1
The impact of immigration on
house prices: Evidence from
2
Statement of Originality
This document is written by Berkant Shaban who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the
text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for
3 Introduction
Many developed countries in Europe, such as United Kingdom, Germany and the Netherlands, have been suffering an immigration inflow during the last decades. Particularly, a simple reason for this immigration would be the fact that people from developing countries tend to move to developed countries seeking better living conditions. At the same time, many developed countries experienced the so called “Housing Bubble”, in which they faced a substantial increase in house prices during the early 2000s until the beginning of the crisis in 2008 (Gonzales and Ortega, 2013, p.37). Also, Gonzales and Ortega claim that the real causes for this bubble are still not understood, since many factors, such as the low interest rates and unemployment seem to have played a role on that (2013, p.37). However, many economists believe that a crucial role in predicting the increase in house prices was played by the increasing number of immigrants in the developed countries.
In principle, the economic theory would suggest that an increase in immigration in a given country would increase the total population, which in turn would increase the demand for housing, given a fixed level of house supply. Combined with an upward sloping supply curve, this would lead to an increase in house prices (Sa, 2014, pp: 2-3). A lot of research has been done on predicting the impact of immigration on the labor markets in many developed countries. However, this paper takes a different approach and tries to investigate the impact of immigration on house prices in the Netherlands. The Netherlands has been number one choice for immigrants from Turkey, Morocco and Indonesia, and respectively those
countries make the highest immigrant population living in the Netherlands. Moreover, the Netherlands is part of the European Union (EU), which faced a substantial enlargement during the last two decades by increasing the number of countries from 14 in 1995 to 27 in 2013. A crucial point is the year 2004, when 10 countries joined the EU in the so called “largest single expansion of the European Union”. Being part of the EU, a person is allowed freely to move across countries to search for better life opportunities, therefore one would expect a higher immigration across countries in the EU. Consequently, this leads to the research question of this paper:
“What is the impact of immigration on house prices in the Netherlands during the last two decades?”
The findings reported by the various researches have been inconsistent. Many economists find
significant results supporting the standard economic theory about the impact of immigration on house prices. On one hand, evidence from Canada suggests a 0.10- 0.12% increase of house prices due to immigration (Akbari& Aydede, 2012, p.1657). The authors also point out that this increase is caused mostly by the immigrants who moved there 10 years ago, meaning that newcomers actually did not play a big role, since they could not afford buying houses immediately. This is also supported by a research from Spain, where authors show that 25% of the total 175% increase in house prices during the Housing bubble is caused by immigrants (Gonzales& Ortega, 2013). On the other hand, Sa finds that immigration flow equal to 1% of the local population actually reduces house prices in England and Wales by 1.7%, thereby showing that his results contradict the standard economic theory (2014, p.30). The main phenomenon behind his results is the fact that immigration generates one-for-one emigration of native people.
4
Netherlands had a stable and substantial economic growth during the last decades, which makes it one of the most powerful economies in the world. Its Gross domestic product (GDP) value per capita has increased substantially by 85% in nominal values since 1995, ranking it currently at 12th place in the world by The International Monetary Fund (IMF) and The World Bank. At the same time, the average purchase price of a house in Netherlands has increased tremendously by nearly 147% since 1995, which is nearly double the increase of its GDP. Akbari & Aydede also point out the same situation that the house prices in Canada sharply increased compared to increases in per capita incomes (2012, p.1645). Moreover, given a data for the same time period from Central Bureau of Statistics Netherlands (CBS), the population of Netherlands has increased by 9.15%, while the amount of immigrants increased on
average 4.16% per year since 1995. In 1995, immigrants entering the Netherlands accounted for 0.62% of the total population, whereas in 2014 they account for 1.07% of the total population.
The structure of the thesis is as follows. Section 2 briefly summarizes the theoretical framework, which includes the literature review and a brief analysis of the main determinants of house prices. Section 3 contains information about the methodology of the investigation and its data, respectively. Section 4 includes the analysis and the results of the model, followed by Section 5, which includes the concluding remarks of the research and its limitations.
Theoretical framework 2.1 Literature review
To begin with, there has not been any specific research made concerning the impact of immigration on house prices in the Netherlands, neither for the labor market. All the following literature review is based on the impact of immigration on house prices in different countries, meaning that any conclusion made from those papers cannot be used as a prediction of the impact of immigration on house prices in the Netherlands. The basic reason for this is the vast differences among the countries and their economies, even though they share similar economic and financial power as the Netherlands.
In his paper “Immigration and housing rents in American cities”, Albert Saiz (2007) analyses the short run impact of immigration inflows in American metropolitan cities on the housing market, while controlling for other variables such as the GDP, Unemployment rate and so on. He uses Census decennial data on the stock of the foreign born, housing rents and home values as a research approach (2007, p.346). He points out the standard economical belief that, in principle new immigrant demand for housing coupled with an upward-sloping housing supply in the metropolitan areas where immigrants settle should yield rising housing rents and prices (2007, p.346). Moreover, the sign and the magnitude of the impact can vary and even might be different than expected, due to the fact that immigration might be associated with offsetting native out-migration (2007, p.346). According to his instrumental-variable estimates, an immigration inflow equal to 1% of the city’s population is associated with an 1% increase in average house prices , therefore suggesting a causal relationship between house prices and immigration, and relevance with the economic theory for the impact of immigration (Saiz, 2007).
Gonzales and Ortega (2013) show similar results with positive impact in their research. They investigate the impact of immigration on house prices in Spain over the period 2000-2010 based on an instrumental-variable estimation. Their basic research indicates that house prices increased by 175% between 1998 and peak of the boom in 2008, declining afterwards during the crisis (2013, p.37). Also, they show that
5
the size of foreign born population increased from 0.5 million to 5 million leading to a growth of the population ranging between 1.5 -2% a year between 1998 and 2008, also declining after 2008 due to the financial crisis (20013, p.38). Their results suggest that immigration had positive significant effects on the housing market in Spain, amounting to 25% from the total increase in house prices, or in other words, a 1% increase in immigration leads to an increase of 1-1.6% of house prices in the following period (Gonzales & Ortega, 2013). Moreover, they show that a 1% increase in immigration leads to 0.8- 1% increase in the number of dwellings built in the following year (2013, p.41). The authors conclude that the reason why the housing boom was larger in Spain than U.S. and the other European countries was due to larger inflows of immigrants, relative to population (2013, p.57).
Akbari & Aydede investigate the impact of immigration on house prices in Canada between 2001 and 2006. They use a panel data based on 1996, 2001 and 2006 population censuses (2011, p.1646). To investigate their research, they use a similar approach as Saiz (2007), whose research as shown above, was about the impact of immigration on house prices in American metropolitan cities. The basic research of Akbari & Aydede suggests that, during that period, the amount of immigrant population in Canada increased by 13.6%, which is about four times the growth of the Canadian-born population (2011, p.1656). Also, they estimate that the house price to income ratio rose from 3.5 in 2001 to 5 in 2006, however, the exact impact of immigration on this change is not known (2011, p.1645). Furthermore, Akbari & Aydede find significant positive impact of immigration on house prices. Their results indicate that a 1% increase in immigration leads to 0.1- 0.12% increase in house prices (2011, p.1657). As can be seen, those results are far less than the results of Saiz (2007), who found that a 1% increase in
immigration leads to 1% increase in house prices in American metropolitan cities. Also, Akbari & Aydede conclude that in fact this positive effect of 0.1- 0.12% is actually caused by immigrants who moved to Canada more than 10 years ago, meaning that newcomers do not really cause a change (2011, p.1657). Moreover, the authors believe that the main cause of this result could be the emigration of natives from the areas that immigrants settle, or the increased supply of housing due to expectations of higher demand in those areas (2011, p.1645).
On the other hand, the author of the article “Immigration and house prices in the UK” Filipa Sa, finds significant negative effect of immigration on house prices in UK. His paper follows the dominant
methodology in the literature on immigration impacts and uses a spatial correlations between changes in the stock of immigrants and changes in house prices in different geographic areas (2014, p. 2). Sa’s basic estimates show that in mid-1990s immigrants accounted for just over 8% of the working age population in UK, whereas today they account for more than 13% (2014, p.1). At the same time, he estimates that house prices nearly tripled between 1995 and 2007, which was then followed by the financial crisis and a recovery in 2010 (2014, p.1). Moreover, Sa believes that there is a substantial variation across local authorities; for example the largest share of immigrants in working age population in 2010 are registered in some London boroughs, where the majority of the population are immigrants. However, for Outside London, there are several local authorities where over 20% of the population is foreign born. (2014, p.1). Also, the highest house prices are registered again in London boroughs (2014, p.1). However, his results show a negative impact of immigration on house prices, which contradicts the main economic theory. He finds that an increase in immigrant population equal to 1% of the local population reduces house prices by 1.7% (2014, p.30). Sa presents an evidence that this negative impact is due to the mobility response of the native population, who have slightly higher income than the others, meaning that they have the opportunity to move to other areas to live in.
6
From the above literature review, one can conclude that, even though the estimates are significant, the sign and magnitude can differ across countries. This means that, we cannot form expectations about the impact of immigration on house prices in Netherlands given the recent literature review of other
countries. However, positive expectations might be formed due to the mobility factor in the EU. In the following subsections, a brief description about the main variables used in the model for Netherlands will be given, namely house prices, immigration, GDP and Unemployment
2.2 House prices in Netherlands
To begin with, the corresponding tables and charts are based on a quarterly data for the period between 1995 and 2015. The average purchase price shows the mean value paid in the period under review for dwellings sold to a private person and intended for permanent residence by a private person (Central Bureau of Statistics Netherlands). The Netherlands has seen substantial increase of nearly 147% in average house prices during the period 1995-2015, which can be seen in Table 2.2:
Table 2.2; Source: CBS; the graphs was made by the author
It is also clear that the increase is stable until the 3rd quarter of 2008, where it reaches its maximum of slightly more than 250000 euros on average per dwelling, and then due to the financial crisis it starts falling. Furthermore, basic analysis shows that the increase in average house prices during the period 1995-2008 was nearly 189%, which is close but higher than the findings of Gonzales & Ortega in their research about Spain, who reported a 175% increase in house prices. Moreover, the decline in the average house prices between the peak in 2008 and the downturn until 2013 was nearly 15.5%, which is about 3.1% percent per year on average. It is worth mentioning that this data includes the overall average price of all kind of dwellings that were available for purchase, meaning that there has not been any distinctions made on the size and the location of the respective dwelling.
However, According to Central Bureau of Statistics Netherlands (CBS), the House Price Index (HPI) is a better indicator of house price movements than the average house purchase price. Simply, the HPI shows the price movement of all dwellings owned by a private person and intended for permanent residence by a private person. According to the article “Why the average dwelling purchase price is not
0 50000 100000 150000 200000 250000 300000 Q 1 1995 Q 4 1995 Q 3 1996 Q 2 1997 Q 1 1998 Q 4 1998 Q 3 1999 Q 2 2000 Q 1 2001 Q 4 2001 Q 3 2002 Q 2 2003 Q 1 2004 Q 4 2004 Q 3 2005 Q 2 2006 Q 1 2007 Q 4 2007 Q 3 2008 Q 2 2009 Q 1 2010 Q 4 2010 Q 3 2011 Q 2 2012 Q 1 2013 Q 4 2013 Q 3 2014 Eu ro s Period
7
an indicator” by the CBS, the average selling price reflects the average price of houses sold in a particular month, however, the houses sold every month are very different and the difference in their
characteristics are not taken into account. For example, in January more expensive houses are sold, while in February cheaper houses are sold; this causes a decrease in average selling price, however, in reality it is not the case. But, the price index of existing houses does take these differences in
characteristics into account. Table 2.2.1 shows the HPI for the Netherlands:
Table 2.2.1; Source: CBS; the graph was made by the author
This table has similar picture as Table 2.2 in its design, however, it shows a 191.5% increase in the HPI on average between 1995 and 2007, which is higher than the result found for average house price increase. The HPI faces a downturn of 17% on average since 2007, and its quarterly average for the period 1995-2015 is 1.1%. Moreover, HPI will be used as the main indicator of house prices in the regression model of this paper.
Moreover, according to Eurostat’s House Price Index (HPI) analysis (CBS Article, 2013), which compares house prices internationally, Dutch house prices fell by more than EU average during the financial crisis. They show that house prices in other EU countries also peaked just before the financial crisis, but had different house price movements afterwards. For example, in Spain house prices continued to fall by 7% on average per year during that period, while in the UK house prices fell and then recovered quickly, and also that the financial crisis hardly seem to had an effect on house prices in Germany.
Furthermore, to give a better insight on the house market in the Netherlands, the number of dwellings sold per year was analyzed. According to the CBS, the number of dwellings sold is the number of registered transactions by the Dutch Land Registry Office (Kadaster) of dwellings sold by a private person, measured at the end of the period under review. The results indicate that on average 171919 houses were sold per year given the period 1995-2015. Table 2.2.2 shows the trend of dwelling sold given the period:
0 20 40 60 80 100 120 Q 1 1995 Q 4 1995 Q 3 1 99 6 Q 2 1997 Q 1 1998 Q 4 1998 Q 3 1999 Q 2 2000 Q 1 2001 Q 4 2001 Q 3 2002 Q 2 2003 Q 1 2004 Q 4 2004 Q 3 2005 Q 2 2006 Q 1 2007 Q 4 2007 Q 3 2008 Q 2 2009 Q 1 2 01 0 Q 4 2 01 0 Q 3 2011 Q 2 2012 Q 1 2013 Q 4 2013 Q 3 2014
8
Table 2.2.2; Source: CBS; made by the author
From the above table, one can see that the Netherlands had a stable volume of dwellings sold per year since 1995 until the late 2007, when the financial crisis begun. Basic analysis on the data from CBS shows that there is a decline of nearly 45% between 2007 and 2013 on the number of dwellings sold,
recovering afterwards.
2.3 Population and Immigration in Netherlands
According to the “Population counter” of CBS, the current population of Netherlands is 16 920 111 people, which is the 64th highest populated country in the world. Also, Netherlands is the 27th most densely populated country in the world, meaning that all the population of the country is concentrated on an area of 41 526 km^2. Given a data from CBS about the population of Netherlands during the period 1995-2015, table 2.3 shows the trend of population:
Table 2.3; Source: CBS; made by the author. 0 50000 100000 150000 200000 250000
Number of dwellings sold
14500000 15000000 15500000 16000000 16500000 17000000 17500000 Po p u lat ion Period
Population of Netherlands
9
Table 2.3 shows that the population growth of the Netherlands was stable and positive during this period. Basic analysis shows that this growth amounts to 9.15% since 1995 until 2015, or in other words the average growth per year was 0.44%. According to an analysis from Eurostat, the relative population growth of Netherlands during the first decade of the 21st century was higher than in the European Union as whole (CBS Web magazine, 2011). However, the analysis also shows that the population growth of the Netherlands was mainly due to natural growth, not immigration. In no other country in the EU 27, the contribution of migration to population growth was as small as in the Netherlands (CBS Web magazine, 2011). On average, migration accounted 81% of the population growth in the EU 27 versus only 22% in the Netherlands; this amount was 74% in Belgium, and 57% in UK.
According to CBS, immigration refers to people who move to the Netherlands from another country and whose arrival results in entries in a municipal population register. However, the criteria for identification is residency for at least four months in the forthcoming six months. Based on the United Nations report “Trends in International Migrant Stock: The 2013 Revision”, The Netherlands is ranked in 28th place in the world by the number of immigrant population living in the country. The report shows that there were 1 964 922 immigrants in the Netherlands, which amounted to 11.7% of the total population, and 0.9% of the total immigration in the world. The leader in the ranking is U.S. with nearly 46 million immigrants, while Germany, U.K. and Spain are ranked respectively in 3rd, 6th and 10th place in the world, which are the top 3 with most immigration in Europe. Moreover, the Netherlands has the 8th highest immigration in Europe. Table 2.3.1 shows the yearly stock of immigrants from all around the world entering the Netherlands:
Table 2.3.2; Source: CBS; made by the author.
Table 2.3.2 indicates a growing trend of the yearly total immigration for the last two decades. It can be seen that there is an upward trend until 2001, after which there is a decline in the yearly immigration, which is followed by an increase after 2005 until now. The data shows an average of 128 218 immigrants entering the Netherlands per year, which amounts to a 4.16% growth per year. Analysis of the quarterly data shows a significant sign of a seasonality, in which every 3rd quarter of a given year has a higher immigration than the other quarters. Possible reasons for that might be the fact that many people come
0 20000 40000 60000 80000 100000 120000 140000 160000 180000 200000 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 IMM IGR A N TS PERIOD
10
for summer jobs in the Netherlands, or students who usually move there in August, since the educational calendar in the Netherlands starts in 1st of September.
Table 2.3.3 divides immigration into 3 different categories, namely Europe, EU and NON Europe. For simplicity, all the immigration from all around the world except Europe, was combined into the NON Europe group. Also, you may assume that the group EU is part of Europe, meaning that EU plus the immigration from other European countries gives as Europe.
Table 2.3.3; Source: CBS; made by the author.
First of all, it is worth mentioning that immigration from Europe is on average 74 789 people per year, whereas it is 53 430 for Non- European, respectively 58.3% and 41.7% of the total immigration.
However, one can clearly see that this ratio was nearly the same during the period 1995-2003, which is then followed by a substantial increase in the volume of immigrants from Europe. The main reason for this increase is the increasing amount of immigration from EU countries since 2004, which is shown by the blue line. This is not surprising given the fact that in 2004, Europe has seen the so called “largest single expansion of the European Union”, in which 10 countries including Czech Republic, Lithuania, Poland and etc., joined the European Union making the total amount of countries 24. Also, this was followed by another enlargement 3 years later, with the newcomers being Bulgaria and Romania, and the last country which joined the EU was Croatia in 2013. The immigration trend from EU countries can be further seen in the following Table 2.3.4:
0 20000 40000 60000 80000 100000 120000 Imm igran ts Period
Immigration by groups
11
Table 2.3.4; Source: CBS; made by the author.
This chart shows a stable trend over the period 1995-2004, following a substantial growth in the volume, which exactly defends our theory that this increase in immigration was possibly caused due to the enlargement of the EU. Basic analysis of the data shows that on average 36 958 immigrants entered the Netherlands from EU, which accounts to 28.8% of the total immigration. Going slightly more in depth, analysis of two periods shows that, in 1995-2004 the average immigration from EU was 19 507 people, whereas between 2005-2014 the average immigration from EU was 54 409 people, which is clearly a substantial increase on the average.
A press release from CBS in 2006, which compares the data from 2005 and 2006, shows that the increase in immigration during that period was due to Dutch emigrants who return back to the Netherlands, and especially the increasing number of Polish immigrants. The results also show that there are hardly any immigrants from other Eastern European countries that joined the EU in 2004, meaning that most of the increase after 2004 in the above graph is due to the Polish and Dutch immigrants, which is quiet
surprising. Moreover, a press release from CBS in 2012 indicates that 4 in every ten immigrants in the Netherlands came from EU countries. The picture is more astonishing in another press release from CBS in 2014. The results indicate that 12 thousand Polish immigrants registered in the municipality for the first quarter of 2014, whereas this amount was only 1.8 thousand in 2005. Also, Poles account for the largest foreign-born group of immigrants from EU in the Netherlands with more than 100 thousand people as of July 2014. Moreover, besides the immigrants from Poland who are registered in the municipality, there were around 80 thousand Poles who came for a seasonal job and were not required to register for residence in 2014. In total, Poles constitute as the 5th largest immigrant group in the Netherlands after residents born in Turkey, Morocco, Suriname and Indonesia. There were a lot of debates about the countries which joined the EU in 2007, namely Bulgaria and Romania, that they might increase the volume of immigrants significantly. Immigrants from those two countries needed a work permit in order to be able to work in the Netherlands, which is usually not the case for countries in the EU. That rule was canceled out in 2013 and since then, Bulgarians and Romanians can freely come to work in the Netherlands (CBS Press release). Recent results from the press release indicate that, the
0 10000 20000 30000 40000 50000 60000 70000 80000 90000 Imm igran ts Period
Immigration from EU
12
number of Romanian immigrants has doubled to just over 2.3 thousand people compared to the first half of 2013, whereas the number of Bulgarians has not risen.
However, a press release of CBS from 2004, indicates that actually the number of emigrants
outnumbered those of the immigrants in the first six months of 2004, which is also considered as the largest net migration since 1950. This fact is also supported in another press release from CBS in 2006, which shows that emigration outnumbers immigration for a third consecutive year besides the fact that immigration is rising since 2004. In 2007, however, the results change and immigration starts
outnumbering emigration. Moreover, the Netherlands has a small fraction of net migration for almost every year for the last two decades, meaning that the yearly difference between immigration and emigration is not that high. However, emigration is not used in the regression model of this paper. If added to the model, then a single variable (net migration) has to be constructed, which would be the difference between immigration and emigration.
2.4 Gross domestic product (GDP) of Netherlands
Even though the Netherlands is a small country, it has one of the leading economies in the world. Central Bureau of Statistics Netherlands (CBS) defines Gross Domestic Product (GDP) as the quantity that
expresses the size of an economy. It is the main indicator for the development of the economy, and economic growth is measured in terms of the volume change in GDP. The Netherlands has a GDP of 866 354$ million as of 2014, which is ranked in the 17th place in the world by The International Monetary Fund and the World Bank. Also, GDP per capita is defined as the GDP divided by the average volume of the population for a given year. The GDP per capita of Netherlands is 51 373$, or 38 501 euros as of 2014, which is ranked in the 12th place in the world by The International Monetary Fund and the World Bank. The following table shows the evolution of GDP per capita of the Netherlands in the previous decades:
Table 2.4; Source: Organization for Economic Co-operation and Development (OECD); made by the author. 0 5 10 15 20 25 30 35 40 45 in t h o u san d e u ro s Years
GDP per capita
13
However, for the research analysis, this paper uses the quarterly change in GDP in real values, since they account for inflationary impacts. The following chart shows the quarterly change of GDP over 1995-2015:
Table 2.4.1; Source: OECD; the graph was made by the author.
Basic statistical analysis given a data from OECD shows that, the change in quarterly Real GDP was 0.46% for the period 1995-2015. The quarterly average change for the period 1995-2007 is 0.725% respectively, while it is -0.125% for the period 2008-2015.
According to a Web Magazine by CBS in 2011, a research made by Eurostat shows that in 2009 and 2010 the Netherlands had the second highest level of prosperity in the EU with a GDP per capita of more than 33% higher than the average in the EU. The highest GDP per capita was in Luxembourg, whereas the country with the lowest was Bulgaria. Even though in the graph above, the Netherlands was not really shining in its yearly growth of GDP during that period, it still had the second highest GDP per capita in the EU. Further, the analysis by Eurostat indicates that, as a result of the crisis, the relative level of prosperity in Greece, Italy and Spain deteriorated by most. Per capita GDP in Greece was 10% below EU average in 2010, whereas it was 6% below in 2009.
Last, GDP is considered as one of the main determinants of house price movements. According to a Web Magazine Analysis from UK, demand for housing is dependent upon income. With higher economic growth and rising incomes people will be able to spend more on houses, which will increase demand and push up prices, meaning that GDP growth and House prices are positively correlated. Moreover, to ascertain future payments towards maintenance of their house, home buyers form their expectation of future income based on current income (Akbari & Aydede, p.1648).
2.5 Unemployment in the Netherlands
Unemployment is an important factor when it comes to taking into consideration the labor force and economy of the country. According to CBS, the unemployed labor force is expressed as a percentage of
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 Q 1 1995 Q 4 1995 Q 3 1996 Q 2 1997 Q 1 1998 Q 4 1998 Q 3 1999 Q 2 2000 Q 1 2001 Q 4 2001 Q 3 2002 Q 2 2003 Q 1 2004 Q 4 2004 Q 3 2005 Q 2 2006 Q 1 2007 Q 4 2007 Q 3 2008 Q 2 2009 Q 1 2010 Q 4 2010 Q 3 2011 Q 2 2012 Q 1 2013 Q 4 2013 Q 3 2 01 4 %
GDP Quarterly (Real)
14
the entire labor force. The unemployed labor force includes all people without jobs (or working less than 12 hours a week), who are prepared to work 12 hours a week or more, are readily available and are actively looking for jobs. Consequently, the unemployment rate is the unemployed labor force as a percentage of the national labor force. The current unemployment rate in the Netherlands is 6.9% as of June 2015. According to Eurostat (CBS Article, 2015), the recent unemployment rate in all European countries is shown as follows:
This chart ranks the Netherlands in 10th place in all Europe, however, one can see that the average
unemployment rate for the EU28 is 9.7, meaning that the Netherlands has lower unemployment than the average. The following table shows the yearly trend for the unemployment in the Netherlands:
Table 2.5; Source: Federal Reserve of St. Louis; made by the author. 0 1 2 3 4 5 6 7 8 9 % Years
Unemployment rate
15
The average unemployment rate during 1995-2015 was 5.686%. According to CBS, from an international perspective, the Dutch unemployment rate is relatively low; and also, that Germany and Austria have the lowest rates across Europe with approximately 5% as for January 2015. Also, since the reunification of Germany, unemployment has always been lower in the Netherlands than in Germany until 2012. Moreover, the Netherlands had the lowest unemployment rate in Europe in 2007 and 2009, with respectively 3.2% and 2.6%, whereas the average of the EU was 7% and 6.9%. In a historical perspective, the unemployment rate is higher now in comparison with the period before the crisis. Also, since 2004, Dutch unemployment has always been under the European average.
Also, unemployment is considered as another important determinant of house prices movements. It is suggested that unemployment is related to economic growth, in a way that when unemployment is rising, less people will be able to afford a house, thereby the house prices would go down. In that case, one would expect a negative correlation between unemployment and house prices movements. Moreover, Akbari & Aydede (2011) point out that unemployment controls the effects of consumption smoothing and uncertainty at the same time.
3. Data and Methodology 3.1 Data
The data which was used for this research and also the charts done in the previous section, was gathered mainly from the Central Bureau of Statistics Netherlands (CBS). The data for Average House prices, House Price Index and Immigration were gathered from CBS, whereas the data for the GDP was gathered from Organization for Economic Co-operation and Development (OECD). Moreover, the model also includes a variable for the interest rates, for which 10-year Government bond yield of the Netherlands was used, and the data for it together with the data for Unemployment, was gathered from the Federal Reserve of St. Louis. All the data available for the model is quarterly, beginning from 1995 until the first quarter of 2015, thereby making it a total 81 observations for each variable. Since CBS is the main statistics organization of the Netherlands, its data is relevant for a research and is up to date. The same holds true for OECD, as well as for the Federal Reserve of St. Louis.
3.2 Methodology
The literature discussed in this paper uses different approaches to calculate the impact of immigration on house prices. They use advanced panel data analysis which mainly relies on supply and demand models, and are combined with instruments. However, the application of such an advanced method is beyond the scope of the Bachelor thesis. The reason behind the fact that the variables GDP,
Unemployment and Interest rates, were chosen as control variables is that those factors are believed to be the main determinants of house price movements in a given economy as discussed in Section 2. The reason why interest rates are included in the model is that, basically interest rates affect the cost of monthly mortgage payments. A period of high interest rates will increase cost of mortgage payments and will cause lower demand for buying a house, thereby indicating a negative relationship between interest rates and house price movements (Web magazine, 2013). Basically, the model of this paper is trying to predict the impact of immigration on house prices, controlling for these explanatory variables. Moreover, for the research analysis the model is going to use the OLS (Ordinary Least Squares) approach, which according to Stock& Watson (2012) consists of three main assumptions, respectively:
16
2. (X,Y) are independently and identically distributed (i.i.d.) 3. Large outliers are unlikely
The OLS multiple regression model is shown as follows:
Y = β0 +β1X1 +…+ βnXn + u Where Y is the dependent variable, and Xi are the independent variable.
Given the OLS approach, there are two regression models that are tested in this paper of this paper:
∆HPI = α + β*(IMM/POP) + γ*(δRGDP) + θ*UNMPL + λ*IR + ε (1) ∆HPI = α + β*(IMM(EU)/POP) + γ*(δRGDP) + θ*UNMPL + λ*IR + ε (2) ∆HPI = Change in House price index of existing own homes;
IMM = Total Immigration; IMM(EU)= Immigration from EU POP = Population;
δRGDP = Real growth of Gross domestic product; UNMPL = Level of Unemployment;
IR = 10-year Government Bond Yield.
First of all, our model uses the change in House price index (HPI) as the dependent variable. Second, the model uses immigration as a proportion of the total population, not the immigration in nominal values. Also, for immigration, our model refers to those immigrants who enter the Netherlands in a given period (quarter), not the immigrant population (stock) living in the Netherlands. As stated above, immigrants are referred to people who are going to stay at least 4 months in the Netherlands. Moreover, since all the independent variables and the dependent variable are in percentage values, their interpretation is as follows: for example if IMM(EU)/POP increases by 1 percentage point (not percent), then HPI increases by β coefficient points.
All the literature discussed in Section 2.1 uses as their main independent variable: Immigrant
stock/Population, not the amount of immigrants entering a given country, such as the model used in this analysis. Unfortunately, our model cannot use the same approach for the independent variable due to the lack of quarterly data on the immigrant stock living in the Netherlands.
Another problem that may arise in the regression model is the omitted variable bias problem. It occurs when an omitted variable is a determinant of Yi and if it is correlated with at least one of the regressors, then the OLS estimator of at least one of the coefficients will have omitted variable bias (Stock & Watson, 2012).
Another important limitation of the analysis could be the endogeneity problem. Simply, endogeneity means that the model used in the analysis and its estimates do not properly capture the causality relation. Endogeneity arises when any one of the following 3-5 points fail to meet:
17 2. ε causes Y
3. ε does not cause X 4. Y does not cause X
5. Nothing which causes ε also causes X
This means that E{ε|X}≠0, or COV(X, ε)≠0. In order to correct the fact that the causation is wrong, a good way is to use panel data with fixed effects regression combined with instruments. Instrumental variables (IV) approach. This approach is used by all the literature discussed in Section 2.1. However, similar ways of regression are beyond the scope of the thesis.
The research consists of main OLS regression analysis for the two models. As a preparatory step, we remove seasonality from our immigration variable and we test for heteroscedasticity at the end:
1) The above charts for the trend of immigration does not show a sign of seasonality because they represent yearly immigration. However, the quarterly data for immigration exhibits seasonality. Looking in a bit detail, the Netherlands has always higher volume of immigrants in the 3rd Quarter than in the other quarters for every year, which is a sign of seasonality. To fix this problem, a Seasonality and Trend analysis is made. The main idea here is that, the proportion of quarterly immigrants given the population is the dependent variable, while we have 3 dummy variables and one independent variable. The dummies are Quarter 1, 2 and 3 respectively, while the independent variable is a Trend variable, which constitutes for the given period, f. e. T (Year)=1 is Q1 1995, T (Year)=1.25 is Q2 1995, T (Year)=1.5 is Q3 1995 and so on. After the regression analysis is done, basically the residuals (ε) of each period are collected, and then added back to the regression model (2), thereby normalizing the seasonality. The models in this regression are given as follows:
IMM/POP = α + β*Q1 + λ*Q2 + θ*Q3 + η*Trend + ε (3) IMM(EU)/POP = α + β*Q1 + λ*Q2 + θ*Q3 + η*Trend + ε (4)
2) The values calculated in regression (3) are used in the main regression of model (1) as the main independent variable IMM/POP (the same holds true for (4) and (2)). Therefore, the main regression analysis is made based on this data and the other explanatory variables used in the model. The most important factor in this analysis is going to be the “β”, which corresponds to the coefficient value of the variable IMM/POP. The alternative hypothesizes for this purpose are respectively H(0): β=0 and H(1): β≠0. Rejecting the former means that there is sufficient evidence to infer that immigration has an impact on change in house price index, whereas rejecting the latter means that there is not sufficient evidence to infer that immigration has an impact on the change of house price index. A point of interest will be the t-values and the p-value. Basically, a p-value< 0.1 will mean that there is significant statistical evidence to infer that immigration has an impact on the house prices in the Netherlands.
3) The model is tested for heteroscedasticity using the Whites test. According to Stock& Watson (2012, p.198), the error term u is homoscedastic if the variance of the conditional distribution of u given X is “constant” for i=1,…,n and in particular does not depend on X, otherwise the error term is heteroskedastic.
18 4. Analysis and Results
This section analyzes the results of this research. It will begin with some Descriptive Statistics and Correlation coefficients, followed by the illustration and results of the Seasonality and Trend analysis. After that, the results of the main regression illustrated, which will be followed by the White’s test for heteroscedasticity.
4.1 Descriptive statistics and Correlation
Variables Observations Mean Std. Deviation Min Max ∆HPI 81 0.01107 0.01788 -0.03987 0.05299 IMM/POP 81 0.00191 0.00033 0.00116 0.00255 GDP growth 81 0.00461 0.00684 -0.02157 0.0182 Unemployment 81 0.05336 0.01474 0.031 0.084 Interest rates 81 0.04065 0.01438 0.00914 0.07557 IMM(EU)/POP 81 0.00026 0.00016 -0.00009 0.00072
Table. 4.1: Descriptive statistics; Period 1995-2015.
Table 4.1 shows the descriptive statistics of all the variables included in the model. Those results are obtained through analysis of their quarterly data, which means that all of the values can be interpreted as quarterly. For example, the main independent variable of the model IMM/POP has a mean of 0.001905 or 0.1905% per quarter. The same way of interpretation holds true for the other variables.
Table 4.1.1 shows the correlation coefficients between the variables:
Change in HPI IMM(EU)/POP δRGDP Unemployment
Interest rate IMM/POP Change in HPI 1 IMM(EU)/POP -0.058 1 δRGDP 0.585 -0.016 1 Unemployment -0.096 0.318 0.136 1 Interest rate 0.649 0.021 0.358 -0.072 1 IMM/POP 0.200 0.794 -0.022 0.018 0.131 1
Table 4.1.1: Correlation coefficients;
First, the variables δRGDP and interest rate show a moderate positive correlation with the dependent variable. This result was expected for the first variable given its positive relation with the house price movements, whereas for the latter, this result is surprising, since we would expect a negative
correlation. Second, Unemployment has a small, but negative correlation with the dependent variable, which was indeed expected. Third, our main variables for the two models indicate different correlations. The proportion of Immigration from EU given Population shows a small but negative correlation,
whereas the proportion of total immigration given population shows a positive correlation. Moreover, the correlation coefficients between the explanatory variables are small, indicating a small chance for multicollinearity.
19
As pointed above, this analysis was made on one purpose: namely, to normalize the seasonality of the quarterly immigration. The regression results are given as follows:
IMM/POP=0.001365 + 1.47E-05*Q1 – 0.0002*Q2 + 0.00076*Q3 + 4.27E-05*Time + ε (3)
(0.000)*** (0.88) (0.06)* (0.000)*** (0.000)*** R^2=66.37%, SE=3.24E-04
IMM(EU)/POP=-1.28E-05 + 3.23E-05*Q1 – 7.6E-05*Q2 + 2.65E-04*Q3 + 4.81E-05*Time + ε (4)
(0.8) (0.53) (0.145) (0.000)*** (0.000)*** R^2=79.26%, SE=1.63E-04
The results of (3) indicate that on average, the proportion of IMM/POP in the 4th Quarter of each year is 0.001408 (0.001365+4.27E-05) with significance level of 1%; and the proportion is 0.0021677
(0.001365+0.00076+4.27E-05) for the 3rd Quarter of each year also with a significance level of 1%, which is the highest among the other quarters. Also, the second quarter has the lowest proportion among the other the quarters with a significance level of 10%. The result of Q1 is not significant.
Using the same structure, the results of (4) indicate that the 3rd Quarter has proportion of 0.0003131 (highest) with a significance level of 1%, whereas the results for the other Quarters are insignificant. In this analysis, the residual of each period is taken and added back to the given model, which then forms the normalized amount of immigration. The following graph shows both, adjusted and not adjusted trend of immigration:
Table 4.2; Seasonality and Trend analysis; It includes both, Model 1 and 2.
From table 4.2, one can see the consequent deviations of Immigration (Orange line, Yellow line) throughout the years, which are mainly caused during the 3rd Quarter of a given year. Moreover, the normalized values which are shown by the Blue and Grey lines respectively, will be used in the main regression models (1) and (2) as the main independent variable.
The impact of immigration on house prices
-0.0005 0 0.0005 0.001 0.0015 0.002 0.0025 0.003 0.0035 0.004
Seasonality and Trend Analysis
20
First of all, a regression based on the first model is made. To get a better insight on the whole situation, there were 4 regressions made given the first model. The first regression consists of only the main independent variable IMM/POP, whereas in each of the next 3 regression one more independent variable was added to the model.
Regression: (1) (2) (3) (4) Variables IMM/POP 4.645 (0.466) 7.986 (0.124) 8.106 (0.112) 7.61 (0.07)* RGDP 1.564 (0.000)*** 1.63 (0.000)*** 1.15 (0.000)*** UNEMPLOYMENT -0.2197 (0.046)** -0.147 (0.106) INTEREST RATE 0.604 (0.000)*** R^2 0.67% 36.19% 39.42% 59.63%
Table 4.4: regression results of model 1; Dependent variable= ∆HPI; Homoscedastic errors. *P-value<0.1; **p-value<0.05; ***p-value<0.01
The following regression table has the same structure as Table 4.4, the only difference is the main independent variable: Regression (1) (2) (3) (4) Variables IMM(EU)/POP -6.57 (0.604) -5.529 (0.593) 1.04 (0.923) -3.03 (0.734) RGDP 1.526 (0.000)*** 1.59 (0.000)*** 1.108 (0.000)*** UNEMPLOYMENT -0.221 (0.062)* -0.14 (0.173) INTEREST RATE 0.61 (0.000)*** R^2 0.34% 34.46% 37.39% 57.9%
Table 4.5: regression results model 2; Dependent variable=∆HPI; Homoscedastic errors. *P-value<0.1; **p-value<0.05; ***p-value<0.01
According to Stock& Watson (2012, p. 161), the R square and the standard error (SE) of the regression measure how well the OLS regression line fits the data. The R square ranges between 0 and 1 and measures the fraction of the variance of Yi that is explained by Xi. The R squares of both models are similar and increase with the addition of a new variable. However, one may say that adding the variables δRGDP and Interest rate to the model increases substantially the R square. This might be due to the high correlation that they have with the dependent variable. However, note that a high R^2 does not mean that the regressors are true cause of the dependent variable, neither there is not omitted variable bias (Stock & Watson, pp.276).
21
To begin with, the variables δRGDP and Interest rate have significant coefficients at 1% level in any of the cases given the two models; also they have similar coefficients in both models. Moreover, adding Unemployment increases the coefficients of δRGDP, whereas adding the variable Interest rate decreases its coefficient in both model 1 and 2. In that case, the interpretation of δRGDP in regression (4) in model 1 is: a 1 percentage point increase of δRGDP increases the HPI by 1.108 percentage points; the same interpretation holds for model 2. At the same time, a one percentage point increase of interest rates increases the HPI by 0.604 percentage points (the same interpretation holds true for model 2). This result is not relevant with the expectations, as one would expect a negative relation between interest rates and house price movement.
The variable Unemployment is added to the models in their respective 3rd regressions. It shows significant negative values in both models at the 10% level in the 3rd regressions, where the model lacks the variable interest rate. However, adding the interest rates increases the p-value of Unemployment significantly in both models, meaning that the models have insignificant coefficients of Unemployment.
The main independent variable IMM/POP shows a positive trend in model 1. First of all, it can be seen that, addition of variables in regressions (2) and (3) increases the coefficient of IMM/POP, whereas addition of Interest rates (4) decreases its coefficient. Moreover, we have only one significant coefficient of IMM/POP which is in regression (4)(main). The result suggests that, a one percentage point increase of the proportion of total immigrants given the population, would increase the HPI by 7.61 percentage points at 10% significance level. This result is quiet interesting given the fact that its coefficient is significantly higher than the main determinant of house price movements- GDP. As pointed above, this might be due to the several limitations of the model used in this paper.
In model 2, the main independent variable IMM(EU)/POP shows negative estimators in regressions 1,2 and 4, and a positive estimator in regression 3. However, in not a single regression its coefficient is significant. We see the same trend as in model 1, meaning that addition of RGDP and Unemployment increases the coefficient of IMM(EU)/POP, whereas addition of Interest rates reduced its coefficient. Also, the fact that IMM(EU)/POP shows a negative coefficient is a bad sign, since one would expect a positive relation with the house price movements given the analysis in section 2.
Given the results of both models, one can clearly see the inconsistency of the estimates. All these results might have occurred due to the fact that the models exhibit endogeneity and omitted variable bias. Panel data analysis with Supply/demand models, and combining instruments would have given completely different and more consistent estimates.
22 Regression: (1) (2) (3) (4) Variables IMM/POP 4.645 (0.389) 7.986 (0.062)* 8.106 (0.051)* 7.61 (0.033)** RGDP 1.564 (0.000)*** 1.63 (0.000)*** 1.15 (0.000)*** UNEMPLOYMENT -0.2197 (0.039)** -0.147 (0.117) INTEREST RATE 0.604 (0.000)*** R^2 0.67% 36.19% 39.42% 59.63%
Table 4.6; regression model 1; dependent variable ∆HPI Heteroskedastacity errors.
Regression (1) (2) (3) (4) Variables IMM(EU)/POP -6.57 (0.606) -5.53 (0.549) 1.04 (0.915) -3.03 (0.709) RGDP 1.526 (0.000)*** 1.59 (0.000)*** 1.108 (0.000)*** UNEMPLOYMENT -0.221 (0.052)* -0.14 (0.166) INTEREST RATE 0.61 (0.000)*** R^2 0.34% 34.46% 37.39% 57.9%
Table 4.7; regression model 2: dependent variable ∆HPI; Heteroskedasticity errors.
Tables 4.6 and 4.7 show the regression of the same models with Heteroskedastic errors. The coefficients of the variables do not change, neither the R^2 of the regressions. However, the p-value and, thereby the significance of the main variable IMM/POP in Table 4.6 differs from the model with homoscedastic errors. It can be seen that, the coefficients of IMM/POP are significant at 10% level in both regression (2) and (3), and significant at 5% level in the main regression (4). Table 4.7 indicates that there are no significant changes on the p-values of the coefficients when using regression with Heteroscedastic errors.
Last, White’s test for Heteroscedasticty is done. The results are given as follows:
Regressions: (1) (2) (3) (4)
Chi2 Chi2 Chi2 Chi2
MODEL 1 7.63 (0.0221)* 10.81 (0.0553) 16.22 (0.0624) 24.13 (0.0442)* MODEL 2 3.96 (0.1382) 9.7 (0.0841) 15.08 (0.0888) 31.13 (0.0053)*
Table 4.8: White’s test for Heteroscedasticity; *p-value<0.05 => reject homoscedasticity.
According to Table 4.8, regression (2) and (3) of both models have homoscedastic errors (including (1) of model 2), therefore the results of tables 4.4 and 4.5 have to be used for interpreting the coefficients. Moreover, table 4.8 indicates heteroskedasticty for regression (4) of both models, meaning that the results of tables 4.6 and 4.7 have to be used for interpretation of the results.
23
In that case, model 1 predicts the following: A one percentage point increase of IMM/POP would increase the HPI by 7.61 percentage points. However, model 2 does not show a significant impact of the variable IMM(EU)/POP on the house price movements.
5. Conclusion
The developments in the living standard of people in the world which started during the late years of the last decade had increased the mobility of people substantially. It is much easier now to buy a ticket and fly to the other part of the world than it was 20 years ago. This has led to an increase in the immigration stock of the world. Basically, people tend to move to live in countries where the opportunities for living are high, which means that many developed countries are subject to a high immigration. Even though it has small size, the Netherlands has been doing well economically and financially during the last 2 decades. The Netherlands had a moderate amount of stock of immigration until 2004, which was followed by an increasing stock of immigrants due to the immigration from EU countries, mainly Poland. Even though the high immigration rate, the Netherlands had also experience high levels of emigration, making it the country with the least contribution of migration to the growth of population in Europe. At the same time, the house prices in the Netherlands have shown stable yearly increase during the last two decades. This result might be the effect of combination of increasing GDP, the lower interest rates until the beginning of the crisis in 2007, and/or the increase of immigration to the Netherlands from EU. Many scientists analyzed the impact of immigration on house prices in given countries all around the world. All of them have found different results, some positive and some negative. Also, there are a lot of literature analyzing the impact of immigration on labor markets in a given country.
This paper analyzed the impact of immigration on house prices in the Netherlands. For this purpose, a multiple OLS regression has been done. Two regression models were included in the thesis, each consisting of 4 regressions. Model (1) used the variable IMM/POP (The proportion of Total immigrants entering the Netherlands given the Total Population); model (2) used the variable IMM(EU)/POP (The proportion of immigrants from EU countries entering the Netherlands given the Total Population). The purpose of the second model was to analyze the impact of the increasing number of immigrants from EU countries since 2004, since that could be a possible exogeneous event. Moreover, Seasonality and trend analysis was made for both models, since the main variable “immigration” showed a sign of seasonality. The normalized values after this analysis were used as the main independent variable in the models. Finally, White’s test for homoscedasticity has been done. For the research purpose, data was obtained from Central Bureau of Statistics (CBS), Organization for Economic Co-operation and Development (OECD) and Federal Reserve of St. Louis. All the data used in the analysis was quarterly, beginning from 1995 until the first quarter of 2015. The results of this paper for model (1) suggest that, a one percentage point increase of the proportion of immigration given population would increase the House Price Index by 7.61 percentage points. However, the results of model (2) indicate no significant effect of the proportion of immigrants from EU given Population.
However, there can be several limitations and problems related to the estimation results of this paper. First of all, endogeneity is a trouble when it comes to analyzing similar topics. All the literature discussed in section 2.1 uses regression with panel data creating advanced supply/demand models with for their analysis. Moreover, they combine those models with instruments, simply because it is a good way of eliminating the endogeneity problem. However, the usage of similar models is beyond the scope of the Bachelor thesis. In that case, a usual multiple OLS regression was made, using only GDP, Unemployment
24
and Interest rates as control variables. However, this could lead to another limitation, the so called omitted variable bias. Moreover, except the fact that there might be other important variables which are omitted from the model, the emigration factor seems to play a role in determining the house prices. This factor was not added to the model, however, Web magazines from CBS suggest that the Netherlands had a substantial amount of emigration, meaning that the net migration is relatively small to cause an impact. In that case, one may believe that, inclusion of emigration to the model could have substantially decreased the coefficient of the main independent variable in our model. Given this fact, I would suggest for further research taking into consideration the impacts of emigration. Last, the literature discussed in section 2 uses the amount of immigrant stock living in a given country, not the amount of immigrants entering a given country in a given period, such as the one used in the model of this paper. This also possibly had an impact on the coefficient of the main independent variable in this paper.
6. Reference list
Akbari, A.H. & Aydede, Y. (2012). ‘Effects of immigration on house prices in Canada’, Applied
Economics, vol. 44 (13).
Dutch house prices fall by more than EU average. (2013, 7 March). Retrieved from: http://www.cbs.nl/en-GB/menu/themas/dossiers/eu/publicaties/archief/2013/2013-01-housepriceindex-art.htm
Dutch unemployment rate lowest in EU. (2007, 17 July). Retrieved from: http://www.cbs.nl/en-GB/menu/themas/arbeid-sociale-zekerheid/publicaties/artikelen/archief/2007/2007-2245-wm1.htm Emigration slows down the population growth. (2006, 10 November). Retrieved from: http://www.cbs.nl/en-GB/menu/themas/bevolking/publicaties/artikelen/archief/2006/2006-104-pb.htm
Gonzalez, L. & Ortega, F. (2013). ‘Immigration and housing booms: Evidence from Spain’, Journal
of Regional Science, vol. 53 (1), pp. 37-59.
Immigration from Eastern Europe remains high. (2012, 12 June). Retrieved from:
http://www.cbs.nl/en-GB/menu/themas/dossiers/eu/publicaties/archief/2012/2012-3589-wm.htm
Immigration increasing again. (2006, 10 May). Retrieved from:
http://www.cbs.nl/en-GB/menu/themas/bevolking/publicaties/artikelen/archief/2006/2006-053-pb.htm
Immigration rising. (2014, 11 August). Retrieved from:
http://www.cbs.nl/en-GB/menu/themas/bevolking/publicaties/artikelen/archief/2014/2014-047-pb.htm
Netherlands unemployment rate. (n. d.). Retrieved from:
http://www.tradingeconomics.com/netherlands/unemployment-rate
Population growth accelerates. (2007, August 10). Retrieved from: http://www.cbs.nl/en-GB/menu/themas/dossiers/vergrijzing/publicaties/artikelen/archief/2007/2007-057-pb.htm
25
Population growth in 2014 nearly 73 thousand: more immigrants and more babies. (2015, 5
February. Retrieved from:
http://www.cbs.nl/en-
GB/menu/themas/bevolking/publicaties/artikelen/archief/2015/population-growth-in-2014-nearly-73-thousand-more-immigrants-and-more-babieshtm.htm
Sa, F. (2014). ’Immigration and house prices in the UK’, The Economic Journal.
Saiz, A. (2003). ‘Room in the kitchen for the melting pot: Immigration and rental prices’, Review
of Economics and Statistics, vol. 85 (3), pp. 502-521.
Saiz, A. (2007). ‘Immigration and housing rents in American cities’, Journal of Urban Economics,
vol. 61 (2), pp. 345-371.
Statistics Netherlands: Employment and unemployment up. (2015, 26 February). Retrieved from:
http://www.cbs.nl/en-GB/menu/themas/arbeid-sociale-zekerheid/publicaties/artikelen/archief/2015/meer-werklozen-en-meer-werkenden.htm
Stock, J.H. and Watson, M.W. Introduction to Econometrics. Pearson Education.
Unemployment differ strongly within Europe. (2012, 27 September).
Retrieved from: http://www.cbs.nl/en-GB/menu/themas/dossiers/eu/publicaties/archief/2012/2012-jeugdwerkloosheid-europees-art.htm
Unemployment in the Netherlands higher than in Germany. (2013, 7 May). Retrieved from:
http://www.cbs.nl/en-GB/menu/themas/arbeid-sociale-zekerheid/publicaties/artikelen/archief/2013/2013-3824-wm.htm
Why the average dwelling purchase price is not an indicator. (2008, 27 May). Retrieved from:
