A BUDGET APPROACH FOR COMPARING HOUSING AFFORDABILITY
Kristof Heylen1
Marietta Haffner2
1)
Higher Institute for Labour Studies (HIVA)
Catholic University of Leuven
Parkstraat 47, 3000 Leuven
tel. +32 16 323112
e-mail: [email protected]
2)
TU Delft
OTB Research Institute for Housing, Urban and Mobility Studies
Jaffalaan 9, 2628 BX Delft
tel. +31 15 2783523
e-mail: [email protected]
Paper to be presented at
Conference: Comparative Housing Research - Approaches and Policy Challenges in a New International Era
24 March 2010 - 25 March 2010 TU Delft, The Netherlands
A joint event between the European Network for Housing Research (ENHR) Working Group on Comparative Housing Policy and the Asia Pacific Network of Housing Research (APNHR). Marking the re-launch of the
European Journal of Housing Policy as the International Journal of Housing Policy.
1
Table of content
Abstract ................................................................................................... 2
1 Introduction .......................................................................................... 3
2 Indicators of affordability of housing ..................................................... 4
2.1 In general ....................................................................................... 4
2.2 Ratio approach ................................................................................ 4
2.3 Residual income and budget approach ............................................. 5
3 Methodology ......................................................................................... 7
3.1 Budget norms in Flanders and the Netherlands ................................. 7
3.2 Residual income and data ................................................................ 8
4 Results ................................................................................................ 10
4.1 Comparison of expenditure-to-income method and minimum budget
method ............................................................................................... 10
4.2 Results according to household characteristics .............................. 12
Conclusions ............................................................................................ 17
References ............................................................................................. 18
Appendix................................................................................................ 20
2
Abstract
In this paper the affordability of housing in the Belgian region of Flanders and the Netherlands is
analyzed, using both the expense-to-income ratio and residual income as indicators of affordability.
Residual income is obtained by deducting net housing expenses from disposable income, whereas the
former is the ratio of net housing expenses to disposable income. For both indicators we apply
affordability standards. In case of residual income we use minimum budget standards – excluding
housing - that allow for decent participation in society. Regarding the expenditure-to-income ratio we
apply the internationally frequently used 30% benchmark. In the methodological section the features
of both methods are compared, resulting in a preference for the combination of residual income and
budget standards. Then the results for both methods are presented, using Flanders and the
Netherlands as our cases. Thirdly, a further analysis using the budget approach is carried out for both
‘countries’ with emphasis on the differences between tenures and socio-economic groups.
Key words: affordability, residual income, budget standards, Flanders, the Netherlands
3
1 Introduction1
In recent years, the traditional method for analyzing housing affordability – the expenditure-to-income
approach – has been subject of continuous criticism for its limitations, more particular its
methodological weakness. However, in comparative research the ratio approach along with the
popular 30% benchmark keeps being used, partly because of a lack of suitable alternatives. The most
recommended alternative in literature – the budget approach for residual income - has often been
regarded as unfit to use in a comparative context. With the budget method, the after housing income
is compared with budgets needed for minimal participation in society. The method originated in the
context of poverty research but is getting more established in housing affordability studies as well
(Bradshaw, 1993; Saunders, 1998; Stone, 2006; Waite & Henman, 2005).
The minimum budgets are generally constructed by experts together with data on average spending
behaviour, for each possible household type. Often they are based on normative guidelines. When a
households income ‘after housing’ falls below the budget standard ‘after housing’, housing is regarded
‘unaffordable’. The budget standards are always drawn for a particular society and loose their
relevance in another time or place. Therefore, in a comparative context, each country should have its
own distinct standards needed for minimal societal participation. However, since there is no standard
method, the way the budgets are calculated differs between studies and hence between countries
(Saunders, 1998; Storms & Van den Bosch, 2009, Bradshaw, 1993). Up to now, no budget standards
were created for more than one country with exactly the same method.
This study outlines the possibilities of the standard budget method regarding the analysis of
affordability in different countries. The neighbouring ‘countries’ Flanders (Belgian region) and the
Netherlands will be our cases, as both have similar welfare levels and a budget standard available to
do the exercise. Affordability will be analyzed, using both the ratio and the budget approach. Next,
affordability according to the budget approach will be analyzed in depth for the two ‘countries’,
stressing the possibilities of an international comparison.
In the next section a discussion of the most commonly used affordability indicators is presented,
stressing the added value of the budget approach. In the methodological part, the application of
budget standards in Flanders and the Netherlands is presented, followed by the description of the
data sources and concepts at stake. In the fourth section, the results are presented.
1 We would like to thank Cor Lamain and Gust Mariën (TU Delft) for doing the calculations for the Netherlands.
4
2 Indicators of affordability of housing
2.1 In general
At least two concepts for measuring housing affordability can be distinguished in the literature. One is
the long-run affordability which is about the long-run ability to finance housing consumption, the user
costs of capital (Hancock, 1993). The more commonly used one, if it is about the financial accessibility
of a dwelling, has been called short-run affordability, as it is about the out-of-pocket expenses of a
household necessary to finance housing consumption at the moment of measurement.
Regardless of whether costs or expenses are used, most definitions in literature are similar in the fact
that reference is made to the connection between housing expenses and household income (Hancock,
1993; Freeman et al, 2000). Stone (2006) expresses it as follows:
‘Affordability expresses the challenge each Household faces in balancing the cost of its actual or
potential housing, on the one hand, and its non-housing expenditure on the other, within the
constraints of its income’.
Among others Hancock (1993) en Freeman et al (2000) argue that the definition of affordability can
not be constrained to the link between expenses and income, but that also standards for quality or
reasonableness could be included. These standards are about what in relation to quality is to be
considered affordable, and what in relation to income. For some parts of the housing market a
dwelling that is too small or lacks necessary comfort or too big may not be a free choice but a
necessity. A suitable dwelling of minimum quality against an affordable benchmark is often out of
reach. Maclennan & Williams (1990) include these aspects in their definition:
‘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’.
Despite the attention that this quality element gets in literature, it remains out of the scope of most
studies on housing affordability. For a theoretical elucidation on the aspect of housing quality related
to affordability we refer to the work of Hancock (1993) and Thalmann (1999). The standard for the
unreasonable burden on household income is yet often included in affordability studies. The
measurement of affordability ‘in the short run’ thus generally includes two elements: a measure for
the relationship between housing expenditure (the indicator) and a standard in relation to income to
judge this measure (Stone, 2006). In the following sections the two most applied and well-known
affordability indicators will be discussed: the ratio approach and the residual income method.
2.2 Ratio approach
The oldest and most well-known indicator for housing affordability is the ratio approach, which
expresses the ratio of the housing expenses to the available household income. In general, this
approach includes an affordability standard varying between 20 and 30%. When a households ratio
falls above this threshold, housing is assumed to be ‘not affordable’. Besides the use in housing
research, the ratio is also applied in the bank and social housing sector (Hulchanski, 1995). In
research the ratio is used to compare means between different groups, to evaluate an evolution over
time or to calculate the size of the group for whom housing is not affordable. In the financial sector
the ratio is applied to evaluate the liquidity of a potential mortgagor, in the form of borrowing
5
constraints. In many countries it is used as an admission criterion for social housing, as an element in
the rent calculation or as an element in the calculation of housing allowances. For instance in the U.S.
and Canada the 30% norm is applied as a maximum threshold as one of the admission criteria for
social housing (De Decker, 2005).
Many authors have however expressed severe criticism concerning the ratio method. A first critic is
that the ratio – since it is a relative measure – is not related to the absolute level of consumption. The
method does not take into account what households have left to spend on other necessary goods and
services. As a result, a same ratio does not have the same meaning for different income levels. It is
obvious that a rate of 30% has a different meaning for a single person earning 1.200 or 3.000 euros a
month. In general, a same ratio is less problematic for higher income groups. When a fixed
percentage is applied as an affordability norm, it would mean that the lower the income, the less one
needs to finance the necessary goods and services to participate in society, possibly not being able to
pay the full price of housing consumption. Applying the same fixed percentage across income groups
will obviously not be useful (Hancock, 1993; Freeman, Kiddle & Whitehead, 2000; Stone, 2006).
With the latter comment we arrive at another often expressed critic, namely that the ratio method
does not have a normative basic assumption. The applied standard (between 20% and 30%) is
chosen arbitrarily or is coming from non-scientific studies to expenditure patterns at the end of the
19th and the beginning of the 20th century. In that period there was a strong belief that the research
methods from the natural sciences were also suitable for the upcoming social sciences. Numerous
studies on housing expenditure patterns emerged with the aim to install ‘natural laws’ regarding
consumption behaviour (Hulchanski, 1995). Hulchanski argues that these studies lacked a conceptual
and statistical basis and the proven ‘laws’ should be regarded with scepticism. Yet, these studies
introduced the saying ‘a weekly pay for a monthly rent’ and thus a normative ratio of 25%. The share
that on average was spent on housing through the years became the norm for what people
‘reasonably’ can spend (Hancock, 1993; Stone, 2006).
A third critic is that the ratio method – using only one standard – does not account for expenditure
and income differences between different household types (Waite & Henman, 2005; Chaplin &
Freeman, 1998). These drawbacks of the ratio method call for cautiousness in applying the method. It
does not mean however that the approach should be totally abandoned. For instance, it can be
suitable to compare means or to analyze trends over time, without making estimate about the relative
size of the ‘groups for whom housing is not affordable’.
2.3 Residual income and budget approach
The residual income method with regard to housing affordability to a large extent overcomes the
criticism on the ratio method. The residual income is calculated by subtracting the housing expenses
from the available household income. Hence, this indictor expresses the consumption possibilities in
absolute terms after paying for housing (Hancock, 1993). The housing situation will be ‘not
affordable’, if the residual income is too low for the consumption of other necessary goods and
services. This reasoning finds his origin in the notion that housing expenses have a different nature
than expenses for e.g. food and clothing. It is for instance less easy – or in the short run even
impossible – to cut in the fixed housing expenditure than in most other expenditure. Housing
expenses make the biggest and least flexible claim on the household budget, which makes it logic to
analyze the income after deduction of rent or mortgage payment (Freeman et al, 2000; De Decker &
Van Dam, 2005).
In contrast to the ratio method, the residual income can be applied using meaningful affordability
standards. A first possibility is the application of norms that the government applies in the field of
poverty or housing policy. In some British studies for instance the norms for the Income Support
6
programme were used. These are suitable norms for residual income (RI) since housing expenses are
not included in determining the level of these benefits. For housing, in the UK, low income households
can rely on ‘housing benefits’ or ‘local housing allowances’ (Freeman, Kiddle & Whitehead, 2000; UK
Government, 2009). In Belgium however the amounts for social assistance benefits (Leefloon) are set
to finance all possible goods and services, and hence also the housing expenses. Therefore they are
not suitable for RI-standards.
A conceptually stronger starting point for drawing RI-standards is the minimum budget method. With
this method – per household type - a basket of minimal necessary goods and services is determined
that allows for minimal participation in society. Once the content of the basket is fixed, prices are
calculated and added to draw up the standards (Stone, 2006; Hancock, 1993; Freeman, Kiddle &
Whitehead, 2000). Clearly, this is an exercise demanding certain normative assumptions, which are
influenced by the society one lives in (Bourassa, 1996). Van den Bosch (1997) argues that ‘as long as
the choices are justified as good as possible, and one is aware that a budget is only valid for a certain
society and era, budget norms deliver interesting information.’ The minimum budget method
originated in the ‘40s of the previous century. Already in 1942 Beveridge applied budget standards
from a study of Rowntree to determine the minimum levels of ‘national assistance’. In 1985 several
prominent researchers of the York University carried out an extensive and influential study on the
methods of minimum budgets, resulting in budget standards for six family types. In a following phase
a difference was introduced between a ‘low cost’ and a ‘modest’ norm (Bradshaw, 1993).
In 1997 Van den Bosch applied the ‘low cost’ method for Belgium. Yet, these results are outdated by
now even if the amounts are indexed. A society is constantly changing, which leads to the
introduction of new goods and services in the ‘minimal basket’ (e.g. internet, mobile phone), whereas
others can be left out. For Flanders we can however rely on a recent study carried out by the
University of Antwerp and Catholic Hogeschool Kempen (Storms & Van de Bosch, 2009). In the
Netherlands, in 2000, the SCP drew up budget standards resulting in a ‘basic needs’ and a ‘modest
but adequate’ variation (Soede, 2006; Soede & Vrooman, 2008).
The budget method has several advantages compared to the ratio method for analyzing affordability:
it makes the link between housing and non-housing expenses explicit; it allows for more accurate
differences across household types and it is more useful to study low income households, as the focus
by definition set on that group. In case of a 30% ratio also high income families might be classified as
having an affordability problem. Waite & Henman (2005), who compared the ratio method (with the
30% norm) with the budget method for income support recipients, concluded that the ‘housing need’
estimates of the budget method better fit with households ‘poverty experience’ than the results from
the ratio approach.
7
3 Methodology
3.1 Budget norms in Flanders and the Netherlands
In the recent minimum budget study for Flanders the researchers took ‘A theory of human need’ of
Doyal and Gough as a starting point (Storms & Van den Bosch, 2009). This theory allows for a
uniform and well-thought drawing of budget standards and is one of the most well-known ‘need
theories’. The basic assumption of ‘A theory of human need’ is that health and autonomy are the two
most universal human needs who have to be fulfilled for a minimal participation in society. These
universal needs can be fulfilled by several products and services such as adequate food, clothing,
housing, health care, safe childhood and leisure time. Contrary to the universal nature of the basic
needs the satisfaction of needs is historically and culturally determined.
The study was performed in several phases. In a first phase the real expenditure of poor Flemish
households was studied, by performing data analyses, a literature study, a case study and several
focus groups with poor people. In the second phase of the study experts designed budgets for the
different ‘baskets’ of the total budget standard. As much as possible, the experts relied on existing
‘normative standards’, such as laws, official guidelines or recommendations concerning certain goods
and services. When no such norms were available, the experts had to evaluate whether the concerned
items are necessary to participate in society in a decent manner. In case of doubt, the inclusion of the
items was discussed in focus groups with low income families. In this study, the average expenditure
patterns were not regarded as a guideline. When the total budget standards per household type were
drawn by the experts, a final round of interviews with poor families was held, to evaluate whether the
amounts were reasonable within the framework of ‘minimal decent societal participation’.
In our study we apply the budget standards of the Flemish study with exclusion of the budget for the
fixed housing expenses. These amounts we will use as thresholds for the residual income.
In the Netherlands, for 2000, minimum budget standards were determined by the SCP (Sociaal
Cultureel Planbureau) on the basis of budgets drawn up by the Dutch National Institute for Budgetary
Information (Nibud). The Nibud norms are calculated by opinions of experts, the availability of goods
and actual consumption patterns at the bottom of the income distribution. Per type of expenditure
(clothing, food, transportation, …) minimum amounts were determined. The method is based on years
of experience and detailed knowledge of the expenses poor households are confronted with. SCP
applied the Nibud data in order to draw up two variants of the poverty line: a ‘basic needs’ and a
‘modest but adequate’ threshold. The terminology refers to the pioneer research of Bradshaw (1993).
The basic needs variant includes the expenditure that can be regarded as minimal necessary in the
Netherlands. It includes virtually unavoidable costs for food, clothing and housing. In 2000 the norm
for a single person was 670 euro per month. The norms for the other household types were calculated
using empirical equivalence factors, drawn by CBS (Central Bureau for Statistics). The basic needs
approach is a minimal one and does not allow for many ‘extras’. The ‘modest but adequate’ variant is
broader and includes normative costs for recreation, membership of library, sports or hobby club, a
subscription to a newspaper or a pet. In this study we will apply the Dutch norms of 2000, with
indexation for 2005. As mentioned before, social norms change over time, which might distort the
theoretical content of the concepts when five year old norms are applied. SCP therefore reconsiders
the content of the minimum budget each year, while linking the indexation to the evolution of median
prices (Soede & Vrooman, 2008).
The SCP method differs from the Flemish method in a few ways: first, there is no normative theory as
a starting point. Nibud applies normative ideas, yet without a theoretical reference framework as in
8
the Flemish study. Second, in Flanders, the minimum budgets are calculated separately for 16 possible
family types, whereas the Dutch study applied an equivalence scale to the budget for a single person.
Third, whereas in both studies the opinion of experts was hired, only in the Flemish study the opinion
of the people at the bottom of the income distribution was heard.
In contrast to the Dutch study, the Flemish study only calculated one budget type, which should allow
people for minimal and decent participation in society. The Flemish budget is conceived as a minimum
budget in order to decently participate in society and is therefore more in line with the Dutch ‘modest
but adequate’ than the ‘basic’ variant. This is also shown by the content of the Flemish budget, which
includes amounts for recreation, membership of sports or hobby clubs, modest travels, DVD and
television.
The budget standards are shown in table 1. The minimum for a single person in the Flemish study
amounts to 916 euros for a man and 917 for a woman, whereas this amount is almost at the same
level for the Dutch modest variant (870 euros). The minimum for a single in the Dutch basic variant is
much lower, with 770 euros. Also for couples the Flemish minimum and the Dutch modest minimum
are almost alike (1216 versus 1190 euros), whereas the basic variant ends up considerably lower
(1050 euros).
A comparison between the two countries is thus from a theoretical point of view preferably carried out
using the Dutch ‘modest but adequate’ approach. This also is the conclusion from a more pragmatic
point of view, given the fact that the prices for many goods and services do not differ much between
Flanders and the Netherlands. The Flemish budget standard and the Dutch modest budget standard
should allow for a considerable similar living standard.
Table 1: Budget standards for disposable income, in euros per month, according to
household type, Flanders and the Netherlands, 2005.
Household type Flanders The Netherlands,
SCP, basic needs
The Netherlands,
SCP, modest but adequate
Single 916 (man)
919 (woman) 770 870
Couple 1216 1050 1190
Sources: Storms & Van den Bosch (2009), Soede & Vrooman (2008)
3.2 Residual income and data
The norms in table 1 are used for calculating the norms of residual income. In both the Flemish and
Dutch study initially a normative rent was included in the minimum budget as owner-occupation is not
considered as a minimal housing situation. In practice, in both studies, the rent amount was not
‘normative’ but based on median or average rent. More in particular, ‘gross’ rent was included in both
studies, meaning that housing allowances were not taken into account. The RI-norms are hence
calculated by taking the gross rent out of the minimum budget. In appendix 1 and 2 the RI-thresholds
are shown for Flanders and the Netherlands respectively.
In our study we relied upon survey data in order to determine the scope of the group that falls below
the standards of residual income. For Flanders we used the Housing Survey 2005, which is a detailed
survey on the housing situation of 5.216 Flemish households. The Dutch data source is WoON 2006, a
9
survey of the Ministry of VROM, including more than 60.000 households. In the WoON 2006 the
income data of the tax administration of 2005 are coupled to the survey data.
Table 2 shows how housing expenses and residual income are calculated for Flanders and the
Netherlands. Net housing costs are taken into account when calculating residual income. Since we
want to calculate how ‘affordable’ housing is, it is logic to take account of subsidies that are directly
linked to the housing situation.
In case of tenants, residual income is calculated by subtracting net rent from disposable income. Net
rent is taking housing allowances into account, which especially in the Netherlands have a significant
impact (Heylen & Haffner, 2009). Concerning owner occupation, the residual income is computed by
subtracting net housing expenses from disposable income. Net housing expenses take the fiscal effect
of the income tax into account, which in both ‘countries’ have a substantial effect. Further, the
housing expenses of owner-occupiers include not only mortgage payments and interests, but also the
owner’s share of maintenance costs and property tax. These latter costs are included in order to make
expenses of homeowners better comparable with expenses in the rental sector. For tenants these
costs are supposed to be included in gross rent. Since the owner’s share of maintenance costs and
property tax are included, also outright owners will have housing expenses.
In our exercise the utility costs (for heating, water, electricity) are not subtracted from disposable
income in order to calculate residual income. This means that the RI standards include normative
amounts for utility costs.
Table 2: 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.
10
4 Results
4.1 Comparison of expenditure-to-income method
and minimum budget method
The aim of this section is to compare the affordability figures for the ratio and budget approach in
both Flanders and the Netherlands. More specifically, we search for mechanisms that are present in
both countries. Table 3 shows the shares of households with a housing-to-income ratio above 30%
and a residual income below the threshold. Whereas for Flanders one standard for the residual income
is used, two thresholds are analyzed for the Netherlands. We start by discussing the Flemish results.
In Flanders, according to the budget method, housing is not affordable for 14,6% of Flemish
households compared to 15,8% using the ratio method. These figures are quite similar, but according
to tenure the differences are remarkable. Following the budget method, social housing is the most
problematic tenure in terms of affordability, as around 39% of social tenants fall below the RI-
standard. Oppositely, according to the ratio method, social housing is the least problematic segment
(12,2%). Since in Flanders social rent is related to income and in 2005 the maximum rent was set at
20% of taxable income, it is logic that relatively few social tenants fall above the 30% norm. As
previous research already pointed out (Heylen et al, 2005), the private rented sector has relatively the
most affordability problems according to the budget method, with 39% above the 30% standard. The
share of private tenants below the RI-norm is considerably lower (27%). Further, the affordability for
owner-occupiers with a mortgage has a better outcome with the budget method (11%) than the ratio
method (23%). To the contrary, affordability for outright homeowners turns out worse when the
budget method is applied.
For the Netherlands, the ratio method classifies relatively more households having an affordability
problem than the modest and minimum budget methods. In case of the latter, housing is only
unaffordable for 8% of households. Following the modest budget method, the results are more in line
with the ratio method, especially in the rented sector. For the reasons outlined before, we will
concentrate on the modest budget results for the Netherlands in the rest of our analyses.
As in Flanders, the ratio and budget method affect the relative position of the rental sectors with
regard to affordability. According to the ratio method, housing is not affordable for 29% and 19% of
tenants in respectively the private and rented sector. When the budget method is applied, the relative
position of private tenants improves, with a mere 19% below the RI-norm, while still 22% of social
tenants are presumed to have an affordability problem. Also like in Flanders, the relative position of
homeowners with a mortgage compared to tenants improves when the ratio method is replaced with
the budget method. As housing is not affordable for 24% of homeowners with a mortgage, this share
falls to 8% when the modest budget method is applied. Finally, regarding the outright owners, the
share of affordability problem is higher with the modest budget method than with the ratio method.
This implies that the relative position of outright owners is worse according to the budget method.
In sum, in both ‘countries’, compared to the ratio method, the budget method improves the relative
position of homeowners with a mortgage and tenants on the private market. Logically then, the
relative position of tenants on the social rented market and outright homeowners is worse.
11
Table 3: Household with a residual income below the RI-norm and housing
expenditure-to-income ratio above 30%, according to tenure, Flanders and the Netherlands, 2005/2006
Flanders The Netherlands
Ratio > 30% RI < RI-norm Ratio > 30%
RI < RI-norm
Minimum budget
Modest budget
Total 15.8 14.6 21.3 8.0 13.9
Owner-occupation 10.4 9.6 21.5 5.8 8.2
With a mortgage 23.3 11.3 24.1 6.1 8.4
Without a mortgage 0.5 8.2 4.5 4.0 6.4
Rented sector 33.0 30.0 21.2 10.6 21.2
Private rented 39.2 27.4 29.4 12.4 19.6
Social rented 12.2 38.6 19.0 10.2 21.6
Test *** *** *** *** ***
N total1 4736 4531 54272 54272/3 54272/3
Sources: Woonsurvey 2005 (Flanders); WoON 2006, OTB/TU Delft calculations (the Netherlands)
²-test: ***p< 0.01
1: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
How can the remarkable difference between the two methods be explained? Therefore we look at the
percentages for whom housing is unaffordable per income quintile and tenure, for both countries.
Table 4 presents the situation for Flanders. When all households are regarded, 25% belonging to the
first quintile is having an affordability problem according to the ratio method. This percentage is
somewhat lower for the second and third quintile, but is still at a level of 13% and 8% in respectively
the fourth and fifth quintile. In the rental market even 18% belonging to the fourth quintile falls above
the 30% threshold.
When the budget method is applied, 59% of households in the first income quintile is presumed to
have an affordability problem. For the rental sector separately, this share even rises to 86%. In the
third quintile the overall share below the RI-norm is not more than 2%, whereas almost no household
in the fourth quintile is having a residual income below the norm.
For the Netherlands the figures are similar. According to the ratio method, 39% of households in the
first quintile and still 17% in the fourth quintile fall below the 30% threshold. As a contrast. using the
budget method, 57% of households in the first quintile are supposed to have an affordability problem,
while this share drops to respectively 3% and 1% in the third and fourth quintile.
These figures indicate that the budget method primarily identifies lower income households as having
an affordability problem, whereas with the ratio method easier classifies higher income households in
the ‘problematic group’. The focus of the budget approach is more on the income side – and is in a
way a device to identify poverty - whereas the ratio approach focuses more on the housing expenses.
The differences between the tenures resulting from the different methods, stem from these income
effects. In the Netherlands and especially in Flanders, the tenants in the social rented sector are
overrepresented among the lower income groups (Heylen & Haffner, 2009), which results in a worse
affordability score when the budget method is applied. On the other hand, homeowners with a
12
mortgage are overrepresented in the higher income quintiles, resulting in a better affordability picture
when the ratio method is applied.
Table 4: Households with a housing expenditure-to income ratio > 30% and a residual
income below the RI-norm (in %), according to tenure and income quintiles,
Flanders and the Netherlands, 2005
Quintiles of equivalent income1
Flanders The Netherlands
All households
Owner-occupation
Rented sector
All households
Owner-occupation
Rented sector
Housing expenditure-to-income ratio > 30%
1 25.1 11.4 50.6 39.4 38.3 39.6
2 19.2 10.5 40.9 22.6 31.4 17.8
3 14.6 10.8 24.8 17.2 25.4 6.1
4 13.1 12.0 18.2 16.5 20.6 2.2
5 8.0 8.3 6.8 10.9 12.0 2.3
Total 15.8 10.4 33.0 21.3 21.5 21.2
Test *** *** *** *** *** ***
Budget method: RI < RI-norm
1 59.0 45.3 85.5 57.4 52.7 59.1
2 11.5 7.5 21.9 7.7 12.1 4.6
3 2.1 1.6 3.5 3.0 4.9 0.5
4 0.9 0.9 0.8 1.0 1.3 0.1
5 0.3 0.3 0 0.4 0.4 0
Total 14.6 9.6 30.0 13.9 8.2 21.2
Test *** *** *** *** *** ***
N2 4530 3364 1099 54272 30299 23972
Sources: Woonsurvey 2005; WoON 2006, OTB/TU Delft calculations
²-test: ***p< 0.01
1: calculated for all households
2: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
4.2 Results according to household characteristics
Table 5 shows the percentage of households below the residual income norm according to activity
status of the head of household, for both Flanders and the Netherlands. The results are remarkable
similar for both ‘countries’. Employment appears to be a crucial factor. In case of an unemployed head
of household, housing is not affordable for respectively 56% and 59% in Flanders and the
Netherlands compared to respectively 8% and 9% in case of employment. The unemployed tenants
are the groups for whom the affordability is worst, with a share below the RI-standard of 71% in
Flanders and 63% in the Netherlands. Further, the share below the RI-norm is relatively high for sick
and disabled people, reaching a level of 33% in both ‘countries’. Thus, the results suggest that the
13
replacement income for the sick and disabled is often not high enough for worthy participation in
society.
The relative position of (early) pensioners is shown to be better in the Netherlands than in Flanders, in
particular in the rented sector. In Flanders, the share of pensioned tenants below the RI-threshold is
20 percent point higher than the share of working tenants below the standard. In the Netherlands this
gap is only 1 percent point. Also in case owner-occupation. the gap between the working and
pensioned people is bigger in Flanders than in the Netherlands.
Table 5: Households with a residual income below the modest RI-norm (in %),
according to tenure and activity status of the head of household, Flanders and
the Netherlands, 2005/2006
Activity status head of household
All households Owner-
occupation Rented sector
Flanders
Working 8.5 6.7 14.7
(Early) pensioned 15.5 10.5 34.4
Unemployed 55.6 38.7 70.6
Sick/disabled 32.2 21.1 46.4
Test *** *** ***
N 4530 3364 1099
The Netherlands
Working 9.5 7.6 13.3
(Early) pensioned 11.2 7.4 14.2
Unemployed 58.7 25.7 63.2
Sick/disabled 32.7 20.4 36.6
Test *** *** ***
N1 54272 30300 23972
Sources: Woonsurvey 2005; WoON 2006, OTB/TU Delft calculations
²-test: ***p< 0.01
1: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
The affordability results according to household type are shown in table 6. Not surprisingly, in both
‘countries’ the couples – and especially those without children – do better than the singles and lone
parents. The lone parents are the group for whom affordability is most under pressure, with a
percentage amounting to 36% and 25% for respectively Flanders and the Netherlands. In Flanders
46% of lone parents in the rental sector fall below the RI-threshold which is 9 percent point higher
than for couples with minimum one child. In the Netherlands this gap is about 10 percent point.
Furthermore, on both the Flemish and Dutch rental market the position of couples with children is
weaker than for couples without children. For homeowners however, the results differ. In Flanders –
as opposed to the Netherlands - housing is relatively more unaffordable for couples without children
than for couples with children. This can be explained by the relatively high share of elderly among the
couples with no children among the Flemish homeowners.
14
Table 6: Households with a residual income below the modest RI-norm (in %),
according to tenure and household type, Flanders and the Netherlands,
2005/2006
Household type All households Owner-
occupation Rented sector
Flanders
Single 16.1 6.3 31.3
Lone parent 36.3 29.4 45.7
Couple, no child 8.5 7.0 14.7
Couple with child(ren) 14.5 4.2 37.0
Test *** *** ***
Total 4530 3364 1099
The Netherlands
Single 21.1 11.8 25.8
Lone parent 25.4 19.0 28.4
Couple, no child 6.2 4.8 8.7
Couple with child(ren) 10.4 7.9 18.6
Test *** *** ***
N1 54273 30298 23974
Sources: Woonsurvey 2005; WoON 2006, OTB/TU Delft calculations
²-test: ***p< 0.01
1: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
Table 7 points out that the educational level is – of course in an indirect way – strongly related to
housing affordability in both ‘countries’. The higher the educational level, the lower the percentage
that falls below the RI-norm. The category of ‘higher education’ is the only one that is perfectly
comparable between Flanders and the Netherlands. These are the heads of household that gained a
polytechnic (‘hogeschool’) or university degree. In both Flanders and the Netherlands this group is
better off than the other educational groups in terms of affordability. For the tenants that only
completed elementary school in Flanders – which to a large extent are elderly – affordability is
strongly under pressure, with a level of 42% below the RI-norm. Here the difference with the owner-
occupiers (14%) is remarkable. Surprisingly, a significant share of the higher educated in the rental
sector fall below the standard for residual income, with respectively 14% and 18% in Flanders and the
Netherlands. Research in Flanders pointed out that about one third of the tenants on the private
market conform to the profile of relatively young, higher educated people that are in a transition
period to owner-occupation (Le Roy et al, 2008). Not for all of them this transition period appears to
be ‘affordable’.
15
Table 7: Households with a residual income below the modest RI-norm (in %),
according to tenure and educational level, Flanders and the Netherlands,
2005/2006
Highest educational level All households Owner-
occupation Rented sector
Flanders
Elementary school 22.1 13.8 41.8
Lower secondary school 16.9 12.8 29.2
Higher secondary school 14.7 9.5 29.4
Higher education 6.0 4.2 14.3
Test *** *** ***
Total 4530 3364 1099
The Netherlands
Low (vmbo) 17.4 10.4 22.5
Middle (havo, mbo, vwo) 13.6 9.1 20.8
Higher education 9.0 5.3 17.7
Test *** *** ***
N1 54272 30301 23973
Sources: Woonsurvey 2005; WoON 2006, OTB/TU Delft calculations
²-test: ***p< 0.01
1: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
Table 8 shows the link between the age of the head of household and the affordability of housing.
When all households are regarded, the results point in a different direction in Flanders and the
Netherlands. In Flanders, the group ’65 and older’ has the highest share below the residual income
threshold, whereas in the Netherlands the youngest group scores the worst. Then when we only
compare the rental sector, the results for both ‘countries’ are rather similar. There are little differences
between the age groups until 64, whereas the oldest group is relatively well-off.
Regarding the owner-occupied sector, in Flanders the youngest group is the least problematic in terms
of affordability, while it is the most problematic in the Netherlands. In Flanders, contrary to the
Netherlands, the oldest homeowner group has the largest share below the RI-threshold (11%). Only
4% of the ‘65-plus’ Flemish homeowners is still paying off a mortgage. This means that at least 7% of
the oldest group has a residual income below the RI-threshold without paying off a mortgage.
16
Table 8: Households with a residual income below the modest RI-norm (in %),
according to tenure and age of head of household, Flanders and the
Netherlands, 2005/2006
Age categories All households Owner-
occupation Rented sector
Flanders
17-34 years 15.9 7.8 24.9
35-44 15.0 10.9 29.2
45-64 12.2 8.2 28.2
65 and more 17.1 11.2 9.5
Test *** *** ***
N 4530 3364 1099
The Netherlands
17-34 years 18.7 10.1 26.0
35-44 14.2 8.8 24.0
45-64 13.0 7.5 22.7
65 and more 11.2 7.1 14.2
Test *** *** ***
N1 54272 30301 24973
Sources: Woonsurvey 2005; WoON 2006, OTB/TU Delft calculations
²-test: ***p< 0.01
1: Weights were applied to the Dutch sample to obtain results for all Dutch households (almost seven
million). The weighted totals are given.
17
Conclusions
The aim of this contribution was to analyze the budget approach for comparing housing affordability
across households. This happens in two ways. First we study the advantages of the approach in
comparison with the ratio approach. Second, we analyze whether the approach can be used
comparatively across countries.
From the discussion of the ways of measuring the price or rent of housing consumption, it follows that
a ratio approach –for example 30% of income spent on housing is considered affordable– regardless
of income level or household size can never be a very precise indicator of whether households can
afford the access to their dwelling. It can however be used to observe whether households in due
course are needing to pay more or less over time for their housing, for example.
A budget standard starts from the other end, determining first what cash amounts different household
types need minimally for their ‘other’ consumption next to housing. Then it determines whether
residual income, the disposable income left after housing expenses, is sufficient for this other
consumption. If not, housing is considered unaffordable for a household.
When applying both standards to Flemish and Dutch housing expenses in 2005, we find that the
results differ. Especially, for the rental sector these results are relevant. With the ratio approach
housing consumption is considered unaffordable for larger shares of private tenants than social
tenants, while with the budget approach, it is the other way around. This effect is stronger in Flanders
than the Netherlands. As one must assume that the budget approach is more precise in signalling
which households cannot afford their housing consumption, it would seem that this result has quite
some policy consequences.
Regarding the relationship between household characteristics and affordability, the results are to a
large extent the same in Flanders as in the Netherlands. The unemployed, low educated, sick/disabled
or lone parents tend to score relatively bad on affordability in both countries. According to age the
results differ. The youngest group (17-34) has the worst affordability position in the Netherlands,
compared to the oldest group (65 +) in Flanders. In both countries, the results significantly differed
according to tenure.
On whether the budget approach can be used for comparison across countries, the conclusion must
be that it in principle cannot be used for direct comparisons, unless the method of composition is the
same. The levels of necessary budgets may differ. However, the budget approach for residual income
can be useful in a comparative setting, even when the methods for drawing the budget standards
differ between the countries. As outlined in our paper for Flanders and the Netherlands, the
comparison can be focused on the differences in relative position of groups, as in this case tenants
and owner-occupiers. Also for comparison of the relative position of groups according to age, activity
status, households status, etc. the approach can be valuable.
Affordability however, is not only about the price of housing consumption and a threshold of
reasonableness in relation to income, it ideally should also be about the quality of housing being
consumed in relation to needs. It would show whether the price and income thresholds are related to
the same or different quality that is being consumed. We did not focus on this aspect in this
contribution. One way to extend the analyses would be to also show (if possible) what type of quality
is being consumed. Ideally, there would be a threshold for decent quality in relation to household size.
In such a complex analysis, one could then distinguish between different groups of households, those
consuming affordable verses not affordable housing in combination with the consumption being of
decent quality verses not decent quality.
18
References
Bourassa. S.C. (1996). Measuring the Affordability of Home-ownership. Urban Studies. Vol. 33. No 10.
pp. 1867-1877.
Bradshaw. J. (1993). Budget standards for the United Kingdom. Aldershot: Avebury.
Chaplin. R. & A. Freeman (1999). Towards an Accurate Description of Affordability. Urban Studies.
Vol. 36. No. 11. pp. 1949-1957
De Decker. P. & Van Dam. R. (2005). De huisvestingsval. of wat huishoudens overhouden na het
betalen van hun woonkosten. In: De Decker. P. Goossens. L. & I. Pannecoucke (ed.). Wonen aan de
onderkant. Garant: Antwerpen – Apeldoorn. pp. 93-1112.
Doms. K. B. van Damme. S. Winters. V. Bilsen & E. Buyst (2001). Op zoek naar eigendom. Een
onderzoek naar de overheidssteun voor eigenaars van woningen. K.U.Leuven. Hoger Instituut voor de
Arbeid.
Freeman. A. Kiddle. C. & C. Whitehead (2000). Defining affordability. In S. Monk & C. Whitehead
(ed.). Restructuring housing systems: from social to affordable housing?. Joseph Rowntree
Foundation. York. pp. 100-105.
Haffner. M. & K. Heylen (2008). Betaalbaarheid van het wonen. Gebruikskosten in Vlaanderen en
Nederland. Steunpunt Ruimte en Wonen. Leuven. 22 p.
Hancock. K.E. (1993). ‘Can Pay? Won’t Pay?’ or Economic Principles of ‘Affordability’. Urban Studies.
30(1). pp. 127-145.
Heylen. K. Le Roy. M. Vanden Broucke. S. Vandekerckhove B. & S. Winters S. (2007). Wonen in
Vlaanderen. De resultaten van de Woonsurvey 2005 en de Uitwendige Woningschouwing 2005.
Departement Ruimtelijke Ordening. Woonbeleid en Onroerend Erfgoed. Brussel.
Heylen. K. & Winters. S. (2008). Betaalbaarheid van wonen voor starters op de woningmarkt.
Bijkomende gegevens op basis van de Woonsurvey 2005. Steunpunt Ruimte en Wonen. Leuven. 48 p.
Heylen K. & M. Haffner (2009). Inkomensongelijkheid en woonuitgaven in Vlaanderen en Nederland.
Tijdschrift voor Sociologie. Vol. 30(2). pp. 143-175.
Hulchanski. D. J. (1995). The Concept of Housing Affordability: Six Contemporary Uses of the Housing
Expenditure-to-Income Ratio. Housing Studies. 10 (4). pp. 471-492.
Le Roy M. Debusschere E. Heylen K. Vandorpe L. en Vandekerckhove B. (2008). Onderzoek naar
woningmerktsegmenten. Steunpunt Ruimte en Wonen. 177 p.
Maclennan. D. & Williams R. (Eds.) (1990). Affordable housing in Britain and the United States. York:
Joseph Rowntree Foundation.
Rowntree, B. S. (1941), Poverty and Progress: A Second Social Survey of York, London: Longmans.
Saunders, P. (1998), Using budget standards to assess the well-being of families, Social Policy
Research Centre (SPRC), Australia, Discussion Paper No 93.
Soede A. & C. Vrooman (2008). Beyond the breadline. A poverty threshold based on a generalised
budget approach. The Netherlands Institute for Social Research SCP. The Hague.
19
Soede. A. (2006). Naar een nieuwe armoedegrens? Basisbestedingen als maatstaf voor een
tekortschietend inkomen. Sociaal en cultureel Planbureau. Den Haag.
Stone. M. (1998): Shelter poverty. in: Van Vliet. W. (ed.). The encyclopaedia of housing. Sage.
Thousand Oaks/London. p. 534-535.
Stone. M. E. (2006). What Is Housing Affordability? The Case for the Residual Income Approach.
Housing Policy Debate. 17 (1). pp. 151-184.
Storms. B. & K. Van den Bosch (eds.) (2009). Wat heeft een gezin minimaal nodig? Een
budgetstandaard voor Vlaanderen. Acco Leuven/Den Haag.
Thalmann. P. (2003). ‘House poor’ or simply ‘poor’?. Journal of Housing Economics. vol. 12. pp. 291-
317.
UK Government (2009). Income Support [http://www.direct.gov.uk].
Van Dam. R. Geurts. V. & I. Pannecoucke (2003). Housing tenure. housing costs and poverty in
Flanders (Belgium). Journal of Housing and the Built Environment. Vol. 18 pp. 1-13.
Van den Bosch. K. (1997). Wat heeft een gezin nodig om rond te komen? Budgetnormen voor drie
typegezinnen. CSB Bericht. UFSIA.
Waite G. & P. Henman (2005). Applying Budget Standards to Assess Housing Affordability. paper
presented at the National Housing Conference. Perth. Australia.
Whitehead C. (1991). From Need to Affordability: An Analysis of UK Housing Objectives. Urban
Studies. Vol. 28. No 6. pp. 871-887.
20
Appendix
Appendix 1 Standards for residual income (=budget standards excluding housing), in
euro per month, according to households type, Flanders, 2005
Number of children Lone parent Couple Single
0 - 837 541 (man)
538 (woman)
1 838 1104 -
2 1059 1316 -
3 1280 1528 -
4 1501 1740 -
5 1722 1952 -
6 1943 2164 -
Source: Storms & Van den Bosch (2009), own calculations
Appendix 2 Standards for residual income (=moderate but adequate budget standards
excluding housing), in euro per month, according to households type, the Netherlands, 2005
Number of children
Lone parent Couple Single
Number of
children < 4 Number of
children >= 4 Number of
children < 4 Number of
children >= 4
0 - - 930 930 620
1 806 930 1116 1240 -
2 992 1240 1302 1550 -
3 1178 1550 1488 1860 -
4 1364 1860 1674 2170 -
5 1550 2170 1860 2480 -
6 1736 2480 2046 2790 -