decision
in a foreign country might be subject to information issues. New investors follow prior
decisions of other companies as an unconventional location could backfire not only financially
but also in a reputational sense (DeCoster & Strange 1993). Essentially leading to a mimicking
of other firms‘ location decisions (Head et al., 1999) which then builds further traction towards
the existing agglomeration.
On the other hand, Fujita et al. (1999) have identified forces that come into play when a
certain level of agglomeration is achieved. Labour cost can be an example of that as it goes up
when the demand for jobs goes up through an influx of incoming investment. This centrifugal
force could be observed in the south-east of China as labour and land prices in Guangdong went
up so dramatically that Foxconn, a supplier for Apple, had to relocate (Pan, 2012).
Further to these centripetal forces, other research indicated further benefits for
subsidiaries and the companies as a whole. Subsidiaries within agglomerations are found to be
more embedded, internationally orientated and autonomous compared to subsidiaries that are
located outside of agglomerations (Birkinshaw & Hood, 2000). Forst (2001) proved that local
context creates an important base of knowledge to draw from for innovative processes of
subsidiaries. While it has been found that subsidiaries in innovative environments are more
likely to generate knowledge benefits over other subsidiaries (Cantwell & Mudambi, 2005). In
this work, subsidiaries are seen to be able to create business advantages, as indicated before,
but also to contribute to the local environment they are established in (Birkinshaw & Sölvell,
2000; Brandstetter, 2006). Considering this, the motivations for firms to invest in
agglomerations should be clear. However, the centrifugal forces emphasized before lead to a
weighing up of the pros and cons of investing in an agglomeration. Companies planning their
FDI strategy are weighting up these centrifugal and centripetal forces to evaluate a location and
13
different entry modes, there are usually different objectives. These propositions are based on
the knowledge-based view (KBV) which is an extension of the resource-based view that
highlights knowledge as a strategic resource that can generate increasing returns and does not
depreciate (Curado, 2006). Furthermore, it can be attributed to the FDI that is motivated by
non-marketable asset seeking from firms. The growth of Silicon Valley as a destination of FDI
was driven by firms behaving accordingly to the KBV to give an example (Almeida, 1996).
However, different kinds of agglomerations might also attract different motives for FDI. It is
expected that urban agglomerations or agglomerations around a widely populated area attract
both efficiency as well as market seeking motives in MNEs. Industrial belts like the “Die Region
der Weltmarktführer” (“The region of the world market leaders”) (Zhang & Warken, 2019) in Heilbronn-Franken, Germany, (GGS, 2019) and innovative agglomerations’ appeal to motives
that concern efficiency and knowledge seeking. Especially, innovative agglomerations with
firms of multiple sectors can be interesting for MNEs as many of them are diversifying and
investing in building a business in different sectors such as Google, but also automotive
companies that now own banks, IT platforms, software centres and many more (e.g. Daimler:
Mercedes Benz Cars (incl. Smart), Mercedes Benz Vans, Mercedes Benz Trucks, Mercedes
Benz Financial Services, Mercedes Benz Bank, Car2Go, et cetera). Essentially any MNE that
is active in more than one sector or industry.
Traditionally, the literature expected that MNEs cause agglomerations by investing in a
region and pulling their suppliers and other supporting industry with them which then creates
an industry agglomeration that creates spillovers for the region and other firms (De Propris &
Driffield, 2006). Much like what happened in Alabama as the state prepared everything in order
to achieve an attractive environment for the Mercedes Benz site in Tuscaloosa. However, De
Propris and Driffield (2006) propose that this chain of events is eventually motivated by a
pre-existing agglomeration that, then, in turn, makes the MNE invest and enrich that agglomeration
14
local firms and also profit from the knowledge and capabilities held in that environment. Given
that MNEs usually have a vast choice of regions and locations to choose from, considering the
entrenchment benefits and opportunity costs, De Propris and Driffield (2006) point appears to
be valid, as the company attempts to gain the maximum benefit from their location decision.
That benefit can best be realized in an environment that is already existing and proven to be
profitable as an agglomerated area of companies and supporting institutions.
2.2 Industry and Country of Origin Agglomeration
The key term of this research – agglomerations – can be differentiated into two sections.
On the one hand, there is industry agglomeration and on the other hand, there is the country of
origin agglomeration. While earlier literature puts an emphasis on industry agglomeration
(Krugman, 1991; Marshall 1920) only focussing on this one type is not representing the full
picture of the effect that agglomeration can have on the location decisions made by MNEs when
investing in a foreign country. Understanding this shortcoming, more recent works included the
country of origin agglomeration as a factor (Chung & Alcacer, 2002; Tan & Meyer, 2011;
Shaver & Flyer, 2000). Clearly, both kinds of agglomerations have their differences in making
a location interesting for a potential FDI involvement. First, Industry Agglomeration in terms
of co-location with companies in the same industry can help the entrant gain benefits concerned
with the field they are operating in such as industry forecasts, and information on local customer
and supplier behaviours (Mariotti & Piscitello, 1995). This knowledge is commonly of a tacit
nature and therefore requires personal interaction to be exchanged (Polanyi, 1962). The
exchange can be enhanced by geographic proximity as an interaction between individuals is
facilitated. Local participation and networks enable firms to follow new trends and technologies
while also reacting to competitors’ actions (Porter, 1998; Almeida & Kogut, 1997). More than
only providing knowledge on customers and suppliers, also resources such as labour force and
15
contrary to the benefits, industry co-location can also be linked to higher cost for the entrant as
he is facing increased competition not only for potentially scare resources but also the
consumers in the market and region. An issue that intensifies with every company that makes
the decision to move in the same area (Folta et al., 2006). Moreover, prior researches such as
Chung and Alcacer (2002) have found that agglomeration might cause knowledge expropriation
by competitors in the same region – causing especially larger and technologically advanced
firms to avoid co-location in a bid to secure their competitive advantage (Shaver & Flyer, 2000).
Second, County of origin agglomeration is related to different benefits than industry
agglomeration that stem from the, often, different fields the companies operate in. Firms that
co-locate in a country of origin agglomeration do this for other motivations than in the case of
an industry agglomeration. For instance, newcomers to a region might find it difficult to trust
local partners due to an increased vulnerability of not understanding the market equally well
(Tsui-Auch & Möllering, 2010). However, when co-locating with firms that have a similar
background these parallels help covering the distance to a foreign environment and increase
trust between FDI compatriots (Table1) (Tan & Meyer, 2011). This comes down to the fact that
individuals often form stronger ties with their same kin (Manev & Stevenson, 2001; Marsden,
1990). Another reason might be a common history of the co-locating firms. Due to common
geographic background firms faced similar challenges in the past such as economic downturns
as well as they could have worked together before. According to Tan & Meyer (2011)
commonalities between expatriates of foreign firms strengthen the bonds between firms as these
individuals socialize with each other in activities like family events. Not only do these ties
provide a potential for exchange of information about the foreign environment but also potential
new business opportunities (Linehan, 2000).
Clearly, the different kinds of agglomerations influence the attractiveness of an
agglomeration according to the entry mode chosen. Greenfield investments of companies with
16
market than others. While an M&A takes over local employees that already have knowledge of
local proceedings and consumers so that they can hit the ground running and face competition
immediately – making industry agglomerations more relevant for them. Lastly, a JV faces
similar circumstances as the M&A as it is a collaboration between two or more firms that are
set up only to achieve predetermined objectives rather than setting itself up for a future in the
market. Making the benefits of an industry agglomeration more appealing for this sort of entry.
Table 1: Industry and Country of Origin Agglomeration (adapted from Tan & Meyer, 2011) Industry agglomeration Country of origin agglomeration Inter firm relationships − Often in direct
competition − Contractual-based
cooperation
− Normally not in direct competition
− Benefit from a higher level of trust
Benefits − Access to local, industry-specific knowledge & resources (such as
specialized labour/inputs, partners, customers and infrastructure
− Access to local market knowledge
− Access to local, home-country-specific resources
− Easier to gain legitimacy
Cost − Competition for scarce
17
2.3 Hypotheses
The following section is concerned with building up the hypotheses that will subsequently
be tested with quantitative data. As mentioned previously, JVs are often established in order to
gain learnings from national or international partners and the local environment. Therefore, it
can be expected that FDI through JV is positively related to location choice towards
agglomerated areas for several reasons.
First, the benefits of agglomerated areas with many businesses and highly skilled and
experienced personnel attract FDI in the form of JVs. Tuan & Ng (2003) found a positive
significant effect of agglomeration on the location decisions of firms in JVs in China confirming
that claim. Especially smaller firms were found to be more responsive to agglomeration
benefits. Second, agglomerations and effects that are favoured through agglomeration such as
skill intension and market concentration are also considered as positive drivers of location
selection of JVs (Zhao & Zhu, 1998). Those factors, market concentration and skill intension,
are not only addressing knowledge-seeking aspects of FDI but also efficiency and market
seeking motives. A JV is, at times, just used for the gain in knowledge, as presented before.
However, it accounts also as a relatively low-risk option to enter a new market as the financial
exposure of collaboration is always going to be lower than building a company by yourself or
buying into an existing company. In that sense the efficiency bonus of being located in an
agglomerated area can be a further pull in engaging in a JV next to the general intentions with
the partner as distribution channels, suppliers and other supporting companies can be explored
for operations in the future. This again is especially true for smaller companies with less
experience in the international business arena and the target country.
Based on the expectations of JVs being created for knowledge exchange the assumption
of proximity as a catalyst for learning. As well as, the results of prior researches (Tuan & Ng,
2003), that mentioned positive relations of the agglomeration effect on JVs, the following
18
H1a: FDI through JV is positively related in its location choice towards agglomerated areas.
More than for JVs the location selection of M&As is predetermined by the selection of
the firm that is taken over. Nevertheless, through the fact that there are more firms in
agglomerated areas than not, the likelihood of one of these to be taken-over is higher. Therefore,
a positive effect of agglomeration on M&A location selection is expected as indicated by Zhang
et al. (2012) from their research into Indian and Chinese M&As in Europe.
Moreover, Brakman et al. (2015) and Guadalupe et al. (2012) found that cross border
M&As in the US tend to concentrate more around agglomerations in comparison to national
M&As. Indicating the effect of liability of foreignness (LOF) that pushes foreign firms to
co-locate with firms from the same country to overcome their disadvantage of higher entry costs.
That potentially puts forward a more significant effect for the country of origin agglomerations.
Nevertheless, also industry agglomerations can assist companies in adjusting to a new
environment by enabling easier connections to those companies around and supporting
networks that already exist thanks to competitors and the target firm in that field and area. Even
if the firm acquires local expertise, it still seeks proximity to other firms in order to optimize
local learning and overcome the LOF as soon as possible. Moreover, efficiency and markets
are also highly relevant here for the investing firm as, for instance, transportation cost can be
lower and lead times quicker if supporting industries such as suppliers are proximate rather than
on the other end of the country. Furthermore, larger agglomerations are found to be more
productive in comparison to smaller ones (Brakman et al., 2015). This also includes the number
of workers is found to be higher in larger agglomerations increasing the potential firm
productivity (Combes et al., 2008) which is key subsequently to an entry into a foreign market.
Especially, considering the disadvantage foreign firms must overcome in comparison to local
competition. The cumulative causation approach, which argues for firms settling near other
firms, (Basile, 2004) implies that there are more potential targets for M&As located within
19
are more firms in those areas. Both the need to overcome the LOF and to improve the
embedding of knowledge and transfer learnings from the new environment and to internalize
in the organization as well as the efficiency benefits of larger agglomerations, leading to the
following:
H1b: FDI through M&A is positively related in its location choice towards agglomerated areas.
As for Greenfield investments, a company building a new entity in a foreign country,
there are fewer restrictive forces in terms of their specific location selection. According to Head
et al. (1995), Shaver and Flyer (2000) and Alcacer and Chung (2007; 2014) the exclusion of acquisitions lead to a sample of FDI that is least constrained in the location decisions. Therefore,
the decisions made by the companies that follow this stream of investment make them sorely
on their own strategical interest. Here the centripetal and centrifugal forces may have the
strongest effect as companies might fear aiding competitors when co-locating (Shaver & Flyer,
2000). LOF is yet again a determinant that has to be overcome. To do so co-location into an
agglomerated area is a tool that can be applied by the strategic decision makers as productivity
is increased (Brakman et al., 2015) along with other benefits mentioned before. Moreover, the
same research found that smaller firms have a bigger interest in co-locating and exploiting the
advantages of agglomeration than their larger counterparts (MNEs) (Shaver & Flyer, 2000). As
can be seen from the results of previous research, the type of firm - in terms of its age and
industry - makes a difference in their co-location decisions (Shaver & Flyer, 2000). Based on
what is said previously, a positive effect of agglomerations’ impact on the decision making of
companies that choose Greenfield investments is expected in this research. Taking into consideration varying results and differences related to companies’ size. Leading to the
following hypothesis:
20
In literature, most tested agglomeration effects on investment decisions for M&As
(Brakman et al., 2015; Böckermann & Letho, 2006), JVs (Tuan & Ng, 2003; Zhao & Zhu,
1998) and Greenfield (Head et al., 1995; Shaver & Flyer, 2000, Alcacer & Chung, 2007) showed similar results. Agglomeration was significant in attracting firms’ investment to a
region with small companies being relatively more affected by it than bigger ones. Given the
different levels of freedom in the location selection, different effect strengths – and even
different effects such as the dispersion found by Guimaraes and Figueiredo (2000) – can be
expected when measuring the effects of agglomeration on the location choices. Therefore, this
part is concerned with reviewing the literature on which of the effects is expected to be the
strongest.
Zhang et al. (2012) found that the relation of M&As from India and China towards Europe
being largely attracted towards agglomerated areas such as cities. However, the expected
outcome of Greenfield investments showed a greater effect of agglomerations on the location
choice, as the investment here is not predetermined by the organisation that is being taken over.
However, this expectation is flawed as it rests on a sample that cannot be differentiated by entry
mode and a share of investments that is calculated by an average of overall investments. This
result, therefore, must be viewed critically. Admittedly, Zhang et al. (2012) made a point
insisting that the freer location decisions by Greenfield investments are expected to be more
attracted by agglomeration than M&As. That ignores the fact that there are more companies
existing in agglomerated areas than there are companies in non-agglomerated areas. Leading to
the expectation that mergers are more likely to occur in agglomerated areas than not, based on
the fact that there are more potential targets for the investing firm in addition to the regular
centripetal forces that apply to all entrants. Despite, for M&As local labour is not as relevant as
the company that is acquired already employs enough or near enough employees to operate
near-maximum capacity (Johnson & Turner, 2003). The existing labour force, as well as
21
major role in the investment decision of the acquiring firm and therefore agglomeration has a
stronger effect – especially when focussing on efficiency – here.
In line with what is said before, the lack of local expertise that comes with a Greenfield
investment sees the biggest need for investing nearer the knowledge that is relevant for the
company entering a market. M&As are bound to have a similar requirement of embeddedness
to overcome the LOF, even if the acquiring firm gains local knowledge and distribution
channels through the deal itself end the personnel of the target company. The main difference
here is that the existing companies that are available as potential targets for merging or takeover
happen to be located more frequently than not in agglomerated areas. Following a cumulative
causation approach which implies that companies tend to locate where others are already
established (Basile, 2004) - a principle on that agglomeration theories are built. Based on the
assumption of agglomeration which describes areas with a high propensity of businesses – in
fact, higher propensity than other areas – this means that the majority of companies in a country
is to be located within an agglomerated area. Indicating that agglomerations might have a bigger
impact on the location selection for M&As in comparison to Greenfield investments. Moreover,
M&A investments tend to prefer competitive industrial relations and avoid more rural areas that
are at times connected with higher unemployment and a worse quality of life (Basile, 2004).
Greenfield investments, on the other hand, might, especially for more low-technology
industries, be more attracted by that low employment and potentially cheaper workforce in an
efficiency-seeking setting. Providing the opportunity to enter a market while saving on
production and transportation cost in comparison to the firm’s headquarter.
Considering the lack of information when entering a new market and a new culture
without any local partner, such as in a JV, and without taking over local knowledge, the
proximity to knowledge hubs, skilled labour pools and a functioning distribution network
22
Lastly, a JV between two companies is often formed for the reason of knowledge
exchange between the partners and joining their efforts in one field of interest. Here proximity
to other firms – potentially rivals – appears as more of a centrifugal force than a centripetal one
on many occasions. Therefore, the following hypothesis is formulated:
H2: Agglomeration is expected to have the strongest effect on M&As over Greenfields and the weakest effect on JVs.
The hypotheses that are developed based on these findings are summarised and visually
presented in the following conceptual model (Figure 1).
23
3 Research Methodology
3.1 Research Philosophy
Irrespective of whether a researcher’s epistemological beliefs have been consciously
examined or discussed, these beliefs trickle-down to crucial parts of a paper such as the phrasing
of its research question and the selection of a methodology to study the data according to
Johnson & Duberly (2000). Therefore, there is growing traction in the importance of
management researchers to reflect on their epistemological beliefs and its implications for
research conducted by them.
The research at hand has been conducted in the assumption that there exists a reality that
is independent of the observer or in other terms - positivism (Easterby-Smith, Thorpe, &
Jackson, 2012). The positivist approach allows data to be applied in order to verify or falsify
predetermined hypotheses. To be able to do that data is stated in quantitative form and a large
sample of transactions is collected. As pointed out by Easterby-Smith et al. (2012), large
samples allow the uncovering of patterns and regularity, while it also admitted that it is a matter
of probability that the data collected will provide for an accurate and universal indication of the
underlying situation.
As a central criticism of the positivist approach, Kim & Donaldson (2016) bring forward
the failure to perceive the objective nature of data and methods. Granting that these methods
are in no way perfect in understanding all underlying processes and significance that people
attach to actions, they are useful in understanding a wide range of the reality at hand while being
of relevance to policy considerations (Easterby-Smith et al., 2012). In the paper at hand, a
quantitative approach enables a swift and straightforward analysis of the data. Although the
research has been conducted using a neutral approach, there is some probability that the researcher’s background and beliefs influenced the interpretation of the data to some extent. It
24
interpretations, while it is hoped that insights generated during this project can help managers
and policymakers gain a better understanding of the underlying aspects that motivate strategic
decision making in an FDI location choice context alongside being incorporated into future
research.
3.2 Data
In order to investigate the location choices made by firms’ investments at a regional rather
than a national level, industrial data on various MNEs were analysed. Numerous investments
were gathered using the database ORBIS/ZEPHYR which is maintained by the Bureau van Dijk
(2014). The bureau compiles data from various government and corporate sources and is
updated periodically (Bureau van Dijk, 2014). In order to focus on the highlighted relevant
equity-based investments only investments that resulted in a 10% ownership at least were
selected, as it is suggested by academical sources (OECD, 2000). Leaving the entry modes of
JV, M&A, and Greenfield as the ones to be observed. While previous research left acquisitions
out of the sample as the location choice here is constrained (Head et al., 1995; Shaver & Flyer,
2000; Alcacer & Chung, 2007; Alcacer & Chung, 2014) this paper addresses the research gap
in the analysis just there and explores how these constrained decisions compare to the
unconstrained Greenfield investments. As Greenfield investments are not included in the
Zephyr database, these were identified by Orbis. New subsidiaries opened between 2016 and
2019, with a parent company in Japan, and no takeover was selected and checked on a
transaction to transaction basis to ensure that the transactions at hand are actually Greenfield
transactions or so-called new plants. The number of firms was drawn from the database Orbis
(Bureau Van Dijk, 2019). The data for FDI was thereafter drawn from the sources mentioned
prior in order to test the hypotheses.
To determine the influence of both industry and country of origin agglomeration on the
25 Formula 1:
𝐿𝐺𝑀𝐽 =
𝛽 + 𝑒𝑋𝑖 + 𝑋𝑐𝑜𝑜 1 + 𝛽 + 𝑒𝑋𝑖 + 𝑋𝑐𝑜𝑜
The industry agglomeration in one location is described by 𝑋𝑖 while 𝑋𝑐𝑜𝑜 defines the
number of firms from the investing companies home country in the specific state is modelled.
Both variables will be further broken down and explained in the independent variable part
below.
In an effort to generate a comparable output the regression was carried out three times
producing three different regression tables and descriptive statistics similar to Basile (2002).
Consequently, the results were evaluated in a comparative approach to evaluate differences and
indicate the significance and strength of the effects for the three different entry mode choices (β) – Greenfield investments, M&A, JV.
3.3 Sample
The USA was used as a geographic sample as its availability of every kind of FDI entry
mode and scope of inward investment promises a sufficient sample. Moreover, the 50 states
provide regional boundaries that can be used for agglomeration calculation and agglomeration
definition.
326 Transactions from Japanese companies into the US were looked at in great detail that
allows controlling for the variables presented in the next section. These transactions were made
by 212 different companies, with some companies making more than one investment during
the time. Moreover, the focus on one country for acquiring or entering firms allowed controlling
for factors such as industry agglomeration as well as the country of origin agglomeration and
its different effects on the location choice much like Alcacer & Chung (2014). The sample
consists of a wide variety of firms to ensure diversity. Overall, there were 128 Greenfield
transactions, 29 JV deals and 169 M&A deals of Japanese firms into the US tested in the
26
well as increases in stakes, were eliminated from the sample. After controlling the sample
further, a number of deals were excluded due to limitations in the data available leaving the
finals sample with 111 Greenfield transactions, 29 JV deals and 162 M&A deals. Accumulating
to 302 transactions to be observed in the regression.
The general timeframe of these transactions was from 2015 up until June 2019 as the data
is less constrained by any crisis that would affect investment behaviour such as the financial
crisis in 2008 (Dornean et al., 2012) (extend how the crisis affects FDI). However, in the case
of JVs transactions, the time period had to be extended to 2010 up to June 2019 in order to
come up with a greater sample size as the original timeframe used for the other entry modes
would have only accounted for 15 cases of JVs. Therefore, it was decided to extend the
timeframe in a bid to gain a more insightful sample to test. 2010 was chosen as a starting date
as data availability before that point in time is limited not only on Zephyr but also other
resources are taken into consideration. The necessity to extend the timeframe for this entry
mode is down to the fact that JVs are being used less frequently in the US than in, for instance,
China because of regulatory differences. While this limited use might indicate that JVs are not
relevant to look at, they are still important as it is the most widely used equity FDI choice in
China (Beijing Municipality, 2019), the country with the – by far – second-highest inward FDI
flow (OECD, 2019b). The number of JVs has been increasing, and thus the amount of attention
that this deal type is receiving in literature is also growing (Lane et al., 2001; Yan & Luo, 2016).
3.4 Variables
3.4.1 Dependent Variable
Location choice of FDI (GF/M&A/JV) in terms of whether foreign firms invest in a location that is defined as agglomerated or not was the dependent variable of the analysis. The
27
maximization it was assumed that each investor chooses the state that provides him with the
highest possible profit. The profits mentioned might come from the availability of inputs into
the firms' production including agglomeration effects from proximate businesses in a similar
industry (Head et al., 1995).
To determine whether or not a firm’s subsidiary is located in an agglomerated area,
industrial data was analysed on a geographical level. Similar to, Holmes and Stevens (2004),
Alcacer and Chung (2014) and Li and Bathelt (2017) industry data from ORBIS was analysed
to determine and detect agglomerations in order to then assess if companies invest in the
respectively relevant agglomerated areas. The location choice of FDI was attributed the values
0 and 1. 0 for the states in the lower 80% of the agglomeration while the ten states with the
highest levels of agglomeration – the remaining 20% - were attributed the value of 1. Each state
and its values for industry and country of origin agglomeration can be retraced in Appendix 1.
The calculation can be reviewed in section 3.4.2 as these values serve as the predictors in the
analysis determining the location decision.
3.4.2 Independent Variables
After determining which states can be considered the host to agglomerations or of a
significant agglomeration the decision was regressed against a number of variables. First, the
independent variables will be introduced before control variables are being outlined.
Industry Agglomeration Levels: Formula 2:
𝐴𝑔𝑔𝑙𝑜𝑚𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑖𝑜 𝑝𝑒𝑟 𝑆𝑡𝑎𝑡𝑒 =𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑖𝑟𝑚𝑠 𝑝𝑒𝑟 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑝𝑒𝑟 𝑠𝑡𝑎𝑡𝑒 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑥 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑖𝑟𝑚𝑠 𝑝𝑒𝑟 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑥
The given formula, adapted from Du et al. (2008), indicates the level of agglomeration in
each of the states included in the sample. The formula includes the industry level to distinguish
28
Calculating this ratio per state for both high- and low-tech industries provided a percentage of
companies located in each state. This percentage was then regressed against the inward FDI
into the states in order to find out to what extent decisions are driven by these levels of
complementary industries.
Country of Origin Agglomeration Levels (Japanese firms in each US state)
Next to industry level agglomeration, country of origin agglomeration plays a great role
in research as it was tested in various studies (Zhang et al., 2012; Head et al., 1995; Tan &
Meyer, 2011).
Formula 3:
𝐻𝑜𝑚𝑒 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝐴𝑔𝑔𝑙𝑜𝑚𝑒𝑟𝑎𝑡𝑖𝑜𝑛 𝑅𝑎𝑡𝑖𝑜 𝑝𝑒𝑟 𝑆𝑡𝑎𝑡𝑒
= 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑖𝑟𝑚𝑠 𝑓𝑟𝑜𝑚 ℎ𝑜𝑚𝑒 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝑜𝑓 𝑓𝑖𝑟𝑚 𝑝𝑒𝑟 𝑠𝑡𝑎𝑡𝑒 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑥 𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑓𝑖𝑟𝑚𝑠 𝑝𝑒𝑟 ℎ𝑜𝑚𝑒 𝑐𝑜𝑢𝑛𝑡𝑟𝑦 𝑖𝑛 𝑦𝑒𝑎𝑟 𝑥
Even considering the centripetal forces that come into play once a certain threshold in the
size of the agglomeration is achieved, larger agglomerations are still considered more attractive
to foreign entrants as the increased productivity that is needed to overcome LOF. This research
follows the expectations brought up by Brakman et al. (2015), and De la Roca and Puga (2012)
that bigger agglomerations are a more important factor in the location choice. Indicating a firm
in a larger agglomeration makes for a more attractive target than a firm in a smaller
agglomeration (for that criteria).
States with high levels of agglomeration were defined by the percentage of the firms in
that state. The ten states hosting the highest percentage of firms (top 20%) in terms of industry
agglomeration was labelled a high agglomeration area for this study. Conducting two separate
analysis for each industry and country of origin agglomeration would be out of the scope for
this paper. Moreover, the states with high levels of industry agglomeration and country of origin
29 3.4.3 Control Variables
Experience and capabilities of firms
It can be expected that firms with increased experience are not in the same need of local
knowledge due to their prior knowledge. Therefore, firm age was used as a proxy to represent
these experiences and capabilities as it is proven to be hard to measure (Cohen & Levinthal,
1990; Kogut & Zander, 1992; Phene & Almeida, 2008). It is assumed that older firms have a
more thorough understanding of the industry and their field and therefore strategic decisions
like location selection. Experienced firms can be expected to recognize the value and risks of
agglomerations better.
Related to the argument that is been made by Shaver and Flyer (2000) that bigger firms
are less likely to co-locate than smaller firms it is to be expected that firm experience plays a
major role in the decision making towards agglomerated areas. Firm experience can come from
multiple sources. One of them is the experience of employees being better trained and more
experienced (Stolzenberg 1978; Brown & Medoff 1989; Kalleberg & Van Buren 1996; Troske
1999; Walace & Kay 2009). Le and Kroll (2017) state that the international experience of
executives is related to exposure towards foreign environments. Those create dissonance in
cognitive schemas of people which stimulates learning. In the long run, these learnings affect employees’ adaptability and performance. Obviously, that experience of employees trickles
down to the experience and capabilities of the organisation as a whole and thereby its capacity
to adapt to different environments more easily than firms who do not have these experiences at
their disposal. Applied to the location selection of a company it is expected that local knowledge
can be predicted and adapted to more easily, making the co-location with other companies and
agglomerated areas overall less relevant for the company in question. That is partly because the
company itself already has some international experience and learnings they can build on and,
thereby, adapt the new challenges at hand and also the experiences of employees explained
30 Knowledge intensity
As companies act within different industries (and some across multiple industries) the
sample was sub-grouped into knowledge-intensive ones and the ones who are not. Hatzichronoglou’s (1997) approach of classifying manufacturing industries based on the
technological intensity was applied and extended. Non-knowledge intensive industries such as
mining, agriculture, forestry, retail, logistics, and medium-low and low-technology
manufacturing activities received a value of 0. Knowledge-intensive activities, as defined by
Hatzichronoglou (1997), and finance, motion pictures, engineering, and architectural services
(Li & Bathelt, 2017) were attributed to the value of 1. High technology industries are identified
by the following Standard Industry Classification codes: 28, 35, 36, 37, 38, 42, 48, 41, 44, 45,
737, 966.
Creating a picture of location choice differences between both kinds of firms and
industries. More than only the locational factors of the investment decision, such as the
technological environment, the setup of the firm is also highly relevant. The intensity of
knowledge a firm processes can have a major impact on the decision for whether it is driven by
innovative considerations or solely on a cost and profit basis (Chung & Alcacer, 2002; Li &
Bathelt, 2017). This is evident in high technology industries that create an environment that is
knowledge-intensive (Li & Bathelt, 2017). In contrast, low technology industries, such as
manufacturing and mature industries where low-cost productions play a major role, are driven
by other aspects such as low labour cost. Consequently, and with the argument that innovative
firms prefer new establishments in agglomeration areas, it can be expected that
knowledge-intensive industries are more attracted by agglomerations than their low technology
counterparts. This comes as no surprise as low technology companies are reliant on low labour
cost, which is unlikely to occur in a high competition area such as an agglomeration. On the
other hand, high technology firms and knowledge-intensive firms can tap into that skilled
31 Subsidiary network
Similarly, to the age of an organisation, the number of subsidiaries also affects its
experience in the international context. This is possible in multiple ways. It can be expected
that MNEs with an existing network of subsidies have advantages in accessing knowledge pools
over firms that have only a few sites. However, there can be also costs associated with these
MNEs as the integration of diverse knowledge becomes problematic. Reasons for that can be
broad. First, each organization has hotspots that are more influential in the organisation than
others. Second, managers are cognitively limited which makes the coordination of many
subsidiaries with sometimes conflicting information problematic and more cost-intensive with
growing networks (Simon, 1947; Meyer et al., 2011). Thereby the search for knowledge
becomes increasingly limited (Laursen & Salter, 2006; Leiponen & Helfat, 2010; Love et al.,
2014; Li & Bathelt, 2017) as a negative relation from the size of a subsidiary network towards
location decisions for agglomerations is expected. The number of international subsidiaries can
be seen as another advantage in their experience of investing abroad. Given the experience in
foreign environments and the adaptability and flexibility caused by this experience, the more
international subsidiaries the smaller the need of investing in an agglomerated area to adapt to
the new market. The more internationalized an MNE is the more local and regional knowledge
and a reduced incentive to co-locate with other companies to gain local expertise (for some of
the motives). Table 2 provides an overview of the variables used for this research and where
32 Table 2 Variable description
Variable type Name Data source Measure
Dependent Location choice Orbis/Zephyr 1= into agglomeration
0= away from
agglomeration Independent Industry Agglomeration
within a US state Orbis As of formula 2 Country of Origin Agglomeration within a US state Orbis As of formula 3
Control Experience and capabilities of a firm (represented by firm age)
Orbis/Zephyr 2019- the year of the foundation of the acquirer
Knowledge intensity Orbis/Zephyr According to SIC code in a list separated into high (1) and low tech (0)
Subsidiary network Orbis/Zephyr Number of
33
4 Results
The following section outlines, first, the preliminary tests for the assumptions of binary
regression to ensure the validity of the results obtained. Second, descriptive statistics and
correlations are analysed. Then, third, the results of the main regression analysis are presented.
4.1 Preliminary tests
Normal distribution was first tested for alongside skewness and kurtosis through a
Shapiro-Wilk test across all levels of the independent variable. The results show that skewness
and kurtosis appear to be present in the data and that the data appears not to be normally
distributed. Despite this, normality of the raw data is not an essential requirement for binary
regression, especially for large sample sizes (Lumley, Diehr, Emerson, & Chen, 2002). To
ensure that this skewness and kurtosis does not create non-parametric, non-linear effects the
Pearson and Spearman correlation coefficients were compared. On the whole these delivered
similar results thus this does not seem to disturb the data. When looking at the different variables
these distributions can be explained. For the number of subsidiaries, a majority of companies is
on the smaller end of that scale when, in contrast, there are only a handful of really large
organisations with hundreds or even thousands of subsidies. While other regressions such as
linear regression require normal distribution for reliable results the binary regression used for
the research at hand does not depend on that. Therefore, transformation is not required.
Next multicollinearity was tested by using the variance inflation factor (VIF). To do that
regression for each of the IVs was executed while all remaining IVs stayed in as predictors. The
VIF and tolerance for all but two variables showed an acceptable range of not higher than 5 and
no lower than .2 respectively (Field, 2009). The only two variables for which this was not the
case were low-technology company-related agglomeration and high-tech company-related
agglomeration. In both cases, the VIF excelled the value of 10 while tolerance stayed above .2.
34
related due to the fact that some areas just host more companies than others and thus have a
higher density of corporations. This might stem from a customer base, resources, or skilled
workers to give just some examples amongst those that lead to agglomeration. The decision
was made to eliminate low-tech agglomeration from the model and run the regression tests with
high-technology agglomeration as a lone measure for industry agglomeration.
Next, in order to test for heteroscedasticity, both a Breusch-Pagan & Koenker test was
performed on all regression models used in the main analysis. For all models of the regression,
heteroscedasticity is not present, therefore it is concluded that it was not going to disturb the
results.
Lastly, Autocorrelation between the error terms was tested for using a Durbin-Watson
test to evaluate the independence of the errors. The test for the regression produced a score of
1.933 which are all within the acceptable range of between 1.5 and 2.5 (Field, 2009) and
indicates not correlated residuals. Therefore, it can be concluded that autocorrelation is not
present.
4.2 Descriptive Statistics
For this research 302 investments from Japan into the USA were assessed and analysed.
These transactions are broken down into 111 Greenfield investments, 29 JVs and 162 M&As.
Table 3 shows how the transactions are distributed over the three different entry modes and
whether or not transactions are made into agglomerated areas or not. Moreover, the kind of
agglomeration that is present in the respective area the investment is made into. The ten states
with the highest percentage of agglomeration for each of the cases, industry and country of
origin agglomeration, are used to define the agglomerated areas. In the case of industry
agglomerations those states are: California (12.3% of businesses within the USA), Texas