Import : An Analytic and Econometric Study of Regions

J

I

B

S

JÖNKÖPI NG UNIVERSIT Y

Import

- An Analytic and Econometric Study of Regions

Paper within Economics Author: Josefine Seydlitz

Tutor: Prof. Charlie Karlsson and Phd Martin Andersson Jönköping June 2007

Magisteruppsats inom Nationalekonomi

Titel: Import – En Analytisk och Ekonometrisk Studie av Regioner Författare: Josefine Seydlitz

Handledare: Charlie Karlsson and Martin Andersson Datum: 2007-02-07

Ämnesord: Import, Export, FoU, Tillväxt, Universitets Utbildning, Regioner

Sammanfattning

Denna unika studie om import är av största vikt, eftersom inga studier på import över regi-oner tidigare utförts. Import är en av de viktigaste drivkrafterna bakom innovatiregi-oner och därför mycket viktigt för en stabil tillväxt. Denna magisteruppsats handlar om förekomsten av interregional import och den rumsliga koncentrationen av import i det svenska import-och exportnätverket.

Fem hypoteser presenteras i den sista sektionen av kapitel två. De variabler som valts ut att användas i analysen är uppdelade i två huvudgrupper, på vilka emiriska test sedan utförts i olika kombinationer av regressions modeller.

Den främsta slutsatsen av denna studie är ett signifikant samband mellan import och de två oberoende variablerna export och företags forskning och utveckling (företags FoU), vilket går helt i linje med uppsatsens teoriavsnitt, regional specialisering i import noder och ex-port noder.

Master’s Thesis in Economics

Title: Import – An Analytic and Econometric Study of Regions Author: Josefine Seydlitz

Tutor: Charlie Karlsson and Martin Andersson Date: 2007-02-07

Subject terms: Import, Export, R&D, Growth, University Education, Regions

Abstract

This is unique study of import is of the greatest importance, since no studies of import across regions have earlier been performed. Import is a driving force of innovations and therefore most important for a stable growth. This master thesis is about interregional im-port, as well as the strong spatial concentration of imports in the Swedish system of net-work.

Five hypothesises are presented in the last section of chapter two. The variables to be used in the analysis are then divided into two main groups, before empirically tested in different combinations of regression models.

The main conclusion of this thesis is a significant correlation between import, and the two independent variables export and firm R&D, and its result goes in line with the theoretical framework of this thesis, regional specialisation in import and export nodes.

Table of Contents

Sammanfattning...1

Abstract ...2

Tables ...4

Introduction...5

Introduction...5

1 Regional Specialisation in Import and Export Nodes ...7

1.1 Theory of Regional and Urban Economies ... 7

– An Import and Export Perspective ... 7

1.2 The Neoclassic Trade Theory... 9

1.3 The New Trade Theory... 9

1.4 Import and Export Nodes... 10

1.5 Product Development ... 12

1.5.1 The Product Life Cycle ... 12

1.6 Hypotheses... 14

2 Data Presentation and Empirical Analysis ...15

2.1 The Features of the Data... 15

2.2 Cross Tab Tests ... 16

2.3 The Empirical Results... 18

3 Conclusions and Suggestions for Further Studies ...22

References ...23

Figures

Figure 1: Bid-Rent Model... 8

Figure 2 : Central Place System... 8

Figure 3: The Role of a Central Import Node... 11

Figure 4: The Growth of an Import Node... 12

Tables

Table 2: Descriptive Statistics... 16

Table 3: The Results of Model 1 and Model 2... 19

Table 4: The Results of Model 3 and Model 4... 20

Introduction

Swedish import is skewed1_{, 2003, 44 percent of all import is concentrated in the Stoc kholm} LA region2 _{of Sweden. This is to be compared to its 22 percent of the Swedish population.} Moreover, the Stockholm LA region has a high share of the country’s firm and university R&D activity, 58 percent, as well as the university-educated population, 31 percent, and the Swedish exports, 28 percent (SCB, 2003). Stockholm was the largest import node of Swe-den, already in the 1600. It then received international impulses primarily through Amster-dam, which was the node to take care of most North European imports. The following century even Gothenburg took care of Swedish imports and in 2003 Gothenburg was the second largest import node of Sweden. It then had 17 percent of Swedish imports to be compared to its 10 percent of the population (Johansson, 1993; SCB, 2003). Why is all this activity concentrated in Stockholm?

The world economy consists of a network of import and export nodes, where there are a numerous of export nodes per import node. Import nodes are often large and located in city areas. It is the responsibility of the import nodes to supply export nodes with all the different kinds of world news. They assess and develop new ideas from the international market and then sell this information and new products further to its export nodes. Im-ports can be consumption, parts of products, and wholesale, as well as retail. Export nodes on the other hand, are often smaller and specialised in production of one or a few goods. Their development and the development of their goods and services are most dependent on the supply of the import nodes. But how do import respectively export nodes come up? How come regions specialise, and what determines its specialisation? These are questions to be answered in this thesis.

When one or a few import companies establishes in a region, it can be seen as knowledge is brought to the region, which increases the market potential of this area. As a result, more companies who want to take part of this knowledge, as well as companies who see business opportunities in the already located companies will also locate their business in the same area. As a result the area grows by itself, because of economies of scale.

R&D activity is a complement to import activity, which often makes them concentrate in the same business area. Imports, as well as R&D are much important for product devel-opment, since they keep consumers more interested and active. Increased aggregate de-mand also enhances the turn over of goods, and therefore wealth increases. An exchange of new ideas, goods and services between regions, i.e. interregional trade, bring about an increase of different varieties in respectively region. As a result, import is important for the wealth of a country, both at a national and interregional level. What determines the location of production, and how is interregional trade made efficient? These are further questions to be answered in this thesis.

Consequently import is of the greatest importance for the world development. Its impor-tance is confirmed by the large field of empirical studies of import. Different national studies suggest that import demand is explained by GDP/real GDP, the price level and the income level (Dutta’s and Ahmed, 1994; Deyak, Sawyer and Sprinkle’s, 1989; Tsionas and

1 _{See Appendix 1 for a Lorenz curve, where import is compared to population and export.}

2 _{Labour Market (LA) Region is defined as a group of municipalities with a common labour market; see}

Christopoulos; 2004). Cole’s study (1995) proposes that import demand is price sensitive and easily substituted.

However, no studies of import across regions are found in databases as EconLit and WOPEC. This is therefore a unique study by its research of different driving forces of im-port across regions, as well as its research of the spatial concentration of imim-port in the Stockholm LA region.

1.1 Purpose

The purpose of this study is to analyse what features that explain the appearance of import across regions, as well as explain the strong spatial concentration of imports in the Swedish system of network.

1.2 Outline

The chapter 2 is a theoretical framework of regional specialisation in import and export nodes, providing an explanation of interregional trade, and the correlation between import and export nodes. It starts off by a presentation of regional and urban economies. This is followed by Neoclassic and New trade theory with clarification of expressions like com-parative advantages, economies of scale and specialisation all in a perspective of import. Moreover, the role and function of import and export nodes, as well as their network are taken up. Furthermore, the product development is explained in a perspective of import, and finally five hypothesises are presented, which follows by the chapter of empirical stud-ies.

Chapter 3 starts off by a presentation of the variables and the models that are used. This is followed by a presentation of the result of the empirical analysis. The final and fourth chapter is a summary of the study and its results. This last chapter does also give sugges-tions for further studies.

1 Regional Specialisation in Import and Export Nodes

– A Theoretical Framework

This chapter provides an explanation of interregional trade by describe the structure of the economy, the implications of different locations and the causes of specialisation.

1.1 Theory of Regional and Urban Economies

– An Import and Export Perspective

This section deals with interregional trade; from the willingness to pay for a certain loca-tions, and the spatial supply of goods and services. The monocentric city model and the central place theory make out the basis of this analysis.

The monocentric city model is developed from the work of Von Thünen. He made up a bid-rent equilibrium to explain the willingness to pay land rent against transport cost, in or-der to minimise cost. The starting point was a flatten area of land with a city centre in the middle. In the city centre the transport costs are minimised, which increases the demand for such location. As a result, the price level increases as the distance to the city centre de-creases. A similar reasoning can be applied on modern economic situations, by substitute a city centre for a central business area, and farmers for commuters. Figure 1, illustrates the willingness to pay for a certain location, according to rent (y-axis) and distance from city centre (x-axis). According to the three different curves in the graph, flower farmers are the most willing to pay high rent for a location in the city centre, and the grain farmers are the least willing. In addition, the outer curve shows that the flower farmer will sell the smallest quantity and the grain farmer the largest. This creates competitive environment, which ac-cording to Von Thünen will lead to an efficient equilibrium. Figure 1b, shows the same principle but is illustrated by a circle, where flower farmers are located in the city centre and the grain farmers the furthest from the city centre. As a result, goods and services will become subject for interregional trade, i.e. goods and services that are not supplied in the city centre have to be imported, as well as goods and services supplied only in city centre will be exported (Brakman, Garretsen and Marrewijk, 2001). The distribution of goods and services will be explained in the following section of the central place theory.

Figure 1: Bid-Rent Model

Source: Brakman, Garretsen and Marrewijk, 2001, page 25

Another well-known model of regional economics is the central place theory. The central place theory gives an understanding to the extraction of interregional trade. It assumes that the variety of supply is determined by the aggregate demand, which depends on the size of the region. This means that smaller regions need to import from larger regions to fulfil the demand of its population. The Figure 2 illustrates the central place theory by a spatial dis-tribution of regions in a recurrent pattern of hexagons. There is a recurrent pattern of large city regions surrounded by regions with gradually decreasing sizes further away from the large city region (Brakman, Garretsen and Marrewijk, 2001).

Figure 2 : Central Place System

Source: Brakman, Garretsen and Marrewijk, 2001, page 33

According to Karlsson and Nilsson (2002) fixed cost is the main explanation to a smaller supply of varieties in small regions. This is so since the demand of a small region will not cover a much variable supply; industries with high fixed costs will need a high demand to cover its fixed costs. Consequently, only daily goods are supplied in the smallest regions.

Village Small city Large city

flowers vegetables

grain b.

Distance from city center Bid rent curves flowers vegetables grain A B

flowers vegetables grain a.

The supply of different varieties grows with the size of the region, which means that the largest regions can afford the most differential supply with rarely consumed goods. And as a result, regions trade to increase the variety of supply.

1.2 The Neoclassic Trade Theory

According to the neoclassic trade theory, trade is explained by comparative advantage due to cross-country differences in factor endowments, for example access to water, altitude, climate, natural resources (Brakman, Garretsen and Marrewijk, 2001). Three different mod-els will be brought up in this section; the Ricardo, the Heckscher-Ohlin and the factor abundance model.

The Ricardo model assumes a world with two countries, home and foreign, and two differ-ent goods. There is an inelastic supply of labour in both countries, which is homogenous in quality, but mobile only within each country. Technology differs within the two countries. There is no cost of transportation, and there is competition in the factor and product mar-ket (Bowen, Hollander and Viaene, 1998). This discussion can be taken down to a regional level, where regions have production within its comparative advantages. For example, re-gions with a relatively high level of technology produce high technological goods and serv-ices that they trade with regions that have low technological endowments.

The Heckscher-Ohlin theorem, based on the same assumptions as the Ricardo model, was developed by two Swedish economists, Eli Heckscher and Bertil Ohlin. The theorem says that a country’s comparative advantages determine its export, and that its scarce resources determine its import (Hirsch, 1967; Neigishi, 1969). This discussion can as the discussion of the Ricardo model, also be applied on interregional trade. As a result, advantages from trade arise from a high aggregate demand of a region’s ample supply (Bowen, Hollander and Viaene, 1998).

The H-O theory was improved by Samuelsson in the 1950’s, and then called the H-O-S theory or the factor abundance model. The factor abundance model extends the H-O the-ory with two new production factors, capital and labour, thus the world consists of two countries that produce two goods, by two different production factors. Accordingly, coun-tries export the good in which it has comparative labour-productivity advantages and im-ports goods in which they have scarce resources (Bowen, Hollander and Viaene, 1998).

1.3 The New Trade Theory

The new trade theory on the other hand, explains imports and exports by economies of scales, specialisation, and the home market effect, as well as the love of variety effect. The concept of economies of scale was first introduced in 1890 by Marshall, and then economists as Ohlin and Krugman have followed his way (Bowen, Hollander and Viaene, 1998; Nourish, 1969). There is a distinction between internal and external economies of scale. The former means that the per unit cost decreases as output increases, and therefore only comes about in monopolistic competition where firms have fixed costs and a constant marginal costs. The latter on the other hand, explains decreased average cost per unit as an industry as a whole increases its production, which primarily due to spillovers of

infra-structure and information technology. External economies of scale explain the cause of in-dustry cluster areas, i.e. a concentration of firms that produce the same kind of goods or services. Thus, large city regions often specialise in diversity and therefore gain from both internal and external economies of scale, thus, economics of scale can be external to the individual firm, and still internal to the whole industry. As a result, economies of scale give such advantages that it is more favourable for export nodes to import from these regions, and keep focus on their area of expertise (Johansson and Karlsson, 2001).

Krugman’s works of 1979 and 1980 have played a vast role for the new trade theory. He developed a new model to explain the occurrence of trade. Opposite to the neoclassic trade theory of comparative advantages this model is based on internal economics of scale and the “love of variety effect”. The love of variety effect means that a one unit increase in supply more than proportionally increases utility, i.e. people love varieties. The starting point of Krugman’s model is that an open economy, with a larger market compared to a closed economy, can produce a larger output. This is so because companies gain from in-ternal economics of scale, and as a result the consumers utilize wider variety in the supply of goods and services (Krugman, 1994; Brakman, Garretsen, and Marrewijk, 2001).

Moreover, Krugman did also develop the concept of the home market effect, to explain export. The home market effect is based on a non-symmetric demand because of positive transport costs, and an increased market sizes as a result of external economics of scale, as well as the love of variety effect. Krugman means that companies will specialize in produc-tion with a high local demand and that they will import products in which their local de-mand is lower (Krugman, 1994; Brakman, Garretsen, and Marrewijk, 2001).

According to any school, regions specialise and as a result import and export nodes de-velop.

1.4 Import and Export Nodes

Import and export nodes are connected through an interregional network by linkages be-tween seller and buyer in respectively region. The import node is responsible for the sur-veillance and development of the world news, as for example technological progresses, and then to supply export nodes with the relevant information. Export nodes on the other hand, are in general smaller in its size, often specialised in just one or a few varieties, and much dependent on the supply of import nodes. The wealth of a node is among other things determined by the rate of in and out flow of a region, which is determined by the ef-ficiency and capacity of the linkages, which is dependent on infrastructure, information technology and transport cost (Johansson and Karlsson, 1991; Jacobs, 1985; Jacobs, 1989; SOU, 1990).

The infrastructure and information technology are most decisive for the possibilities to overcome different hindrances, such as mountains or rivers, trade barriers as customs, as well as cultural and language difficulties that are prevented with affinity. The former, a well-developed infrastructure, facilitates factor mobility and decreases transport cost, and could be for example harbours, airports, and storages. The latter, information technology, pro-vides contact intensive environments that reduces transaction costs with for example tele-phone and Internet (Johansson and Karlsson, 1991; Johansson and Karlsson, 2001; Jo-hansson, 1993; JoJo-hansson, 1988).

Additionally, Adam Smith was first to discover the advantages of locations close to water, and therefore the importance of transportation costs, which today is fundamental of geo-graphic economics. The transport cost is determined by the distance between seller and buyer, and the properties of their interaction link. Transport cost is often observed as an iceberg effect; if an iceberg is to be transported some units of the iceberg will melt and not reach the target destination, and these units could be looked upon as transportation cost. One commonly used measure of transport cost is the difference between costs, insurance and frights (c.i.f.) and free on board (f.o.b.). The c.i.f. is the total value of imports, inclusive cost associated with trade that leaves a destination, while the f.o.b. is the value of the im-port without any costs associated with transim-port (Krugman, 1994; Brakman, Garretsen and Marrewijk, 2001; Johansson and Karlsson, 2001).

In brief, this discussion lead to the same conclusion as Eli Heckscher once added up to; import is much more changeable than export; see Figure 3, (Johansson and Karlsson, 1991).

Figure 3: The Role of a Central Import Node Source: Johansson and Karlsson, 1991, page 57

As illustrated in Figure 3, central import nodes should be flexible to be able to assess all different news of the international economy, as well as distribute the news to all different specialised, export nodes in the periphery (Johansson and Karlsson, 1991).

Import nodes do often develop in areas with good market potential, for example close to water. Establishments of new import companies in an area increase the market potential since knowledge about import is brought to the region. As a result, the demand of estab-lishment in this area increases in order to gain from external economies of scale. In this way there is a continuous inflow of new knowledge to the region, and the import node grows by itself. This relationship is illustrated in Figure 4.

International sources

Figure 4: The Growth of an Import Node

Source: RTK (1998), p. 47.

Import nodes are not only attractive for import companies, but also for companies that can supply import companies with goods and services so that import companies can focus on their speciality.

It is often central import nodes that have the major concentration of local research and de-velopment (R&D), since they are good complements. The R&D process is most appropri-ate in larger regions, where there is a larger aggregappropri-ate demand to cover the fixed costs of many different varieties (Johansson and Karlsson, 1991; Johansson, 1988; Ohlin, 1933; RTK, 1998; Johansson, 1993; SOU, 1990). An efficient R&D activity requires skilled la-bour. Firstly, skilled labour is necessary to be able to assess the international impulses, sec-ondly, to find out about customer demand, and thirdly, to develop products that makes customers active and makes this process profitable.

1.5 Product Development

Product development, and new product cycles do often come up through import substitu-tion. Import substitution is a process that start new product cycled by replace imports by local production. This could be imitations of products or production methods, or even de-velopment of new complement goods. An already existing demand of what is imported does more or less secure a continuous demand even from local production (Jacobs, 1985; Jacobs, 1969).

1.5.1 The Product Life Cycle

The development of a product is often explained by the theory of the product life cycle, which divides the development process into three different phases; the new phase, the growth phase, and the mature phase. In the theory of the product life cycle, a new product is defined as Kuznets definition “a recent invention or unfamiliar developments” produced

Geographic market potential

Attractive for companies with knowl-edge about import and its networks

Establishment of import companies Market and import

with new production technology (Hirsch, 1967, pp 18). The first phase of product devel-opment, the new phase, is labour intensive and requires skilled labour. Production is often out sourced to specialist firms, which set the production structure and minimizes bugs. Moreover, production of the first phase is often located in cluster areas; first because the development of products requires much R&D. Secondly, the first phase of product devel-opment is costly and much dependent on external economies of scale. Thirdly, products in the first phase of the product life cycle are mostly distance sensitive, i.e. they require an in-tensive contact between seller and buyer, which might include inspection, negotiations, and contract discussions, as well as legal consultation. The demand of the product at this time of the development is determined by accessibility, as well as the quality and price of possi-ble substitute goods. If there are no substitutes, it means that competition is low and the price can be set high, which bring about mark up profits that support the high start up costs (Vernon, 1966; Hirsch, 1967).

When the product enters the second phase, the growth phase, and the demand remains even though the patent contract runs out and competition increases, production the spe-cialised and labour intensive production will be exchanged for capital-intensive mass pro-duction. External economics of scale plays less role and internal economies is of greater importance. As a result the demand becomes more price elastic (Vernon, 1966; Hirsch, 1967).

When the market becomes sutured, the product will enter the last and mature phase. The manufacturing plant will grow, production will become even more capital intensive, and most skilled labour will be replaced by unskilled and semiskilled labour. As a result, the H-O theorem can be applied, since it becomes interesting to out source production to labour intensive regions with lower rental costs etc, for example export nodes. However, products can reach the mature phase but still require special inputs from the local market, such as skilled labour and spare parts, which make them appropriate for regions in the periphery. Still some of the production that do not require such inputs can be out sourced to locations with lower wage.

In brief, standardised products are sold with more routine compared to products in the early stages of development. Developed routines, with less contact between seller and buyer, means decreased transaction costs and greater potential for trade (Johansson and Karlsson, 2001; Johansson and Karlsson, 1991). This discussion can be summarised by the Leontief paradox. It says that advanced countries that invent labour saving capital will ex-port them in their first phase of the product life cycle and imex-port them in the later phase of the cycle (Vernon, 1966; Hirsch, 1967).

1.6 Hypotheses

This section presents five hypothesises, based on the theoretical framework just presented, for the following empirical chapter to answer. They are as follows:

Hypothesis 1:

The import value is dependent on the value of Gross Regional Product (GRP)

According to Jane Jacobs, the wealth of a nation is improved by efficient in and out flows, i.e. efficient import respectively export.

Hypothesis 2:

The import value is dependent on the size of its region

According to theory, import is a contact intensive activity, which gains from economies of scale, well-developed infrastructure and information technology. These factors are mostly supplied in large regions with high density, and are the source to self-growth.

Hypothesis 3:

Import is dependent on export value

An efficient assessing of the international news creates innovations and new product cycles in the import node before production is outsourced to regions in the periphery.

Hypothesis 4:

Import is dependent on R&D

Imports facilitate R&D activity, i.e. import is a complement to R&D activity.

Hypothesis 5:

Import is dependent on the supply of university educated people

The process of assess and distribute the international news of import is improved by skilled labour.

2

Data Presentation and Empirical Analysis

The first section of this chapter presents the variables and their features. The second sec-tion, presents the composition of the models in the regression analysis as well as the em-pirical results.

2.1 The Features of the Data

The most appropriate data for this study was found at Statistics of Sweden (SCB), year 2001 and 2003. The data is aggregated over labour market regions (LA-regions), which ac-cording to NUTEK are defined as a group of municipalities with a common labour market. The variables that are included in this analysis are presented in the Table 1, where import is the dependent variable.

Table 1: The Dependent and Independent Variables

Abbreviation Explanation Unit Year

Imp Import value SEK 2003

GRP Gross Regional

prod-uct Value

SEK 2003

Pop Population No of people from

0 – 100 years old

2003

Exp Export Value SEK 2003

U R&D Accessibility3 _to

Uni-versity R&D No of man years 2001

F R&D Accessibility to Firm

R&D No of man years 2001

U Ed University Education No of people 2001

As seen in the table above, the most explanatory variables are from 2003, however three of them are from 2001, the effect of this is assumed to be negligible and therefore ignored. In Table 2, the descriptive statistics of the above variables are presented.

3 _{Within the same LA region}

Table 2: Descriptive Statistics

Imp GRP Pop Exp U

R&D FR&D U Ed N* 81 81 81 81 81 81 81 Max 271114530703 13 1929766 224311605766 99871 289651 261771 Min 365383 6 2981 5326146 0 0 131 Mean 7660381858 9 110653 9675932781 2212 6042 10500 Med 986501958 9 42438 2461873286 0.3 30 3664 SD** 32257254999 1 244369 28727440270 11926 34335 30902 Skewness 7 0.2 6 6 7 8 7 Kurtosis 58 0.3 40 43 58 61 56

* Number of observations **Standard Deviation

Comparing the maximum and minimum values, a large difference between the observa-tions is observed. Moreover, a comparison between the mean and median value shows that all variables have larger mean values than median value. This is confirmed by scatter plots that show three significant outliers, Stockholm, Gothenburg and Malmö. The distribution of all variables, except for the GRP variable, is skewed with a right hand tail, since the criti-cal values of a normal distribution are between minus two and two. Moreover, the values of the kurtosis of all variables, except for the GRP variable, are far above two, which is the critical value for a normal distribution. This means there is as concentration in the centre compared to the tails. Tests show that the kurtosis values decrease gradually as N is de-creased; however, there is no large change when N is reduced with one unit at the time. The following test to be presented is a correlation test called Cross Tab test.

2.2 Cross Tab Tests

This section will present the result of two cross tab test, where import is the dependent variable and export and firm R&D are the two independent.

The test divides the 81 LA-regions of Sweden, into three equally large shares, where the 27 regions with the top highest value makes out the “high”-share column respectively row. The 27 regions with the middle values makes out “middle”-share column respectively row, and the last 27 regions with the lowest values makes out the “low”-share, column respec-tively row.

The table is divided into nine main (bold) cells where the three different groups of the two variables interact with each other. Each of these cells is divided into three interior cells, with one left column that has two rows, and one right column. The two rows in the left column present the count respectively the expected value. The count value represents the amount of regions within that share, high, middle or low, that the two variables have in common. The expected values, is the number representing equal distribution between the

three shares, high, middle and low, of the two variables. Moreover, the right column of each cell, presents the ratio between the count and expected value. If this value is equal to one, there is an equal distribution of the two variables. Moreover, if there is a diagonal, with values above one between the ratios, from the cell high/high down to the low/low, there is a clear correlation between the distributions of the two variables, across LA-region. In Table 4, the results of the cross tab test between import and export is tested.

Table 1: Cross Tab – Import / Export

Import High Middle Low

Export Ratio Ratio Ratio

Count 24 3 0 High Expected 9 2,23 9 0,62 9 0,00 Count 3 20 4 Middle Expected 9 0,54 9 1,92 9 0,60 Count 0 4 23 Low Expected 9 0,00 9 0,50 9 2,62 Pearson Chi-square 91,778 (0,000)

There are 24 LA-regions of the 27 LA-regions within both the high share of import and high share of export. This should be compared to the expected value of 9 for equal distribution, and gives a ratio of 2,23. Furthermore, there are 20 of the 27 regions within the middle share of import and export, which have both middle values of import and middle values of ex-port. This gives a ratio of 1,92. Furthermore, there are 23 regions of the 27 regions within the lowest share, which have both the lowest value of import and export. This gives a ratio of 2,62.

As a result, there is diagonal of the ratios from high/high, to low/low, with higher values than one. These cells do also have the three highest ratios in the whole table. This means that there is a clear correlation between the regions with highest import and highest export, between the regions that have middle import and middle export, as well as between the re-gions that have the lowest import and lowest export.

Table 2: Cross Tab – Import / F R&D

Import High Middle Low

F R&D Ratio Ratio Ratio

Count 20 7 0 High Expected 9 2,22 9 0,78 9 0,00 Count 6 15 6 Middle Expected 9 0,67 9 1,67 9 0,67 Count 1 5 21 Low Expected 9 0,11 9 0,56 9 2,33 Pearson Chi-square 53,778 (0,000)

There is a clear correlation between the regions with the highest import value and the re-gions with the highest F R&D value, due to the ratio of 2,11 > 1 for the high share of im-port and F R&D. The same is current for the middle and low share of imim-port and F R&D, where 1,67 > 1 and 2,33 > 1. As a result, there is a diagonal of rations greater than one, from the high/high cell, to the low/low cell.

The following section will present the empirical result of the regression analysis.

2.3 The Empirical Results

There are two different main models of this empirical analysis; Model 1 and Model 3, the best results of these two models are then summarised, in the Model 5. This is primarily an attempt to avoid multicollinearity. The models and their results are presented in numerical order. The Model 1 and Model 2 are presented below and their results in Table 3.

Model 1

Log Imp = a + _(Log Pop) + _(Log Exp) + _(Log U R&D) + _(Log F R&D) + _(Log U Ed) + u Model 2

Log Imp = a + _(Log Pop) + _(Log Exp) + _(Log U R&D) + _(Log F R&D) + _(Log U Ed) + u

Some tests have been performed with other variables than population to explain size. For example have population per squaremeter been used in an attempt to achieve a measure for density. However the results where similar, and population was chosen as the explanatory variable for the size of a region.

Table 3: The Results of Model 1 and Model 2

Results of Model 1 Results of Model 2

Constant 0.873* (0.161) 0.318* (1.005) LogPoP 0.015* (2.495) 0.02* (3.277) LogExp 0.006* (2.827) 0.01* (4.233) LogUR&D 0.584* (0.550) LogFR&D 0.002* (10.107) 0.0001* (25.985) U Ed 0.019* (2.395) R_ adjusted 0.997 0.996 Df** 80 80

*significance value, (t-value) **Degrees of Freedoms

Significant variables of the Model 1, population, export and firm R&D, are moved over to the Model 2, where the constant still is insignificant. The Model 3 and Model 4 are pre-sented below and their results are prepre-sented in Table 4.

Model 3

Log Imp = a + _(Log GRP) + _(Log Exp) + _(Log U R&D) + _(Log F R&D) + _(Log U Ed) + u Model 4

Table 4: The Results of Model 3 and Model 4

Results of Model 3 Results of Model 4

Constant 0.230* (1.211) 0.044* (2.045) LogGRP 0.178* (1,358) 0.015* (2.477) LogExp 0.001* (5.815) 0.001* (5.276) LogUR&D 0.154* (1,438) LogFR&D 0.001* 9.715 0.001* (24.816) U Ed 0.027* (2.258) R_ adjusted 0.996 0.996 Df** 80 80

* Significance Value, (t-value) **Degrees of Freedom

In the same way as in Table 3, the significant variables from model three, GRP, export and firm R&D, are moved over to model 5. Table 5 presents the final model, Model 5, with the most significant values, export and firm R&D.

Model 5

Table 5: The Results of Model 5 Results of Model 5 Constant 0.08* (2.708) LogExp 0.01* (7.593) LogFR&D 0.01* (25.458) R_ adjusted 0.996 Df** 80

* Significance Value, (t-value) **Degrees of Freedom

The final model does still have problems with multicollinearity despite several attempts to get rid of them. The correlation between export and firm R&D is high; the R_ adjusted value is 0.943, when export is dependent and firm R&D is independent. Different compo-sitions of the explanatory variables have been tried in order to avoid multicollinearity, for example GRP per capita, and population per square meter. The population and GRP vari-ables are separated, since they could explain the same things. In addition and lastly, one ex-planatory variable is taken away from the model at the time, while the tolerance value is studied. However, Edling and Hedström (2003) suggest that multicollinearity is not a problem as long as the estimates of the coefficients are not used.

On the other hand, there are no other signs that Model 5 should be an inadequate regres-sion model. In addition to the above analysis, cross tab tests are performed on the final Model 5, to confirm its results.

3 Conclusions and Suggestions for Further Studies

It is clear that import is a driving force of innovations, and that this unique study of import across regions therefore is of the greatest importance for a deeper understanding of import. The purpose of this study was to analyse what feature that explain the appearance of im-port across regions, which also was aimed to provide an explanation of the strong spatial concentration of imports in the Swedish system of network.

The main conclusion of this thesis is a significant correlation between the dependent vari-able of import, and the two independent varivari-ables export and firm R&D. This result goes in line with the theory, presented in the theoretical framework part of this thesis.

Jacobs (1985 and 1969) argues that import is much important for the wealth and growth of an economy, which means that regions with high import also should be rich. She means that it is important to follow the cycles of product development to keep a stable growth, as well as have a balanced trade where export pays for import. Economies of scale are the fundamentals of new trade theory and explain interregional trade by specialisation. Import nodes specialise in import activity, which implies assess international impulses and distrib-ute them to the regions in the periphery. This means that import nodes are specialised in the early phases of product development. According to the theory of the product life cycle, early phases of product development require R&D and highly educated labour. In accor-dance with the home market effect developed by Paul Krugman, the R&D activity is clus-tered in an area where the import companies create a relatively high demand for R&D. The monocentric city model and the central place theory can be interpreted as the import nodes’ specialisation within the early phases of the product life cycle, means that they have to import mature goods and services from the later phases of the product life cycle. As a result, all imports to the import nodes needs to be compensated with a relatively high ex-port that pay for its imex-ports, which is in line with the empirical results of this thesis. Moreover, if a high import is correlated with a high export, there is a high turnover of goods and services in these regions, a high turnover of goods and services is an important driving force of growth, and again the theory is in line with the result of a positive correla-tion between import and growth.

In brief, this study confirms the theory of imports, and its great importance for growth. The result of the empirical analysis suggests that import is best explained by export and firm R&D, these variables are spatially concentrated, it emphasis the importance of effi-cient network, based on infrastructure and information technology. Suggestions for further studies are to investigate ways of how to put import improvements into action. Moreover, to performance similar studies of other countries, to guarantee the result of this study, as well as to see whether the same goes for developing countries, and if import activity could b e a g o o d w a y f o r t h e m t o a c h i e v e g r o w t h .

References

Brakman, S., Garretsen, H. and Marrewijk, C. (2001), An Introduction to Geographical

Econom-ics, Cambridge; University Press

Bowen, H.P. Hollander, A. & Viaene, J-M. (1998), Applied International Trade Analysis, Bas-ingstoke; Macmillan

Cole, I. M. (1995), Japanese Import Price Elasticises by Region of Origin: 1968-1988, Japan

and the World Economy 7, pp. 131-138

Dutta, D. and Ahmed, N. (2004), An Aggregate Import Demand Function for India: A Co Integration Analysis, Applies-Economic-Letters Vol. 13 pp. 607-613

Deyak, T. A., Sawyer, W. C., and Sprinkle, R. L. (1989), An Empirical Examination of the Structural Stability of Disaggregated U.S. Import Demand, Review-of-Economics-and-Statistics; 71(2): 337-341

Edling, C., and Hedström, P. (2003) Kvantitativa metoder – Grundläggande analysmetoder för

sam-hälls- och beteendevetare, Lund; Studentliteratur

Hall, P. (1994), Innovation, Economics and Evolution: theoretical perspectives on changing technology in economic systems, New York; Harvester Wheatsheaf

Hirsch, S. (1967), Location of Industry and International Competitiveness, Oxford; Clarendon

Husted, S. and Melvin, M. (1998), (4th _{ed), International Economics, Boston; Addison-Wesley}

Jacobs, J. (1985), Cities and the Wealth of Nations – Principles of Economic Life, New York; Vin-tages Books

Jacobs, J. (1969), The Economy of Cities, New York; Random House

Johansson, B. (1988), Innovation Processes In the Swedish Network of Export and Import Nodes, CERUM Vol. 15

Johansson, B. (1993), Ekonomisk dynamik i Europa – Nätverk för Handel, Kunskapsimport och

Innovationer, Malmö; Boktryckeriet

Johansson, B. och Karlsson, C. (1991), Från Brukssamhällets Exportnät till Kunskapssamhällets

Innovationsnät, Karlstad; Länsstyrelsen i Värmlandslän

Johansson, B. and Karlsson, C. (2001), Geographical Transaction Costs and Specialisation Opportunities of Small and Medium – Sized Regions: Scale Economies and Market Exten-sion, Berlin; Johansson, B., Karlsson, C. and Stough, R., Theories of endogenous regional growth: lessons

for regional policies (Ed.), Springer, pp. 150-180

Karlsson, C. and Nilsson, R. (2002) Agglomeration, Economies of Scale and Specialisation in a Central Place System, Cheltenham; Johansson, B, Karlsson, C and Stough, R.R., Regional

Policies and Comparative Advantages, E. Elgar

Krugman, P. (1994), Rethinking International Trade, Cambridge, Mass; The MIT Press

Neigishi, T. (1969), Marshallian External Economies and Gains from Trade between Simi-lar Countries, The Review of Economic Studies, Vol. 36, No 1, pp 131-135

Ohlin, B. (1933), Interregional and International Trade, Cambridge, Mass.; Harvard University Press

RTK (1998), Regioner, Handel och Tillväxt Marknadskunskap för Regionen, Stockholm; Region-plane- och trafikkontoret, Katarina Tryck AB

RTK (2003), Växande Branscher – om Stockholmsregionens Samspel med Övriga Landet, No. 8, Stockholm; Regionplane- och trafikkontoret, Katarina Tryck

SoU (1990), Stadsregioner i Europa – Underlagsrapport, No.34, Stockholm; Gotab

Tsionas, E. G. and Christopoulos, D. K. (2004), International Evidence on Import De-mand, Empirica, No 31, pp. 43-53

Vernon, R. (1966), International Investment and international Trade in the Product Cycle,

Appendix 0 10 20 30 40 50 60 70 80 90 100 0 50 100 150 200 250

Population Imports Exports

Diagram 1: Lorenz Curve

Source: SCB, 2003.

Municipalities ranked in ascending order according to size Cumulative Percentage