The Externality Effect of Mosques on House Prices
A study of anti-Islamic and populistic right voting behaviour29-06-2018 Master Thesis
Msc Business Economics: Dualtrack Corporate Finance & Real Estate Finance Supervisor: Dr. M.I. Dröes
Second Supervisor: Dr. F.P.W. Schilder Author: Victor Meertens, V.T.R. Student No. : 11920467
2
Abstract
Mosques become a more integral part of the Dutch landscape, representing the Islamic community in the Netherlands. This is a much debated subject in politics, especially so since the times of rising right populism in Europe. To shed light on the perception towards, and effect of, mosques in neighbourhoods, this research examines the externality effect of mosques on surrounding house prices. Using difference-in-difference models on a dataset concerning mosques and house transactions amongst 16 municipalities in the Netherlands between 2001-2016, this research discovered a decrease in prices of 2.62% for houses within a range of 500 meters around the mosque, on average. Findings indicate an anticipation effect starting at the announcement of the arrival of the mosque and worsening when the construction period starts. An adaption effect is excluded, indicating a permanent effect for the first 10 years after opening. Furthermore, above average size mosques and mosques in rural and sub-urban areas reveal negative externality effects of the highest magnitude. By contrast, smaller, urban located, and traditional looking mosques in forms of a minaret or chapel reveal even positive externality effects. Therefore, zoning policy is advised whereby rather several smaller mosques are build, instead of one large mosque. Since the appearance of mosques could have captured other omitted variables, no distinction by cause in nuisance or perception could be made. Furthermore, results on voting behaviour show that municipalities primarily voting anti-Islamic and populistic radical right, are associated with the largest decrease in house prices with the arrival of mosques. This is followed by parties of Christian nature, whilst left-winged parties show the opposite with even positive effects on house prices. The relevance of research on this subject is emphasised by the negative trend demonstrated over the recent 16 years. Whereby the externality effect of mosques on house prices was still positive in 2001-2004, however rapidly decreased to negative figures in the more recent years. Where externality effects were on average -9.48% in 2013-2016.
Statement of Originality
This document is written by Student Victor Meertens, 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 the supervision of completion of the work, not for the contents.
3
Content
1. INTRODUCTION... 4
2. LITERATURE REVIEW ... 8
2.1HEDONIC PRICING STUDIES ... 8
2.2DIFFERENCE-IN-DIFFERENCE STUDIES ... 13
2.2.1 Simple and more complex methods ... 13
2.2.2 Robustness of Difference-in-Difference methods ... 14
2.2.3 NIMBY and LULU ... 15
2.3PREVIOUS RESEARCH ON HOUSES OF WORSHIP ... 17
3. ISLAM IN THE NETHERLANDS ... 21
3.1HISTORY AND ISLAMIZATION ... 21
3.2THE RISE OF POPULISTIC PARTIES ... 24
4. HYPOTHESES ... 26
5. DATA... 29
5.1GATHERING OF DATA ... 29
5.1.1 Dataset on Mosques ... 29
5.1.2 Dataset on Housing Transactions ... 31
5.1.3 Dataset on Voting Behaviour ... 31
5.2DESCRIPTIVE STATISTICS ... 32
5.3DEPENDENT AND INDEPENDENT VARIABLES ... 38
5.3.1 Dataset on Mosques ... 38
5.3.2 Dataset on Housing Transactions ... 40
5.3.3 Dataset on Voting Behaviour ... 42
6. METHODOLOGY ... 43
6.1AVERAGE TREATMENT EFFECT ... 43
6.2GEOGRAPHICAL RESPONSE PROFILE ... 45
6.3DYNAMIC RESPONSE PROFILE ... 45
6.4HETEROGENEITY AND ROBUSTNESS ... 46
7. RESULTS ... 47
7.1AVERAGE TREATMENT EFFECT ... 47
7.2GEOGRAPHICAL RESPONSE PROFILE ... 50
7.3DYNAMIC RESPONSE PROFILE ... 53
7.4HETEROGENEITY AND ROBUSTNESS ... 56
7.4.1 Heterogeneity on Mosque Characteristics... 57
7.4.2 Heterogeneity on Voting behaviour ... 59
7.4.3 Heterogeneity per Municipality ... 61
7.4.4 Heterogeneity per Timeframe ... 64
7.4.5 Robustness without Control Group ... 65
8. LIMITATIONS AND FURTHER RESEARCH ... 69
9. CONCLUSION ... 72
REFERENCES ... 77
4
1. Introduction
“Growing resistance against new mosques” (Nieuwsuur, 2015) and “Neighbourhood massively ‘sells’ house out of protest mosque” (Metro, 2016) were headliners in the media. However, also “A new and very big mosque in North – and barely protest” (NRC, 2017) was published about the arrival of one of the biggest mosques in the Netherlands. In many regions there has been protests about the arrival of mosques by cause of local residents being afraid of parking overload, disruption of the quietness of the neighbourhood and a possible decline of the value of their homes. However, the announcement of a new mosque in Amsterdam-North did not come with protests. Suggesting that the points of view towards mosques differs by time and between regions in the Netherlands.
Research on mosques is of great importance since it represents the Islamic society in the Netherlands, of whom the segregation and policy is a popular and present subject in the political landscape over the recent years. Moreover, mosques function as public facilities. In the past, much research has been conducted on the spatial externality effect of public facilities such as hospitals, public transport and schools. However, only little research has been done on the presence of houses of worship, and specifically mosques in the Western countries. Subsequently, the literature that has been conducted on the consequences of mosques in neighbourhoods was not conclusive. Brandt, Maennig and Richter (2014), stated that mosques in Berlin have positive externality effects on house prices up to 1,000 meters. However, Leijten (2016) stated the opposite, a negative externality effect of mosques on house prices in Amsterdam. Although these are the only researches done on the subject of externality effects of mosques, resembling research on other houses of worship were neither conclusive. Whereby researchers found both negative (Do, Wilbur, & Short, 1994; Babawale & Adewunmi, 2011), as positive (Carroll, Clauretie & Jensen, 1996; Brandt et al.,2014) externality effects of churches on house prices. Since there is evidence of a gap of knowledge in the literature on the effect of houses of worship and particular of mosques, the research question of this thesis will be as follows:
What is the externality effect of mosques on surrounding house prices?
Furthermore, previous literature on externality effects on the housing market suggest that externalities, besides the price, can also be translated in change in asset liquidity, residential mobility or bargaining power (Nelson, 1982; Jud, Winkler, & Kissling, 1995; Harding, Rosenthal, & sirmans, 2003). In this research, these factors are examined by measuring the externality effect of mosques on the time on the market, as an analogy for the asset liquidity and residential mobility, and the mark-up which is the difference between the ask and selling price, also as an analogy for the asset liquidity and moreover the bargaining power.
5 It is clear that the stigma in the media about mosques, and moreover the Islam, has worsened in the Netherlands in the last few decades. This is due to countless factors. For instance, the diplomatic conflict with Turkey and speculations of the influence of the Middle East in the Netherlands through mosques and Imams. Or the stigmatization of the Islam due to the rise of Islamic State and the recent terrorist attacks that took place in Europe. This stigmatization might have caused residents to anticipate on the arrival of a mosque in their neighbourhood. Therefore, it is of importance to examine the possible impact of anti-Islamic and populistic right voting behaviour on the externality effects of a mosque and to, if necessary, control for it on a municipality level.
Since previous literature has been conducted on specific segments of the housing market and each covers only one particular metropolitan area, the results lack external validity. This research includes data of 46 mosques across 16 municipalities in the Netherlands. The data on mosques is subsequently distinguished on physical and visual characteristics to discover heterogeneity in the externality effect of mosques. From the same municipalities multiple characteristics of the housing transactions over the period 2001-2016 are amassed from the database of the Dutch Association of Realtors. Lastly, data on the local share of votes per political party for every national election between 2001 and 2017 is collected. The timeframe chosen for this research starts at the change of the political landscape in these recent years, when populistic parties started to emerge, such as; Lijst Pim Fortuyn (LPF) in 2002 and the Party For Freedom (PVV) in 2006.
Using the Difference-In-Difference (DID) and the hedonic pricing model, statistical analysis is conducted to examine the research questions. The DID model uses a treatment group, houses that lie within a certain distance of a mosque, and compares it to a control group that lies out of this range to distinguish the effect from other time variant and time in-variant trends. The hedonic pricing model controls for the house-specific characteristics. In this research the reach of the externality effect of mosques on house prices, thereby the optimal range of the treatment group and the control group are discovered using geographical response profiles and regressions with several bandwidths. As the descriptive statistics indicate, mosques are not randomly allocated, but namely located in neighbourhoods with smaller and lower priced houses. To take these and all other unobserved time invariant differences between treatment- and control group into account, location fixed effects are controlled for on a 6-digit postal code level. Moreover, the study also controls for yearly time fixed effects to address unobserved time variant differences that are linked to the specific locations of mosques and houses. Lastly, since the DID model is based on the assumption of a parallel trend between treatment- and control group in absence of the treatment and the descriptive statistics indicate a selection bias, more statistical analyses disregarding the control group are conducted. Using
6 only the timing of the mosques’ opening years in a normal hedonic pricing model, robustness of the conventional results is tested.
The results of the statistical analyses show a negative externality effect of mosques on house prices in the range of 500 meters of -2.62% on average. Even though most municipalities show negative effects, some municipalities show the opposite. The results on the impact of mosques on the time on the market and the mark-up do not reveal figures significantly different from zero. This substantiates that the externality effect of mosques is fully, and only, incorporated in house prices and not in asset liquidity, residential mobility or bargaining power. The geographical response profiles show that the effect reaches up to 500 meters, therefore mosques only appear to have local impacts on house prices. Moreover, the dynamic response profile demonstrates an anticipation effect in house prices starting five years prior to the opening of the mosques, correspondent with the average announcement date of new mosques, and significantly reaches a decrease of 1.0% in house prices two years prior to the opening of the mosque, correspondent with the start of the construction period. The first five years after the opening of the mosques the price-drop stabilizes around 4.0%. Due to insignificance, the DID model cannot conclude whether an adaption effect, which would potentially cancel out the negative externality effect over time, exists. The robustness check using only the hedonic pricing model significantly concludes that the negative effect is permanent over the first 10 years after the opening.
Regarding the dependency of the externality effect on the anti-Islamic and populistic voting behaviour the results show that municipalities that belong to the 75th centile of the local share of votes for the
PVV experience negative effects of mosques on house prices of the highest magnitude (2.48%). Subsequently, the results indicate that even before the existence of the PVV, the other populistic party LPF also shows a negative effect of 5.62%. Moreover, other parties like the Christian parties show a more modest negative effect between 1.83% and 2.05% and the more left-wing parties reveal even positive externality effects.
This research distinguishes itself from others by also examining the potential causes of the found effects. The results on the heterogeneity of the externality effect indicate that the traditional appearance of mosques in forms of a minaret or chapel is correlated with a positive externality effect, suggesting that the perception towards Islam only has a minimal contribution to the negative impact on house prices. However, omitted variables such as the difference between newly build and existing mosques or the tolerance of the neighbourhood against the Islam and therefore the likelihood of the presence of a minaret or chapel in the construction permit, could have biased these results. Furthermore, mosques above average in size revealed severe negative externality effects of 11.8%, against even positive effects of 2.50% for smaller mosques. However, no conclusive interpretation can
7 be given, whether the negative effect found is caused by nuisance or perception. Analyses on the level of urbanisation show that mosques in urban areas have positive externality effects of 4.6%, in contrast to -10.5% in sub-urban and rural areas. This might be explained by the more open-minded viewpoints and multicultural composition of urban areas, as appointed by Steenbekkers, Simon and Veldheer (2006). Moreover, the results indicate a negative trend in the effects over time, stating positive externality effects between 2001-2004 turning into negative effects of 9.48% on average in the recent years of 2013-2016. The robustness tests by using only the mosques’ opening years and putting aside the parallel trend, indicate that the conventional results mentioned above are conservative measures of the true estimate.
The results of this thesis demonstrate the importance of research on this subject. Where trends show that the externality effect of mosques has worsened over the years and is worse off in rural areas or regions where the local share of votes is primarily of populistic right and Christian nature. This research does not indicate that policy makers should minimise the welfare loss by placing mosques in only urban or politically left-winged areas. However, it does provide tools to discover on which level mosques generate negative effects and how to overcome them. A solutions based on the results is to construct multiple smaller mosques than rather a large one. Further research should lead to an answer on whether the policy should focus on minimizing the nuisance or the negative perception towards the Islam, as possible causes for the negative externality effect
This research is structured as follows. Firstly, in Section 2 the literature on other externality studies and previous studies performed regarding, houses of worship and the housing market, is reviewed. In this part the relevance of this research, its position in the literature and the motive for the right methodology will be explained. Secondly, in Section 3, the history of the Islam in the Netherlands and the rise of populistic right parties will be addressed. Thirdly, Section 4 formulates the hypotheses. Fourthly, in Section 5 will the gathering of the data from different databases be put apart, including explanations about the multiple variables and their descriptive statistics. Fifthly, the right methodology to test the hypotheses is described in Section 6. Sixthly, both the main results as the robustness test results are analysed in Section 7. Seventhly, the conclusion, implications and ethics of this research are covered in Section 8. Lastly, Section 9 concludes with the limitation of this research and suggestions for further research.
8
2. Literature Review
The externality effect of mosques on house prices is central for this research. Therefore it is of importance to look at the methodologies and limitations that are enlightened in previous literature. To start, this section reviews the literature on the emerge of hedonic price models for measuring externality effects and elaborates further on studies relevant for this research that used hedonic pricing models. Next will the studies on simple and more complex DID methods be put apart and will their limitations be discussed. Additionally, the phenomenon’s Not-In-My-Back-Yard (NIMBY) and Locally-Unwanted-Land-Use (LULU) will be highlighted to give more context to the possible negative externality effects of mosques. At last, the previous literature on the specific topic of externality effects of houses of worship on house prices will be discussed, in the hope to connect findings, methodologies and their limitations to those of more general studies on externality effects, to come to the best methodology to use for this research.
2.1 Hedonic Pricing Studies
Already in the beginning of the 70’s, research has been done on the measurement of externality effects using hedonic price models. In the research House Prices and the Measurement of Externalities, Wilkinson (1973) tries to investigate the environmental characteristics that could attribute to the structure of the house prices. Wilkinson highlights the lack of objective methods to measure these effects. He points out that older statistical models only show correlations between house prices and environmental characteristics, but due to high correlation between the two, these results lack significance. In this case the models do not show the rationale behind the results and therefore make it impossible to generalise. Contributions to structures of house prices could differ between social classes and other geographic and demographic variables. Important in his research is the separation of housing characteristics into dwelling- and location specific variables. Wilkinson used 1255 observations between 1960-1964 from Leeds (U.S.), which would today point out to be a rather small sample. The conclusion drawn in the paper, is the lack of significance for simple hedonic pricing models, due to omitted variables and high correlation. This paper was, with its 17 variables used for dwelling- and location specifications, a precursor for follow-up research (Wilkinson, 1973).
Another important research in the evolution of measurement of externality effects and hedonic price models is the one performed by Li and Brown (1980). They point out that previous research using hedonic price models does not, or use only few variables concerning geographic-specific characteristics. In their research a division into three types of contributions to the structure of house prices has been made; proximity, pollution and aesthetics. In previous research the proximity was
9 based on the bid-rent curve that is used to explain prices in urban areas, where distance to the centre is the key variable. However, the methodology used by Li and Brown explains that accessibility also depends on more local factors like proximity to non-residential use of land. Such land use comes with positive externalities from proximity but also negative externalities due to noise and air pollution, traffic congestion and deterioration of sight. Li and Brown were the first to not only look at the positive externalities of non-residential use of land such as grocery stores, schools and open parks, but also examined the negative micro-level externalities. This resembles with the research performed in this thesis, whether the externalities of public facilities - such as a mosques- are existence in de structure of house prices. The parameter ‘distance’ to a certain non-residential use of land is an important variable used in the regressions of Li and Brown, which is also of importance for showing a causal relationship between mosques and house prices in this thesis. Furthermore, Li and Brown’s findings on non-residential land use differs between specific uses. They found that deterioration of sight indeed attributes to lower house prices, same as for noise- and air pollution, and distance to industry- and commercial areas. However, on the contrary, proximity to schools did not show a significant attribute to the structure of house prices, indicating that the externality effect could differ between various forms of public facilities (Li & Brown, 1980).
Following on both studies of Wilkinson (1973) and Li and Brown (1980), the literature on studies regarding externality effects evolved. Poon (1978) studied the externalities of railways using a hedonic price model. Notable is the fact that Poon (1978) ,in correspondence to the research of Li and Brown (1980), used characteristics of externalities such as air- and noise pollution and deterioration of sight. In addition, he describes the negative effects on the human health and the physical conditions of their homes . The value of the social welfare loss is not directly measurable and therefore estimated by examining the house prices surrounding the railways, assuming that the negatives effects are - partly - capitalised into the house prices (Poon, 1978). Data is used from the municipality London (Canada), consistent of 285 observations among which 85% single-family dwellings and only 15% multi-family dwellings, not randomly chosen. This could immediately be called the limitation of this research, since single-family dwellings are overrepresented and the sample is not randomly assigned. Important is the fact that Poon, besides using physical and geographical characteristics for its hedonic price model, also uses variables such as; financial terms and mortgages, tax characteristics, ask and actual sales price and date of sales. A side note, this research concerns only mosques in the Netherlands and since tax and financial terms are almost constant for the whole sample, these are neglected. This corresponds with the outcome of the research of Poon (1978). To conclude, the outcomes of the research of Poon show a discount for house prices, implicating that this method could be a good measurement to estimate social and economic costs. Poon emphasises the importance of the results for allocation of
10 these public facilities for city planners, which corresponds with the relevance of this thesis, seen in a more present perspective.
Further in time, multiple studies are performed on the externalities of transportation facilities due to air- and noise pollution. One of these, is the study performed by Nelson (1979) on noise pollution of six airports in the U.S. In his research he reflects on the different methods used in the literature to measure the externality effect and exposes that multicollinearity- and specification biases are recurring issues in these studies. He therefore quotes the article by De Vany (1979), which offers a solution to this problem. De Vany not only adds distance as a variable to his regressions, he also divides them into dummies for every distance “ring” from the specific airport. This way he eliminates the rent gradient and found a way to measure the externality effect. This method for overcoming the problem of multicollinearity and specification bias used in the study of De Vany, is the inspiration for the methodology used for this thesis. In Section 6.2, more will be elaborated on the distance “ring” dummies that will be used in the regressions.
Another study on the externalities of infrastructure has been performed by Hughes and Sirmans (1992).The main goal of their paper was to measure the externality effect of traffic density in neighbourhoods. Their results indicate that an actual negative externality effect of high traffic density on house prices is indeed noticeable, indicating a discount on surrounding house prices of respectively 9,0% and 4,9% for urban and suburban locations. Furthermore, two lessons can be learned from their research. Firstly, Hughes and Sirmans revealed that not only on macro- but also on micro (street) level externality effects are capitalised into house prices, indicating that housing markets are efficient enough to do so. Secondly, the externality effect differs between urban and rural locations (Hughes & Sirmans, 1992). Both should be considered for this thesis, since it shows that the externality effect of mosques on house prices could indeed be significant at micro (street) level and that this effect could differ between different regions and locations. Moreover, this thesis will try to explore the reason behind this spatial difference.
The next paper that will be discussed, is again from the author Nelson, however published in 1982. In this paper, he reviews nine researches done on the externality effect of highway noise on surrounding house prices. What stands out in his research, is the examination of the assumptions made by the different papers. He states that one of the most important assumption is the one of freedom of movement. Most of his reviewed articles assume that residents are free to move between houses and regions, although this is not always true. Nelson formulates that moving comes with physical cost such as relocation of belongings, and transaction- and bidding costs such as; real estate agents, advertising and notarial costs (Nelson, 1982). Nelson highlights the value of taking the residential mobility of the
11 region into account, since omittance could result in underestimated externality effects. Nelson suggests the time on the market as indicating variable for the residential mobility.
However, the literature also shows other interpretations of the time on the market. In the study performed by Jud et al. (1995) the time on the market is examined as measurement for asset liquidity. Whereby they define liquidity as the time needed to convert an asset in to cash (Munn, Garcia, & Woelfel, 1991). In contrast to the money markets, where the time on the market is close to zero, is the housing market highly illiquid and characterized by time on the markets of weeks or months. Studies have shown that housing characteristics and neighbourhood- and market conditions affect the time on the market (Haurin, 1988; Kluger & Miller, 1990; Asabere, Huffman, & Mehdian, 1993; Kalra & Chan, 1994) and is therefore a good reason to examine in this research.
Subsequently, Munn et al. (1991) have measured the liquidity of the housing market by the spread between the ask- and selling price, defined as the mark-up. This can be positive, in case of overbidding in overheated housing markets, or negative in less attractive markets. In addition, Harding et al. (2003) connect the mark-up to the bargaining power of the seller and buyer. They make the assumption that competition, reflecting the willingness-to-pay by the buyers, drives the bargaining power for the seller, and vice versa. Subsequently, they state that increasing heterogeneity in house and neighbourhood characteristics causes the spread to increase. In case of this research could the arrival of a mosque near a house create more heterogeneity, assuming the spread to increase. This factor is difficult to valuate and therefore will the bargaining power from both the seller as buyer have a bigger contribution to the selling prices. Both the mark-up as the time on the market will be taken into account in the statistical analyses of this thesis, as put apart in Section 5.3.2.
To look further into the literature on externality effects, a subject more in line with mosques is further discussed. Both Clotfelter (1975) and Jud (1985) performed studies on the externality effect of public schools on house prices and tried to link these effects to the racial mix of the schools. Schools show resemblances with houses of worship since the society agrees on the need of such facilities, however would people rather not have them near their homes due to the negative externality effect. This phenomenon is often referred as “Not-In-My-Backyard” (see section 2.2.3). Both Clotfelter (1975) and Jud (1985) link independent variables such as the racial mix to their methodology, to indicate whether the externality effect of public schools is dependent on the people attracted into the neighbourhood. Both studies reflect data samples from the 70’, in times where Afro-American people where looked down upon, which could be resembled on some level with the Islam community in the Netherlands, who are seen as inferior by extreme right populistic groups nowadays. Although they both found a negative relationship with house prices, a very important note amplified in their studies; not the racial
12 composition of students is causing a drop in surrounding house prices, the perception towards other minorities is feeding the negative externality effect. Important for this research on the externality effect of mosques is the inclusion of people’s perception. Rather than to assume it is solely caused by traffic- and parking density, noise pollution or cultural differences. The studies of Clotfelter (1975) and Jud (1985) show that cultural differences in combination with unfounded prejudices towards minorities should not be underestimated and can indeed have effects on the economic and social welfare of a country. This indicates that policy makers and governments should prioritise desegregation, which amplifies the relevance of the outcome of this thesis.
Another view on the contributors to the structure of house prices comes from Luttik (2000), who examined the externality effect of environmental factors. In his research the terms from the paper of Wilkinson (1973), such as aesthetics and quality of sights, come forward. Luttik has tried to measure the contribution to house prices by the presents of trees, water and open space. His results show that quality of sight creates a premium for house prices. However, on the one hand, correction for the effect of surrounding nature will be redundant for this thesis, since these factors are assumed to stay almost constant over time and therefore already taken into account in the difference-and-difference regression used in this thesis. On the other hand, the paper of Luttik shows that quality of sights is an important factor for the structure of house prices. In this thesis the “quality of sight” can also be interpreted as the actual visibility of a mosque in the neighbourhood. Therefore in this thesis, both physical and visible characteristics of mosques will be taken into account in the statistical analyses. In Section 5.3.1, more is explained regarding these variables.
To summarize the findings from the literature review on hedonic pricing models used to estimate externality effects, the following can be concluded: Poon (1978) shows that railways have significant negative externality effects but more important emphasises on the relevance of externality studies for policy makers. Next, Nelson (1979) and De Vany (1979) introduce the distance dummy variable that will be used for this thesis, to overcome the problem of multicollinearity- and specification bias. Hughes and Sirmans (1992) showed that externality effects can be measured on micro (street) level due to efficient housing markets. Furthermore, they highlight that these effects can differ between regions, which will be examined in this thesis. The next paper of Nelson (1982) introduces the importance of residential mobility that can be measured by the time on the market. Others complement that time on the market could be an measurement of the asset liquidity, also measured by the up (Jud et al., 1995) . Subsequently, Harding et al. (2003) emphasise measuring the mark-up as analogy for the bargaining power of the seller and buyer. In addition, Clotfelter (1975) and Jud (1985) point out that the perception towards minorities in a country can be of importance for the structure of house prices, which is something different than an actual causal effect of the minority
13 population. It highlights the importance of this thesis by stating that segregation could be measured by externality studies and desegregation should be prioritised in governmental policies. At last, the paper of Luttik contributes to this thesis by amplifying on the fact that physical and visible characteristics significantly contribute to the structure of house prices. All findings mentioned above are processed in this thesis.
2.2 Difference-in-Difference Studies
In contrast to the hedonic models, in this section more is elaborated on the evolution of difference-in-difference models to measure externality effects. First the history of difference-in-difference-in-difference-in-difference studies, and thereby the inspiration for the methodology used for this thesis, is put apart. Second, the limitations and liabilities of this method are discussed and there will be illustrated how to counteract them. At last, the two most important terms that involve externality studies are pointed out, giving more context to the relevance and implication of this thesis.
2.2.1 Simple and more complex methods
The literature on difference-in-difference models used for research already started in 1854 by Snow. In this research he has tried to examine whether the illness Cholera was transferred through air or water. His sample consisted of two neighbourhoods close to each other, to be sure that most variables, among which the air quality, were the same. Since both neighbourhoods had different water suppliers, because one neighbourhood switched water supplier through time, and only one of them was contaminated with Cholera, he could measure the effect of contaminated water supply (Snow, 1854). By comparing the difference in death rates over time between the two neighbourhoods, before and after the switch of supplier, Snow could trace the effect of contaminated water. The results illustrated the effects of bad water quality and had major implication on public health policies.
Later in time, simple DID models were used to measure effects of regulations and nation laws, such as minimum wages (Obenauer & Von der Nienburg, 1915) (Lester, 1946) (Card & Krueger, 1994), working environments (Rose, 1952) (Ashenfelter, 1978) and immigration on labour markets (Card, 1990). These early difference-in-difference models highlighted that this method did not need highly sophisticated computers to succeed and was easy to interpreted. However, this convenience was simultaneously also the weakness of the method, which will be further discussed in the next section.
Since the turn of century, the methodology of difference-in-difference methods evolved. As of today, it is called one of the most important methods used, partly by how it tackles problems such as omitted
14 variables and non-random sampling. The most important inspiration for the methodology used in this research on the externality effect of mosques is the recent study of Dröes and Koster (2016). In their research they tried to find the externality effect of wind turbines on surrounding house prices. The first dataset contained location - and several other - characteristics of 1,898 wind turbines that were placed between 1982-2012. The second dataset contained prices and characteristics of two million housing transactions between 1985-2011. Covering almost 70% of the transactions that took place in that period. In their research they found that the arrival of wind turbines caused house prices, in a range of 2 kilometres, to drop with 1,4%. Their methodology fits this thesis best for different reasons. First of all, they combine the simple difference-in-difference method with the hedonic pricing model. By controlling for housing characteristics, the average treatment effect cannot be biased by differences in housing characteristics between the control- and treatment group. Secondly, they discovered that the externality effect was heterogeneous in various ways. They included area- and appearance characteristics and found that the externality effect was greater for taller wind turbines and wind turbines located in more urban areas. Therefore in this thesis the appearance of the mosques and the level of urbanisation of the areas are taken into account. To further examine the heterogeneity in the externality effect of mosques, anti-Islamic and populistic right voting behaviour is taken into account as an explanatory reason for the differences between regions. Thirdly, they used different distances between houses and wind turbines to estimate the best control- and treatment group, to come to the most significant results. Their results illustrate that control groups chosen to close to wind turbines, still capture some of the externality effect, underestimating the average treatment effect. In this thesis the right distance set for the control- and treatment group is similarly tested by taking several ring distances, looking at the most significant results. At last, the authors included statistical analyses to measure the anticipation and adaption effect, since this reveals robustness of the results and discovers whether the externality effect holds over time. In Section 6.3 more will be elaborated upon these effects and how they are tested in this research.
2.2.2 Robustness of Difference-in-Difference methods
As stated above, the simplicity of the difference-in-difference methods comes with weaknesses. An important paper in the literature on the robustness of difference-in-difference methods is the one of Bertrand, Duflo and Mullainathan (2004). In their research they try to highlight that results in the literature are too often seen as significant and causal. In their paper; How much should we trust Difference-in-Difference estimates? (2004), they use fictional laws, placebo laws as they call them, to show that even events that don’t exist, can give significant results using the difference-in-difference methodology. The results indeed illustrate that placebo laws at state level show significant effects.
15 Several explanations and solutions to the problem are stated in their paper. Primarily, Monte Carlo simulations were conducted for the different methods used to correct for the problem. The most common problem stated, is the use of difference-in-difference methods for too small samples. In case of too small sampling, the serial autocorrelation understates the standard deviation, causing a to high t-statistics and wrong significance levels. Among others, such as splitting times series into “pre” and “post” periods, their main solutions is to correct for serial autocorrelation in the regressions. This is incorporated in the methodology of this research by clustering the standard errors.
Another important paper for revealing weaknesses of the difference-in-difference method is the one from Abadie (2005). In this research, the author questions whether the assumptions of the Difference-in-Difference method holds. Among which; would the treatment- and control group indeed have followed the parallel trend if the treatment effect was neglected? He highlights that this cannot be the case if characteristics of the treatment group are correlated with the outcome of the treatment effect. For instance, in the framework of this thesis, mosques could be placed in neighbourhoods where house prices were already lower. Their solution to the problem is to correct for location fixed effects, since this will tackle the differences between the control and treatment group before, during and after the treatment.
Two lessons can learned from the paper Bertrand, Duflo and Mullainathan (2004) and Abadie (2005), which are also incorporated into the research of Dröes and Koster (2016). First, caution is needed when interpreting “significant” results and biased results in case of smaller samples can be prevented by including serial autocorrelation into the regressions. Second, mosques are potentially not randomly located and differences between control- and treatment group could possible exist. Both problems can be solved by including location fixed effects at street level. This coincides with corrections that are used in the research of Dröes and Koster (2016), which makes the use of their methodology an even better fit for this research.
2.2.3 NIMBY and LULU
In this thesis the externality effect of arriving mosques on surrounding house prices is tried to be identified. Questions arise on what this externality effect is and how it originates. In this section the literature on the phenomenon that certain land use can cause a negative externality effect is elaborated. In the literature two terms on this subject are distinguished; “Not-In-My-Back-Yard” and “Locally Unwanted Land Use” (further referred to as NIMBY and LULU).
16 One of the first and more important researches that goes into the meaning, implications and solutions of NIMBY and LULU is the one of Brion (1988). On the one hand, he explains LULU’s as land uses that are not wanted in the neighbourhood. This could refer to examples such as waste storages or the arrival of a 50 story flat building. LULU refers to the land use itself, that is owned by the government or an private party. Although, even in case of privately owned LULU’s the role of the government in terms of permissions, zoning and others, is indispensable. On the other hand, he explains NIMBY as a phenomenon where neighbours of the LULU’s complain about the arrival of particular land use. In this case society agrees on the need of those land uses for the community to function accordingly. However, dislike the presence of those facilities near their own homes. In contrast of the owners of LULU’s, are NIMBY’s usually common among residential owners. The different interests between the two creates a certain tension field. Where the private owners or the government has to locate their LULU’s and the residents has to keep them out of their neighbourhoods. Brion (1988) states that this is where the problem comes in. First, the government cannot withhold their objective function to allocate LULU’s. Politics won’t choose for the utilitarian approach, where the best location for the greatest group is stated, but depend their choice on the balance of power in society. In a case study on the construction of a waste storage in Boston, Brion (1988) shows that in best case, logistically seen, an incinerator should be located in a more wealthier neighbourhood. However, politically seen, this was never achievable. The individual political potential gains from favouring wealthier residents outweigh the efficiently loss for society for locating the incinerator in less suitable neighbourhoods, householding economically and politically less powerful residents. Second, Brion (1988) highlights the fact that LULU’s were placed in poorer neighbourhoods, whereby the judicial system is failing in compensating the neighbours for the negative externality effects. For this research, both problems are of importance. Firstly, should mosques be located in neighbourhoods that are best fit, where the negative externality effect is minimized? Or is this the same as locating mosques where people are less powerful, poorer or unable to speak up for their rights. Finding a negative effect for mosques on house prices for different regions and voting behaviours will, as seen in the paper of Brion (1988), raise questions for policy makers whether this effect should be minimized. Question has to be answered such as; is compensation of neighbours for negative externality effects an illusion which doesn’t hold in practice? Should the decision on locating mosques be made on the welfare loss for the whole society, where poorer neighbourhoods will be the victim of? Or should they look at other characteristics such as multicultural diversity or other physical and locational characteristics? Brion (1988) introduces the importance and relevance for policy makers on subjects such as this thesis and highlights that the potential outcomes could help in the problem of locating LULU’s.
17 A more recent study comes from Schively (2007), where she explains more about the recent literature on both phenomenon’s. In this paper LULU’s are categorised into public service facilities, which correspond with the fear of loss in the quality of life and the property value, and into facilities that have consequences for the environment and the human health. For this research the first one is applicable, where neighbours are afraid of overcrowd, traffic density and decrease in house prices. Schively (2007) illustrates that the perception of these effects of LULU’s and the scientifically proven magnitude of these effects, do not correspond. In correspondence with the studies performed by Clotfelter (1975) and Jud (1985), Schively (2007) states that the externality effect found on house prices should not be blindly seen as a causal effect from physical characteristics, but also as an outcome from perceptions. She highlights that policy makers underestimate the impact of perception and amplifies that scientific results can shed light on those misinterpretation. Measuring the existence of the externality effect of mosques on surrounding house prices is therefore not the only goal of this research, defining and explaining the effect is of even more importance.
2.3 Previous Research on Houses of Worship
It is safe to say that there is not much literature written on the externality effects of houses of worship, whereby only one academic paper has been published that includes the impact of mosques. The four most important studies are analysed in this section. Moreover, a smaller thesis that is relevant for this specific research is briefly disclosed. Through analysing these researches, limitations can be identified and processed in the design and methodology of this research, and the positioning and relevance of this research can be clarified.
The first study regarding the externality effect of houses of worship is published by Do et al. (1994). In their research they tried to identify whether churches have negative or positive effects on surrounding house prices. The dataset consisted of transactions of 469 single-family dwellings over the time of 1991-1992, in the metropolitan area Chula Vista (U.S.). Using a hedonic pricing model, consistent of multiple housing characteristics and the distance to the church - as variable to measure the magnitude - the externality effect is examined. The results indicate a negative effect of -3.01% on house prices for households in a range of 50 feet, which fades out to 850 feet, where the effect appeared to be 0,00%. In their research, the relevance to debate on the consequences of public land use is clearly present. On the more public level, they argue that zoning and location planners should take externality effects into account. On the more private level, they argue that residents should stand up for their rights to be compensated, even if lawsuits are necessary to accomplish so. Although back in the 90’s this research was important for encouraging the debate, nowadays this research shows several limitations.
18 Firstly, although the hedonic pricing model corrects for non-church related effects by including housing characteristics. It did not exclude other factors that could bias the externality effect found. For example, the effect could also be caused by the arrival of a bus station one block away. Secondly, a hedonic pricing model does not allow for reverse causality. For instance, churches could be placed in neighbourhoods where house prices were already lower. Third and last, the external validity of the research is hard to check. It entails only single family dwellings, in just one metropolitan area. Solutions to all three limitation will be incorporated in this thesis and further put apart.
In contrast to the findings of Quang Do et al. (1994), the researchers Carroll et al. (1996) found a positive externality effect of churches on surrounding house prices, on average. Their research is of bigger magnitude and used a dataset of over 5,000 property transactions and 32 churches between 1986-1990 in Henderson (U.S.). Though, their results also showed negative externality effects for several churches underscoring the relevance for further research, since the outcomes from this and previous research is not conclusive. The outcomes of both studies could have differed because of the methodologies used, or because of the difference between regions and neighbourhoods used in the datasets. Furthermore, the results state that larger churches, measured by lot size, have positive effects on the magnitude of the externality effect. The research of Carroll et al. (1996) is still relevant since it both shows that; the direction of the effect can differ between houses of worship among different regions, and physical characteristics of churches influence the externality effect found.
After a gap in the literature of about 15 years, Babawale and Adewunmi (2011) published their research on the externalities of churches on apartments in Lagos, Nigeria. The dataset used in this research consists of 450 rental apartments and three major churches in the metropolitan area of Lagos. Using a hedonic pricing model the researchers try to find whether the negative externalities outweigh the positive ones of churches in the neighbourhood. They distinguish positive values such as; the symbolism of morality and a point for social gatherings, and negative values such as; church bells and noise pollution, traffic congestion and parking shortages. The results show, in correspondence to the research of Do et al. (1994), a negative externality effect. However, this research shows a lot of limitations. First, It is based on a small dataset, covering only 450 dwellings, three churches, and only one metropolitan area. Secondly, it examined only apartments, which reduces the sample even more if we look at unique locations for the dwellings. Thirdly, besides the fact that a hedonic model can’t correct for all external factors, the variables used in this research cover only the number of bedrooms, bathrooms, the availability for parking and accessibility, which is a very scarce list of characteristics. Overall, much reason to say that further research of more academic level should be conducted on the subject.
19 In contrast to Babawale and Adewunmi (2011), the most recent research is written in a more broader spectrum. In the research of Brandt et al. (2014) the externality effect of different types of houses of worship on house prices is examined. This research is the first to perform a research on this subject in Europe. Hereby the distinction is made between Islamic and Christian houses of worship, and therefore comes closest to the research of this thesis. In their research, data is used of almost 5,000 households in the entire city of Hamburg (Germany) between 2002-2008. Using a hedonic pricing model, predictions are made towards differences in the externality effects between Islamic and Christian houses of worship. Results show that both Islamic and Christian houses of worship have positive externality effects within a range of 1,000 metres. If the effect is resembled with households beyond this threshold, a positive effect is found of 4,6% for households within the 100-200 meter distance from the house of worship. Within the 100 meter range, no significant result has been found. Important for the outcome of this research is that it shows that residents in Hamburg do not see the arrival of mosques as deterioration of their neighbourhood, even as a positive contribution. The main limitation of this research is the fact that the dataset is based on only one metropolitan area, specific of the one of Hamburg. Since Hamburg is known for its open-minded and liberalistic point of view, generalising this research for the whole of Europe or specific the Netherlands is not applicable. Therefore further research on the externality effect of mosques in the Netherlands could shed light on the question whether perceptions towards mosques is generalisable for comparable regions across Europe.
Although it is not a research paper from one of the journals, the thesis of Leijten (2016) is worth to be mentioned in this literature review. Leijten tried to find an externality effect of mosques on house prices in Amsterdam (the Netherlands). His methodology is similar to the one used in the externality study of Dröes and Koster (2016), whereby the difference-in-difference method is combined with the hedonic pricing model. The results show that the arrival of a mosques in Amsterdam decrease house prices with 2.3% on average within the range of 150 meter. The used dataset consisted of 70% of the transactions from 1985-2015 (86,000 observations) and almost 50 mosques, which make the results credible. However, the main limitation of the research is the limited external validity for the rest of the Netherlands or the rest of Europe. Another limitation is the lack of reasoning behind the effect found by Leijten (2016). No variables are included in the methodology to ascertain the possible explanations for the negative externality effect of mosques found. The external validity and heterogeneity of the average treatment effect are tried to be improved in this research on the externality effect of mosques on surrounding house prices.
20 To summarize, it becomes clear from previous literature on houses of worship that there is still no conclusive result found. On the one side, Do et al. (1994) and Babawale and Adewunmi (2011) state a negative externality effect of churches. On the other side, Carroll et al. (1996) found a positive externality effect of churches and Brandt et al. (2014) a positive effect of both churches and mosques. The overall limitations in these researches are the following. The specific data samples covering only one metropolitan area and only one sort of dwellings, question the external validity. The use of simple hedonic pricing models is vulnerable for biased result due to omitted variables, non-random sampling and reverse causality. At last, the lack of examined factors to explain the externality effects, cause absence of explanatory power of the results. These limitation will be taken into account in this thesis by; i) combining the difference-in-difference model with the hedonic pricing model, as conducted by Leijten (2016) and Dröes and Koster (2016), ii) by expanding the dataset for different regions such as rural, sub-urban and urban ares, and different sorts of dwellings and iii) by finding explanatory factors such as the anti-Islamic and populistic right voting behaviour, to measure the perception of its residents towards the Islam, and other physical characteristics of mosques, to explain other heterogeneity in the externality effect. All will be elaborated in more detail in the methodology, Section 6.
21
3. Islam in the Netherlands
This chapter firstly addresses the history and arrival of the Islam in the Dutch and European landscape and secondly the rise of populistic right parties since the year 2000. Besides the causal relationship that is examined between mosques and surrounding house prices in this research, the most important sub-question is whether there is a causal relationship between anti-Islamic and populistic right voting behaviour and the externality effect of mosques on house prices. Therefore, this section covers the reasoning to use the local share of populistic right votes as a measurement for the anti-Islamic point of view and specifically the most prominent anti-Islamic party in the Dutch political landscape, the Party for Freedom (PVV).
3.1 History and Islamization
The first encounter with the Islamic culture was through colonisation in several parts of the world, among which the Dutch East Indies. The governance and policy towards this religious belief differed greatly among the colonists. The colonists encountered two different groups op Islamic followers; the ‘bad’ Muslims that resisted the colonisation of their lands and uphold fanatic following of the Islamic beliefs; and the ‘good’ Muslims that embraced Western and modern culture and practiced the Islamic belief in a more shallow way (Maussen, 2009). Their Islamic leaders and houses of worship were sponsored in order to withhold more control. However, the Dutch governance towards the Islam was of different nature. During the leadership of Snouck Hungronje, the Dutch regime believed the Islam should be approached with caution, resulting in the implementation of a policy with more respect towards the Islamic belief where the Islamic law, their freedom of religious worship and pilgrimage were sustained. However, similar to the French, the Dutch did encourage populations that imbedded more Western culture to reduce the gap between the two countries (Maussen, 2009).
In contrast to the more neutral policy of the Dutch, a system of pillarization and exploitation of the local community still existed. At first, when Indonesia claimed independence in 1945, almost 300.000 people, mostly Moluccan, fled to the Netherlands. This was primarily caused by the Dutch promise for protection and an independent state for locals that fought for the Dutch colonists in Indonesia at the time of revolt (Focus Migration, 2007). After World War II the Dutch cities were ruined, houses had to be rebuild but employment was still hard to find because of the many losses of the war, causing recruitment of Southern European guest workers. After the economy flourished in the late 60’s also guest workers from North Afrika and Turkey arrived in the Netherlands for employment (Dutch Language Union, sd) (Focus Migration, 2007). In contrast to policies of other European countries, in the Netherlands the foreign guest workers were managed by the institution for social workers who did
22 Source: Centraal Bureau van de Statistiek (CBS), Islamieten en Hindoes in Nederland, 1 januari
not have any experience with foreign cultures. In their governance no distinction was made between guest workers from North-Africa or Southern and Eastern Europe, causing even more pillarization (Maussen, 2009). Around 1975-1980 the second peak of immigrants was caused by the independence of Suriname, leading to resistance by Dutch civilians due to the already existing shortage on the housing market.
Remarkable is the scarce attention in the Netherlands for the Islam until 1970. The reason for this late change in policy could be explained by institutions like the Foundation for Foreign Workers, who were designed to only foresee in the basic needs such as accommodation and healthcare, in contrast to religious needs. Where in the beginning the stay of guest workers was indeed temporary, it changed from 1970 onwards when the Muslim guest workers united and the Netherlands realised that temporary guest workers would probably become permanent Dutch citizens (Maussen, 2009). This resulted in the first debates concerning the construction of mosques in Almelo and Rotterdam in the late 80’s. In line with the separation of Church and State the government stated that local municipalities were not obligated to subsidies and finance houses of worship, but should solely embrace the rise of the Islamic belief (Maussen, 2009). Together with the debate on the Islamic independence, several new laws were implemented concerning immigrants, among which the law of family reunification (Focus Migration, 2007) (Dutch Language Union, sd). In contrast to the Southern European guest workers, who returned to their home countries in this period, the Moroccan and Turkish population increased rapidly in the Netherlands. The main causes for this phenomenon were the families reunions, the high birth rates in these population groups and their economical- and political unstable home countries (Focus Migration, 2007). The more recent history also shows immigration waves from the Dutch Antilles, but since only 0,31% of their population follows the Islamic beliefs this is of less relevance for this research (EU Business School - Study Country, sd).
FIGURE 1:ISLAMIST IN THE NETHERLANDS 1971-2004
23
Source: Centraal Bureau van de Statistiek (CBS), Bevolking naar Migratieachtergrond. Own Translation.
The history of guest workers and immigrants in the Netherlands is still clearly visible in the multicultural society in the Netherlands. Figure 1, Figure 2 and Figure 3 clearly show how immigrants emerged over time and that the Surinamese, Moroccan, Moluccan (Indonesian) and Turkish immigrants indeed cover the biggest portion of Dutch immigrants. It is therefore not surprising that almost all mosques in the Netherlands can be divided in these four movements; the Surinamese, Moroccan, Moluccan and Turkish mosques, which is also visible in the data shown in Section 5. The next section will elaborate more on the most recent immigration of refugees to the Netherlands starting from 2000 and its effect on the pillarization of Islamic minorities in the Netherlands.
FIGURE 2:MIGRATION IN NETHERLANDS 1996-2016
It shows the neto immigration of four of five most important immirgations waves of the Dutch history as discussed. From the graph it save to conclude that thetraditional immigration from those countries almost settled around zero nowadays.
FIGURE 3:DUTCH POPULATION REFLECTED IN MIGRATION BACKGROUNDS
It shows the division of 1st and 2nd immigrants in the Dutch population as of 1 januari 2016. It becomes clear that besides
German, the Turkish, Maroccan, Surrinam and Indonesian (Mollucan) immigrants are the greatest represented.
24
3.2 The Rise of Populistic Parties
In the more recent years the perception towards the Islam changed. This can be explained by to role of the media, politics and events that occurred globally. One of the biggest events that changed people’s lives were the terrorist attacks on 09-11-2001 on the World Trade Centre in New York. The United States declared war on terrorism and held Osama Bin Laden and his terrorist organisation Al Qaida responsible. Fear among civilians for the rise of extremists in the Middle East emerged and its consequences on the debate on immigration policies in Europe are still visible (Boswell, 2007). Around the time of the attacks, the rise of the radical right in the political playfield began.
In 2002 the socialistic Prime Minister of France, Lionel Jospin, was defeated by Jean-Marie Le Pen in the first round of the presidential election (Norris, 2005). Le Pen, who represented the radical right, was described as someone who expressed herself in anti-Semitic and racist ways. Though a wave of bewilderment went through Europe, not much later in 2002, the Lijst Pim Fortuyn was founded in the Netherlands. Pim Fortuyn was assassinated three months before the elections and was known for his open speech of the integration problem of the Islam in the Netherlands (Norris, 2005). Ever since, we have seen a more open speech about the Islam and integration of Dutch immigrant in the Dutch society.
Moreover, the globalisation and evolution of the media caused the negative context the Islam was framed in to spread even faster. Theories such as the “News Value Theory” point out that news worthiness depends on news values, among which: Conflicts, magnitude, references to negativity and geographical proximity (Boukes & Vliegenthart, 2017). In recent media these elements are visible in news feeds such as those of Islamic State and the recent terrorist attacks in Europe. In particular the most recent news about the financing of Dutch mosques by the Middle East and possible terrorist linked organisations (NRC & Nieuwsuur, 2018),could be linked to the News Value Theory. It coincides with the theory since it concerns the Netherlands (proximity), is negatively charged, covers the magnitude of the terrorist actions of the past and exposes the conflict of Islamisation in the Netherlands. On the other hand, little attention is payed to the good aspects of Islamisation in the Netherlands. The danger of framing used by media is the misrepresentation of opinions in a country. The image could thus be outlined that most of the people disregard the Islam, which is the contrary if we look at the voting behaviour in the Netherlands. In the more recent years, political right parties in Europe responded to fear and framing, and used the more recent immigrant problem from the Middle East as foundation for their political agenda.
In the Netherlands this resulted in the foundation of the “Partij Van de Vrijheid” (PVV, also Party of Freedom) by Geert Wilders in 2006. This thesis uses the voting behaviour for the PVV as an explanatory
25 factor for the differences in externality effects of mosques on surrounding house prices per region. This is under the assumption that PVV belongs to a Populistic Radical Right Parties (PRRP) as defined in the literature. In this assumption the PVV is assumed to represent radical right residents that rebel against the coming of the Islam. Looking at the literature, the minimum definition of Political Radical Right Parties (PRRP) represented by Mudde (2007) shows that these parties are recognised by nativism, consistent of nationalism and xenophobia. Mudde points out that nationalism represents the love for, and creation of, one state for its own people. However, it does not necessarily represent hate towards others. In addition, nativism represents the combination of nationalism and xenophobia. Xenophobia, which stands for the fear and hate against strangers, comes along with prejudice that is represented in the ingroup-outgroup theory (Brewer, 2002). Hereby the own group is to be found superior to the other groups. Comparing the definition of Mudde (2007) to the ideas of the PVV, it becomes clear that the PVV indeed falls in the category of a Populistic Radical Right Party. What emerges from the political policy of Geert Wilders is this xenophobia, the hate and fear towards the Islam and the Islamic immigrants in the Netherlands.
It is therefore logical to hypothesize that the externality effect of mosques on surrounding house prices per region is dependent on the populistic radical right point of view of their residents, that could be represented by local share of votes for the PVV.
26
4. Hypotheses
In this section the hypotheses, tested in this research using statistical analyses, are summarized. At first, by analysing the literature on previous research on the externality effects of houses of worship, and in particularly on the ones of mosques, it becomes clear that the externality effect of newly arrived mosques on surrounding houses prices is not conclusive. Therefore will the first and main hypotheses (H1) be testing if there is a significant positive or negative externality effect on surrounding house prices to be found. Since a recent thesis concerning the externality effect of mosque in Amsterdam concluded that the externality effect was negative (Leijten, 2016), this thesis predicts an negative externality effect on average.
H1: There is a significant negative externality effect of mosques on house prices.
Secondly, since the previous literature on the externality effects of houses of worship, and particularly mosques, consequently uses datasets concerning only one metropolitan area, no answer has been found on whether the externality effects differs per region or municipality. To create more robustness and external validity for the results to be found, the second hypothesis (H2) will test for the heterogeneity of the externality effect per region and municipality.
H2: The effect of the opening of a mosque on house prices varies for different municipalities.
In the next five hypotheses will the heterogeneity and robustness of the possible externality effect of mosques on surrounding house prices be further revealed. As pointed out in Section 3 of this research, the history and current islamisation in the Netherlands is a sensitive and very present-day subject. In this section the connection has been made between the perception towards the Islam in the Netherlands and the rise of anti-Islamic and populistic right parties from the past decade. To further examine how this possible heterogeneity in externality effects of mosques, as tested in the second hypothesis, can be clarified, will the third hypothesis (H3) test whether the heterogeneity in externality effects of mosques on surrounding house prices per region is dependent on the anti-Islamic and populistic right voting behaviour of the region. In this research the voting behaviour for the Party of the Freedom (PVV) is used as an analogy for this anti-Islamic and populistic right voting behaviour, as substantiated in Section 3.2.
H3: The heterogeneity in the externality effect can be explained by the local share of anti-Islamic and populistic radical right votes
