Predicting Rental Prices in the Non-Regulated Housing Market of Amsterdam University of Amsterdam
FEB
Stijn In den Kleef 10517022
Supervised by: Mr. V. Nelidov 09 February 2018
Table of Contents Page Number Table of Contents………... 2 Abstract………... 3 List of tables………... 4 List of Figures………. 5 List of Appendices……….. 6 Introduction………. 7
1.1.Private Investment in the Dutch Rental Housing Market…………. 7
1.2.Research Method……….. 9
1.3.Research Questions……….. 10
1.4.Rationale of the Research……… 11
1.5.Structure of the thesis………... 11
2. Literature Review ……… 12
2.1.The Dutch Housing Market………. 12
2.2.The Housing Sector in Amsterdam……….. 14
2.3.Related Studies………. 16
3. Research Methodology………. 20
3.1. The Regression Model………... 21
3.2.The Variables……… 21 3.3.Data Collection………. 22 4. Data Analysis……… 23 4.1.Results………...………. 23 4.2.Discussion ………. 25 4.3.Limitations ……….………... 27 4.4.Conclusion……….. 28 Appendix 1……… 30 References………... 31
Abstract
This thesis investigates the rental levels in the non-regulated rental market in the Netherlands. The Dutch government started encouraging private investment in the housing market, thus it would be of high importance to study how rent prices are determined. The thesis seeks to examine whether rent prices can be predicted in an accurate way. A regression model is applied to establish a correlation between the rent price of a property and a set of variables. By establishing such a correlation, the thesis seeks to provide landlords and tenants with an estimation of rental prices based on house attributes and location.
List of Tables
Table Page Number
Table 1: Rental Prices in Amsterdam…….………. 24 Table 2: Overview of the Continuous Variables in Amsterdam ……. 25
List of Figures
Figure Page Number
Figure 1: Amsterdam Housing Market……… 15 Figure 3: Factors Affecting Rental Prices ………….………. 17 Figure 4: Conceptual Model……… 20
List of Appendices
Appendix Page Number
1. Introduction
1.1. Private Investment in the Dutch Rental Housing Market Over the last decade, there has been a growing investment rate in the Dutch rental housing market. The market offers valuable opportunities for private investors. As of 2014, there has been an unprecedented interest in the Dutch housing market from national and international investors alike. According to a report published by Capital Value—an advisory firm in the field of real estate— there were more than 200 international investors ready to invest in the residential housing market in 2014 (Capital Value, 2014). The same report indicated that local investors were to invest €2 billion in the rented housing, twice the amount invested by international investors. The report indicates that the rental housing market in the Netherlands started to witness a steady recovery following the housing market crisis of 2008. This recovery was marked by private investors making their entrance for the first time into the Dutch rented housing market. This has had significant implications for the private investors and the Dutch Housing associations alike.
The housing market is easily affected by socio-economic changes, such as demographic changes, income development, consumer behavior, and price fluctuation. The Dutch Housing market was badly hit by the 2008 crisis. Prior to the crisis of 2008, the presence of private investors in the housing market was insignificant. More than 85 % of the market was under the control of the Dutch Housing associations (Huisman, 2016). During the crisis, the returns from the housing market dropped to its lowest levels and the housing association started selling a significant part of their residential blocks, both regulated and non-regulated (Kroot & Giouvris, 2016). According to Huisman (2016), the
Dutch rental market makes up 40 % of the total housing stock. 50 % of the rental market is now dominated by private investors. Apart from the 2008 crisis, the participation of the private sector in the rental market has been facilitated also by the tendency of the
government to encourage the non-regulated housing sector.
The increasing interest in the Dutch rental housing is the most remarkable shift in the housing market in the last decade (Huisman, 2016). This increasing interest can be ascribed to a number of socio-economic factors (Jonkman, Janssen-Jansen & Schilder, 2015). These factors include are conditions on the housing market, such as increasing prices and shortage of the residential real estate market, what causes the rent levels to rise (Jonkman, Janssen-Jansen & Schilder, 2015, p.511). Commercial investors are turning their attention to rental apartment blocks mainly because they have low management costs and vacancy risks. In particular, private investors tend to invest in new urban areas because in most these dwellings are either non-regulated or are likely to be deregulated later (Huisman, 2016). This is mainly because non-regulated rental housing is more lucrative. Due to the governmental control in the regulated sector, investors have limited control over rent prices. It is in the non-regulated market that investors can enjoy the benefits of the free renting market.
The regulated and the non-regulated rental housing market offer different business prospects for investors. In the regulated market, the rent level is measured by valuation points (De Boer & Bitetti, 2014). Valuation points are allotted to every aspect of the dwelling, including "floor space, energy index, heating, toilet, lavatory, renovation, size of a kitchen" (Boelhouwer & Priemus, 2014). Rent price within the regulated rental market must be in accordance with the market rents. On the other hand, in the
non-regulated market, private investors enjoy the benefits of the free renting market. Van Bragt et al. explains (2015), "in the period 2016 – 2018 the boundary rent level of € 710 p/m is frozen so every year fewer points are needed to reach the maximum rent level of € 710 p/m" (p.100). This opens up endless opportunities for commercial investors to set the rent price. Since there is no government regulation, the valuation methods vary from one investor to another. The valuation methods may have no relation to the market demand or scarcity, nor do they reflect the market rents. Accordingly, finding a reliable way of predicting rent prices is inevitable. It is in this context that the scope of this paper falls. It seeks to explore the question of how to predict rent prices in the non-regulated housing market of Amsterdam.
1.2. Research Method
This thesis will apply an OLS regression model to determine the appropriate valuation method in the non-regulated rental market in Amsterdam. Through regression the value of a home can be determined by separating the different aspects of the home - the number of bedrooms, number of bathrooms, proximity to the city center by adding these variables into a model (Brunauer et al., 2010, p.405). Since it is theoretically difficult to predict rent prices in the non-regulated market, the OLS model will help provide an approximate rent price of rental unit. The model will analyze a number of independent variables that affect rental levels in Amsterdam. These variables include the specific features of the house, and the location variables. The proposed model aims at analyzing and understanding the structure and the drivers that might be of help in predicting rent prices.
The literature available suggests that rent levels are key factors in property valuation method. Commercial investors use them to determine the value of a property and its prospective investment value. As a result, an OLS regression model will provide real estate investors with a valuable tool to estimate the rent of residential properties (Brunauer et al., 2010). This will be a good asset for private investors who wish to reinforce their investment policy. In other words, an OLS regression model will help investors determine which residential properties have the highest rent levels and the highest potential returns. Moreover, the proposed model will provide the different stakeholders with the necessary understanding to predict rent levels in the free renting market. This will help them determine the profitability of the property and any extra services/facilities they may provide to the tenants.
1.3. Research Questions
Through the application of an OLS regression model, this thesis seeks to answer the following research question. The research question in this thesis is:
Can rental prices in the non-regulated housing market of Amsterdam be predicted more accurately?
A review of the literature available reveals that there a disparity between the real value of a property and the rent level. Accordingly, the thesis seeks to contribute to the literature on the Dutch non-regulated rental market. Given the fact that the non-regulated market in the Netherlands is relatively new, data on rent levels and house attributes is still scarce. Accordingly, an OLS regression model is likely to provide a way to predict rent prices in Amsterdam non-regulated rental market.
1.4. Rationale of the Research
The increasing interest in the Dutch rental housing is the most remarkable shift in the housing market in the last decade. Following the 2008 crisis, the real estate market has been marked by the entrance of the private investors on the one hand, and by the tendency of the government to encourage the non-regulated rental market on the other (Wahlen, 2016). The academic interest in the Dutch non- regulated rental market is relatively new. Accordingly, the literature on the subject is still scarce. It is this context that the scope of this thesis falls. The relevance of this thesis will be to propose a valuation model that can be used to predict rent prices. According to the Dutch Centraal Planburea (CPB, 2017) real estate corporations rent out luxurious dwelling below market value, leading to waiting lists. Its major contribution will be to provide real estate
investors with a valuable tool to predict rent levels of residential properties in the non- regulated sector.
1.5 Structure of the Thesis
The thesis is made up of four inter-related chapters. The first two chapters establish the theoretical background of the research. They introduce the concept of non-regulated rental housing in the Netherlands and review the existing literature. The first chapter is a general introduction to the thesis where the research problem, the research methods, and research question is formulated. The second chapter reviews the literature available in the field of the Dutch Housing Market. This chapter contextualizes the research problem by narrowing it down to Amsterdam. The importance of this chapter lies in the way it establishes the theoretical framework for the next chapters by reviewing
the previous studies. Chapter 3 details the research methodology, while chapter 4 presents the results of the OLS regression model and discusses the findings.
2. Literature Review
2.1. The Dutch Housing Market
The governmental regulation of the housing market seems to be a distinctive feature. Despite the government's encouragement of the non-regulated housing sector in the last years, the market is not completely free. Subdivision and a strict spatial planning policy have helped the government to maintain its control of the market. This regulation has caused home ownership to be low in the Netherlands (De Wit, Englund & Francke, 2013). According to the current urban planning policy in the country, house building is only possible in the suburbs (Lu & Stead, 2016). All these reasons contribute to keeping home ownership at a rather low level despite some progress (Lu & Stead, 2016).
The government's control of the housing sector extends to the rental market as well. It has followed a rigorous policy of providing social housing to people with low-incomes and of keeping a firm grasp of rental market (Lu & Stead, 2016). The Housing Act of 1901 has set the tone for the government's provision of social housing and the continual improvement of the housing conditions of the people. In pursuance of this, the government pays housing allowances to people of low-income which can go up to €699/ month. This is the threshold between regulated and non-regulated rental sectors.
However, the determination of the maximum rent is not a random matter. The Netherlands uses a points system known as woningwaarderingsstelsel (Property Valuation System). It assigns points to a property based on the characteristics of the house, the location and its type (De Wit, Englund & Francke, 2013). However, the
market for non-regulated real estate is still relatively small. The low participation of private investors in the rental market is due not only to the valuation system but also the restrictions on home ownership imposed by the government (Agnello & Schuknecht, 2011). The early private participation in the housing sector goes back to the year 2012 (Hochstenbach & Boterman, 2015). This coincides with the economic recovery after the 2008 crisis, but also with the government's continuing encouragement of the
non-regulated rental market.
It is also important to note that the government's regulation of the housing market can affect the value of the property. In other words, one might ask here: does the point valuation system really reflect the market conditions? The literature available suggests that there is a discrepancy between the rent levels in the regulated market and those in the non-regulated market. The Dutch points system is meant to determine the rent level for social housing, but also to serve as a guide for the free market sector. However, to be legible for social housing, applicants should have an income of no more than 34.000 Euro per year (Hekwolter et al, 2017). In practice, this means the long
waiting time that is between 3 and 10 years (Hekwolter et al, 2017). This also means that, while on the waiting list, many people have to rent in the free market and end up paying more than €699/ month. Accordingly, it can be argued that the rent level for social housing does not reflect the market conditions. As stated by Hekwolter et al. (2017), the average rent level in Amsterdam is estimated at €1.900.
Moreover, there is a perceived gap in the value of the property (Hekwolter et al, 2017). In practice, a value gap means that the price of houses is very high but the rent level is very low. In the Dutch context, the value gap is caused by the government's social
housing, the regulation of the rent and the housing allowance program (Kroot & Giouvris, 2016). In its attempt to increase homeownership, the Dutch government
sponsors those houses that are occupied by the owner themselves (Hekwolter et al, 2017). This can be in the form of subsidies or through deducting mortgage rate. The value gap in the housing continues in the wake of the 2008 financial crisis. The crisis has caused the price of houses to decline by 20% and the rent level to soar by 5 %. Because of this value gap, the rent level in the Netherlands does not reflect the market conditions.
The last 5 years have witnessed a shift in the Dutch housing market. The shift consists of the increasing presence of private investors in the market. This coincides with the efforts of the government to encourage private investment. According to Hekwolter et al, (2017), the government tries to increase the non- regulated rental housing sector by limiting the taxes for house owners and stimulating the transformation of vacant office space into non- regulated real estate (p. 39). However, the presence of private investors is still insignificant compared to other European countries. In 2016, it was estimated that there were around 7.6 million houses in the Netherlands, 40% of which is rented (Hekwolter et al, 2017). The same study revealed that social housing made up 39 % of the total housing market while the non-regulated sector accounted only for 3 %. When compared to Europe, the Netherlands has the smallest free housing market. (Hekwolter et al, 2017).
2.2. The Housing Sector in Amsterdam
Of all the Dutch cities, Amsterdam witnesses a rapid demographic growth. The city is usually hailed as the premier destination for expats and tourists. According to Van Schuiling (2005), around 50 % of the population in Amsterdam has at least one foreign
parent, which makes it one of the most cosmopolitan cities in Europe. This has significant consequences on the housing sector. Over the last two years, the city has seen a sharp housing shortage. The demographic changes in the city are not related only to the number of new settlers. Changes in the type of households contribute to this housing shortage. The city is witnessing an increase in the number of single households (Hekwolter et al, 2017). Moreover, given the percentage of expats in Amsterdam, the rental market has gone up significantly. Today, the rental market in Amsterdam is far above the national level (Hekwolter et al, 2017). The rental market in Amsterdam represented more than 70 % of the total housing sector and around 45 % of it is non-regulated (Hekwolter et al, 2017).
Overall, the housing market in Amsterdam is divided between owner-occupied houses, social housing, and private rent houses. Owner-occupied homes represent a small
percentage of the housing sector. According to Hekwolter et al, 2017, there are 480.000 households in Amsterdam by 2106. The following graph represents the distribution of
households in Amsterdam.
Figure 2: Amsterdam Housing Market 30.20%
25% 44.80%
Amsterdam Housing Market
2.3.Related Studies
Rent price is of growing interest to real estate experts in the Netherlands (Tu, De Haan, & Boelhouwer, 2017; Lennartz, Haffner & Oxley, 2012; Huisman, 2016). In the last few years, the Dutch housing market has been marked by the entrance of private investors into the market. Since the government exerts no direct control on the rent prices in the non-regulated housing market, investors find it difficult to adequately predict the rent price (Lennartz, Haffner & Oxley, 2012). According to Huisman (2016), the difficulty to predict rent prices is mainly due to the fact that there is no fixed ratio between the value of a rental unit and the rent cost. There are many drivers that interfere with the rent price. Some of these drivers are related to the condition of the house, some to its location and some to the characteristics of the house and its features (Tu, De Haan, & Boelhouwer, 2017).
There has been substantial research on the drivers of the residential housing market. Overall, these drivers can be channeled in two directions, namely demand, and supply (Tu, De Haan, & Boelhouwer, 2017). The first direction includes the economic and the demographic factors, while the second direction is made up of such factors as construction and real estate stock (Blessing, 2015). These drivers are interconnected and they affect each other in the short and the long run. For the sake of consistency, this part of the thesis reviews the literature available on the factors of the free rental market based on the house attributes. However, it would be methodologically appropriate to give a summary of the different drivers of the residential market. The number of factors that can affect the rental price vary from the regulated to the non-regulated market.
A review of the literature available reveals that there have been numerous attempts at finding a way of predicting house prices (Tsolacos, Brooks & Nneji, 2014; Engsted & Pedersen, 2015). These studies have sought to determine the different factors that affect rent prices. For instance, (Engsted & Pedersen, 2015), explore the predictive power of the rent-to-price ratio in determining the rent price. Their work has proved that there is a correlation between the value of the house and the rent price. They argue that an increase (decrease) in the ratio signals a future increase (decrease) in returns (Engsted & Pedersen, 2015, p. 257). Even though the primary objective of their study was to determine whether it is better to buy or to rent a house, the findings of their research relate to the scope of this thesis in that they establish the house value and the location as drivers to determine rent asking value. Tsolacos, Brooks & Nneji (2014), used a probit model and a Markov-switching model to predict the trends in rent prices for two years. Their approach to rent prediction was based on the market behavior in office, retail, apartments, and warehouses (Tsolacos, Brooks & Nneji, 2014).
Brunauer et al. (2010) made a hedonic approach to price modelling based on spatial heterogeneity. The research was conducted in the city of Vienna and sought to establish a correlation between rent levels and the locational variable. The hedonic model developed in this study that the location of the house and the postal code affected the rent level (Brunauer et al., 2010). In the model, closeness to the city center had a positive effect on the price level. The application of hedonic models to real estates was first introduced in 1984. He carried out a hedonic regression model to explore the variation in rent levels in the regulated and the non-regulated Canadian housing market. Hedonic models can contribute to the understanding of rent levels by establishing a causal effect
for a number of variables. However, this model can only be reliable if it includes a comprehensive list of variables, which is out of the scope of this thesis. What this thesis seeks to do is to establish correlations between rent level on the one hand, and a set of variables on the other hand.
Rental prices have been regressed on a number of variables in the literature available. Coates & Matheson (2011) explore rent variations in relation to two sets of variables: constant variables and temporary variables. They use mega-events, such as the Olympic Games as an example of temporary variables (Coates & Matheson, 2011). They introduce mega events as time on the market variable. Constant variables are those variables that relate directly to the characteristics of the house, its location and its situation. The findings of the research point to a causal effect between the time on the market variable (the mega-events) and the rent price. Time on the market seems to increase the rent level significantly compared to the other variables. McCord et al. (2014) made a similar approach to rent price when they developed a linear regression model to predict rental prices against location and time on the market. Their study was carried out in the UK and it has established a correlation between the regressed variables. The studies reviewed so far in this thesis, were all conducted outside the Netherlands. Their relevance to the research lies in the fact that they establish the specific characteristics of the house, the house features, location and time on the market as factors to predict rental prices.
In the Dutch context, there have been a few attempts at predicting rental prices. Priemus (2011) studied the Dutch housing market in comparison to its American counterpart. The study concluded that the rent price in the Netherlands is difficult to
predict in the non-regulated market for two reasons: the lack of direct governmental control and the scarcity of relevant literature (Priemus, 2011). Don (2009) maintained that the current policies of the Dutch housing market results in a gap between the value of the property and the rental price. Such a gap, he argues, needs to be filled by a clear agenda on how to predict the house prices in the country, which should include a comprehensive list of valuation points. Boelhouwer & Priemus (2012) investigated the situation of the rental market in Amsterdam and concluded that, given the current points valuation system, the rental prices do not reflect the market condition. The effect of this is a significant range of variations in the rental prices. What these studies have sought to do is an analysis of the current market condition in the Netherlands. They seem to agree that predicting the rental price in the non-regulated housing market is difficult.
3. Research Methodology
This chapter explains the OLS regression model used in the thesis. It starts by designing the theoretical model of the regression (Figure 3). Then it moves to explain the model and the relevant variables.
Figure 4: Conceptual Model Rent Level Asking
House Specific Characteristics House Features
Size Bedroom Bathroom Garden Roof Terrace Balcony Bath Shower Fireplace Air condition Furniture Sharing Parking place Garage Elevator Storage Pool Location Center East West North Westpoort New-west
3.1. The Regression Model
This paper seeks to determine if rental prices can be predicted in Amsterdam’s non-regulated housing market. To help address rent level, the thesis has analyzed a total of 3.152 rental units in the non-regulated market in Amsterdam. The rent asking analyzed extends over the period between Q1 2011 and Q2 2014. The regression model includes 26 variables and correlates them with the rental prices. A linear regression model will be applied to try and answer the research question. The model developed in this paper regresses the dependent variable rent on three independent set of variables as illustrated in Figure 4. The regression model can be represented as follows:
𝑅𝑒𝑛𝑡𝑖 = 𝑓(𝑆𝑖𝑗, 𝐹𝑖𝑗,𝐿𝑖𝑗,)
In this model rent is the monthly rent for one rental unit, S is a number of j house specific characteristics, F is a number of j house features and L is a number of j locations.
The base hedonic model developed in this thesis is performed as follows: Rental Price Amsterdam, i
= 𝛽0 + ∑ 𝛽𝑎 𝑆𝑎,𝑖 𝑎 𝑎=1 + ∑ 𝛽𝑏 𝑏 𝑏=1 𝐹𝑏,𝑖 + ∑ 𝛽𝑐 𝑏 𝑐=1 𝐿𝑐,𝑖
where rent is the monthly rent for one rental unit, S is a number of j house specific characteristics, F is a number of j house features and L is a number of j locations. 3.2. The Variables
In this section of the thesis, the relevant variables are explained. To start with, the paper uses the monthly rent asking in the free rental (non-regulated) market as its
dependent variable. It is shown in the table both as a logarithmic function and as a
The first set of independent variables is referred to as house specific characteristics. These characteristics are all drawn from existing literature. According to the literature available, all house specific characteristics impact the increase rent level, except the for sharing variable. The category of rental units available for sharing comprises those units designed for students’ occupation, which is likely to decrease the rent asking price. As it can be seen from the conceptual model in figure 4, one of the house characteristics is furniture. It can take two forms: furnished or shell. The second set of independent variables is referred to in the thesis as house features. This includes the presence of a parking place, a garage, an elevator, a storage and a swimming pool. Based on the existing literature, all house features are expected to increase the rent price. The last set of variables is location. This includes distance to center, East, West, North, Westpoort and New-West.
3.3. Data Collection
The major source of data in this thesis is Pararius.nl. The source is very reliable as it is considered as the premium source of rental housing. This is mainly because
Pararius.nl is connected to a large network of Dutch real estate agencies. Since its foundation in 2006, this database offers substantial data on the housing market in the country, both regulated and free. Almost every type of data is available there, ranging from house features, rent levels, zip codes etc. This thesis analyzes 3.152 residential units. These rental units were listed on Pararius.nl in the period extending from Q1 2011 and Q2 2014. There is no data available on house specific characteristics prior to 2011 on Pararius.nl.
Since the scope of this thesis falls within the non-regulated rental housing market, the data extracted from Pararius was filtered. In a first step, data related on the regulated market had to be deleted from the corpus. Any rental unit with a rental level below the monthly price of social housing was eliminated. Second, all the details related to house types, time on the market have not been considered as they are out of the scope of the paper. Only the details of the house specific characteristics house features and location variables were processed.
4. Data Analysis
In this section of the thesis, the major regression results of Amsterdam as full sample are discussed, of both the linear and the semi-log models. The regression results are for Amsterdam as a full sample are presented in Table 2. In the regression model, the rent price is used as a continuous variable in an attempt to account for rent variations in the non-regulated rental market. Moreover, the regression model is performed with the market conditions in mind. This is the reason the model uses quarterly time effect rather than monthly dummies.
4.1.Results
This part of the thesis introduces the results of the OLS regression model. The observations made in the model included 3.152 rental units in Amsterdam and the suburbs. Table 1 represents the results of the OLS regression model in Amsterdam. The data extract from Prarius.nl was later processed statistically to come to an estimated mean rent price in Amsterdam. The statistical description is represented in Table 2. To show the application of the regression model, a case study is given in Appendix 1.
The data extracted from Prarius.nl was then statistically processed to calculate the mean rental price per rental unit in Amsterdam. The statistical description was designed to establish some correlations between these variables and the rent level. The results of the statistical study were used to test the validity of the results of the regression. Table 2 summarizes the statistical description of the data.
Table 2: Overview of the Continuous Variables in the Amsterdam 4.2. Discussion
As far as the regression results of the linear model are concerned, there were 3.152 observations made, which accounts for 76% of the overall variance. This almost corresponds to the adjusted R-squared. The findings of the regression model are compatible with the existing literature. For instance, the size of the unit seems to be a determining factor of rent level. As it can be seen from Table 1, the size factor translates into a €15,86 marginal increase for every square meter extra in the unit size. The size factor can be seen also in the variation in the number of bedrooms. The more bedrooms the higher the rent level. The regression shows that rent asking level for rental units with one bedroom decrease the rent level with € 10,91 while that of four-bedroom rental units
Amsterdam
Variable Mean Medium Min Max
Independent Variables: Size 92 85 14 675 #Bedrooms 2 2 1 8 #Bathrooms 1 1 1 5 Dependent Variables: Rent €1.605 €1.555 €670 €13.497 Rent per m2 €19 €19 €5 €50
increased with €511,14. The bathroom variable was also important in determining rent level. The presence of an extra bathroom results in an increase of €61,21 when everything else stays the same.
As expected, the other house specific characteristics yielded similar results. They all impact the rent asking level. For instance, the presence of a garden and a roof terrace increased the rent level per rental unit by € 44,65 and € 66,23 respectively. The other characteristics, also play a significant role in increasing or decreasing the rent level. Two variables seem to decrease the rental price. First, there is a significant rent variation according to the state of the house, namely furnished or not. The regression has showed that when a rental unit is leased in a shell condition, its price will be €389,29 less, all else the same. Second, rent units available for sharing are €59,19 lower. Most of the rental units in the available for sharing category are intended for students, which explains the rent level variation.
The results of the regression in relation to the second set of variables, house features, are also significant. There are correlations between house features and the rental price. The existence of a swimming pool is the only variable that brings an increase in the rent level. It brings a marginal increase of €452,17 per rental unit. All the other feature of the house: parking place, elevator, storage etc. have a negative effect on the rent level. The storage, the parking place and the elevator bring the rent level down by around €42, €29 and €13 respectively.
The findings of the regression demonstrate, also, that there exists a correlation between the location of the rental unit and rents. The model tested variables that describe the location of the property either in the Center, West, East, North, Westpoort and
New-West. The results show that rental units located in the center are highly-valued as they increase the rent by around €65. The North and the New-west areas are the farthest to the city center and accordingly, they decrease the rental price by €550 and €561 respectively.
The statistical data shown in the table suggest that house specific
characteristics, house features and location increase/decrease the rental prices. The correlation between these variables and the rental prices is illustrated in the results of the linear regression above. The statistical data demonstrate that the mean rental price in Amsterdam is €1.605 for a 2-bedroom and one-bathroom house for 92 square meters. The study as attached in Appendix 1 shows how close the predictions using the results are to
the actual rent.
5.1.Conclusion
This thesis has investigated rental prices in the non-regulated rental market. It sought to determine whether it is possible to predict rental prices in the free rental market. The findings of the regression model demonstrate that there is a correlation between rental price and the regressed variables. The thesis has tried to provide an understanding of how rent prices can be determined in the free rental housing market. Upon reviewing the literature available, there seems to be a lack of data on the Dutch non-regulated housing market. The OLS regression model has helped to analyze of the rent levels and try to predict them. The analysis of the data has taken place at two levels in the thesis. In the first level, a regression model was performed to Amsterdam full sample. The
regression model sought to answer the following question: Can rent prices be predicted more accurately?
Based on the results of the regression model, a number of variables seem to correlate rent levels in the non-regulated market. The regression model performed on Amsterdam full sample accounts for 76% of total variance. The results of the model are compatible with the findings of previous studies in the field. The variables in the model play a significant role in determining the rent level. For instance, the size of the rental unit seems to have a positive impact in the regression. The rent level appears to be higher in the city center and lower in the suburbs despite the presence of extra features, such as the garage, the parking place and an elevator. At a second level, a statistical description of the data was provided in order to determine the mean rental price in the city center. A case study was conducted on a 2-bedroom and one-bathroom apartment to test the validity of the correlations. The study shows the model created in this thesis yields accurate results predicting rental levels.
5.2.Limitations
This thesis has investigated rental prices in Amsterdam non-regulated housing market. Its main goal was to come to an understanding of how rent prices are set. A regression model was developed and performed to correlate a number of variables with the rent level. The model has yielded significant correlations in this regard. However, this paper lays no claims to perfection. There are a few shortcomings to the research, which can be improved in further research.
The first limitation of the research is related to the variables. There are far more variables that can be regressed against rent level. The thesis has focused on only three sets: house specific characteristics, house features and location. However, variables such as house type, time on the market can all contribute to the findings. Moreover, the
regression has only established correlations between the variables. The findings of the regression model cannot ultimately be used to determine rental prices. In order to do this, the causal effect of the variables needs to be established. For future research it is
recommended to take more variables into account. By broadening the scope of the research, better conclusions can be drawn based on energy labels, marketing of the property or age of the property e.g..
Furthermore, this paper only focuses on Amsterdam, it is recommended to take more areas into account for future research.
Appendix 1: A Case Study Example
In this case study, the focus is on location as an independent variable and rent as a dependent variable. A hypothetical two-bedroom, one-bathroom unfurnished apartment is moved between the city center and New-West to see how the price changes. Accordingly, the rental price can be predicted based on the role of the locational variable. According to the statistical description presented in Table 2, the mean base price is €1.605, excluding location. The mean price is the sum of all the marginal prices related to house specific characteristics and house features, but it excludes the locational variable.
If this apartment is located in Rosmarijnsteeg, in the city center, then the
regression predicts that the rental price for this apartment would be €1.670 a month. If we keep everything else constant, and we move this apartment slightly away from the city center, the price will decrease based on the direction. For example, if this apartment is located in the Jan van Zutphenstraat, New West the price of the apartment would be €1.044. A search was made on Prarius.nl for an apartment similar to the hypothetical apartment and the following results were found. An apartment in Rosmarijnsteeg, with two bedrooms, one bathroom and unfurnished was advertised at a price €1.675 per month. That is just €5 higher than the hypothetical apartment. Another apartment in the Jan van Zutphenstraat was found with a rental price of €1.040, matching the hypothesis. The case example shows how the regression model developed in this thesis can be used to predict that the rental price can vary according to the location and the distance to the city center. Various other apartments have been studied with similar results.
References
Agnello, L., & Schuknecht, L. (2011). Booms and busts in housing markets:
Determinants and implications. St. Louis: Federal Reserve Bank of St Louis. Blessing, A. (2015). Public, Private, or In-Between? The Legitimacy of Social
Enterprises in the Housing Market. Voluntas, 26(1), 198-221.
Boelhouwer, P., & Priemus, H. (2014). Demise of the Dutch Social Housing Tradition: Impact of Budget Cuts and Political Changes. Journal of Housing and the Built Environment, 29(2), 221-235.
Brunauer, W. A., Lang, S., Wechselberger, P., & Bienert, S. (2010). Additive Hedonic Regression Models with Spatial Scaling Factors: An Application for Rents in Vienna. Journal of Real Estate Finance and Economics, 41(4), 390-411.
Capital Value (2014), An analysis of the Dutch Residential (investment) Market. Retrieved from: https://www.capitalvalue.nl/nl/onderzoek
Coates, D., & Matheson, V. A. (2011). Mega-Events and Housing Costs: Raising the Rent While Raising the Roof? The Annals of Regional Science, 46(1), 119-137. De Boer, R., & Bitetti, R. (2014). A Revival of the Private Rental Sector of the Housing
Market? Lessons from Germany, Finlad, the Czech Republic and the Netherlands. Paris: Organization for Economic Cooperation and Development (OECD).
De Wit, E., R., Englund, P., & Francke, M. K. (2013). Price and Transaction Volume in the Dutch Housing Market. Regional Science and Urban Economics, 43(2), 220. Don, H. (2009). Agenda for the Housing Market. De Economist, 157(2), 251-264 Engsted, T., & Pedersen, T. Q. (2015). Predicting returns and rent growth in the housing
market using the rent-price ratio: Evidence from the OECD countries. Journal of International Money and Finance, 53, 257.
Hekwolter, M, Nijskens R & Heeringa, W (2017) The Housing Market in Major Dutch Cities. Amsterdam: De Nederlandsche Bank N.V.
Hochstenbach, C., & Boterman, W. R. (2015). Navigating the Field of Housing: Housing Pathways of Young People in Amsterdam. Journal of Housing and the Built Environment, 30(2), 257-274.
Huisman, C.J (2016). A silent shift? The Precarisation of the Dutch Rental Housing Market. Journal of Housing and the Built Environment, 31 (1), 93-106. Jonkman, A., Janssen-Jansen, L. & Schilder L. (2015) Rent Increase Strategies and
Distributive Justice: The Socio-Spatial Effects of Rent Control Policy in Amsterdam. Journal of Housing and the Built Environment,1 (4), 1-25.
Kroot, J., & Giouvris, E. (2016). Dutch Mortgages: Impact of the Crisis on Probability of Default. Finance Research Letters, 18, 205.
Lennartz, C., Haffner, M., & Oxley, M. (2012). Competition Between Social and Market Renting: A Theoretical Application of the Structure-Conduct-Performance Paradigm. Journal of Housing and the Built Environment, 27(4), 453-471
Lu, P., & Stead, D. (2016). Understanding the Notion of Resilience in Spatial Planning: A Case Study of Rotterdam, the Netherlands. Cities, 35, 200-212
McCord, M., Davis, P. T., Haran, M., McIlhatton, D., & McCord, J. (2014).
Understanding Rental Prices in the UK: A Comparative Application of Spatial Modelling Approaches. International Journal of Housing Markets and Analysis, 7(1), 98-128.
Priemus, H. (2011). Renting in the United states: A Dutch Perspective. Cityscape, 13(2), 159-162.
Tsolacos, S., Brooks, C., & Nneji, O. (2014). On the Predictive Content of Leading Indicators: The Case of U.S. Real Estate Markets. The Journal of Real Estate Research, 36(4), 541-573.
Tu, Q., de Haan, J., & Boelhouwer, P. (2017). The Mismatch Between Conventional House Price Modeling and Regulated Markets: Insights from the Netherlands. Journal of Housing and the Built Environment, 32(3), 599-619.
Van Bragt, D., Francke, M. K., Singor, S. N., & Pelsser, A. (2015). Risk-Neutral Valuation of Real Estate Derivatives. Journal of Derivatives, 23(1), 89-110. Van Schuiling, D. (2005). The Amsterdam Housing Market and the Role of Housing
Associations. Journal of Housing and the Built Environment, 20(2), 167- 181 Wahlen, S. (2016). Crisis, Inequality and Consumption - a Dutch perspective. Italian
