Evidence on the Relationship between Unaffordable Housing and Poor Health

Follow the recent Global Financial Crisis, the role of Australian Governments in providing assistance to households unable to access, afford, or maintain adequate housing has come increasingly into focus. Australia is experiencing ongoing and substantial housing affordability decline, and our cities have some of the world’s highest rates of housing un-affordability. Housing affordability directly affects the type, quality, and security of housing that individuals can access. This paper examines the relationship between poor health and poor housing affordability for Australians, to answer two essential questions for Australian policy makers: Does poor health lead to unaffordable housing? And does unaffordable housing affect people’s health? Analysis was based upon two large Australian datasets, the Household, Income and Labour Dynamics in Australia (HILDA) Survey and the General Social Survey, (GSS). We highlight the populations most vulnerable to affordability-related poor health such as lone parents and their children and older renters. This study contributes empirical evidence, allowing us to examine if there is a fundamental bi-directional relationship between poor housing affordability and health even when demographic and socio-economic factors have been, to a large extent, accounted for. INTRODUCTION – HOUSING AFFORDABILITY AND HEALTH The Universal Declaration of Human Rights (1948 Article 25) allocates to all people “the right to a standard of living adequate for the health and wellbeing of himself and his family”, and identifies housing as a central component of that standard of living. Like other similar policy and legislative documents, the Universal Declaration is based upon the intuitive understanding that housing and health are related, but 60 years after the Declaration, surprisingly little is still known of the detail of that relationship. This paper examines an important and overarching part of the housing and health relationship, the way that housing affordability and health are related. How exactly are housing and health related? Are there particular aspects of housing e.g. affordability, quality, tenurethat are related to health? This paper considers the relationship between one overarching dimension of housingaffordability and its relationship with health and well being. We focus on the bi-directional housing affordability and health relationship for Australians to establish how housing affordability may influence health, and how health may influence housing affordability. Among the essential housing rights afforded by the Universal Declaration, housing affordability has a dominant role. Affordability directly affects the quality, security, and appropriateness of the housing that individuals are able to obtain. Further, housing costs, which are commonly the largest household expenditure for Australians (ABS, 2007), also affect the ability of low income households to meet other essential needs (such as food security Kirkpatrick and Tarasuk, 2011; or health care Pollack et al 2010). The necessity to meet housing costs means that many households ‘trade off’ essential needs to meet their housing costs, this impacts on the broader lives of the individuals within them, most critically, on the ability to secure employment or education, access basic services and maintain their connections to family and friends. It should be further noted that housing is “both a consumption and an investment good” (Mandic and Cirman 2011) that can be transferred inter-generationally, this means that current inequalities of access to affordable housing can be compounded over time (and inherited by subsequent generations). The importance of housing as a central determinant of the health and wellbeing of individuals has been well established (though we note, not fully elucidated) across academic and policy literatures, and a connection between these two major areas of Australian life predicates the shape of much government expenditure and welfare policy. A number of studies have examined and established a clear relationship between housing affordability and health (for example, Bentley et al, 2011; Shaw et al. 1999; Taylor et al. 2007; Ford and Burrows 1999). Some focus on the particular health effects of affordable housing, for example, recent work by Bentley, et al (2011) found a small but significant causal relationship between housing affordability and mental health. Importantly, in this longitudinal study of more than 15,000 Australians a mental health effect from housing affordability was only found for individuals living in low incomes households. Similarly, Taylor et al. (2007) found a relationship between mental health and the pressures of meeting housing costs, especially for heads of households. Their study is additionally interesting as they established a gender-related difference in the health effects. The impact of housing affordability on health has also been observed in qualitative work. For example, Hulse and Saugeres (2008) who found that housing affordability impacts on mental health resulting in heightened stress and anxiety, and in the earlier longitudinal research by Nettleton and Burrows (1998) they found an association between mental health and mortgage indebtedness. Studies of the relationship between dwellings quality and health are relevant as affordability is implicated in the quality of the dwelling that people can afford. For example, heating has been shown to affect physical health (Howden-Chapman et al.2007; Gemmel 2001; Naughton et al. 2002), further, warmer houses were shown by Howden-Chapman et al. (2007) to be associated with improved self-assessed health and fewer GP visits. Similarly, damp has been shown to be associated with respiratory illness (Bonnefoy et al. 2003; Shaw, 2004; Bornehag et al. 2001; Shenassa et al. 2007). Considering the features and layout of dwellings, noise exposure was found to be related to overcrowding (Evans et al. 2003), and was statistically linked to worse mental and physical health (Howden-Chapman & Wilson 2000, pp. 140–4). Recently, Cutts et al (2011) in their study of housing insecurity found that overcrowding was associated with food insecurity for children. Affordability is also likely to affect health indirectly via tenure. Research examining the health outcomes of tenure consistently finds owner-occupation to be the ‘healthiest’ tenure (for example, Smith et al. 2004, p. 579; Macintyre et al. 2001, p. 29; Macintyre et al. 2003), associated with various health benefits (Cairney & Boyle 2004, p. 161), for example higher psycho-social wellbeing (Kearns et al. 2000), and lower risk ratios for mortality (Breeze et al. 1999). In comparison, rental tenure has been associated with negative health measures, such as poorer self-reported health (Windle et al. 2006), coronary heart disease (Woodward et al. 1992) and risky health behaviours such as smoking (Kendig et al. 1998). A recent study by Pollack, Grinnin and Lynch, (2010) of housing affordability, tenure and health, not only found increased odds of poor selfrated health among individuals in unaffordable housing, but they also found that the health effects of poor affordability were heightened for individuals who were renting rather than owning. Importantly, it is likely that a substantial proportion of the explanation for these health differences is related to the way that tenure and affordability are bound together. Finally, not only does the location of the housing that individuals occupy affect affordability, but affordability affects the location that individuals are able to access housing in. This locational effect has been shown by many authors (such as Acevedo-Garcia et al. 2004; Macintyre et al. 2003) to influence individual health by the access it provides (to social connections, or green spaces), as well as distance from perceived crime (Ross & Mirowsky 1999; Stafford et al. 2007). This paper seeks to answer two questions, aimed at exploring the bi-directional relationship between housing affordability and health for the Australian population: 1. Does poor health lead to unaffordable housing? And 2. Does unaffordable housing influence individual health? METHODS This paper is based upon a quantitative analysis of two large surveys which measure aspects of housing and health in the Australian population. Though a range of datasets can be used to measure housing affordability or health across the Australian population, few reliable large-scale datasets allow the measurement of both housing affordability and health characteristics. This analysis is therefore based upon two large datasets: the General Social Survey (GSS) and the Household, Income and Labour Dynamics in Australia (HILDA) survey. Together, these provide information which allows us to explore the unaffordable housing and health relationship The GSS is a large-scale Australian Bureau of Statistics survey specifically designed to provide reliable estimates at the national level and for each state and territory. It collects data on a range of personal and household characteristics of people aged 18 years and over resident in private dwellings, throughout nonremote areas of Australia every four years. The most recently available dataset, collected in 2006, provides data on a nationally representative sample of 13,375 households (ABS 2006). The GSS was used in this analysis to measure unaffordable housing characteristics across a number of population characteristics. The HILDA Survey is a household-based longitudinal survey conducted annually since 2001 (Watson 2008). It collects a wide range of data on households and the individuals living within those households by surveying adult members of participating households every year via face-to-face interview and a selfcompletion questionnaire. For this analysis we examine demographic, socio-economic, housing and health data collected between 2001 and 2007 from 12,968 survey respondents (59,233 observations over seven survey waves). The HILDA dataset was used to examine our primary research questions on the relationships between housing and health. Because it is a longitudinal survey, the HILDA dataset also provided us with the ability to examine whether poor health in preceding years predicts current unaffordable housing, and whether unaffordable housing predicts physical or mental health after taking into account prior health status. In other words, the longitudinal nature of the data allowed us to look at possible causality in both directions between housing affordability and health. In order to assess housing affordability across individuals and their households, two measures of housing affordability were used in this analysis. The first was housing unaffordability which used the well established 30/40 rule, where individuals living in households with an income in the lowest 40 per cent of the national income distribution, and paying more than 30 per cent of equivalised disposable household income in housing costs (rent or mortgage), were classified as being in unaffordable housing. This ratio approach has been widely used in Australian research and policy, (National Housing Strategy (1992); Harding et al (2004); Commonwealth of Australia (2008); Seelig & Phibbs (2002)) and shown to be a robust measure allowing housing affordability to be examined over time (Nepal et al 2010). The HILDA dataset was used to derive this measure and the prevalence in the Australian population was estimated for 2006. A second measure, housing stress, was chosen to reflected self-reported housing affordability problems. This was based on responses to a question in the GSS, which asked respondents if they had had difficulty in paying rent or mortgage and utilities (electricity, gas, telephone bills) in the past 12 months. The prevalence of housing stress was also estimated for 2006. In order to examine health broadly, three measures of health were used: 1. Self-assessed health: An overall measure of health, in which an individual rated his/her health on a five-point scale from 1=excellent to 5=poor. 2. and 3. Physical and Mental Health: Both assessed using the widely recognised Short Form 36 (SF-36) tool, which takes responses to 36 questions about health and collapses these to subscales measuring different aspects of health. These are then used to calculate separate summary scores for mental and physical health, both of them on a standardised scale from 0– 100. A higher score reflects better health Prevalence estimates were calculated using survey estimation commands in Stata 11.0, and weighted to the Australian population. To examine demographic and socio-economic predictors of unaffordable housing, we used logistic regression models, adjusted for age. To examine poor health as a predictor of unaffordable housing, we averaged health scores across the previous three years and then included quintiles of this average score in a logistic regression model adjusted for age, sex, country of birth, educational attainment, highest occupation level in household, disposable income and household structure. These models estimated odds ratios (and 95% confidence intervals) for the association with unaffordable housing for each quintile of health relative to the healthiest quintile (top 20% of scores). To examine unaffordable housing as a predictor of physical and mental health, we included unaffordable housing in a linear regression model of health scores adjusted for baseline health, age, sex, country of birth, educational attainment, highest occupation level in household, disposable income and household structure. Regression coefficients obtained from these models represent differences in SF-36 summary score comparing individuals living unaffordable housing with those living in affordable housing. DESCRIPTIVE RESULTS Based on our analysis of HILDA data, just below 6 per cent of Australian adults (around 780,000 individuals), are living in households that are both low income and paying a large proportion of income in rent or mortgage costs (Table 1). This overall prevalence is consistent with previous findings using the same measure (for example, ABS, 2004, Catalogue No. 1370.0). We further find that the prevalence of unaffordable housing is slightly higher within urban Australia, than Inner Regional areas. More remote parts of Australia have a lower prevalence. Interestingly in this table, we show that while around 6 per cent of households are strictly defined as being in unaffordable housing, much higher proportions (more than double) report that they had been unable to meet mortgage, rent or household bill payments on time in the last 12 months. Table 1 shows that this is around 12 per cent, or almost 2 million Australians reported experiencing this form of housing stress. TABLE 1: ESTIMATED PREVALENCE OF INDICATORS OF UNAFFORDABLE HOUSING BY SEX & LOCATION, 2006. Unaffordability (30/40 rule) Housing stress (unable to pay rent, mortgage, bills on time in past 12 months)