The Impact of Decreased Interest Rate Deductibility on The Residential Real
Estate Market: Empirical Evidence from The Netherlands
A Master Thesis by
Rogier M. S. C. van Griensven
11083255
University of Amsterdam MSc Real Estate Finance
July, 2016
Supervisor: Ryan C. R. van Lamoen 2nd Reader: ___
Statement of Originality
This document is written by student Rogier van Griensven, 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
Acknowledgements
I would like to thank my supervisor, Ryan van Lamoen, for the time he made to coach me on writing the masters thesis. Even though he is working a full time job at DNB, he managed to be always quick of response and showed real dedication as a supervisor. His knowledge on both the subject and the analytical methods used in this thesis created a strong basis for his coaching, which truly helped me improve on my thesis.
Furthermore, I would like to thank my friends and family, who were always up for a discussion on the Dutch housing market. These discussions gave me the insights that I
Abstract
This study investigates the effects of a change in interest rate deductibility regulations on the housing prices and number of transactions in The Netherlands. Numerous research has been conducted in this topic in liberal markets such as the US and the UK. This study provides the first empirical results of the effects that the change in interest rate deductibility of January 2013 had on the strongly regulated housing market in The Netherlands. To analyse these effects, panel data from the Dutch Household Survey (DHS) of 2001 till 2015 and aggregated data on the national level from the Central Bureau of Statistics (CBS) of 1995 till 2016 are used. The effect of the changes in interest rate deductibility regulations on the housing prices is analysed by a static panel regression, and the effect of the changes in interest rate deductibility regulations on the number of transactions in the market is analysed by an OLS time series regression. Both models are extensively tested on significance and robustness and a significant negative effect of the change in interest rate deductibility regulations on housing prices is found.
Table of Contents
Statement of Originality ... 2
Acknowledgements ... 3
Abstract ... 4
Table of Contents ... 5
1. Introduction ... 6
2. Related Literature ... 9
2.1 Interest Deductible Housing Markets in the World ... 9
2.2 The Influence of Interest Rate Deductibility on the Housing Market ... 11
3. The Dutch Mortgage Market ... 14
3.1 Interest Rate Deductibility ... 14
3.2 The Non-‐‑Profit Social Housing Institutions and the Strong Tenant Protection & Rental Control ... 16
3.3 Restrictive Regulatory Zoning Regime for Construction & Development ... 17
3.4 Mortgage types ... 17
3.5 Hypotheses ... 19
4. Data & Descriptive Statistics ... 20
4.1 Data Preparations ... 21
4.2 Summary Statistics ... 23
5. Methodology ... 25
5.1 Modelling the Effect of Changed Interest Rate Deductibility Regulations on Prices . 26 5.2 Modelling the Effect of Changes in Interest Rate Deductibility Regulations on Transactions ... 30
6. Results ... 32
6.1 The Effect of Changes in Interest Rate Deductibility Regulations on Prices ... 32
6.2 The Effect of Changes in Interest Rate Deductibility Regulations on Transactions .. 36
7. Robustness ... 38
7.1 The Robustness of the Estimated Effect of Changes in Interest Rate Deductibility Regulations on Prices ... 38
7.2 The Robustness of the Estimated Effect of Changes in Interest Rate Deductibility Regulations on the Number of Transactions ... 45
8. Discussion and Conclusion ... 46
9. References ... 50
10. Appendices ... 54
10.A Descriptive Statistics and Summary Statistics ... 54
10.B Regressions ... 58
1. Introduction
The financial credit crisis caused a lot of turbulence in many economies around the globe (Kolb, 2010). It has taught us that markets are more intertwined than we expected before. It has also taught us that the negative consequences of debt usage can become real. The biggest asset class worldwide is real estate and it is highly leveraged by debt (Bennigson, 2004). The leverage ratio in real estate as an asset class is generally higher than, for example, in companies, because it is seen as a relatively safe investment. This counts for both commercial real estate, where investors leverage their assets with debt and for the housing sector, where homeowners can leverage their home with a mortgage. The crisis showed us that also real estate markets can collapse. Numerous studies have determined the effects of the crisis and discovered the cause of it. Mendoza and Quadrini (2010) discuss the impact of financial globalization on the amount of leverage used in assets. They also discuss whether increased leverage was a cause of the financial crisis. Brunnermeier (2009) explains the cause of the losses in the mortgage market and how this amplified into turmoil on the financial markets during 2007 and 2008. As a result of this crisis, governments tightened credit regulations for financial institutions, in order to prevent excessive borrowing behaviour. Several studies focus on this matter, also in the specific area of mortgage debt. Regulation and deregulation on the mortgage market have been researched extensively in the United States. Campbell (2013) discusses mortgage reforms and lessons learned from the credit crisis in the US. As some countries have already phased out the interest rate deductibility, The Netherlands followed them since January 2013 (Eigenhuis, 2016). New interest rate deductibility regulations have become active in The Netherlands since then. The deductibility rate will decrease with 0.5% on a yearly basis, which means that it will decrease from 52% in 2012 to 38% in 2041 (Eigenhuis, 2016). The maximum loan-‐‑to-‐‑ value (LTV) is gradually decreases with 1% a year from 105% maximum in 2013 to 100% maximum in 2018. Furthermore, the interest only mortgages are declared non-‐‑ deductible since January 2013.
The main goal of this study is to estimate the effects of the tightened interest rate deductibility regulations in The Netherlands on housing prices and transactions. There is still little research conducted outside of the US or the UK on the topic of interest rate
deductibility. The Dutch housing market is one of the highest leveraged housing markets worldwide (NVB, 2014). Solely from a leverage point of view this is internationally perceived as a high risk, as leverage works as a multiplier on both profits and losses, it increases the volatility of the returns and thus increasing risk. In fact, there is more to this subject than what is commonly known. The Dutch housing market is shaped by four dominant forces; the interest rate deductibility for homeowners, non profit social housing institutions, strong tenant protection and rental control, and a restrictive regulatory zoning regime for construction and development (NVB, 2014). Adding to the memo of the NVB (2014) and prior studies on the matter of a change in interest rate deductibility regulations, this study will give more context and new results to increase the available knowledge on the effects of changes in interest rate deductibility regulations. As the four dominant forces point out, the Dutch housing market is very different from other housing markets that have been studied. Combining this with the fact that there is only little research conducted on the effects of a change in interest rate deductibility regulations in The Netherlands makes studying these effects for The Netherlands is all the more relevant.
Hendershott, Pryce and White (2002) research the effects of the gradually phased out interest rate deductibility in the UK. They explain that depending on the country and its economy the phase-‐‑out could have different effects, where the most probable outcome is that it will in a decrease in loan-‐‑to values (LTV’s), higher cost of debt, lower housing consumption by households and a decrease in housing prices. The latter, housing prices, will be the main point of interest in this study. The research question following from this is: “Did the changes in the interest rate deductibility in The Netherlands affect the residential real estate prices?”. If it did; what is the sign and magnitude of this effect? What is the expected long-‐‑run effect of these changes? And did the changes in the interest rate deductibility also affect the number of transactions in the market? This study is the first one to provide empirical results that show the impact of the change in interest rate deductibility in The Netherlands on the residential real estate prices and transactions by analysing panel data with a static pricing model.
In order to investigate the effect of the decrease in the interest rate deductibility on housing prices, this study will use the data of the Dutch Household Survey (DHS) from 2001-‐‑2015 and CBS data on the Dutch economy and housing market. The DHS survey
data contains information of over 2000 households. The combination of both economic fundamentals (GDP, interest rate and inflation) and property characteristics as control variables will form the core of the analysis in this study. The analysis will consist of a static model and will be tested on its robustness by examining a dynamic panel data model in the robustness chapter. Also the effect on the number of transactions will be estimated.
The results of this study gives insight to homeowners on the effect of the changes in interest rate deductibility regulations on the price of their house. As this topic has been a point of much discussion in The Netherlands, the results show that the effects are either not as large as homeowners would have expected, or larger than they had expected. Either way, households benefit from knowing the effect of the change in interest rate deductibility regulations, as it heavily influences their financial situation and retirement planning. Furthermore, the results are of value to the government and other policy makers that want to further restructure the mortgage market. As the change in the interest rate deductibility regulations was loaded with discussion during the last decade, it is interesting for policy makers to know whether these changes cause market turbulence in the short-‐‑run or that they phase down gradually. Furthermore, this study helps these policy makers to accurately measure the potential costs and benefits of new regulations for the government on an ex ante basis. In an international perspective, the results are valuable to other countries that want to change the interest rate deductibility regulations too. Adding the results of a highly regulative market to the more liberal markets that have been studied often (UK and US), gives a broader view on the expected effects on the market.
In chapter 2, the literature on changes in the interest rate deductibility regulations will be discussed. Chapter 3 will provide information on the Dutch mortgage market. Chapter 4 will will discuss the descriptive statistics and summary statistics of the data sample. This will give some insight in the representativeness of the sample for the Dutch market. Chapter 5 will discuss which methods are used in this study. Besides this, chapter 5 will provide prior studies to discuss the relevance of different control variables. In chapter 6, the results chapter, the outcomes of the analyses will carefully be interpreted and put into context. In chapter 7 several robustness analyses will be presented. Chapter 8 will discuss the consequences of the outcomes, their relevance to academic literature, and
their relevance to society. Additionally, recommendations for future research subjects or for improvement of this study are stated in chapter 8.
2. Related Literature
This chapter discusses the existing literature on the effects that changes in the interest rate deductibility regulations can have on housing markets. First an overview of different interest rate deductible markets in the world will be provided. Second, a literature review of the literature on the effects that changes in the interest rate deductibility regulations can have on housing markets will be discussed. Arguments provided by existing literature on how to model and analyse the different relationships on this matter, will be provided in the Methodology chapter.
2.1 Interest Deductible Housing Markets in the World
The Netherlands is not the only country that has an interest rate deductibility system. The UK has had a tax deductibility system in place for mortgage interest, but gradually
phased this out during the latter part of the 20th century. There are many other
economies that still have an interest rate deductibility system in place. For instance, the US, Australia, Canada, Sweden, Norway, Finland, Denmark, Italy, Germany, Austria, Belgium, France and India still have an interest rate deductibility system in place. First, a short background will be provided for the few countries that will be referred to in the literature review of subchapter 2.2 in order to create a context for the information provided later on. Additionally, the countries that have experienced changes in the interest rate deductibility regulations will be briefly highlighted.
United Kingdom
The United Kingdom experienced a long period of interest rate deductibility. The UK income tax was introduced in 1842 in which all household interest payments were deductible (Hills, 1991). During past century the UK phased out their interest rate deductibility system, which was effectuated by two major changes (Hendershott et al., 2002). In 1974 a ceiling was introduced on the size of the mortgages eligible for
deduction. Afterwards the UK decreased the deductibility rate to 0% in four steps from 1993 to 1999.
United States of America
The United States has an interest rate deductibility system in place that contains several limitations. The income tax was introduced in the US in 1894. All forms of interest rate deductibility were possible at that time but were declared unconstitutional soon after. In 1913, however, the interest rate deductibility has been reintroduced (Lowenstein, 2006). Lowenstein (2006) also stated that the interest rate deductibility was not a “stepping-‐‑stone to middle-‐‑class home ownership” as the income entry level was higher than the amount of income which 99% of the inhabitants earned at that time. From 1986 onwards, the regulations were altered. The first regulation states that the deductibles have to exceed the standard deduction in order to reduce tax. Second, deduction is only allowed for principal residences or second homes. Third, interest rate deductibility is only allowed on the first million dollar of debt used for housing (USC, 2016).
Canada
Canada has implemented interest rate deductibility in a different way. Canada always wanted to be of a free trade agreement and postponed the introduction of an income tax for that matter. In 1917, income tax has been introduced in Canada (Brown & Marshall, 2008). The interest rate deductibility in Canada is not allowed for interest on loans secured by a personal residence. Homeowners that use their property as an investment for rental income can deduct the mortgage interest like any other cost as business expense. From 2001 onwards, there is a possibility for private homeowners to deduct their mortgage interest (SCC, 2001). This is possible with an asset swap, where a person sells his investments to buy a house partially or fully, takes out a mortgage and buys back the original investments afterwards.
Australia
Australia does not have an interest rate deductibility system that is widely applicable for homeowners. In Australia interest is deductible either when the property is rented out, or when the property is partly used as a place to do business, meaning a formal business
practice has to be registered and a clearly identifiable part of the property has to be allocated to this professional use (ATO, 2016).
Sweden
In the past, in Sweden the interest on loans was fully deductible at the marginal tax rate of a homeowner. In 1983, Sweden decreased the interest rate deductibility and in 1991 it was capped at 30% (Kochhar & Kramer, 2009).
Norway
In Norway, all interest on loans is deductible from the tax statement (Nordisk eTax, 2016). In the business sector, however, some changes have been introduced on January
1st 2014 (Saastad & Kinden, 2013). The mortgage interest for homeowners is still fully
deductible.
2.2 The Influence of Interest Rate Deductibility on the Housing Market
The effects of changes in credit regulations differ quite a lot throughout the different studies that were published over the past 25 years. This is mostly because many effects are possible and depending on each particular economy some effects dominate others. In this subchapter, an in-‐‑depth literature review will be provided on the different possible effects of changes in interest rate deductibility regulations.
Glaeser & Shapiro (2002) show that interest rate deductibility is a poor instrument to encourage homeownership. In the United States the ownership subsidy (deductibility) moves with the inflation, which varied a lot from 1960 onwards. However, the homeownership rate stayed quite constant during this period. Hilber and Turner (2014) agree on this and add that there is a difference per region, according to the degree of regulation tightness. The tighter the regulations, the stronger and more adverse the effect on the rate of homeownership. However, interest rate deductibility does have a positive effect on the homeownership rate in regions with less regulation and households with higher incomes. Wolswijk (2005) adds to this that there is a significant positive relationship between deregulation and the demand for mortgage debt, as pointed out by his pooled regression model on 15 EU countries. If the demand for mortgage debt is impacted by a change in interest rate deductibility, this could impact housing prices too. Wolswijk (2005) points this out with a Granger-‐‑causality test that
shows no significant evidence to reject the hypothesis that mortgage debt does not have a causal relationship with housing prices.
Earlier research on the subject has taught us that tax reforms have rather little impact on the housing market. Gale (1997), for example, pointed out that the tax reforms in Britain during the second half of the last century have had nearly a noticeable effect on the housing market. Follain & Melamed (1998) estimated the potential increase in tax income for the US. They show that the expected increase in tax revenues is lower than other researchers at that time argued. They estimated that the US tax income would increase with 10 bln $, which is 25% of what others stated. Bruce & Holtz-‐‑Eakin (1999) used a numerical simulation model to estimate the short-‐‑term and long-‐‑term effects of tax reform in the United States. The results do not suggest a large impact. Bruce & Holtz-‐‑ Eakin (1999) argue that this is probably due to their methodology, which embodies forward-‐‑looking household behaviour in the demand for housing.
More recent empirical research by Hendershott et al. (2002) shows that the phase-‐‑out of the interest rate deductibility in the UK had a strong negative impact on the housing market. Hendershott et al. (2002) study the effects of the interest rate deductibility phase-‐‑out in the UK and the tax penalty that was introduced in order to prevent households from mortgage prepayment. They explain different scenarios that could happen when the interest rate deductibility is reduced, of which a combination of housing price decline and decreased demand for housing are most probable. Hendershott et al. (2002) find that the LTV’s in the UK would decline with approximately 30% in case the partial interest rate deductibility would be abolished for new mortgages. They expect that the older households with older mortgages will be the most volatile towards the new policy, and will pay down their mortgage as they generally have higher incomes and therefore experience a more severe impact by the regulation due to their higher income tax bracket. This finding is in contrast with earlier research presented in this chapter. Housing finance can severely change which has a significant impact on the market. However, these changes lack behind of what was estimated by Hendershott et al. (2002). The UK implemented tax penalties on early prepayments to slow down this effect. Hendershott et al. (2002) also explain an intuitive calculation to estimate the impact of the LTV change on housing consumption, house prices and the homeownership rate. They do this with the weighted average cost of
capital (WACC). This calculation model will be translated for the Dutch market and will be used in the results chapter to compare the results of this research with the potential estimated impact the change in interest rate deductibility regulations could have on housing prices in The Netherlands. Concluding from Hendershott et al. (2002) there is serious reason to believe that interest rate deductibility regulation changes can heavily impact housing prices, consumption and the ownership rate. In their opinion the change in LTV’s has a buffer effect in this case.
Damen, Vastmans & Buyst (2014) show with their alternative and intuitive model to what extent the ability to pay a mortgage is related to housing prices. They correct for interest rate deductibility and different payoff schemes for different mortgage types. They tested the model on 8 OECD countries and nearly all markets have shown significant cointegration between the two. As a conclusion, Damen et al. (2014), show that housing prices are highly influenced (elasticity of about 1) by the ability to pay the mortgage on a house in the long-‐‑run. For this, it can be expected that a decrease in interest rate deductibility decreases the ability to pay a mortgage in the long-‐‑run and thus decreases housing prices. Sommer, Sullivan & Verbrugge (2013) add to this that lower interest rates, relaxed lending standards and higher incomes in the US account for half the increase in housing prices between 1995 and 2006.
Ortalo-‐‑Magné and Rady (2004) describe a theoretical view on the determinants of changes in the number of transactions. They state that fluctuations in transactions are primarily caused by demand. Andrew and Meen (2003) studied microeconomic panel data to analyse the effects of changes in income distribution on a household level. Their study adds to Ortalo-‐‑Magné and Rady (2004) that especially changes in the income distribution away from young households decreases the demand for housing and thus the number of transactions. To place this in the context of a change in interest rate deductibility regulations, the earlier discussed expectation of Hendershott et al. (2002) is that a decrease in the interest rate deductibility decreases the demand for housing. According to the theory of Ortalo-‐‑Magné and Rady (2004) this decreases the number of transactions in the housing market. Damen et al. (2014) furthermore state that a decrease in the households’ ability to pay their mortgage, decreases demand. According to the theory of Ortalo-‐‑Magné and Rady (2004) this again means that the number of transactions on the market is expected to decline. According to Andrew and Meen
(2003), the effect on demand will be larger if the ability to pay of young households is decreased.
Concluding from the literature discussed in this subchapter, there is significant evidence in existing literature that changing the interest rate deductibility can have a divergent impact on the housing market. According to Hendershott et al. (2002) a combination of a decline in housing prices and a decrease in demand for housing are the most probable. This research adds significant value to this subject by estimating actual results of the change in interest rate deductibility regulations on housing prices for The Netherlands. Evidence on this matter is still scarce and is not analysed for The Netherlands on an ex post basis before. The next chapter will broaden the background on the Dutch mortgage market and translates the literature presented in this chapter to the context of the Dutch housing market. The literature has shown that many outcomes are in fact possible and it depends on the specifics of different economies and markets what effects will materialize. In order to derive hypotheses that explain expected outcomes of regulation changes for the Dutch market, the next chapter will combine both background and existing literature.
3. The Dutch Mortgage Market
According to the Dutch association of banks (NVB, 2014), the Dutch housing market is shaped by four strong forces; the interest rate deductibility, a high share of non-‐‑profit social housing institutions in the rental market, strong tenant protection and rent control, and a restrictive regulatory zoning regime for construction and development. Partly caused by these shaping forces, the Dutch residential market contains one of the highest debt shares worldwide. The following subchapters will carefully explain the Dutch housing market by these four forces. This will provide the background information necessary to place the literature on interest rate deductibility regulations in the context of the Dutch housing market.
3.1 Interest Rate Deductibility
The Netherlands has had a firm income support for homeowners through the interest rate deductibility. It started in 1893, when the first income tax came into force. At first,
the potential rent one could get when owning a house was added to the income statement, and the costs were all deductible from this income statement, including interest on a mortgage loan. Little change have been made throughout the years, but the interest, since then, has always been fully deductible (Kromhout & Oving, 2008). It was not a tool to support homeownership until after the second World War. During the rebuilding period only 28% of the housing stock was owner-‐‑occupied. By that time, mortgagors required buyers to pay at least 33% of the home with their household equity (Kromhout et al., 2008). In 1956 the National Mortgage Guarantee was introduced. This caused the mortgagors to drop the requirements as the government covered the potential losses mortgagors could incur. Since then, the owner-‐‑occupied market grew to 56% in 2013 (NVB, 2014). It grew bigger than the rental market in the year 1997 (Kromhout et al., 2008). The NVB (2014) agrees that the interest rate deductibility has been an important driver for homeownership growth. However, the NVB (2014) also states that other factors like enduring undersupply, increasing incomes, decreasing interest rates and households being able to get mortgages based on two incomes instead of one, also caused the homeownership rate to increase. This raises the question whether the interest rate deductibility really had a significant influence on the homeownership growth. In fact, The Netherlands still has a lower owner-‐‑occupancy rate than the average in the Eurozone (NVB, 2014). This is also due to the fact that The Netherlands has an extensive pension fund system. Because of this system, there is less need for households to make pension savings by for example buying a house.
Around the year 2000, the debate on interest rate deductibility started to rise. According to Rouwendal (2007), the so-‐‑called lost income tax revenue was less than 6 billion euros in 1995 and nearly doubled since then, while the imputed rent tax remained about 2 billion euros. Rouwendal (2007) puts this in context by stating that the interest rate fell substantially over those years, which implies that interest payments should have decreased during those years. However, due to a constant price growth during this period and the increasing homeownership rate, debt also increased, which most probably caused the interest payments to rise.
In 2001, the first tightening regulation on the mortgage market was enacted. From 2001 onwards, people are only allowed to deduct tax for a maximum of 30 years. This also applied to new mortgages. The already existing mortgages were declared deductible for
30 years from that point in time. Close to the start of the credit crisis, a discussion arose about the interest rate deductibility. Rouwendal (2007) explains the arguments pro and con interest rate deductibility in The Netherlands. He explains that the most commonly used argument pro interest rate deductibility is the increase in homeownership rate. He argues that due to the pressure this system puts on the income tax base, it is likely that the interest rate deductibility must be limited or abandoned.
The consequences of limiting or abandoning can be quite severe, as stated in chapter 2.2. The Netherlands has seen a constant price growth from 1981 to 2008, according to the NVB (2014), who researched the CBS data of those years. Following the conclusions from Hendershott et al. (2002), prices can be heavily affected by a change in the interest rate deductibility regulations. According to Hilber et al. (2014), it depends on the degree of regulation tightness how severe this effect will be.
Due to the high rate of deductibility the debt share in The Netherlands is very high. According to the DHS data used in this research, the average initial LTV on houses in The Netherlands was 93.9% before the interest rate deductibility regulations changed. From 2013 onwards this is 88.0%, which is much higher than the euro zone average. However, the losses incurred on mortgages are one of the lowest in the euro zone (NVB, 2014). This is partly due to the monotonous housing price growth for nearly three decades, which is no problem when prices are still rising, but starts to incur losses when prices are dropping. This is most probably the reason why the loss rate grew during and after the crisis years.
3.2 The Non-‐‑Profit Social Housing Institutions and the Strong Tenant
Protection & Rental Control
The Dutch rental housing market is characterized by its social housing system. About one out of every three houses is owned by a housing corporation (NVB, 2014). The NVB (2014) states that these corporations are non-‐‑profit institutions which are highly regulated by the government. While housing prices in The Netherlands boomed over the last three decades, the rent controls ensured that the renting segment stayed relatively affordable (NVB, 2014). According to the NVB (2014), it is due to this strongly controlled rental segment, that a buy-‐‑to-‐‑let segment is one that hardly exists in The Netherlands. Because of this, housing prices can not be influenced much by investors’ valuation
calculations. The last decade, the market was liberalized partly by new housing regulations (NVB, 2014). However, last decade has also shown an increased amount of fraud incidents in the housing institutions, for which multiple sets of regulations were enacted in order to increase the governmental supervision on these social housing institutions.
3.3 Restrictive Regulatory Zoning Regime for Construction & Development
The Dutch housing market stock consists of about 56% of owner-‐‑occupied houses. This is below average for the euro-‐‑zone according to NVB (2014), which is partly due to a permanent undersupply, which is caused by restrictive zoning regulations that cause the new stock added each year to be less than 1 % according to NVB (2014). Municipalities have the ability to strongly control the cities in The Netherlands with these zoning regimes. The negative result is the permanent undersupply. The upside, however, is that the government can potentially control neighbourhood developments and intervene in case neighbourhoods are evolving into deprived areas. To give a recent example, it is most probably for this matter that the municipality of Amsterdam made a deal with AirBnB that houses cannot be put for rent more than 60 days a year (Milikowski, 2016). The municipality would like to control which real estate function dominates which area of the city.
3.4 Mortgage types
The Dutch residential mortgage market consists of about ten types of mortgages, of which the mortgage without repayment, the improved life insurance and the annuity mortgage are the vast majority. Until January 2013 the mortgage without repayment was fully deductible from the income tax statement, like all others. Table 3.1 shows the different mortgages that are active in the Dutch housing market. This table clearly shows the impact of this regulation on the household preferences with regard to mortgage types. From January 2013 onwards, the mortgage without repayment (interest only) decreased in market share from 44.5% to 31.3%.
-‐‑TABLE 3.1-‐‑
MORTGAGETYPES IN THE SAMPLE: MARKETSHARES OF THE DIFFERENT MORTGAGE TYPES USED IN THE NETHERLANDS
These tables show the different mortgages that are active in the Dutch housing market.
The difference between table 1 and 2 is that table 1 contains main (1st) mortgages on
houses bought before 2013 and table 2 contains main (1st) mortgages on houses bought
from January 1st 2013 onwards. Specifically, the first table shows the market for
mortgage types before the interest rate deductibility regulations changed, and the second table shows the market for mortgage types after the interest rate deductibility
regulations changed. The 2nd column shows the frequency of the mortgage in the dataset
used for this research and the 3rd column shows the market share calculated with regard
to this data. Note that the frequency is noted in number of observations in the sample and the market share is noted in percentages.
Mortgage Types Frequency Market share
Annuity mortgage 718 8.33
Traditional life-‐‑insurance mortgage 480 5.57
Improved traditional life-‐‑insurance mortgage 2 23.20
Linear mortgage 146 1.69
Endowment mortgage 82 0.95
Investment mortgage 780 9.05
Mortgage without repayment 3,836 44.49
Mortgage with life-‐‑annuity construction 77 0.89
Lifelong mortgage with death insurance 84 0.97
Bank saving mortgage 281 3.26
Other mortgage 138 1.60
Total 8,622 100.00
Mortgage Types Frequency Market share
Annuity mortgage 427 13.30
Traditional life-‐‑insurance mortgage 259 8.07
Improved traditional life-‐‑insurance mortgage 857 26.69
Linear mortgage 80 2.49
Endowment mortgage 42 1.31
Investment mortgage 368 11.46
Mortgage without repayment 1,005 31.30
Lifelong mortgage with death insurance 25 0.78
Bank saving mortgage 103 3.21
Other mortgage 5 0.16
Total 3,211 100.00
3.5 Hypotheses
Concluding from all literature presented in chapter 2 and 3, hypotheses can be stated for the expected outcomes of this study. Hendershott et al. (2002) state that a change in interest rate deductibility regulations can impact the housing market in different manners, where a combination of a decrease in prices and demand for housing are the most probable. Hilber et al. (2014) adds to this that the negative is expected to be more severe in an economy with a high degree of regulation tightness. Furthermore, Damen et al. (2014) state that housing prices are highly influenced by the ability of a household to pay the mortgage obligations. Sommer et al. (2013) and Wolswijk (2005) add to this that relaxed lending standards cause housing prices to increase. Gale (1997) however, argues the opposite of Hendershott et al. (2002) by stating that the tax reforms in Brittain had nearly no noticeable effect on the housing market. Furthermore, Bruce et al. (1999) state that interest rate deductibility changes can have smaller effect than Hendershott et al. (2002) and Damen et al. (2014) argue, as the effects are usually implemented gradually, or households preventively adjust to the potential future situation, which would partially nullify the effect. Although not all research is aligned on this matter, the majority of the most recent studies argues and shows that tightening interest rate deductibility regulations will cause housing prices to fall. Concluding from these studies, the following hypothesis can be drawn:
Hypothesis 1: the expected effect of the change in interest rate deductibility regulations in The Netherlands will have a negative effect on housing prices in the Dutch market. Hendershott et al. (2002) also state that the negative effect on housing prices goes hand in hand with a decrease in demand for owner occupied housing. This aligns Damen et al. (2014) as they argue that the demand for owner occupied housing is highly influenced by the ability to pay a mortgage for households, which is directly influenced by the interest rate deductibility. Following Ortalo-‐‑Magné and Rady (2004) and Andrew and Meen (2003), this decrease in the demand for housing will decrease the number of
transactions in the market. Concluding from these studies, the following hypothesis can be drawn:
Hypothesis 2: the expected effect of the change in interest rate deductibility regulations in The Netherlands will have a negative effect on the amount of transactions on the housing market.
4. Data & Descriptive Statistics
In order to derive a housing price model that controls for as many characteristics and determinants of housing prices as possible, a dataset containing information on as many housing characteristics and other determinants of housing prices as possible, is needed. In this study, data from De Nederlandsche Bank: the Dutch Household Survey (DHS) will be used. This is an extensive panel dataset, provided by Centerdata, containing data on more than 2000 households over the period 2001 to 2015. The data contains information on purchase prices, house types, building period, size of living room, size of garden, number of rooms, garage, national mortgage guarantee, initial mortgage amount, mortgage types, initial LTV, initial mortgage amount, current outstanding mortgage balance, current LTV, urbanization degree and provinces. These variables are either included or deducible from the information present in the dataset. Unfortunately, the DHS data do not contain actual price data. As the data tracks households over time, there are not enough transactions in the sample. An other way to derive prices for existing houses over the years is used. The data from the Central Bureau of Statistics (CBS) keeps track of the Dutch economy and statistical data. Among these data are price indices on provincial levels from 1995 onwards, which make it possible to index all purchase prices of the DHS data with a purchase date from January 1995 or more recently. CBS also has data on long-‐‑term interest rates, inflation, GDP and housing supply. This information will be merged into the DHS dataset in order to include the macroeconomic characteristics in the models, or will be used to estimate the macroeconomic model for transactions. The dataset contains approximately 3500 observations between 2001 and 2015 for which all this information is known, except for the housing supply. The housing supply only contains data from 2012 onwards. The dataset used to estimate the effect of the change in interest rate deductibility regulations
on housing prices will be the yearly data from 2001 till 2015. The dataset used to estimate the effect of the change in interest rate deductibility regulations on the number of transactions will be the monthly data from January 1995 till May 2016.
The following subchapter will explain how the data is prepared for the analysis. It will show how the data is altered, how outliers are dealt with, and how new variables are generated. Chapter 4.2. will provide the summary statistics and interprets the values presented. Note that the numbers discussed in this chapter are rounded numbers.
4.1 Data Preparations
Alterations
First the different yearly panel sets on five different subjects have been appended to one another in order to create yearly panel sets that include all information needed from the DHS survey. Subsequently, these yearly panel sets have been merged into one set with the CBS data.
In order to obtain meaningful results in the analysis phase, the data has been cleaned. Specifically, all variables needed for either a regression or for creating a new variable have been checked on consistency throughout the years, fill in errors, and reasonable and / or possible observations. An example of the consistency is the ‘I don’t know’ answer that has a different numeric code throughout the years. In a few years, this also differs per variable. An other example of the consistency is the difference in the codes corresponding to provinces in The Netherlands. Along the sample period these codes changed. All ‘I don’t know’ answers have been changed to ‘missing’. The provincial codes have been corrected. Furthermore, the data also contained clear fill in errors. For example, the survey asks for the purchase price of a house in thousands of euros. Some observations showed to be between 5,000 and 2,500,000 euros. These have been divided by 1,000 in order to fit the data as they were meant. This has only been done for
observations that were clearly incorrect.1
1 For example, debt to income ratio can show whether the amount of mortgage debt was a clear fill in