FDI & Ports

1

Abstract

This paper focuses on the question whether and to what the presence of a port affects the

greenfield FDI in the European NUTS-2 regions. The hypotheses have been derived from

existing international business and economic geography research context. In this paper

truncated negative binomial model and random effects model have been performed as

empirical methods of analysis and data regarding greenfield FDI from fDi Markets and freight

from Eurostat for the period 2003 - 2010 have been used. The findings of this study show that

there is no or little evidence that the presence of a port in a European region lead to more

greenfield investments in the same region.

1. Introduction

Jungnickel (1993) proclaims the most noticeable growth in trade took place in the 1970s.

Subsequently, the augmentation in foreign direct investment (FDI) has gone beyond of both

- national outputs and international trade. The second half of 1980s is noted with 33 per cent

annual growth rate of FDI (UNCTAD, 1993). Over short period of time the world share of the

latter has doubled. As remarkable as such event may be, the increase and distribution of FDI

can be accounted for only an insignificant portion of all international capital flows coursing

through highly integrated world financial markets (Banuri and Schor, 1992).

More than 20 years ago Dunning (1993) defined the “ownership–localization–internalization

(OLI) paradigm” which outlined the activities of multinational enterprises (MNE’s) as the ones

derived from the relationship between the competitive advantage of MNE’s and countries

and their ability to generate value-added activities. Firms prefer to invest abroad when

ownership advantages can be obtained by capitalizing possessory assets/capabilities and by

profiting from the location-specific advantages rendered from internalizing cross-border

activities in receiving country. Setting apart the definition of horizontal and vertical FDI’s,

Dunning (1993, 1998) specified four key determinants for the companies to internationalise

the operations and emphasize the specific-location advantage: (1) the natural resource -

seeking incentive (admission to natural resources); (2) the market - seeking incentive

(admission to new markets); (3) the efficiency - seeking incentive (reorganization of

operations in order to diminish the labour, machinery and material costs of production and

2

increase efficiency); (4) the strategic - asset - seeking incentive (admission to strategically

related created assets). Both, international business and economic geography research, focus

their interest on the notion of location - where and why the companies opt for explicit

destination of their activities (e.g. Nachum & Wymbs, 2005; Porter, 2001; Krugman, 1991a;

Markusen, 1996). The success of any location, would it be a country, a region or a city

predominantly relies on its attractiveness compared to other locations in terms of availability

of local resources (Robinson, 2002). The ports have traditionally served as means to connect

domestic and international markets and impact the trade.

The ports in their essence contribute to the value of the shippers and to the third-party service

providers; the value projection is based on the consumer segmentation and targeting, thus

facilitating the ports to obtain value not only for themselves, but also for the chain those are

implanted in. As the status of ports changes from sites with distinct functions to elements of

value-driven chain systems, a paradigm shift takes place (Robinson, 2002). The role of the

ports as a transportation or production hubs has been greatly neglected by the academics

focusing on the world-wide commodity or production chains, regardless their contribution to

the global allocation of goods (vessels carry about 90 per cent of world trade volumes) (Jacobs

et al., 2010).

This paper is dedicated to the analysis of greenfield FDI, focusing on the ports of European

NUTS-2 regions during the period 2003 - 2010. The main concern is whether and to what the

presence of a port affects the FDI in European regions. The paper is organized as follows: in

Section 2, the focus is given to the theoretical framework and the presentation of the

hypothesis. Section 3 provides introduction of data and empirical methods used for analysis.

Section 4 continues with the discussion of the latter and the results obtained. Section 5

contains the conclusions and suggestions for the future research topics.

3

2. Literature review

2.1. Background

To what extent the ports keep having an influence on the hosting city is a question of general

concern. Their relationship and interdependence seem to be decreasing both in Europe and

in Asia, where the comparison has been drawn and different reasoning for each continent

concluded (Ducruet, 2006). The costs of conducting a port activity, accompanied by higher

levels of pollution and congestion, are increasing and since transportation costs decrease,

negative externalities expand into the hinterland (Hall, 2007). Both, costs and profits, are

spread over a wider region, immersing into the latter.

Nowadays, a new stream of port development has appeared - ‘port regionalization’ - due to

the growing importance of logistics networks and emerging locations outside the traditional

port territory. Within the port regionalization concept, the activities of the port are partially

relocated into the hinterland, where opportunities to improve capacity and freight circulation

due to affordable and extensive spatial capabilities arise. Such process is predominantly

market-driven and requires high levels of efficiency gains (Notteboom et al., 2005). Although,

the topic of port regionalization has not been well-studied from a spatial econometric

perspective, Bottasso et al. (2014) found that ports possess tendency to increase GDP of the

region where located and have positive effect on the GDP of neighbouring regions.

Ballou (2007) in his research has indicated that under the feasible hypothesis regarding a

general trend of increased globalisation and free trade in combination with increasing

outsourcing activity, the logistics activity will gain more importance and growth, which in turn

indicates an inflow of FDI due to the internationalization of the economy. Thus, it is not

surprising that the region of Lombardy (Italy) received 43% of inward FDI of the Italian logistics

sector in 2010 (Maggi et al., 2011), while it does not have a direct access to the nearest

seaport (located in the region Liguria) and Milan, the regional capital, is the main economic

agglomeration of the region Lombardy.

The topic of spatial distribution of the economic activity has been well-studied prior. The

neoclassical location theory presupposes the idea that the market will converge to an

equilibrium. In the long-term, the economic activity will locate ubiquitously through the

territory due to the constant returns to scales. Hence, the more productive industries will

4

locate in the core regions, where costs are higher. In the early 1990s a theoretical framework

of “New Economic Geography (NEG) models” has emerged (e.g. Krugman, 1991b; Puga,

2002). These models consider the chance that diversifications in the accessibility and

attractiveness of the locations caused by FDI into infrastructure could possible contrary

impact on the spatial distribution of economic affairs. The models vary from the preceding

ones because of the market seeking (market size) and cost saving (production cost)

imbalances. As such, this indicates convergence of economic affairs also encouraged by

trading activities of the regions. Thus, agglomeration economies, affected by firms’ location

decisions, become endogenous (Ottaviano, 2008). Additionally, locations in a proximity to

gates and hubs are favoured due to efficient transport infrastructure (Krugman, 1993).

Traditionally, economic theory has been proposing that transport infrastructure positively

affects industry productivity predominantly by diminishing transport and time costs, leading

to decrease of production costs, increase in productivity, specialization, trade and

competition advancement, expansion of compatible markets (Aschauer, 1989); better

accessibility and low transportation costs assist in escalation of market capabilities of various

places (Niebuhr, 2006; Condeco-Melhorado et al., 2011). The local and regional government

officials are known to have supporting attitude towards transport infrastructure (where ports

own one of key roles) as it is a determining component to promote territorial cohesion,

economic development and concentration, moderate economic inequalities (Notteboom et

al., 2005). For example, for the period 2007 - 2013 around 30% of the EU Regional

Development Funds (ERDF) and Cohesion Funds have been dedicated to organize investments

with transport infrastructure (intended at enhancement of regional benefits and finalizing

trans-European Transport networks (TEN’s)) receiving a notable portion of the budget

(ESPON, 2009).

2.2. Hypothesis

Port as a magnet for regional FDI in Europe

Nowadays, MNEs more and more regard the continent Europe as a moderately unified

territory instead of a group of independent countries. Thus, regions in Europe with analogous

features located in different countries are frequently recognized as comparable replacements

than unalike regions of the same country (Basile et al., 2009). The regional competition has

5

increased due to the process of European integration, which has (1) shaded national

boundaries; (2) steadily eliminated economic, social and cultural dissimilarities among the

countries; (3) promoted free movement (capital, goods and workers )(e.g. Cheshire and

Gordon, 1995; Budd, 1998; Gordon, 1999; Lever, 1999; Markusen and Nesse, 2007).

Consequently, the groups embodying economical concerns of the area are competing, not

the regions themselves (Gordon and Cheshire, 1998). The local and regional government

officials under stress of the elections participate in the competitive ventures to impact the

employment and protect business premiums. Besides that, the officials desire to be

recognized for their dynamic role in exhilarating local and regional economic development

(Turok, 2004; Markusen and Nesse, 2007). The modern zeals to lure the FDIs in the region

taken by the officials include enhancement and application of regional assets associated with

specific labour pools, university research and societal background (Raines, 2003; Turok, 2004).

The stimulus is focused on the affinity of inward investments by powering emphasis and

support of the peculiar robustness of the location (Raines, 2003).

The regional trade (i.e. economic) activities significantly rely on the accessibility of port

facilities and services. As many industry experts and researchers note, one of the primary

instruments to develop port and its’ hinterland is through FDI inflow. Those can positively

impact economic health of the regions as well as of the country (Cho and Ha, 2009). As the

previous capital instalments of the investees (in contrast to the purchase of the present firms)

do not impact the location choices related to greenfield investments, the latter are beneficial

in the research of regional competition (Burger et al., 2012). Thus, the presence of the port in

a region can have a positive effect on the FDI inflow of the same region.

H 1: There is a positive relationship between the presence of the port in the European

region and the number of greenfield investments in the same region

Port, World City and regional FDI in Europe

Megacities and global cities are two different concepts that can be wrongly interchanged. The

first one is just a big agglomeration of people, such as Calcuta, in India, or Chongqing, in China.

What make them different from a world city is that both are missing cosmopolitanism and

interconnectedness environment (Goerzen et al., 2013). Sassen’s (1991) concept of “Global

6

capacity” inspired a hierarchical theoretical classification of 10 alphas, 10 bettas, 35 gammas

and 67 delta global cities (Beaverstock et al., 1999). Sassen (2000) concluded that the

globalization of capital was bringing to such cities as Tokyo, Singapore and London a

significant share of corporate powers. These cities are taking, for instance, the power of the

international finance (Beaverstock et al., 1996). Additionally, these cities are also strategic

locations for the internationalization of the advance product services (APS) firms, due to the

high educated labour pool and for being close to the clients (Bagchi-Sen, 1997).

As prior mentioned, Dunning (1998) has established a link between FDI flow and MNE’s

location strategies. Hence, the companies desire to locate in places maximizing corporate

objectives. A global city alone may be limited to receiving investments from one specific

service/industry sector, while port city may be the goal destination of the other, as both

locations show different characteristics. However, global and port cities together would have

a greater range of characteristics and options to offer, with different and complementary

labour pools, which could attract more firms from different sectors whose aims are

completely diverse.

H 2: The positive relationship between the port in the European region and the number of

greenfield investments is stronger in the presence of a world city in the same region

Port, Infrastructure and regional FDI in Europe

Ports have arisen as important places for logistics since supply chain management has

become one of the key components of the competitive advantage for firms and companies

(Li et al., 2006), as important places of trade and exchange of transport modes for

commodities and manufactures. As it has been said, ports are important for the management

of physical flows, transporting and storing, and by doing so, ports need physical infrastructure

connected with the city and the region. Transport infrastructures (defined by Rietveld (1994)

as “facilities such as railway lines and stations, highways, canals, sea- and airports”) are known

to be useful to speed up the convergence process between regions and economic growth. For

instance, Démurger (2001) found that variation between Chinese provinces’ economic

performance could be partly explained by transport infrastructure and telecommunication

facilities, while Boopen (2006) performed an analysis in Africa, providing also positive

7

relationship. But to take advantage of the port, both systems should be complementary and

integrated, and ports should have good connectivity to spread their economic potential.

The ports can be classified, according to their transhipment incidence, as: pure transhipment

hub (where containers are stored temporarily after being handled), hub (where transhipment

is combined with export-import activity), regional gateway (where the presence of

transhipment is lower than the export-import activity), and gateway/feeder (mostly focused

in export-import activity). The first kind of ports do not to have a great economic impact on

the surrounding areas (Musso et al., 2004), examples of these are Gioia Tauro and Algeciras,

located in Calabria (Italy) and Cadiz (Spain), two of the most depressed regions in Europe

where the percentage of transhipments is higher than 90%. On the other extreme, gateway

and feeder ports have a lower percentage of transhipments, for example the cases of the

biggest European ports (Rotterdam, Hamburg and Antwerp) the transhipment ranges

between 29 and 36%, so they must have good connections with hinterlands, good examples

are Rotterdam (The Netherlands) and Los Angeles (USA) (Rodrigue et al., 2013). In such

circumstances, ports may act as engines for the companies located close by or may attract

firms to the region, as those can make use of the offered infrastructure. Nevertheless,

congestion of transportation infrastructure may have negative effect on the total travel time,

also known as the “Braess paradox” and suggests that a rise in the supply of infrastructure

has a negative impact on the productivity (Sheffi, 1985). Thus, to maximize the profits the

port needs to be well connected (in combination with manageable traffic situations) and

integrated in a regional and national network. In the cases when the connectivity by road,

train or air is better, the economic impact would be higher and thus the port would attract

more FDI.

H 3: The positive relationship between the port in the European region and the number of

greenfield investments is stronger in the presence of good infrastructure in the same region

Port, Business Activity and regional FDI in Europe

Ports have been important places to relocate physical goods from one transport mode to

another for international trade. These factors used to facilitate investment and urbanization

of cities. However, ports have become irrelevant for the prosperity of cities (Fujita and Mori,

8

1996), as negative externalities associated to them, such as congestion, have arisen,

weakening the relationship between ports and port cities (Zhao et al., 2017).

Since the wave of national deregulations started in the 1970s and the proceeding

globalisation, there has been in a process of industrial relocation which has been studied by

two streams of analysis, called World Network Cities (WNC) and Global Commodity Chain

(GCC). While the first ones are analysis of the networks of information and advanced product

services (APS), this is, financial companies and law firms; the second are the analysis of the

production units connected through flows of commodities, where ports have an important

role. These flows seem to have an uncertain relationship (Jacobs et al., 2010), as APS tend to

follow the urban hierarchy. Cases about this can be found in Canada (Slack, 1989) and

Australia (O’Connor, 1989). Three main types of locations have been identified by O’Connor

(1989): firstly, port cities providing basic daily services; secondly, port cities managing long-

term contracts and thirdly, international cities managing the overall worldwide shipping

industry. On the other hand, ports are becoming important in the process of the supply chain,

thus they have the necessity to provide logistic facilities (Pettit and Beresford, 2009). This can

lead to expectation that surrounding regions of a port may receive FDI from companies that

take part of supply chains, which are companies linked between each other to add value in a

product that should be delivered to a customer (Christopher, 1992). Thus, manufacturing,

transport and logistic firms may be expected to locate in these port regions to take advantage

of the facilities of the seaport. The clustering of such companies then may attract FDI from

the same sectors and industries. On the contrary, APS firms will locate in global cities

specialized in information sectors that don’t need to be close to the physical flow of goods.

H 4: The positive relationship between the port in the European region and the number of

greenfield investments is stronger for the logistic sector projects of the same region

3. Data & Methodology

3.1. Data and Variables

The primary data sources that have been used, are the fDi Markets database and Eurostat

database for the timeframe 2003 - 2010, both on NUTS-2 level. The fDi Markets database

(26.995 observations) contains data on project level about the greenfield investments

between different sectors/business activities and the presence of different world cities. The

spatial distribution of total number of investments across two hundred sixty NUTS-2 European

9

regions (EU-25, Switzerland, Norway, Cyprus) is provided in Figure I. The investments are not

evenly distributed, it can be noted that regions including the world cities receive greater share

of FDIs.

Figure I - Total number of investments across 260 NUTS-2 European regions

The Eurostat dataset (2.120 observations) contains data about regional specific effects, such

as GDP per capita, long-term unemployment rate, percentage of people with high education,

accessibility, etc. Another dataset contains data (1.101 observations) about the amount of

freight (in thousand tons) on NUTS-2 level and is also coming from Eurostat. The importance

of this data is justified due to the reason that it shows in which region maritime transport

activity is present and thus indicating the existence of the ports.

To make the data suitable for the empirical analysis, some modifications have been made to

the datasets. During the merger of the three datasets, the number of observations went down

to 1774 observations due to several reasons. Firstly, the regions (NUTS-2 level) with no

specific regional data (GDP, employment, etc.) were deleted from the database, because for

those regions it was not possible to control for place specific effects. Secondly, the database

has been collapsed based on the NUTS-2 regions, to show the amount of investments per

region (dependent variable) and different business activities. In this way detailed information,

such as company name, of no additional contribution to the model has been deleted. In the

10

obtained dataset of greenfield investments within European regions, there were regions with

zero greenfield investments and, moreover, for some of the regions no regional data was

available, resulting in exclusion of such data.

Dependent variable

For the empirical analysis, multiple dependent variables will be used. The first one is the

number of greenfield investments per region (NUTS-2) per year, which has been called

‘NumberofInvestments’. The second dependent variable is the logarithm of the ‘Investment

total’, which is the total amount of investments in a certain region in a year. Besides these

dependent variables, attention is given to the specific effect on different business activities

within the relationship between the presence of ports and greenfield investments. To analyse

this effect, several other dependent variables have been made to categorize different

business activities, namely “Business Services”, “Headquarters”, “Logistics”, “Production”,

“R&D”, “Sales & Marketing” and “Support & Servicing”. They indicate the amount of

greenfield investments per year for specific business activities within a region.

Independent variable

To show the effect of ports in a region on the greenfield investments, a dummy variable

‘PortinRegion’ has been introduced based on the freight data of Eurostat. If there was freight

transport in a certain region, the dummy will get a 1 and otherwise a 0.

Control variables

To control for factors that may influence the relationship between the existence of a port in

a region and greenfield investments, we use multiple control variables. In the article of

Karreman et al. (2017), three categories of control variables have been used, namely demand

factors, supply factors and external economies. These categories show the attractiveness of

European regions. In this paper, the first two categories have been used. External economies

category has been left out from the model since there is no accessible data (Table 2). The

control variables are measured on the NUTS-2 regional level and the corporate tax variable -

on the country level. In the correlation matrix (Table 1), where possible multicollinearity has

been checked for.

Table 1 - Correlation Matrix

11

Demand factors

The demand factors look at market-seeking incentives for companies with possible greenfield

investments. These demand factors consist of GDP per capita, accessibility by air and rail/road

and world cities. GDP per capita is included in the model, because it is expected that this has

positive influence on the relationship between ports and FDI, and companies will more easily

invest in regions which are doing economically well. The accessibility factors are coming from

the ESPON research by Spiekermann and Wegener (2006). Accessibility is important from the

fact that the attractiveness of a region increases when the region is better accessible. The

world city control variable indicates specific investments in a year by region, what type of

world city is present in that region. Introduction of this control variable to the model, enables

to control for the effect that a company will establish itself in a region not only for a port, but

because of a large city in the region which could have multiple benefits, like good company

climate. The world city variable has been divided into 5 groups, namely Alpha+, Alpha, Beta,

Gamma and Rest.

Supply factors

The category supply factor consists of the long-term unemployment, university degree and

corporate tax rate. The long-term unemployment and university degree (education) show the

viability of the labour market (Head and Mayer, 2004). The employment compensation

control variable has not been used, for the reason of high correlation with the GDP per capita

factor. The unemployment factor is coming from Eurostat dataset with NUTS-2 level data. The

university degree factor is measured following the International Standard Classification of

Education (5–6) [ISCED]. It is expected that both these variables will influence the effect of

Port in Region GDP per capita World city

reference

Long-term

unemployment University

Accessibility by

air

Accessibility by

rail and road Congestion costs Corporate tax

Port in Region 1

GDP per capita 0.103 1

World city

reference -0.010 0.407 1

Long-term

unemployment -0.151 -0.439 -0.065 1

University 0.153 0.566 0.346 -0.205 1

Accessibility by

air -0.134 0.574 0.549 -0.171 0.378 1

Accessibility by

rail and road -0.379 0.314 0.104 0.024 0.091 0.603 1

Congestion

costs 0.057 0.557 0.396 -0.271 0.485 0.57 0.209 1

Corporate tax 0.044 0.284 -0.043 0.048 0.106 0.245 0.45 0.053 1

12

ports on FDI in a positive way. With long-term unemployment, there are more people in the

labour market available for jobs and with more skilled people (University degree), they will

attract FDI to the region. The corporate tax rate is a good addition to the model, because if

countries have different tax rates, maybe companies will invest in the country with lower

taxes. The corporate tax rate data is deducted from the EY International Tax database

(Brienen et al., 2010). The congestion costs are added as a control variable to the model,

because it could have a negative effect on the relationship of ports on FDI. Congestion cos ts

variable is derived from the ESPON research by Spiekermann and Wegener (2006).

Table 2 - Descriptive Statistics of the Variables in the Models

3.2. Methodology

To examine the effect of the presence of a port(s) in a region on the number of greenfield

investments in that region an appropriate model must be chosen. The dependent variable is

a count variable, thus an OLS model will not be adequate, because count data is discrete and

will normally violate one of the assumptions of OLS, namely the assumption of equal variances

(Gardner et al., 1995). The mean and the variance of count data are linked to each othe r, as

Name Description Mean SD

Port in Region Dummy variable if there is maritime

(freight) transport in the region with ports 0.38 0.49

GDP per capita Regional GDP (in millions of euros) 22292.6 11047.5

World city

classification

Alpha+ (5), Alpha (4), Beta (3), Gamma (2)

and Rest (1) 1.43 0.93

Congestion

costs Level of congestion 0.22 0.22

Long-term

unemployment Long-term unemployment rate 3.18 2.75

University Percentage of the workforce between 25

and 64 with tertiary (ISCED 5–6) 0.28 0.09

Corporate taxes Statutory corporate tax rate 29.16 6.08

Accessibility by

air

Number of people that can potentially be

accessed by air (in millions) 96.81 31.97

Accessibility by

rail and road

Number of people that can potentially be

accessed by rail and road (in millions) 98.1 61.22

13

higher outcomes lead to higher variances. When working with count data, in this case the

number of greenfield investments made in the region, there are mainly two models

appropriate: a Poisson model or a negative binomial model. An important assumption of the

Poisson regression is that the mean is equal to the variance of the count data. However, in

our data this is not the case. The variance is much larger than the mean of the number of

greenfield investment in the regions and therefore this assumption is violated (see table

below). This phenomenon is called overdispersion. Therefore, the Poisson model is not the

right model in this study. In case of overdispersion, the negative binomial model is preferred

and thus this model will be used to examine the effect of the presence of a port in a region

on the number of greenfield investments in that region.

Table 3 - Total number of investments

Variable Observations Mean Standard Deviation

Total number of investments 1891 13.939 23.076

Another common problem with count data is that the data is skewed due to the presence of

many outliers in the data. This could occur due to the presence of multiple high values, but

most of the time skewness with count data is caused by many zeros in the data (Atkins et al.,

2013). In the latter case a zero-inflated negative binomial model must be chosen. However,

in the data obtained in this study there are no values of zero. Consequently, this leads to a

new situation in which the model must correct for the fact that there are no zeros in the count

variable. This is possible using the zero-truncated model, which is used when count data is

used and a value of zero is not present in the dataset. Keeping in mind that overdispersion is

present in the count data, the definitive model is a zero-truncated negative binomial.

TotalNumberofInvestmentsit = β0 + β1PortinRegionit + β2GDPpercapit +

β8LongtermUnemploymentit + β3Universityit + β5AccessibilitybyAirit + β6AccessibilitybyRailRoadit

+ β7CongestionCostsit + β8CorporateTaxit + β4WorldCityReferenceit + εit

Also, in this study a model is used to estimate the effect of the presence of a port on the total

amount of greenfield investments in that region. Since the total amount of investments is a

continuous variable a negative binomial model would not be logical, instead a fixed effects or

14

random effects model is more appropriate. The fixed effects model will omit all the time -

invariant variables, which will not be beneficial for the model (Allison, 2009). For example,

both infrastructure variables will be dropped as well as the world city classification. More

importantly, the main variable of interest, namely the presence of a port in the region, will be

omitted in this model, because of its time-invariant character. The fixed effects model will

therefore not be the right model to use in the present study. Based on this large disadvantage,

the random effects model is chosen to estimate the effect of the presence of a port on the

total amount of greenfield investments in that region. It is expected that this estimation yields

the same sign and significance for the main variable, namely the presence of the port. In both

above mentioned models several control variables will be included.

Thereafter, three models will be run to test hypothesis two and three. All these random

effects model will have an interaction variable, which will indicate if the hypothesis two and

three hold. In the first of these models an interaction variable is created with the presence of

the port in a region and the world city classification to test hypothesis two. In the latter two

of these models an interaction variable is created with the presence of the port in a region

multiplied by on the one hand accessibility by air and on the other accessibility by rail and

road. Hence, the outcome will show if the hypothesis for infrastructure holds. Furthermore,

for all random effects models the standard errors are made robust to enhance the structural

validity.

The effect of the presence of a port in a region on number of greenfield investments for

different business activities will also be researched. Seven random effects regressions will be

performed in which the different types of business activities will be used as dependent

variable. Following this methodology will lead to the results and hence helps to answer our

hypothesis and consequently the main question.

4. Results

In this section, the results of the truncated negative binomial model concerning the effect of

the presence of a port in a region on the number of greenfield investment in that same region

will be discussed firstly. Secondly, the random effects model regarding the effect of the

presence of a port in a region on the total amount of investments in that same region will be

addressed. In the random effects models, multiple interaction effects are used to check

15

whether hypothesis 2 and 3 hold. To examine the effect of different business activities on the

presence of ports in a region on FDI, multiple random effects model will be used. This model

will conclude the section and afterwards the results will be discussed in the discussion.

Table 4 shows the results of the truncated negative binomial model and two random effects

model with the world city interaction effect. In the first model, the dependent variable is the

TotalNumberofInvestments. The independent variable, which indicates if there is a port in the

region, is presented as a dummy variable and takes the value of ‘1’ if a port in present in the

region and a value of ‘0’ if this is not the case. The results show that the presence of a port in

a region positively affects the number of greenfield investments in that same region. This

finding is significant at a 1% significance level and means that the logs of expected counts

would increase with 0.18, ceteris paribus. This finding supports our first hypothesis.

Moreover, almost all control variables are significant at a significance level of 1% in this first

estimation, only long-term unemployment is not significant.

However, there are some unexpected signs regarding the control variables. For example, the

variable GDP per capita has a negative sign, where a positive sign was expected. When the

GDP per capita increases in a region, it has been expected that this would attract greenfield

investments due the fact that there will be more demand in the region. However, the results

argue otherwise. Also, the variable congestion costs show a peculiar effect. The coefficient

shows that an increase in congestion costs will lead to an increase in the number of greenfield

investments. This is peculiar, as when reasoned logically, higher costs usually make a region

less attractive for investments. Again, the results provide evidence in the contrary. The

coefficients of other variables are also notable given their large positive coefficients, namely

the dummy variable of the world city classifications. In this model, the effect of an Alpha+ city

classification leads to an increase in the logs of expected counts of 2.86, ceteris paribus.

The results of the random effects model (model 2) which estimates the effect of the presence

of a port in a region on the total amount of greenfield investments in that same region.

Correspondingly with the results of the truncated negative binomial model in Table 4, the

random effects model (model 2) also shows a positive and significant effect of the variable

port in the region.

16

Table 4 – Results TNB Binomial model & random effects models

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

17

Table 5 - Random effects models by accessibility

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

The GDP per capita variable shows the expected effect in the random effects model, namely

a positive effect on the total amount of investments. However, long-term unemployment and

corporate tax show an opposite effect compared to the negative binomial model. This is also

contradictory with our expectation. The rest of the variables are compliant with our

expectations, such as congestion costs (negative effect).

18

The third model is included with the interaction effect between port(s) in a region and

different levels of world cities. All the interaction effects for every level of world city are

insignificant. Comparing model 3 with model 2, hardly any changes can be discovered and

eventually hypothesis 2 can be rejected.

Table 5 includes the fourth model, where an interaction effect with port in the region and

accessibility by air is added; and the fifth model, where an interaction effect with port in a

region and accessibility by rail & road is included. All these interaction effects are insignificant.

Comparing model 4 and 5 with model 2, hardly any changes can be discovered and eventually

hypothesis 3 must be rejected.

Table 7 shows the results of the seven random effects models, with the different business

activities as dependent variable. The estimates of the effect of the presence of a port in a

region are of main interest, since this outcome will show if hypothesis 4 holds. The results for

the variable presence of port are mostly insignificant, however for the logistics sector the

variable is positive and significant at a significance level of 1%. So, when a port is present in a

region an increase in logistics greenfield investments is expected. This partially aligns with the

expectations mentioned in the literature review. Table 6 is a presentation of the shares of

business activities in the overall amount of investments. It shows the distribution of

investments in our dataset.

Table 6 - Function distribution of greenfield investments in European regions

19

Table 7 - Random effects models by business activities

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

The control variables show some different effect, then in the first random effects model. GDP

per capita shows mixed results as well as congestion costs. However, other variables show

predominantly the same sign and significance. For example, the education variable,

university, is positive and significant for six of the seven models. The same goes for the world

city classifications variables. All coefficients are positive and only 3 out of 28 coefficients are

not significant. The corporate tax level is negatively significant for all models. Long-term

unemployment, accessibility by air and rail and road are predominantly insignificant.

20

5. Conclusion and Discussion

When firms want to invest in other countries and/or regions, they look at several

determinants of location choices. A possible factor in these location choices is the presence

of a port. In this paper, we present evidence for the relationship between the presence of a

port and foreign direct investments in European region. In detail, the main research question

namely whether and to what the presence of a port affects the greenfield FDI in the European

NUTS-2 regions, has been studied from four angles. The first angle, is the positive presence of

a port in a region and foreign direct investments. Second, we address the effect of a world

city in the region with the presence of a port and relation to FDI. Third, we check whether the

accessibility by air and rail & road have a positive influence of the relationship. Finally, we

made a distinction between multiple business activities of the investments and the presence

of a port.

The main results of the empirical analysis have a mixed outcome. The presence of a port in a

region has in model 1 till 3 a positive significant influence on investments in that same region,

but for models 4 and 5 there is no significant evidence on this effect. Focussing more on detail

with also the presence of a world city and accessibility by air and rail & road in the region, we

don’t find evidence that these factors have an influence on the relationship between ports

and FDI. Looking at the differences in business activities and investments, we find only a

positive outcome for logistics. This is an expected outcome, following the fact that ports are

important nodes in the supply chain of many firms (Pettit and Beresford, 2009).

Concerning the control variables some outcomes were contradictory to the expectation. For

example, the GDP per capita in the negative binomial model is negatively significant, where

the expectation would be that higher a higher GDP per capita would lead to more greenfield

investments in that same region. Since the average employee compensation in the dataset is

highly correlated with the GDP per capita, a possible explanation could be that higher wages

deter greenfield investments. Another unexpected result in the negative binomial model was

the positive and significant coefficient for the congestion costs. Logically, higher costs lead to

a lower attractiveness of a region, however maybe due to the popularity of region, these costs

are taken for granted. In the random effects model, other remarkable results were obtained.

The long-term unemployment variable was negative and significant. Higher unemployment

21

means a wide offer of labour forces. However, this could also be a sign that the region is not

performing that well. Another peculiar effect is that the corporate tax rate variable shows a

positively significant effect. Lower taxes usually lead to a higher attractiveness of a region. A

possible explanation could be that higher corporate tax rates are correlated with better legal

systems, moreover the infrastructure was also somewhat correlated with the level of

corporate taxes. These results can serve as the basis for future research.

The overall answer to the research question is that there is no or little evidence in this study

that the presence of a port in a European region lead to more greenfield investments in the

same European region. None of the hypothesis can be fully accepted and some even need to

be rejected. A limitation in this research is regarding the size of the port in a region. In a

future research, it would be good if an independent variable would be created with different

levels of the size of ports and their effect on greenfield investments in the same region.

Another limitation is that there are no control variables in the field of external economies, in

which we control for previous investments in the same region. Also, concerning previous

investments in specific business activities, such as logistics. Future research could be focussed

on other continents, since this research has only included European regions.

22

6. References

 Allison, P. D. (2009). Fixed effects regression models (Vol. 160). SAGE publications.

 Aschauer, D.A. (1989). Is public expenditure productive? J. Monetary Econ. 23, 177–200.

 Atkins, D.C., Baldwin, S.A., Zheng, C., Gallop, R.J., & Neighbors, C. (2013). A tutorial on

count regression and zero-altered count models for longitudinal substance use data.

Psychol Addict Behav. 2013 Mar; 27(1): 166-77. doi: 10.1037/a0029508. Epub 2012 Aug

20.

 Bagchi-Sen, S. (1997). Service employment in large, medium, and small metropolitan areas

in the United States. Urban Geography, 18(3), 264-281.

 Ballou, R. H. (2007). The evolution and future of logistics and supply chain management.

European business review, 19(4), 332-348.

 Banuri, T., Schor, J. (eds.) (1992). Financial openness and national autonomy:

Opportunities and constraints. Oxford. Clarendon Press.

 Basile, R., Castellani, D., & Zanfei, A. (2009). National boundaries and the location of

multinational firms in Europe. Papers in Regional Science, 88: 733–748.

 Beaverstock, J. V.& & Smith, J. (1996). Lending jobs to global cities: skilled international

labour migration, investment banking and the City of London. Urban studies, 33(8), 1377-

1394.

 Beaverstock, J., Smith, R., & Taylor, P. (1999). A roster of world cities. Cities, 16(6): 445-

458.

 Boopen, S. (2006). Transport infrastructure and economic growth: evidence from Africa

using dynamic panel estimates. The empirical economics letters, 5(1), 37-52.

 Bottasso, A., Conti, M., Ferrari, C., & Tei, A. (2014). Ports and regional development: a

spatial analysis on a panel of European regions. Transportation Research Part A: Policy

and Practice, 65, 44-55.

 Brienen, M. J., Burger, M. J., & van Oort, F. G. (2010). The Geography of Chinese and Indian

Greenfield Investments in Europe. Eurasian Geography and Economics, 51(2).

 Burger, M. J., van der Knaap, B., & Wall, R. S. (2012). Revealed competition for greenfield

investments between European regions. Journal of economic geography, 13(4), 619-648.

 Budd, L. (1998). Territorial competition and globalization: Scylla and Charybdis of

European cities. Urban Studies, 35: 663–685.

 Cheshire, P. C., & Gordon, I. R. (1995). European integration: the logic of territorial

competition and Europe’s urban system. In J. Brotchie, M. Batty, E. Blakely, P. Hall, P.

Newton (eds) Cities in Competition: Productive and Sustainable Cities for the 21st Century,

pp. 108–126. Melbourne: Longman.

 Cheshire, P. C., & Gordon, I. R. (1998). Territorial competition: some lessons for policy. The

Annals of Regional Science, 32: 321–346.

 Cho, H., & Ha, Y. (2009). Determinants of FDI Inflow in Regional Port with Resource-Based

View and Institutional Theory: A Case of Pohang-Yeongil Port. The Asian Journal of

Shipping and Logistics, 25(2), 305-331.

23

 Condeco-Melhorado, A., Gutierrez, J., Garcia-Palomares, J.C. (2011). Spatial impacts of

road pricing: accessibility, regional spillovers and territorial cohesion. Transportation

Research Part A 45, 185–203.

 Démurger, S. (2001). Infrastructure development and economic growth: an explanation

for regional disparities in China?. Journal of Comparative economics, 29(1), 95-117.

 Ducruet, C. (2006). Port-city relationships in Europe and Asia. Journal of International

logistics and trade, 4(2), 13-35.

 Dunning, J. H. (1993). Multinational enterprises in the global economy. Wokingham, UK:

Addison-Wesley.

 Dunning, J. H. (1998). Location and the multinational enterprise: a neglected factor?

Journal of International Business Studies, 29: 45–66.

 European Spatial Planning Observation Network – ESPON, 2009. Accessibility Project

Reports. <www.espon.eu>.

 Fujita, M., Mori, T. (1996). The role of ports in the making of major cities: self-

agglomeration and hub-effect. J. Dev. Econ. 49 (1):93–120.

http://dx.doi.org/10.1016/0304-3878(95)00054-2.

 Gardner, W., Mulvey, E.P., & Shaw, E.C. (1995). Regression analyses of counts and rates:

Poisson, overdispersed Poisson, and negative binomial models. Psychol Bull. 1995 Nov;

118(3): 392-404.

 Goerzen, A., Asmussen, C. G., & Nielsen, B.B. (2013). Global cities and multinational

enterprise location strategy. Journal of International Business Studies, 44 (5), The

Multinational in Geographic Space (June/July 2013), pp. 427-450.

 Gordon, I. R. (1999). Internationalisation and urban competition. Urban Studies, 36: 1001–

1016.

 Hall, P. V. (2007). Seaports, urban sustainability, and paradigm shift. Journal of Urban

Technology, 14(2), 87-101.

 Haralambides, H.E. (2012). Ports: Engines for Growth and Employment. European Ports

Policy Review: Unlocking the Growth Potential. European Commission, Brussels, 25 -26

September 2012.

 Head, K., & Mayer, T. (2004). Market potential and the location of Japanese investment in

the European Union. Review of Economics and Statistics 86 (4), 959 - 972.

 Jacobs, W., Ducruet, C., & De Langen, P. (2010). Integrating world cities into production

networks: the case of port cities. Global networks, 10(1), 92-113.

 Jungnickel, R. (1993). Recent trends in foreign direct investment, in Bailey, P., Parisotto, A.,

Renshaw, G. (eds.) (1993). Multinationals and employment: The global economy of the

1990s. Geneva. ILO.

 Karreman, B., Burger, M. J., & van Oort, F. G. (2017). Location Choices of Chinese

Multinationals in Europe: The Role of Overseas Communities, Economic Geography, 93:2,

131-161, DOI: 10.1080/00130095.2016.1248939

 Krugman, P. (1991a). Geography and trade. Cambridge, MA: MIT Press.

24

 Krugman, P. (1991b). Increasing returns and economic geography. J. Polit. Econ. 99, 483–

499.

 Krugman, P. (1993). The hub effect: or, threeness in international trade. In: Ethier, W.,

Helpman, E., Neary, P. (Eds.), Theory, Policy and Dynamics in International Trade.

Cambridge University Press, Cambridge.

 Lever, W. F. (1999). Competitive cities in Europe. Urban Studies, 36: 935–948.

 Li, S., Ragu-Nathan, B., Ragu-Nathan, T. S., & Rao, S. S. (2006). The impact of supply chain

management practices on competitive advantage and organizational performance.

Omega, 34(2), 107-124.

 Maggi, E., & Mariotti, I. (2012). Globalisation and the rise of logistics FDI: the case of Italy.

Foreign Investment: Types, Methods and Impacts, 29-60.

 Markusen, A. (1996). Sticky places in slippery space: A typology of industrial districts.

Economic Geography, 72(3): 293–313.

 Markusen, A. R., & Nesse, K. (2007). Institutional and political determinants of incentive

competition. In A. R. Markusen (ed.) Reining in the Competition for Capital, pp. 1–42.

Kalamazoo, MI: W.E. Upjohn Institute for Employment Research.

 Martin, C. (1992). Logistics and supply chain management, (pp. 4). Financial Times,

London.

 Musso, E. (2004). Porti, lavoro, economia: le regioni portuali di fronte alla rivoluzione

logistica. CEDAM.

 Nachum, L., & Wymbs, C. (2005). Product differentiation, external economies and MNE

location choices: M&A in global cities. Journal of International Business Studies, 36(4):

415–434.

 Niebuhr, A. (2006). Market access and regional disparities: new economic geography in

Europe. Ann. Regional Sci. 40, 313–334.

 Notteboom, T. E., & Rodrigue, J. P. (2005). Port regionalization: towards a new phase in

port development. Maritime Policy & Management, 32(3), 297-313.

 O’Connor, K. (1989). Australian ports, metropolitan areas and trade-related services. The

Australian Geographer, 20(2), 167-72.

 OECD, (2014). The Competitiveness of Global Port-cities. OECD Publishing, Paris

http://dx.doi.org/10.1787/9789264205277-en.

 Ottaviano, G. (2008). Infrastructure and economic geography: an overview of theory and

evidence. Eur. Bank Invest. Pap. 13, 8–35.

 Pettit, S. J., & Beresford, A. K. C. (2009). Port development: from gateways to logistics hubs.

Maritime Policy & Management, 36(3), 253-267.

 Porter, M. (2001). Regions and the new economics of competition. In A. Scott (Ed), Global

city-regions: Trends, theory, policy: 139–152. Oxford: Oxford University Press.

 Puga, D. (2002). European regional policies in light of recent location theories. J. Econ.

Geogr. 2, 373–406.

25

 Raines, P. (2003) Flows and territories: the new geography of competition for mobile

investment in Europe. In N. Phelps, P. Raines (eds) The New Competition for Inward

Investment, pp. 119–136. Cheltenham: Edward Elgar.

 Rietveld, P. (1994). Spatial economic impacts of transport infrastructure supply.

Transportation Research Part A: Policy and Practice, Elsevier, vol. 28(4), pages 329-341,

July.

 Robinson, R. (2002). Ports as elements in value - driven chain systems: The new paradigm.

Maritime Policy and Management, 19: 3–8.

 Rodrigue, J. P., Comtois, C., & Slack, B. (2013). The Geography of Transport Systems (pp.

142), Hofstra University, Department of Global Studies & Geography.

 Sassen, S. (1991). The global city: New York, London, Tokyo. Princeton, NJ: Princeton

University Press.

 Sassen, S. (2000). The global city: strategic site/new frontier. American studies, 41(2/3),

79-95.

 Sheffi, Y. (1985). Urban transportation networks. Englewood Cliffs, NJ: Prentice Hall.

 Slack, B. (1989). Port services, ports and the urban hierarchy. Tijdschrift voor Economische

en Sociale Geografie, 80(4), 236-243.

 Spiekermann, K., & Wegener, M. (2006). Accessibility and spatial development in Europe.

Scienze Regionali 5 (2), 15 - 46.

 Turok, I. (2004). Cities, regions and competitiveness. Regional Studies, 38: 1069–1083.

 UNCTAD, (1993). World Investment Report 1993: Transnational corporations and

integrated international production. New York. United Nations.

 Zhao, Q., Xu, H., Wall, R. S., & Stavropoulos, S. (2017). Building a bridge between port and

city: Improving the urban competitiveness of port cities. Journal of Transport Geography,

59, 120-133.