Affordability of housing Flanders and the Netherlands compared in the short-term and long-term

Affordability of housing Flanders and the Netherlands compared in the short-term and long-term Marietta Haffner 1 Kristof Heylen 2 1) OTB Research Institute for Housing, Urban and Mobility Studies, Delft University of Technology, Jaffalaan 9, 2628 BX Delft, The Netherlands, m.e.a.haffner@tudelft.nl; Steunpunt Ruimte en Wonen 2) Higher Institute for Labour (Hiva), Catholic University of Leuven, Parkstraat 47, 3000 Leuven; Steunpunt Ruimte en Wonen 1

Abstract Worsening affordability of housing seems to be old news. Recently it has reached the housing agenda again in many countries, also in Flanders and the Netherlands. Rents and house prices in Flanders have been rising more than inflation, as have house prices in the Netherlands. In the Dutch rental sector waiting times have increased. In this contribution the most recent available data on affordability in both countries from 2005/6 are presented in two ways, for the short term and for the long term. The outcomes for both concepts of affordability will be presented in terms of tenants and owner-occupied households according to income group. One of the main conclusions concerning short term affordability is that the residual income is on average lower in the rental sectors than in the owner-occupied sector This gap is bigger in Flanders than in the Netherlands. In addition, in both countries residual income is on average higher in the private than the social rented sector, despite an average lower net rent in the social rented sectors. Also, in all sectors in both countries the income inequality between the first and third income tertile is higher after net housing expenditure is taken into account. Thus short-term affordability is relatively better for higher-income households than for lowerincome households. The interest cost of owner s equity and the mortgage together with the estimated expected house value change determine largely the level of the user cost or the long term affordability in Flanders and The Netherlands. In both countries, the yearly expected value gain surpasses the annual costs for each of the (equivalent) income quintiles, meaning that owner-occupation yields a profit regardless the income level. This is in contrast to the rented sector. Key words: Affordability, housing, Flanders, residual income, The Netherlands, user cost 2

Introduction Worsening affordability of housing seems to be news of all times (Bramley, 1994; Gabriel et al., 2005; Moore & Skaburskis, 2004; Quigley & Raphael, 2004; Whitehead, 1991). Affordability has reached the housing agenda again in Flanders, one of the three Autonomous Regions of Belgium, and The Netherlands. In Flanders, private rent has been rising more than inflation and house prices have doubled since 1995 (De Decker, 2007). Also in the Netherlands house prices have been rising more than inflation (Ministerie van VROM, 2007). In the Dutch rental sector waiting times for a social rental dwelling have increased. This must be regarded as the signal for the tightening of the (social) rental market, as rising rents in response to tighter markets are mostly impossible as a result of rent regulation in most of the rental market. Also it is often argued in both countries that the problems of first-time movers accessing the housing market are increasing in Flanders and The Netherlands. Therefore time has come to compare the most recent available data on affordability in both countries. Opportunity to do so is convenient, also because new data is available for both countries. For Flanders the new information on housing affordability became available for 2005. For The Netherlands the reference date for the new data is 1 January 2006. As first-time movers are considered to be in trouble on the housing market on the one hand, and house prices have been rising enormously in both countries on the other hand, the question about housing affordability can be asked in two ways: what is housing affordability in the short-term; what is it in the long-term? Stated differently: what cash flows (expenses) do households need to pay for their housing in the short-term when they move to a (new) house; what are the housing costs in the long-term taking into account also expected house value developments? With the presentation of short-term and long-term results for housing affordability in two countries we intend to combine the results of two standard ways which can be found in the literature for explaining affordability. Mostly they are separately used (Gabriel et al., 2005; Bramley & Karley, 2005) sometimes both are used at the same time in order to explain different aspects of affordability (Doms et al., 2001; Elsinga & Conijn, 2001; Haffner, 2000; Quigley & Raphael, 2004). It will be argued that each definition of affordability has it own advantages and uses. In order to present our results in two ways, in the next Section the theoretical explanations can be found for the reasons of exploring affordability of housing in the short-term and long-term. Next we present how we put the definitions into practice for Flanders and The Netherlands. In order to understand the affordability results in their context, the next Section focuses on the housing market and relevant housing policy. The results follow before the content of this contribution is put into conclusions. Affordability: short run versus long run Concepts The concept of affordability of housing is a vexed one (Wilcox, 1999). It means different things to different people (Quigley & Raphael, 2004: 191/2): [it] jumbles together in a single term a number of disparate issues: the distribution of income, the ability of households to borrow, public policies affecting housing markets, conditions affecting the supply of new or refurbished housing, and the choices that people make about how much housing to consume relative to other goods. 3

Even if affordability is only about the consumption of housing of individual households, different definitions of the concept can be found. From the perspective of an investment good approach, some of the confusion can be explained, however. In (neo)classical economics a dwelling is an investment good that generates a flow of services (Hall and Jorgenson, 1967), being housing services in the case of housing. It is these housing services that are consumed, not the dwelling itself. Affordability of the housing services can be expressed in two standard ways. On the one hand housing services have a cost, the so-called user cost of (housing) capital (e.g. Conijn, 1995; Hall and Jorgenson, 1967; Van Order and Villani, 1982; Hendershott 1988; Poterba, 1984). The second way affordability of housing services can be expressed is by way of the expenditure or cash flows for the finance of the service (Hulchanski, 1995; Stone, 2006a). Table 1 and 2 show the different components of the two concepts, as they are used in this contribution. Table 1 Components of housing expenses in relation to the income of the tenant and the owner-occupier Tenant Owner-occupier Gross rent Gross housing expenses* - Housing allowances +/- Fiscal effect income tax = Net rent = Net housing expenses Disposable income Disposable income - Net rent - Net housing expenses = Residual income = Residual income *) Gross housing expenses include not only mortgage expenses, but also owners costs, such as owner s share of maintenance costs and property tax. Table 2 Components of user cost for tenant and owner-occupier* Tenant Owner-occupier Gross rent Mortgage interest - Housing allowances - Fiscal effect income tax = Net rent + Interest on owner s equity - Fiscal effect income tax + Interest and tax effect on transaction costs + Costs of operation +/- Expected value change of the dwelling = User cost of capital = User cost of capital *) See Appendix A for the assumptions and the calculation of the user cost for the homeowner. The principal difference between the two concepts arises with homeownership because the two roles of owner and consumer are not separated. It is the owner of the dwelling that takes out the loan (finance) and pays for the loan (interest and repayment, including possible tax effects). The owner also calculates what the cost of the housing service will be; this would also include the opportunity costs of owner s equity and not only the costs of loan financing, including income tax effects. Cost would also include operational or management cost such as 4

depreciation and insurance and cost would allow for expected value changes of the dwelling. 1 It is the housing consumer that pays the cost: To an economist housing rent, is the opportunity cost of using housing capital for one period (Quigley & Raphael, 2004: 195). Depending on which roles are compared in which tenure, affordability across renting and owning can be expressed as user cost on the one hand, and as cash flow on the other (Elsinga & Conijn, 2001; Haffner, 2000). On the one hand user cost of renting (rent) is compared to user cost of homeownership. In both cases the costs of the consumption of housing services are compared. On the other hand rent, as cash flow needed for consumption, is compared to the cash flows that an owner(-occupier) needs to finance homeownership. An explanation can be found in the distinction between a long-run versus a short-run dimension of affordability (Hancock, 1993). Short-run affordability is about liquidity of households and their possibility to finance the access to either a rental or owner-occupied dwelling at a certain moment in time. Long-run affordability is about the ability to pay the cost of housing services in the long-run. The long-run concept, the user cost concept, is the one that allows price/cost comparisons between one unit of housing service and another. The short-run concept allows a comparison of liquidity, based on cash flows. Definitions of affordability contain three elements as the often referred to definition by Maclennan & Williams (1990; Whitehead, 1991; Hancock, 1993: 129; Freeman et al, 2000) shows: Affordability is concerned with securing some given standard of housing (or different standards) at a price or a rent which does not impose, in the eyes of some third party (usually government) an unreasonable burden on household incomes. The three elements are: a standard of housing quality, a price or rent and a standard or norm for reasonability of the price or rent in relation to household income. The standards of quality and of reasonability of price or rent appear necessary mainly from a policy point of view whether a certain price-quality standard for housing is considered to be achieved and whether it is considered to be reasonably affordable. In some definitions quality standards pertain to housing needs of households and are seen as different from affordability (Gabriel et al., 2005; Whitehead, 1991). One way to look at both affordability concepts in relation to the standard of reasonability is using the earlier mentioned concept of opportunity cost. Whitehead (1991: 873) defines affordability as the opportunity cost of housing vis-à-vis other goods and services. (See also Stone, 2006b). Opportunity cost of housing is then about what has to be foregone in order to obtain housing and whether that which is foregone is reasonable or excessive in some sense. (Hancock, 1993: 129). In the case of cash flows/liquidity a price or rent standard in relationship to income would be considered a logical opportunity cost, or rather expense, as affordability indeed is about using current household income or household income that is foregone by paying for housing. 1 Another way of arranging the components of user cost is to present the direct return on investment after income tax and the indirect return. The indirect return consists of the appreciation of house value, while the direct return consists of the user cost minus the costs of operation. Indirect return is usually included in user cost because it can be considered as an increase in income potential of the owner. This increase in economic power can be argued based on the accretion concept in American literature (Goode, 1977). 5

As the user cost concept is a long-run concept, permanent income may be more useful to express the cost that is foregone in order to consume housing services. When the determination of housing subsidies is the issue, a benchmark based on some kind of neutrality concept may be chosen. Housing subsidies may aim for an equal policy treatment of dwellings from e.g. a tax or tenure neutral point of view (Haffner, 2002; Hancock & Munro, 1992; Hills, 1991; Poterba, 1984, 1992, Thalmann, 2007). For measuring liquidity there are two well-known indicators relating the price or rent to income. Stone (2006a) following Baer (1976) calls this the indicator. First, there is the expenditure-to-income ratio which is about taking up part of the household budget. Secondly, the residual income is about the income left for non-housing needs after paying for housing (Gabriel et al., 2003; Hulchanski, 1995). Stone (2006b: 455) states that The opportunity cost language is essentially the logic of residual income. The advantage of using residual income is that a standard of reasonability can be designed, while for the former no standard is available, as the proportion spent on housing does not give any indication whether too much is spent on housing in relation to other expenditure. As a reasonable standard for residual income official social minima for non-housing needs (e.g. Income Support in the UK) are applicable (Hancock, 1993; Freeman et al., 2000). In addition, the family budget approach is referred to as a suitable method for composing adequacy standards (Stone, 2006a). Application of concepts In our study of the short term affordability, we apply the residual income concept as opposed to the ratio-approach. The residual income method gives us an idea of consumption possibilities in absolute terms after deducting housing expenses. Using the ratio-approach, a ratio of 20% does not have the same meaning for a low income than for a high income household. The residual income approach overcomes this conceptual weakness. Another argument for applying the residual income approach is the fact that housing expenses stand for a fixed claim on the household budget. They have a different nature than expenditure on for example food or clothing, which are more easily cut down in the short run (Freeman et al, 2000; De Decker & Van Dam, 2005). Given this fixed claim on income, it is interesting to analyze income after deduction of housing expenses. In this respect Van Dam et al. (2001) showed that the relative poverty rate in Flanders is higher in the case of residual income than when disposable income is taken into account. These results are in line with Schwabe s law, which states that the share of housing expenses in the disposable income decreases as disposable household income rises (Hulchanski, 1995). Also in the Netherlands recent data indicate that this law applies (Ministerie van VROM, 2007: 143). As mentioned before, another advantage of the residual income approach is that it allows for meaningful standards. However, since no recent studies on family budget standards are available for Flanders and the social minima in Flanders and the Netherlands are taking housing costs into account, no adequacy standards are used in this study. As an alternative, we calculate mean residual incomes and compare them between housing market sectors and income tertiles. In our study, the residual income is calculated as disposable household income minus net housing expenses. As table 1 shows us, net housing expenditure in the rented sector (net rent) is computed by subtracting the housing allowance from gross rent. The net housing expenses in the owner-occupied sector are the result of taking the fiscal effect into account. The latter is the sum of the mortgage payment (interest plus capital repayment), property tax and the 6

owner s share in dwelling maintenance. The two latter expenditure posts are included in order to make expenditure in the rented and owner-occupied sector more comparable. For Flanders the Housing Survey 2005 is used for calculating the affordability measures. This survey has been conducted by the Kenniscentrum voor Duurzaam Woonbeleid and includes 5216 Flemish households. A weighting factor is applied for adjusting the distribution of dwelling type and district. For The Netherlands we relied upon the Dutch Housing Survey, the WoON 2006, a survey of the Ministry of VROM. In the WoON 2006 the income data of the tax administration of 2005 are coupled to the survey data. For the WoON 2006 more than 60.000 households have been interviewed. Weighting factors are applied to the sample, also to get results for all Dutch households, almost seven million. Based on the raw data, we set our minimum number of necessary observations at 50 per cell, also for the Flemish Housing Survey, in order to safeguard the statistical reliability and accuracy of the results. Regarding the analysis of the short term affordability, we need to take the group of recently moved households into account. For these are the households that deal with the current situation on the housing market and the current housing policy. Therefore, we analyzed the group of households that moved during a period of five years before the survey-date. For The Netherlands this is the period 2001-2006; for Flanders, the period 2000-2005. Concerning the analysis of the owner-occupied sector, only the situation of homeowners that have taken out a mortgage is taken into account. An affordability analysis in the sense of financial accessibility is less relevant for the group without a mortgage, since gross housing expenses only include property tax and homeowner s maintenance expenses. The long term affordability analysis is carried out for the households that moved during a 10- year period before the survey date. This restriction is set because the Flemish Housing Survey 2005 does not include information on house prices for the group that moved before 1995 (for the sake of memory bias). In this analysis, all owner-occupiers are included. With regard to the user cost the difference between homeownership with or without mortgage is not relevant, since the cost of equity is considered also a component of the user cost and not only the mortgage interest. The components of user cost to be calculated can be found in Table 2. The components are the costs of capital net of income tax, the management costs or costs of operation and what we call the expected value change of the dwelling. We used values for the relevant year (2005) or calculated estimates, as Appendix A explains. Most of the data, assumptions and calculations are relatively straightforward. Less straight forward may be the expected value change of the dwelling. The expected value change of a dwelling is to be regarded as the so-called specific or pure price change. The specific price change is the inflationary component of the value change (Conijn, 1995; Wigren, 1996). It is the price change of a dwelling with a constant-quality package of services. Price changes reflecting quality changes in the dwelling, such as improvements, are excluded from the price change. For the Flemish data, we did not make such a correction, as quality improvements were not detected. For the Dutch data the repeated sales index was used to calculate the specific price change: the price change of individual dwellings is being followed. In order to have more of a long-run view, price changes were averaged for the last five years. 7

Housing market and policy Flanders and The Netherlands may be considered well chosen, as countries with similar signals about affordability problems, but with different housing markets and also different housing policies. Flanders may be the country in northwestern Europe with the possible exception of Germany that has the smallest social rental sector expressed as a share of the total housing stock, while The Netherlands has the biggest social rental sector (Department of the Environment, Heritage and Local Government (Ireland), 2004). Table 3 shows these differences: almost 6% for Flanders in 2005, more than 35% for The Netherlands. The counterpart can be found in the market share of owner-occupation which amounts to more than 74% in Flanders and almost 56% in The Netherlands, leaving more than 19% of the market in private landlords hands in Flanders and nine percent in the hands of Dutch private landlords. Owner-occupiers with and without a mortgage also is a distinguishing aspect between the two countries : 45% of Flemish owner-occupiers are still paying for a mortgage loan to finance their home in comparison to twice that percentage (almost 90%) of Dutch owner-occupiers. Table 3 Size of housing market sectors (in %) in Flanders (2005) and The Netherlands (2006) Tenure Flanders The Netherlands Rented sector total 24,1 44,2 Social rented sector 5,6 35,2 Private rented sector 18,5 9,0 Owner-occupied sector total 74,4 55,8 With a mortgage 33,6 48,5 Without a mortgage 40.9 7,3 Free housing 1,5 Total households 100 100 Source: Flemish Housing Survey 2005/Steunpunt Ruimte en Wonen and WoON 2006/OTB Research Institute for Housing, Urban and Mobility Studies The types of mortgage loans that are used to finance homeownership are also different between both countries. In Flanders they are mostly repayment loans, either annuity loans or fixed capital repayment loans, while in The Netherlands other forms of mortgage loans dominate, such as endowment loans and interest-only loans. For both of the loan types the full amount of interest is paid during the loan term, but no repayment takes place. For the endowment loan instead of repaying the loan, the borrower will pay into a savings or stock account. Popular loan types can be derived from the treatment of owner occupied dwellings in income tax, sometimes in combination with a reaction to the situation on the housing market. Examples are the interest-only loans in The Netherlands that became quite popular in more recent years, because of house price appreciation. House prices rose with almost 30% in the period end of 2000 to the end of 2005 (Kadaster, http://www.kadaster.nl), while household incomes could not keep up the pace (Ministerie van VROM, 2007). The average house price almost reached 223,000 euro; the range covered 165,000 euro for apartments to 370,000 euro for a detached single family dwelling. 8

In Flanders the average house price of 162,400 euro in 2005 is lower than in the Netherlands. However, price increases on average outpaced income increases as well. The price of small and middle-large dwellings even rose more sharply during the period 2000-2005. In 2000 the price on average was 89,500 euro; in 2005 already 139,600 euro. Also the price of apartments rose strongly, with an average of 93,000 in 2000 which rose to 147,500 in 2005. In contrast, the mean price of large dwellings in 2005 (272,300 euro) is not considerably higher than in 2000 (262,700 euro). In 2005 29% of sold dwellings were categorized as apartments, 55,6% as medium-size dwellings and 15,4% as large dwellings (Stadim, 2006). Even though house prices rose strongly in both countries, only in The Netherlands exotic mortgage types appeared, because of the influence of the personal income tax system: endowment loans and interest-only loans in the Netherlands are attractive. Mortgage interest, which is at the maximum during the loan term for these mortgage types, is fully deductible (after the taxation of imputed rent is taken into account) in income tax against marginal tax rates. This is not the case in Belgian federal law on income tax; thus repayment mortgage types are popular there. Owner-occupiers with a mortgage receive a substantial deduction in income tax with regard to their mortgage payments. In a first step the deductible sum is calculated; in a second step this sum is deducted from the yearly taxable income. In the fiscal system that was in force from 1989 to 2005 the deductible sum consisted of three parts: the regular and additional interest deduction and the capital deduction. The level of these three components depended on several conditions the height of the mortgage sum and the household composition being the most important factors - and was bounded by some upperlimits. Furthermore, the fiscal effect in Belgium was limited in time. In the rental sector housing policy shows its different effects in both countries in three ways, implicit or explicit bricks and mortar subsidization, implicit or explicit income related housing allowance, and rent regulation (Elsinga et al., 2007). In the Flemish social rented sector the social housing associations are entitled to build and renovate social dwellings, using subsidies in the form of low-interest loans floated by the VMSW, which is an autonomous agency entitled by the Flemish authority to finance social housing associations. By means of this bricks and mortar subsidy, the social rent is set lower than the market rent. Rent setting is differential which can be considered an implicit housing allowance (Kemp, 1997). The rent level is related to the household income and composition on the one hand and to the actual building cost on the other. Affordability gets an extra guarantee by an upper rent-limit (1/60 th of the yearly taxable household income). The eligibility criteria income boundaries are rather selective. In 2005 about 12% of Flemish households were eligible for social housing. Of all tenants on the private market 39% of households is entitled to social housing in 2005. From 2008 onwards the social rent setting system changed, in order to make it more transparent for tenants. Yet, the link with affordability and dwelling quality is still present. Next to differential rent setting, there are also explicit housing allowances available in the Flemish rental sector. The housing allowance system in Flanders is a selective one. In 2005, only low income households that moved from a poor quality rental dwelling to another rental dwelling in good condition were eligible. The Housing Survey points out that in 2005 only two percent of tenants received this allowance, with an average of 120 euro per beneficiary. In the Netherlands the instruments of bricks and mortar subsidies are designed differently than in Flanders: bricks and mortar subsidies nowadays are an implicit or non-government contribution and housing allowances are an explicit or government contribution. For the bricks and mortar subsidies the year 1995 was an important year for the social housing associations. At this time the ties with the central government were cut, meaning that government granted no more periodic object subsidies (see also Ouwehand & Van Daalen, 9

2002). Nevertheless, social housing associations are still able to set the social rent below the cost price rent due to uneconomic investment which is in line with their non-profit status. This way, social tenants are implicitly subsidized by housing associations. One Dutch tenant out of three received a housing allowance in 2006 (Ministry of VROM, 2007). On average this boiled down to 148 Euro per month per recipient. In order to set the allowance level the household income and composition as well as the rental price (maximum 615,01 euro per month in 2006) are taken into account. A final housing policy instrument in the rental sector is rent regulation. In Flanders it applies to running contracts in the private rental sector; in The Netherlands to new and running contracts with a rent level below a certain maximum. In Flanders, the Rental Act (Huurwet) is one of the remaining policy competences of the Federal Government apart from fiscal policy in the field of housing. All other housing policy is formulated by the three Autonomous Regions. The Rental Act is considered a liberal law, as it is characterized by a strong protection of the right of ownership, a contractual freedom and assumed equality of parties. In Belgium no system of rental price regulation exists for new contracts in the private rental sector. Contracts of sitting tenants are regulated with inflation for the duration of the contract (maximum of nine years). In The Netherlands the dwelling assessment system and the yearly rent adjustments are important instruments that regulate the rented sector. The yearly rent adjustments are decided by the Parliament and hold for the regulated part of the rental market, which counts for 95% of the total rented sector. The dwelling assessment system is a scoring system for evaluating the quality of the dwellings in the regulated rented sector. The score determines the maximum rent for a certain dwelling at the start of the tenancy agreement and is based on relevant features of the dwelling and the neighborhood. Short-term affordability Table 4 shows the results of the short-term affordability calculations for the different housing market sectors in Flanders and The Netherlands. In the rented sector in Flanders in 2005 the average net rent of the recent movers does not diverge much from the average gross rent. We already mentioned that the span and thus the average amount - of the explicit housing allowance in Flanders is limited. In Flanders the net social rent on average is 258 euro, which is considerably lower than average net rent in the private sector (453 euro). Nevertheless, because the mean disposable income was considerably higher for private tenants, the residual income in social housing is significantly lower than in the private market. Given that the gap in average rent between the private and social rented sector approaches 200 euro and social rented dwellings on average do not have less rooms than private rented dwellings, the bricks and mortar subsidy involved in combination with the differential rent setting in the Flemish social rented sector is substantial. The fiscal effect for owner-occupiers with a mortgage in Flanders on average amounts to 84 euro per month. Assessing this figure, we should take into account that Belgian households only receive their tax return after a period of more or less two years. Consequently, we hypothesized that the owners that have moved in the year of the survey or the year before (2004/5) have not yet used their tax deduction. Further, the average net housing expenditure in Flemish owner-occupation (746 euro) is by far higher than the average net rent (421 euro). However, the average residual income is more than 1200 euro higher (factor 1.9) in the owner-occupied sector because of considerable income differences between renters and owner-occupiers. 10

Table 4 Affordability according to tenure status and type of rental, for households moved during five recent years, means in euro per month, Flanders/The Netherlands, 2005/2006 Gross rent/ gross housing expenditure Housing allowance/ fiscal effect Net rent/ net housing expenditure Disposable Income Residual income Flanders Rented sector 423-2 421 1814 1393 Private rented sector 453-1 452 1884 1432 Social rented sector 258-2 256 1433 1177 Owner-occupied 830-84 746 3349 2603 sector* The Netherlands Rented sector 421-47 374 1631 1257 Private rented sector 501-19 482 1927 1445 Social rented sector 394-56 338 1533 1195 Owner-occupied sector* 1074-308 766 3127 2361 *owner-occupiers with a mortgage Source: Flemish Housing Survey 2005; WoON 2006/OTB Research Institute This tenor of the conclusion is also valid for the Netherlands on January first, 2006 (Table 4). The ratio of mean disposable income between social rent and owner-occupation is 2.0 (3127 euro versus 1533 euro); the ratio remains 2.0 for the residual income (2361 versus 1195 euro). In Flanders this ratio decreases from 2.3 to 2.2, with the step from disposable to residual income. The factor for the difference in mean disposable income for owner-occupiers and private renters is 1.6 in the Netherlands and 1.8 in Flanders, before and after housing expenses. Thus on average the difference in short-term affordability between the rental subsectors and homeownership is bigger in Flanders than in the Netherlands. For the rental sector as a whole, the average difference with owner-occupation is 1.9 in both countries. This equal result comes about because the averages for rent are closer to the bigger rental subsector in a country: private rent in Flanders, social rent in the Netherlands. The extreme low mean social net rent of 256 euro in Flanders is remarkable, compared to the Netherlands (421 euro). Differential rent setting in the much smaller Flemish than Dutch social rental sector in Flanders apparently has a big effect, as Flemish housing allowances on average are negligible. A final difference between both countries is the average fiscal effect of the recently moved owners. It is remarkably higher in The Netherlands than in Flanders (308 versus 84 euro). However, this higher government intervention in The Netherlands is compensated by higher gross housing expenditure (1074 versus 830 euro), which results in a similar average amount of net housing expenditure in both countries. In The Netherlands, gross housing expenses on average are higher than in Flanders as a result of average higher house prices and the possibly higher loan-to-value rates. The results in Table 4 for the short-term analyses for the different tenures in both countries thus show that on average housing is much more affordable in the short run among homeowners with a mortgage than tenants. The residual income is higher for homeowners with a mortgage than for tenants, even though the average net housing expenditure is higher 11

for homeowners than for tenants. The on average higher disposable household income for owner-occupiers than for tenants allows for these results. In addition, in Flanders as well as The Netherlands the residual income is on average significantly higher in the private rented than in the social rented sector. These results did not take into account the possible differences in household types in the different groups. One example is that in 2005 there are more single-family households (including one-parent households) in the Flemish social rental sector than in the private rental sector. The reverse (only singles) is true for The Netherlands. To account for household type differences in the different tenures, Table 5 presents the shortterm affordability figures for Flanders according to tertiles of equivalent income, for both tenants and owner-occupiers with mortgage who recently moved. As to be expected, the gross rent for tenants and the gross housing expenditure for owners rises with income. For both tenures, the average in the third tertile is 40% higher than in the first tertile. Logically, also disposable and residual income is highest in the third tertile. Because of the income boundaries applied to housing allowances in Flanders, the average housing allowance decreases with income. The average fiscal effect on the other hand is higher in the second than in the first tertile as a result of certain characteristics of the Belgian tax system. Because of the built-in upper-limits of the system, the average effect in the third tertile is not significantly higher than in the second, even though gross housing expenditure is on average more than 240 Euro per month higher in the third than in the second tertile. For both the rented and owner-occupied sector with a mortgage income inequality between the first and third tertile on average increases after deduction of the net rent or net housing expenditure; it increases when taking the step from disposable to residual income. In the case of the rented sector the inequality-ratio increases from 2.3 to 2.8, in the case of owneroccupation from 2.0 to 2.2. Inequality on average is higher in the rental sector and also increases more in the rental sector, implying that housing is relatively less affordable in the rental sector than for owner-occupation with a mortgage. Table 5 Affordability according to tenure status and (equivalent) disposable income, for households moved during five recent years, means in euro per month, Flanders, 2005 Gross rent/ gross housing expenditure Housing allowance/ fiscal effect Net rent/ net housing expenditure Disposable Income Residual income Rented sector, disposable income tertiles 1 358-4 354 1125 771 2 418-1 417 1645 1237 3 488 0 488 2617 2129 Ratio 3 rd tertile/1 st tertile 1.4-1.4 2.3 2.8 Owner-occupied sector with a mortgage, disposable income tertiles 1 699-60 639 2325 1686 2 769-92 677 3111 2434 3 1011-98 913 4540 3627 Ratio 3 rd tertile/1 st tertile 1.4 1.6 1.4 2.0 2.2 Source: Flemish Housing Survey 2005 12

Table 6 is the Dutch counterpart for the Flemish table (Table 5) with the short-term affordability results according to tertiles of equivalent income, for both tenants and owneroccupiers with mortgage who recently moved. The drift of the observations for Flanders also goes for the Netherlands: gross and net housing expenses and income rise with equivalent income tertile; the average amount for housing allowances decreases with income while the inequality of income between the first and third tertile increases after housing expenses (residual income). Furthermore, in both countries the ratio for average residual disposable income between the third and the first tertile is higher for the rental sector than for the owneroccupied sector. The short-term affordability turns out relatively worse for tenants than for owner-occupiers. In contrast to the Flemish situation, the average fiscal effect for the Dutch owner-occupiers with a mortgage increases between the second and third tertile, meaning that no upperceilings are implemented with regard to the fiscal effect. There are other differences between both countries as well. The ratio of housing expenses between the first and third tertile in Flanders is not affected by traceable government intervention, meaning the effect of housing allowances and income taxation does not change the average ratio. In the Netherlands this ratio changes in the opposite direction for tenants and homeowners. For renting the ratio increases; for homeowners with a mortgage it decreases. Furthermore, in the Netherlands the income on average is more unequally distributed between the first and third tertile (ratio of 2.5 for renting and 2.1 for homeownership with mortgage) than in Flanders (respectively 2.3 and 2.0). Income inequality increases after deduction of housing expenses. In absolute terms the inequality of income increases most in the Flemish rental sector (+0.5), while it increases least for Flemish homeowners with a mortgage (+0.2). For Dutch homeowners with a mortgage the inequality ratio increases with 0.4, for Dutch tenants the increase is 0.3. On average the inequality in residual income turns out equal in both countries for the tenants (2.8), but unequal for owner-occupiers with a mortgage (2.2 in Flanders versus 2.5 in the Netherlands). This again shows that short-term affordability in both countries for tenants on average is worse than for owner-occupiers; for Flemish owner-occupiers with a mortgage better than for the Dutch ones. 13

Table 6 Affordability according to tenure status and (equivalent) disposable income, for households moved during five recent years, means in euro per month, The Netherlands, 2006 Gross rent/ gross housing expenditure Housing allowance*/ fiscal effect Net rent/ net housing expenditure Disposable Income Residual income Rented sector, disposable income tertiles 1 370-91 279 974 695 2 407-44 362 1518 1156 3 485-5 480 2401 1921 Ratio 3 rd tertile/1 st tertile 1.3 0.1 1.7 2.5 2.8 Owner-occupied sector with a mortgage, disposable income tertiles 1 901-191 710 2129 1419 2 1024-282 742 2862 2120 3 1297-450 847 4390 3543 Ratio 3 rd tertile/1 st tertile 1.4 2.4 1.2 2.1 2.5 *) Housing allowance received is based on income in reference period. That income may not be equal to the income measured in the survey. Source: WoON 2006/ OTB Research Institute Long-term affordability Table 7 presents the yearly user cost of housing in Flanders and the Netherlands, for households that moved during the last ten years (before the year of the survey). The results are shown according to tenure status and equivalent income. Figure 1 and 2 show the yearly user cost for owner-occupiers per income quintile divided according to cost (positive) and gain (negative) components, for respectively Flanders in 2005 and The Netherlands in 2006. For Flanders, when all user cost components are included, the user cost of owner-occupation in 2005 turns out to be a return: owner-occupation on average yields a yearly profit of 7637 euro (Table 7). This yield does not significantly vary between the four lowest income quintiles. In the fifth quintile this profit is remarkably lower (1103 euro on average). The user cost has turned negative for 2005 because of the strong influence of the expected value change. The average gain from the 2005 expected value change for owner-occupiers is 19703 euro (see Figure 1). When the expected value change component is excluded, the yearly user cost for owneroccupation in Flanders is on average 12066, meaning a cost of 1006 euro per month. This cost is considerably higher than the average user cost in the Flemish rental sector (4936 euro per year). In order to calculate the expected value change, we multiplied the present value of the dwelling with the average yearly increase in house prices in the 5 years preceding the surveyyear (2005). For Flanders this rise is differentiated according to province and dwelling type. The average yearly rise is 8,2%. The expected value change is related to the household income in two opposite ways. First, higher income households on average occupy dwellings of higher value than lower incomes, which contributes to a greater value change. The more value one has, the more change is possible. Because of this effect, the expected value change increases from the first through the third income quintile in Flanders (Figure 1). On the other hand, the average yearly price-rise 14

of large dwellings in Flanders in the 5 year period (1,1%) is far less outspoken than the pricerise of medium-size dwellings (10,7%) or apartments (9,3%). Logically, high income households own relatively more large dwellings than households with a lower income. Therefore, as Figure 1 indicates, the expected value change in the fifth income quintile is below average, resulting in a significantly lower user gain in the fifth than in the other quintiles. Table 7 Flanders Owner-occupation, incl. expected value change Owner-occupation, excl. expected value User cost of housing according to tenure status and (equivalent) disposable income, for households moved during ten recent years, means in euro per year, Flanders/The Netherlands, 2005/2006 Total Quintiles of equivalent disposable income 1 2 3 4 5-7637 -8863-9819 -9036-9469 -1103 12066 9914 11025 12630 11597 15190 change Rented sector 4936 3705 4478 5164 5133 6064 Private rented sector 5328 4438 5217 5184 5475 6187 Social rented sector 2998 - - - - - The Netherlands Owner-occupation, incl. expected value change Owner-occupation, excl. expected value change -4128-3093 -3292-3431 -3504-5660 12290 13315 10962 10968 11373 14293 Rented sector 4437 3387 4325 4889 5504 7107 Private rented sector 5653 4148 4843 5360 6382 8602 Social rented sector 4092 3251 4216 4735 5094 5519 Source: Flemish Housing Survey 2005, Stadim 2008, WoON 2006/ OTB Research Institute When the expected value change is not taken into account, the 2005 user cost is significantly higher in the higher income quintiles of Flemish owner-occupiers, mainly because higher incomes on average occupy higher valued dwellings, yielding more interest costs on owner s equity and mortgage. Furthermore, also management costs are higher among higher income Flemish home-owners. These trends are clearly visible in Figure 1. For Flanders, the management costs include maintenances costs, property tax and depreciation of the dwelling. In addition, the user cost in the (private) Flemish rented sector strongly varies with equivalent income. The difference with regard to user cost between the private and the social rented sector is considerable. For the Dutch owner-occupiers total user cost per year excluding the estimated expected value change at the value of 12290 Euro is on average slightly higher (more than 200 Euro) than the Flemish total, as Table 7 shows. User cost including expected value change on average are negative as well, boiling down to a yield earned by living in ones own dwelling. 15

Thus in both countries the average estimated expected value change exceeds the total of the other components of user cost, such as the interest costs of owner s equity and debt and the management costs. For the Netherlands the average expected value change was estimated as the average value change over the five years before the year of the survey calculated for the twelve provinces and the single-family versus the multifamily dwelling. The averages for the multifamily dwelling ranged from 4.3% in Flevoland (new land province) to almost 12% in the north of the Netherlands (Groningen en Friesland); for the single-family dwelling from 3.9% in the south (Limburg) to almost ten percent, also in the south (Zeeland). In general house prices have been decreasing towards the end of the period and changes have even turned slightly negative in the last year in Flevoland for the apartments (-1.2%). This is contrary to the Flemish house price rises which were still increasing at the end of the five year period. Figure 1 Composition of user cost of Flemish owner-occupiers and according to quintile of equivalent income, 2005 Source: Flemish Housing Survey 2005 16

Figure 2 Composition of user cost of Dutch owner-occupiers and according to quintile of equivalent income, 2006 Source: WoON 2006/ OTB Research Institute The yield of owner-occupation increases in absolute terms from the second quintile on with equivalent disposable income. This is in contrast to the Flemish user cost which varies with income all the way. The most remarkable difference is the one of the fifth quintile where in absolute terms the largest average estimated expected value change can be found for the Netherlands, in accordance with expectation. For Flanders it is the smallest one, as explained before. As in Flanders Dutch owner-occupiers are in fact earning a yield on living in their own dwelling, while tenants are paying their rent, lower for social tenants than for private tenants. Apparently the combination of differential rents, possible bricks and mortar subsidies and housing allowances turns out on average much lower in Flanders (2998 Euro) than in the Netherlands (4092 Euro), although the annual user cost for the private tenant is also lower in Flanders (5328 Euro) than in the Netherlands (5653 Euro). Figure 2 shows the components of long-term affordability for owner-occupiers in the Netherlands according to quintile of equivalent income. As could be derived from Table 7, again the prominent effect and the distribution of the estimated expected house price rise can be observed. The components of user cost that are important as well are the interest costs on owner s equity and on debt. The Dutch interest costs on debt are higher on average than the Flemish ones (compare with Figure 1); Mortgage amounts on average must be higher because of higher house prices. The same then goes for the tax deduction on mortgage interest. For interest costs on owner s equity (including transaction costs) the situation is reversed. Finally, the managements cost which for the Netherlands consist of maintenance cost, property tax, depreciation of the dwelling, costs of the ground lease and the building insurance, on average come to more than 40% of the user cost excluding the estimated expected value change of the dwelling. The exception to the Dutch rule that the amount of mortgage interest on average is higher than the opportunity costs of owner s equity is in the first quintile. Those are the households which have been owner-occupier for a longer time and have repaid part of their mortgage. Also, their marginal income tax rate which is relevant for mortgage interest deduction is lower 17

than in the other quintiles. For the interest cost on owner s equity there is no difference between the quintiles as the tax rate is proportional (30%) to the return on equity. The results show that for both countries for each quintile of equivalent disposable household income the estimated expected value change of the dwelling surpasses the other cost components, making the user cost turn out negative, resulting in a yield from living in the own dwelling. The period of five years that was used for the calculation of the expected value change of the dwelling may be considered too short for the aim of the calculation of the longterm affordability, but it does give an indication of the size of the effect. If the long-term value change of the dwelling is less, than the one estimated here (see Boelhouwer, 2000; Eichholtz, 1997), user cost will be higher than shown here, and may become positive at some point. The argument of working with long-term averages may be valid for other variables as well. For instance, the interest rates used, may be relatively low from a historic perspective. Finally, it is important to realize that the calculations here were done in nominal terms, meaning interest rates and expected value change were not corrected for general price inflation. If they were corrected, the effects would cancel out in the formula used. However, the costs of homeownership are sensitive to inflation, if included in the analyses, and if nominal interest payments are deductible in income tax. User cost turns out lower with higher inflation and higher with lower inflation (see Quigley & Raphael, 2004). Conclusions Affordability problems in housing have risen high on the agenda, especially of the media, in Flanders and The Netherlands. First-time movers are considered to be in trouble on the housing market on the one hand, and house prices have been rising enormously in both countries on the other hand. If cash flows of households necessary to access a dwelling are compared among households, outcomes may be different than if housing costs are compared. In the latter concept the opportunity costs of owner s equity and the expected house price appreciation are included. Thus the question about housing affordability can be asked in (at least) two ways depending on the aim of the comparison: what is housing affordability in the short-term; what is it in the long-term? Stated differently: what cash flows (expenses) do households need to pay for their housing in the short-term when they move to a (new) house (accessibility); what are the housing costs in the long-term taking into account also expected house value developments? The results for the short-term analyses for the different tenures show that on average the difference in short-term affordability between the rental subsectors and homeownership is bigger in Flanders than in the Netherlands. This effect mainly occurs because of the larger differences in disposable income between the housing market sectors in Flanders than in the Netherlands. These income differences are related to certain policy differences (e.g. income boundaries in social housing) and the different weight of the sectors in the housing market in Flanders and The Netherlands. This all adds up to the fact that in both countries housing is much more affordable in the short run among homeowners with a mortgage than tenants. In addition, regardless of a lower net rent in social housing, in Flanders as well as the Netherlands the residual income is on average higher in the private rented than in the social rented sector. The short-term results according to tertiles of equivalent disposable income, thus according to income corrected for household composition, show that disposable income is more strongly rising with equivalent income than net housing expenses. If we consider the ratio of the 18