Professor Marybeth Shinn is a community psychologist who has conducted research on homelessness in the United States for over two decades. Studies have included longitudinal and cross-sectional studies of risk factors for homelessness, and experimental studies of interventions to end it for individuals with mental illness and families. She has consulted with government agencies and policy groups in the United States and Canada. In July this year she gave a keynote address on homelessness at the Trans-Tasman Community Psychology Conference and was a keynote presenter at the APS Roundtable on Homelessness.
Homelessness is on the rise throughout the developed world, including in countries like Australia that has the resources to put an end to it. Prime Minister Kevin Rudd has called homelessness a "national obscenity" and the Australian Government has committed resources to reduce homelessness in a recent White Paper (Homelessness Taskforce, 2008), so prospects for action here are much better than in my own country (the United States). Ending homelessness requires an understanding of its prevalence, and also its origins in social policy, social exclusion and individual vulnerabilities. Homelessness is not primarily a psychological phenomenon, but psychologists can help to end it.
Prevalence of homelessness
The prevalence of homelessness depends on how it is defined. The White Paper adopts the classification scheme by Chamberlain and MacKenzie (1992) that distinguishes primary homelessness (rough sleeping or living in improvised dwellings), secondary homelessness (living in specialist homeless services or with friends and relatives) and tertiary homelessness (living in boarding houses or caravan parks). Both primary homelessness and living in specialist homelessness services would be called literal homelessness in the United States. Europeans tend to adopt a broader definition focusing on tenuous or inadequate ties to housing.
Prevalence depends even more on whether one looks at homelessness on a single night (point prevalence) or over some longer period. For example, in New York City, which has excellent records of the number of people staying in homeless shelters (specialist services), over 10 times as many people (3.27% of the City's population) stayed in shelter over a five-year period in the 1990s as on a single night (Culhane et al., 1994). Lifetime prevalence is even higher, as shown by telephone surveys of adults living in conventional dwellings even though such surveys miss people who are currently homeless, and people in mental hospitals or prisons who have high rates of prior homelessness. A nationally representative survey found that 7.4 per cent of the adult population in the United States had been literally homeless at some point in their lives (Link et al., 1994).
No comparable telephone survey has been done in Australia, but it is safe to assume that the number of Australians who have been homeless over a period of years or even a lifetime is far higher than the one-night figure of 105,000 cited in the White Paper. Indeed, one can surmise as much from information scattered through the White Paper. The one night census in the Supported Accommodation Assistance Program (SAAP) is 48,200; the average episode of supported accommodation is 50 days; half of new requests for accommodation are turned down; only 19 per cent of people who are homeless have contact with SAAP on any given day. These data suggest that at least a quarter of a million Australians experience literal homelessness over a year.
The difference between one-night counts and longer periods is critically important. The White Paper outlines plans to build up to 50,000 affordable rental homes, to build up to 2,700 homes for people who have been homeless or are at risk, and to build or upgrade up to 9,000 homes in remote Indigenous communities, for a total of up to 61,700 units over the next decade. That might well resolve the problem of homelessness for the 105,000 people estimated to be homeless on a single night, but it will not solve the problem for those who are homeless over the course of a year, much less a decade, and will do little for the 1.1 million households (10% of all households in Australia) that were in financial stress in 2007. Psychologists versed in epidemiology can help policy makers and the media understand the difference between point prevalence and period prevalence, and the implications for ending homelessness.
Origins of homelessness
Social policy
Comparison of rates of homelessness across developed countries can elucidate how social policy influences homelessness. Toro and colleagues (Toro et al., 2007, and others cited by Shinn, 2007) have used telephone surveys in a number of countries to examine lifetime rates of homelessness as shown in the first row of Table 1. The United States and the United Kingdom have substantially higher rates than several countries on the European continent. The subsequent rows in the table suggest that rates of homelessness are related to rates of income inequality and social benefits.
The second row in the table shows the Gini coefficient, the most widely used index of inequality, from the latest report of the United Nations Development Programme (2007). In general, the English speaking countries, led by the United States, have higher levels of inequality than countries on the European continent or Japan. Perhaps more relevant to homelessness is the percentage of income or consumption by the lowest 10 per cent of a country's population, shown in the third row. The United States is worst, but Australia is not far behind.
Inequality is a joint function of market income and of tax and transfer policies. The fourth row of the table shows social benefits and other transfer programs as a percentage of gross domestic product (GDP), as calculated by political scientists Alesina and Glaser (2004). Unfortunately, their work does not include Australia. The United Kingdom and especially the United States are less generous than continental Europe in providing a social safety net for poor citizens. The last row, also from Alesina and Glaser (2004), shows government expenditure on family programs. The high rates of homelessness among families with young children in the United States may well stem from the low levels of expenditures to help families. Social housing is even harder to compare across countries, because housing assistance takes so many different forms, but lack of housing that is affordable for poor people has certainly been implicated in the rise of homelessness in the United States.
In democracies, social policies reflect underlying social attitudes. Alesina and Glaser (2004) show that across 30 nations, the average belief that poverty is society's fault (in survey data) explained 43 per cent of the variance in social expenditures; among wealthier countries with per capita GDP above $15,000 in 1998, it explained 82 per cent. Racial heterogeneity also explained 66 per cent of the country-to-country variance in spending. The cross-national data suggest that when other people look more like us, we are less likely to blame them for their poverty and more generous in offering help. Psychologists can attempt to understand and change these social attitudes. Among the most important are attitudes underlying social exclusion.
Social exclusion
Every study that has looked has found higher rates of homelessness among socially excluded groups, although not every socially excluded group is affected. In the United States, African Americans and Native Americans have high rates of homelessness (Burt et al., 1999). In Japan it is minorities such as the Ainu, Koreans, Okinawans and groups such as Eta and Hinin who fall outside of the main social classes (Okamoto, 2007). In France, Africans and people from overseas departments are more likely than others to become homeless (Firdion & Marpsat, 2007). And in Australia, Aboriginals and Torres Strait Islanders are at special risk (Homelessness Taskforce, 2008).
Psychologists can help to identify the mechanisms of social exclusion, in order to combat them. In the United States I suggest that four are centrally related to homelessness: current discrimination in income and employment, past discrimination leading to disparities in wealth, current discrimination in housing, and differential rates of imprisonment. For example, median family income for Black Americans is only 55 per cent as high as for White Americans and median household net worth is only one eighth as high (Conley, 1999). Because much of wealth is tied up in housing, this means that African American households have less secure housing and fewer housing resources to offer relatives or friends. Residential discrimination can be examined by sending paired testers of different races, who are otherwise comparable in employment, education, income, assets, debt levels and family circumstances, to apply for the same housing units. In the latest tests, African American renters received less favourable treatment than white renters over a fifth of the time (Turner et al., 2002). Finally, rates of imprisonment in the United States are the highest in the world and black men are 7.1 times as likely as their white counterparts to be incarcerated (Harrison & Karberg, 2004), depriving African American families of wage earners and providing further grounds for discrimination in housing and employment after men are released (Shinn, 2007). Australian psychologists might identify, and counteract, the mechanisms by which social exclusion leads to higher rates of homelessness for minority groups in Australia.
Individual vulnerabilities
Of course, individual vulnerabilities matter to homelessness too, but I do not start there, because the fundamental attribution error suggests that people attribute too much explanatory power to individual actions and characteristics, and too little to environmental circumstances. Introductory research methods taught us that correlation is not causality. Individual characteristics may lead to homelessness or homelessness may create individual problems. For example, Johnson and Chamberlin (2009) found that homeless people in a large Melbourne sample were as likely to develop mental illnesses after becoming homeless as before. Other variables, such as poverty, may cause both homelessness and other adverse conditions. A final more subtle point is that characteristics of people who are homeless at a particular point in time are likely to differ from characteristics of people who are homeless over a longer period. A point-in-time survey will include more people with problems that lead to difficulty extricating themselves from homelessness.
Individual characteristics associated with homelessness involve economic, human and social capital, and life transitions. With respect to economic capital, it is no surprise that homeless people are poor. They are also more likely to have grown up poor. One study in New York City found that growing up in poverty was related to homelessness for families, but did not predict housing stability after families received subsidised housing from the City, suggesting that lack of resources, rather than some sort of ‘culture of poverty' was at issue (Shinn et al., 1998).
Individuals who experience homelessness often have low levels of human capital, including educational and occupational attainment. Disability caused by mental health, physical health and substance abuse problems can also be thought of as a deficit in human capital, leading to difficulties in earning a living. According to the Australian Government White Paper, 36 per cent of SAAP clients had mental health or substance problems, which makes this group an important minority. Both sorts of problems are more common among single individuals than among families who become homeless.
Like mental illness and substance abuse, social capital, in the form of positive social ties, has a bi-directional relationship with homelessness (e.g., Firdion & Marpsat, 2007). People who lack a social safety net, or who wear out their welcome with family and friends, are more likely to become homeless, but homelessness also disrupts social ties. Social ties may be particularly important for groups who are dependent on others, such as older adults and adolescents.
Negative relationships including domestic violence and experiences of childhood abuse or foster care are also frequently reported by homeless people. The White Paper suggests that domestic violence may be particularly important for women in Australia, afflicting 55 per cent of women with children and 37 per cent of young single women at SAAP. However research also suggests that the more carefully one measures violence against women, the higher the prevalence rates are among women who are poor, and the less difference there is between women who become homeless and poor women who stay housed (e.g., Bassuk et al., 1996). Reducing domestic violence is important in its own right, but without comparison groups it is difficult to say how much violence contributes to homelessness.
Transitions can be times of risk. In the United States, episodes of poverty frequently begin with the birth of a child (Waldfogel, 2001), so it is not surprising that episodes of homelessness often do as well (Weitzman, 1989). In Japan, loss of lifetime employment puts older men at risk; when they lose jobs, they frequently lose company housing (Okamoto, 2007). In Australia, youth leaving foster care and adolescents leaving home have been at particular risk, although the White Paper suggests that risk among this group may be decreasing.
Relationships among risk factors
How are these risk factors at different levels of analysis related? One way of thinking about the relationship is that income inequality, social welfare programs and the affordability of housing influence the overall rates of homelessness, whereas social exclusion and individual vulnerability influence who becomes homeless. The children's game of musical chairs provides a useful analogy. The players are poor households and the chairs are the housing units that they can afford. If there are more poor households than housing units they can afford, some people will be left homeless when the music stops. The particular people who fail to secure housing will be those suffering social exclusion or individual vulnerabilities.
A second way to think about combining levels is risk amplification. Individuals who become homeless may start with some individual risk factors, but these are amplified the longer that they are without homes. People engage in more risk behaviours, experience more victimisation and become more entrenched in homelessness the longer they spend in that state. An implication is that it is important to intervene quickly to re-house people who fall into homelessness.
There can also be interactions across levels of analysis, such that factors at one level can protect against factors at another. For example, homelessness among families is sometimes attributed to single parenthood. International comparisons show that Belgium and the United States have very similar rates of single parenthood, but the United States has a far higher rate of family homelessness. It is likely that family policy in Belgium makes the difference: single parents are less likely to be poor. Similarly, housing subsidies in New York City (a policy intervention) compensated for all the individual risk factors that led families to become homeless, leading formerly homeless families to the same levels of stability as other poor families (Shinn et al., 1998).
Roles for psychologists
Although the causes of homelessness are not primarily psychological, there are many roles for psychologists in ending it. The first is in educating policy makers, the media and the public. One important issue is the difference between point and period prevalence, and the vastly greater efforts that are needed to house the numbers of people likely to experience homelessness over a decade compared with a single night. A second is in the implications of the fundamental attribution error. It is wonderful that the Australian Psychological Society wants to do its part in ending homelessness, but it is important not to imply that the causes of homelessness are essentially psychological, letting policy makers off the hook. However, psychologists can work to understand attitudes towards social expenditures and the mechanisms by which social exclusion is associated with homelessness.
Psychologists can also design, implement and study programs to prevent and end homelessness for particular groups. For many poor people, affordable housing may be enough. Women additionally need safety from domestic violence. Adolescents need programs that respect their needs for autonomy and relatedness, and that provide support if they cannot be reunited with families. The French Foyer Model in which young people receive shared accommodation with supports on condition that they participate in education, training or employment (Homelessness Taskforce, 2008) seems promising. Adults with mental illness need ongoing assistance, such as supported housing. An especially successful and cost-effective model in the United States is called Housing First. Individuals with serious mental illness and co-occurring substance problems receive their own apartments with private landlords, directly from the streets, and ongoing services from interdisciplinary Assertive Community Treatment teams (e.g., Gulcur et al., 2003). Critical Time Interventions (e.g., Susser et al., 1997) that help to stabilise people in communities can be helpful for individuals leaving institutions. Older adults may need ongoing care in specialist facilities.
Psychologists should use their skills as researchers to evaluate intervention programs with rigorous designs. In the US, we have too often failed to invest in research to determine what programs work for whom, or have conducted studies that were not designed to yield meaningful information. Australia should do better.