University of Groningen Faculty of Economics and Business

(1)

University of Groningen

Faculty of Economics and Business

The labour market in the European Union

Moving in the right direction?

January 2012

Abstract

This paper investigates whether differences in income levels and unemployment rates influence migration flows among countries in the European Union. Using a Panel OLS estimator, it is tested whether changes in migration flows are influenced by income and unemployment rate differences and by changes in those differences. No large influence of those differences on changes in migration flows has been found in this paper. Other factors, like Schengen membership, neighbour, language and population do not seem to influence these changes very much either.

Key words

European Union, Labour Market, Migration.

(2)

2 Section 1 – Introduction

According to the Optimal Currency Area (OCA) theory of Mundell (1961) economies of areas or countries that adopt the same currency should move together. So, when one country in the currency area faces a negative demand shock other countries of that same currency area should face a comparable demand shock. This is important for a currency area, because monetary policy is centralised, so differences in demand shocks cannot be smoothened by devaluation or revaluation of currencies. Under asymmetric demand shocks one area will face lower production and higher unemployment, while the other faces higher production and lower unemployment.

A low number of asymmetric shocks is important for the stability of a monetary union. The fact that countries are in the same monetary union and the number of asymmetric shocks between those countries, are not completely exogenous. Two different views exist about the influence of a monetary union on the number of asymmetric shocks. The Krugman view (Krugman and Venables, 1991) argues that the number of asymmetric shocks in the European Union (EU) will increase after countries lower trade barriers. Krugman and Venables reason that lower trade barriers lead to more trade integration in the union, which leads to economies of scale opportunities due to a larger market and to regional concentration of industries to benefit from these opportunities. This

concentration process increases the risk of asymmetric shocks. For example, assume that the Netherlands concentrates on making cheese and Germany specialises in making

sausages, because low trade barriers allow them to do so. The moment demand for cheese, for whatever reason, collapses relative to the demand for sausages, the economy of the Netherlands will be hit, while the economy of Germany will relatively boom. If both countries produce both sausages and cheese, both economies will be equally hit by a fall in demand for cheese. So, trade integration and specialisation leads to asymmetric shocks.

The European Commission (EC) (De Grauwe, 2003, page 23) argues that trade

(3)

3

means that European countries import and export comparable products. In the example above, both Germany and the Netherlands produce and trade sausages and cheese, but German cheese is different from Dutch cheese. As a consequence, a fall in demand for cheese hits both economies, so shocks will be more symmetric. Furthermore, even under the assumption of regional specialisation, demand shocks are not necessarily asymmetric between countries. If European economies fully integrate, borders will be less and less important. Accordingly, specialised regions are not by definition located in one European country, but can be cross-border. In the example, the area that specialised in producing cheese consists both of Dutch and of German regions. In this scenario, a drop in the demand for cheese influences both countries symmetrically. The recent financial and economic crises, however, showed that asymmetric shocks among countries in the EU do still occur at this moment. A currency area that faces different shocks between its regions could, however, still be an optimal currency area. There are three different ways to deal with asymmetric shocks.

Regions that face a positive shock can transfer money to support the economies of

regions that face a negative shock. This is the so called public support, which has recently been used to support the economies of Greece, Portugal and Ireland by the other member states. Public support can, however, be a politically sensitive solution when money transfers turn out to be necessary for a longer time with the danger of moral hazard around (De Grauwe, 2003, page 10). As a requirement for public support, Greece has been told by the other members to reform regulation and to lower government debt. After receiving the money, Greece can choose to cancel these economic reforms and

adjustments and speculate that other European countries will support them again in the future. Public support of regions with low output by regions with high output becomes permanent, as happened between regions within countries like Belgium, Italy and

(4)

4

A more structural solution to solve the problems of asymmetric shocks is a flexible labour market. Flexibility in the labour market can be implemented in two ways: flexibility of wages and mobility of workers (De Graauwe, 2003, page 7). Under wage flexibility, Greece workers accept lower real wages so that a negative output shock does not increase the unemployment rate. The lower Greek wages will increase the

competitiveness of the Greek products and will stimulate its demand. When the economy of Greece consequently starts growing again, the wages of the Greek workers will

increase accordingly. Although countries in the euro area started harmonising their regulations of the labour markets since the Maastricht Treaty had been signed (De Grauwe and Mongelli, 2005), this solution is not very realistic. In general, national politicians do not like to abolish or change laws about job and wage protection. The fear for loosing elections is usually too large to make major changes in these kinds of laws.

The labour market can also deal with asymmetric shocks between regions when labour in a monetary union is mobile (De Grauwe, 2003, page 7). For example, if Spain faces high unemployment due to a negative demand shock and at the same time unemployment is low in the Netherlands, Spanish people can migrate to the Netherlands to work. The migration both lowers the unemployment rate in Spain and reduces the upward pressure on wages and accordingly on inflation in the Netherlands. Hence, politically sensitive measures as money transfers to other countries and lowering wage protection are avoided. In this paper I will concentrate on the mobility of labour in the European Union. The main question I will investigate is:

Do income and unemployment rate differences influence the direction and the size of migration flows among countries in the EU?

(5)

5

economic asymmetries. In case migration in the EU does differently, apparently other factors play a more important role in the migration decision of EU citizens.

Seven sections follow this introduction. I will start with a short history of the creation of the EU, followed by a literature review. In section 4 I explain the economic model I use to test my research question and in section 5 I will elaborate on the dataset I use. My econometric model is explained in section 6 and the empirical results are presented in section 7. I will finalise this paper with some conclusions in the last section.

Section 2 – Some Historical Facts about the European Union

The process of European unification was very much stimulated by the Second World War. The European countries Belgium, Italy, France, West-Germany, Luxembourg, and the Netherlands started working closer together in the 1950s, to prevent such a war from happening ever again and as a remedy against the strong nationalistic feelings in those days. As a result the European Coal and Steel Community (ECSC) was founded in the Treaty of Paris (1951) to pool the heavy industries of the countries mentioned above. The cooperation was extended when the Treaty of Rome (1957) was signed and the European Economic Community (EEC) was created separately from the ECSC. The goal of the EEC was to create a common market for the six nations.

(6)

6

fiercer competition. The recognition of the economic advantages by politicians was the basis of a compromise to integrate the economies of the member nations first and wait with the integration in the political sphere. This compromise led to the concentration on the creation of a single European market by stimulating international trade, mobility of capital and labour among the European countries and consequently led to a couple of agreements.

The Schengen Agreement, signed in 1985 by West-Germany, Luxembourg, Belgium, the Netherlands and France, allowed Europeans to travel without passport within the borders of the Schengen area. Due to the necessity of constitutional changes, the Schengen Agreement was implemented only in 1995 and by that time also Portugal and Spain had signed the agreement and joined the area. Two years later Austria and Italy also

implemented the Schengen Agreement. At this moment the Schengen area exists of 24 of the 27 European Union countries plus the non-EU countries Switzerland, Norway and Iceland (for a complete list, see Appendix 1).

In the Maastricht Treaty, signed in 1992, the name European Union (EU) was officially established and the European currency (euro) was created. This was another step in the creation of a single European market and the stimulation of cross-border trade and mobility. Due to the euro, physically introduced in 2002, the costs of exchanging one currency for another (transaction costs) disappeared, the price transparency increased as cross-border prices were easier to compare and uncertainty due to exchange risk

disappeared (De Grauwe, 2003, page 59). Disadvantages were a loss of flexibility by member states to react to asymmetric shocks and the loss of freedom by countries to choose their monetary policy preferences with respect to inflation and unemployment, i.e. the choice between a policy to keep inflation low and a policy to keep unemployment low.

In the Maastricht Treaty, the member states agreed to implement the Maastricht

(7)

7

the monetary union. The main ideas of the Maastricht criteria were adopted in the Stability and Growth Pact (SGP) in 1997 to make certain that this low inflation policy would also be continued after the introduction of the euro. The SGP allows the members of the euro area to have maximum deficits of 3% and maximum government debts of 60% of GDP. Higher deficits or debts can lead to incentives and political pressure to increase inflation in the euro zone to lower the governments’ burdens in real terms.

The main reason to criticise the SGP has been a lack of flexibility, because the pact does not allow governments to support their economies in times of recession. During a

recession the government may be the only one willing to invest and if the government decides not to do so or even lowers investment, the economic downturn is likely to be deeper and to take longer. For that reason, but on the cost of its credibility, the SGP criteria were relaxed in 2005, ironically enough under pressure from its biggest initiator Germany (together with France).

Historically speaking, the unification of and cooperation within Europe have been very important to prevent another World War II from happening ever again and to stimulate economic growth. Because unification in the political sphere needs more time, the unification in the economic sphere, i.e. the creation of a single market, developed faster. Now the Euro has been introduced, an absorption mechanism is necessary to deal with asymmetric shocks. Labour mobility can be this absorption mechanism and thus is very important in a well-functioning monetary union.

Section 3 – Literature Review

A European Commission report states that on average only 2% of residents in an EU country has a nationality from another member state, which is extremely low as

(8)

8

the percentage of EU citizens of 2% that lives in another member state is rather low. So apparently inhabitants of the EU have a strong preference to move within the same country over moving to another member state.

Although the mobility in general is lower in the EU as compared to the United States, some distinctions have to be made. Krieger et al. (2006) found that the mobility for some European countries is equal to or even larger than the mobility in the United States. Krieger et al. claim that a mobility comparison is very well possible, because the influence of larger distance differences in the United States can be compensated by the smaller differences in culture, language and institutions as compared to the European Union. They find that the differences in mobility are not that spectacular. 21% of the inhabitants of the EU has ever lived in another member state than their own, which is lower than the 30% of inhabitants in the United States that currently live in another state than their state of birth. These two percentages are not perfectly comparable, but direct comparable numbers are very hard to find. It should be noticed that there are major differences among the different European countries. From the population of the Nordic countries, 40% has ever lived in another European country, which is way higher than the 15% of the population in the southern countries (Krieger et al., 2006).

The mobility levels of the EU’s member countries are expected to be far from

(9)

9

near future, but if unemployed, approximately 35% of the EU-25 citizens is willing to move to another country to increase job opportunities (European Communities, 2006).

The level of education influences the level of migration and the intention to migrate of the European population significantly. Low and average educated people are less mobile than high educated people. Of the former category 17-20% ever lived in another member state, while of the latter 33% did so (Krieger et al., 2006). Also, migration intentions of the lower educated do not seem to be responsive to income inequality. For the higher educated the relationship is positive, with lower mobility intentions at low levels of income inequality and higher intentions as inequality increases (European Communities, 2006).

In 2005 the European Commission investigated the mobility of European citizens by using 24,500 questionnaires of Europeans from all member states. In this questionnaire not only the actual movements and the duration of these moves were documented, but also the intentions of people to move in the future. Compared to the European Barometer of 1995, the intentions to migrate changed differently per country. Spain’s willingness decreased as compared to 1995, while the willingness in general did increase for the other European countries, especially for Poland. In general younger people have on average a higher willingness to move to another member state and also, as said above, the level of education has a positive effect on these intentions.

(10)

10

Differences in labour institutions do create hesitations to migrate (European Commission, 2005).

Zimmerman (2009) found no indication that labor force mobility has increased in Europe in the recent years. Poor language skills and cultural differences are the main reasons and also current barriers in the form of e.g. non-transparent markets, the absence of cross-border recognition of professional qualifications and the non-transferability of social entitlements have negative effects on mobility in the EU (Zimmermann, 2009). The rise in female participation in the work force results in double income households and makes mobility a more complex issue and a long distance move less probable.

At last, Krieger et al. (2006) found that weather is an important reason to migrate from one country to another for inhabitants of the EU-15. This explains migration from richer northern European countries to the poorer countries in the south like Spain and Italy. Krieger et al. find that post-retirement and leisure migration is an important part of the total migration in the EU-15 as compared to the new EU members. For the new members, mostly economical differences are a reason to migrate.

Mobility is both a challenge and an opportunity for the EU. If used well mobility could lead to efficient use of the labour force (brain gain). The challenge is to prevent a brain drain, which happens if the sending region only loses valuable workers who are not integrated well in the receiving region and thus not used efficiently. In other words, they are more valuable for the EU as a whole if they stayed in their home region. The

migration from new Schengen members could lead to a brain drain, as mostly high educated people are willing to migrate and they migrate to countries that are already highly educated. (Krieger and Fernandez, 2006)

(11)

11

War II, labour mobility between these two regions was rather low, even though the north was clearly richer and there were no restrictions for migration. For a long time however, European workers were preferred over workers from the southern states to fill vacancies in the northern states. Labour mobility only increased from the south to the north when the World War II caused a drop in migration from Europe. So a major historical event was necessary to initiate migration from the south to the north in the USA. Likewise, a major event may be necessary to initiate migration in the EU.

Section 4 – Economic Model

In this paper I will investigate whether migration in the European Union is influenced by differences in incomes and unemployment rates. Furthermore, I will test whether

migration is a function of the variables presented below in Equation 1.

Migrationijt = f[(yit- yjt), (uj(t-1)- ui(t-1)), Schengenijt, Neighbourijt,

Languageijt, Populationit,, Populationjt] (Eq. 1)

where i ≠ j.

Migrationijt is the number of people that move from country j to country i in year t. I

expect that the migration between these countries depends on a couple of variables. The first variable is the difference between the income of country i and country j. Under a well-functioning labour market, migration flows should go from countries with lower incomes to countries with higher incomes. In other words, this variable should have a positive influence on migration.

(12)

12

that unemployment people first try to find a new job in their own country before looking for one abroad.

Schengen, Neighbour and Language are three dummy variables, which I expect to influence the net migration positively. If two countries are Schengen members,

neighbours or speak a comparable language these dummies equal one. Countries speak a comparable language if their languages belong to the same language group. I use the following language groups: Germanic languages (German, English, Dutch, Danish, Swedish), Roman languages (Spanish, Italian, Romanian, French, Portuguese), Slavian languages (Polish, Czech, Slovenian, Slovakian, Bulgarian) and Uralic languages (Finnish, Hungarian).

Migration is the number of people that migrate from one country to the other. I expect that this number is larger for countries with relatively many inhabitants. If a person can choose between moving to a large country or a small country, he will ceteris paribus choose for the large country, because more people means more job opportunities in absolute terms. For that reason the variable Populationit, the population of country i, is an

explanatory variable.

Also, for a random country the number of immigrants from a large country is ceteris paribus likely to be higher than the number of immigrants from a small country.

Intuitively, when two countries have a comparable share of migrating people relatively to their populations, the absolute number of people migrating from a larger country is higher. The variable Populationjt, the populations of country j, has been added to the equation for

that reason.

Section 5 – Data Description

(13)

13

members and that is why the dataset is limited to the following countries: Germany, Netherlands, Austria, Denmark, Sweden, Spain, Italy, Czech Republic, Slovakia, Slovenia, Poland, Lithuania, Latvia and Romania. For the countries Belgium, France, Bulgaria, Greece, Hungary and Portugal, data are only available of emigration to the countries mentioned above. Furthermore, annual migration data are only available for the period from 1998 until 2009.

Graph 1a, Total Migration Graph 1b, Change Total Migration

The graph 1b shows the first difference of the total migration. Interesting are the peaks in the years 2004 and 2007, which are two years in which the Schengen area was enlarged. Also the drop of total migration in 2008 is quite noticeable and is very likely caused by the financial crisis.

(14)

14

an individual unit root can not be rejected as the results of the other tests show (Table 1). In other words, the migration between country x and country y over time does have a unit root. The result of this unit root test is quite surprising and counterintuitive as migration is the number of people that moves from one country to another in a certain year. Hence, why would migration of one year influence migration of the next year? The result of this unit root test can probably be explained by the effects of globalisation, which caused an upward trend in migration during the period of my sample. If data had been available for a longer period, migration would probably not have had a unit root. As a consequence I will use the first difference of migration in my econometric model in the next section. In other words, I will test which variables influence the change of migration between countries.

Table 1, Unit Root Test Migration

Unit Root Test Intercept and Trend Intercept

Statistic Probability Statistic Probability

Levin, Lin & Chu -18.836 0.000 -2.920 0.001

Im, Pesaran and Shin W-stat 2.491 0.993 6.205 1.000

ADF – Fisher Chi-square 540.293 0.718 491.974 0.982

PP – Fisher Chi-square 579.768 0.2730 512.964 0.923

Graph 2 shows the comparison between the average migration between two countries that speak a comparable language and the average migration between two countries that do not speak a comparable language. The graph seems to confirm the intuition as described in the previous section, that a comparable language has a positive influence on migration between two countries. The peak in 2007 for the comparable language line can be

(15)

15

Graph 2, Influence Language on Migration

Graph 3, Comparison migrations between neighbours and non-neighbours

(16)

16

For migration I used data of annual direct migration flows from one country to another. This means that I do not know the migrants’ nationalities which these flows consist of. For example, in case of migration from Spain to Germany I do not know if and how many people are African immigrants that use Spain as a first stop to get a working permit and then move on to Germany. If a large part has an African nationality, the income difference is much larger for these individuals than for a Spanish individual, implying a much stronger incentive to move to Germany.

Also unknown are the ages of the migrants, so I cannot distinguish between workers and non-workers like children and retired people. Obviously non-workers care less about wage differences than workers. Between the old member states there is quite a lot of after-retirement migration (Krieger et al., 2006), for which wages are obviously less important and for example the climate matters much more.

For income differences I use the per capita PPP (in US Dollars) average wages of a country. Since high-educated people migrate relatively more (Krieger et al., 2006) the average per capita wage could be a less suitable parameter. Wages for high educated people can be different in two countries even if the average wages of these two countries are the same due to differences in income inequalities. For more mobile high educated people the differences in the wages they can earn are more interesting than the average wage differences among countries.

(17)

17 Section 6 – Econometric Model

In this section, two equations are presented to test which variables influence the change in migration. For both equations a Panel OLS estimator will be used. With Equation 2, the influence of the variables, explained in Section 4, on the change of migration will be tested.

Δ(Migrationijt) = α0 + α1(yit- yjt) + α2(uj(t-1)- ui(t-1)) + (yit- yjt) * (α3Neighbourijt +

α4Languageijt + α5Δ(Schengenijt)) + (uj(t-1)- ui(t-1)) * (α6Neighbourijt +

α7Languageijt + α8 Δ(Schengenijt)) + α9 Δ(Schengenijt) + (Eq. 2)

α10Neighbourijt + α11Languageijt + α12Populationit + α13Populationjt + εijt,

where i ≠ j.

In my economic model in Section 4, I explained the meaning of the individual variables of this equation. In this econometric model, however, the relationship between the variables and the change in migration is tested, which changes the meaning of the

relationship as compared to Section 4. With this equation is tested, whether the growth of migration between two countries is relatively larger, if there is a relatively large

difference between incomes and unemployment rates. Furthermore, I test if the fact that countries are neighbours or speak a comparable language cause differences in the change in migration.

(18)

18

rates between two neighbours, the larger the effect on migration. The same applies to the Language dummy variable.

Ultimately, the Δ(Schengenijt) variable is somewhat different from the Schengen dummy

variable explained in Section 4. Δ(Schengenijt) is a dummy variable that equals one when

a country enters the Schengen area and is zero the periods before and after. This way the effect of the entry to the Schengen area by a country on migration is tested and not so much the membership itself.

Equation 3 is used to test the influence of the change in income and unemployment rate differences on the change of migration.

Δ(Migrationijt) = α0 + α1Δ(yit- yjt) + α2Δ(uj(t-1)- ui(t-1)) + Δ(yit- yjt) *

(α3Neighbourijt + α4Languageijt) + α5(yit- yjt) * Δ(Schengenijt) +

Δ(uj(t-1)- ui(t-1)) * (α6Neighbourijt + α7Languageijt) + (Eq. 3)

α8(uj(t-1)- ui(t-1)) * Δ(Schengenijt) + α9Δ(Schengenijt) +

α10Neighbourijt + α11Languageijt + α12Populationit + α13Populationjt + εijt,

where i ≠ j.

The intuition here is that if the income or unemployment rate difference between two countries grows faster, the migration will grow faster in order to smooth economic differences. If this is the case changes in migration react to changes in income and unemployment rate differences between countries, and thus function as a stabilising factor. For the combined variables, also the change in income and unemployment rate differences are used for the same reason. In combination with the Schengen entry dummy, the absolute difference between incomes and unemployment rates is used, because for all migration flows it consists just out of one moment in time, which makes the use of a change over time of income and unemployment rate differences irrelevant.

(19)

19

In this section Equation 2 and 3 will be tested in three different ways. I will start to test the models for all the countries in my sample together. Then I will divide the countries in two groups: Group 1, the countries that signed the Schengen agreement before 1998, which exists of Germany, Netherlands, Austria, Denmark, Sweden, Spain and Italy, and Group 2, the countries that signed the Schengen agreement in 2004, existing of Hungary, Poland, Latvia, Lithuania, Slovakia and Slovenia, plus Romania and Bulgaria, which signed in 2007. The countries in Group 1 are all members of the monetary union of the Euro, except for Denmark, but that country connected its currency to the Euro. The Group 2 countries are not members of the currency area, except for Slovakia, but they are members of the EU and will in the future probably join the monetary union too. I will accordingly test the models for the changes in migration flows between Group 1 and Group 2 and for migration flows within Group 1. Dividing the countries in two groups, allows me to compare the migration streams of the old and new EU members.

Subsection 7.1 – All countries together

Table 2, Migration of the complete sample

Variable Coefficient Eq. 2 (Standard Error)

Variable Coefficient Eq. 3 (Standard Error) α0 -266.2841 (179.6373) α0 -247.0758 (186.5423) (yit-yjt) -0.015199 (0.010785) Δ(yit-yjt) -0.044902 (0.113396) (uj(t-1)-ui(t-1)) 25.52740 (27.77919) Δ(uj(t-1)-ui(t-1)) 80.48179 (80.28554) (yit-yjt) * Neighbourijt -0.002206 (0.024624)

Δ(yit-yjt) * Neighbourijt -0.054386

(0.326495)

(yit-yjt) * Languageijt 0.123414***

(0.039413)

Δ(yit-yjt) * Languageijt 0.825981**

(0.327545)

(yit-yjt) * Δ(Schengenijt) 0.187729***

(0.029903)

(0.029267)

(uj(t-1)-ui(t-1)) * Neighbourijt 20.54108

(52.94329)

Δ(uj(t-1)-ui(t-1)) *

Neighbourijt

(20)

20

(uj(t-1)-ui(t-1)) * Languageijt -85.03945

(62.65198)

Δ(uj(t-1)-ui(t-1)) *

Languageijt

581.2826*** (213.6053)

(uj(t-1)-ui(t-1)) * Δ(Schengenijt) -244.7395***

(75.91976) (uj(t-1)-ui(t-1)) * Δ(Schengenijt) -218.4783*** (74.10902) Neighbourijt -57.76398 (284.4981) Neighbourijt -110.0821 (290.8829) Languageijt 215.2219 (277.4235) Languageijt 334.5193 (282.9349) Δ(Schengenijt) 1512.180*** (377.7270) Δ(Schengenijt) 1489.056*** (384.9297) Populationit 7.527523 (5.138733) Populationit 5.825151 (5.139327) Populationjt 4.619066 (4.993091) Populationjt 4.540257 (4.978421)

Total Observations 2925 Total Observations 2825

Total Cross-sections 315 Total Cross-sections 315

Adjusted R-squared 0.0200 Adjusted R-squared 0.023

Durbin-Watson 2.5368 Durbin-Watson 2.546

F-statistic 5.5658 F-statistic 6.2276

Significance F-statistic 0.000 Significance F-statistic 0.000

* significant at 10% level, ** significant at 5% level, *** significant at 1% level.

The income and unemployment rate differences do not influence the changes in migration. Only in combination with entry to the Schengen area and speaking a comparable

language the income difference plays a significant role. The unemployment rate difference only plays a significant role in combination with entry to the Schengen area and surprisingly this relationship is negative, which suggests a reaction in the wrong or destabilising direction. The Schengen entry dummy is also significant. However, the adjusted R-squared is very low implying the exogenous variables in the equation hardly explain changes in migration.

(21)

21

can hardly be explained by the exogenous variables in both Equation 2 and 3. In other words, changes in migration flows do not seem to be influenced by absolute differences in incomes and unemployment levels and neither by changes in income and

unemployment rate differences.

Subsection 7.2 – Migration between Group 1 and Group 2

In Equation 2, the variable (uj(t-1)-ui(t-1)) * Languageijt correlates with (yit-yjt) * Languageijt,

hence is left out of the equation.

Table 3, Migration between old and new members

Variable Coefficient Eq. 3 (Standard Error) α0 -524.4224 (444.1846) α0 -726.8987 (448.7698) (yit-yjt) -0.023158 (0.020547) Δ(yit-yjt) -0.065435 (0.204517) (uj(t-1)-ui(t-1)) 51.44974 (57.25390) Δ(uj(t-1)-ui(t-1)) 76.87312 (165.9974) (yit-yjt) * Neighbourijt -0.008885 (0.045510)

Δ(yit-yjt) * Neighbourijt 0.135034

(0.612492)

(yit-yjt) * Languageijt 0.166264**

(0.083685)

Δ(yit-yjt) * Languageijt 6.344849***

(2.159348)

(0.054384)

(0.051975)

(uj(t-1)-ui(t-1)) * Neighbourijt 55.99259

(130.4469)

Δ(uj(t-1)-ui(t-1)) *

Neighbourijt

106.9552 (427.7575)

(uj(t-1)-ui(t-1)) * Languageijt - Δ(uj(t-1)-ui(t-1)) *

Languageijt

6.344849*** (2.159348)

(uj(t-1)-ui(t-1)) * Δ(Schengenijt) -375.5362**

(22)

22 Languageijt 2313.100 (1483.240) Languageijt -1415.716 (1524.031) Δ(Schengenijt) 2996.457*** (851.8906) Δ(Schengenijt) 3120.208*** (842.1849) Populationit 14.90434 (12.61113) Populationit 13.38551 (12.31475) Populationjt 5.074144 (12.96471) Populationjt 14.58342 (12.46632)

The results from this sample are quite comparable to the results of the complete sample in Subsection 7.1. Also here the differences in incomes and unemployment rates are only significant in combination with the Language and Schengen dummies. The same applies to the changes in income and unemployment rate differences. Compared to the results in the previous subsection, the adjusted R-squared values are somewhat higher, but still they only explain a small part of the changes in migration flows.

Subsection 7.3 – Migration within Group 1

For this group of countries the Schengen variables are left out of the equation, since all countries are Schengen members for the whole sample period. Concerning Equation 2, the variables (yit-yjt) * Neighbourijt and (uj(t-1)-ui(t-1)) * Neighbourijt are correlated and also

correlate with the variables (yit-yjt) * Languageijt and (uj(t-1)-ui(t-1)) * Languageijt, which

again are also correlated. For those reasons I decided to leave all these variables out of the equation. Ultimately, (yit-yjt) correlates with (uj(t-1)-ui(t-1)) so the latter one is also left

(23)

23

For Equation 3, the changes in differences in incomes and unemployment rates correlate with the combinations between the changes in differences in incomes and unemployment rates and the neighbour and language dummies. So, these combination variables are left out of the equation.

Table 4, Migration within the group of old Schengen members

Variable Coefficient Eq. 3 (Standard Error) α0 -63.89684 (59.14668) α0 -63.80616 (58.37290) (yit-yjt) 0.003357 (0.008643) Δ(yit-yjt) 0.064036 (0.039311)

(uj(t-1)-ui(t-1)) - Δ(uj(t-1)-ui(t-1)) 119.1534***

(32.58676) Neighbourijt 136.0944* (74.73430) Neighbourijt 139.0428* (73.78835) Languageijt 89.18937 (66.20599) Languageijt 91.61595 (65.36798) Populationit 1.236999 (1.262993) Populationit 1.033470 (1.153491) Populationjt 1.812060 (1.165293) Populationjt 1.785199* (1.077593)

(24)

24

which suggests that the labour market functions well. However, the adjusted R-squared values are again very low and thus hardly have any explanatory value.

Section 8 – Conclusion

In this paper I explained why labour mobility in a monetary union is important, especially when the economic integration is much ahead of the political integration as is the case in the EU. Labour mobility can only work as a stabilising factor when migration goes in the right direction, i.e. from countries with low income and high unemployment to countries with high income and low unemployment. As data for migration are only available for a limited period of time and show a unit root in that period, I tested for the change in migration.

The results in this paper show some significant influence of income and unemployment rate differences and changes in those differences on changes in migration. Especially when two countries speak a comparable language and enter the Schengen area, income differences and changes in income differences play a significant role. However, the explanatory value of these variables is low, which suggests that other variables not used in this paper explain a large part of the changes in migration flows. Hence, changes in migration flows do not seem to act as a stabilising factor for neither economic differences, neither changes in economic differences among countries in the EU.

It should be noticed that in the period between 1998 and 2007 the average incomes of the member states were growing and unemployment rates of the highly unemployed

countries, like Spain and Italy, were decreasing. Consequently, inhabitants of these countries were not really forced to react on differences or changes in differences in incomes or unemployment rates with other countries. Because, why would a person move away from family, friends and a familiar culture if jobs are available and wages

(25)

25

As said in Section 3, Eichengreen (1991) found that in the United States a major event had to happen before migration from the poor south to the rich north started. Maybe, the EU also needs such an event before migration really boosts and starts reacting to

differences and changes in differences in economic circumstances. The economic and financial crisis, which started in 2008, could be such an event, although Graph 1a in Section 4 shows that total migration in the EU decreased in 2008. However, now countries like Greece, Italy and Spain turn out to be hit harder by the crisis than for example Germany, it is likely that there will be an enlargement of economic differences among European countries. Consequently, migration to the countries that where hit less fierce will have to increase to smooth the crisis. Future research will tell if the crisis turns out to be the major event that caused a boost in migration.

In this paper I did not find evidence that changes in migration flows are much influenced by income and unemployment rate differences. Migration flows, however, possibly are more influenced by those factors. Although Mundell argued that migration is an

important factor in a monetary union already in 1961, data are only available from the mid-1990s on and not even for all EU countries. With the availability of data over a longer period, problems with serial correlation will likely disappear and research can be done about the factors that influence migration flows, instead of changes in those flows. Furthermore, a more detailed data set, for example with information about ages and nationalities of migrants between two countries, would make research more accurate.

(26)

26

Since migration is important for the sustainability of the European Union, especially in times of crisis, it is important that migration is as easy as possible for inhabitants of the EU. Therefore, common regulation in the labour market, for example for retirement insurance, of the EU is very important. On the other hand, the inhabitants should be made ready to migrate, so education in general and education of languages (English) in

particular should be at a sufficient level in all member states. Further research should be done to find out which factors further stimulate migration.

(27)

27 Appendix 1

Countries that signed the Schengen Agreement

Year Countries

1985 Belgium, France, Germany, Luxembourg, Netherlands

1990 Italy

1992 Greece, Portugal, Spain

1995 Austria

1996 Denmark, Finland, Iceland, Norway, Sweden

2004 Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta,

Poland, Slovenia, Slovakia, Switzerland

(28)

28 References

De Graauwe, P. (2003). Economics of Monetary Union. Oxford: Oxford University Press.

De Grauwe, P. and F. P. Mongelli (2005). Endogeneities of optimum currency areas;

What brings countries sharing a single currency closer together. European Central

Bank, Working Paper Series No. 468.

Eichengreen, B. (1991). Is Europe an Optimum Currency Area? NBER Working Papers Series, Working Paper No. 3579.

European Commission (1990). One Market, One Money. European Economy, No. 44, October.

European Commission (2005). Eurobarometer 63, Public Opinion in the European

Union. September.

European Commission (2008). Labour market and wage development in 2007, with

special focus on the economic impact of immigration. European Economy, No. 5.

European Communities (2006). Europeans and mobility: first results of an EU-wide

survey. Eurobarometer survey on geographical and labour market mobility.

Krugman P. (1991). Geography and Trade. Cambridge, MA: MIT Press.

Krugman P. and A. Venables (1993). Integration, specialization, and adjustment. National Bureau of Economic Research.

Krieger, Hubert and Enrique Fernandez (2006). Too much or too little long-distance

(29)

29

on worker mobility. European Foundation for the Improvement of Living and Working Conditions.

Mudell, R. A. (1961). A Theory of Optimum Currency Areas. The American Economic Review, Vol. 51, No. 4. September, pp. 657-665.

Zimmermann, Klaus F. (2009). Labor Mobility and the Integration of European Labor