Toward Understanding Homelessness: The 2007 National Symposium on Homelessness Research. Incarceration and Homelessness. Empirical Basis for Defining the Issue


The basis of the link between incarceration and homelessness is the degree to which there is overlap among the populations  whether it is measured from the perspective of the prevalence of homelessness among an incarcerated population or prevalence of incarceration among a homeless population. That a substantial overlap exists should not be surprising given the similarities in profiles between the incarcerated population and the single adult homeless population, where incarceration is most prevalent. Both are both predominantly male, young, and minority (Langan & Levin, 2002; Burt et al., 1999; Mauer, 1999; Culhane & Metraux, 1999). People in both populations are typically poor and undereducated and possess few job skills (Western & Beckett, 1999; Lichtenstein & Kroll, 1996; Burt et al., 1999). Both populations are characterized by the research literature and the mainstream media as having high rates of disability, especially involving mental illness and substance abuse (Burt et al., 1999; Freudenburg, 2001; Conklin et al., 2000; Lamb, 1998; Peters et al., 1998).

Experience of Homelessness Among the Prison Population

Prisons are run by state or federal government entities. In contrast to jails, prisons incarcerate persons who are convicted of more serious offenses and who serve considerably longer sentences. Prisons are typically located at considerable distances from where incarcerated individuals lived prior to their conviction. In 1999, the average time served for state prisoners was 34 months (Hughes, Wilson, & Beck, 2001), and in 2002 the average time served in federal prison (felony convictions) was 49 to 50 months (U.S. Sentencing Commission, 2004). Most persons are either released from prison on parole, meaning that the last part of their sentence is served while they are in the community and supervised by a parole board, or are released without supervision after serving their full sentence in prison.

Lengthy periods of incarceration in remote locations often attenuate the social and family ties that are crucial for successful reentry into the community. Regained economic and residential stability almost always requires that a person receive, upon release from prison, support from family, social service agencies, faith-based organizations, or other parties interested in facilitating a smooth transition for the released individual. In the absence of such supports (and in some instances the absence of any type of effective discharge plan), individuals released from prison are at high risk for homelessness as well as other undesirable outcomes.

Only a handful of studies examine the overlap of prison and homelessness, and the extant literature has limited comparability due to variation in the study populations and the time frames used. However, taken together, the research suggests that about a tenth of the population coming into prisons have recently been homeless, and at least the same percentage of those who leave prisons end up homeless, for at least some period of time.

These studies include a Bureau of Justice Statistics (BJS) study (Hughes, Wilson, & Beck, 2001), which found that, among a nationwide survey of state prisoners expecting to be released in 1999, 12 percent reported being homeless at the time of their arrest. Another nationwide BJS study (Ditton, 1999) found that in 1998, 9 percent of state prison inmates reported living on the street or in a shelter in the 12 months prior to arrest. A California study (California Department of Corrections, 1997) reported that in 1997, 10 percent of the states parolees were homeless. This study also found that in urban areas such as San Francisco and Los Angeles, an estimated 3050 percent of all parolees were homeless. A 1999 Urban Institute three-site study of 400 returning prisoners with histories of drug abuse found that 32 percent had been homeless for a month or more at least once in their lifetimes, and 18 percent reported they were homeless for at least a month in the year after they were released from prison (Rossman et al., 1999). The Massachusetts Housing and Shelter Alliance (Hombs, 2002) reported that 9.3 percent, 10.5 percent, and 6.3 percent of all people exiting state prisons in Massachusetts in 1997, 1998, and 1999, respectively, went directly to shelters after release. In The Urban Institutes four-site Returning Home study (Visher, 2006), anywhere from 2 percent (Maryland, Ohio, and Texas) to 5 percent (Illinois) of respondents slept at a shelter during their first night out of prison. Another 3 to 4 percent slept at a hotel, motel, or rooming house the first night out.

Research by Metraux and his colleagues used administrative data to not only assess shelter use among a cohort of persons released from state prisons, but also to assess factors associated with higher likelihoods of shelter use following release. Metraux and Culhane (2004), looking at people exiting the New York State prison system to New York City locations, found that, with incomplete pre-incarceration data, 6.6 percent had a history of shelter use in the two-year period prior to incarceration and, with more complete post-incarceration data, 11.4 percent had an episode of shelter use in the two-year period subsequent to release. Metraux (2007), looking at persons released from state prisons to Philadelphia locations, found the rate of shelter admissions within two years to be 4.3 percent. The later study found that a proxy measure for mental illness was associated with a substantial increase in the likelihood of a shelter stay. In both studies, an indicator of a history of shelter use prior to incarceration, although incomplete, was a strong predictor of subsequent shelter stay in both studies. Increasing age was also significantly linked to higher likelihoods of post-release shelter use (and decreasing likelihood of reincarceration) in both studies, suggesting that as persons age out of criminal activity their risk for homelessness increases. Finally, the studies showed conflicting results on the effect of parole on homelessness, with the New York study showing release on parole to increase the likelihood of shelter stay, while the Philadelphia study showed a significant decrease in this likelihood.

These two studies, which merge data from multiple and large administrative datasets and use multivariate regression methods to assess various factors and their associations with the likelihood of shelter use, go beyond simply reporting rates and permit some insight into risk factors for homelessness among persons released from prison. For example, both studies confirm that shelter use prior to prison entry is the strongest predictor of post-release shelter use, a finding that lends itself well to being incorporated into a simple screening mechanism for targeting persons at-risk for homelessness. However, more studies of this type of sophistication, using other types of data, are needed to build a base of evidence for the role of such key factors as mental illness or parole supervision on the risk for homelessness after release from prison.

Experience of Homelessness Among the Jail Population

In contrast to prisons, most people are in jail for lesser offenses and only for a short time  the median stay is one day. Quick release commonly occurs when persons post bail or serve minimal time for minor offenses or charges are dropped. People will stay in jail longer when they are unable to post bail and remain in jail while awaiting trial or, following conviction, when persons convicted for lesser offenses serve their remaining time in jail. Some defendants are given split sentences, which involve a period of probation supervision after jail time is completed.

Persons serving longer jail sentences may have similar reentry issues as their imprisoned counterparts. However, even short-term incarcerations may disrupt lives and interfere with the ability to maintain employment and housing. Few jails have pre-release programs that provide case management services to link prisoners leaving jail to community services (Steadman & Veysey, 1997) and/or housing. Those on probation may have a number of court-ordered probationary conditions that make it difficult to return to live with family or friends or to find appropriate housing. Probation clients mandated to find employment right after release may be pressured to find a job regardless of how far the job is from their intended housing. While the housing options may be fewer given probation restrictions, being on probation may provide structured support to assist a released prisoners search for housing.

The few studies on homelessness among jailed populations suggest that the rates of homelessness for those exiting jails are lower and more loosely coupled with the jail release than they are for those exiting prison. However, because the jail population is much larger that the prison population, the number of persons exiting jails who become homeless is much larger. Metraux and Culhane (2003) found that, among 76,111 persons released from New York City jails in 1997, 5.5 percent entered New York City shelters for single adults in the subsequent two-year period. A recent BJS survey of jail inmates (James & Glaze, 2006) found that for jail inmates without a mental heath problem, 9 percent reported homelessness in the year before jail entry, as compared to 17 percent of those who had a mental health problem. In a sample count of jail inmates in Salt Lake City in July 2005 and January 2006, nearly 10 percent identified themselves as homeless (Reentry Policy Council, 2006). One study of frequent jail users found that 82 percent of repeat users of jail in a metropolitan area in the South were transient or homeless at jail intake (Ford, 2005).

McNiel, Binder, and Robinson (2005) looked at homelessness and mental illness among a jailed population. The study found that for the almost 13,000 jail episodes that were examined, in 16 percent of the episodes the person in question was homeless at the point of arrest, and in 18 percent of the episodes the person in question was diagnosed with a mental disorder. This rate of mental disorder was 30 percent among the episodes involving homelessness. Furthermore, homelessness and a dual diagnosis of severe mental disorder and substance-related disorders were associated with longer jail episodes.

As with examining prison to homelessness, the literature here is sparse and offers only a sketch of the nexus between jail and homelessness. The extent and dynamics here need further exploration and need to incorporate other dynamics such as was done in the study by McNiel and colleagues. Furthermore, given that both homelessness and incarceration, especially in jails, disproportionately impact impoverished, minority males (Harrison & Beck, 2006; Culhane & Metraux, 1999), it is unclear how much more elevated the rates of homelessness are among persons released from jail when compared to a comparable group of persons who have not been jailed.

Experience of Incarceration Among the Homeless Population

Just as homelessness is a common experience among persons incarcerated in jails and prisons, having had an incarceration experience, be it jail or prison, is a common occurrence among single adults who are homeless. Conversely, the studies that examine incarceration histories among homeless populations are also difficult to compare, but judging from the results it appears that upwards of 20 percent of a single adult homeless population can be assumed to have been incarcerated at some point.

Examples of such studies include Metraux and Culhanes (2006) examination of a sheltered single adult population in New York City. In this study, 23.1 percent experienced at least one incarceration episode in the two-year period prior to the date examined. This included 7.7 percent with a prison stay and 17.0 percent with a jail stay. According to the 1996 National Survey of Homeless Assistance Providers and Clients (NSHAPC), 49 percent of homeless adults reported at least one lifetime experience of having spent five or more days in a city or county jail, 4 percent had spent time in a military lock-up, and 18 percent had been incarcerated in a state or federal prison (Burt et al., 1999). A recent study of 1,426 community-based homeless and marginally housed adults found that 23.1 percent of study participants had a history of imprisonment (Kushel et al., 2005). Schlay and Rossis (1992) summary of twenty studies conducted in the 1980s found that, depending on the study, 4 percent to 49 percent of the homeless population report serving time in prison. The mean across the studies was 18 percent.

When focusing on persons diagnosed with mental illness, the intersections between homelessness and incarceration appear to be intensified (Metraux & Culhane, 2004; Ditton, 1999). In contrast, however, Solomon and Draine (1999) found more tenuous links between criminal justice history and homelessness in a sample of 325 psychiatric probation and parole clients. Other studies examining homelessness and criminal justice-related risk factors among persons with mental illness focus primarily on arrests, without examining incarceration specifically. Several studies here have found housing instability to be associated with an increased likelihood of coming into contact with police and of being charged with a criminal offense (Brekke et al., 2001; Clark, Ricketts, & McHugo, 1999; Martell, Rosner & Harmon, 1995).

Metraux and Culhane (2006, 2004) also present evidence suggesting that the trajectories between homelessness and prison and homelessness and jail vary. The links between prison and homelessness are much more immediate, with an episode of homelessness being most likely to occur within 30 days of a prison release. This suggests that homelessness among persons released from prison is a reentry issue. This is consistent with research that shows persons released from prison to be at greatest risk for a variety of undesirable outcomes during this time period (Nelson, Deess, & Allen, 1999; Travis, Solomon, & Waul, 2001). Furthermore, Metraux and Culhane (2004) find that shelter use increases, albeit modestly, the risk for a subsequent reincarceration. In contrast, Metraux and Culhane (2006) found that shelter and jail use tended to follow a more sequential pattern featuring multiple stays in each system and a more prolonged pattern of residential instability.

View full report


"report.pdf" (pdf, 242.57Kb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®