Comparison between port of Rotterdam and port of Amsterdam

(1)

Simon Majtlis 10281738 Thesis: Economics & Finance 1

Comparison between

Port of Rotterdam

and

Port of Amsterdam

Simon Majtlis, 10281738, Economics and Business – Economics & Finance

Supervisor: Swapnil Singh

(2)

Simon Majtlis 10281738 Thesis: Economics & Finance

Abstract

This paper is a comparison between the Port of Rotterdam and the Port of Amsterdam. Both are ports of significant economic value within the Hamburg – Le Havre port region. The Port of Rotterdam and the Port of Amsterdam act as important main ports for Western Europe in the fast moving and rapidly changing globalizing markets. Nevertheless, until now, relatively little (numerical) research on these ports has been conducted and published. For this paper we are researching whether the Port of Rotterdam and the Port of Amsterdam are creating synergy or negative competitiveness for one another. This research question is derived from a current OECD working paper by Merk & Notteboom (2013). The influence of general economic variables on port-specific variables is tested over the timespan 1980-2013. Data has been retrieved from the ports’, OECD’s and EUROSTAT’s databanks. By far not all data proofed usable for the concluding residual value correlation and suggestions for further research are made. Our research concluded that the answer to our main research question seems to be more of an expected future trend than a clearly outlined currently visible situation. Synergy between the Port of Rotterdam and the Port of Amsterdam is likely to occur as both ports aim to cooperate and in the future will be under even more competitive pressure.

(3)

3

1) Introduction

Throughout past centuries the Netherlands has been a country of traders. From the Golden Age onwards this meant trading across oceans with own ships and ports. Nowadays, in the 21st century, this is still the case and the Port of Rotterdam and the Port of Amsterdam play an important role in Western-European transport economics. In the last fifty years the maritime industry has changed the shape of the world economy (Stopford, 2003). Revolutions such as mechanized bulk transport and containerization have in an impressive way contributed towards globalization. The Port of Rotterdam and the Port of Amsterdam have both, irrespective of their size, profited from these developments. Throughout the last decades they have rapidly grown in size. Due to this growth both ports are now running up against their limits in terms of size (Port of Rotterdam, Wiegmans 2011) . With fierce competition in the Le Havre port range and changing global economic landscapes it is key for these ports to keep on developing and innovating. Not only for their own company profits but also for their contribution to local employment, regional economic impact and their fair contribution to the Dutch economy. To gain more market shares the Port of Rotterdam and the port of Amsterdam could compete each other or search for ways of synergy. As noted in Rotterdam’s Port Compass 2030: “Collaboration between the port of Rotterdam and other ports realizes advantage for customers and the community: this can be achieved by maximizing the use of resources and optimizing the usage of suitable waterways” (Port Compass 2030, 2011). This sounds positive and beneficial, nevertheless little is known about possible synergy or negative competition actually occurring between the Port of Rotterdam and the Port Amsterdam. The scope for this research will therefore be a competitiveness comparison between the Port of Rotterdam and the Port of Amsterdam. The specific research question is if the ports of Rotterdam and Amsterdam create synergy or negative competitiveness for one another?

This study contributes in an empirical way to an OECD research by Merk & Notteboom (2013). Merk & Notteboom’s working paper tries to give an evaluation of the current performance of the ports of Rotterdam and Amsterdam. They also try to quantify the effect of the two ports on economic and environmental questions. Summarized, the main scope of their research consists of “attempting to identify the impact of ports on their territories and possible policies to increase the positive impacts of ports on these territories” (Merk & Notteboom, 2013). Besides the OECD research, even though maritime academic research is

(4)

actively developing, little research has been performed on comparisons between the Port of Rotterdam and the Port of Amsterdam. Taking into account the effect both ports have on the Dutch economy this is remarkable. The lack of relevant research data in combination with the major positive economic effect of both ports gave cause for this thesis study. As will be elaborated further on, a general economic research method was applied to compare the ports of Rotterdam and Amsterdam. This general economic research method was chosen due to a lack of available port specific (statistical) research methods. By conducting extensive port-specific literature research a good direction was given on what data to collect in order to perform the research and answer our research question. The port-specific literature clarified which port-specific and general economic variables play an important role in measuring port performance. From there on, the necessary port-specific data was gathered by contacting the business intelligence units of the Port of Rotterdam and the Port of Amsterdam. Next to that, general economic variables used as independent variables were taken from databanks provided by the OECD and EUROSTAT.

With all the data provided, this study step by step worked towards a residual value correlation determining similarities between both ports. When reviewing these final outcomes of our research, statistics showed that both the Port of Rotterdam and the Port of Amsterdam achieve most of their overall transhipment, and income, through the port-specific incoming transhipment variable. The relevant individual regressions and correlated residual values showed strong dependency on the Dutch imports. The Port of Rotterdam and the Port of Amsterdam can thus mutually benefit from increasing Dutch imports. The exact (numerical) results of the research will be presented throughout the study; the remainder of this paper is therefore build up as follows. Section 2 discusses current academic maritime research and data. Section 3 presents the related literature. Section 4 gives the relevant variables and their underlying data. Section 5 provides the descriptive statistics, followed in section 6 and 7 with respectively the correlations and regression model. Section 8 concludes, discussing limitations of this study and giving suggestions to future research.

(5)

5

Research question

Do the ports of Rotterdam and Amsterdam create synergy or negative competitiveness for one another?

Sub Research question

Do the Port of Rotterdam and the Port of Amsterdam behave in the same cyclical direction when looking at important economically determining variables?

2) Current academic maritime research and data

As stated by Palls, Vitsounis & de Langen (2010), in our rapidly growing globalizing economy there is an increase in maritime (port-specific) economic research taking place. This increase in maritime (port-specific) economic research can be explained by rapid developments in the port industry (increasing world trade, new technologies and increasing private involvement). This academic research is often of immediate relevance and challenges the conventional organization and classification of ports. However, despite the currently increasing amount of research, many studies still only focus on a specific port or terminal. Also, within these academic studies, only a small amount of research is actually based on commodities other than containers (Palls, Vitsounis & de Langen, 2010). Little attention is being given to the different possible (macro) economic effects of ports performing in the same region or country, for instance economic effects between ports (negative competitiveness vs. synergy) and other economic effects with a more general economical undertone. The latter for example being: the mutual effect of Dutch GDP on transhipment for both the Port of Rotterdam and the Port of Amsterdam.

A relevant current research for this thesis is the OECD study by Merk & Notteboom (2013), which tries to explain port competitiveness between the Port of Rotterdam and the Port of Amsterdam. They do so by emphasizing on the economic impact of both of these ports within their regions and on the Dutch economy. To clarify: data provided by the Port of Rotterdam for instance states that the Port of Rotterdam accounts for 3 % of the Dutch national income. Merk & Notteboom (2013) also speak of the importance of possible synergy between ports in a certain region or country. How the existence of such a synergy should be measured is however not explained or emphasized. Finding an appropriate way of researching this would thus definitely be a contribution to research and a useful scope for this research.

(6)

Looking for a maritime economic framework to represent such a study, the following is of interest. When one takes Stopford’s (2003) writings on maritime economics, a significant amount of the graphical representations is derived from time-series correlations. Researching similar port-specific variables on a numerical statistical basis and from there on creating the necessary times-series correlations thus seems relevant. Our statistical research will consist of a similar outline. Eventually we will also be comparing residual values (discussed later) making way for a commonly used researching method within maritime economics.

Surprisingly few extensive data banks for collecting port specific data were publicly available, especially when one takes the size of the maritime economy into account. To gather my numerical research data contact therefore had to be made with both the Port of Rotterdam and the Port of Amsterdam. Both their business intelligence units and data collecting departments confirmed the relevance of, and showed enthusiasm towards, this research.

3) Literature Review

To come closer to eventually modelling our research question we will first have to look into some of the key drivers and characteristics of ports in general. More specifically, the Port of Rotterdam and the Port of Amsterdam will of course be emphasized and analysed. This literature review is not only background research on these ports, but goes further and also looks into determinants of port performance and competitiveness in general.

The role of ports in our transport system

Before focusing on specific characteristics and other background information on the Port of Rotterdam and the Port of Amsterdam, we will shortly look into the (evolving) role of ports in our transport system. Stopford (2003) states that ports are the essential interface between land and sea and that their main function consists of providing a secure location where ships can berth. A modern-day, high-efficiency port must be capable of handling all sorts of cargoes (bulk, containers, wheeled vehicles, general cargo, etc.). Being able to transport these different types of bulk through the port’s hinterland by railways, roads and domestic waterways also requires sufficient and functional infrastructure. Even though ports show many similarities, there is no such thing as a typical port (Stopford, 2003) This is something that will become clear throughout the following description and analysis on the Port of Rotterdam and the Port of Amsterdam.

(7)

7

Rotterdam

The Port of Rotterdam is well known for being a diversified, all-round hub with containers and liquid bulk being its main cargo categories (Port of Rotterdam, 2013). This results in Rotterdam being connected to a large number, if not all main container-shipping lines. According to research performed by Cesar Ducruet, based on data of Lloyd’s Maritime Intelligence Unit, there are 40 ports worldwide whose strongest direct traffic link is with Rotterdam. Rotterdam is thus one of the most central hubs in the container and liquid bulk industry (Merk & Notteboom, 2013). Over land, the Port of Rotterdam strikes an area of 10,500 ha and due to the so-called “Maasvlakte”, the port can accommodate the world’s largest freight vessels at any time. Rotterdam is in an outstanding economic position because of its expansion possibilities. Currently creating the “Tweede Maasvlakte” or Maasvlakte 2, and thus providing facilities for even more and larger vessels, Rotterdam is one of the few expanding ports in Europe (Wiegmans, 2011). Merk & Notteboom state that because of this, the Port of Rotterdam has the best nautical accessibility profile in northern Europe (2013). In 2010, the value added of the Port of Rotterdam represented 10.3 billion euros. This economic value represented quite a share of the regional GDP: 10.5% in 2008. Regional GDP, better known as Gross Regional Domestic Product (GRDP) is the aggregate of gross value added of all resident producer units in the region. It is a split up of the national GDP into several regions of a country focussing specifically on the agriculture, industry and transport sectors. The expansion of port value added by the Port of Rotterdam was actually quite small in the time span 2002-2010: growing by only 0.4% on average per year. This is far below the growing rates of port volumes that reached 4.1% over the same time span (Merk & Notteboom, 2013).

When looking into employment figures, it is interesting to note that port-related employment growth has basically remained flat for the last decade. Employment levels in the Port of Rotterdam reached 73.529 jobs in 2010. Of these jobs, the value added per worker is 90% higher compared to the metropolis Rotterdam (Merk & Notteboom, 2013). Concluding on their port analysis Merk & Notteboom (2013) state that despite Rotterdam’s dominant position in the intra-European and Far-Eastern market, Rotterdam has not proven capable of developing into a leading maritime centre, proving unable to attract mayor maritime law, finance and consultancy firms. These days shipbuilding is for instance often taking place in South Korea and China and no longer in cities throughout Europe. A specific reason for this is

(8)

not given. Relatively higher European loans could however very well be the answer to why this shift has taken place.

Amsterdam

As opposed to many other ports, the Port of Amsterdam’s main supply comes in the form of liquid bulk and dry bulk. Comparing the amount of liquid bulk between Amsterdam and Rotterdam, these quantities are quite similar. Very limited container traffic actually comes into Amsterdam. Amsterdam is a big player in some specific markets such as coal, cocoa and petrol. Being specialized in such a manner does make the Port of Amsterdam a unique port. Combining this specific infrastructure with the fact that the government of Amsterdam just recently (27-11-2014) has approved building a new sea lock gives the Port of Amsterdam an interesting market position in the Le Havre – Hamburg range for the upcoming decades (Port of Amsterdam, 2014).

Currently, especially when compared to Rotterdam, the diversity of the Port of Amsterdam’s maritime connections is relatively limited. Most of its port and coastal carrier connections are within Europe. (Merk & Notteboom, 2013). Unlike the Port of Rotterdam, which has almost finished developing land for a new port terminal in the North Sea (the so-called: “Tweede Maasvlakte”), Amsterdam is not growing away from the city centre. The port authority in Amsterdam is thus stabilizing the amount of land available. Meanwhile the city council has set significant targets for building houses in these same port areas (Wiegmans, 2011). Together, the port authority and city council hope to increase maritime traffic by enhancing the productivity of the current available port land (Port of Amsterdam, 2008).

The expansion of port value added was actually not substantial in the time span 2002-2010: growing by 2.3 % on average per year. Keep in mind that this is below the growing rates of port volumes that reached 5.6% over the same time span (Merk en Notteboom, 2013). The port region of Amsterdam accounts for almost 60000 jobs (direct and in-direct). For the last 10 years there has been zero measured growth in the amount of jobs available (Port of Amsterdam, 2014). Even though the growth in the amount of jobs available has been limited once again, the Port of Amsterdam distinguishes itself (as does the Port of Rotterdam) in their job density ratios. These levels are around 40% higher than their metropolitan areas (Merk & Notteboom, 2013)

(9)

9

Determinants of port performance, synergy and competiveness

As noted above, the ports of Rotterdam and Amsterdam prove to be specialized quite differently, each in their own type of bulk supply. These differences could create a situation/opportunities of/for synergy between both ports. Negative competiveness on the other hand could also very well be possible. According to Merk & Notteboom synergies will exist if the total outcome through cooperation of the two ports creates more net benefits than in the case where both ports operate in isolation (2013). A synergy is thus only possible if both ports are different to each other in some manners. This statement is further clarified in earlier research performed by Notteboom (2009) in his methodological note on complements and substitutes within gateway ports. From a neoclassical perspective Notteboom states that: “it can be argued that two load centres in the same multiport gateway region are perfect

substitutes for a port user if that user is willing to substitute one load centre for another at a

constant rate. Two load centres are perfect complements if they are always consumed together in fixed proportions by a port user” (2009). Measuring the complementary vs. substitutionary differences between these ports can relate to several areas including: scale, diversification, room for growth, location, nautical access, foreland orientation, hinterland orientation and exchanges over land (Merk & Notteboom, 2013, p 33). Specifications on several of these key drivers have already been discussed throughout the past literature review.

Ports prove to be very important assets for countries with high export and import ratio’s. The further a port is developed, the higher the profits for firms with high export and import ratios. To give an example: if the Port of Amsterdam and the Port of Rotterdam would be at a development level similar to ports in Brazil, the maritime transport cost would be 15 % higher than is currently the case (Merk & Notteboom, 2013), emphasizing the value for the Dutch economy with these ports as such valuable assets.

Apart from the previously discussed measurement variables of port performance, the ports of Amsterdam and Rotterdam have a significant economic impact on for instance the Randstad region. As noted by Merk & Notteboom, port economic impact studies generally look at valued added and employment data (2013).

(10)

4) Relevant variables and their underlying data

So far several aspects and the necessary background information on the Port of Amsterdam and the Port of Rotterdam have been described. It is now time to start aiming more at the exact research that will be conducted. As we recall, the research question we are trying to answer is whether the Port of Rotterdam and the Port of Amsterdam create synergy or negative competitiveness for one another? Answering this question on a quantitative statistical manner requires working only with variables useful for both ports. This way one eventually can create an equation of several variables applicable to both ports. The variables provided below are available for both the Port of Amsterdam and the Port of Rotterdam. This was a criterion in order for the variables to be useful for this research.

Between the chosen variables a distinction is made for port specific variables and more general underlying economic variables. For now, an explanation will only be given on why certain variables are chosen and what data lie behind. Applications of the data (log, growth size, dummy variables, etc.) and the specific statistical framework used for performing our analysis will be discussed separately. Statistical and graphical analyses will be performed on all data collected. Due to the difference in port size between Rotterdam and Amsterdam it could however very well be possible that our residual comparison (the concluding part of our research) will not consist of all the data collected up to this moment. All the collected variables will be summarized below and for the graphical and statistical analysis of the variables eventually not used in answering our research question we would like to refer to the appendix (part A).

Port specific variables and their underlying data

The relevant data underlying the port specific variables chosen have been obtained by directly contacting the Port of Amsterdam, the Port of Rotterdam and the European Sea Ports Organization (ESPO). Certain data have also been found online in the Eurostat and OECD databases. Most data are on a yearly basis, but interestingly enough the Eurostat database also provided quarterly data (1997-2013) on certain port specific variables.

(11)

11

Transhipment (1980-2013, x 1000 tonnes)

By definition transhipment is the offloading of a good onto another ship or means of transportation, like rail, road, etc (OECD, 2002). Since this is the pre-eminent function of a port it is quite useful for any port-specific or logistics research that total transhipment has become a measurable variable. Transhipment is a broad concept containing for example dry bulk, liquid bulk, containers, roll-on, roll-off, motor vehicles and other general cargo’s. Transhipment as such comprises an interesting bundle of goods because it contains many of the things we consume in modern society (food, energy resources, and consumption goods). In our research we will be comparing how transhipment correlates to for example Dutch GDP. For the Port of Rotterdam and the Port of Amsterdam, transhipment data is available and measurable on a yearly basis between 1980-2013.

Far more extensive research is possible when one analyses and compares each of the different transhipment subcategories separately, for example by looking at the amount of imported liquid bulk (oil) and comparing this to the oil price in the same period. Such extensive research is however beyond the scope of this bachelor thesis research and will be one of the suggestions to be made for further research.

Incoming, Outgoing Transhipment (1980 -2013)

Apart from total transhipment data, data is also available on incoming and outgoing transhipment. Incoming transhipment can be seen as a part of (Dutch) imports. Outgoing transhipment could on the other hand very well be taken into account as an indicator of country’s exports. Given that both the Port of Amsterdam and the Port of Rotterdam have provided me with these data, this could possibly result in interesting outcomes when compared to import and export numbers in the same 1980-2013 period. Incoming and outgoing transhipment thus gives a subtle deepening distinction for analysing transhipment.

(12)

Containers (1980-2013)

The amount of containers shipped through a port can be seen as a very precise indicator of fluctuations in the amount of goods coming into and leaving a port. Far more adjustable in size than bulk, containers seem to be by outset short run indicators on how an economy is performing. Containers are well measurable and thus a useful port-specific variable. Containers are generally measured in two ways either as a twenty-foot equivalent unit (TEU) or in gross weight. Yearly data on containers has been provided for both the Port of Amsterdam and the Port of Rotterdam in the period 1970-2011.

Amount of incoming maritime ships (1980-2013)

The amount of incoming ships, independent of their size or type of freight have been measured and made available for respectively the Port of Rotterdam and the Port of Amsterdam since, 1900 and 1980. On a first glance it is interesting to see how certain of the main historical events of the 20th century had their direct influence on the amount of ships coming into for example the Port of Rotterdam. During the World War II this average over 5 years was 974. Compared to 12.026 ships in 1939. These outliers do not only have to be of general historical importance. Within port development, certain infrastructural projects can take years if not decades to develop and put in use. Taking for example Rotterdam’s “Tweede Maasvlakte”, this progressive project started off in 2008 and will be fully finished and in complete use no earlier than 2030 (Port of Rotterdam, 2014). Sudden numerical changes in demand or supply therefore do not always require a general economical substantiation. Port specific developments as described above can also have their significance on the amount of incoming maritime ships. Nevertheless, the amount of incoming maritime ships should give a usable port specific variable different to transhipment and containers as earlier mentioned above. For our research, it is important that one takes into account the existence of such outliers. Taking all of these import events into account statistically is unfortunately beyond the scope of this thesis and will be considered as another suggestion made for further research.

(13)

13

General economic variables and their underlying data

The main parts of our research will consist of analysing to what extent the port specific variables of Rotterdam and Amsterdam correlate. It will, however, also be necessary to make proper comparisons with important general economic variables. Doing so will hopefully provide an explanation as to why certain fluctuations occur. Determining what are the most useful general economic variables to choose from in this case is important, especially because of the broad range of possibilities. Just as with the port specific variables, below one can find explanations of all the general economic variables taken into account for this research. For reaching our conclusion and accordingly answering our research question, certain of the economic variables have not been taken into account. Accordingly, the statistical meaning of these variables and their graphical and underlying numerical data can be found in the appendix (part A).

Gross Domestic Product – Dutch

According to a summarized definition provided by the OECD: “GDP is the sum of the final uses of goods and services within an economy measured in purchasers prices” this can be either on an annual or quarterly basis. In both cases, the GDP gives a broad impression on how an economy is performing. This is relevant for our research because hypothetically we could expect certain growth variables of ports to be behaving in the same direction as certain GDP’s. As mentioned before, the Port of Rotterdam is the main traffic link for 40 ports worldwide. Therefore, looking for a GDP variable that takes into account for example the world’s strongest exporting countries could also be of relevance to our research and is thus taken into account as a suggestion for further research.

International Trade (consisting of main economic indicators (MEI))

International trade justifies the existence of ports. Comparing fluctuations in international trade with certain port specific variables could thus possibly also be of relevance for answering our research question. The measurable data on international trade comes from the OECD and is divided up into imports and exports. The OECD makes a distinction and only takes so called main economic indicators into account when determining import and export numbers. In all OECD countries statistics on imports and exports (merchandise trade) are seen as important economic statistics. This dataset itself contains international trade statistics measured in billions of US dollars. Once again we have the possibility of using

(14)

annual data provided since 1980 and thus setting up time series data and according with relevant port specific variables.

5) Descriptive Statistics

By now we have done background research on our research subject, explained the variables we will be using and also collected the necessary data for these variables. We have therefore arrived at a point where our data set is going to be placed into a statistical framework. This process will consist of several steps leading to a multiple regression where we will eventually be analysing and comparing the residuals of both ports with regard to the variables found suitable. Explanation of the latter will follow further along this paper.

Steps in our statistical research: 1) Create logarithmic values

2) Note the descriptive statistics per variable (port specific, general economic) 3) Determine the correlation level between similar variables

4) Run the multiple regression 5) Collect the residuals

Create logarithmic values

The Port of Amsterdam is significantly smaller in size when being compared to the Port of Rotterdam. This is something that already came up in our literature review and even more clearly came up when analysing the provided data files. The answer to our main research question, i.e. do the ports of Rotterdam and Amsterdam create synergy or negative competitiveness for one another, can be found when statistically comparing the Port of Amsterdam with the Port of Rotterdam. This becomes difficult when only taking the nominal values into account. The following example should state why this is the case.

Taking the total transhipment data (x1000 tonnes) and randomly picking the year 1998 (in the range 1980-2013), we see respectively 313.663 (Rotterdam) and 55.808 (Amsterdam) as the given values. Once again this shows the greater magnitude of the Port of Rotterdam. Nevertheless, as Merk & Notteboom already mentioned in their OECD research, both the Port of Amsterdam and the Port of Rotterdam are of economic significance for their own region and mutual hinterland in providing jobs and transporting goods (2013). Comparing these ports

(15)

15

with one another could therefore show them being significant to for instance the Dutch economy.

So how does one compare data when the sizes differ so much of one another? To do this we first adjust our variable to become logarithmic. As stated by Stock & Watson: “another way to specify a nonlinear regression function is to use the natural logarithm of Y and/or X. Logarithms convert changes in variables into percentage changes, and many relationships are naturally expressed in terms of percentages” (2012)

Port-specific variables:

The port specific variable abbreviations stand for the following logarithmic variables:

- ISR (ISA) stands for incoming ships Rotterdam (Amsterdam). - TSR (TSA) stands for total transhipment Rotterdam (Amsterdam) - ITSR (ITSA) stands for incoming transhipment Rotterdam (Amsterdam) - OTSR (OTSA) stands for outgoing transhipment Rotterdam (Amsterdam) - TCR (TCA) stands for total containers Rotterdam (Amsterdam)

- TUER (TEUA) stands for total twenty-foot equivalent units Rotterdam

(Amsterdam)

ISR ISA TSR TSA ITSA OTSA ITSR OTSR TEUA TEUR TCA TCR

Obs ₃₄ ₃₄ ₃₄ ₃₄ ₃₄ ₃₄ ₂₅ ₂₅ ₂₄ ₃₄ ₃₄ ₃₄

Mean _10.336 _9.015 _{12.661 10.901 10.615 9.495} _{12.461 11.306 11.306 15.463 11.000 15.024} Std. Dev. _0.0447 _0.083 _0.189 _.456 _.444 _.510 _.124 _.270 _.670 _.572 _.516 _.520 Min _10.272 _8.833 _{12.363 9.764} _9.480 _8.367 _{12.293 11.052 10.468 14.469 10.365 14.086} Max _10.459 _9.131 _{12.998 11.469 11.118 10.292 12.654 11.786 12.986 16.290 12.495 15.787}

(16)

General-economic variables:

The two tables above give a summary of the, so-called, summary statistics used in this research. All values are logarithmic and based on the original data collected from the port of Rotterdam, the port of Amsterdam, and the online databanks provided by the OECD and EUROSTAT. All data is based on yearly time series in the period 1980-2013.The economic variables considered are GDP, Import and Export.

Time-series data

tsset Years, yearly

time variable: Years, 1980 to 2013 delta: 1 year

In time series data, a variable generally is correlated from one observation or date, to the next. In practice, economic time series often show growth that is approximately exponential (Stock & Watson, 2012). For this research all our data is represented as time series regression and computed as logarithms or changes in logarithms (growth rates).

GDP IM EX Obs 34 34 34 Mean 13.057 5.215 5.291 Std. Dev. 0.248 0.755 0.786 Min 12.666 4.108 4.171 Max 13.379 6.398 6.509

(17)

17

6) Correlation

The correlation between two variables (for example X and Y) is:

𝑐𝑜𝑟𝑟 𝑋, 𝑌 = 𝑐𝑜𝑣 𝑋, 𝑌

𝑣𝑎𝑟 𝑋) 𝑣𝑎𝑟(𝑌 = 𝜎𝑋𝑌 𝜎𝑋 𝜎𝑌

So why is correlation relevant to our research question which tries to answer whether or not the Port of Amsterdam and the Port of Rotterdam create synergy or negative competitiveness for one another? To answer this question explaining the relevance of correlation we will be looking at two aspects. First we should look at our sub research question asking if the ports of Rotterdam and Amsterdam behave in the same cyclical direction when looking at important economically determining variables? As stated by Berk & de Marzo (2013), correlation is: “a unit-free measure of the extent to which two random variables move together, or vary, together” This could very well be seen as a synonym for moving in the same cyclical direction.

So why is correlation so relevant to our main research question looking at possible synergies between these Dutch ports? An empirical investigation by Berkovitch & Narayanan (1993) gives a possible outcome. Their research looks into motives for takeovers and suggests that correlation between firm’s outputs should be positive if synergy is the motive for a takeover. At this point a takeover between the Port of Rotterdam and the Port of Amsterdam would seem highly unrealistic. Positive correlation between their output factors (for example: transhipment) would however imply possible synergy when two firms merge or entail a possible joint venture.

If correlation rate is: © Pearson Education

+.70 or higher Very strong positive relationship +.40 to +.69 Strong positive relationship +.30 to +.39 Moderate positive relationship +.20 to +.29 weak positive relationship +.01 to +.19 No or negligible relationship -‐.01 to -‐.19 No or negligible relationship -‐.20 to -‐.29 weak negative relationship -‐.30 to -‐.39 Moderate negative relationship -‐.40 to -‐.69 Strong negative relationship -‐.70 or higher Very strong negative relationship

(18)

Determine the correlation level between similar variables

When analysing time-series data sets, a starting point is plotting the data. In the following part these plots will be given for logarithmic data and (logarithmic) growth data. The specific variables for the Port of Amsterdam and the Port of Rotterdam have been grouped together and graphed with an accordingly relevant general economic variable. Apart from the graphical time-series representations, the necessary correlation levels have also been given. To get a clear understanding of what all the gathered data could mean for this research, initially a broad set-up had to be chosen. Graphical representations and correlations levels have therefore been designed for all the previously mentioned variables. This was necessary to be able to make a rational trade off, selecting which grouped variables would be able to provide sufficient economical explanatory values when trying to answer our research question. When analysing these graphical representations and correlations levels, two aspects are taken into account: 1) the correlation level for the specific port specific variable between the Port of Rotterdam and the Port of Amsterdam, and 2) the correlation level between the port specific variable and the general economic variables provided (for both the Port of Rotterdam and the Port of Amsterdam).

Determining the grouped variables for the residual research

To narrow down the research, analysing all the outcomes and determining the most relevant grouped variables was the next step. The chosen grouped variables for the eventual residual research are the following: 1) the incoming transhipment for Rotterdam and Amsterdam, compared with Dutch Imports and GDP as general economical variable, and 2) the outgoing transhipment for Rotterdam and Amsterdam with Dutch Exports and GDP. These two groups were chosen due to primarily their similar (high) logarithmic correlation levels and similarly positive logarithmic growth rates. The latter consisting of a bit more fluctuation, especially when looking at the logarithmic growth rates of outgoing transhipment for Rotterdam and Amsterdam compared to the general economic variables.

The grouped variables that did not meet up to the set combination of expectations can be found in the appendix (part A). Analysing these grouped variables is still worthwhile. The recent 2008 financial crisis is for instance clearly linked to graphical representations of the grouped variables.

(19)

19

Incoming Transhipment Rotterdam, Incoming Transhipment Amsterdam, Dutch Imports, Dutch GDP (1980-2013)

Correlate ITSR ITSA IM GDP (obs = 34)

ITSR ITSA IM GDP ITSR 1 ITSA 0.9362 1 IM 0.9759 0.9611 1 GDP 0.9126 0.9900 0.9493 1 4 6 8 10 12 14 1980 1990 2000 2010 2020 Years ITSR ITSA IM GDP

(20)

Correlate GITSR GITSA GIM GGDP (obs = 25)

NB: GITSR GITSA GIM and GGDP are yearly growth rates.

-. 2 0 .2 .4 .6 1980 1990 2000 2010 2020 Years GITSR GITSA GIM GGDP

GITSR GITSA GIM GGDP

GITSR 1

GITSA 0.5721 1

GIM 0.6238 0.4785 1

(21)

21

Outgoing Transhipment Rotterdam, Outgoing Transhipment Amsterdam, Dutch Exports, Dutch GDP (1980-2013)

correlate OTSR OTSA EX GDP (obs = 25)

OTSR OTSA EX GDP OTSR 1 OTSA 0.9678 1 EX 0.9351 0.9187 1 GDP 0.8345 0.8270 0.9513 1 4 6 8 10 12 14 1980 1990 2000 2010 2020 Years OTSR OTSA EX GDP

(22)

correlate GOTSR GOTSA GEX GGDP (obs = 25)

-. 2 0 .2 .4 .6 1980 1990 2000 2010 2020 Years GOTSR GOTSA GEX GGDP

GOTSR GOTSA GEX GGDP

GOTSR 1

GOTSA 0.3775 1

GEX 0.0912 0.1514 1

(23)

23

7). Regression model

“A (multiple) regression model permits estimating the effect on 𝑌_! of changing one variable 𝑋!! while holding the other regressors, in our case only 𝑋!!, constant” (Stock & Watson, 2012).

Single variable regression model: 𝑦 = 𝑋_!!𝛽 + 𝜀 Multiple regression model: 𝑦 = 𝑋_!!𝛽 + 𝑋_!!𝛽 + 𝜀

Regressions applicable to our research will give the effect of different general economic variables on the explanatory ports specific variable. A distinction will be made between singular and multiple regressions. Single variable regressions mean calculating the explanatory effect of independent general economic variables separately. Multiple regressions consist of a combination of both independent general economic variables trying to explain the dependent port specific variable. A possible multiple regression is for instance the effect of Dutch import and Dutch GDP on incoming transshipment for the Port of Amsterdam (either for the logarithmic variant or the growth rates).

Setting up a multiple regression does not mean it is directly useable for further use in this research. It needs statistically to be of significance and the general economic variable(s) need(s) to explain the port specific variable. The benchmark for this significance test will be put at a 5% for a two-tailed t-test, providing the following hypothesizes (𝐻_! vs. 𝐻_!).

𝐻_!: The independent general economic variable does not explain the dependent

port-specific variable.

𝐻_!: The independent general economic variable does explain the dependent port-specific variable.

Performing the regressions in STATA gave us a general ANOVA tables and from it we picked the indicators necessary to determine if the independent variable indeed showed explanatory value to the dependent variable. These indicators can be found in the tables below. Most important for determining significance in this case is the p-value.

Also the coefficient of determination is provided, making up for the squared correlation values (𝑅). The coefficient of determination (𝑅!_{) gives us the percentage of data points that}

(24)

fall on the regression line. The more data points that fall on the regression line, the higher the value of 𝑅!_{becomes. It is a useful measure of prediction in the case we are adding more data} points to the model and wondering on a percentage basis where they will end up. This is of interest due to the modest size of our data sets. Keep in mind however that an increase in the sample size will always lead to a certain increase of the 𝑅!_{values and is not always} significantly correct.

When taking the grouped variables provided earlier and running all the (multiple) regressions for either the Port of Rotterdam or the Port of Amsterdam, the singular logarithmical regressions seem (highly) significant for both ports. On the other hand, the grouped multiple regressions and the growth rates that looked promising when taking the previous correlations into account appear not to be significant in all cases an will therefore not be taken into account for the determination of the residual values. The multiple regressions with both GDP and import/export as independent variables have all been placed in the Appendix Part B. What could be possible explanations for this absence of sufficient significant regressions? There could be numerous reasons, one of them being a lack of data. The research method applied does seem to fit the scope of this research. A suggestion for further research would therefore also be, applying the same method but with a wider data range that will hopefully become available as maritime academics and statistical databanks keep on rapidly developing. For now we will be focusing on the logarithmical singular regressions provided below. It will be these logarithmical regressions that will be used for explaining the residual values and thus possible overlapping characteristics of both ports. The latter needed for answering our research question.

(25)

25

Incoming Transhipment separately regressed with Dutch GDP and Dutch Imports (1980-2013)

This table represents the results of the regression analysis, using only singular logarithmic regressions. The table contains the data of 4 separate regressions each proving to be significant through their p-values and thus usable for further residual value correlations. The regression coefficient mentioned is the change in response per unit change in the predictor. The higher this number, the more influence this independent variable has on changing the dependent’s variable outcome either positive or negative. Interesting to see is the comparison of GDP vs. IM coefficients.

The standard errors are the standard errors of the regression coefficients. They can be used for hypothesis testing and constructing confidence intervals (Stock & Watson, 2012). Referring once more to the p-values, compare the values here with the grouped regressions placed in the appendix and note the high significance of the singular logarithmic regressions. Finally, the 𝑅!_{is very high and suggests predictive capability. Still, as}

mentioned earlier, 𝑅!_{should always be used in combination with other testing parameters to present reliable}

results.

ITSR Rotterdam Amsterdam

Coefficient (Std. Err.) p-value 𝑅! Coefficient (Std. Err.) p-value 𝑅! GDP .667 (.062) 0.000 0.8328 1.615 (0.136) 0.000 0.8161 IM .210 (0.010) 0.000 0.9523 .504 (0.053) 0.000 0.7354

(26)

Outgoing Transhipment independently regressed with Dutch GDP and Dutch Exports (1980-2013)

This table represents the results of the regression analysis, once again using only singular logarithmic regressions. For the Port of Rotterdam and the Port of Amsterdam Dutch GDP and export (EX) data have been regressed with the port specific outgoing transshipment data. Relevant to mention is the difference between coefficients when comparing GPD vs. EX for both Rotterdam and Amsterdam. GDP seems to be of a greater influence than EX in for both the Port of Rotterdam and the Port of Amsterdam. Also interesting to look at is how the standard errors are build up or the different coefficients, showing more fluctuation for the GDP coefficient in comparison with EX. The p-values are again (highly) significant and these regressions are all thus being taken into account for the residual value correlation worked out below.

Residual value correlation

One of the first made statements of this research emphasized the difference in size of the Port of Rotterdam and the Port of Amsterdam. Nevertheless our research question and sub-research question consisted of working towards a data set were one could compare and analyze the ports with one another as if they are equal (to a certain extend). To do so we have

been working with statistical applications and adjustments that will let us compare the Port of Rotterdam with the Port of Amsterdam. During this process, quite some of our

OTSR Rotterdam Amsterdam

Coefficient (Std. Err.) p-value 𝑅! Coefficient (Std. Err.) p-value 𝑅! GDP 1.323 (.182) 0.000 0.6964 1.751 (0.190) 0.000 0.7266 EX .423 (0.033) 0.000 0.8744 .570 (0.055) 0.000 0.7354

(27)

27

starting data eventually proved not to be sufficiently usable. Incoming and outgoing transshipment data has shown to be of use and will be continued to work with. Hopefully, if the research method shows of relevance, for future research with more extensive data the same methodology can be applied. So what does this methodology entail?

The final methodology will consist of setting up a correlation between residual values taken from the different regressions. With these residual outcomes and together with the background information presented throughout this research, possible synergy between the Port of Rotterdam and the port of Amsterdam might be displayed. Before presenting the research results some information will be given on the research method applied.

After in the previous section having determined the regression statistics we will now be focusing on the correlation between the residual values coming forth from these regressions. Combinations made will be for instance, the effect of GDP on the residual values of incoming transshipment for the port of Amsterdam and the port of Rotterdam.

The residual value can be seen as that what the independent variable failed to explain. The residual values for both the Port of Rotterdam and the Port of Amsterdam will be correlated with each other. High correlation between these residuals might indicate an extra common significant variable that both the ports have in common. Even though the port of Rotterdam and the port of Amsterdam differ in many ways, whatever possibly correlates between the residual values relates the ports and is what makes them unique and comparable. If residual correlation levels are high then there is a possible other variable that could correlate with both ports. Keep in mind that this is off course a modeling and thus a simplified version. The same principle should however be able to be applied to a much further extent, when cautiously adding more general economic variables.

The following syntax has been computed to determine the residual values necessary for the correlations found below. Different general economic variables (‘var’) were used to run the different regressions. A combination of both (“IM & GDP” or “EX & GDP”) trying to further explain possible correlation showed significance problems in the multiple regression and has therefore been, as mentioned before, left out as a possibility.

(28)

Analyzing the residual correlation results

Incoming Transhipment Rotterdam/Amsterdam, Dutch GDP (1980-2013)

correlate ITSA_ ITSR_res (obs = 25)

The residual correlation level when taking Dutch GDP as the independent variable gives a moderate positive relationship. The general correlation between incoming transhipment Rotterdam, incoming transhipment Amsterdam and Dutch GDP provided earlier in this research showed very strong positive relationships. Generalized this could suggest that GDP takes a reasonable part of the fluctuations for the incoming transhipment of the Port of Rotterdam and the Port of Amsterdam for its account.

ITSA_res ITSR_res

ITSA_res 1

ITSR_res 0.3553 1

generate ITSA_res = . //ITSA residual generate ITSR_res = . //ITSR residual foreach var of varlist ITSA ITSR {

regress `var' //IM or GDP ( predict temp, resid

replace `var'_res = temp drop temp

}

correlate ITSA_res ITSR_re © Swapnil Singh

generate OTSA_res = . // OTSA residual generate OTSR_res = . // OTSR residual foreach var of varlist OTSA OTSR {

regress `var' //EX or GDP predict temp, resid

replace `var'_res = temp drop temp

}

correlate OTSA_res OTSR_res © Swapnil Singh

(29)

29

Incoming Transhipment Rotterdam/Amsterdam, Dutch Import (1980-2013).

correlate ITSA_ ITSR_res (obs = 25)

The residual correlation level when taking Dutch Import as the independent variable gives no or no negligible relationship between the residuals. The general correlation between incoming transshipment Rotterdam, incoming transshipment Amsterdam and Dutch Import provided earlier in this research showed very strong positive relationships. This would suggest that Dutch Import takes almost all mutual effects of incoming transshipment for its account, given the fact that the correlation between both residuals is now almost zero. When reasoning there rests logic in suggesting that Dutch import should represent a considerable part of the incoming port transshipments. Some caution with this statement can however do no harm and further research with a more extensive data set would give more clarification on this interesting result.

Outgoing Transhipment Rotterdam/Amsterdam, Dutch GDP (1980-2013)

correlate OTSA_ OTSR_res (obs = 25) ITSA_res ITSR_res ITSA_res 1 ITSR_res -0.0202 1 OTSA_res OTSR_res OTSA_res 1 OTSR_res 0.8909 1

(30)

Outgoing Transhipment Rotterdam/Amsterdam, Dutch Export (1980-2013)

correlate OTSA_ OTSR_res (obs = 25)

In both cases analyzing the two different outgoing transshipment residuals (Dutch GDP: 0.8909 Dutch Export: 0.7713) the residual variables show a high level of positive correlation. What do these high residual value correlations tell us at this moment? In earlier correlations dealing with outgoing transhipment we also came across high positive relationships. Given the fact that even after cutting out respectively Dutch GDP and Dutch export these levels are still relatively high (just a slight decrease) probably suggests that there is another variable hidden in the residual values. Determining what variable this requests extensive research and has to be applied very subtle. In the Appendix (Part B) it for instance shows how the creating multiple regressions clearly changes significance levels. It makes earlier significant (significant) variables incapable of explaining the dependent variable when part of a multiple regression.

8) Conclusion

The main purpose of this research was setting up a competitiveness comparison between the Port of Rotterdam and the port of Amsterdam. The relevance for doing this came from a recent OECD research by Merk & Notteboom (2013). Their research discusses the situation in the Netherlands where two ports serve the same economy and a similar hinterland. In their paper the possibility of an occurring synergy between these ports is shortly mentioned. Interestingly enough however no proof or way to determine such a synergy is given. Trying to determine such a synergy gave scope to my research and led to the following main research question being: do the ports of Rotterdam and Amsterdam create synergy or negative competitiveness for one another? To eventually come to answering this main

OTSA_res OTSR_res

OTSA_res 1

(31)

31

research question, answering a sub research question was necessary: do the Port of Rotterdam and the Port of Amsterdam behave in the same cyclical direction when looking at important economically determining variables? The proper general economical variables and port specific variables to take into account came forward throughout the literature research and were mainly collected after contact with the Port of Rotterdam and the Port of Amsterdam. The relevant data was graphed on 1980-2013 time series and correlated with one another. When analysing the time-series data behaviour in the same cyclical direction seems evident for both the Port of Rotterdam and the Port of Amsterdam. A suggestion for future research on this aspect would be the addition of important port specific events into the time-series. A critical note to be put in place is the fact that we are dealing with yearly data. Where general economic data is widely available on a quarterly, monthly or even daily basis, the port-specific data is currently only provided on a yearly basis. The regressions would have yielded more accurate results when the port specific variables would have been available on a more frequent basis. This is something that will probably be possible in the near future due to rapid developments in the area of maritime economics (Palls, Vitsounis & de Langen, 2010). Coming to our main research question the answer to it comes from a combination of findings in the literature and the statistical results. Also, the answer to our main research question seems to be more of an expected future trend than a clearly outlined currently visible situation. Synergy between the Port of Rotterdam and the Port of Amsterdam is likely to occur and its origin is most likely found within the incoming transhipment data. This is twofold. First of all, both the Port of Rotterdam and the Port of Amsterdam achieve most of their overall transhipment, and income, through incoming transhipment. The relevant individual regressions and correlated residual values showed strong dependency on the Dutch imports. Both ports therefore benefit from increasing Dutch imports.

Second of all, literature suggests increasing competition between ports in the Le Havre Range. This increasing competition could possibly mean diminishing Dutch imports. The port of Rotterdam and the port of Amsterdam seem to counteract this by encouraging co-operation and joint investments on several different business aspects (explained throughout there so called Port Compass 2030). Taking the previously two mentioned factors into account in many ways suggest a highly more probable synergy than negative competition. Off course this is something that will have to be proven by research taking place in the near future.

(32)

Limitations, suggestions and discussion

Apart from certain limitations to my own research, this concluding part will also provide suggestions for future research and give way for discussion about maritime academic data. The main limitations of this research lie in the fact that we have only been working with yearly data. Even though it dated 1980 onwards, the fact that it is yearly data gave certain boundaries. We are speaking of boundaries because nowadays especially for certain macro economic variables very precise data is available. Given the economic size, the amount of employment and the revenue made by both the Port of Rotterdam and the Port of Amsterdam it is therefore surprising they are still only working with yearly data. Due to innovation this is also probably bound to change.

Referring back to our research, there are many more extensive ways to perform the data steps we have taken. Interesting distinctions can for instance be made between when transshipment is looked at per commodity. In that case, crude oil transshipment could for example be correlated with the oil price over a certain times series. Doing this for different commodities is at this point, unfortunately, beyond the scope of this research.

Finally, pointing out the specific statistical way this research has been computed; one notes that it is very well applicable on other ports. So forth making it a very scalable way of doing research. On top of that the residual value correlation also seems very useful in numerically tracking policy changes. This does not specifically have to mean changes by ports, but can once again be applied on al sorts of macro economical questions. When applying the residual value correlation in those cases addition of more explanatory variables would be of contribution. As seen in this research, one has to keep in mind that this is something that has to be done subtle. Otherwise unbalancing the regression up to that point.

(33)

33

Bibliography

Balassa, B. (1978). Exports and economic growth: further evidence. Journal of development Economics, 5(2), 181-189.

Berk, J., & DeMarzo, P. (2007). Corporate finance. Boston: Pearson Addison Wesley.

Berkovitch, E., & Narayanan, M. P. (1993). Motives for takeovers: An empirical investigation. Journal of Financial and Quantitative analysis, 28(03), 347-362.

Merk, Olaf, and Theo Notteboom. "The Competitiveness of Global Port-Cities: The Case of Rotterdam/Amsterdam – the Netherlands." OECD Regional Development: Working

Papers (2013): 1-114.

Pallis, A. A., Vitsounis, T. K., & De Langen, P. W. (2010). Port economics, policy and management: Review of an emerging research field. Transport Reviews, 30(1), 115-161

Stock, J., & Watson, M. (2012) Introduction to econometrics (3rd ed., p. 832). Boston: Pearson/Addison Wesley.

Stopford, M. (1997). Maritime economics (2nd ed., p. 593). London: Routledge.

Theo Notteboom, “Complementarity and substitutability among adjacent gateway ports”

Enviroment and Planning 41.3 (2009): 743 – 762

Wiegmans, Bart. "Changing Port–city Relations at Amsterdam: A New Phase at the Interface?" Journal of Transport Geography 19.4 (2010): 575-583.

(34)

News articles

De Kiewit, A. (2014, November 27). Gemeenteraad zegt ‘Ja’ tegen nieuwe sluis. Economic Significance. (2014). Port of Amsterdam: Factsheet Edition 3, 4-4.

Havenvisie 2030: Voortgangsrapportage 2014. (2014). Port Compass, 1-44. Port of Rotterdam

Havenvisie 2030 (2011). Port Compass., 1-110. Port of Rotterdam

Haven in cijfers: 2011-2012-2013. (2014). Havenbedrijf Rotterdam NV, 20-20.

Slimme Haven. (2008). Havenvisie Gemeente Amsterdam 2008 – 2020, 44-44.

Schrijver, J., Op de Beek, F., Bleijenberg, A., Immers, B., De Kieviet, M., Passier, G., Tavasszy, L (2008). Visies op verkeer en vervoer 2020 - 2033. Tweede Maasvlakte, (1), 0 - 48.

Rotterdam: The most important port in Europe. (n.d.). Retrieved January 15, 2015, from: https://www.maasvlakte2.com/en/index/show/id/682/economic-importance

(35)

35

Appendix

PART A.

Transhipment Rotterdam, Transhipment Amsterdam, GDP (1980-2013)

correlate TSR TSA GDP (obs=34) TSR TSA GDP TSR 1 TSA 0.8333 1 GDP 0.9303 0.9039 1 10 11 12 13 14 1980 1990 2000 2010 2020 Years TSR TSA GDP

(36)

Simon Majtlis 10281738 Thesis: Economics & Finance correlate GTSR GTSA GGDP (obs=33) GTSR GTSA GGDP GTSR 1 GTSA -0.0441 1 GGDP 0.4681 0.1875 1 -. 2 0 .2 .4 .6 1980 1990 2000 2010 2020 Years GTSR GTSA GGDP

(37)

37

Incoming Transhipment Rotterdam, Incoming Transhipment Amsterdam, Dutch Imports, (1980-2013)

correlate ITSR ITSA IM (obs=25) ITSR ITSA IM ITSR 1 ITSA 0.9362 1 IM 0.9759 0.9611 1 4 6 8 10 12 1980 1990 2000 2010 2020 Years ITSR ITSA IM

(38)

correlate GITSR GITSA GIM GGDP (obs = 25)

GITSR GITSA GIM GGDP

GITSR 1 GITSA 0.5721 1 GIM 0.6238 0.4785 1 GGDP 0.4004 0.5681 0.5118 1 -. 2 0 .2 .4 .6 1980 1990 2000 2010 2020 Years GITSR GITSA GIM

(39)

39

Outgoing Transhipment Rotterdam, Outgoing Transhipment Amsterdam, Dutch Exports (1980-2013)

correlate OTSR OTSA EX (obs=25) OTSR OTSA EX OTSR 1 OTSA 0.9678 1 EX 0.9351 0.9187 1 4 6 8 10 12 1980 1990 2000 2010 2020 Years OTSR OTSA EX

(40)

correlate GOTSR GOTSA GEX (obs=24)

GOTSR GOTSA GEX

GISR 1 GOTSA 0.3775 1 GEX 0.0912 0.1514 1 -. 2 0 .2 .4 .6 1980 1990 2000 2010 2020 Years GOTSR GOTSA GEX

(41)

41

Incoming Ships Rotterdam, Incoming Ships Amsterdam, Dutch Imports (1980-2013)

correlate ISR ISA EX (obs=34) ISR ISA IM ISR 1 ISA 0.0576 1 IM 0.0376 -0.0555 1 4 6 8 10 1980 1990 2000 2010 2020 Years ISR ISA IM

(42)

correlate GISR GISA GEX (obs=33)

GISR GISA GIM

GISR 1 GISA 0.1096 1 GIM 0.4482 0.1003 1 -. 3 -. 2 -. 1 0 .1 .2 1980 1990 2000 2010 2020 Years GISR GISA GIM

(43)

43

TEU’s Rotterdam, TEU’s Amsterdam, Dutch GDP (1980-2013)

correlate TEUR TEU GDP (obs=24) TEUR TEU GDP TEUR 1 TEUA 0.2884 1 GDP 0.9723 0.2523 1 0 5 10 15 20 1980 1990 2000 2010 2020 Years TEUR TEUA GGDP

(44)

correlate GTEUR GTEUA GGDP (obs=24) GTEUR GTEUA GGDP GTEUR 1 GTEUA -0.0741 1 GGDP 0.4701 0.2521 1 -1 -. 5 0 .5 1 1.5 1980 1990 2000 2010 2020 Years GTEUR GTEUA GGDP

(45)

45

PART B

Multiple regressions

Outgoing Transhipment Rotterdam Amsterdam, pooled Dutch Exports and Dutch GDP (1980-2013)

OTSR Rotterdam Amsterdam

Coefficient (Std. Err.) p-value 𝑅!_{: 0.9062} Coefficient (Std. Err.) p-value 𝑅!_:0.7712 GDP -0.918 (.336) 0.012 -.239 (0.829) 0.775 EX .6727 (0.096) 0.000 .644 (0.262) 0.020

(46)

Incoming Transhipment Rotterdam Amsterdam, pooled Dutch Exports and Dutch GDP (1980-2013)

ITSR Rotterdam Amsterdam

Coefficient (Std. Err.) p-value 𝑅!_{: 0.9543} Coefficient (Std. Err.) p-value 𝑅!_:0.8261 GDP -0.102 (.106) 0.345 2.389 (0.594) 0.000 IM .239 (0.031) 0.000 .261 (0.195) 0.191

(47)

47

Growth Rates Incoming Transhipment Rotterdam, pooled Dutch Imports and Dutch GDP (1980-2013)

NB: Notice the high p-value explaining the low explanatory value of the variables. Making them un-useful in a certain way.

GITSR Rotterdam Amsterdam

β (Std. Err.) p-value 𝑅!_{: 0.3980} β (Std. Err.) p-value 𝑅!_:0.0875 GGDP -0.265 (.476) 0.583 2.193 (1.309) 0.104 GIM .248 (0.265) 0.009 -0.257 (0.232) 0.277