• No results found

European Migration Flows: The Impact of the Schengen Agreements, the Anti-Immigrant Sentiment, and the Financial Crisis

N/A
N/A
Protected

Academic year: 2021

Share "European Migration Flows: The Impact of the Schengen Agreements, the Anti-Immigrant Sentiment, and the Financial Crisis"

Copied!
31
0
0

Bezig met laden.... (Bekijk nu de volledige tekst)

Hele tekst

(1)

European Migration Flows: The Impact of the

Schengen Agreements, the Anti-Immigrant Sentiment,

and the Financial Crisis

Jana Krolik

S2019493

Van Heemskerckstraat 28a University of Groningen

9726 GL Groningen, NL Faculty of Economics and Business

(2)

2

Abstract

The aim of this paper is to determine how the Schengen Agreements, the anti-immigrant sentiment, and the financial crisis have influenced European migration flows. Europe poses a unique research area, due to its history and political and economic structure. The dataset consist of eight countries and a time span from 1996-2011. The gravity model was applied to determine the effect of the three variables. The Schengen membership of the sending country has a positive immediate effect and a negative lagged effect. The membership of the receiving country has a lagged negative and positive immediate effect. The financial crisis also has a negative effect on migration flows. The anti-immigrant sentiment in the host country is positively related to migration flows. Additionally, the membership of the receiving and sending country in the EU was determined. The impact of EU-membership of the sending country is insignificant. The receiving country EU-EU-membership has an immediate negative and a lagged positive effect. Possible explanations for these results are given; however, future research should look into it further.

(3)

3

Table of Content

1. INTRODUCTION ... 4

2. LITERATURE REVIEW ... 6

2.1. Migration ... 6

2.2. Schengen Agreements and European Migration ... 7

2.3. The Anti-Immigration Sentiment ... 9

2.4. The Financial Crisis and European Migration ... 12

(4)

4

1. Introduction

In 2013, there were more than 230 million people living outside their home country whereas this number was only at 154 million in 1990 (UN, 2013). The number of people migrating is constantly increasing and one of the most popular destinations for immigrants is Europe (UN, 2013). This paper will take a closer look at the migration patterns of a number of countries within Europe. Europe poses a very special case in terms of internal immigration since “what can be regarded as

the European migration regime is something absolutely unique and distinct, incomparable in the contemporary world in terms of scale or the legal and institutional framework” (Kaczmarczyk et

al., 2012). Kaczmarczyk et al. present six unique patterns that characterize European migration, namely:

 Common European history  Migration mini-systems

 Large-scale recruitment of migrant workers

 Compensation for deficit of increase and ageing of population  Majority from highly developed countries

 Propensity to emigrate

The common history of Europe and its countries has created a special bond between the different countries that promotes migration more than in other parts of the world (Kaczmarczyk et al., 2012). Several features play a role, such as revolutions, wars, colonization, and crises, which have created a special bond among European countries that is unique to this part of the world (Kaczmarczyk et al., 2012). It has also led to the creation of mini-systems or clusters, such as Netherlands and Belgium or Germany and Poland, where bilateral migration within the country pair is very pronounced over longer periods of time (Kaczmarczyk et al., 2012). That is, the migration flows from country A to country B and vice versa are very pronounced and show the intra-regional interdependencies that exist within these mini-systems (Kaczmarczyk et al., 2012). Furthermore, European migration was largely impacted by active recruitment of migrant workers in the 1950’s and 60’s (Kaczmarczyk et al., 2012). There were times when the richer countries were in dire need of workers and they started recruiting immigrants on a large scale to compensate for labor gaps (Kaczmarczyk et al., 2012). This occurred in several European countries and is not only fairly unique to Europe, but also still influential today (Kaczmarczyk et al., 2012). A fourth aspect to consider is that many countries in Europe (e.g. Germany) suffer from an ageing of the population as well as a deficit of population increase (Kaczmarczyk et al., 2012). To compensate for these two issues, migration into these countries has played an important role (Kaczmarczyk et al., 2012).

1950's: Economic Union (Germany, Italy, France,

The Netherlands, Luxembourg, Belgium) 1960's: Denmark, Ireland, and UK join 1980's: Portugal, Spain, and Greece join 1990's: Austria, Finland, and Sweden join 2000's: Czech Republic, Cyprus, Estonia, Latvia, Lithuania, Hungary, Malta, Poland, Slovenia,

(5)

5 Another unique pattern of European Migration is that the majority of migrants come from developed rather than developing countries, which implies that the reasons for migration are different as well (Kaczmarczyk et al., 2012). Finally, there seems to be a European culture that creates a willingness to leave one’s country, that is the willingness to emigrate seems to be stronger among Europeans than among other groups of people (Kaczmarczyk et al., 2012). This last point is slightly controversial, since the fact that people tend to emigrate more might be due to the geographical situation of Europe. There are many small countries that are very close together, unlike for instance North America, where there are only two extremely large countries. So the possibilities for emigration are very unique in Europe compared to other parts of the world, which might explain the propensity of people to emigrate (Kaczmarczyk et al., 2012).

Europe’s present political, economic, and cultural peaceful state is unprecedented which makes it a unique area to study especially in terms of migration since the stability of the union and its increasing integration can be assumed to impact the movement of people largely. The inception of the European Union (to find out more about the history of the EU, see Box 1) led to the creation of a new European migration space in which new migration patterns emerged that were influenced by a number of events (Kaczmarczyk et al., 2012). The goal of this paper is to analyze these migration flows in the light of three of the most noteworthy events/developments of the past decades. In other words, the guiding research question is: how have the Schengen agreements, the financial crisis, and the anti-immigrant sentiment influenced migration flows among European countries between 1996 and 2011? Due to the widespread communism in Europe prior to 1990, data is unavailable for most European countries until the mid-1990’s. Unfortunately, this limits the time-span that can be considered in this research, which is the reason why the analysis starts only in 1996 and not earlier.

Box 1 The History of the European Union

(6)

6 However, the impact of all three factors can be analyzed in this time span. First of all, the Schengen area has made travel across borders much easier and it is a specific feature of Europe. Even though it was founded much earlier, we can still observe new additions to the Schengen area in recent years. Not only are the Schengen agreements very specific to Europe, but they are also a highly influential factor on terms of the movement of people. Second of all, the financial crisis that started in 2008 was a major global event whose impact could be felt (and still can be) in various areas. Due to its importance and strong impact, it should not be overlooked in the research of migration patterns that falls into the same period since it can be expected that the crisis also left its mark in this area. Thirdly, the anti-immigrant sentiment that has recently been spreading among Europe is considered. Unlike the two previous variables, the anti-immigrant sentiment is not a one-time event, but rather a development that has been gaining in momentum over the last couple of years. While the EU has been growing and established itself as a political entity, many European countries have seen the emergence of Eurosceptic, right-wing political parties that represent an anti-immigrant sentiment as a reaction to European integration.

2. Literature Review

2.1 Migration

Before we get into more detail about the main variables of interest, let us discuss the question why migration happens in the first place. The topic of migration and its origins is a well-researched topic and several studies have identified a number of push and pull factors that are the basis of migration. Ramos and Surinach (2013) summarize the migration pull and push factors that have been identified in the literature. First, let us go over the push factors which push people out of the country. These push factors fall into three different categories: social, historical, and cultural; economic; and political (Ramos & Surinach, 2013). The first category, comprising the social, historical, and cultural push factors, is the largest one with five factors (Ramos & Surinach, 2013). The first one of these is family reunification, meaning that the reason people leave their home country is to be reunited with family members that already live in another country (Ramos & Surinach, 2013). Another factor is diaspora migration, which refers to the movement of people that belong to a same kind of group, for instance Jews living in Israel (Ramos & Surinach, 2013; Merriam-Webster, 2014). A third factor is that people might seek the freedom from the discrimination they might experience in their home country (Ramos & Surinach, 2013). The last two push factors in this category are first a common language between the host and home country and secondly a colonial relationship between the two countries (Ramos & Surinach, 2013). The second category deals with economic push factors, which are often believed to be the main reason for immigration by the general population (Ramos & Surinach, 2013). The three factors in this category are the prospects for higher wages, the potential to improve one’s living standards, and professional and/or personal development (Ramos & Surinach, 2013). In terms of the last category for push factors, which refers to political factors, there are two: seeking political freedom as well as safety and security in another country (Ramos & Surinach, 2013). Overall, these push factors are pretty much self-explanatory. These are, in general, the reasons why people develop the wish to leave their home country. It is not only just one or all of these reasons that applies to migrants, but rather a combination of various factors.

(7)

7 (Ramos & Surinach, 2013). As mentioned before, many European countries are facing a decline in population growth and an aging population, which makes immigration an attractive method to combat these effects (Kaczmarczyk et al., 2012). In terms of geographic factors, the distance between the home and host country as well as the existence of a common border are important factors to consider (Ramos & Surinach, 2013). The category for social, historical, and cultural factors includes discrimination and human rights abuses, which people want to avoid or escape from (Ramos & Surinach, 2013). The economic pull factors are poverty, unemployment, low wages, and a lack of education and basic health in the home country, that people want to escape from in the host country (Ramos & Surinach, 2013). This leaves us with the political pull factors of migration, which include not only corruption and violence/conflict, but also poor governance (Ramos & Surinach, 2013).

As we can see, there are various different factors that lie at the core of migration. In the end, the reason for migrating is a complex combination of the factors above and most likely additional ones, that differ from person to person. In an effort to generalize the main reasons, the above list was generated and can serve as an attempt to explain why migration happens in the first place. The aim of this paper is now to determine in how far this complex migration behavior is influenced by specific mechanisms and events. The first one of these is the Schengen Agreements.

2.2 Schengen Agreements and European Migration

This section will briefly discuss the impact of the Schengen Agreements on European migration. Please refer to Box 2 to learn more about the background and history of the Schengen Agreements. Just like above, there is a timeline of all the additions to the Schengen area with the relevant countries in bold. In the migration literature, no research has been done on the impact of the Schengen Agreements on migration. However, the connection between these two things seems obvious since the creation of the Schengen area is a migration policy to facilitate the international movement of people. So it can be assumed that an effect is definitely present, however, its extent (also compared to other factors) still needs to be determined.

The Schengen area is a globally unique phenomenon that is generally considered to be a success (Collett, 2011). Nevertheless, this international cooperation does not come without its troubles, one of them being a problematic disconnect between the principle of free movement and the national immigration policies of the member countries (Collett, 2011). For instance, Spain’s decision to evict large numbers of unauthorized immigrants sparked concern on the French side of the border that a large influx of immigrants would follow (Collett, 2011). Other issues concern the internal and external borders (Collett, 2011). The internal borders have been softened and controls have been

1985: Founding of Schengen Area by: France, Germany, Belgium, Luxembourg,

and the Netherlands

1995: Spain and Portuga l join 1997: Austria and Italy join 2000: Greece joins 2001: Denmark, Finland, Sweden, Iceland, and Norway join 2007: Czech Republic, Estonia, Hungary, Latvia, Lithuania, Malta, Poland, Slovakia,

and Slovenia join

2008: Switzerl

(8)

8 removed to allow free movement within the Schengen territory (Grabbe, 2000). Due to this internal connection of many countries and the security implications that go along with it, the external borders are the exact opposite, namely hard and entry into the area is very strictly controlled (Grabbe, 2000). This creates more tensions between the different member countries in the sense that the security standards among the countries differ greatly which leads to concern about the security in some countries (Collett, 2011). In the future, the member states have to continue cooperating and finding agreements in terms of immigration policies and regulations (Collett, 2011). Overall, however, the Schengen Agreements have been met with more positive than negative response and it would be interesting to determine how large its impact on immigration flows among European countries is.

Overall, there are four possible combinations in terms of Schengen-membership of the sending and the receiving country. These are illustrated in Figure 1. The first hypothesis relates to cells one and three. The Schengen agreements were designed to facilitate migration within the area, so in that sense it should have a positive impact on migration. If the sending country is inside the Schengen area, then it could be assumed that movement within the area is facilitated and will therefore increase. At the same time, migrants can leave the Schengen area, which is not too difficult. This brings us to the first hypothesis regarding the relationship between European migration flows and the Schengen Agreements.

H1a: Migration from country A to country B goes up if the sending country (country A) is in the Schengen area.

The next hypothesis refers to cells one and two. It is important to keep in mind that the Schengen area has very hard external borders, which makes entering the area difficult. This could potentially limit migration flows. On the other hand, people might be more inclined to go to a Schengen country rather than a non-Schengen member due to certain characteristics that the Schengen countries fulfill. That is, to become a member of the Schengen area, a country has to comply with certain rules and fulfill certain requirements which could make it a more attractive destination for immigrants looking to improve their living situation. If we assume that the Schengen-membership of a receiving country makes it a more popular destination which outweighs the difficulty of entering that country, we get to the following hypothesis:

H1b: Migration from country A to country B goes up if the receiving country (country B) is in the Schengen area.

Figure 1: Schengen Matrix

Schengen Country A Non-Schengen Country A Schengen Country B (1) Both countries are

Schengen members

(2) Country A is not a Schengen member, country B is a Schengen member Non-Schengen Country B (3) Country A is not a

Schengen member, country B is a Schengen member

(9)

9 In a similar vein, the next set of hypotheses relates to the membership in the EU of the two

countries in question. Again, there are four possibilities which are equivalent to those presented in Figure 1 but relating to EU rather than Schengen membership. A similar reasoning can be applied too. If the sending country is an EU member, then migration within the Union as well as leaving it is relatively easy. Therefore we expect the following:

H2a: Migration from country A to country B goes up if the sending country (country A) is an EU member.

Similar to countries of the Schengen area, the countries that are in the European Union have to fulfill certain requirements. Also, the EU is a strong economic region which makes it attractive for immigrants. Therefore, it can be assumed that an EU member country is a popular destination which increases migration flows.

H2b: Migration from country A to country B goes up if the receiving country (country B) is an EU member.

A final note on the EU and the Schengen agreements: Both of these are specific to the geographical area that is Europe. Only European countries are part of the EU and only European countries are part of the Schengen area. To become a member country of either one, the candidate country has to fulfill a list of requirements which are similar for both unions. That means that the same type of countries will qualify. This means that there is an overlap between countries that are EU-members and countries that are Schengen members. However, these two things are two separate agreements.

2.3 The Anti-Immigration Sentiment

It is undeniable that recent years have seen the emergence of a strong anti-immigrant sentiment among many European countries. It appears that the “recent anti-immigrant sentiment and a

backlash against multiculturalism within European societies stems from frustration and concern over the demographic change and the increasing cultural diversity brought about by migration inflows on the one hand, and the economic crisis on the other” (Kaczmarczyk et al., 2012). It is

often assumed that immigrants simply exploit the social systems of their host countries and represent a burden for society even though empirical results show that the economic effect of immigration is either not significant or very small (Kaczmarczyk et al., 2012; Hainmueller & Hiscox, 2007). Additionally, it has been found that due to differences in values and beliefs, higher

Box 2 The History of the Schengen Agreements

(10)

10 educated people are less racist and appreciate cultural diversity and its benefits (Hainmueller & Hiscox, 2007).

In the following section, we will highlight political parties that represent the anti-immigration sentiment that has been spreading across Europe for each country of the dataset. A brief overview will be given for the largest or most successful right-wing parties for each of the countries. The list is not exhaustive, that is not every single party will be taken into consideration, only the most noteworthy one(s) for each country. This analysis of the political landscape in terms of right-wing, anti-immigrant political parties will serve a basis for the data selection later on.

Hungary. The anti-immigrant sentiment that has recently been spreading across Europe is

represented by right-wing political parties that are known beyond their borders due to their nationalist agendas. One of the most internationally well-known parties is the Hungarian party Jobbik. Jobbik was created out of a movement in 2003 by a group of students and its current leader is Gábor Vona (Jobbik, 2010). The party can be found on the far right of the political spectrum and they are internationally coined as a neo-Nazi and anti-Semitic organization (Paterson, 2014). The party has raised concerns internationally on a large scale that reaches Brussels and Washington due to not only its political agenda but also due to the popularity of Jobbik in Hungary (Paterson, 2014). In the 2014 election, Jobbik attained 20% of the votes which makes them the strongest radical party in Europe (Paterson, 2014).

Prior to Jobbik’s existence, so before 2003, the MIÉP (The Hungarian Justice and Life Party) was the most right-wing political party in Hungary. The MIÉP was founded in 1993 and was the leading radical party until the elections in 2009 when Jobbik entered the picture (McKinney, 2002; NSD, 2014).

Poland. Poland has seen the emergence of various right-wing political parties, three of which are

highlighted here. In 1992, the right-wing political party “Self Defense of the Republic of Poland” (SRP) was born out of disappointment over the changes after the fall of the Iron Curtain (Piskorski, 2004). After only representing rural interests at first, the party that attracted attention via loud demonstrations also targeted the working class (Repa, 2006). The party has been anti-EU and former leaders even “praised Adolf Hitler’s economic policies” (Repa, 2006).

In 2001, a new right-wing party entered the political arena in Poland and achieved immediate success: Law and Justice (PiS) (PiS, 2014). The PiS was only 0.7 percentage points behind SRP who came in fourth at the 2001 elections (NSD, 2014). Over time, the PiS established itself as one of the biggest parties in Poland (NSD, 2014).

Three years ago, in 2011, the far-right party “New Right” (NP) entered the picture (NSD, 2014). This party represents even more radical views that the SRP or the PiS. Among other things, the leader Korwin-Mikke wants to reestablish monarchy, end democracy, and turn the European Commission building in Brussels into a brothel (Day, 2014).

Luxembourg. The most right-wing, anti-immigrant party in Luxembourg was founded in 1987 as an

action committee that only promoted the single issue of pension justice (Cochrane & Nevitte, n.A.). It evolved into the “Alternative Democratic Reform Party” (ADR) and has not only broadened its agenda but also gained in popularity (Cochrane & Nevitte, n.A.). Concerning the issue of immigration, ADR state that immigration should be restricted in so far that only a small number should be allowed into the country and be fully integrated (Luxemburger Wort, 2013). According to the ADR unrestricted immigration will lead to a rate of immigrants that “exceeds Luxembourg’s

ability to successfully integrate” them (Luxemburger Wort, 2013). Overall, the party represents

(11)

11

Spain. According to Hernandez-Carr (2012), “Spain is one of the few European countries which in recent decades hasn’t experienced the emergence of a successful “new extreme right” party into its political system. However, Spanish society and its party system are experiencing a series of rapid changes that could be opening a political space for these kinds of parties”. When looking at the

major political parties in Spain over the last two decades, there is no party that has the extremist right-wing and anti-immigrant views that can be found in some of the other countries, such as Jobbik in Hungary. However, there are some minor parties that represent these views and were born out of the Spanish Fascist movement led by Franco’s leading party “Falange Española Tradicionalista” (Oxford Dictionaries, 2014). Whereas Franco’s party was abolished in 1977, a number of parties have emerged that follow the same ideologies, such as the “Falange Española Auténtica” and the “Falange Española Independiente” (Oxford Dictionaries, 2014; AXT, 1997). As the introduction of this paragraph suggests, neither one of these parties achieved great political success in recent years. “Falange Española Auténtica” had some votes, which were, however, always under the 1% threshold and places in the “Others” category (NSD, 2014).

Contrary to the above observations, there have been recent developments in Spain that paint a different picture. A new far-right party called “VOX” has been launched earlier in 2014 (Short, 2014). This party is also reminiscent of Franco’s ideologies and was founded by members of various political parties, one of them being the ruling party PP (Short, 2014). It has to be seen in future years whether this right-wing party will achieve popularity in Spain that is equivalent to other European right-wing parties.

Netherlands. In 1918, the “Reformed Political Party” (SGP) was founded which makes it the oldest

political party in the Netherlands (SGP, 2014). According to their official website, the party’s views are based on the Bible (SGP, 2014). Its current leader is Kees van der Staaij (SGP, 2014). Similarly to the ideology of the ADR of Luxembourg, the SGP states that immigration should be limited and restricted (SGP, 2014). The SGP has been getting a small number of votes in all of the last elections.

In recent years, however, a more notably right-wing political party in the Netherlands has been the “Party for Freedom” (PVV), with its leader Geert Wilders (PVV, 2014). The PVV is a Eurosceptic and anti-immigrant party (PVV, 2014). The PVV is one of the internationally most well-known right-wing parties and comparisons have been made between the party leader and Adolf Hitler (Calabria, 2014). Since its introduction, the PVV has by far outperformed the SGP, with 15.45% being its biggest success in 2010 versus the 1.74% achieved by the SGP (NSD, 2014).

A final party worth mentioning is the political party “Leefbaar Nederland” which existed between 1999 and 2003 (Parlement, n.A.). One of its leaders was Pim Fortuyn who was assassinated in 2002 by an animal rights activist for his anti-immigrant views (BBC, 2014). The party never received enough votes to gain seats in parliament.

Norway. The most successful right-wing party in Norway is in fact one of the biggest parties in

Norway overall, namely the “Progress Party” (FRP) (NSD, 2014). It was founded in 1973 and has evolved into the second largest party of Norway (Fremskrittspartiet, 2013). The party has been warning against the consequences of immigration and multiculturalism. Despite their strong opposition to immigration, the FRP are not extreme-right, but slightly more moderate, which makes them even more popular (Nome, 2013). In fact, the party’s membership has doubled over the last decade (Fremskrittspartiet, 2013). Since its inception, the FRP has been the most successful right-wing and anti-immigration party in Norway (NSD, 2014).

Denmark. Since the elections of 1998, the “Danish People’s Party” (DF) has been gaining in

(12)

anti-12 immigrant sentiment, stating that they refuse to accept a multicultural society in Denmark (Dansk Folkeparti, 2002). Furthermore, the party is Eurosceptic (Dansk Folkeparti, 2002).

Before the establishment of the DF, the FRP (Progress Party) was the most right-wing party in Denmark (NSD, 2014). It was out of this party, that Pia Kjærsgaard formed the new and now much more successful DF (Bjørklund, n.A.).

Germany. The final country in this study that will be considered is Germany. The largest and most

notable right-wing party of recent years is the “National Democratic Party of Germany” (NPD) (NSD, 2014). After having been established in 1964, the party faced struggle such as attempted prohibition of its existence to growing success in recent years (NPD, 2013). Not only is the party Eurosceptic, but it is also highly anti-immigration especially concerning Islamic people (NPD, 2013). According to the NPD, immigrants are responsible for the unemployment of natives and immigrants only wish to exploit the social and financial support system of Germany (NPD, 2013). Compared to the major political parties of Germany and also to the extremist parties of other countries of this dataset, the NPD has had very little success in elections, with a maximum of 1.6% in 2005 (Zicht, 2013).

Based on this analysis, here is the hypothesis:

H3: The greater the anti-immigrant sentiment in the receiving country, the smaller immigration to that country will be.

2.4 The Financial Crisis and European Migration

One very recent event that largely impacted numerous areas of societies all over the world is the financial crisis of 2008 (for more information, see Box 3). First and foremost, the crisis affected the banking industry and financial markets, but also sectors, such as the automobile industry (Britannica, 2014). But the effects can be felt in almost all areas including the movement of people:

“The global financial crisis can be viewed as having a deeper and more global effect on the movement of people around the world than any other economic downturn in the post-WWII era of migration” (Fix et al., 2014). It appears that immigrant workers are hit harder than non-immigrant

workers (Fix et al., 2014; OECD, 2013). This is due to certain disadvantages they have compared to the nationals, such as language skills and educational background (Fix et al., 2014). This could lead to the assumption that migration flows might decrease due to the financial crisis (Fix et al., 2014). The recession even led to so-called return migration in which immigrants have to return to their home countries, because they cannot make a living in their host country (Fix et al. 2014).

(13)

13 the Netherlands, Norway, and Spain there is an overrepresentation of immigrants in temporary employment (OECD, 2013).This brings us to the following hypothesis:

H4: Immigration to country A goes down if the country was affected by the financial crisis.

3. Methodology

In the following section, the methodological part of the research will be discussed. First of all, the gravity model will be explained. This is followed by the additional hypotheses and the final model for this research. Next, the data will be discussed and as well as the estimation technique used.

3.1 Gravity Model

The gravity model is based on the works of Newton and was first applied to the topic of trade by Tinbergen about 50 years ago (Chaney, 2011). The basic version of the gravity model looks as follows:

(1)

where i and j are two countries, GFij is the gravitational force between i and j, Mi and Mj are the masses of the two countries, and Dij is the distance. Oftentimes, GFij is measured in terms of trade of flows or exports and the masses are measured in terms of GDP (Reinert n.d.). Furthermore, the gravity model tends to be estimated in log form, so that everything that was previously multiplied is now added and the division become distractions (Reinert n.d.). When applied to trade (and in this case migration) the following basic model is formed:

Box 3 The Financial Crisis of 2008

(14)

14 (2)

We can now adjust this basic model to the needs of our research. This is the basic equation to which further variables will be added. This is an established model that has been applied to the topic of migration before. Also, it includes the economic mass and the distance of each country pair, which are both important aspects in the study of migration. It therefore is a base model which gives reliable results and is suitable for the purpose of this study. Previous studies have applied it before (Ramos & Surinach, 2013).

One important aspect to consider in terms of the gravity model when applied to trade is the zero problem, which occurs when observations in the data set show zero trade flows (Haq et al., 2011). To deal with this problem, researchers have in the past either omitted those observations or replaced them with small positive values, which is necessary because the gravity model tends to be estimated in a log form and taking the log of zero is not possible (Haq et al., 2011). Research has shown that

“whenever zero trade flows are prominent, (...) selection bias should be taken into account in estimating the gravity model” (Haq et al., 2011). This problem can also occur in terms of migration

flows. In our case, there are very few observations that have zero migration flows from country A to country B. During the regression analysis, these observations were omitted. Since the number of omitted observations in relation to the total number of observations is so small and therefore not prominent, this does not pose a problem in terms of a selection bias and is therefore of no concern for this research.

3.2 Hypotheses

The main variables under consideration have already been introduced above. Additionally, the base variables of the gravity model as well as a number of control variables are also included in the final model, which will now be discussed.

GDP. The first independent variable is a crucial element of the gravity equation, namely GDP. We

will include the GDP of the sending country and the GDP of the receiving country. This will measure the mass of each of the economies. The data was taken from the World Bank. For the GDP of the receiving country, a positive relationship is expected. That is, the higher the GDP of the receiving country, the higher the migration inflows. Secondly it is expected that the smaller the GDP of the sending country, the larger the migration outflows.

H5a: There is a positive relationship between the GDP of the receiving country and immigration to that country.

H5b: There is a negative relationship between the GDP of the sending country and emigration from that country.

Distance. The second crucial variable is distance, which measures the geographical distance

between the countries in question. To measure this, the great circle approach was used. Here, we expect the dependent variable to go down as distance increases.

H6: There is a negative relationship between the distance of country A to country B and immigration from country A to country B.

Difference in GDP per capita. This variable measures the difference in GDP per capita between the

(15)

15 GDP of the receiving and the sending country is simply calculated. It is represented as a percentage of the other GDP per capita. It is expected that as the difference goes up, migration goes up.

H7: There is a positive relationship between the difference in GDP per capita between country A and country B and immigration from country A to country B.

Unemployment Rate. The next variable included is the unemployment rate of the sending country.

People need a regular income, so having a job is crucial. If the job market conditions in the home country are bad, seeking a job elsewhere might be a major motive to emigrate. This variable is also related to the fear of many that immigrants merely come for the jobs. Consequently, it is expected that there is a positive relationship between the unemployment rate of the sending country and migration flow.

H8: The greater the unemployment rate of the sending country, the larger emigration will be.

Common Border. This variable is a dummy variable, in that it indicates whether the two countries in

question share a common border (0=no common border; 1=common border).

H9: Emigration from country A to country B goes up if there is a common border between the two countries.

Common Language. Again, this is a dummy variable (0=no common language; 1=common

language). We consider the official language at country level and discard any regional dialects, additional languages spoken etc.

H10: Emigration from country A to country B goes up if the two countries share a common language.

A final note on the main variables relating to membership of the EU and Schengen agreements, the financial crisis, and the anti-immigrant sentiment: it might be the case that none of these variables have an immediate effect but rather a delayed effect on migration flows. Therefore, lagged versions of these variables will also be included in the model to account for this possible delay. The hypotheses remain the same.

3.3 Model

By inserting the variables into the equation, the final model looks as follows:

ln migrationijt = β0 + β1 ln GDPSit + β2 ln GDPRjt + β3 ln distanceijt + β4 ln DGDPCijt + β5 borderijt + β6 languageijt + β7 ln UESit + β8 EUSit + β9 lagEUSit + β10 EURjt + β11 lagEURjt + β12 SchengenSit + β13 lagSchengenSit + β14 SchengenRjt + β15 lagSchengenRjt + β16 ln AISjt + β17 ln

lagAISjt + β18 FCt + β19 lagFCt + eijt

As you can see, the log form was used and all variables were plugged in. An error term was added at the end of the equation. The subscripts i and j stand for the sending and receiving country, respectively. The term t indicates the time.

3.4 Data

(16)

16 problematic. That is, the data set had to be reduced based on the GDP data availability in the given time period. In the next data collection round, the migration flow data was collected, which posed even more challenges. The data for many countries is either not available at all, or only available for outflows (we need inflows as well outflows for each country), or only available for a very limited number of years. Therefore, the final sample only consists of eight Eastern and Western European countries. The selection of these countries is merely based on data availability. The following countries are included in this study: Germany, Denmark, Norway, The Netherlands, Poland, Luxembourg, Spain, and Hungary. Since the gravity model is used in this research, each country is paired with each of the other countries with migration flows going in either direction. Considering the time span from 1996 to 2011, this leaves us with a total of 56 pairings and 896 observations. The data for our first variable, which is Migration, was taken from the World Bank Global Bilateral Migration Database. For the dependent variable, flow data is used rather than stock data; however, there is a lack of flow data on European migration patterns, especially for the earlier part of the period in question. We will therefore have to accept missing data and we would propose to repeat the research once more complete data is available on the subject.

The three variables relating to GDP also had data gaps. These, however, were all cleared up when the majority of countries was excluded from the research. The data was also taken from the World Bank database and is in current US dollar (see reference list for databases used). Even though current US dollar is not corrected for inflation, this is no problem since all monetary variables are measured this way so relatively speaking the relationships are not impacted.

Distance was calculated using the greatest circle approach, whereby the air distance (as the crow flies) between two points is calculated. We used the capitals of all countries as reference points. The variable Common Border is straight forward: do the two countries in question have a shared border or not? Only land borders were taken into consideration, meaning if there is water between two countries and they are not directly connected, there is no common border. For instance, Iceland (not part of this study) has no common border with any other country.

For the variable Common Language the official languages in each country are considered. Dialects and languages that might be common in the country, but are not official, were excluded.

The data for the variables concerning EU and Schengen membership were taken from the official website of the European Union and were complemented by the official website of the Norwegian government, since Norway is not a member of the EU (see reference list for databases used). First of all, we have the dummy variables for EU-membership of the sending country (0=sending country is not EU member country; 1=sending country is EU member country) and EU-membership for the receiving country (0=receiving country is not EU member country; 1=receiving country is EU member country). We only consider membership and not candidate countries. The second Europe-specific variable is whether the sending and receiving country are in the Schengen are: Schengen-membership of the sending country (0=sending country is not in the Schengen area; 1=sending country is in the Schengen area) and Schengen-membership of the receiving country (0=receiving country is not in the Schengen area; 1=receiving country is in the Schengen area).

The data for the unemployment rate of the sending country was taken from the World Bank database on unemployment (see reference list for databases used).

(17)

17 each country in the sample for the given time-period were analyzed and the most right-wing and anti-immigrant parties were identified. Identifying these parties was a two-step process. First of all, the policies of all parties were analyzed using their official websites as well as articles by various news agencies, such as the Guardian or the BBC, to identify the most right-wing and anti-immigrant parties. For each election year, the party that had the highest percentage of national votes (not percentage of seats) out of that group was used. Alternatively, an aggregated number could have been used: however there were two things that made this solution unfeasible. First of all, there is always the risk of not including all relevant parties, which would lead to a biased result, especially for countries with very large number of political parties, such as Spain. Second of all, oftentimes, data is unavailable for smaller parties or those that did not receive enough votes to get a seat in parliament. Therefore, there would have been large data gap. That is why only the party with largest percentage of votes was considered.

All countries were impacted by the financial crisis of 2008, some more than others. However, this research does not consider the extent or impact on each country that the crisis had, but only whether the crisis was taking place or not. Since the crisis started in 2008, there is a dummy variable that indicates 0 for all years prior to 2008 and a 1 from 2008 onwards. It is expected that, generally speaking, the crisis has a negative impact on migration flows. Since this variable has the same values for each year for every country, no distinction between sending and receiving country needs to be made.

3.5 Estimation Technique

To estimate panel data using the gravity model, one has several possibilities, such as the fixed model, the random effects model, and the pooled model. The fixed model is used when one is

“interested in analyzing the impact of variables that vary over time” (Torres-Reyna, n.A.). This

means that time-invariant features are excluded from the analysis (Torres-Reyna, n.A.). Applied to this research, this means that a number of meaningful variables get omitted from the model. That is, variables such the dummy variable for the EU and Schengen membership get omitted which makes the output useless. Therefore, fixed effects could not be used. A second estimation technique is the random effects model. In the random effects model, “unlike the fixed effects model, the variation

across entities is assumed to be random and uncorrelated with the predictor or independent variables included in the model” (Torres-Reyna, n.A.). Here, the time-invariant variables remain

(18)

18 not did the two countries in question neither share neither a border nor a language. One outcome that is noteworthy is that for the difference in GDP per capita, which shows a minimum value of 19.21% and a maximum value of a staggering 520.70%. These results were generated when country pairs such as Luxembourg-Hungary or Norway-Poland were formed. These results are accurate and the mean reflects well the overall relationships between the variables. It shows that overall, migrants went to host countries that had an average of an additional 29.25% in GDP per capita than their home country. Finally, the three standard variables of the gravity model, namely the GDP of the sending country, the GDP of the receiving country, and the distance between the sending and the receiving country do not show any remarkable results.

There are a number of assumptions regarding this method, which need to be checked for violation. They are normality, multicollinearity, heteroskedasticity, and autocorrelation (Hill, Griffiths & Lim, 2012).

To ensure that the assumption of normality is fulfilled, the data for migration can be visually observed in a histogram. Since the data for migration flows is always positive, it can be assumed that the data will be positively skewed. When graphing the log version of the variable, however, the data appears to be normally distributed (see Appendix 2).

Secondly, we need to consider the issue of multicollinearity. When multicollinearity exists, it means that variables move together in systematic ways (Hill, Griffiths & Lim, 2012). A first look at the correlation matrix does not reveal any variables that are too largely correlated. Obviously, there is correlation between the linear variables and their log counterparts (as well as lags), but this is of no importance. In terms of correlations of other variables pairings, the largest correlation exists between the dummy variable EU-membership of the receiving country (Lagged) and the Schengen-membership of the receiving country (Lagged) (0.6150), which is reasonable since the Schengen area and the European Union are closely related. To test for multicollinearity, the VIF test is applied. For all the results, see Appendix 3. The first table in Appendix 3 shows the VIF output for the entire model. We can see that the VIF output is slightly higher for some variables. For instance, the variable with the largest value is the receiving country in EU (Lagged) with a VIF of 7.51. However, this outcome should not concern us for two reasons. First of all, most of the variables with a high VIF are indicator variables, which will automatically lead to a higher VIF, since the values the variable can take are very limited (Allison, 2012). Secondly, we are using lagged variables, which means that there is already a close relationship between many of the variables (Allison, 2012). If we run the model without the lags (see second table in Appendix 3), the VIF are much smaller and of no concern whatsoever. Overall, the multicollinearity can be safely ignored in this case.

Thirdly, heteroskedasticity needs to be considered. When heteroskedasticity exists, the variances for all observations are not the same (Hill, Griffiths & Lim, 2012). To test this, the Breusch-Pagan Test is used (see Appendix 4). The outcome of the test shows that the chi-squared value is rather large whereas the p-value is rather small. This indicates that heteroskedasticity is present.

(19)

19

4. Empirical Results

Table 1 shows the results for the model. The first column shows all the variable names as well as the constant, the F-value/Wald chi2-value, and the R-squared for the model. Column two gives the regression results for the model excluding the main variables relating to Schengen, the EU, the financial crisis, as well as the anti-immigrant sentiment. The third and fourth columns give the results for the pooled model using White’s Standard Errors and the random effects model using White’s Standard Errors, respectively. All three models are highly significant and have a relatively high R-squared.

Table 1: Regression Results

Ln(Migration) 1 2 3

GDP Sending Country (Log) .9999587*** 1.093163*** .5221733 *** GDP Receiving Country (Log) .8587452*** 1.047025*** .5022337*** Distance (Log) -.7570453*** -.6333227*** .0845827 Difference in GDP per Capita (Log) -1.56147*** -1.561236*** -1.001483 *** Unemployment Rate in

Sending Country (Log)

.7063878*** .5342253*** .0350242 Common Border .0256029 -.0087502 1.849046*** Common Language .8289185*** 1.05443*** -.6673859 Sending Country in EU -.1549351 -.2853555 Sending Country in EU (Lagged) .3058873 .1702941 Receiving Country in EU -.8280128*** .1250469 Receiving Country in EU (Lagged) .6019747** .4702774** Sending Country in Schengen Area -.9061404*** -.4093868*** Sending Country in

Schengen Area (Lagged)

.4911099** .3683391** Receiving Country in

Schengen Area

.266641 .2560009** Receiving Country in

Schengen Area (Lagged)

-.4518878* -.2812671** Anti-Immigrant Sentiment (Log) .1704706*** .0875693* Anti-Immigrant Sentiment (Log, Lagged) -.0249101 -.0236917 Financial Crisis -.6234202*** -.0781852

Financial Crisis (Lagged) .0350641 .0097653 Constant -39.08581*** -46.5274*** -24.72928 ***

F-value 418.91*** 183.42***

Wald chi2 283.64***

R-squared 0.7765 0.8100 0.6717

(20)

20 First of all, we can see that there are some differences for the overall pooled model and the random effects model (both using White’s Standard Errors). This indicates that the model might be too much estimation technique dependent. It is hard to say which model is more accurate; however, the pooled model has similar results to the first base model, which might indicate that these results are more accurate. Furthermore, the pooled model explains an additional 3.35% of the dependent variable as compared to the reduced model, whereas the random effects model explains less, namely 10.48% less. In the first model, all variables are highly significant at a 0.01% level except for the Common Border variable. The same is true for all of these variables in the second model. The third model differs from these results. Here, the distance variable is insignificant as well as the common language dummy, and the unemployment rate of the sending country, whereas the common border dummy is highly significant. The fact that the distance variable is insignificant when it is a base variable of the gravity equation could be an indicator that the random effects model is not the best choice to estimate the model. The first variable is the log form of the GDP in the sending country, which has a positive sign in all three models. The coefficient for this variable is much smaller in the third run as compared to the first two runs. This result does not confirm hypothesis H5b. Moving on to the second variable, which is the log version of the GDP of the receiving country, we see that all models generate a positive sign for this variable and again the coefficient is much smaller for the random effects estimation. This confirms hypothesis H5a. The next base variable of the gravity model is the log version of the distance between the two countries. In the first two runs, the sign is negative and the coefficient decreased slightly in the second run. This confirms hypothesis H6. However, the coefficient is much smaller and insignificant in the random effects model, which would not confirm the hypothesis. For the next variable, which is the difference in GDP per capita between the two countries, the coefficient remained virtually the same in the first two runs and decreased slightly in the third run. The negative sign in the models comes as a surprise, meaning that hypothesis H7 is not confirmed. The log version of the unemployment rate in the sending country decreased in the second model as compared to the first. These two runs show a positive sign for this variable, which confirms hypothesis H8. The third model generates an insignificant result which does not confirm the hypothesis. Since the coefficients for Common Border are not significant in the first two models, hypothesis H9 cannot be confirmed by them. The variable is, however, highly significant in the third run and the positive sign confirms hypothesis H9. The final control variable is Common Language, which has a positive value in the first and second model with a slightly higher coefficient in the second run. The positive sign of the models confirms hypothesis H10. The variable is insignificant in the third run, so in this case the hypothesis cannot be confirmed.

(21)

21 well the lagged form, which are significant at a 0.01% and 0.05% level respectively for both models. Again, we can see that the sign is negative at first but positive for the lagged variable, which confirms hypothesis H1a. In terms of the receiving country being part of the Schengen area, the normal form of this variable is insignificant, whereas the lagged version is significant at a 0.05% level in the pooled model. The random effects model generates two significant results, the signs are the same. The sign for the lagged form is negative in both models, so hypothesis H1b is not confirmed. It can be confirmed in its normal form for the random effects model. Next, we have the log version and its lagged counterpart of the anti-immigrant sentiment in the receiving country. Whereas the regular form is significant in both models, the lagged version is insignificant for both models. The former has a positive sign which means that hypothesis H3 is not confirmed. Finally, the results for the financial crisis variables render insignificant results for the random effects model. For the pooled model, the normal version has a negative sign and is highly significant. The lagged form is insignificant. Hypothesis H4 is confirmed for the pooled model, but not for the random effects model. As we can see, the results differ for some variables between the pooled and the random effects model and are similar for others. It is therefore hard to say what the correct interpretation should be.

5. Discussion

Analysis

(22)

22 more than an immediate effect. The second variable related to the Schengen area is about the receiving country being part of it. Again, we looked at the immediate as well as the lagged effect. For this variable, the normal version of the variable generated insignificant results for the pooled model but significant results for the random effects model. But both models generated significant results for the lagged form. The outcome of the lagged version goes against the assumption that migration will go up if the receiving country is in the Schengen area. This can be explained as follows: as mentioned before, the borders within the Schengen area are very soft whereas the outside borders of the Schengen area are extremely hard. In other words, migration within the Schengen area is easy, but migrating into the Schengen area from the outside is very hard. In this case, it was found that overall (inside and outside) migration inflows to a country go down if that country is within the Schengen area. It would be interesting to look at this issue in a more in-depth study to determine whether this effect is indeed positive for Schengen to Schengen migration and negative for non-Schengen to Schengen migration. Overall, it was in this case expected that migration to Schengen countries would still be more popular, even though the external borders are very hard. But it appears that the borders are indeed so hard, that immigration from non-Schengen to Schengen countries negatively outweighs the within Schengen immigration. So in that sense, the Schengen Agreements that were designed to facilitate migration within the area actually reduce overall migration flows due its hard external borders. Overall, it appears that for the EU and Schengen related variables, the lagged versions generate the more plausible results given the reasoning applied in this paper. However, most outcomes for the normal forms of the variables also generated significant results, but with different signs. This outcome is very puzzling and should be looked into further in future research.

The next main variable that was included in the model is the anti-immigrant sentiment in the receiving country. The lagged version of this variable did not render significant results in either model which is rather surprising. A more abstract concept as the anti-immigrant sentiment could be expected to have a more delayed effect, but this is not what was found here. Furthermore, the immediate effect is a positive one in both models, which is also not what was expected. A possible explanation for this puzzling outcome is that the relationship between the anti-immigrant sentiment and inflows of immigrants has in fact a reverse relationship. That is, the more immigrants enter a country, the stronger the anti-immigrant sentiment in that country will be and not vice versa. This is of course just a theory and will have to be researched further in future studies.

(23)

23

Limitations

First of all, the sample is very limited in terms of number of years and number of countries. This is due to data being largely unavailable. It would be interesting to repeat the research once a more extensive dataset can be collected and to see how the outcomes would change.

Secondly, migration flows are not only influenced by the variables used in this research. There is much research on other variables that have been found to have an influence. Adding more variables to the model might also affect the outcome and lead to a higher accuracy of the model.

Lastly, there are alternative ways to measure some of the variables. For instance, the size of an economy can also be measured in terms of population and the distance does not necessarily need to be measured sung the great circle approach.

Conclusion

(24)

24

References

Allison, P. (2012). When Can You Safely Ignore Multicollinearity?. Available: http://www.statisticalhorizons.com/multicollinearity. Last accessed 1st Jul 2014.

AXT. (1997). Spain. Available: http://www.axt.org.uk/antisem/archive/archive1/spain/index.htm. Last accessed 20th Jun 2014.

BBC. (2014). Dutch free killer of anti-Islam politician Pim Fortuyn. Available: http://www.bbc.com/news/world-europe-27261291. Last accessed 6th Jul 2014.

Bertoli, S., Brücker, H. and Fernández-Huertas Moraga, J. (2013). The European Crisis and Migration to Germany: Expectations and the Diversion of Migration Flows. Available: http://ftp.iza.org/dp7170.pdf. Last accessed 26th Jun 2014.

Bjørklund, T. (n.A.). Extreme Right Parties in Scandinavia An overview. Available:

http://www.politik.uni-mainz.de/ereps/download/scandinavia_overview.pdf. Last accessed 21st Jun 2014.

Britannica. (2014). The Financial Crisis of 2008: Year In Review 2008.Available:

http://www.britannica.com/EBchecked/topic/1484264/The-Financial-Crisis-of-2008-Year-In-Review-2008. Last accessed 25th Jun 2014.

Calabria, S. (2014). Right-Wing Dutch Politician Geert Wilders May Face Legal Action After Promising ‘Fewer' Moroccans In The Netherlands. Available:

http://www.huffingtonpost.com/2014/03/20/dutch-politician-moroccan_n_5001762.html. Last accessed 20th Jun 2014.

Chaney, T. (2011), The Gravity Equation in International Trade: An Explanation. University of Chicago, NBER and CEPR.

Cochrane, C. & Nevitte, N. (n.A.). Support for Far-Right Anti-Immigration Political Parties in Advanced Industrial States: Insiders, Outsiders and Economic Disaffection. Available:

http://individual.utoronto.ca/cochrane/farright.pdf. Last accessed 20th Jun 2014.

Collett, E. (2011). Faltering Movement: Explaining Europe's Schengen Struggle. Available:

http://www.migrationpolicy.org/article/faltering-movement-explaining-europes-schengen-struggle. Last accessed 24th Jun 2014.

Dansk Folkeparti. (2002). The Party Program of the Danish People's Party.Available:

http://www.danskfolkeparti.dk/The_Party_Program_of_the_Danish_Peoples_Party. Last accessed 21st Jun 2014.

Day, M. (2014). EU elections 2014: the Polish party that wants to turn EC building in Brussels into brothel. Available: http://www.telegraph.co.uk/news/worldnews/europe/poland/10836329/EU-elections-2014-the-Polish-party-that-wants-to-turn-EC-building-in-Brussels-into-brothel.html. Last accessed 20th Jun 2014.

(25)

25 Europa. (2009). The Schengen area and cooperation. Available:

http://europa.eu/legislation_summaries/justice_freedom_security/free_movement_of_persons_asylu m_immigration/l33020_en.htm. Last accessed 23rd Jun 2014.

Fix, M., Papademetriou, D.G., Batalova, J., Terrazas, A., Lin, S.Y. and Mittelstadt, M. (2009). Migration and the Global Recession. Available: http://www.migrationpolicy.org/. Last accessed 26th Jun 2014.

Fremskrittspartiet. (2013). History . Available: http://www.frp.no/nor/The-Progress-Party/History. Last accessed 21st Jun 2014.

Grabbe, H. (2000). The Sharp Edges of Europe: Extending Schengen Eastwards. International Affairs. 76 (3), 519–536.

Hainmueller, J. and Hiscox, M.J. (2007). Educated Preferences: Explaining Attitudes toward Immigration in Europe. International Organization. 61 (2), 399-442.

Haq, Z., Meilke, K. and Cranfield, J. (2011). The Gravity Model and the Problem of Zero's in Agrifood Trade. Available: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2294005. Last accessed 2nd Jul 2014.

Hernandez-Carr, A. (2012). Right-wing extremism in Spain: An approaching breakthrough?. Available: http://extremisproject.org/2012/08/272/. Last accessed 20th Jun 2014.

Hill, R.C., Griffiths, W.E. & Lim, G.C. (2012). Principles of Econometrics. Asia: John Wiley & Sons Pte Ltd.

Jobbik. (2010). The Movement For a Better Hungary. Available: http://www.jobbik.com/about_jobbik. Last accessed 19th Jun 2014.

aczmarczyk, P., Lesi ska, M. Ok lski, M.. (2012). New Patterns of European Migration: A Broad View from a Multidisciplinary Perspective. ANU Centre for European Studies Briefing Papers. 3 (10), 1-31.

Luxemburger Wort. (2013). All about the Alternativ Demokratësch Reformpartei (ADR).

Available: http://www.wort.lu/en/community/all-about-the-alternativ-demokratesch-reformpartei-adr-4f60c61be4b047833b93b9d8. Last accessed 20th Jun 2014.

McKinney, W.C. (2002). Present State of Politics in Hungary.Available: http://www.unc.edu/~wmckinne/politics.html. Last accessed 19th Jun 2014 Merriam-Webster. (2014). Diaspora. Available: http://www.merriam-webster.com/dictionary/diaspora. Last accessed 10th Jul 2014.

Nome, M. (2013). The Norwegian shift to the right: 10 short lessons on The Progress Party.

Available: http://www.opendemocracy.net/magnus-nome/norwegian-shift-to-right-10-short-lessons-on-progress-party. Last accessed 21st Jun 2014.

(26)

26 NSD. (2014). European Election Database. Available:

http://www.nsd.uib.no/european_election_database/country/. Last accessed 19th Jun 2014. OECD. (2013). International Migration and the Economic Crisis: Understanding the Links and Shaping Policy Responses. Available: http://www.oecd.org/els/mig/46292981.pdf. Last accessed 26th Jun 2014.

Ortega-Argilés, R., Piva, M. Potters, L, and Vivarellli, M. (2010). Is Corporate R&D Investment in High-Tech Sectors More Effective?. Contemporary Economic Policy. 28 (3), 353-365.

Ortega-Argilés, R., Potters, L. and Vivarelli, M. (2011). R&D and productivity: testing sectoral peculiarities. Empirical Economics. 41 (1), 817-839.

Oxford Dictionaries. (2014). Falange. Available:

http://www.oxforddictionaries.com/definition/english/Falange. Last accessed 20th Jun 2014. Parlement. (n.A.). Leefbaar Nederland. Available:

http://www.parlement.com/id/vh8lnhrouwy7/leefbaar_nederland. Last accessed 6th Jul 2014. Paterson, T. (2014). Hungary election: Concerns as neo-Nazi Jobbik party wins 20% of vote. Available: http://www.independent.co.uk/news/world/europe/concerns-as-neonazi-jobbik-party-wins-20-of-hungary-vote-9244541.html. Last accessed 19th Jun 2014.

PiS. (2014). History of Law and Justice. Available:

http://www.pis.org.pl/unit.php?o=partia&p=desc. Last accessed 20th Jun 2014.

Piskorski, M. (2004). Self Defense of the Republic of Poland - Who we are and what we stand for. Available: http://www.samoobrona.org.pl/pages/20.english/index.php?document=presentation.html. Last accessed 20th Jun 2014.

PVV. (2014). Partij Voor de Vrijheid. Available: http://www.pvv.nl/. Last accessed 20th Jun 2014. Ramos, R. and Surinach, J. (2013). A Gravity Model of Migration Between ENC and EU.

Available: http://www.iza.org/en/webcontent/publications/papers/viewAbstract?dp_id=7700. Last accessed 26th Jun 2014.

Reinert, K.A.. (n.A.). Gravity Models. Available:

http://press.princeton.edu/chapters/reinert/19article_reinert_gravity.pdf. Last accessed 15th Apr 2014.

Repa, J. (2006). Polish nationalism resurgent. Available:

http://news.bbc.co.uk/1/hi/world/europe/4754079.stm. Last accessed 20th Jun 2014. Rogoff, K. (2011). The Second Great Contraction. Available:

http://www.theglobeandmail.com/globe-debate/the-second-great-contraction/article589111/. Last accessed 2nd Jul 2014.

(27)

27 Short, V. (2014). New far-right party VOX launched in Spain. Available:

http://www.wsws.org/en/articles/2014/02/25/spai-f25.html. Last accessed 20th Jun 2014. Torres-Reyna, O.. (n.A.). Panel Data Analysis Fixed & Random Effects. Available: http://dss.princeton.edu/training/Panel101.pdf. Last accessed 4th Jul 2014.

UN. (2013). International Migration Press Release. Available:

http://esa.un.org/unmigration/wallchart2013.htm. Last accessed 21st Jun 2014.

Zicht, W. (2013). Ergebnisse der Bundestagswahlen. Available:

http://www.wahlrecht.de/ergebnisse/bundestag.htm. Last accessed 21st Jun 2014.

Databases Used:

Database Variable(s) Access Date Link

(28)

28

Appendix

Appendix 1: Summary Statistics

Variable Observations Mean Standard

Deviation

Min Max

Migration 851 3.870246 15.26164 0 164.705 GDP Sending

Country

896 6.47e+11 8.58e+11 1.85e+10 3.62e+12 GDP

Receiving Country

896 6.47e+11 8.58e+11 1.85e+10 3.62e+12

(29)

29 Appendix 2: Sktest of ln(Migration) to check visually for normality

Appendix 3: VIF Test

VIF Test including all variables

Variable VIF 1/VIF

(30)

30 Sending Country in Schengen Area 5.94 0.168455 Receiving Country in Schengen Area (Lagged) 5.88 0.170189 Sending Country in Schengen Area (Lagged) 5.87 0.170270 Difference in GDP per Capita (Log) 2.89 0.346578 Financial Crisis 2.74 0.364410 Financial Crisis (Lagged) 2.67 0.373889 Common Border 2.38 0.420996 Unemployment Rate in Sending Country (Log) 2.21 0.452436 GDP Receiving Country (Log) 1.93 0.518081 Distance (Log) 1.85 0.539772 GDP Sending Country (Log) 1.81 0.550992 Common Language 1.45 0.689752 Mean VIF 4.59

VIF Test excluding lagged variables

Variable VIF 1/VIF

(31)

31 GDP Receiving Country (Log) 1.88 0.531257 Distance (Log) 1.85 0.540178 GDP Sending Country (Log) 1.78 0.561691 Receiving Country in EU 1.75 0.571358 Anti-Immigrant Sentiment (Log) 1.72 0.580100 Sending Country in EU 1.71 0.583589 Common Language 1.45 0.691268 Financial Crisis 1.37 0.731686 Mean VIF 2.01

Appendix 4: Breusch – Pagan Test

Breusch-Pagan / Cook-Weisberg test for heteroskedasticity Ho: Constant variance

Variables: GDP Sending Country (Log), GDP Receiving Country (Log), Distance (Log), Difference in GDP per Capita (Log), Unemployment Rate in Sending Country (Log), Common Border, Common Language, Sending Country in EU, Sending Country in EU (Lagged), Receiving Country in EU, Receiving Country in EU (Lagged), Sending Country in Schengen Area, Sending Country in Schengen Area (Lagged), Receiving Country in Schengen Area, Receiving Country in Schengen Area (Lagged), Anti-Immigrant Sentiment (Log), Anti-Immigrant Sentiment (Log, Lagged), Financial Crisis, Financial Crisis (Lagged)

chi2(19) 72.62

Prob > chi2 0.0000

Appendix 5: Woolridge Test

Wooldridge test for autocorrelation in panel data

H0: no first-order autocorrelation

F( 1,55) 12.525

Referenties

GERELATEERDE DOCUMENTEN

Uit de resultaten bleek dat stress geen direct effect had op pornocraving, maar dat de relatie tussen stress en pornocraving inderdaad werd gemediëerd door cortisol,

Paragraph 5.1 will report on the results regarding the degree of internal democracy regarding the selection process of lobby points within the refugee

Politieke leiders kunnen een grote verscheidenheid aan motivaties hebben om te kiezen voor een bepaalde strategie in het kader van hun schuldmanagement middels

Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of

temporary reintroduction of border controls at the Swedish internal borders in accordance with Article 23 and 24 of Regulation (EC) 562/2006 establishing a Community Code on

To analyse the impact of the GFC this paper re-calibrated/re-estimated the six-equation model of Jacobs, Kuper and Ligthart (2010) for the period 1980Q1–2009Q4, and investi- gated

Bank risk-taking is defined as the ratio of risk assets to total assets and the bank-level lending rate is defined as the ratio of interest income to total loans.. A regression line

While investigating the impact of the East Asian crisis (1997-1998) on the capital structure of emerging market firms, Fernandes (2011) finds that while total