Unofficial policy: access to housing, housing information and social services among homeless drug users in Hartford, Connecticut

Background Much research has shown that the homeless have higher rates of substance abuse problems than housed populations and that substance abuse increases individuals' vulnerability to homelessness. However, the effects of housing policies on drug users' access to housing have been understudied to date. This paper will look at the "unofficial" housing policies that affect drug users' access to housing. Methods Qualitative interviews were conducted with 65 active users of heroin and cocaine at baseline, 3 and 6 months. Participants were purposively sampled to reflect a variety of housing statuses including homeless on the streets, in shelters, "doubled-up" with family or friends, or permanently housed in subsidized, unsubsidized or supportive housing. Key informant interviews and two focus group interviews were conducted with 15 housing caseworkers. Data were analyzed to explore the processes by which drug users receive information about different housing subsidies and welfare benefits, and their experiences in applying for these. Results A number of unofficial policy mechanisms limit drug users' access to housing, information and services, including limited outreach to non-shelter using homeless regarding housing programs, service provider priorities, and service provider discretion in processing applications and providing services. Conclusion Unofficial policy, i.e. the mechanisms used by caseworkers to ration scarce housing resources, is as important as official housing policies in limiting drug users' access to housing. Drug users' descriptions of their experiences working with caseworkers to obtain permanent, affordable housing, provide insights as to how access to supportive and subsidized housing can be improved for this population.

Research has shown that substance use problems afflict anywhere from 28 to 67% of homeless individuals [3][4][5][6][7] and that substance abuse increases individuals' vulnerability to homelessness [8][9][10]. Others have argued that structural changes, for example the loss of manufacturing jobs and affordable housing stock in inner-city neighborhoods, are the causes of the increase in homelessness over the past two decades [11,12]. More recently researchers have argued that both are important considerations. While personal characteristics, such as drug use, may not in themselves cause homelessness, they make certain individuals more vulnerable to homelessness given an increasingly competitive housing market [13][14][15][16][17][18]. Structural factors determine why pervasive homelessness exists in this historical time, while individual factors explain who is least able to compete for scarce affordable housing.
Structural factors that may contribute to drug users' greater vulnerability to homelessness include official and unofficial housing policies that determine eligibility for and access to various housing and welfare subsidies. The effects of housing policies on drug users' access to housing have been understudied to date. Official policies include the federal "One Strike and You're Out" law (P.L. 104-120, Sec.9) passed in 1996 that allows federal housing authorities to consider drug and alcohol abuse and convictions of people and their family members when making decisions to evict them from or deny access to federally subsidized housing. Many states, including Connecticut, have opted out of this law. Flat line funding of federally subsidized housing programs, such as the Housing Choice voucher program (formerly known as Section 8), and Shelter Plus Care, have limited the number of subsidies available. Both programs allow recipients to choose their own apartments on the competitive market and pay a proportion of the rent depending on recipients' income. While Connecticut does not consider drug convictions in decisions to deny applications for housing vouchers, criminalization of drug use affects drug users' access to housing in other ways, as criminal background checks are routine in many apartment rental applications. Other policies which have impacted drug users' access to housing include the Personal Responsibility and Work Opportunity Act of 1996, popularly known as Welfare Reform, in particular the elimination of the SSI Addiction Disability and a ban on receiving welfare benefits for convicted drug offenders [19][20][21].
Less understood are the effects of "unofficial" policy on drug users' access to housing. In this paper, unofficial policy is defined as the way in which official policy is implemented or enforced, or not, and the operating policies of organizations or individuals. This definition of unofficial policy borrows from Lipsky's [22] idea of "street level bureaucrats." For Lipsky, low-level employees who directly interact with the public, for example social workers, police officers, or unemployment counselors, ought to be viewed as policy makers rather than implementers of policy. As Lipsky puts it, the "decisions of street level bureaucrats, the routines they establish, and the devices they invent to cope with pressure, effectively become the public policy they carry out (xxii)." The pressures of work faced by street level bureaucrats include an almost infinite demand for services by the public along with inadequate resources available to workers to meet these demands. Street level bureaucrats use a number of strategies to ration services, including limiting access to information about services, creating categories of clients, exercising discretion in distributing benefits and sanctions, and increasing the costs of applying for services. Lipsky does not fully consider, however, the ways official policy may shape unofficial policy. For example, an official policy that cuts federally subsidized housing may create periods of relative scarcity, which may have direct effects on the pressures and coping mechanisms street level bureaucrats use.
Unofficial policy may help explain research that has shown that substance users are significantly less likely to exit homelessness [23] or access social services [24,25] than non-substance abusing homeless. For example, Zlotnick [23] and colleagues found that exit from homelessness was associated with greater social support and greater contact with service providers for homeless without a current substance abuse disorder, but not for homeless with current substance abuse. They suggest that this may be because substance using homeless persons may be more focused on obtaining and using drugs than gaining access to services, or that they may be unable to mobilize their social support networks. An alternative explanation, consistent with Lipsky's view of "street level bureaucrats" is that service providers may choose to devote more of their limited resources to homeless individuals without substance abuse problems whom they may see as more "deserving" or as having a greater chance at success in maintaining their housing. Supporting this second explanation is a study by Dohan and colleagues [26] that found that welfare workers generally applauded welfare reform's renewed attention to deservingness, including program emphases on client self-sufficiency and personal accountability.
Some drug users face multiple barriers to accessing and maintaining stable housing, including long-term substance abuse, mental health issues, and histories of arrest. Such individuals have been identified by researchers and advocates as "chronically homeless" [27][28][29][30]. As a result, alternatives to emergency shelters to house this population have begun to be proposed, including the Housing First Model, and supportive housing programs [30]. The Housing First model advocates for the provision of housing to drug addicted or mentally ill homeless that is not contingent on their "readiness," i.e., completing residential treatment programs or maintaining sobriety for a period of time. Rather, they advocate for housing with supportive services attached, including mental health services, addiction services, and assistance in budgeting, obtaining employment or maintaining an apartment. This is in contrast to the traditional Continuum of Care model that consists of several components including outreach, treatment and transitional housing, then supportive housing. Continuum of Care seeks to enhance clients' "housing readiness" by requiring sobriety and compliance with psychiatric treatment before placement to more permanent housing [30]. Connecticut has funded several supportive housing projects that provide affordable, service enriched rental housing for homeless and at-risk populations, many of whom are coping with mental illness, histories of substance addiction, or HIV/AIDS [31]. Some of these supportive housing programs follow the Housing First model and allow residents to choose which, if any, supportive services they wish to utilize. Other programs require residents to fulfill program requirements, such as active involvement in job training or substance abuse treatment. The effects these differing philosophies have on the ways in which service providers implement programs, i.e. the unofficial policy of these programs, has not been studied. This paper will look at the "unofficial" policies that affect drug users' access to housing. Using longitudinal, indepth interviews with both housed and homeless drug users and key informant interviews with housing caseworkers in Hartford, Connecticut, we will look at the process by which drug users receive information about different housing subsidies and welfare benefits, and their experiences in applying for these.

Design
We conducted longitudinal in-depth interviews with active drug users to explore their housing status and stability over time, and barriers and facilitators drug users face in accessing housing. Eligibility criteria included being over 18 years old and having used cocaine, crack or heroin within the last 30 days at the first interview. We sought to recruit active users of heroin and cocaine because previous research conducted by our research team indicated that these were the illicit drugs most frequently abused in Hartford, and that users of these substances had steadily increasing rates of homelessness over the past thirteen years [32][33][34]. Purposive sampling was used to identify and recruit drug users in various housing situations, including: 1) supportive housing, 2) subsidized housing, 3) non-subsidized housing, 4)"doubling up" with family or friends, 5) homeless in shelters, and 6) homeless on the street. We defined "doubling up" as the practice of temporarily moving in with family or friends.
In addition, we conducted key informant interviews and focus group interviews with service providers including shelter, supportive housing, and substance abuse treatment staff, and housing advocates in order to obtain service provider perspectives on the barriers and facilitators drug users face in accessing information, housing and services.

Participants
Sixty-five drug users were interviewed at baseline. Fortysix percent of the sample was African American, 46% Puerto Rican, 8% non-Hispanic white, and 46% women. Participants were ethnically similar to other research projects conducted with active drug users in Hartford, although women were oversampled [33,35]. Fifty were located for follow-up interviews at three months. Of those who were not located, four were confirmed to be in jail, and one was confirmed to have moved out of state. Excluding those individuals who were in jail or had moved results in an overall retention rate of 83%. Fortyone were located for interviews at 6 months. Of those who were not located at 6 months, two were deceased, two were confirmed to have moved out of state, and five were in jail. Excluding those who had died, gone to jail, or moved out of state resulted in an overall retention rate of 73.2%. The refusal rate was less than 5%.
We conducted key informant interviews with six service providers including staff at three area shelters, leaders of groups advocating for low-income housing or to end homelessness, and staff at a substance abuse treatment organization. Two focus group interviews with three and four participants each were conducted with staff from an additional shelter and staff at an organization administering several supportive housing programs. These were originally designed and intended to be key informant interviews. However, staff at each organization expressed interest in being interviewed together so that they could share and compare their perspectives and experiences. Key informants and focus group participants included staff in different positions within their organizations, including the executive director of one organization, supervisors and caseworkers with more direct, daily interactions with clients. Participants were 60% female, 60% white, 30% African American, and 10% Latino. The refusal rate among service providers was approximately 50%. Most refusals were due to time constraints or scheduling problems.

Procedure
Participant recruitment for the drug using sample was achieved through a combination of direct street recruitment and referral from other projects. For participants who were directly recruited, we targeted recruitment in locations where populations of drug users with differing housing characteristics could be found. Drug users who were homeless were recruited from each of Hartford's seven shelters or soup kitchens. Outreach staff approached potential participants in these settings, distributed HIV prevention materials such as bleach kits and condoms to initiate a general discussion about risk behaviors and assess their general eligibility for the study. Those participants who appeared interested and eligible were given an appointment card for full screening. Drug users who were doubled up with family or friends or housed in subsidized, non-subsidized or supportive housing were similarly recruited through street outreach, or from prior knowledge of their situation from ethnographic research in other research projects working with active drug users. We attempted to recruit equal numbers of drug users (approximately 10 or 11) from each of the housing statuses. In practice it was much easier to recruit participants in some housing categories than others (e.g. homeless in shelter and participants doubled up with family or friends were easier to identify and recruit than homeless on the street or drug users in supportive housing). Therefore, throughout the course of recruitment, when participants in any particular housing category became overrepresented, recruitment for that housing category was stopped and outreach and recruitment efforts focused on finding drug users in under-represented housing situations. Table  1 shows the housing status of participants at baseline, 3 and 6 months. Participants received a $25 incentive for completing each interview and a $15 bonus for completing all three interviews. Interviews were approximately 1 1/2 hours in duration. Written informed consent was obtained from all participants, both drug users and service providers, and the research protocol was approved by the Institutional Review Board at the Institute for Community Research.
All in-depth interview guides were project developed. Baseline interviews with drug users explored participants' housing histories over the previous two years, focusing on: reasons for moves, evictions or housing changes; types of public assistance, social services and housing subsidies applied for and accessed; the amount of time elapsed between application for housing and other social services and receipt or denial of housing or other services; and reasons given for denial of housing programs or apartment applications. To help participants construct their housing histories, we asked them to describe their current living situations and then moved back in time.
Three month follow-up interviews explored changes in housing status and access to housing programs. Baseline interviews were reviewed prior to follow-up interviews so that interview questions could be focused on participants' specific situations. If housing status changed since baseline, interviewers explored reasons for moves, eviction or housing changes, and any new applications to public assistance, social services or housing subsidies. The status or outcome of applications made or planned at baseline were explored. Six month follow-up interviews used the same interview guide as three-month interviews. Again, three-month interviews were reviewed so that questions followed up on any housing changes planned or made. Three-month interviews also included a brief quantitative survey to collect basic demographic information including age, income, length of time living in Hartford, educational level, and quantity and frequency of use of a variety of different drugs. This brief survey was added after it was determined that it was difficult to quantify such information from qualitative interviews. Six-month interviews also included brief demographic surveys that collected information on income, quantity and frequency of drug use in the last 30 days.
Service providers for key informant and focus group interviews were selected to represent a variety of organizations that may be directly or indirectly involved in assisting drug users to obtain housing. A list that included local homeless shelters, soup kitchens, drug treatment centers, mental health organizations, housing and homeless advocacy groups, and supportive housing programs, was compiled from staff knowledge, internet searches and networking with housing advocates and service providers. Potential staff members to target for interviews were also identified in an attempt to represent the ethnic and pro- fessional diversity within organizations. Project ethnographers then directly contacted staff at the organizations, explained the purpose of the study and invited staff to participate in a key informant interview, or contacted supervisors within the organization to explain the purpose of the study and ask permission to contact other staff members to participate in a key informant interview. Written informed consent was obtained from service providers. The length of interviews was 1/2 hour to 45 minutes.
Interviews with service providers focused on the facilitators and barriers that drug users face in accessing independent housing and in maintaining stable housing. We asked service providers about the characteristics of their clientele, including how clients are referred to their organization, to determine initial barriers or facilitators to accessing social or housing services for active drug users. We then asked them to describe the types of housing programs available, other services provided by the organization, and the eligibility requirements for housing and other services for their clients. We also asked them to describe the process through which they try to obtain housing or other services for their clients, the clients whom they have the most difficulty assisting in accessing housing, the strategies, if any, they use to overcome barriers in accessing housing for these difficult clients, and the clients who are the easiest to assist in accessing housing. Finally, we asked providers to describe reasons drug using clients have difficulty maintaining housing and the kinds of support services they feel are necessary to keep drug users in stable housing.

Analysis
All interviews were tape recorded and transcribed verbatim. All text data were coded and analyzed for key themes and patterns of response using Atlas.ti software [36]. Interviews were coded for type of interview (key informant, drug user baseline, three or six month). Interviews with drug users were further coded for demographics and housing status at the time of interview. Data were then coded a first time for content. The coding tree was developed in an iterative process by the research team and applied to in-depth interviews with drug users and keyinformant interviews. This first level of analysis coded for broad categories, e.g. social service application process, caseworkers, housing subsidies, shelter, or eviction. After this first level of coding, interviews were coded a second time to further refine categories and emerging themes. For example "creaming", "silting" "costs of applying for services," and "service provider discretion/priorities" were themes that emerged during this second level of coding. Excerpts presented in this paper were chosen to reflect these themes. All names of persons or organizations used in the paper are pseudonyms. Finally, in-depth interview with drug users were analyzed to capture changes over time. After all interviews that a participant had completed were coded, summaries were written for each participant that described his or her housing history, and welfare or other benefits received. Each participant's changes in housing status and the housing subsidies or other benefits applied for or received were then quantified by filling out a Housing Summary Checklist. These data were entered into SPSS and analyzed to show changes in housing status and stability, receipt of welfare or health benefits over time, and associations between housing status and applications to housing programs.

Sample characteristics
Demographics for participants were collected at threemonth interviews.

Shelters as point of access to housing programs
In comparison with homeless participants who stayed on the street or who doubled up with family members or friends, homeless participants who stayed in shelters reported receiving more information about different housing programs available, particularly supportive housing programs, and were more likely to have applied for these and housing subsidies such as Shelter Plus Care or Section 8. Eight out of the 21 participants who stayed in a shelter at some point during the study period applied for or received supportive housing as compared to 2 out of the 27 participants who were homeless on the street or doubled up with family or friends but had not stayed in a shelter during the study period (p = .013). All the area shelters employed caseworkers whose job it is to help shelter residents access more permanent housing and other services, such as mental health or drug treatment. The shelter that was offering this particular housing program closed during the summer, so outreach workers visited other area shelters to find residents who had stayed in that shelter in the winter months. Because space and funding for these programs is limited, however, shelter caseworkers often do not do much outreach, and shelter residents may lose opportunities to apply for or receive new housing programs simply because they are not at the shelter when applications are being accepted. A few participants (N = 7) reported learning about housing programs or subsidies from caseworkers at inpatient or outpatient substance abuse treatment programs, or methadone maintenance, and three reported learning about programs by word of mouth from friends or acquaintances. For the vast majority, however, shelter staff were the primary referral agents to accessing information about housing programs.

Caseworker Priorities: "Creaming" versus "Silting"
While all shelters had full time staff dedicated to helping shelter residents obtain permanent housing, staff from different shelters or even staff within the same shelter often had differing philosophies that affected how they processed clients. Some viewed their role as "referral agent," i.e., they referred their clients to organizations administering various supportive housing programs or housing subsidies for which they might be eligible as Mrs.
Roberts described in an in-depth interview. Those who viewed their role as mainly referral often described a process by which they referred clients whom they thought had the best chance at success as Mrs. Roberts described. Lipsky (1980) described this process as "creaming" and argued that this was one way that service providers cope with the demand for services being outmatched by resources. If resources are limited, then creaming is a rational strategy to ensure that resources are not wasted. Mrs. Roberts described the greatest challenge to her job as being the magnitude of the homeless problem and the limited resources available to confront the problem. Because staff at this shelter explicitly wished to serve as a positive example of the Housing First model, they engaged in what one staff member jokingly referred to as the opposite of creaming, "silting." If success can be demonstrated with even the most difficult cases, then that provides a stronger justification for increasing funding for such programs.

Mrs. Roberts
Carla: Our goal has been "Let's look at the people who have the worst histories that nobody else will ever house," and that's really the approach we take, and we have created policies around that. You know we don't rule people out. We also feel strongly, and this is again the Housing First model, that you don't fix people first. They don't need to be fixed. They don't need to be ready. They just need to be housed and then you work from there and you work with intensive support.
These differences in priorities and mandates result in differences in the ways that caseworkers assist shelter residents in obtaining housing. Whereas Mrs. Roberts described making sure that clients were "ready" for housing by referring them to substance abuse or mental health treatment programs before working with them on housing, staff at St. Mark's insisted that sobriety was not a precondition for housing. Mike who obtained housing through this program confirmed this. Silting, however, may result in another bias whereby less difficult clients, i.e. those without mental illness or chronic substance abuse, are not considered "prime candidates" for the program or actively recruited. One shelter resident in fact complained that "trouble makers" had received housing through this program while he, who always followed shelter rules, had not. Lipsky (1980) argues that another way of rationing services is to increase the costs of applying for them. However, he also argues that increasing the cost of applying for services will only marginally affect demand, since those seeking many of the services that street level bureaucrats administer, e.g. housing subsidies, or welfare, need the services and cannot access them anywhere else. In other words, those with other options would not suffer the costs of seeking services. He argues that because of this, those seeking services from street-level bureaucrats should be considered "involuntary clients."

The Cost of Applying for Services
Shelter residents could be considered involuntary clients because they have few options to obtain affordable, permanent housing other than through housing subsidies or supportive housing programs. However, many participants in this project described weighing the costs of applying to various programs with the benefits they expected or hoped to receive. Whether or not a participant decided to apply for supportive housing or housing subsidies depended in part on their felt need. Those who were doubled-up with family members or friends, in addition to receiving less information about services, also may not have been as inclined to seek out information or apply for programs because their felt need was not as great, as described by Don who usually stayed in his girlfriend's apartment but occasionally stayed in a shelter when he had conflicts with her. He felt like he could use Section 8 because he often had difficulty paying his rent but had never applied. Other costs include providing the paperwork that is required to apply for the programs or subsidies that can be difficult for persons who have been homeless for prolonged periods of time and may have lost many of their important documents. Other times, clients may not want to disclose some of their personal documents, such as arrest or medical records, which they find too personal to share and may question the necessity or relevance of such documents to their applications. Participants often expressed the feeling that these were capricious demands of caseworkers in order to delay or deny them access to housing programs. Jennifer described having to meet with a caseworker monthly, be involved in daily sessions for her drug treatment, and attend weekly job readiness trainings. This particular program seems to have followed the Continuum of Care model, in that participating in several supportive services was mandatory and that applicants were expected to begin their participation in supportive services before they actually received their housing subsidy.
If the need for services is great enough, however, participants described being willing to put up with the costs of applying and expended a great deal of energy to get their needs met. Long waiting lists also increase the cost of applying for housing subsidies such as Section 8. While waiting lists are determined by the amount of funding available, other practices, such as requiring that applicants make all requests in writing, increase costs even more.
Chris (Puerto Rican, 48): I applied for Section 8 and they said I'm in list number thousand and something but that's been for years and I call them. They say they don't take information on the phone. You have to do it through correspondence and stuff, so I didn't bother, but sometimes they do send you a letter once in a great while and they tell you what number you're at and stuff.
All but three of the participants who reported applying for Section 8 were homeless when they applied, either residing in shelters or doubled up with family members or friends. The likelihood that an applicant will still be living in the address listed on an application when correspondence is sent or subsidies become available is therefore very small, particularly since shelters place limits on the length of time a client can stay. When letters are sent to inform applicants that they have received a subsidy, they have a very limited time in which to accept the subsidy and find an apartment. If applicants do not respond because they never received the letter, they are placed again at the bottom of the waiting list, or their application is thrown out. Many participants reported applying for housing subsidies years before the baseline interview and having received no information regarding their applications since then. Many assumed that their applications were still in effect. Others found out that this was not the case only after they began working on their housing needs with caseworkers at other organizations. State public assistance, such as food stamps and Medicaid, also require recipients to respond to correspondence requesting updated information twice a year to determine continued eligibility. Again, participants without a permanent address reported that they often did not receive their correspondence and frequently had their food stamps and medical benefits cut off temporarily until they could fill out the paper work and get them reinstated. As this could take between one to several weeks, it worsened their already precarious economic situation and decreased their chances of obtaining or maintaining stable housing. Lipsky (1980) argues that while eligibility for public service benefits often may seem cut-and-dried, a considerable part of eligibility depends on the service providers' discretion. He further argues that the assignment of benefits or sanctions to clients is negotiated through "interpersonal strategies and implicit maneuvering." This negotiation, however, occurs in a context in which the service provider has much greater power over the definition of the situation and control over its outcome than the person seeking services.

Distributing Benefits and Sanctions: Favoritism versus Discretion
Nine of the participants who stayed in shelters complained of "favoritism" in terms of who gets a bed at the shelter, how long they are allowed to stay, and who gets help with services. Sometimes this favoritism seemed based on past personal relationships that staff had with residents. Shelter staff also reported sometimes extending shelter stays for certain residents, which they described as reflecting staff discretion and program flexibility to meet client needs. They reported that they based these decisions on how "compliant" clients were to their treatment plan, or how actively they were trying to work on their goals. Residents could also have their stay shortened if they failed to follow shelter rules. The decision about whether or not to extend a resident's stay at a shelter can have profound effects on whether he or she is able to access more permanent housing through subsidies or supportive housing programs. As mentioned above, most people access services through caseworkers. In addition, changing address, as residents are forced to do if their shelter stay is not extended, makes it more difficult to follow through on applications for housing subsidies.

Conclusion
The data presented in this paper illustrate that unofficial policy is as important in understanding drug users' vulnerability to homelessness and housing instability as official policy. In addition to the limits imposed by eligibility criteria and under-funding of housing subsidies, both housing caseworkers and drug users described a number of mechanisms that limit drug users' access to housing information and services. These include limited outreach regarding housing programs and subsidies to the homeless or marginally housed who avoid shelters. In addition, caseworkers prioritized clients in order to make decisions about how best to expend their limited resources and energy, "creaming" versus "silting." Another way that caseworkers rationed services is by increasing the costs of applying for them. Drug using participants described bureaucratic red tape and being treated disrespectfully by caseworkers. Finally, housing caseworkers are able to exercise considerable discretion when processing applications and serving their clients. This discretion was perceived as "favoritism" by the drug using participants interviewed in this project, and as program flexibility by caseworkers.
Housing caseworkers and advocates act as "street level bureaucrats" and have developed these mechanisms as rational ways of coping with the limited resources available to perform their jobs.
Unofficial policies that are used by caseworkers to ration scarce resources help explain the relationship between structural factors (the lack of affordable housing and under-funding of subsidized housing) and personal vulnerabilities (drug abuse, arrest and mental illness) that are alternatively hypothesized to cause homelessness. Housing caseworkers perform their jobs within the constraints of the larger socio-political context operating within the United States. The Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) and the budget flat lining or cutting of many federally funded housing programs has created an increased scarcity of resources to those whose job it is to provide permanent housing to the homeless and precariously housed urban poor. At the same time, rhetoric used to justify the imposition of severe time limits on lifetime welfare benefits focus on the pernicious effects of welfare on the individual, breeding dependency and sexual immorality, and the return to "personal responsibility" and independence that revoking benefits enforced [26,37]. Drug users are particularly vilified within this system, as they constitute the undeserving poor, who have only their selfish consumption to blame for their poverty [26] and specifically targeted in official housing policy in the "One Strike Law" that bans drug users' and their families from receiving federally subsidized housing.
The service providers interviewed in this project, and whom homeless drug users must petition in order to access housing subsidies and supportive housing programs, use these same discourses to understand the reasons poor people are seeking services, and the kind of help to which they are entitled. Some saw their jobs as reform-ing the individuals seeking their services by providing job training assistance, mental health or drug treatment services; at the same time they blamed some clients for their homelessness because of their drug addictions, or poor work ethics. However, the passing of PRWORA also created a drastic decrease in services as state welfare offices were expected to empty welfare rolls and return recipients to work as soon as possible. Many state welfare departments disqualified welfare recipients for relatively minor infractions [38]. Like has been seen in other states [37], Connecticut Department of Social Services (DSS) caseworkers were pressured to ration services to the greatest extent possible. In the current political climate, therefore, it is hardly surprising that only one of the participants in this project reported being referred to other needed services by their DSS caseworkers, and nearly all reported having food stamps and state medical benefits discontinued for not filling out the proper re-determination paper work.
Housing caseworkers and advocates who worked in the shelters also operated in this political context. Many described clients as compliant or non-compliant, and used their discretion in order to determine who received beds on any particular night, who would be allowed to stay for an extended period at the shelter, and to whom they would devote their energy to try to help exit homelessness. The compliant homeless included those who were actively working on their "problems" by entering drug treatment or searching for a job, while the non-compliant included those who refused to comply with shelter rules or program requirements. Homeless drug users described interactions with caseworkers in overwhelmingly negative terms. They saw caseworkers as showing favoritism to some and felt that their demands of clients were unreasonable and capricious. Similar to results in this study, research on the health and social service needs of HIV infected persons has found that these persons had overwhelmingly negative experiences with service providers and case managers [39][40][41].
Shelter staff are also constrained by budget cuts that limit the resources available to help their clients, but have even less control than low-level DSS or HUD employees in determining who gets access to these. Shelter caseworkers used different strategies to manage their jobs under these difficult circumstances. First, shelter staff expended little energy in outreach to homeless who slept on the street or doubled up with family and friends. This is a rational strategy considering that there were too few resources to assist even those shelter residents who actively sought their services in obtaining permanent housing. Another strategy used is similar to Lipsky's "creaming" as shelter caseworkers expended time and energy assisting clients whom they thought had the greatest likelihood of success.
Those deemed as having the best chance of success were those who were employed, and therefore less likely to have been incarcerated, while those who deemed unlikely to succeed included those who were actively using drugs, or with a serious mental illness. The final strategy, following the Housing First model, was "silting," in which shelter caseworkers attempted to house those deemed to be the most difficult cases, the long term homeless including chronic substance abusers, and persons with significant histories of mental illness and incarceration. By showing the success and cost-effectiveness of housing "pathological" individuals, they resisted dominant conservative discourses and advocated for expanding services to these individuals.
There are many ways of lessening the impact of the unofficial policies that serve to limit drug users' access to housing and other services. Increasing outreach to those in need of housing services who do not reside in shelters would improve access to information and housing programs to homeless persons who avoid shelters. Other valuable changes could decrease the costs of applying for services. For example, shelter staff and DSS caseworkers could receive on-going training regarding communication skills and a "customer service" approach to clients to address the lack of respect perceived by some participants. These trainings could explore and challenge caseworkers' implicit attitudes about homeless drug users and would be particularly important for organizations whose mission is to provide supportive housing to the chronically homeless. Alternative methods for the homeless to inquire about the status of their applications for housing subsidies or welfare benefits other than by mail could further decrease the costs of applying. Finally, caseworker discretion in distributing sanctions and benefits could be minimized by formalizing criteria by which decisions are made regarding extending a shelter resident's stay or who receives housing services. Improved communication regarding decision-making criteria might decrease shelter residents' perceptions of staff favoritism.
These mechanisms, however, are unlikely to be effective without devotion of significant resources and political will to solving the housing crisis. Lipsky (1988) argues that street level bureaucrats effectively become policy makers as they implement policy. Official policy, however, imposes constraints on caseworkers' ability to perform their job by defining the amount of resources available. Housing caseworkers have little incentive or power to eliminate barriers to drug users' access to housing, information and services in the current political economy. That an alternative model exists in the Housing First Model is a hopeful sign and a challenge to dominant discourse about impoverished drug users. Such challenges need to be continued and expanded in order to find solutions for chronically homeless substance abusers.
This paper is one of the first to explore how unofficial policy limits drug users' access to housing, information and service. Qualitative research is particularly well suited to explore processes such as the mechanisms housing caseworkers and advocates use to ration services. This study is additionally strengthened by its use of in-depth interviews with a larger sample of drug users than those typically included in qualitative research, the inclusion of drug users in various housing situations, the high proportion of women, and its longitudinal design. Limitations to the study include the small number of housing service providers interviewed and the lack of inclusion of non-drug using low-income, homeless or marginally housed participants. Additional qualitative and quantitative research is needed to explore ways that unofficial policy limits drug users' and non-drug users' access to housing defined as the sources of information about, perceived eligibility, application for and receipt and denial of housing programs and subsidies, welfare and social services.