Spatial analysis of housing stress estimation in Australia with statistical validation.
Rahman, Azizur ; Harding, Ann
1. INTRODUCTION
Housing stress has emerged as a widely discussed public policy
issue among politicians, academics and policy makers in Australia. With
the unprecedented growth in housing prices - and rents - throughout the
past decade, many Australians are increasingly finding housing
unaffordable (Rahman, 2011; Yates, 2011). Between 1995 and 2005, real
house prices in Australia increased by more than 6 percent per year,
with an average annual increase of almost 15 percent from 2001 to 2003
(Yates, 2011). This was well above the average annual increase in the 20
years to 1995 of just 1.1 percent and the 50-year average (from 1960 to
2010) of 2.5 percent per year. These data are illustrated in Figure 1
and contrast with the significantly slower growth in Gross Domestic
Product (GDP) per capita and average earnings over much of the period. A
significant increase of the real house prices is marked from 2001
onwards.
[FIGURE 1 OMITTED]
Compared with other economically advanced nations, Australia is
often reported as having experienced relatively rapid growth in real
house prices over the past 20 years or so (Tumbarello and Wang, 2010).
Just over the five year period from 2000 to 2004, Australia had the
third highest rate of house price inflation among Organisation for
Economic Co-operation and Development (OECD) member countries, ranking
behind only Britain and Spain (Productivity Commission, 2004; The
Economist, 2011). Moreover, a recent report of the Australian Bureau of
Statistics (ABS) shows that established house prices increased by an
average of 33 percent between 2002-03 and 2006-07 (ABS, 2008). Within
this time period house rents have also increased rapidly. For instance,
within only a 12 month period ending in August 2007, house rents
increased in Perth by 36.4 percent, Melbourne by 23.4 percent,
Australian Capital Territory by 22.7 percent, Sydney by 18.8 percent,
and Brisbane by 13.5 percent (Pearson, 2007). So, housing stress has
become an important financial challenge for households, especially for
low and middle income groups and an important public policy concern for
the national, state and local governments.
About 1.7 million people in this country are in housing stress
(Sandel and Wright, 2006). Households with relatively low income and
housing costs greater than a certain proportion of household income (for
instance, more than or equal to 30 percent) are typically defined as
being in housing stress (Rahman, 2009). The concept may also be extended
to describe inadequate housing for a proportion of the population. Most
of the policy debates on housing stress to date have been confined to
the national or state level (Wood et al., 2005; Harding et al., 2004;
Nepal et al., 2010; Rahman, 2011; Flood, 2012). This is largely due to
the ready availability of data at this coarse geographic level in the
sample survey files available from the ABS. However, methodological
advances in spatial microsimulation modelling mean that it is now
possible to generate synthetic spatial micro-population data (Rahman et
al., 2010a).
As in many other countries, substantial spatial differences in
socioeconomic growth and wellbeing exist across Australia (Chin et al.,
2005; Harding et al., 2006; Stimson et al., 2008). Australian housing
programs include subsidising housing costs and rent assistance; mortgage
subsidies; and land development planning for housing. All of these
policies have had significant impacts on individuals and their living
standards, experiences, choices, constraints, decisions and lifestyle
preferences (Melhuish et al., 2004; Kelly et al., 2006; Rowley and Ong,
2012; Rahman et al., 2013). In addition, housing acts as a proxy for a
host of other factors relevant to economic disadvantage and social
inequalities at small area levels. Small area level housing stress
statistics also vary with the demographic and socioeconomic conditions
of households - and with geography (Rahman, 2011). So, there is a keen
interest in understanding who is struggling to afford to buy or rent a
house and the impact at small geographic area levels.
This paper studies a spatial analysis of the estimation of
statistical local area (SLA) level housing stress in Australia. One of
the arguments frequently evoked in the literature is that
microsimulation modelling technology based small area estimation lacks
vigorous tests of statistical reliability for the microsimulated
estimates. So this paper also offers a new statistical approach for
validating the results of small area housing stress statistics.
2. A REVIEW OF THE LITERATURE
Typically housing stress describes a financial situation of
households where the cost of housing--either as rental, or as a mortgage
repayment is considered to be significantly high relative to household
income. A range of definitions for describing the situation of housing
stress are available in the literature. The following subsections will
discuss all methods of measuring housing stress and compare different
definitions.
Measures of Housing Stress
Housing stress can be measured by combining two basic quantities -
the income and expenditure of a household. A household can be considered
under housing stress when it is spending more than an affordable
expected proportion of its household income on housing. The affordable
expected cut-off point of housing expenditure can vary with the
circumstance of households as well as location of dwelling.
As a general rule of thumb, a household spending at least 30
percent of its income on housing can be considered under housing stress
(see King, 1994; Landt and Bray, 1997). Some researchers use a different
threshold of housing expenditure by restricting the definition to
households within different income quintiles. For example, an income
threshold of more than 25 percent for housing costs is used by the
National Housing Strategy (1991) and Foard et al. (1994). Additionally a
commonly used definition of housing stress is specified in Harding et
al. (2004), where a threshold of more than 30 percent of housing costs
was used, but only for those households having income in the bottom 40
percent (lowest two quintiles) of the equivalised income distribution.
Another definition restricts the designation of 'being in housing
stress' to those households spending more than 30 percent of their
income on housing and belonging to the bottom 10th to 40th income
percentile of the income distribution (ABS 2005). It is noted that any
threshold-based definition is an arbitrary slice through a continuum,
meaning that small area level estimates of a percentage of households in
housing stress would be better treated as estimates of small areas with
the greatest percentage of households in housing stress. More
explicitly, if an area has a very high percentage of households
suffering from housing stress under one of the above definitions, the
area probably ranks highly on percentage of households suffering from
housing stress however defined.
The residual income approach to housing stress measure looks at
what different household types can afford to spend on housing after
taking into account the other necessary expenditures of living (Stone et
al., 2011). Although it is an alternative to benchmarking the income and
expenditure ratio measures of housing stress commonly used in Australia,
this approach requires an operationalised residual income standard that
is not only difficult to quantify but also arbitrary according to
varying circumstances of households. This means that a household has a
housing related financial stress problem if it cannot meet its
non-housing related needs at some minimum level of adequacy after paying
for housing (Stone, 2006a). The appropriate indicator of the tension
between housing costs and incomes is thus the difference between them -
the residual income after paying for housing, rather than the ratio of
costs to income.
Defining a residual income standard involves use of a
socially-defined standard of adequacy for non-housing items. Thus, while
the residual income logic has some conceptual broadness, a particular
residual income standard is not universal, but socially grounded in
space and time (Stone, 2006b; Stone et al., 2011). Issues involved in
selecting such a standard for non-housing necessities can be difficult
and complex.
Both the ratio approach and the residual income approach suggest
that as the housing costs behaviourally tend to make the first claim on
disposable income, a household has a housing stress problem if, after
paying for housing, it has insufficient (residual) income to meet its
nonshelter needs at some normative level of adequacy. The difference
between the two approaches is how they define the normative level of
adequacy for non-shelter items. The ratio approach defines it as a
fraction of income: traditionally 75 percent. More recently 70 percent
has been defined as the minimum share of income that must be available
after housing costs in order to avoid hardship in meeting non-shelter
needs (Nepal et al., 2010; Rahman, 2011). By contrast, the residual
income approach defines the normative level of adequacy for non-shelter
items as a monetary amount that is independent of income but very
dependent upon household composition and the non-housing cost of living
as a function of time and place (Burke et al., 2010).
Types of Ratio Measures
A rationale for the use of the 30/40 rule based ratio measure is
given in this subsection. It is noted that this ratio measure not only
provides continuity with traditionally used measures, but also it is
simple to apply and easy to understand.
The definitions of housing stress by three 'rules'-based
ratio measures are as follows:
1) 30-only rule: A household is considered to be in housing stress
if it spends more than 30 percent of its disposable or gross income on
housing costs;
2) 30/40 rule: A household is considered to be in housing stress if
it spends more than 30 percent of its disposable or gross income on
housing costs and the household also belongs to the bottom 40 percent of
the equivalised disposable income distribution; and
3) 30/(10-40) rule: A household is considered to be in housing
stress if it spends more than 30 percent of its disposable or gross
income on housing and falls into the bottom 10th to 40th income
percentile of the equivalised disposable income distribution.
Although the cut-off point of housing costs for all these
definitions is the same, there are some concerns associated with each of
these rules. For example, is gross income or disposable income the
appropriate base income to calculate housing costs for measuring housing
stress? (Gross income is the income of a household from all sources
before deducting tax and the Medicare levy, whereas disposable income is
the income that remains to a household after deducting the estimated
personal income tax and the Medicare levy from gross income.) If a
researcher uses 30 percent of gross income as a base, then after
possible deductions that figure may be around 40 to 45 percent of actual
disposable income. Hence, 30 percent of gross income should equate to a
reasonably high proportion of actually received income for housing and
other costs. In addition, the 30/40 and 30/(10-40) rules both restrict
the definition to those households that are within the bottom 40 percent
of the equivalised income distribution. The issue here is: why is the
cut-off point at the lowest 40 percent of income distribution? For the
latter rule, why are households in the bottom 10 percent of the
equivalent income distribution being omitted?
In general, when the individuals have a higher income, they have
greater choice in how to spend it. For lower income households, almost
all of their income may be spent on basic necessities, including food,
clothing and housing. This group is at higher risk of not being able to
afford increasing housing costs or they may not have any choice on
housing. For the higher income households, paying more than 30 percent
income on rent or a mortgage is more likely to be a choice, perhaps to
live in a more convenient or desirable area, or to pay off extra on the
mortgage to shorten the term of payment. However, there is a possibility
that the households in the third quintile (40th to 60th income
percentile) of the income distribution - who usually are known as middle
class earners - may also have financial hardship in meeting high housing
costs, and may have only limited choices to do with housing. By choosing
the bottom 40 percent of income distribution as the cut-off, the middle
class earning households are excluded from the definitions.
Although middle class income households are at a lower risk of
housing stress than low income households, they may be at a level of
'marginal housing stress' because a substantial rise in
interest rates, housing prices, or job loss etc. may cause the middle
class income households to fall into housing stress. Moreover the 40
percent cut-off is the same regardless of the area in which the
individual or household unit is living. Hence no account is taken of
housing costs which vary with location; for example the high rents of
Canberra and Sydney compared to the low rents of Adelaide are not taken
into account in these definitions.
A very severe form of housing stress is the risk of homelessness
and may apply to households in the lowest 10 percent of income
distribution. This group is quite vulnerable to rising housing costs.
Note that many homeless are homeless due to a situation of financial
hardship where individuals are unable to afford housing costs or to keep
a place to live. Rapidly increasing housing costs could force more of
the lowest earning households into homelessness. So the exclusion of
households within the lowest income decile from the 30/10-40 rule may
overlook this severe form of housing stress. In addition, this
definition cannot be used as a means of strategic policy intervention
for poverty and housing assistance programs due to its exclusion of the
most disadvantaged households. However, some studies do argue that the
reported incomes of households in the bottom 10 percent of the income
distribution do not always accurately reflect their living standards,
and their inclusion in the definition may overestimate housing stress
(see ABS, 2005), which is why the ABS argues for the 30/10-40 rule.
A Comparison of Various Ratio Measures
A comparison of the three rules of measuring housing stress is
provided in Table 1. Note that none of these definitions takes into
account the fact that housing costs vary according to area. The
specified rules use relative income of household and the general rule
(30 only) uses the absolute household income.
The 30/40 rule is the widely used definition of housing stress in
Australia. Although this definition may ignore marginal housing stress,
it acknowledges the size of the household income unit by using the
equivalised household income distribution. Whereas, the 30/(10-40) rule
is also based on equivalised household income distribution, it is more
restricted and occasionally uses a definition that ignores both the
severe and marginal forms of housing stress. Nevertheless the
availability of suitable data, methodological tools and specific
research interests in each of these definitions is useful.
It is noted that, in all the definitions, households with negative
and nil incomes have been removed from the analysis. In survey data, few
households have reported nil or negative incomes. These are often
excluded from any analysis related to income distribution and financial
well-being, as research from the ABS has shown that the expenditure of
these households is similar to that of households earning much more, so
these incomes are considered an unreliable measure of a household's
standard of living (ABS 2005).
Moreover, the distributions of housing stress measured by the three
different rule-based variants are presented in Figure 2. It is obvious
from the figure that not only does the percentage of households in
housing stress vary under different definitions, but also the density of
the SLAs varies with the percentage of housing stress across Australia.
[FIGURE 2 OMITTED]
The graph of the '30/40 rule'- based variant of housing
stress shows that approximately 67 percent of the SLAs have housing
stress households of 7 to 11 percent, with a mean of 9.52 percent and a
coefficient of variation (C.V.) of 34.95. In addition, the graph of the
'30/40-10 rule'-based variant shows that most SLAs (about 87
percent) have housing stress households of 3 to 7 percent, with a mean
of 4.91 percent and a C.V. of 41.85. The '30 only rule'
variant of housing stress reveals that about 51 percent of the SLAs in
Australia have households with a rate of housing stress of 13 to 17
percent, with a mean of 14.68 and a C.V. of 36.71.
According to Karl Pearson the C.V. is a very powerful tool for
comparing the variability of two or more series of variants (Gupta and
Kapoor, 2008), where a variant having the lowest C.V. is considered to
be more consistent than the others. In this regard, since the C.V. for
the '30/40 rule'-based variant of housing stress estimation is
the lowest compared with the variation measures for the other two
variants, this variant ('30/40 rule'-based definition) of
housing stress estimation is more consistent than the others.
Furthermore, in terms of the distributional pattern of these three
curves, the '30/40 rule'-based housing stress variant also
shows a more rational pattern towards the usual normal curve, while the
'30/(40-10) rule' and '30 only rule'-based variants
resemble leptokurtic and platykurtic curves respectively. From the
statistical point of view, the '30/40 rule'-based housing
stress estimation is more consistent and appropriate at small area
levels in Australia.
The '30/40 rule'-based definition is also accountable and
valid for using socioeconomic policy analyses that link with the housing
stress issue. For instance, one of the significant policy implications
of this definition is that this rule is widely used as the basis for
determining household eligibility for entry to public rental housing
and/or receipt of commonwealth rent assistance (CRA). Moreover, the
definition has been used by many researchers and public and private
organizations including the National Housing Strategy (1992), ABS
(2002), Harding et al. (2004), Yates and Gabriel (2006), and recently in
estimating figures used by the Australian Prime Minister and
subsequently published by the Australian Government Department of
Families, Housing, Community Services and Indigenous Affairs (FaHCSIA,
2008). Therefore, this paper uses the '30/40 rule'- based
variant to define households in housing stress as those with equivalised
household gross income in the lowest two quintiles (bottom 40 percent)
of all household incomes in Australia, who are spending more than 30
percent of their gross household income on either renting costs or
mortgage repayments.
3. METHODOLOGY
This section briefly presents the research methodology - which is a
spatial microsimulation modelling technology (MMT) approach of small
area estimation. The method is rapidly becoming popular in the developed
world and has now a wide range of applications (see for example, Rahman,
2011; Rahman et al., 2013; Rahman and Harding, 2014) including
simulation of the small area impact of changes in income taxes and cash
transfers (Ballas and Clarke, 2001; Harding et al., 2009); the
development of small area measures of poverty and social exclusion
(Tanton et al., 2009; McNamara et al., 2007; Miranti et al., 2011); the
small area modelling of activities of daily living status and/or the
need for different types of care (Williamson, 1996; Lymer et al., 2008);
the development of the SimObesity model to examine small area obesity
among children (Procter et al., 2008); small area health-related
conditions (Ballas et al., 2006a; Rahman and Harding, 2011; Rahman and
Harding, 2013) and the socio-economic impacts of major job gain or loss
at the local level (Clarke, 1996; Ballas et al., 2006b).
Spatial-level Microdata Generation
Creation of a synthetic micropopulation dataset at the small area
level, such as the SLA level in Australia, is very challenging. Small
area estimation technologies have become useful tools to overcome this
challenge. Although there are two methods (statistical and geographic)
in small area estimation for generating small area microdata, this paper
uses the geographic approach also known as spatial microsimulation
modelling (SMM). A detailed description of various methods, their
properties, suitability and applications are reported in other studies
(Rahman, 2009; Harding and Tanton, 2011; Rahman and Harding, 2014). The
MMT approach of microdata simulation involves some complex procedures,
whose gradual evolution has been described in detail in other research
(see for example, Chin and Harding, 2006; Rahman et al., 2010b; Cassells
et al., 2010; Rahman, 2011; Rahman et al., 2013).
To produce SLA level housing stress estimates in Australia, a SMM
was designed that uses a range of datasets that come from the Australian
Bureau of Statistics. These datasets have custom designed tables from
the Census. In summary, the ABS sample survey in question is reweighted
to match the small area Census benchmark tables, resulting in unit
records for households and individuals for each SLA in the model.
General discussion about these datasets and various steps of microdata
generation are contained in Rahman (2011). The model generates
reasonable microdata (by an accuracy index criterion (AIC) illustrated
in Rahman, 2011) for 1 397 SLAs which contain more than 99.9 percent
households. Among 1 422 SLAs across Australia, the model did not produce
reasonable microdata for only 25 SLAs (non-convergent SLAs as per the
AIC), which had very small or no populations and were typically located
in very remote areas. The overall microdata generation process is
depicted in Figure 3.
[FIGURE 3 OMITTED]
Clearly, the process starts by using the SAS language to run the
general model file, which contains the path to all input data files and
the GREGWT algorithm. The main calculations in the iteration process for
the GREGWT algorithm operate separately for each id number of small
areas (that is SLA codes). This complex process tracks numerous matrix
and/or vector calculations towards achieving convergence for each SLA in
the minimum number of iterations. In addition, it also does analysis for
extreme data units to determine whether the extreme units have effects
on the overall calculations. However, the output keeps records on only
the top 30 extremes.
Although the GREGWT program follows the Newton-Raphson approach of
iteration, the entire execution process of the model follows just a few
successive algorithmic steps, which can be described as:
Step 1: Read in the general model file.
Step 2: Read in benchmark tables, Census data and microdata records
from Survey of Income and Housing-Confidentialised Unit Record Files
(SIH-CURFs) with SIH-linkage file mentioned in the general model file.
Step 3: Query the individual records within the microdata according
to the classifications of the general model file.
Step 4: Change original weights to a new set of weights following a
truncated Chi-Square distance function for an appropriate allocation of
households/individuals towards the small area benchmarks.
Step 5: Apply the Newton-Raphson method of iteration to determine
the best set of new weights by minimising the total distance between the
new-synthetic weights and original weights.
Step 6: When convergence has been achieved and/or predefined number
of iterations reached, the corresponding new set of synthetic weights is
retained by the process and considered as the best reweights.
Spatial Microsimulation Model Outputs: The 1st Stage
Basically there are three outputs from this initial phase of the
model. First of all, the core output is the file of synthetic household
weights by SLAs in Australia. This file is considered as the most
significant output of the model because of its usefulness in the next
computational stage of the model (for getting small area microdata and
the estimates). The second and third outputs of the model are,
respectively, details about residual estimates of the synthetic weights
and a convergence report of the model. These two outputs are associated
information about the synthetic weights produced by the model. For
example, the residual estimates file shows the accuracy of the new
weights according to various benchmark classifications. In the spatial
microsimulation process, a modeller's expectation is to minimise
the overall residual estimates as much as possible, to ensure the
consistency and reliability of the synthetic weights. In addition, the
convergence report provides information about whether or not the GREGWT
reweighting algorithm has converged to the benchmarks for a specific
SLA. When the convergence rate seems reasonably low, then the modeller
may need to revisit the specification of the model for modification.
Note that the "synthetic weights" file (see Table 2) is
the central requirement in the MMT approach of small area estimation.
The synthetic weights output file is often known as the synthetic or
simulated spatial microdata new-weights, and it is the only output to be
used in the next stage of the model for producing ultimate small area
estimates. If this stage of the model can generate more accurate
synthetic weights at small area levels, then the final small area
estimates of interests are likely to be statistically more reliable.
Model Outputs: The 2nd Stage
To produce small area estimates of housing stress we have to run
the second stage of the housing stress model. This section describes
various parts of the 2nd stage of the model for SLA level housing stress
estimation.
Typically, three input files are essential for the second stage of
the housing stage model. They are
1) SIH-CURFs;
2) Synthetic weights; and
3) The Consumer Price Index (CPI) file.
These three input files are connected by a SAS program file that is
known as the second stage program file. This SAS file not only contains
all the linkage paths towards the input files, but also it programs the
definition of the housing stress measure, various logic operations and
codes of summary statistics for small area estimates. It also indicates
a pathway to an outputs folder where the demanded small area estimates
could be stored.
The output from the second stage model is the ultimate file for
small area housing stress estimates in Australia. This research
considers the SLA in Australia as a small area. So, the ultimate output
file will contain a range of data for the SLA level housing stress
estimation. In particular, the file contains data for the following
attributes presented in Table 3.
The output file provides household level estimates of total numbers
as well as percentages for each characteristic in the above table. The
model can also produce persons' level small area estimates for
these variables.
4. RESULTS AND DISCUSSION
This section reports on a selection of the outputs which are
produced by the model.
Households and Housing Stress by Tenure Type
The distributions of Australian households and housing stress by
tenure are given in Figure 4. About 70 percent of households are living
in their own house, with half of them being buyers. Nearly 27 percent of
households are renters, with about 22.5 percent being in private
rental. Only 2.9 percent of Australian households are living in other
tenures, such as hospital beds, military housing, hotels/hostels etc.
Figure 4b reveals that one-third of buyer households (33.2 percent) in
Australia are in housing stress. It seems an indication that a
proportion of low income households buying their house with the support
of first home owners' grant is associated with a high house price,
and very low levels of housing supply in many areas, especially in the
inner city areas. Additionally, about 59.6 percent private renter
households experience housing stress, while just 6.9 percent public
renters are in housing stress. So, housing stress estimates for private
renters have not only significant influence on the housing stress
estimates for renters and overall households, but also have an effect on
spatial scales where housing supply is very limited and the demand as
well as costs of housing are high for a proportion of low to middle
earner households (Rahman, 2011).
Although in theory, households living in public housings are paying
less than 30 percent of their assessable income in housing rent (AIHW,
2009), in the equivalised household gross income amount they may be
paying more than 30 percent of their income in housing costs. The
Commonwealth Rent Assistance eligibility is dependent on recipients
being on some form of government transfer payment which is also the
primary source of income for public housing households. However, as very
low income households, these tenure groups are likely to be in housing
stress. For instance, in 2005-06, the proportions of public housing
households in Australia with an older resident was 28 percent and with a
member with a disability was 29 percent, while substantial percentages
(about 29 and 33 percent of households with an older tenant or tenant
with a disability respectively) of them were still in housing stress,
after the Commonwealth Rent Assistance had been received (see for
example, SCRGSP, 2007; AIHW, 2008).
Estimates for Different States and Territories
The model estimates a total of 7 128 035 households in Australia,
of which 10.9 percent (i.e., 773 073 households) are in housing stress
(Table A1 in Appendix). One-third of Australian households are located
in NSW of which about 11.6 percent of households are in housing stress,
and the estimated housing stress number for private renters (i.e., 164
089 households) is almost twice the estimated number for buyers (83 894
households). Victoria is the residence of a quarter of Australian
households with about 10.4 percent of households being in housing
stress, most of which are buyers and renters. Nearly 11.3 percent of 1
387 069 households in Queensland are estimated to be in housing stress
with almost 27.9 percent being private renters.
Although Western Australia contains 701 116 households, of which
about 9.9 percent are in housing stress, the estimates for public
renters are much lower in WA and Tasmania compared to the estimates for
other states and territories. The overall rate of housing stress is also
higher in South Australia. About 10.1 percent of 181 666 households are
experiencing housing stress in Tasmania. Moreover, only 6.6 and 9.2
percent of households located in the Australian Capital Territory and
Northern Territory are in housing stress, with the highest prevalence
rate (i.e., approximately 20 percent) in the public renters.
Housing Stress by Statistical Division
Table 4 presents the results of housing stress estimates for
various statistical divisions (SD) in Australia. An estimated number of
163 655 (21.2 percent) and 135 702 (17.6 percent) households are
experiencing housing stress in Sydney and Melbourne SDs. A relatively
smaller but significant number of housing stress households are in other
major capital city SDs--such as Brisbane: 66 718 (8.6 percent), Perth:
53 766 (7.0 percent) and Adelaide: 46 749 (6.1 percent).
Thus, Sydney, Melbourne, Brisbane, Perth and Adelaide collectively
account for about 60.5 percent of the total number of households in
housing stress for Australia. In comparison, only 2.4 percent of housing
stress households reside in Hobart, Canberra and Darwin. The remaining
37.1 percent of households reside in non-capital SDs. Seven south-east
coastal SDs such as Hunter, Illawarra, Mid-North Coast and
Richmond-Tweed in the NSW and the Gold Coast, Sunshine Coast and Wide
Bay-Burnett in Queensland - have relatively higher estimates than other
noncapital SDs (ranging from 11 991 to 25 787 households) and
collectively contain 15.8 percent of all housing stress households in
Australia.
Estimates for Various Statistical Subdivisions
To get a much better view at the regional level, the results at the
statistical subdivision (SSD) level show that a significantly large
number of 20 990 households experiencing housing stress is in the port
city Newcastle (Table A2 in Appendix). There are several main
geographical regional parts where housing stress is concentrated at SSD
level in Sydney, Melbourne, Perth, Adelaide and coastal regions in New
South Wales and Queensland. Twelve SSDs making up the western, south
western, northern and inner parts of Sydney collectively contain an
estimate of 150 775 (19.5 percent of total) housing stress households in
Australia. The Fairfield-Liverpool SSD in western Sydney individually
has the highest proportion of 16.9 percent households in housing stress.
Although Western Melbourne SSD has the third highest estimated
number of 17 098 households, the area's rate of 11.5 percent is
relatively low. The Greater Dandenong, Hume and Frankston cities and
inner Melbourne have housing stress rates of 14.9, 14.1, 12.6 and 12.3
percent respectively. In addition, several SSDs in north, east and
south-east metropolitan Perth and the northern, southern, western and
eastern parts of Adelaide have noticeably large estimates of housing
stress. Some other major coastal centres such as Wollongong,
Richmond-Tweed and Hastings in NSW; Gold Coast, Sunshine Coast, Wide
Bay-Burnett and Cairns city in Queensland; and the Hobart SSD also have
significant estimates.
It is noticeable that low income households residing in the
attractive and a high demand Gold Coast region are more prevalently (an
average rate of 14.0 percent) in housing stress. This may be because of
a very high level of house prices or rents in the Gold Coast areas.
SLA Level Estimates of Housing Stress across Australia
The spatial analysis depicts estimates by SLAs. Typically, the
spatial units of analysis vary greatly in population size and presenting
results for the estimated number of households in housing stress usually
does not mean a great deal when looking at which areas have housing
stress. Thus, only the percentage estimates are considered in spatial
analysis, and the spatial graph is depicted in Figure 5. For mapping,
the quantile classification is used for geographic distribution of the
housing stress (but those SLAs that did not meet the accuracy criterion
in the microdata simulation process are treated as missing). This option
examines the relativity of all SLAs in Australia. In view of the fact
that city areas are very condensed and unseen in the main map, they are
presented in separate boxes.
[FIGURE 5 OMITTED]
Findings of the spatial analysis reveal that most of the SLAs in
the eastcoast and some SLAs in the west-coast regions in Australia have
a relatively higher rate (over 11.2 percent) of households in housing
stress. Although many SLAs in inland remote regions throughout the
country have the lowest rates of housing stress households, small areas
across the mining-boom regions in inland Queensland and Western
Australia illustrate relatively higher percentage estimates.
The map also reveals that a number of SLAs located within some
major capital cities of Australia have significantly high rates of
housing stress (ranging from 16.81 to 28.00 percent). Some SLAs in inner
locations of Melbourne, Canberra and Adelaide have the highest
percentage estimates. For example, SLAs of inner city in Melbourne and
Canberra have estimates of 27.0 percent and 23.2 percent respectively.
Perhaps, these results are due to the fact that housing in inner city
SLAs is always preferable to many high income households who are in
housing stress by choice. Housing supply is very much limited in inner
city areas. So the house price and rents are too high, and consequently
unaffordable to a high proportion of low to middle income households.
Nevertheless, many SLAs from Brisbane and Sydney, with some others
from coastal cities in Queensland and NSW, also have the highest rates.
It is evident that few SLAs in Sydney: Fairfield (C) - East, Canterbury
(C), Bankstown (C) - North-East and Auburn (A) have a significantly high
proportion of housing stress. This is because a large number of
households live in these SLAs, with a sizable representation of them
from the low income households. Also, small sample size problems appear
to exist within many SLAs in Brisbane, where the number of households
experiencing housing stress is very low, but the percentage estimate is
significantly high due to the small value of the denominator.
5. VALIDATION TOOLS
Validation and the creation of measures of the statistical
reliability of small area estimates by microsimulation modelling are
challenging (Ballas and Clarke, 2001; Hynes et al., 2006; Edwards and
Clarke, 2009; Rahman, 2009; Rahman et al., 2010a). At small area levels,
the estimated data are typically unavailable from another source.
Accordingly, some researchers have suggested re-aggregating the small
area estimates up to larger levels, where reliable data are available to
compare the results (Ballas and Clarke, 2001; Kelly, 2004), while others
have attempted to use alternative methods to determine the accuracy of
their model estimates (Hynes et al., 2006; Edwards and Clarke, 2009).
Discussions about various validation methods used by researchers are
outlined in detail in other studies (i.e. in Rahman, 2011; Rahman et
al., 2013; and Rahman and Harding, 2014). This section offers a new
validation tool for testing the accuracy of SLA level housing stress
estimates in Australia which are produced by the microsimulation
modelling technology.
Absolute Standardised Residual Estimate (ASRE) Analysis
In this approach to validation, we first have to calculate an
absolute standardised residual estimate (ASRE) for a small area (in this
case SLA level housing stress estimation), and then analyse the values
of the ASRE to make a decision about the accuracy. The mathematical
formulae for the ASRE use the following standard notations:
[[??].sub.ij] is an observed household total in the jth data at the
ith small area;
[Y.sub.ij] is the total households in the jth population at the ith
small area;
and
[m.sup.r] is the number of small areas in a rth region and r >
i.
The ASRE can be defined as
ASRE = ([[delta].sub.ij])/[square root of AEMSE])
where [[delta].sub.ij] = [absolute value of [Y.sub.ij] -
[[??].sub.ij]] and AEMSE = 1/[m.sub.r] [[sigma].sub.m] [([Y.sub.ij] -
[[??].sub.ij]).sup.2] where the
AEMSE is the Average Empirical Mean Square Error (see for example,
Gomez-Rubio et al., 2008 and Rahman, 2011).
The decision criterion for this validation technique is: 1) when
the value of ASRE is close to zero or less than 2 for a SLA then the
synthetic household estimate is acceptable (i.e. the performance of the
model estimate is good); and 2) when the ASRE value is at least 2, then
it is usually considered as a large error (Field, 2000) suggesting that
unexplained errors exist in the model estimates and/or the
microsimulated datasets.
Results from the ASRE Analysis
Results of ASRE analysis for overall households in housing stress
confirm that for 1 205 SLAs out of 1 278 (94.3 percent) in Australia,
the model determined very accurate housing stress estimates (Figure 6).
There are 73 SLAs that have an ASRE measure of at least 2, and many of
these SLAs are located in the capital cities and coastal centres such as
Wollongong, Newcastle, Coffs Harbour, Tweed Heads, Gold Coast, Hervey
Bay, Mackay etc. For instance, a few SLAs in Ipswich show a high value
of ASRE, which indicates that the model has produced statistically
non-significant housing stress estimates in this area. In particular the
SLA: Ipswich (C) - Central shows an ASRE value of 5.6, which is much
bigger than 2. So, for this small area, the estimate of housing stress
is not statistically accurate using the ASRE measure.
Ipswich is one of the fastest growing regions in Brisbane and the
population characteristics are quite different to the Australian
average. In particular, a significantly large number of working
population families (about 60 percent) are Technicians & trades
workers, Community & personal service workers, Clerical &
administrative workers, and Labourers, who tend to have lower incomes
(ABS, 2007). But the housing costs in this area are relatively high. The
supply of housing in this area is also inadequate with growing housing
demand for increasing populations. As a result, the model simulates
significantly high estimates of housing stress for the region by
considering the micro-level attributes.
[FIGURE 6 OMITTED]
To get an idea of why a non-significant value of ASRE arises for
some of these small area estimates, we may check detailed micro-level
results for an SLA (such as Petermann-Simpson in Alice Springs, NT)
along with its geographic characteristics. For the Petermann-Simpson
SLA, the ASRE value of 8.5 has revealed that the model overestimated the
housing stress for overall households. It is noted that
Petermann-Simpson is one of the functional economic and strategic
growing areas in rural central Australia (ABS, 2007; Rahman, 2011).
Economic growth in this SLA results from the flow-on effects of
providing regional support services to major national projects such as
tourism, culture and heritages conservation, mining development, defence
construction, forestry and horticultural trials, and a transport and
logistics hub servicing the central Australia railway. However,
residential land release and housing supply is not consistently adequate
in this remote area with its growing population. High demands for
housing increase the house price and rents in the area that increase
noticeably the money allocated to housing for lower income households
and perhaps skew the estimate of housing stress. Sharply increasing
housing costs (the average annual change for 2008-09 is estimated as 27
percent) for a large group of low income households (having median
weekly income of 961 AUD) residing in Petermann-Simpson has influence
over a high rate of housing stress.
6. CONCLUSIONS
This paper has empirically examined the statistical local area
level housing stress estimates across Australia using a synthetically
simulated micro-dataset and analysed the results. It has also
demonstrated a new method for validating the results of small area
housing stress statistics.
According to our findings housing stress estimate is greatest
within several-hotspot areas in Australia. One of the key findings using
outputs from the spatial microsimulation model was that in 2011 around
one in ten Australian households were experiencing housing stress, with
large numbers of these households residing in the east coast states of
New South Wales, Victoria and Queensland. When looking at housing stress
at a higher geographic disaggregation, findings from the model outputs
have revealed that households experiencing housing stress were mostly
residents of the Sydney, Melbourne, Brisbane, Perth, Adelaide, Gold
Coast, Hunter, Illawarra, Mid-North Coast statistical divisions, along
with some other statistical divisions located across the coastal centres
of New South Wales and Queensland. The Canberra, Hobart and Darwin
statistical divisions all have relatively low housing stress levels.
Breaking the geographic classifications down to a finer level, we
find greater heterogeneity in housing stress estimates, but still the
households are concentrated in these main locations or spots. Areas with
a high proportion of households living in housing stress were those
concentrated in the outer fringes of capital cities along the east coast
of Australia. Of particular interest was Newcastle, which has the
largest estimated number of households (20 990) in housing stress among
all of the statistical subdivisions in Australia. More explicitly, the
range of estimated numbers of housing stress was from 1 886 households
for Newcastle (C) Outer West to 2 826 households for Newcastle
(C)--Inner City among the nine SLAs in this statistical subdivision.
Although the estimated number is the highest for Newcastle, the
percentage estimate (about 11.4 percent) was relatively lower than in
many hotspot SSDs within the capital and non-capital cities. Some other
non-capital coastal cities--such as Wollongong, Richmond-Tweed, Hastings
and Clarence etc in New South Wales and Gold Coast, Sunshine Coasts,
Wide Bay-Burnett and Cairns City in Queensland - have spatial
subdivisions with much higher rates of housing stress. In addition, many
statistical subdivisions within capital cities have also demonstrated
large estimated figures. Basically, these regional subdivisions are
located in the greater western and northern regions of Sydney, in the
western, inner, eastern middle, southern and northern outer regions of
Melbourne, in the north-west, south-east and Logan City regions of
Brisbane, in the north, east and south-east metropolitan regions of
Perth, as well as in the northern, southern, western and eastern regions
of Adelaide.
Breaking the geographic scale down even further to one of the
smallest and administratively helpful areas--the SLA--we can really see
which small areas are suffering the most from housing stress. Findings
have demonstrated that a large number of SLAs in the New South Wales
coastal cities, including Sydney, had the highest numbers of households
in housing stress. Most of the SLAs in Melbourne, Adelaide, and Hobart
also had significantly high estimates. Moreover, the rapidly growing
mining areas around inland locations in different states have resulted
in many SLAs with relatively higher estimates of housing stress. This
could be because of a significant lack in the supply of housing within
these quickly growing mining areas, which in turn creates a high demand
of housing and then increasing housing costs for mainly low and middle
income households. In contrast, significantly large numbers of SLAs in
Brisbane, Canberra and Darwin have much lower numbers of households in
housing stress. This is probably because these SLAs are not only small
in size but also have relatively smaller household populations. The
results of the percentage estimates reveal somewhat opposite results to
the number count estimates: that is, many small SLAs with few households
show high percentages of households in housing stress, but there are
actually only a few households in stress in these locations.
Nonetheless, various SLAs in different capital cities indeed confirm
significantly large values in housing stress for both number counts as
well as percentages.
The validation tool outlined in this paper is the ASRE analysis,
where an ASRE for the SLA level housing stress estimate has been
calculated and then analysed using a standard cut-off criteria for
making a decision. Results have demonstrated statistically accurate
estimates for a very high number of SLAs (about 94.3 percent). There are
a number of SLAs with statistically insignificant values of ASRE, and
most of them are geographically located in the capital cities, including
Melbourne, Brisbane, Canberra and Darwin, as well as major coastal
centres in the Eastern part of Australia. Additionally, findings suggest
that the proposed validation tools can not only check the statistical
validity of an SLA level estimate, but can also identify and describe
the possible features of the SLAs that may have insignificant results.
The SLAs with ASRE values significantly bigger than 2 demonstrate
inaccurate housing stress estimates for the respective SLAs. In such a
case researchers would undertake further analysis of these micro-level
data for these SLAs, along with their geographic attributes.
Looking at future research directions, we are currently finalising
estimates of SLA level housing stress estimates by tenure types within
eight major capital cities in Australia, comparing the estimates of
housing stress between the cities as well as looking at different SLAs
within a specific major city. In addition, a proposed technique for
estimating confidence intervals around the housing stress estimates will
also be explored. Finally, using groupings of various housing costs such
as 0-10, 10-20, 20-30 percent etc of the households' income, a new
study would estimates the housing stress for different income deciles
and then map the estimates within these groups at a chosen spatial scale
such as local government area.
APPENDIX
Table A1. Number of Households and Housing Stress Estimates by
Tenure Types for the States and Territories in Australia, 2011.
States Overall Owners Buyers Public Private Other
& Total Renters Renters tenure
Terri- HH (1) HH HH HH HH HH
tories (HS% (2)) (HS) (HS) (HS) (HS) (HS)
NSW 2328200 836696 760241 114423 548464 68376
(11.57) (0.098) (11.04) (17.84) (29.92) (0.135)
VIC 1781601 665595 649015 57158 364009 45824
(10.42) (0.074) (11.00) (17.23) (28.53) (0.103)
QLD 1387069 452587 480441 49455 362374 42211
(11.29) (0.127) (9.80) (15.66) (27.90) (0.142)
WA 701116 226922 270603 29681 151063 22847
(9.91) (0.087) (8.90) (14.91) (26.94) (0.153)
SA 583284 208924 208090 42311 104603 19356
(10.54) (0.064) (9.94) (15.19) (32.66) (0.103)
TAS 181666 70923 62269 10912 32428 5134
(10.11) (0.059) (10.31) (14.04) (31.96) (0.136)
ACT 116911 35567 45761 9453 24101 2027
(6.59) (0.008) (4.52) (20.05) (15.44) (0.000)
NT 48188 8432 18174 4533 14668 2380
(9.35) (0.43) (7.08) (19.32) (15.67) (0.042)
AUS 7128035 2505646 2494594 317926 1601710 208155
(10.85) (0.091) (10.30) (16.72) (28.74) (0.13)
Note: (1) No. of Households; (2) Proportion of Households in Housing
Stress. Source: the Authors.
Table A2. Lists of the Thirty-Five SSDs with the Highest Estimated
Numbers, and Highest Percentages of Households, Experiencing Housing
Stress across Australia, 2011.
ID SSD Name HS1 %
11005 Newcastle 20990 11.4
10525 Fairfield-Liverpool 17464 16.9
20510 Western Melbourne 17098 11.5
50515 North Metropolitan 16090 10.1
10520 CanterburyBankstown 15935 16.1
40505 Northern Adelaide 15626 11.9
10540 CentralWestern Syd. 15352 15.2
10515 St George-Sutherland 14748 9.8
10505 Inner Sydney 14589 12.1
10570 Gosford-Wyong 14365 13.0
20505 Inner Melbourne 14264 12.3
50525 South Eastern Metro. 13417 11.0
20565 Southern Melbourne 13338 9.1
40520 Southern Adelaide 12689 10.0
20550 Eastern Middle Melb. 12316 8.3
30715 Gold Coast West 11732 14.1
10545 Outer Western Syd. 11640 11.2
10553 Blacktown 11322 13.2
30905 Sunshine Coast 11195 14.0
11505 Wollongong 11142 11.6
50520 South Western Metro. 11003 9.9
30710 Gold Coast East 10889 15.5
20580 SuthEast Outer Melb. 10446 11.9
40510 Western Adelaide 9800 11.6
30507 Nrthwest Outer Bris. 9339 8.4
20530 Northern Mid. Melb. 9199 10.1
10555 Lower Northern Syd. 9140 8.2
50510 East Metropolitan 8934 10.1
10530 Outer SuthWest Syd. 8837 11.9
10560 Central North Sydney 8815 6.6
40515 Eastern Adelaide 8634 9.8
10510 Eastern Suburbs 8568 9.8
30511 Sutheast Outer Bris. 8345 10.5
60505 Greater Hobart 7856 10.3
20555 Eastern Outer Melb. 7826 9.1
ID SSD Name HS %2
10525 Fairfield-Liverpool 17464 16.9
12501 Coffs Harbour 3055 16.7
10520 Canterbury-Bankstown 15935 16.1
30710 Gold Coast East 10889 15.5
12007 Lismore 1758 15.4
10540 CentralWestern 15352 15.2
Sydney
12005 Tweed Heads&Coast 3611 15.1
20575 Greater Dandenong 6384 14.9
City
12010 RichmondTweed SDBal 7311 14.9
12503 Port Macquarie 2338 14.6
20535 Hume City 6453 14.1
30715 Gold Coast West 11732 14.1
12505 Clarence(excl. 5146 14.0
CoffsHarb)
31507 Hervey Bay City 2589 14.0
Part A
30905 Sunshine Coast 11195 14.0
30705 Gold Coast North 2533 13.9
30520 Caboolture Shire 6324 13.8
30545 Redcliffe City 2806 13.6
12510 Hastings(excl.Prt 5238 13.5
Macqu)
30530 Logan City 7670 13.4
10553 Blacktown 11322 13.2
31505 Bundaberg 2954 13.2
14515 Lower South Coast 3362 13.0
30910 Sunshine Coast 3066 13.0
SD Bal
10570 Gosford-Wyong 14365 13.0
14003 Bathurst 1381 12.7
20585 Frankston City 5484 12.6
11507 Nowra-Bomaderry 1433 12.6
35005 Cairns City Part A 5485 12.5
23005 Mildura Rural City A 2110 12.4
30720 Gold Coast SD Bal 633 12.4
30501 Inner Brisbane 4227 12.4
20505 Inner Melbourne 14264 12.3
24005 Greater Shepparton A 1948 12.1
10505 Inner Sydney 14589 12.1
Note: (1) Arranged by No. of Households Experiencing Housing Stress,
and (2) Arranged by Percentage of Households Experiencing Housing
Stress. Source: the Authors.
ACKNOWLEDGEMENTS: This paper utilises the methodological research
that has been undertaken as part of the PhD of Dr Rahman, based on the
three prestigious scholarships: an E-IPRS from the Commonwealth of
Australia, the ACT - Land Development Agency Postgraduate Research
Scholarship from the Government of Australian Capital Territory (ACT)
and the Australian Housing and Urban Research Institute (AHURI) and the
NATSEM Top-Up Scholarship from the University of Canberra (UC). Special
thanks are due to Robert Tanton and Shuangzhe Liu at UC, and Mark
Morrison, Kenneth Russell and peoples involved in the CSIRO workshop at
CSU in Australia for their valuable comments and stimulus.
REFERENCES
ABS (2002). Measuring Australia's Progress. ABS Cat. No.
1370.0, Australian Bureau of Statistics, Canberra.
ABS (2005). Housing occupancy and costs of Australia--2002-03. ABS
Cat. No. 4130.0.55.001, Australian Bureau of Statistics, Canberra.
ABS (2007). The 2006 Census Community Profile Series. Cat. No.
2001.0, Australian Bureau of Statistics, Canberra.
ABS (2008). Year Book Australia. ABS Cat. no. 1301.0, Australian
Bureau of Statistics, Canberra.
AIHW (2008). Housing Assistance in Australia. Catalogue no. HOU
173, Australian Institute of Health and Welfare, Canberra.
AIHW (2009). Public Rental Housing 2007-08: Commonwealth state
housing agreement national data report. Housing assistance data
development series, Catalogue no. HOU 187, Australian Institute of
Health and Welfare, Canberra.
Ballas, D. and Clarke, G. P. (2001). Modelling the Local Impacts of
National Social Policies: a Spatial Microsimulation Approach.
Environment and Planning C: Government and Policy, 19(4), pp. 587-606.
Ballas, D., Clarke, G., Dorling, D., Rigby, J. and Wheeler, B.
(2006a). Using geographical Information Systems and Spatial
Microsimulation for the Analysis of Health Inequalities. Health
Informatics Journal, 12(1), pp. 65-79.
Ballas, D., Clarke, G.P. and Dewhurst, J. (2006b). Modelling the
Socio-Economic Impacts of Major Job Loss or Gain at the Local level: A
Spatial Microsimulation Framework. Spatial Economic Analysis, 1(1), pp
127-146.
Burke, T., Stone, M.E. and Ralston, L. (2010). Residual Incomes in
Australia: Methodology and Exemplification. Positioning Paper,
Melbourne: Australian Housing and Urban Research Institute.
Cassells, R., Harding, A., Miranti, R., Tanton, R. and McNamara, J.
(2010). Spatial Microsimulation: Preparation of Sample Survey and Census
Data for SpatialMSM/08 and SpatialMSM/09. Technical Paper No 36, NATSEM,
University of Canberra.
Chin, S. F. and Harding, A. (2006). Regional Dimensions: Creating
Synthetic Small-Area Microdata and Spatial Microsimulation Models.
Online Technical Paper--TP33, NATSEM, University of Canberra.
Chin, S. F., Harding, A., Lloyd, R., McNamara, J., Phillips, B. and
Vu, Q. N. (2005). Spatial Microsimulation Using Synthetic Small Area
Estimates of Income, Tax and Social Security Benefits.
Australasian Journal of Regional Studies, 11(3), pp. 303-335.
Clarke, G. P. (1996). Microsimulation for Urban and Regional Policy
Analysis. Pion, London.
Edwards, K. L. and Clarke, G. P. (2009). The Design and Validation
of a Spatial Microsimulation Model of Obesogenic Environments for
Children in Leeds, UK: SimObesity. Social Science and Medicine, 69(7),
pp. 1127-1134.
FaHCSIA (2008). Making housing affordable again. Online version
accessed 26 July 2009,
http://www.facs.gov.au/internet/facsinternet.nsf/vIA/housing
affordable/$File/making housing affordable again.pdf.
Field, A. (2000). Discovering Statistics Using SPSS for Windows.
SAGE Publication Ltd, London.
Flood, J. (2012). Housing Stress: Keep or Discard. Paper presented
at the 6th Australasian Housing Researchers ' Conference, Adelaide,
8-10 Feb.
Foard, G., Karmel, R., Collett, S., Bosworth, E. and Hulmes, D.
(1994). Public housing in Australia. Australian Institute for Health and
Welfare, Canberra.
Gomez-Rubio, V., Best, N. and Richardson, S. (2008). A Comparison
of Different Methods for Small Area Estimation. NCRM Working Paper, ESRC
National Centre for Research Methods, unpublished. Gupta, S.C. and
Kapoor, V.K. (2008). Fundamentals of Mathematical Statistics. New Delhi,
Sultan Chand and Sons.
Harding, A. and Tanton, R. (2011). Policy and People at the Small
Area Level: Using Microsimulation to Create Synthetic Spatial Data. In
R. Stimson (Ed) Handbook in Spatially Integrated Social Science Research
Methods, Edward Elgar, Sydney
Harding, A., Lloyd, R., Bill, A. and King, A. (2006). Assessing
Poverty and Inequality at a Detailed Regional Level: New advances in
spatial Microsimulation. In M. McGillivray and M. Clarke (Eds),
Understanding Human Well-Being, UN University Press, Tokyo. Harding, A.,
Phillips, B. and Kelly, S. (2004.) Trends in Housing Stress. Paper
presented at the National Summit on Housing Affordability, Canberra, 28
June.
Harding, A., Vu, N. Q., Tanton, R. and Vidyattama, V. (2009).
Improving Work Incentives and Incomes for Parents: the National and
Geographic Impact of Liberalising the Family Tax Benefit Income Test.
Economic Record, 85(Sept), pp. s48-s58.
Hynes, S., Morrissey, K. and O'Donoghue, C. (2006). Building a
Static Farm Level Spatial Microsimulation Model: Statistically matching
the Irish national farm survey to the Irish census of agriculture. Paper
presented at the 46th Congress of European Regional Science Association,
Volos, 30th August to 3rd September.
Kelly, D. (2004). SMILE Static Simulator Software User Manual
Teagasc. Rural Economy Research Centre (RERC), Teagasc Athenry
Publication, Galway.
Kelly, S., Phillips, B. and Taylor, E. (2006). Baseline Small area
Projections of the Demand for Housing Assistance. AHURI Final
Report no. 92, Australian Housing and Urban Research Institute,
Melbourne.
King, A. (1994). Towards Indicators of Housing Stress. Monograph
Series no. 2, Canberra, Department of Housing and Regional Development.
Landt, J and Bray, R. (1997). Alternative Approaches to Measuring
Rental Housing Affordability in Australia, Online Discussion
Paper--DP16, NATSEM, University of Canberra.
Lymer, S., Brown, L., Harding, A. and Yap, M. (2008). Predicting
the Need for Aged Care Services at the Small Area Level: the CAREMOD
Spatial Microsimulation Model. International Journal of Microsimulation,
2(2), pp. 27-42.
McNamara, J., Tanton, R. and Phillips, B. (2007). The Regional
Impact of Housing Costs and Assistance on Financial Disadvantage: Final
Report. Australian Housing and Urban Research Institute, Melbourne.
Melhuish, T., King, A. and Taylor, E. (2004). The Regional Impact
of Commonwealth Rent Assistance. AHURI Final Report no. 71, Australian
Housing and Urban Research Institute, Melbourne.
Miranti, R., McNamara, J., Tanton, R. and Harding, A. (2011).
Poverty at the Local Level: National and Small Area Poverty Estimates by
Family Type for Australia in 2006. Applied Spatial Analysis and Policy,
4(2), pp 145-171.
National Housing Strategy (1991). The affordability of Australian
housing. Australian Government Publishing Service, Canberra.
National Housing Strategy. (1992). National Housing Strategy:
Summary of papers, Canberra, Australian Government Publishing Service.
Nepal, B., Tanton, R. and Harding, A. (2010). Measuring Housing
Stress: How Much do Definitions Matter? Urban Policy and Research,
28(2), pp 211-224.
Pearson, C. (2007). House Rents Rising in Australia. Life in
Australia. Online version accessed 25 September 2007,
http://lifeinaustralia.wordpress.com
Procter, K. L., Clarke, G. P., Ransley, J. K. and Cade, J. (2008)
Micro-Level Analysis of Childhood Obesity, Diet, Physical Activity,
Residential Socioeconomic and Social Capital Variables: Where are the
Obesogenic Environments in Leeds? Area, 40(3), pp 323-340.
Productivity Commission. (2004). First Home Ownership. Productivity
Commission Inquiry Report No 28, Productivity Commission, Melbourne.
Rahman, A. (2009). Small Area Estimation Through Spatial
Microsimulation Models: Some Methodological Issues. Paper presented at
the 2nd General Conference of the International Microsimulation
Association, Ottawa, 8-10 June.
Rahman, A. (2011). Small Area Housing Stress Estimation in
Australia: Microsimulation Modelling and Statistical Reliability. PhD
thesis, NATSEM, University of Canberra.
Rahman, A. and Harding, A. (2011). Social and Health Costs of
Tobacco Smoking in Australia: Level, Trend and Determinants.
International Journal of Statistics and Systems, 6(4), pp. 375-387.
Rahman, A. and Harding, A. (2013). Prevalence of Overweight and
Obesity Epidemic in Australia: Some Causes and Consequences. JP Journal
of Biostatistics, 10(1), pp. 31-48.
Rahman, A. and Harding, A. (2014). Small Area Estimation and
Microsimulation Modelling. Chapman and Hall/CRC, London.
Rahman, A. Harding, A. Tanton, R. and Liu, S. (2010a).
Methodological Issues in Spatial Microsimulation Modelling for Small
Area Estimation. International Journal of Microsimulation, 3(2), pp.
3-22.
Rahman, A., Harding, A., Tanton, R. and Liu, S. (2010b). Simulating
the Characteristics of Populations at the Small-Area Level: New
Validation Techniques for a Spatial Microsimulation Model in Australia.
In JSM Proceedings, Social Statistics Section, Alexandria, VA: American
Statistical Association, pp. 2022-2036.
Rahman, A., Harding, A., Tanton, R. and Liu, S. (2013). Simulating
the Characteristics of Populations at the Small-Area level: New
Validation Techniques for a Spatial Microsimulation Model in Australia.
Computational Statistics and Data Analysis, 57(1), pp. 149-165.
Rowley, S. and Ong, R. (2012). Housing Affordability, Housing
Stress and Household Wellbeing in Australia. AHURI Final Report No.192.
Melbourne: Australian Housing and Urban Research Institute.
Sandel, M. and Wright, R. J. (2006). When Home is Where the Stress
is: Expanding the Dimensions of Housing that Influence Asthma Morbidity.
Archives of Disease in Childhood, 91(11), pp. 942-948. SCRGSP. (2007).
Report on Government Services 2006. Productivity Commission, Melbourne
(Steering Committee for the Review of Government Service Provision).
Stimson, R., Robson, A. and Shyy, T. K. (2008). Modelling the
Determinants of Spatial Differentials in Endogenous Regional Employment
Growth and Decline Across Regional Australia, 1996-2006. Paper presented
at the ANZRSAI 32nd Annual Conference /ARCRNSISSNational Conference,
Adelaide, 30 Nov. -3 Dec.
Stone, M.E. (2006a). What is Housing Affordability? The Case for
the Residual Income Approach. Housing Policy Debate, 17(1), 151-84.
Stone, M.E. (2006b) A Housing Affordability Standard for the UK.
Housing Studies, 21(4): 453-76.
Stone, M.E., Burke, T. and Ralston, L. (2011). The Residual Income
Approach to Housing Affordability: the Theory and the Practice.
Positioning Paper, Australian Housing and Urban Research Institute,
Melbourne.
Tanton, R., McNamara, J., Harding, A. and Morrison, T. (2009). Rich
Suburbs, Poor Suburbs? Small Area Poverty Estimates for Australia's
Eastern Seaboard in 2006. In A Zaidi, A Harding and P Williamson (Ed.),
New Frontiers in Microsimulation Modelling, Ashgate, London.
The Economist. (2011). The Economist House-Price Indicators,
Interactive overview of global house prices and rents. Online version
accessed 18 March 2012,
http://wwweconomist.com/blogs/freeexchange/2010/10/global hou se
prices.
Tumbarello, P. and Wang, S. (2010). What Drives House Prices in
Australia? A Cross-Country Approach. IMF Working Paper (WP/10/291),
International Monetary Fund, Washington, D.C., USA, pp. 1-24.
Williamson, P. (1996). Community Care Policies for the Elderly,
1981 and 1991: a Microsimulation Approach. In Clarke G (Ed)
Microsimulation for Urban and Regional Policy Analysis, Pion,
London, 64-87.
Wood, G., Forbes, M. and Gibb, G. (2005). Direct Subsidies and
Housing Affordability in Australian Private Rental Markets. Environment
and Planning C: Government and Policy, 23(5), pp. 759-83.
Yates, J. (2011) Housing in Australia in the 2000s: On the Agenda
Too Late? In H Gerard and J Kearns (Eds), The Australian Economy in the
2000s. Proceedings of a Conference, Reserve Bank of Australia, Sydney,
pp 261-296.
Yates, J. and Gabriel, M. (2006). National Research Venture 3:
Housing Affordability for Lower Income Australians. Australian Housing
and Urban Research Institute, Melbourne.
Azizur Rahman
Lecturer, School of Computing and Mathematics, Charles Sturt
University, Wagga Wagga, NSW, 2678, Australia. Email:
azrahman@csu.edu.au
Adjunct Associate Professor, University of Canberra, Canberra, ACT
2601, Australia. Email: azizur.rahman@canberra.edu.au
Ann Harding
Professor, National Centre for Social and Economic Modelling
(NATSEM), University of Canberra, Canberra, ACT, 2601, Australia. Email:
ann.harding@natsem.canberra.edu.au
Figure 4. Distribution of Households and Housing Stress Estimates
by Tenure Types in Australia, 2011.
a. Australian Households.
Owners 35.15
Buyers 35.00
Other tenure 2.92
26.93
Renters public 22.47
Renters private 4.46
b. Households in Housing Stress.
Renters public 6.88
Buyers 35.00
Renters private 59.55
0.34
Owners 0.30
Other tenure 0.04
Source: the Authors.
Note: Table made from pie graph.
Table 1. A Comparison of the Different Measures of Housing Stress.
30 only rule 30/40 rule 30/10-40 rule
General definition-- Specified More specified
a household is in definition--'a definition--a
housing stress if it household is in household is in
spends more than 30 housing stress if it housing stress if it
percent of its income spends more than 30 spends more than 30
on housing costs'. percent of its percent of its
income on housing income on housing
costs and the and places into the
household also bottom 10th to 40th
belongs to the income percentile of
bottom 40 percent of the equivalised
the equivalised income
income distribution'.
distribution'.
Assessing all forms of Ignores any marginal Ignores both the
housing stress in one housing stress. marginal and severe
flag. housing stress.
Only the absolute The relative income The relative income
household income is of the household is of the household is
considered. taken into account. used.
It is free from It is based on It is based on
equivalised household equivalised equivalised
income cut-off. household income household income
cut-off by the between 10 to 40
bottom 40 percent. percentiles.
Has been used in the Widely used in Used on a few
past. Australia. occasions.
No account is given to Proper treatment is Proper treatment is
the size of income given to the size of given of the size of
unit. the household income the household income
unit. unit.
Source: Rahman, (2011).
Table 2. An Illustration of Households Synthetic Weights Produced by
the GREGWT Algorithm for SLA level Microdata at in Australia.
Turning the national level household Household (HH)
weights in the Survey of Income synthetic weights
and Housing (SIH)--CURFs data into for the SLA
levels microdata
Unit HH Wkly Wkly Other HH NSW NSW NSW Other
record ID income rent variable weight SLA1 SLA2 SLA3 SLA
1 1 7 3 ... 1029 0 10.2 0 ...
2 2 11 4 ... 157 0 0 0 ...
3 3 11 4 ... 157 0 0 0 ...
4 4 11 4 ... 157 0 0 0 ...
5 5 11 0 ... 1003 2.45 9.64 16.38 ...
6 6 11 0 ... 1003 2.45 13.54 16.38 ...
... ... ... ... ... ... ... ...
... ... ... ... ... ... ... ...
... ... ... ... GREGWT ... ... ...
Re-
weight-
ing
53220 ... ... ...
8.4 12465 25853 27940 ...
million
No. of No. of households in
HHs in SLAs
AUS
Source: Rahman, (2011).
Table 3. Attributes of the Final Outputs file of the Model.
* SLA ID; * Renter private * Buyer in housing
* Total number households; stress;
of households; * Other tenure type * Renter public in
* Fully owner households housing stress;
households; (i.e., hospital, hostel, * Renter private in
* Buyer military tenure etc); housing stress;
households; * Total housing stress; * Other tenure
* Renter public * Owner in housing households in
households; stress; housing stress.
Source: the Authors.
Table 4. Housing Stress Estimates by the Statistical Division in
Australia, 2011.
ID SD (1) Name HS (2) %
105 Sydney 163655 21.17
205 Melbourne 135702 17.55
305 Brisbane 66718 8.63
505 Perth 53766 6.95
405 Adelaide 46749 6.05
307 Gold Coast 25787 3.34
110 Hunter 24764 3.20
115 Illawarra 17058 2.21
125 Mid-North Coast 15777 2.04
309 Sunshine Coast 14261 1.84
120 Richmond-Tweed 12680 1.64
315 Wide Bay-Burnett 11991 1.55
210 Barwon 9783 1.27
350 Far North 9055 1.17
320 Darling Downs 8011 1.04
605 Greater Hobart 7856 1.02
510 South West 7742 1.00
145 South Eastern 7716 1.00
805 Canberra 7700 1.00
240 Goulburn 7339 0.95
235 Loddon 6794 0.88
130 Northern 6654 0.86
345 Northern 6654 0.86
140 Central West 6568 0.85
255 Gippsland 5959 0.77
220 Central Highlands 5621 0.73
330 Fitzroy 5609 0.73
615 Northern 5339 0.69
150 Murrumbidgee 5234 0.68
410 Outer Adelaide 4500 0.58
ID SD Name HS %
340 Mackay 4368 0.57
155 Murray 4292 0.56
135 North Western 4204 0.54
620 Mersey-Lyell 3912 0.51
230 Mallee 3404 0.44
245 Ovens-Murray 3339 0.43
215 Western District 3203 0.41
705 Darwin 3171 0.41
250 East Gippsland 3016 0.39
312 West Moreton 2825 0.37
420 Murray Lands 2657 0.34
435 Northern 2637 0.34
425 South East 2153 0.28
535 Central 1870 0.24
515 LowerGreat South 1848 0.24
415 YorkeLower Nrth 1612 0.21
225 Wimmera 1486 0.19
525 Midlands 1423 0.18
710 NT -Bal 1334 0.17
610 Southern 1266 0.16
530 South Eastern 1245 0.16
430 Eyre 1147 0.15
160 Far West 727 0.09
545 Kimberley 685 0.09
325 South West 575 0.07
355 North West 529 0.07
540 Pilbara 449 0.06
520 UpperGreat South 430 0.06
335 Central West 224 0.03
000 Australia 773073 100
Note: (1) Statistical Division; (2) Total No. of Households in
Housing Stress. Source: the Authors.
