(Re)Opening the door to the Dutch housing market for first-time homeowners:

(1)

(Re)Opening the door to the Dutch housing

market for first-time homeowners:

Estimating the costs of a Dutch housing saving scheme

Author: Jorus van Erkel

Supervisor: prof. dr. M.A. Allers

MSc Economics

University of Groningen

Faculty of Economic and Business

January 2013

MSc Economics

Abstract:

We briefly discuss the current state of the Dutch housing market and investigate the root of the current downturn in the housing market. We argue that a housing saving scheme can stimulate the Dutch housing market and increase the housing market accessibility. By extending Yalta’s theoretical model, we have shown that a housing saving scheme increases housing market accessibility. We then estimate the costs of a housing saving scheme for the government, in terms of forgone tax income, using micro data on Dutch households from the WoON(2009) dataset.

JEL: D61, H22, R22

(2)

1. Introduction

In the wake of the housing market crash in America, questions have been raised about the stability of the Dutch housing market. In the World Economic Outlook of 2008 the IMF dedicated a chapter to the housing market developments in advanced economies. One of the major insights was an estimate that the Dutch housing market was overvalued by 30%. In addition, as reported by the bulletin from the Dutch central bank of the 6th of December 2011, the total amount of mortgage debt of Dutch households stands at €644 billion in mid-2011. Consequently, the total amount of mortgage debt amounts to approximately 110% of the Dutch gross

domestic product. In addition, the Dutch housing market is facing a prolonged

downturn, caused by an accumulation of various factors. In short, the housing market downturn is caused by the combination of the economic recession, uncertainty about the future of mortgage loan interest deductibility and stricter mortgage loan

regulation.

The Dutch tax system heavily favors financing homeownership through mortgage loans because mortgage loan interest payments are tax deductible. During the 1990’s through to 2008 house prices experienced a continuous appreciation and during the same period banks became more generous with mortgage loans. The ratios of the amount of mortgage debt in relation to the house value (Loan-to-Value (LTV)) and the amount of mortgage debt in relation to income (Loan-to-Income (LTI)) increased significantly during this time period. It was not uncommon to receive a mortgage loan the size of six times one’s gross yearly income and/or of more than 110% of the house value. In the light of the crash of the American housing market and these trends on the ratios for LTV and LTI, the Dutch mortgage loans were no longer deemed sustainable. In order to reduce both the systemic risks as well as the risks for individual households resulting from the accumulation of mortgage loans, a series of policy changes have been implemented by the Dutch Government.

As of August 2007 the Dutch Authority for Financial Markets (AFM) and the Dutch Banking Association agreed upon a code of conduct, the ‘Gedragscode

(4)

interest rate deductibility will be limited, more on this in the institutional context section.

Although we acknowledge that the restrictions on the LTV and LTI ratios are prudent and necessary to ensure long term stability and sustainability of the housing market, they come with some side effects. Any reduction in mortgage loan interest tax deduction will, at least in the transition period, lead to an increase in the net cost of housing. This, in combination with stricter mortgage loan regulation, will decrease the accessibility of the housing market, particularly for young and low to middle income households. This has a knock-on effect on the market as a whole. The

previously mentioned households form the majority of the first-time homebuyers and provide fresh blood to the market. In contrast to the current homeowners, (potential) first-time homebuyers do not have to worry about selling their current residence. Problems now arise as a significant group of potential first-time homebuyers face difficulties to finance any house purchase, which leads to a stagnation of the in-flow. This influences current homeowners because when they want to move, they face a higher risk of an overlap in having bought a new house and having to sell the old house. This results in extra costs as the household would have to finance two mortgage loans. Facing the risk of these extra costs, the dynamic arises where it is optimal for individual homeowners to postpone buying a new house until they have sold their old one, which leads to a stagnation of the housing market.

In addition, the housing price decline creates a lock-in effect for some

households. This lock-in effect describes the decrease in residential mobility resulting from a mortgage loan exceeding the value of the house. Schilder and Conijn (2012) found that approximately 500.000 households are ‘underwater’, in other words, have a higher mortgage debt than the value of their house. These households would have to finance residual debt on top of the mortgage of a new house when moving. It is likely that these households exhibit a lower residential mobility.

(5)

who shows a reduction in household mobility due to falling house prices. On the other hand, Engelhardt (2003) and Molloy et al. (2010) show that falling house prices do not constrain mobility. However, it is likely that mobility is lower for households in The Netherlands compared to their international counterparts. In the same study as mentioned earlier, Schilder and Conijn (2012) surveyed mortgage advisors on the current mortgage loan market. Approximately a third of the respondents indicated that they expect that households with a negative housing equity are unable to refinance their mortgage loan to purchase a new house. In addition, two thirds of the

respondents indicated that a residual debt of €5.000 could be financed in a new mortgage. In other words, it would be difficult for current homeowners who have a mortgage debt exceeding the value of their home to finance the purchase of a new home. As of January 2013, the residual debt is eligible for mortgage interest rate deductibility for 10 years. Although the financial loss resulting from residual debt may have decreased due to this new regulation, household may still experience psychological barriers for residential mobility.

Due to stricter regulation on the Loan-to-Value ratio, a significant group of households will no longer be able to fully finance their house purchase (including purchasing costs) through mortgage loans. In addition, a specific problem arises for the young and low to middle income households. These households face significantly stricter regulation on the amount they can borrow in relation to their income. As a result, these households face a gap between the price of their desired house and the maximum allowed mortgage loan. For some households, this gap may be large

enough for them to refrain from purchasing a house altogether or choose to settle for a smaller house. This decreases the welfare of some households as they might have been willing and able to pay more for a house. Yet, due to the financial restrictions, they face lower (housing) consumption possibilities. Those households who still plan to buy a house have to provide some of the financing themselves, which would most likely be through savings. The welfare loss resulting from the financial regulation could warrant a government intervention.

In general, the government may feel the need to stimulate homeownership for a number of reasons: due to market externalities which lead to a non-optimal

(6)

government stimulates homeownership is through the tax deductibility of mortgage loan interest. This heavily incentivizes borrowing to finance a house purchase and favors mortgage loans with little or no amortization.

In order to mitigate the negative impact of recent housing market regulation, in terms of stricter LTV and LTI ratios, we investigate the policy option of fiscally stimulating saving for housing purchases. The main goal would be to increase the housing market accessibility for low to middle income households. In addition, these households can stabilize the housing market by increasing demand for the lower segments which can lead to a domino-effect of transactions. This domino-effect can reduce the burden for current homeowners of selling their house.

This paper will focus on housing market accessibility and specifically on the decision on whether to purchase a house for (potential) first-time homeowners. The aim of this paper is to provide estimates of the costs of these savings schemes for varying scenarios and objectives.

(7)

2. Institutional context

In this section we provide background information on the Dutch housing market and the Dutch tax system.

2.1 Housing supply elasticity

The Dutch housing market is characterized by a relatively inelastic supply by international comparison (OECD, (2011)). As discussed by Vermeulen and

Rouwendaal (2007), housing prices in The Netherlands hardly influence the supply of houses. This can be explained by the combination of a general shortage of land, the high population density and the strict zoning laws in the Netherlands. In contrast, the demand for housing has increased during the last few decennia through a steady population growth, in combination with an increase of the number of both single person households and dual-earner households.

2.2 Income tax

In the Netherlands, income is taxed differently depending on the source of income. The Dutch tax system distinguishes between income from ‘work and housing’ (Box 1), ‘substantial business interest’ (Box 2) and ‘savings and investments’ (Box3). For income originating from either ‘substantial business

interest’ or ‘savings and investments’, tax is levied at a flat rate. The income from box 3 is based on a notional yield of 4%, which then is taxed at a flat rate of 30%. There is a tax-free allowance of €21.139 (in 2012) for Box 3. In contrast, income from Box 1 faces a progressive tax rate as presented in table 1.

Table 1: Marginal Tax Rate for

income from Work and Housing

Taxable Income Percentage Until € 18.945 33,10% from € 18.946 to € 33.863 41,95% from € 33.864 to € 56.491 42%

(8)

2.3 Mortgage interest deductibility

The Dutch government intervenes in the housing market in a number of ways. The most important government intervention is that the interest paid on mortgage loans is deductible from taxable income in Box 1. This, in combination with a relatively high marginal tax rate, creates a significant incentive to borrow for the purchase of a house. As of January 2013, regulation regarding mortgage rate

deductibility has been changed. The maximum marginal tax rate at which the interest can be deducted will be reduced from the highest marginal tax rate (52% in 2012) to the second highest marginal tax rate (42%). In 2013, this reduction will be

implemented gradually. Each year the maximum marginal tax rate at which the interest can be deducted from will be lowered with 0,5% each year. Furthermore, new mortgages will have to be repaid within 30 year in order to be eligible for mortgage interest rate deductibility. As a result, among others, interest-only mortgage are no longer eligible for interest rate deductibility.

2.4 House transactions Tax

In addition, a tax is levied on house transactions. On the 1st of July 2011 the transaction tax on houses in the Netherlands was temporarily lowered from 6% to 2% of the purchase price. As of the 1st of July 2012 the reduction in transaction taxes was made permanent.

2.5 Code of conduct for mortgage financing

(9)

Consequently, the purchasing costs including transaction taxes, are currently roughly 6% of the house value.

Nonetheless, much was still left to the discretion of the banks through a comply or explain clause. Banks were allowed to deviate from the imposed limits on the LTV and LTI, however they would have to clarify why they deviated from the limits. In 2009, the Dutch Authority for the Financial Markets (AFM), in cooperation with the Dutch National Bank, conducted a study on the implementation of the code of conduct. The results revealed that from a sample of mortgage loans between 25% and 33% of new mortgages were provided under the comply and explain clause. From these cases 65% were deemed irresponsible deviations from the LTV and LTI ratio limits by the AFM. Based on these results, the AFM concluded that self-regulation of the mortgage loans by banks had failed and that stricter regulation was necessary. As of August of 2011 a more stricter version of the code of conduct was put into place, with the most notable change being a more clear definition for the cases eligible for the comply or explain clause. In addition, the AFM would fine any irresponsible deviation from the rules stated in the code of conduct.

2.6 Various (local) taxes related to housing

In the Netherlands, a number of taxes are related to homeownership. First of all, homeowners pay income tax in Box 1 over their notional rental income, which is a percentage of the estimated value. This is a ‘progressive tax’ in the sense that a higher percentage of the housing value is taxed for higher estimated house values, as is shown in table 2. The value is based on the annual assessment done by the

municipalities (known as the ‘WOZ-waarde’). Homeowners also pay property taxes based on these estimated values to the municipalities as well as water boards. Both of these tax rates are determined locally.

(10)

2.7 National Mortgage Guarantee

In addition, mortgage loans in the Netherlands are insured through the

National Mortgage Guarantee (Nationale Hypotheek Garantie (NHG)) if they fulfill a number of criteria. In principal, the guarantee is covered by a fund which is financed by the debtors by a one-off payment of 0.7% of the mortgage value. However, in the end the fund is backed by the government. This fund covers the mortgage payments in case of divorce, worker disability, unemployment and/or death of the partner. Until the 1st of July 2009 the fund would guarantee mortgages on houses with values up to € 265.000, however, in light with the economic recession, the ceiling had been raised temporarily to € 350.000 to increase confidence in the Dutch housing market. As of the 1st of July 2012 the guarantee ceiling has been lowered to € 320.000. After which it will be decreased in yearly steps to € 290.000 and € 265.000 respectively.

2.8 Starter loans

In the Netherlands municipalities try to bridge the financing gap between the maximum mortgage loan and the house price. Approximately 200 of the 415

municipalities offer the possibility of ‘starter loans’, which are loans specifically designed for first time homeowners. These loans are designed to function as a secondary loan in addition to the mortgage loan. These loans can increase the maximum amount a household can borrow due to favourable payment conditions. Although size, terms and conditions of such a loan vary between municipalities, they do have a number of common characteristics. For the first three years the starters loan is interest- and amortization-free, maturity of the mortgage loan is a maximum of 30 years and the house purchase has to be guaranteed by the National Mortgage

(11)

2.9 Rental Market

The Dutch government also strongly influences the Dutch rental-housing market. For example, low-income households receive rental subsidies. Single person households may receive rent subsidies if their annual income is below €20.600. Households consisting of multiple persons may apply for these rent subsidies with a maximum income of €27.950. In addition, public housing corporations provide social housing for below market rent, primarily for low income households. 90% of these houses have to be allotted to households with a maximum income of €34.085.

Now that we have discussed the context of the Dutch housing market, we will continue with a discussion of the relevant literature.

3. Literature Review

This section will provide an overview of the relevant literature. We will focus on the reasoning behind imposing LTV-ratio limits, the discussion on the benefits and externalities associated with homeownership and we will discuss literature on the housing consumption decision in relation to financial restrictions. We focus on these subjects because they provide a better understanding of why government intervention on the housing and mortgage market would be warranted

3.1 LTV-ratios

(12)

estate market developments, they find that reductions in LTV limits were followed by significant drops in the house price appreciation rate.

A study by the IMF (2011), based on data from 19 advanced OECD countries during the time period 1980 until 2010, shows that countries with high LTV ratios, tend to have deeper downturns,. This leads to a macroeconomic risk in view of the fact that Case et al. (2005) show that the wealth effect of changes in house prices is much larger than a similar shock in other asset prices. This is in line with the notion of Borio and Shim (2007) and Borio, Furfine and Lowe (2001) that LTV ratios are pro-cyclical as lenders tend to relax LTV ratios in good times in response to competitive pressures and perceptions of declining risk.

In summary, there is empirical evidence suggesting that high LTV ratios have a causal link with housing price volatility and the depth of economic downturns. Consequently, it seems prudent to restrict the LTV ratio of mortgage loans, as this limits macroeconomic risks, the financial risk for individuals and for the financial market as a whole.

3.2 Benefits Homeownership

There exists a wide ranging literature on the subject of the social and financial benefits of homeownership. First of all, homeownership has been linked to increased self-esteem, a better quality home environment and fewer child behavior problems among home owning parents by Haurin et al. (2002). In addition Gatzlaff et al. (1998) report a positive relationship between homeownership and dwelling quality. Other studies link homeownership to better health (Robert and House 1996), greater participation in social networks (Rohe and Stegman 1995), success in the labor market (Fisher and Coulson 2002) better educational outcomes for children (Boehm and Schlottmann 1999) and increased political participation (Gilderbloom and Markham 1995).

In addition, homeownership may result in external effects. In the context of a review on the welfare benefits of home ownership, van Ewijk et al. (2006) and CPB (2010) discuss the externality effects of homeownership. They report that

(13)

Rosen (1984) and Glaeser, Shapiro (2002) and CPB (2010) conclude that these

positive effects are relatively small. In addition, both CPB (2010) and van Ewijk et al. (2006) discuss the negative external effect that stimulating homeownership might have through increased congestion. In addition, the CPB notes that homeownership may result in a decrease in residential mobility, although, according to Van Vuuren and Van Leuvensteijn (2007), this can be attributed to the transaction tax.

Summaries on the extensive homeownership externality literature are provided by Rosen (1985), Rohe et al. (2002) and Dietz and Haurin(2003). In addition, Dietz and Haurin discuss the proper methodology given the endogenous nature of

homeownership as a variable, as addressed by DiPasquale and Glaeser (1999).

Even though homeownership is correlated with various positive factors, at least some of these relations may be attributed to heterogeneity and may not pose a causal link, as discussed by Dietz and Haurin. Furthermore, the positive externality effects of homeownership appear to be relatively small. In conclusion,

homeownership would not warrant any government intervention based on market failure arguments.

3.3 Housing/consumption decision

A large number of studies have been done on the housing decision of

households. Yoshikaw and Ohtaka (1989) studied the relation between land prices and the savings of prospective homeowners. For their study they used Japanese cross-sectional data and the results show that higher prices result in some potential buyers to become permanent renters, while those who keep house purchasing plans have higher saving rates.

Engelhardt (1994) reports a similar discouraged saver effect in a study in Canada. The empirical results show a marginally significant negative relationship between house prices and asset accumulation. In addition, the propensity to participate in the previously discussed Registered Home Ownership Savings Program(RHOSP) is lower for higher house prices.

(14)

attributed to the methodology, which depended on homeownership rates which do not vary much across cities in the United States.

In addition, the housing decision is influenced by financial regulation. Based on micro data on 30.000 individuals from 14 OECD countries, Chiuria and Jappelli (2003) show that higher down payment ratios force young households to postpone homeownership, thus influencing the owner occupancy rates distribution among age groups.

Ortalo-Magné and Rady (2006) used a life-cycle model of the housing market with a property ladder and a credit constraint and find that credit constraints delay some households’ first home purchase and force other households to buy a smaller home than they would like. They also provide empirical evidence from data from the UK and US to support their findings.

By imposing required down payments or financial restrictions on mortgage loans in general, some individuals are restricted in their consumption possibilities. This can lead to a welfare loss for some individuals. Given the earlier mentioned LTI and LTV ratio limits, this would most likely be young individuals with low to middle incomes who are forced to refrain from or postpone homeownership due to the regulation. Even though the welfare loss is the result of warranted artificial

restrictions, the government may feel inclined to help these households overcome the financial barriers for equality reasons.

4. (Housing) Saving Schemes

(15)

4.1 Dutch savings schemes

In the following section a number of current and former Dutch saving schemes will be discussed with the aim of providing a framework for a possible housing savings scheme. In recent years two savings schemes were available, namely a wage savings scheme (spaarloonregeling) and the life-course savings scheme

(levensloopregeling). In 2013 these two savings schemes will be replaced by the vitality savings scheme (vitaliteitsregeling).

For the wage savings scheme, individuals were allowed to deposit a maximum of €613 per year on a special savings account. The amount saved is deductible from taxable income in Box 1 and exempted from tax from Box 3 under the condition that it is fixed for four years. In addition, the maximum for total savings at any point in time in the wage savings scheme is €2.452. Under some circumstances the money on the savings account may be spent within the mandated four fixed years. One of these exceptions was using the money from the savings account to purchase a house.

The Life-course savings scheme aimed to fiscally stimulate income smoothing. People can yearly save a maximum of 12% of their gross income to a maximum of 210% of gross income. The amount saved can be used solely for unpaid leave. The amount saved can be deducted from taxable income in the relevant year. Upon withdrawal, the savings have to be added to taxable income from work and housing (Box 1). In other words, this savings scheme entails a delay of taxation on income from work and housing. Just like the wage savings scheme, the amount saved in the life-course savings scheme is exempted from taxes over savings in Box 3. At any moment in time an individual could only participate in either the wage savings scheme or the life-course savings scheme.

(16)

A review of the current and historic savings schemes reveals a few common characteristics. For instance, all the discussed savings schemes provide savings tax (Box3) exemption. In addition, the amount of savings is tax deductible from taxable income from work and housing for all saving schemes. However, the savings schemes differ in whether the savings must be added to taxable income upon usage or not and on the flexibility in terms of the goal of the savings. Another important difference lies in the maximum set on the allowed annual and total savings.

The wage savings scheme is the closest to the intended housing savings scheme. However the wage savings scheme sets a relatively low maximum for both the yearly savings and the total savings which makes it inappropriate for the goal of stimulating housing market accessibility.

4.2 Housing savings schemes abroad

In other countries there are a number of comparable savings schemes which are in line with the idea of housings savings. We focus on the savings schemes which encourage saving before the purchase of a house.

One example is the Registered Home Ownership Savings Plan in Canada, which aimed to promote homeownership in Canada and specifically to help young renters to save for home purchase. The Registered Home Ownership Savings Plan (RHOSP) was introduced in 1974 and ended in 1985. Due to a strong increase in real house prices in the 1970s it became harder for many young renters to save for home purchase. Through the RHOSP, renters could establish a special account, which allowed for a tax deduction for the amount saved in a year. The allowed annual tax deduction was capped to $1000 with a total maximum of $10.000 plus earnings from savings. These limits were doubled for married couples and the interest rates earned on these accounts were exempted from tax.

A study by Engelhardt (1997) investigates whether RHOSP increased home ownership. Based on the empirical results Engelhardt concludes that the program increased the annual rate of transition from renting to owning for younger households by 20%.

In Germany, homeownership is encouraged through ‘Bausparkassen’, which encourage households to save for a down payment. The idea is that households

(17)

a predestined total savings amount is reached. Then the household is eligible for a predetermined loan at a low interest rate. Thus in essence, (part of) the funding for the mortgage loan is provided by the future borrowers. In addition the savings are tax deductible or generate a tax credit for low to medium incomes.

4.3 (Dutch) Housing Saving Scheme

This section will discuss the possibilities to fiscally stimulate housing savings, followed by a discussion on the goals for such a savings scheme.

The objective of the housing savings scheme is to increase accessibility of the housing market for the outsiders. It is neither feasible nor realistic to ensure housing market accessibility for all income levels. Therefore, we have to impose a cut-off point at which housing market accessibility is feasible. In the Netherlands people with an income up to €34.000 are eligible for government rental housing support either via social housing or rental subsidies, which would make this a natural cut-off point. Consequently, the outsiders are defined as those individuals with an income

exceeding €34.000 who are not a homeowner. In addition, we define housing market accessibility as being able to purchase an apartment. We chose an apartment as the target type of housing because data from the Cadaster shows that this is the least expensive type of housing. Therefore this type of housing would serve as an ideal starting point on the housing market for first time homeowners. This premise is supported by data from the WoON(2009) survey, which shows that a clear majority of first time homeowners in 2009 owned an apartment. Lastly we would design the housing saving scheme as such that an individual with an income as low as €34.000 is able to purchase an apartment after 10 years of participating in the housing saving scheme.

The housing saving scheme can be customized based on the questions, Who should be able to buy what, and within which timeframe? These are political

questions and this paper merely tries to provide cost estimates for various scenarios.

(18)

savings is deductible from taxable income in Box 1 and that the amount saved in the saving scheme is exempted from savings tax (Box 3).

Now that we have defined the housing saving scheme we will focus on estimating the participation and costs of the housing saving scheme in the following section.

5. Economic Model

This section aims to provide a qualitative analysis of the effects of stricter mortgage loan regulation on the housing decision of potential first time homeowners. For this purpose we develop an economic model in line with the model used in Yalta (2011). In addition the model will be extended with the aim of allowing for

government intervention through tax deduction, in order to study the effect of a housing saving scheme. We will first discuss Yalta’s model combined with our graphical interpretation of the model.

5.1 Yalta model

As mentioned earlier, homeownership is at the very least correlated with numerous benefits. For the model we assume that transitioning from renting to owning a home results in a positive stream of utility. For the model we assume that this positive stream of utility resulting from owning a house, U(H), which is separable from the utility of consumption, U(C), with diminishing marginal utility from both consumption and housing. Here, consumption is defined as the consumption of all goods and services including consumption of housing services out of income. The purchasing costs for housing usually not only consist of the value of the house but also of transaction costs, e.g. closing costs, notary costs and transfer tax. In addition people incur additional costs in terms of moving costs and redecoration costs. Both the transaction costs and additional costs can reasonably be assumed to be

(19)

In this model the required down payment, D, is defined as the amount a household has to pay out of pocket in order to purchase the preferred house. In other words, the required down payment is the difference between the purchasing costs for housing and the maximum amount of borrowing allowed. In some countries there exists a fixed required down payment which can be an absolute amount, but usually is a percentage of the loan value. Other countries impose loan-to-value limits which implicitly results in a required down payment if the purchasing costs exceed this loan-to-value limit.

In accordance with the required down payment, households have to give up present consumption, C, to save for a certain duration of time in order to purchase a house. As mentioned earlier, consumption in this model includes housing expenses. In other words we assume that both renting and owning a house feature the same

recurring costs. Consequently, there exists a trade-off for the timing of the housing purchase against the forgone consumption. The projected time of purchase,τ _,

depends on the amount saved each period, S, which is the part of income, Y, which is not consumed, and the size of the required down payment. Income is taken to be constant over time and we assume that the real interest rate earned on savings is zero. As a result the projected time of purchase is:

(1)

=

Households are assumed to live forever, thus the total utility gain from housing purchase is an infinite stream of utility starting atτ _.

Households make a cost-benefit analysis, weighing the utility loss of saving to buy a house againt the utility gain from owning a house. By dividing utility gain from homeownership, u(H), by the total utility loss as a result of the forgone consumption for duration τ, u(Y)-u(C), and accounting for differences in time preference,

, we obtain the relative utility gain function:

(2)

=

∗

(20)

Where β is the rate of time preference bound by 0<β<1. When g>1 utility gains from owning a house outweigh the costs of the forgone consumption in order to buy the house. In case g ≤ 1, households will not consider purchasing a house and will continue renting instead. Following equation (2) there exists an optimal level of consumption (and thus saving), which gives the maximum return obtainable from this saving scheme.

Taking the derivative with respect to consumption, combined with the constraints 0<C<Y and u(H )≠ 0 results in the first order condition:

(3)

1 −

!

"

# = $!% − !#&′

In line with Yalta (2011), we assume a Constant Absolute Risk Aversion (CARA) utility function:

(4)

!# = −

() _*

, 0 < , < 1,

Where α is the coefficient of absolute risk aversion. Using CARA preferences we can find max[g] from the optimum rate of consumption. Combining equations 3 and 4 yields:

(5)

1 −

=

_*

1 −

*

′

Using the Taylor expansion - = ∑ -/

0! 2

034 and further derivation result in an

optimum rate of saving as presented in equation 6. For the step-by-step derivation we refer to Yalta (2011).

(6)

5 = 67 8

_*

9

(21)

time preference and the degree of risk aversion. As a result of credit regulations resulting in a required down payment a household faces the choice on whether to save (and thus giving up part of their consumption) or continue renting.

With the aim of qualitatively studying the effect of an increase in the required down payment the model is summarized as in table 3. As presented in equation (2), each household determines whether the benefits of homeownership exceed the utility loss of forgone consumption. When the benefits exceed the costs, households will save according to the optimum savings rate as presented in equation (6). Households save the same amount because income is fixed and the optimum savings rate does not depend on variables that change over time. Consequently the time of purchase, τ, is defined as in equation (1). The required down payment in this model is defined as a percentage, θ, of the Housing value. For simplicity we assume that there exist no additional costs of moving which have to be included in the mortgage loan. Hence the amount to be borrowed is equal to the housing price.

Table 3: Summary Yalta’s model

(I) :_; = :< (II) =_; = > ∙ :_; (III) > = 6@A B (IV) C_; = :_;− =_; (V) D = D< (VI) @ = E ∙ D (VII) F = @/= (VIII) H=_; = ∑ =;_;3I _; %J = KLMNO PJ= PQRSL T 5 = NUVSO!O WQV NX TQRSL , = QY!TZNV WST[ QRWTSNL \ = WQV NX VSO UWXWLM #J = #NLT!OUVSNL 7 = ]^!SW_ 7N`L aQbOLV c = cN!TUWSM = dSO NX a!WMℎQT dP = dNVQZ PQRSL T

(22)

Figure 2 shows the effect of an increase in the required down payment.

A household facing tighter loan regulation may react in a number of ways, these are denoted by the points A-D, which represent archetypes of households. According to the household’s original plan, it would be able to purchase house H with the optimal saving rate resulting in 7 at . When confronted with a higher required

Time

τ

dPJ 7 Total Savings, Required Down Payment

Figure 1: Yalta’s Model: Basic Situation

Time Total Savings, Required Down Payment

τ

dPJf dPJg

τ*

B C D A 7g 7f

(23)

down payment, the household alters his savings rate as shown in equation (6).

Depending on a household’s absolute risk aversion (α) and rate of time preference (β), it will end up somewhere along the line between points B and C with a corresponding time of purchase.

It is also possible that, due to the higher required down payment, the

household becomes a discouraged saver. As shown in equation (2), a household will purchase a house if the total utility gain of homeownership exceeds the loss of utility due to forgone consumption, which is when g>1. This tradeoff is negatively related to the required down payment through its influence on the time of purchase. When the increase in the required down payment is such that g≤1, the household would no longer consider purchasing a house.

In reality, the household might opt to stick with his savings plan and buy a smaller house at (point A) or combine the options of increasing its savings rate, delaying the time of purchase and/or down size the savings goal. However, for simplicity, the savings goal, house H, in this model is exogenously given. This limits the model by excluding the endogenous link between the price of the house, which reasonably reflects quality and/or size, and the utility gained from homeownership. Yalta’s model has shown us that an increase in the required down payment will lead to a delay in the purchase of a house or even to the discouragement of homeownership. We will now investigate whether the effect of required down payment on homeownership can be mitigated by a housing saving scheme.

5.2 Extended model

In order to allow for government intervention via a housing saving scheme we have to extend the Yalta model. First of all we make a distinction between Net

Income and Gross Income by introducing a proportional income tax where income is again constant over time. Consequently the equation for income equals:

(7)

%

_J

= %

_Jh

= %

_Ji

− j

_J

(24)

(8)

j

_J

= o%

_Ji

− d7

_J

Where

δ

_{is the tax rate and}d7_J_{is the amount of tax deductibles at time t. In this}

model all savings are by definition destined for the purchase of a house, thus TD_l is equal to S_l. Henceforth, we replace the notation of TD_l with S_l. The amount of savings in each period now depends on the optimum rate of saving and net income, thus:

(9)

P

_J

= 5 ∙ %

_Jh

By combing equations (7) and (8) and the fact that TD_l = S_l we can show that the net income is:

(10)

%

_Jh

=

s

st

%

Ji

When a households is saving to purchase a house, 5 > 0, if not 5 = 0. Hence, the net income of a household participating in the saving scheme exceeds the net income of non-participating households.

These changes in Yalta’s model result in a model as presented in table 4.

Table 4: Summary extended model

(25)

The most important effect of making housing savings tax-deductible is the effect on the decision on whether to buy a house or not as represented in equation (2). The tax deductibility influences the housing decision through , the time of purchase. To accommodate for the model extension we rewrite equation (1) by combining equations (1), (VIII) and (IX) resulting in a new equation for the time of purchase:

(11)

=

_s_}_~

,

Where

< 0

An increase in the tax deductibility decreases the time of purchase, which in turn influences the housing decision. Following equation (2), reprinted below for convenience, the impact is twofold.

(2)

=

∙

,

Where

< 0

The present value of the utility gained from owning a house, as presented by the top half of the first part of equation (2), increases due to the decrease in the time to purchase. On the other hand, the loss of utility due to forgone consumption, as

denoted by the bottom half of the first part of equation (2), decreases. As a result, g increases for all individuals as decreases. If g>1, the individual will plan to purchase a house. Consequently, more people will become homeowners due to tax deductible savings for house purchase.

For those individuals who would have bought a house even without tax deductibility, the tax deductibility will accelerate their time of purchase. As shown in equation (9), the savings in each time period depend on the optimal savings rate, σ, and the net income, Y_ln. As shown in equation (6), the optimal savings rate is not influenced by tax deductibility. However, tax deductibility increases net income. As a result, the savings in each period increase, resulting in a lower time before purchase.

This section provided us with a theoretical foundation that supports the notion that a housing saving scheme will decrease the time of purchase and stimulate

(26)

amount of borrowing allowed, discourages homeownership. This supports our earlier observation that the restriction placed on mortgage loan by the code of conduct discourages homeownership. Furthermore, by extending Yalta’s model we show that making housing savings tax deductible will stimulate homeownership.

In the subsequent sections we focus on estimating the costs of implementing such a housing scheme in the Netherlands.

6. Data

This paper uses data from the Dutch ‘Housing Need Survey’ 2009 (WoON 2009) which is a survey among 78.071 Dutch households taken in the period

September 2008 until May 2009. This is the most recent edition of a survey which is conducted every 3-4 years. At the time of writing WoOn 2012 was not yet available for this study. The ‘Housing Need Survey’ provides information on household characteristics including age, education level, income, assets, mortgage details and other liabilities. In addition, the data set provides data on dwelling characteristics ranging from price, size of the house to number of rooms. The data is provided by the households themselves through the surveys, though some data is also provided by governmental agencies like the Dutch fiscal authority, the Central Bureau for Statistics and the Dutch Cadaster. The dataset also contains weight variables which can be used to draw population-wide conclusions. Each household in the survey is assigned a person weight variable based on the characteristics of the head of the household and a household weight variable, thereby indicating the representativeness of the person or household for the entire Dutch (household) population based on the population statistics from the Central Bureau for Statistics. These weighting variables are based on person and household characteristics like age, gender, marital status, education level, and income, among others.

(27)

to 74.518 households. As a consequence the sum of the weight factors no longer adds up to the total population of 2009.

7. Methodology

This section will discuss the methodology used to provide an estimate of the costs of a housing savings scheme. We use the aforementioned data from the ‘Housing Need Survey’ which provides data on a representative sample of Dutch households. With the help of the weighting factors we can make inferences on the population as a whole.

When estimating the costs of the housing saving scheme we focus on the costs resulting from the tax deductibility for income from Box 1. For these calculations we use the marginal tax rates from 2012, as presented earlier in table 1. This study does not take into account the recent changes in regulation regarding the maximal

marginal tax rate for interest rate deductibility and the general changes in the

marginal tax rates. In addition, we will not consider the forgone tax income resulting from the exemption of the housing savings from tax for Box 3 income.

The data used for the estimation of the costs provides information on the characteristics of the household population in 2009. We assume that the composition of the population and the characteristics as portrayed by the data will remain constant over time. This enables us to make inferences on future participation of the housing saving scheme.

(28)

We consider the LTI ratio as a binding restriction and not the LTV ratio. The reasoning is that a household facing a LTV restriction could still be a homeowner if its income is still sufficient to find alternative financing. In addition, it is possible for the household to have opted for a less expensive house. It is more ambiguous whether the LTI is a definite restriction for homeownership. Again it is possible that the household could opt for a house with a lower value, however there are also problems with general affordability for the households that exceed the LTI limit. The LTI limit can be defined as a maximum mortgage loan factor, which is the maximum factor of one’s income someone may borrow. We assume that those households whose LTI-ratio exceeds the maximum mortgage loan factor by more than 0.2 would no longer be homeowners and are eligible for the housing savings scheme as artificial renters. We incorporate this margin as some people might have still been able to become a homeowner if the mortgage loan restrictions were in place. A too strict enforcement of the LTI limit would then lead to an overestimation of the number of artificial renters. We assume that these artificial renters rent an average apartment and

accordingly face net rent expenses of €5.400 a year, which is based on the average net rent expenses for apartments in the dataset.

The costs of any housing savings scheme will depend on both the maximum yearly tax deduction and on participation. We will first discuss how we determine the maximum yearly tax deduction, which is followed by a discussion on which

households we deem eligible for the savings scheme. Lastly we make assumptions on which eligible households will actual participate in the housing savings scheme.

7.1 Tax deductibility

(29)

income because this would be the most relevant from a policymaker perspective which we will briefly clarify.

Currently, the Dutch government stimulates the housing market by providing support for both renters and owners of houses. However, we observe a policy gap where some households are neither eligible for rent subsidies nor able to purchase a house. This could be deemed unfair in terms of equality, as well as unfeasible for the functioning of the housing market. Individuals with an income up to approximately €28.000 can apply for rent subsidies and up to an income of €34.000 individuals are eligible for social housing. Those individuals with an income exceeding €34.000 are expected to find housing on the private market, either by renting or purchasing a home. However, problems arise due to the restrictions on the LTI ratio for mortgage loan because not all households with an income higher than €34.000 are able to purchase a house. This leaves a policy gap where some households are not eligible for government support for affordable housing, either through rental subsidies or mortgage interest tax deductibility.

We know from the data provided by the Cadaster that the price of an average apartment in July 2012 is roughly €176.000. On average the purchasing costs of buying a house accumulate to 6% of the house value, thus the total costs of buying an average apartment would be €187.000. Given the LTI limit, someone with an income of €34.000 facing an mortgage loan interest rate of 5% can borrow approximately €153.000, this results in a gap of €34.000 that needs to be financed alternatively if this person wants to buy an average apartment. We set the maximum tax deductibility equal to the amount this person would have to save yearly for 10 years in order to buy an average apartment, which would be equal to €3.400. For simplicity, we assume that the house prices will remain constant in the long run at the level as observed at July 2012.

7.2 Eligibility

(30)

We also impose an upper limit on gross income. We impose this maximum income for eligibility because the housing savings scheme is aimed specifically for those households who might be discouraged from homeownership due to the LTI ratio requirements. We determine the upper limit based on for which incomes the

participation is no longer deemed feasible from a policymaker perspective. We assume a person to be self-sufficient if he is able to fully finance the purchase of a row house based on its income. The average price of a row house (including

purchasing costs) is €216.000. Given the LTI ratio limits, someone with an income as low as €46.000 would still be able to finance their house through a mortgage loan. Consequently we impose a limit of €46.000 on gross income for eligibility for the housing savings scheme.

7.3 Participation

We do not expect that the whole eligible group would participate in the housing savings scheme. As discussed in the section on the economic model, it may be optimal for an individual to rent even though they are able to buy a house. This would depend on their preferences in terms of utility derived from homeownership and the rate of time preference.

In determining who would participate in a housing saving scheme, we first consider those who would be able to set money aside as savings. For this we estimate a Willingness To Pay (WTP) for housing. The WTP will be a measure how much a person is willing to spend on housing relative to its income. For homeowners this would be the net housing expenses, for renters this would be the net rent expense. In the economic model we discussed earlier the optimal savings rate depends on the rate of time preference and the degree of absolute risk aversion. This implicitly entails that each individual has its own optimal savings rate depending on its preferences.

However, due to data limitations we assume that all individuals have the same Willingness to Pay, expressed as a fixed percentage of income.

(31)

after-tax income, thus excluding effects of after-taxes and/or subsidies from housing on income. Based on the dataset, we find that the average WTP is 26% of the VROM-income. In order to determine the Capacity to Save (CTS) for individuals we assume that all individuals have the same WTP, as expressed as a percentage of their VROM-income. We assume that the WTP for all individuals is equal to the estimated 26%. The CTS is the difference between the WTP of an individual and his net housing expenses. For simplicity, we do not allow for partial participation in the housing savings scheme, thus individuals will either participate for the full yearly amount for 10 years or they will not participate at all. In addition, we assume that all eligible individuals whose CTS is equal to or exceeds the yearly tax deductible amount will participate, with the exception of those individuals older than 55.

Based on these assumptions we can identify the group of individuals who would participate at the start of the housing saving scheme which we define as ‘Group 0’. This Group 0 represents the initial inflow to the saving scheme.

Contrary to our expectations, the frequency distribution of participation is not single-peaked. We would expect the frequency of participation to increase until a certain age after which we expect a decline. The reasoning is that beyond a certain age, older individuals are less likely to be willing to save for 10 years and purchase a house because these individuals will have less time to experience the benefits of owning a house. There are two explanations for the fact that the estimated distribution does not match the expected distribution. First of all, this could result from a sample size which is not large enough, this, combined with under sampling of certain age groups, could result in this type of distribution. Another explanation is that we do not account for an individual’s preferences. Especially for older individuals, the fact that they are not a homeowner and have a high CTS may signal that this individual favors the consumption of other goods over homeownership. Consequently we may

overestimate the participation of older individuals.

(32)

estimate the inflow for the years following the first year of the housing saving scheme.

We will base our estimate of the inflow on the participation distribution of group 0. We first calculate the average participation of age groups to smooth

unrealistic fluctuations in participation with age. This results in average participation levels of the age groups as presented in figure 3.

We base our estimate of the inflow on the difference in average participation of the age groups. For example, the average participation for the age groups 18-25 and 26-35 are 1.875 and 7.927 respectively. The difference, 6.052, is the inflow which occurs in age group 26-35. We assume that this inflow is uniformly distributed with age, thus an inflow of 605 can be attributed to each age cohort in the age group 26-35. The total inflow of the age group 18-25 is 1.875 and would result in an inflow of 234 per age cohort. The long run participation distribution over age is presented in figure 4. 0 1.000 2.000 3.000 4.000 5.000 6.000 7.000 8.000 9.000 18-25 26-35 36-45 46-55

(33)

In comparison with the participation distribution of group 0, we observe that the long run participation distribution is single-peaked and shows a significantly lower participation for those individuals of forty years and older. In addition, total participation in the long run (80.700) is significantly lower than the participation of group 0 in the initial year. An explanation for this difference is that the participants of group 0 have not had the opportunity to participate in the saving scheme in the past. As a result, there is a large inflow in the initial year, followed by a relatively low inflow for the subsequent years as other individuals become willing and able to participate over time.

The following section will present and discuss the results from the estimation of the costs and participation of the housing savings scheme.

8. Results

In this section we will discuss the results of the estimation of the standard scheme and the dynamics at play. In the subsequent section, we perform a sensitivity analysis by analyzing the results of estimations under alternate assumptions and specifications.

We allow for a yearly maximum tax deductibility of €3.400 for savings with 0 1.000 2.000 3.000 4.000 5.000 6.000 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56

Figure 4: Frequency Distribution

participation in the long run

(34)

of at least €34.000 to purchase an average apartment after 10 years of saving. In addition, we impose a limit of €46.000 on the gross income for participants because anyone with a higher income should be able to finance a house purchase without support. With this specification and the assumptions we impose as discussed in the methodology section, we are able to estimate the cost of such a saving scheme.

Figure 4 shows how total participation and costs evolve over time. In addition table 5 provides a brief summary of the results. For a full overview of the year by year participation and costs we refer to table 6 of the appendix. At the time of introduction of the saving scheme, year 0, we observe an inflow of a relatively large initial group we call ‘Group 0’. In year 0 approximately 234.000 individuals will participate in the saving scheme and in each subsequent year there will be 8.000 new participants. By assumption, participants will partake in the saving scheme for the full 10 years. Consequently, after 10 years we observe a large outflow of the initial group and for the following years the outflow will be equal the inflow, resulting in a constant participation and costs in the long run.

0 50.000 100.000 150.000 200.000 250.000 300.000 350.000 €-€100.000.000 €200.000.000 €300.000.000 €400.000.000 €500.000.000 €600.000.000 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Time

Figure 4: Standard Scheme, total

participation and costs over time

(35)

Table 5: Results Standard Scheme

Year Total participation Total costs housing savings scheme

0 234.109 € 371.962.131

9 306.109 € 487.628.270

10 80.000 € 115.666.139

Age Yearly inflow per cohort Total yearly inflow of age group

18-25 250 2000

26-35 600 6000

36-45 0 0

46-55 0 0

Total yearly inflow 8000

To put these results in to perspective, Koning et al. (2006) estimated that in 2006 the mortgage interest rate deductibility resulted in approximately €11,75 billion of forgone tax income. They also report that the total of local taxes on housing in 2006 is € 2,25 billion and the transaction taxes totaled 3 billion on the same year. In addition, data from the Cadaster show that over the year 2011 a total of approximately 120.000 house transactions have taken place. Thus the housing saving scheme is relatively small compared to the other housing market interventions and can be

described as a small and targeted approach to stimulating housing market accessibility and demand for the housing market.

9. Sensitivity analysis

In this section we analyze the sensitivity of the results to the specification of the housing saving scheme and the assumptions made. The reason to do this is

twofold. First of all this provides a better understanding of the range of potential costs of the saving scheme. In addition, we would like to provide estimates for alternate specifications, for example, changes in the yearly maximum tax deductibility. These models feature the same dynamics as the standard scheme. To summarize these characteristics, we observe a relatively large inflow at the year of introduction, year 0, followed by a continuous yearly inflow. The costs peak in the tenth year, year 9, after which a sharp drop will occur due to the outflow of the large initial group of

(36)

9.1 Changes in the yearly maximum tax deductibility

In the standard scheme we set the yearly maximum deductibility to €3.400, which allows an individual with a gross income of €34.000 to be able to purchase an apartment after 10 years of saving. Now we investigate the changes in participation and costs for different specifications of the maximum deductibility. Table 7 provides a summary of the results, a full overview of the results is presented in table 8 of the Appendix.

(37)

Table 7: Comparison standard scheme with schemes for alternative limits on yearly tax deductibility

Higher Deductibility Standard Scheme Lower Deductibility

Able to purchase average apartment

with income € 33.000 € 34.000 € 35.000

Yearly maximum

deductability € 3.800 € 3.400 € 2.900

Year Total participation

Total costs housing

savings scheme Total participation

Total costs housing

savings scheme Total particiaption

Total costs housing savings scheme

0 178.133 € 291.799.571 234.109 € 371.962.131 329.089 € 490.875.490

9 229.433 € 383.892.972 306.109 € 487.628.270 435.289 € 647.144.664

10 57.000 € 92.093.401 80.000 € 115.666.139 118.000 € 156.269.173

Age Yearly inflow per cohort

Total yearly inflow of

age group Yearly inflow per cohort

Total yearly inflow of

age group Yearly inflow per cohort

Total yearly inflow of age group

18-25 150 1200 250 2000 350 2800

26-35 450 4500 600 6000 900 9000

36-45 0 0 0 0 0 0

46-55 0 0 0 0 0 0

Total yearly inflow 5700 8000 11800

(38)

9.2 No income limit for eligibility

In the standard scheme we limit eligibility to those individuals with incomes lower than €46.000 in order to provide the tax support to those individuals who would need it the most. Table 9 shows the results of a saving scheme without such a limit. In addition, table 10 in the appendix will present the year by year costs and participation. We observe a significant increase in participation and costs. A closer inspection reveals that the majority of the increase in participation can be attributed individuals in the age groups 36-45 and 45-55. This can be explained by the fact that income generally increases over age which results in a higher probability to participate as their CTS will be higher.

Table 9: Results Standard scheme and alternate estimation

without income cap for eligibility

Standard Scheme No maximum income for eligibility

Year

Total participation

Total costs housing savings scheme 0 234.109 € 371.962.131 639.666 € 1.153.714.658 9 306.109 € 487.628.270 852.966 € 1.550.968.165 10 80.000 € 115.666.139 237.000 € 397.253.507 Age Yearly inflow per cohort

Yearly inflow per cohort

18-25 250 2000 400 3200

26-35 600 6000 1550 15500

36-45 0 0 0 0

46-55 0 0 500 5000

(39)

9.3 Higher Willingness-to-Pay

We now explore the effect of assuming a higher WTP on the estimation of costs and participation. In the standard scheme we assume a WTP of 26% of the gross income (VROM-income), which is based on the net housing expenses of the

individuals in the Dutch ‘Housing Need Survey’ 2009 dataset. However, based on their preferences, some individuals are willing to spend more on housing in relation to their income than others. In addition, especially young individuals might be willing to cope with relative higher housing expenses in anticipation of increasing income. Therefore, we explore the effects of a higher WTP on the costs and participation by arbitrarily setting the WTP equal to 30% of the gross income (VROM-income). The results are presented in table 11, furthermore, a full overview of the estimation is added as table 12 of the appendix. The results show a large increase in participation and costs due to the assumption of a higher willingness-to-pay. This reveals that the results of the estimation of the costs are strongly influenced by the assumption on the Willingness-to-pay and Capacity-to-Save. In the standard scheme we assume 26% based on the average living expenses in relation to income of the households in the dataset. However, this sensitivity analysis shows that a significant larger group may participate, depending on individuals’ true preferences.

Table 11: Comparison standard scheme and alternate scheme with

assumption of Willingness-to-Pay of 30%

Standard Scheme Willingness-to-Pay assumption of 30% Year

Total participation

Total costs housing savings scheme 0 234.109 € 371.962.131 443.658 €612.233.137 9 306.109 € 487.628.270 589.458 €813.510.041 10 80.000 € 115.666.139 162.000 €201.276.904 Age Yearly inflow per cohort

(40)

9.4 Lower participation based on age

Another crucial assumption is the probability distribution of participation over age. In the standard scheme we restricted participation to individuals younger than fifty-six. However, following our methodology without age restriction, some individuals aged fifty-six or higher would be willing to participate. However, it is deemed unlikely that these individuals would still be willing to save for 10 years and purchase a house. First of all, these individuals will have less time to experience the benefits of owning a house. In addition, the fact that these individuals are still not homeowners may be a signal that their preferences favor renting over owning a house. This same line of reasoning can be applied to other age groups. For this alternate estimation we assume that of those individuals we identified as willing to participate from the age group 46-55 years-old only 30% will actually participate. For the age group 35-46 we assume this percentage to be 50%.

Table1 13 shows the results of this alternative estimation in comparison to the standard scheme. In addition, table 14 is added to the appendix, which provides a full overview of the year by year costs and participation of these specifications. We observe a decrease in initial participation by a third compared to the standard scheme. However, the inflow of participants in upcoming years remains unchanged because the inflow in the standard scheme already consisted of individuals aged 35 or below.

Table 13: Results standard scheme and alternate scheme with lower

participation with age assumption

Standard scheme Participation decreasing with age Year

Total participation

Total costs housing savings scheme 0 234.109 € 371.962.131 146.253 € 232.654.256 9 306.109 € 487.628.270 218.253 € 347.298.524 10 80.000 € 115.666.139 80.000 € 114.644.268 Age Yearly inflow per cohort

18-25 250 2000 250 2000

26-35 600 6000 600 6000

36-45 0 0 0 0

46-55 0 0 0 0

(41)

10. Discussion

In this section we pay attention to the limitations in the methodology which might lead to inaccuracy in the estimations. Given the available data, we provide a fairly accurate estimation of the potential participation and costs of a housing saving scheme. However, the actual participation and hence costs might deviate from the predicted participation and costs.

As discussed earlier, there is a significant group of potential participants which is not accounted for. It is impossible to identify the number of (working) adults living with their parents based on the Dutch ‘Housing Need Survey’ 2009. This leads to an underestimation of the actual participation.

In addition, the introduction of a housing saving scheme might alter the rental housing decision of individuals. Individuals might choose to temporarily downsize their current housing or live with their parents with the aim of increasing their capacity to save in order to enable themselves to participate in the saving scheme. If this behavior would occur then the actual participation would exceed the predicted participation.

Furthermore, this study may have inaccurately determined an individual’s capacity to save due to a possible endogenous relationship between WTP and current net living costs. Those individuals with a high WTP value housing over the

(42)

The costs of the housing saving scheme as estimated in this study are the loss of tax income due to the tax deductibility of the saving scheme. However, it is

important to note that the use of the housing saving scheme also decreases the amount which would be borrowed to purchase the house. This results in a tax gain in terms of a decrease in the use of the mortgage loan interest tax deductibility.

Not every housing saving scheme participant would provide these tax gains as these tax gains arise from the comparison to what the individual would otherwise have done. By its nature the housing saving scheme would result in housing purchases which would otherwise not have happened. In these cases no tax gains arise from a decreased use of interest tax deductibility. For the instances in which a house purchase would have occurred, the tax gains depend on the type of mortgage, the amortization schedule and the interest rate.

We illustrate this with an example. When someone participates in the housing saving scheme and saves €34.000 over 10 years, assuming a tax rate of 42%, this entails a loss of tax income for the government of €1.428 each year for the 10 years the individual participates. Now we assume that the individual’s mortgage loan is €34.000 lower compared to what it would otherwise have been. We assume an interest rate of 5% and the most simple, yet extreme mortgage type, an interest-only mortgage. The yearly interest expenses would have accumulated to €1.700 and due to its tax deductibility would have resulted in a loss of tax income for the government of €714. Consequently, for each subsequent year of the mortgage loan the tax gains would amount to €714.

It is difficult to quantify the total amount of tax gains that would result from the housing saving scheme and therefore this is not accounted for in assessing the total costs for the government of a housing saving scheme. Consequently, the

estimated costs should be interpreted as a worst case scenario in which none of these tax gains arise.

Lastly, this study makes use of a household survey taken in the period September 2008 until May 2009. At the time of writing households are being

(Re)Opening the door to the Dutch housing market for first-time homeowners: