Most Rental Assistance Recipients Work, Are Elderly, or Have Disabilities
July 17, 2013
As administrators of federal rental assistance programs face strained resources and mounting needs for affordable housing, it’s reasonable to consider whether they should develop and implement policies to encourage and assist families they serve to gain skills and work experience that could help them find jobs and increase their earnings. Increased employment and earnings might eventually enable some families to afford market-rate rental units and could also reduce the costs of the rental assistance programs by reducing the size of the rental subsidies some tenants need.
Our analysis finds that 88 percent of households that received rental assistance in 2010 were elderly, disabled, working (or had recently worked) or likely have access to work programs under the Temporary Assistance for Needy Families (TANF) program.
Among households receiving housing assistance through the Department of Housing and Urban Development’s (HUD) major rental assistance programs in 2010 that were not elderly or disabled, nearly three-quarters worked that year or recently or participated in a program through which they likely were subject to a work requirement. The percentage was higher than three-quarters among households participating in the housing choice voucher and public housing programs, and less than three-quarters in the Section 8 project-based rental assistance program. These percentages almost certainly were lower in 2010 than in a typical year; in 2010, the unemployment rate stood at 9.6 percent, its highest level in 27 years.
Nearly half of the non-elderly, non-disabled households in each of the three major rental assistance programs that were not attached to the labor force or subject to a work requirement in another program included either a pre-school age child or an individual (other than the head or spouse) with a disability. Responsibilities to care for these individuals can make it difficult for single heads of households to work without child care subsidies or other assistance.
Federal policy should encourage efforts to increase earnings among HUD-assisted tenants, which could reduce rental subsidy costs and help families afford private market rents. This is particularly important in high housing-cost areas where non-elderly, non-disabled households are likely to receive rental assistance for a longer period.
Policies are most likely to succeed at significantly increasing tenants’ incomes or overcoming substantial barriers to work if the policies meet three criteria: (1) the strategies are evidence-based; (2) agencies with work-focused expertise — rather than housing agencies — lead the local initiatives; and (3) partnerships are forged among human services and housing providers that operate across the local labor market, rather than requiring individual housing agencies or owners to each have to try to establish such partnerships on their own.
Such policies should be part of larger strategies to help low-income individuals improve their employment prospects, rather than focused solely on housing assistance recipients or funded by diverting funds away from housing assistance and thereby leaving more families with severe housing-cost burdens (and likely causing an increase in homelessness as a result). There is no evidence that allowing housing agencies to shift rental assistance funds to support work-related services increases earnings enough to be cost-effective. (Moreover, further deregulation of housing agencies would not address the larger lack of employment among households in the privately administered project-based rental assistance program.)
This paper uses 2010 administrative data submitted to HUD by state and local housing agencies and private owners that administer rental assistance under HUD’s three general purpose programs: the Housing Choice Voucher (HCV), public housing, and project-based rental assistance (PBRA) programs. Together, these programs provide rental assistance to 96 percent of HUD-assisted households. We use these data to analyze the extent to which households receiving rental assistance are attached to the labor market or subject to work requirements in other programs, and consider strategies to boost tenants’ employment and earnings.
Large Majority of HUD-Assisted Households Are Elderly, Disabled, Attached to the Labor Market or Subject to a Work Requirement
Federal rental assistance programs enable about 5 million low-income households to afford modest homes.
In 2010, some 88 percent of the households that received rental assistance through HUD’s three general purpose programs were elderly, disabled, working or had worked recently, or were likely subject to a work requirement through TANF participation. Only 251,000 HCV, 116,000 public housing, and 155,000 PBRA households were not elderly, disabled, working (or worked recently), or likely subject to a work requirement. (See Figure 1.)
The total share of households in these categories is broadly similar across the major rental assistance programs, although the relative share of households in some of the categories differs. (See Figure 2.) Most significantly, PBRA households are more likely to be elderly or disabled (66 percent of PBRA households fall into this category, compared with 49 percent of HCV and 54 percent of public housing households), and PBRA households not headed by an elderly or disabled person are less likely to have earnings than public housing or HCV households. (See “Non-Elderly, Non-Disabled Households in Private Owner-Administered Program Less Likely to be Attached to the Labor Market” section below for more on this latter finding.)
Nearly Three-Fourths of Non-Elderly, Non-Disabled HUD-Assisted Households Work or Are Likely to be Subject to Work Requirements
Concerns about work effort among recipients of housing assistance typically focus on families that are not elderly or disabled. Such families made up 45 percent of the households that received housing assistance through HUD’s major programs in 2010. The substantial majority (73 percent) of these households worked in 2010 or recently, or participated in a program through which they were likely subject to a work requirement. (See Figure 3.)
In the HCV and public housing programs, which are administered by public housing agencies (“PHA assisted”), three-fourths or more of non-elderly, non-disabled households were working, worked recently or were subject to a work requirement, compared with 62 percent of households in the PBRA program, with which a public agency generally has no direct role (“private owner assisted”). The non-elderly, non-disabled households can be divided into three groups:
- Households attached to the labor market: In 2010, nearly two-thirds of the non-elderly, non-disabled PHA-assisted households were working or had worked recently. Some 51 percent of private owner-assisted households fell into this category. Other data indicate that the weak economy in 2010 dampened these work rates. HUD reported a five percentage point higher rate of work among public housing “prime working group” households in 2007 than in 2010. Some 74 percent of working-age, non-disabled adults that received some type of rental assistance from federal, state or local governments in 2006 reported earnings in that year or 2007 (or both), before the recession caused unemployment to spike.
- Households likely subject to work requirements: TANF participation among non-elderly, non-disabled households without earnings in the three programs was identical: nearly 11 percent received assistance from state TANF programs that impose work requirements on the vast majority of adult recipients. A small share of households — fewer than 2 percent of non-elderly, non-disabled households in each program — reported receiving General Assistance (GA) benefits in 2010. Non-disabled individuals receiving GA benefits are generally subject to work requirements under state or local rules, but we do not have sufficient information to quantify this and therefore do not classify GA recipients as subject to work requirements, although many of them are.
Households neither attached to the labor market nor subject to TANF work requirements: Nearly half of the remaining non-elderly, non-disabled households in each of the three programs included a pre-school age child or a member (other than the head or spouse) with a disability. Responsibilities to care for these individuals may make it difficult for single heads of households to work without child care subsidies or other assistance.
Little is known about these households’ ability to work, and unfortunately the available data do not shed much light on that. Research on assisted families has shown that many households may have complex problems such as poor mental and physical health, lower educational attainment levels, substance abuse, and domestic violence, all of which can act as barriers to employment. The large majority of these households had some income from sources such as child support, pensions, or foster care payments. Fewer than 1 percent of voucher and public housing households report having no income at all (compared with 10 percent of non-elderly, non-disabled PBRA households).
|Table 1 |
Key Characteristics of Non-Elderly, Non-Disabled Households in 2010, by Program
|Voucher||Public housing||Project-Based |
|Attached to the Labor Market||65.6%||64.8%||51.6%|
|Worked in 2010||54.9%||56.3%||44.4%|
|Received UI in 2010||7.5%||6.0%||5.1%|
|Worked in 2009||3.2%||2.5%||2.1%|
|Likely Subject to TANF Work Requirements||10.6%||10.6%||10.3%|
|Caring for Child Under 6 or Disabled Person||11.5%||11.5%||18.6%|
|Source: CBPP analysis of HUD administrative data|
Non-Elderly, Non-Disabled PHA-Assisted Households Have Similar Labor Market Characteristics
Non-elderly, non-disabled households in the two PHA-assisted programs — Housing Choice Vouchers and public housing — have similar labor market characteristics. In 2010, 55 percent of HCV households and 56 percent of public housing households had earned income. (See Table 1 above.) The median earnings for these households in the public housing program ($15,503) were slightly higher than those for households in the voucher program ($14,947) and only slightly more than the pay from full-time, year-round minimum wage work in 2010. Some households had other income in addition to earnings, such as child support, bringing the median total income for non-elderly, non-disabled households to slightly more than $17,000 in both programs.
Non-Elderly, Non-Disabled Households in Private Owner-Administered Program Less Likely to Be Attached to the Labor Market
In 2010, 44 percent of non-elderly non-disabled households in the PBRA program were working, and another 7 percent had worked recently. Greater shares of working-age, non-disabled households in the PHA-operated programs were employed in 2010 (55 percent of HCV households and 56 percent in the public housing program). These households also were more likely than those in the PBRA program to have worked recently (10.7 percent of HCV households and 8.5 percent in the public housing program).
The reasons for the variation between the publicly and privately managed programs are unclear. Geography may be one contributing factor. A larger share of private owner-assisted households than other assisted households live in Ohio, Illinois, and Michigan, which had higher unemployment rates than the national average in 2010. This means there was generally less work in these states and potentially fewer jobs available for lower-skilled individuals, although the state distribution of assisted units in each program does not fully explain the different level of work participation among private owner assisted-households compared with PHA-assisted households.
The available data on demographic characteristics show modest differences among the working age, non-disabled households in each of the three major programs, although none seem likely to significantly affect work rates or income. The heads of these PBRA households are slightly younger than the comparable HCV and public housing households (median age 30 to 34, compared with a median age of 35 to 39 for the PHA-assisted households). As noted above, nearly half of the households that are not attached to the labor force or receiving TANF benefits (and therefore likely subject to a work requirement) are caring for pre-school age children or a disabled household member in each program, so it’s unlikely the younger age of PBRA heads of household correlates with greater child care responsibilities that could interfere with work. The administrative data HUD collects do not include information on educational attainment or prior work history, two factors that may affect households’ work rates and income.
Policy differences among the programs also are unlikely to explain the significant differences in work and incomes between the public and private programs. HUD rules allow private owners and PHAs the same flexibility to give working households preference for admission, and income eligibility rules differ only slightly among the programs. Similarly, households in all three programs generally pay 30 percent of income toward rent. Some researchers have hypothesized that project-based subsidies may isolate families from available jobs, but if that were the case one would expect to see a higher rate of employment among HCV households, who have mobile subsidies, and similar rates among public housing and PBRA households. Our analysis shows, however, that work rates in 2010 were nearly identical among HCV and public housing non-elderly, non-disabled households, and that public housing tenants were more likely to be employed than working-age, non-disabled PBRA tenants.
Some may offer the Moving to Work (MTW) demonstration — which allows public housing agencies serving 12 percent of PHA-assisted households to vary rent rules, impose work requirements and time limits, and shift funds from housing assistance to services to support work — as another explanation for higher levels of work and income among PHA-assisted households. But we found this was not the case. In fact, employment rates were slightly higher for households at non-MTW agencies in both the HCV and public housing programs than at the MTW agencies reporting data to HUD in 2010.
Of course, the combination of various small effects could add up to the significant difference in work rates among PHA- and private owner-assisted households we found in our analysis. But another explanation seems to us to also be an important factor: private owners do not do as good a job as public housing agencies at connecting tenants to work opportunities or work supports. There are two primary reasons to think this may be the case. First, as private — and predominantly for-profit — entities, owners of PBRA properties may be less likely to be aware of or have relationships with organizations that provide job training or work supports, such as child care subsidies. Unlike private assisted owners, public housing agencies often participate in local Workforce Investment Boards, which decide how to spend and target federal job-training funds, and many PHAs are affiliated with, or even part of, city, county or state government, increasing the chances that staff become familiar with how their tenants can access necessary work supports such as child care.
Second, owners of PBRA properties have no incentive under the program’s funding policy to encourage families to increase their incomes, as HUD pays the full amount of the gap between tenant contributions and the approved unit rents. In contrast, public housing agencies’ incentives are two-fold: they may be able to provide housing vouchers to more families (and earn more in administrative fees) if tenants’ incomes increase, and higher rental payments from public housing families help agencies meet their operating costs when HUD’s subsidy payments fall short, which commonly occurs. These incentives may prompt some agencies simply to target more families who are already working when they first receive assistance (which would also contribute to higher work rates among PHA-assisted households). But some agencies may also respond by encouraging and helping current recipients to increase their earnings.
HUD has one initiative to promote work among PBRA households and tenants of properties with federally insured mortgages. Known as Neighborhood Networks, the program began in 1995, when few low-income families owned personal computers, to encourage private owners to establish computer centers in their properties. Some Neighborhood Networks centers partner with local agencies or raise private funds to offer assistance to residents to search for jobs or to enhance their job readiness. (HUD provides no additional funds for the program, though owners are allowed to tap some reserve funds for this purpose.) As of 2008 — the latest information on HUD’s website — HUD reported that there were approximately 1,500 Neighborhood Networks centers. But it is unclear how many families use the centers or their programs or what effect they have on employment rates and earnings.
To the best of our knowledge, no other federally funded job-training or work support programs are targeted at private owner-assisted households. PBRA tenants are not eligible to participate in HUD’s Family Self Sufficiency (FSS) Program, which funds service coordinators to help HCV and public housing families develop five-year self-sufficiency plans and access services that can help them increase their earnings and become independent of welfare assistance. Participants accumulate savings based on the increased rent they pay as their earnings rise, which agencies set aside for them in special escrow accounts. Completing the program enables families to receive their escrowed savings. Studies have shown substantial increases in earnings for families that complete the FSS program. In March 2013, Senators Jack Reed (D-RI) and Roy Blunt (R-MO) introduced the Family Self Sufficiency Act, which would expand eligibility for the FSS program to include tenants assisted by the PBRA program, as well as to streamline the program and stabilize funding for FSS coordinators.
Despite the lack of federal support for efforts to promote work among their residents, some private assisted owners provide such services because they see it as part of their mission or as good for the stability of their property. They partner with local agencies, raise funds from foundations, or use their own funds for this purpose. But it is likely that such efforts are not as widespread as in the public housing and voucher programs, perhaps contributing to the larger share of non-elderly, non-disabled households in the PBRA program who are not employed.
Most Non-Elderly, Non-Disabled Households Receive Housing Assistance for Five Years or Less, With Longer Stays in Highest Rent Areas
Despite modest incomes, the majority of non-elderly, non-disabled households who lived in assisted housing at the end of 2010 had been receiving rental assistance for no more than five years. The median length of stay ranged from three years (36 months) for households in the PBRA program to 39 months for public housing households and 48 months for HCV participants. In contrast, elderly and disabled households stay in the housing assistance programs substantially longer. Of the non-elderly, non-disabled households that had received rental assistance for five years or more, a majority relied primarily on earnings.
High Rents Drive Duration of Receipt of Rental Assistance
Households that had received assistance for five years or more in 2010 were more likely to work than those that had received assistance for shorter periods. They were more likely to live in high-cost areas where they would struggle to maintain stable housing without assistance. For each of the major rental assistance programs, we divided the non-elderly, non-disabled households into housing-cost quartiles based on the two-bedroom HUD Fair Market Rent (FMR) for the area in which they lived. (The FMR reflects the cost of renting modest housing in a metro area or non-metropolitan county.) In each program, a much larger share of non-elderly, non-disabled households living in areas with the highest rental costs had received rental assistance for more than five years than of households living in areas with lower rents.
Among households that had received rental assistance for more than five years and lived in the lowest quartile of area rents, a smaller share reported receiving any earnings in 2010 than of such households in the highest rent quartile, as shown in Figure 5. It is likely that in lower rent areas, households able to earn modest incomes move off of rental assistance in less time because they can find other housing they can afford. As a result, relatively few working households have longer stays in lower rent areas. In higher rent areas, by contrast, families earning modest incomes would struggle more to afford housing without assistance. For example, the median annual income for working families in the highest rent quartile that had participated in the HCV program for more than five years was $21,136 in 2010. To remain in their homes without a voucher, the typical family in this group would have to pay 81 percent of its income for rent and utilities. Such an extremely high rent burden is unlikely to be sustainable. For these households, continued receipt of rental assistance supports their continued employment as well as providing a stable home.
Policymakers Should Consider Proven Strategies to Increase Earnings of Assisted Housing Tenants
While the large majority of the working age, non-disabled adults in assisted housing programs work, their earnings often are, as just noted, too low to afford housing in the private market, particularly in higher cost areas. Helping assisted tenants work and increase their earnings could both improve their lives and reduce subsidy costs or free up units for new households in need of housing assistance. For some individuals, services to increase their skills could lead to higher pay, and services to provide more reliable child care or transportation assistance or improve health status could make it possible for them to work more regularly throughout the year. We still have much to learn about how to increase earnings among disadvantaged populations, but a number of evidence-based approaches could be targeted or expanded to serve more assisted housing residents.
A minority of households that have no recent work history are likely to need more intensive services to overcome barriers to work or help them connect to job opportunities, particularly if they are not receiving TANF assistance or otherwise lack access to such services. Enabling these households to work steadily or to afford stable housing without assistance is particularly challenging. Few initiatives have proven effective even for a portion of this target population (i.e., the households that have the most serious obstacles to employment), and those interventions that are effective are generally costly. Housing assistance programs provide a vital defense against homelessness and destitution for these families and individuals. Strategies to increase work among these households must avoid depriving them of the safety net that housing assistance provides.
It is possible that changes in HUD’s income-based rent rules or other housing policy changes, separately or in combination with these types of services, would increase earnings. For example, today a family with a voucher whose earnings rise to the point where it is no longer eligible for a rent subsidy loses the voucher after six months. If a worker in the family is then laid off and the family can no longer afford the rent, the family must go to the bottom of the waiting list and may not receive assistance again for many years, potentially exposing it to serious hardship and even homelessness. This risk could deter some families from increasing their earnings in the first place. Allowing “zero subsidy” families to hold onto their voucher for longer than six months would have no cost, but could encourage more families to take the risk of shifting to higher-paying work. In addition, as noted above, findings on the success of HUD’s Family Self-Sufficiency program are promising, and rigorous research is now underway that will help policymakers determine if the program should be substantially expanded. HUD also is launching a demonstration to determine whether changes to the rent rules would increase work and earnings.
While assisted housing recipients potentially could benefit from increased additional investments in programs aimed at increasing their employment and earnings, it is important that these programs not be funded with the limited resources available to provide low-income housing assistance. Funding for affordable housing is already highly inadequate to meet the need. Only one in four families eligible for federal rental assistance receives it. More families than ever before face severe housing affordability problems: from 2007-2011, the number of families with “worst case” housing needs increased by 43 percent, and homelessness among families with children is increasing. Diverting scarce housing resources to employment-related services would make sense only if that use of funds did not increase the number of households lacking an affordable place to live, and we do not have evidence at this point that indicates such an outcome can be achieved.
Federal Policy Should Encourage Use of Evidence-Based Practices
While research shows that many interventions aimed at increasing earnings have little effect, we have rigorous evidence of several strategies that, when implemented well, can produce significant increases in earnings among low-skilled individuals. In various places across the country, three types of strategies have been found to be effective:
(1) Certain “mixed approaches” that allow individuals to participate in a mix of employment and training activities including short-term training, education, and job search;
(2) Sectoral employment; and
(3) Life skills education.
Mixed approaches. This strategy uses a mix of approaches to match individuals’ job readiness and interests instead of requiring all individuals, regardless of their level of job readiness, to participate exclusively in a job search program. It provides job search assistance for individuals who are job ready and can pursue immediate employment, but it also encourages and supports participation in education and training programs designed to increase participants’ employability. Individuals with significant employment barriers such as mental health or physical health issues receive services to address these issues before participating in, or in conjunction with, other employment services. For example, Oregon’s early welfare-to-work program that used this approach increased earnings by 25 percent over a five-year period.
A housing-based strategy to increase earnings — the “Jobs-Plus” demonstration — used an individualized approach to help recipients increase their skills and find jobs, provided financial incentives to work through modifications of HUD’s standard rent policies, and helped participants access work supports like child care and transportation. A rigorous study of Jobs-Plus found that it increased average earnings among non-disabled, working age public housing tenants at the agencies that implemented the multi-faceted program well by $1,300 per year (16 percent) for the seven years of the follow-up study. The study did not include a detailed cost-benefit analysis, but MDRC estimates the cost of the package of interventions at roughly $1,800 per person per year for the four years of the demonstration. At this cost, the increase in incomes — and rent payments — would have to continue for a number of years for the demonstration to pay for itself.
Sectoral employment. This strategy focuses on training programs specifically tailored to the types of industries in the local area. These programs prepare unemployed and underskilled workers for skilled positions and often are developed in close cooperation with local employers. The length of training depends on the sector. A rigorous study of different sectoral employment programs in three cities — Boston, Milwaukee and New York — found that participants’ earnings were 18 percent higher over two years than those not selected to participate, with most of the earnings increases in the second year after the completion of training.
Life skills education. Individuals that face significant personal and family challenges (such as mental health issues) often need more support to find employment than most programs targeted to low-income individuals provide. A particularly successful, though costly, example of this “life skills” approach is the Building Nebraska Families program. TANF recipients in several Nebraska counties were randomly selected to receive home visiting and life skills education services, in addition to the usual TANF employment services (primarily job search assistance). In the second year of follow-up, participants with the most significant barriers earned an average of $135 more per month — an increase of 41 percent — than those not selected to participate. Participants were significantly more likely to be employed for 12 consecutive months and to move from a lower to a higher wage job.
It is important to note, however, that even though participants’ employment outcomes were better than those not selected to participate in the program, many of the program participants never found steady employment — only 46 percent were employed for 12 consecutive months during the 30-month follow-up period. In addition, the program was expensive to operate — an average of $7,400 per participant and more for the participants who were very hard-to-employ.
Lessons for Housing Policy
Increasing earnings among assisted housing tenants should be a focus of federal policy, but the limited evidence of what works — and the substantial cost of modestly successful interventions — suggest that policymakers should proceed cautiously with regard to housing policy in this area. While important, employment services for assisted housing residents should not be funded out of dollars appropriated for housing assistance unless we are certain that the provision of such services will result in a net increase (or, at least, in no decrease) in the number of low-income families that are able to afford housing.
Policies that seek to increase employment and earnings through the efforts of individual public housing agencies and private owners of assisted properties are unlikely to succeed, and risk undermining the primary mission of the rental assistance programs to provide affordable homes. Instead, federal policy should encourage partnerships among welfare, workforce, and housing agencies, and facilitate the formation of such work-oriented partnerships across all of the housing agencies and private owners of assisted housing located in the same labor market area.
Agencies With Expertise Should Lead Efforts to Promote Work
Substantial expertise and resources are required to develop and operate successful initiatives to significantly increase the earnings of recipients of housing assistance. Few entities in the housing delivery system have the requisite expertise to lead such efforts.
Federal policy should encourage partnerships among welfare, workforce, and housing agencies, rather than vesting responsibility in housing agencies to operate work programs directly. (The Family Self-Sufficiency program recognizes this limitation and funds only service coordination, along with a financial incentive, not the direct provision of services.) As HUD stated in a recent report to Congress: “The goal of achieving higher levels of work participation for recipients of HUD’s rental assistance can be best achieved through a multi-systems approach that addresses the multi-faceted barriers to employment that low-income families in our nation face.”
To encourage collaboration among housing providers and workforce, welfare-to-work, and health care systems, federal grant programs in these non-housing domains could reward (or require) agencies to prioritize the delivery of a portion of their services to households that receive federal housing assistance. Such targeting is supported by evidence showing the greater positive impacts that some work programs have on extremely poor families that also receive housing assistance.
HUD Should Encourage Labor-Market-Wide Partnerships
HUD should do more to encourage housing providers to develop partnerships with human services and health care agencies at the local and regional level as a companion strategy to federal policies to encourage prioritization of federally assisted tenants for these services. HUD should facilitate coordination of work-oriented partnerships across all of the housing agencies and private owners of assisted housing located in the same labor market area. Such coordination is essential to overcome the two barriers that most housing providers face: the limited number of tenants appropriate for any particular work-related initiative, and the lack of funding for the necessary staff time to develop such partnerships. Small public housing agencies have limited budgets for administrative costs, and HUD gives private owners of assisted housing no extra funding to administer their programs. Without funding, many will be unwilling to take responsibility for helping their tenants increase their earnings.
Several Congressional proposals would promote labor-market-wide partnerships among housing, welfare, and workforce agencies. Draft legislation released by the leadership of the House Financial Services Committee in April 2012 included two proposals that would require HUD to encourage such partnerships. That legislation, the Affordable Housing and Self-Sufficiency Improvement Act of 2012 (AHSSIA), would have required HUD both to report annually to Congress on public housing agencies that have entered into agreements with welfare, workforce, and other agencies to promote their tenants’ economic security, and to give extra credit in assessing agency performance to those PHAs that increase the scope and size of their Family Self-Sufficiency programs.
Another proposal would broaden access to public housing agency-administered Family Self Sufficiency (FSS) programs, which typically access services for their participants throughout a city, metropolitan area or county. In addition, the Reed-Blunt bill mentioned above that would make PBRA tenants eligible for FSS would enable private owners to tap into the partnerships that FSS has already developed. If it isn’t feasible for PBRA tenants to join an existing FSS program, the bill would enable owners to access federal funds held in a reserve account to provide staff to coordinate services for families in the program.
Expansion of the Moving to Work Demonstration Not a Panacea
Some policymakers believe that housing assistance programs — without any additional funding — can help tenants increase their earnings and obtain housing they can afford through an expansion of the Moving to Work (MTW) demonstration. The Moving to Work demonstration allows HUD to grant broad waivers of federal statutes and regulations and establish alternative funding policies for 39 of the nearly 4,000 agencies that administer the public housing and Section 8 voucher programs.
However, the program’s track record lacks evidence of such success. Without such evidence, policymakers should proceed cautiously with an expansion.
Despite its name, MTW is a broad deregulation initiative — not an employment-focused program — and only some of the agencies in MTW have focused on promoting household self-sufficiency. HUD has permitted MTW agencies to implement strategies to promote work and increase earnings without regard to whether the strategies are evidence-based or operated in a manner that enables their effectiveness to be assessed. To date, MTW has not identified policy changes or other initiatives that have been found to result in increased earnings for assisted families or in more rapid and successful transitions to the private market. Recent reports from the Government Accountability Office (GAO) and HUD have found that MTW’s effects “could not be conclusively identified.” A HUD report and other assessments of MTW have highlighted some best practices, but as the GAO pointed out, “In most cases, the practices chosen were based on the opinions of HUD or contracted staff and largely involved anecdotal (or qualitative) data rather than quantitative data.” In addition, as noted above, our analysis found that employment rates of non-elderly, non-disabled households assisted by MTW agencies in 2010 were somewhat lower than at non-MTW agencies.
Some MTW agencies are using their flexibility to shift funds from rental assistance to providing work-promoting services or financial incentives to work. Yet any shift of resources that does not pay for itself will exacerbate the already severe shortage of rental subsidies, and there is no evidence that agencies have chosen cost-effective strategies. A small number of MTW agencies have also imposed work requirements, time limits on receipt of rental assistance, or both; there is no rigorous evidence that these policies have increased work or earnings either, no assessment of the policies’ administrative costs, and no evidence regarding the potentially harmful consequences of these policies for families that lose assistance as a result. (The demonstration has not required agencies to track families that leave the housing programs.)
For example, the Chicago Housing Authority invested considerable funding in the Chicago Family Case Management Demonstration, which provided home visits and intensive supportive services (including mental health services), together with Transitional Jobs and other employment services to several hundred of the most vulnerable families displaced by the redevelopment of two large public housing developments in Chicago. The Urban Institute evaluated the demonstration, but could not assess impacts definitively due to the lack of a control group. After two years, self-reported employment had increased 10 percentage points, with those who participated in the Transitional Jobs program or had a high school degree (or GED) more likely to be employed. But there was no increase in average incomes, wage rates did not improve, and the reported level of public assistance receipt remained unchanged. By 2011, the year after the demonstration ended, a significant share of participants reported improved physical and mental health, but the overall rate of reported employment had declined slightly to a level that was not a significant change from 2007. The average cost per household was $8,700 over three years, with apparently no increase in rent revenues for the housing authority as a result.
If MTW were expanded in its current form, it would be no more likely than it has been to date to test work-related initiatives effectively and would raise other major risks, as we explained in a recent analysis of an MTW expansion proposal. Some expansion of the number of housing agencies allowed some flexibility from federal rules based on their high performance can be justified if appropriate protections are in place and it is necessary to pass other important rental assistance reforms. Such an expansion could facilitate learning on the effectiveness of various policies in increasing earnings or enabling families to transition successfully out of housing assistance programs, if Congress requires rigorous evaluations and funds them. But expanding MTW is not essential to test these approaches (since this could be done through a more narrowly targeted demonstration) and would not substitute for the proposals described above to improve links to the workforce development and other systems that have the resources and expertise to effectively deliver work promoting strategies for housing assistance recipients, including those that are not assisted by PHAs.
Increasing earnings among housing assistance recipients will benefit families and potentially enable the rental assistance programs to serve more households. Our analysis shows that in 2010, most assisted households that could be expected to work were employed, even at a time of unusually high unemployment, although their earnings often remained too low to afford housing without subsidies. Helping predominantly low-skilled individuals achieve higher earnings is challenging; many interventions have had little effect, and those that have been successful are often costly. Federal policy should encourage strategies to increase earnings of housing assistance recipients that are evidence-based and led by entities with appropriate expertise, along with work-oriented partnerships that involve housing providers throughout a labor market area. Housing agencies rarely have the requisite expertise or assist sufficient families to operate workforce programs efficiently, and housing funds are already too scarce to be diverted to such efforts without the certainty that the provision of such services will result in a net increase (or, at least, in no decrease) in the number of low-income families that are able to afford housing.
This analysis uses U.S. Department of Housing and Urban Development (HUD) administrative data on households that participated in the Section 8 Housing Choice Voucher (HCV), Public Housing, and Section 8 Project-Based Rental Assistance (PBRA) programs during calendar years 2009 and 2010. The data are an extract from the Family Report Form HUD-50058 (for HCV and public housing households) and HUD-50059 for PBRA households.
Our total sample for 2010 consisted of 4,285,252 active household records with income and demographic data. We excluded 54,483 households with blank or conflicting demographic information and non-elderly non-disabled households with blank income fields. The number of households in HCV (2,072,975) corresponds closely to the 2,084,649 households that received HCV assistance, on average, in 2010, based on CBPP analysis of data from HUD’s Voucher Management System (“Voucher Utilization Data, 2004-2010,” Center on Budget and Policy Priorities, National Fact Sheet, October 19, 2011, http://www.cbpp.org/files/9-21-11hous-US.pdf ). Due to lack of reporting, our HCV sample excludes six housing agencies participating in the Moving to Work (MTW) demonstration (Delaware State Housing Authority, Orlando Housing Authority, Champaign (IL) Housing Authority, Lincoln (NE) Housing Authority, Portage County (OH) Metropolitan Housing Authority, and Seattle Housing Authority).
The number of household records in the Project Based Rental Assistance program (1,191,281) is 96.5 percent of the 1,234,822 PBRA units in HUD's Multifamily Assistance and Section 8 Contracts Database, as of December 2010. The 3.5 percent discrepancy could be entirely due to vacancies. (The PBRA records do not include the approximately 34,000 units assisted under the Section 8 Moderate Rehabilitation program administered by HUD’s Office of Public and Indian Housing.)
The number of household records for Public Housing (1,020,996), however, is significantly short of the total units funded. We do not have access to a reliable count of occupied public housing units in 2010. In 2011, HUD provided funding for 1,117,954 public housing units. See http://www.cbpp.org/files/4-13-11hous-US.pdf. We do not know whether the difference of approximately 10 percent between the likely number of funded units and reported households is due to vacancies, under-reporting, or both.
Major Categories and Data Adjustments
This section explains how we categorized the data and what adjustments we made.
Households Working or Recently Attached to the Labor Market
For the purpose of this analysis, non-elderly, non-disabled households that received income in 2010 from wages or self-employment (1,025,418) or from Unemployment Insurance (UI) (127,989) are defined as households attached to the labor force. In addition, we considered as recently attached to the labor force the 54,591 non-elderly, non-disabled households that reported earnings as their primary or sole source of income in 2009 but did not have income from employment or UI in 2010. About 40 percent of these families received TANF benefits in 2010, but for purposes of this analysis they are counted in the “attached to the labor force” category. We were able to match 838,245 of the 987,320 non-elderly non-disabled households that relied primarily on earnings in 2009 with the 2010 dataset. In 2010, 118,392 of these households exited the rental assistance programs, and 20,008 became elderly or disabled households. Most of the remaining households in the matched sample were still working or received UI benefits in 2010.
Households Subject to a Work Requirement
We considered households likely to be subject to work requirements if they are not included in the “attached to the labor force” category discussed above and received TANF benefits in 2010 that were counted in determining their rent obligation (202,561), or if they were subject to the special “imputed welfare income” rent rules due to a work-related sanction (1,629). If the TANF benefits were not counted in the determination of the household’s required rent contribution, we assumed that the payment was a type of benefit that would not carry work requirements. We did not include 12,920 non-elderly, non-disabled families reported as receiving TANF benefits (and not working or receiving UI benefits) in the number of households subject to work requirements because of this assumption.
We retained the TANF categorization for 12,336 non-elderly, non-disabled households without minor children, as it is possible that an adult would continue to receive TANF benefits for a child in foster care or otherwise temporarily absent from the home. If the source of income for these households is not actually TANF, however, and was improperly categorized by local staff, they may not be subject to a work requirement.
Income Types and Amounts
Percentages of non-elderly, non-disabled households receiving certain types of income are based on the records of 1,934,514 households for which data on income sources were provided. The data provide a maximum of eight income sources per household, so may omit some amounts and sources of income for households with nine or more income sources (which are very rare). There are 44,686 households missing income information that we excluded from our estimates regarding income but not from other portions of the analysis.
General Assistance and TANF-Ω Adjustments
We made two types of adjustments for households reported as receiving income from a General Assistance (GA) program. For households that included children, if there is no GA program in their state or the state program does not serve families with minors, we considered them as receiving TANF benefits. We shifted 55,721 non-elderly, non-disabled households into the TANF category under this adjustment. For other households living in a state that does not offer a GA program, we changed the source of income to “other countable income,” except when the benefits were not counted in the determination of the household’s required rent contribution. If the housing agency did not count the payment in determining the family’s rent contribution (“total tenant payment”), we assumed that determination was correct and considered the income as “other excluded income.” As noted above, we made the same assumption for TANF income that was not included in the tenant rent calculation.
We discuss the following types of households.
Non-Elderly, Non-Disabled Households
We defined a household as non-elderly, non-disabled when the sole member, or the head and the co-head or spouse (if present), are younger than 62 and are not receiving Supplemental Security Income (SSI) benefits or identified in the administrative data as having a disability.
These are households where the sole member, or the head or co-head or spouse, meets the HUD definition of “elderly” or “disabled”. This is consistent with HUD’s definition of “elderly family” and “disabled family” at 24 C.F.R. §5.403. (The presence of a child or another adult with disabilities does not make a household a “disabled family.”) This category includes 91,596 households in which the head, co-head, or spouse is elderly or disabled but the other member of the couple is not.
Under the U.S. Housing Act and HUD’s implementing regulation (24 C.F.R. §5.403), an individual meets the definition of a “person with disabilities” if he or she has been determined to meet the disability standard established by the Social Security Act for receipt of Social Security Disability Insurance or Supplemental Security Income (SSI) benefits, or is determined by the responsible entity to have a physical, mental, or emotional impairment that is expected to be of long-continued and indefinite duration and substantially impedes the individual’s ability to live independently, but more suitable housing conditions could improve the individual’s ability to live independently. Individuals with developmental disabilities also are included in the HUD definition of disability.
To determine disability status, we used receipt of SSI benefits by individuals younger than 62 or a check in the “disabled” field on the HUD form. (We did not consider other income sources that may be tied to disability in some cases, such as Social Security, other pension income or General Assistance.) Our methodology appears to have shifted about 25,200 HCV households, 10,600 PH households and 12,300 PBRA households into the “elderly/disabled” category compared to HUD’s practice of relying solely on whether the “disabled” field on the form is checked for a particular household member. Given the frequent difficulty of assessing whether a person has disabilities and the legal limitations on housing agencies or providers seeking information related to disability status unless it is essential to program administration, the HUD practice likely undercounts the number of “disabled” households.
We used HUD’s definition of elderly as a person 62 years or older.
The HUD file does not flag households based on whether they are headed by one or two adults. We used one or more of the sex, age, disability status, and income variables to identify the presence of a co-head or spouse in the household. If data were entered for a co-head or spouse in any of these fields, we assumed the presence of a co-head or spouse in the household.
Presence of Minors in the Household
The data have a flag variable to identify households with minor children; however, this flag sometimes contradicts other children variables (age, disability, and income). To address this contradiction, we created our own flag, which considered all the child variables to identify households as having minors.
High- and Low-Cost Areas
We divided our sample of non-elderly, non-disabled households in each program into four quartiles based on HUD’s FY 2012 two-bedroom Fair Market Rent for the area in which they resided. We matched the county where each household resides with its two-bedroom FMR. Households residing in New England were matched to the two-bedroom FMR for their town.
Duration of Stay
We subtracted the admission date from the date of action in the 2010 data file to obtain the number of months in the program. We excluded 105,711 households that appeared to have erroneous data for either or both of these variables. Most of the excluded households are in the PBRA program, which may affect the reported findings on duration of stay for PBRA households. As with the rest of this analysis, we excluded from the estimates of duration of stay any households that have an “end of participation” record in the 2010 data. This provides a clearer picture of the lengths of stay of families continuing to participate in the program.
 CBPP Research Associate Thyria Alvarez-Sanchez analyzed the data, former intern Liza Getsinger worked on the initial draft, and Vice President for Family Income Support Policy LaDonna Pavetti contributed to the recommendations.
 For a description of key features of each of the major rental assistance programs, see “Policy Basics: The Housing Choice Voucher Program,” http://www.cbpp.org/cms/index.cfm?fa=view&id=279; “Policy Basics: Introduction to Public Housing,” http://www.cbpp.org/cms/index.cfm?fa=view&id=2528; and “Policy Basics: Section 8 Project-Based Rental Assistance,” http://www.cbpp.org/cms/index.cfm?fa=view&id=3891, Center on Budget and Policy Priorities, updated January 25, 2013.
 The technical appendix explains how we analyzed the data and key definitions. This paper builds on the analysis of data on participants in the Housing Choice Voucher program in 2010 in Barbara Sard and Thyria Alvarez-Sanchez, “Large Majority of Housing Voucher Recipients Work, Are Elderly, or Have Disabilities,” Center on Budget and Policy Priorities, December 2, 2011, http://www.cbpp.org/cms/index.cfm?fa=view&id=3634.
 Center on Budget and Policy Priorities, “Policy Basics: Federal Rental Assistance,” January 25, 2013, http://www.cbpp.org/cms/index.cfm?fa=view&id=3890. HUD programs provide rental assistance to 4.7 million of the 5 million households that receive federal rental assistance. (This figure does not include the homeless assistance programs because detailed income data are not available for these households.) The Rural Housing Service of the U.S. Department of Agriculture provides rental assistance for another 270,000 households, for whom detailed income data also are not available. HUD provides rental assistance through several small programs to close to 190,000 additional households, most of whom are elderly or have disabilities. National and state-level data on the households assisted under each of the rental assistance programs are available at http://www.cbpp.org/cms/index.cfm?fa=view&id=3586. Other federal programs provide grants, tax credits, or reduced-interest loans to build or rehabilitate rental housing, but without rental assistance based on family income, such programs generally do not make units affordable to poor families.
 “Worked recently” is defined here as receipt of unemployment insurance in 2010 or reported earnings in 2009.
 Some Section 8 PBRA properties were funded exclusively to house elderly or disabled households, under the precursor to the current Supportive Housing Programs for the Elderly (section 202) and Disabled (section 811).
 Of the households attached to the labor market, 6.6 percent received unemployment insurance in 2010. To receive these benefits, individuals must have amassed a significant recent work history. Chad Stone and William Chen, “Introduction to Unemployment Insurance,” Center on Budget and Policy Priorities, February 6, 2013, http://www.cbpp.org/cms/index.cfm?fa=view&id=1466.
 Based on data reported by agencies that did not participate in the Moving to Work demonstration, HUD reported to Congress that 59 percent of public housing households with one or more non-disabled adults aged 21 – 55 had income from work in 2007, compared with 54 percent in 2010. U.S. Department of Housing and Urban Development, Office of Public and Indian Housing, Office of Program, Policy, and Legislative Initiatives, “Tenth Annual Report to Congress on the Impact of the Quality Housing and Work Responsibility Act of 1998 on the Public Housing and Section 8 Voucher Program: 2000 to 2010,” Table 4.
 CBPP analysis of data from months 24 and 36 of the 2004 Survey of Income and Program Participation.
 Federal law requires states to meet strict work participation rates for “work-eligible” individuals in their TANF programs. Job search and training activities can meet these requirements to some extent. Work-eligible individuals include most adults receiving assistance, with a few exceptions. Between April and June 2011, states exempted about 15.5% of TANF work-eligible individuals from work requirements and excluded another 8 percent based on federal criteria (primarily caring for a child under the age of one). Nearly all states eventually deny benefits to recipients and their children if they do not comply with work requirements.
 Liz Schott and Clare Cho, “General Assistance Programs: Safety Net Weakening Despite Increased Need,” Center on Budget and Policy Priorities, December 19, 2011. See Table 1, p. 4.
 Only a very small share of these households had two adults: 4 percent in the HCV program, 10 percent of the public housing residents, and 7 percent of the PBRA tenants.
 Mary Corcoran and Colleen Heflin, “Barriers to Work Among Recipients of Housing Assistance.” Cityscape, 16:2, 2003, http://www.huduser.org/periodicals/cityscpe/vol6num2/1barriers_work.pdf; Susan J. Popkin and Elizabeth Davies, “Improving the Lives of Public Housing’s Most Vulnerable Families,” Urban Institute, January 2013, http://www.urban.org/UploadedPDF/412763-Improving-the-Lives-of-Public-Housings-Most-Vulnerable-Families.pdf; Joe Parilla and Brett Theodos, “Moving “Hard to House” Residents to Work,” Urban Institute, 2010, http://www.urban.org/UploadedPDF/412256-Moving-Hard-to-House-Residents.pdf.
 It is possible that additional households worked but their earnings were not included in the HUD administrative data. This could occur legitimately for two reasons. First, HUD rules allow public housing agencies to delay required reporting of income changes until families’ regular annual income recertification. Some agencies elect this option as a work incentive or to reduce administrative costs. Second, the administrative data used for this analysis include only one record per family per year. If the employed family member lost her job during 2010 and requested a rent reduction, only the most recent record showing no earned income would be available.
 The vast majority of these households (93 percent for HCV and 95 percent for public housing) relied on earnings as their primary source of income.
 Certain public housing program rules encourage higher-income families to remainin the program, particularly the option to pay a “flat” or ceiling rent that is less than 30 percent of family income and may be lower than rents for comparable units in the private market. (HUD’s 2014 budget would require such rents to be no lower than 80 percent of the metro-wide or county-level Fair Market Rent set by HUD to reflect the cost of modest rental housing in the area.) By contrast, families in the HCV program always are required to pay at least 30 percent of their income for rent, and if this amount exceeds the rent for their unit plus the utility allowance, the family’s subsidy is terminated (after a six-month suspension period).
 There is more flexibility in the public housing program than in the other programs to admit applicants with incomes between 51 – 80 percent of the local median income, which agencies could use to admit more working households. Private owners are not allowed to skip over applicants who applied earlier based on the amount of their earnings.
 As noted above (see note 15), the public housing program allows higher income households to pay a flat or ceiling rent that may be less than 30 percent of income for rent. Certain (although not most) public housing tenants also receive an earnings disregard in the first two years after they go to work. Non-disabled tenants in the HCV program and all PBRA households are ineligible for any disregard of their earned income in determining their rent obligation.
 See, e.g., D. A. Reingold, “Does inner city public housing exacerbate the employment problems of its tenants?” Journal of Urban Affairs, 19(4), 469 (1997).
 At non-MTW agencies in 2010, 66 percent of non-elderly, non-disabled HCV households and 65 percent of public housing households worked or had worked recently, compared to 64 percent of HCV households and 61 percent of public housing households served by MTW agencies. The appendix discusses the underreporting of data by MTW agencies. It is not clear whether underreporting affected these findings.
 HUD, “Frequently Asked Questions on Neighborhood Networks,” http://www.hud.gov/offices/hsg/mfh/nnw/aboutnn/whatisnn/nnwfaq.cfm, accessed June 13, 2013.
 Congress may have omitted PBRA tenants from eligibility for FSS because as originally enacted in 1990, the obligation to operate an FSS program applied to administration of additional assisted units and by that time Congress had ceased to fund new PBRA contracts.
 Barbara Sard, “The Family Self-Sufficiency Program: HUD’s Best Kept Secret for Promoting Employment and Asset Growth,” Center on Budget and Policy Priorities, April 2001, http://www.cbpp.org/cms/?fa=view&id=174.
 Robert C. Ficke and Andrea Piesse, “Evaluation of the Self-Sufficiency Program: Retrospective Analysis, 1996 to 2000,” HUD, April 2004, http://www.huduser.org/Publications/pdf/selfsufficiency.pdf; and Lalith de Silva et al, “Evaluation of the Family Self-Sufficiency Program: Prospective Study,” HUD, February 2011, http://www.huduser.org/portal/publications/familyselfsufficiency.pdf. See also Nandita Verma et al, “Working Toward Self-Sufficiency: Early Findings from a Program for Housing Voucher Recipients in New York City,” MDRC, December 2012, http://www.mdrc.org/publication/working-toward-self-sufficiency, regarding an evaluation of a modified FSS program, which found early positive results for the poorest households. A HUD-funded rigorous controlled evaluation of FSS is now in process.
 Some studies have found higher rates of employment and higher incomes among households in private assisted housing. E.g., Sandra Newman, C. Scott Holupka and Joseph Harkness, “The Long-Term Effects of Housing Assistance on Work and Welfare,” Journal of Policy Analysis and Management, Vol. 28, No. 1, 81-101 (2009). Such different findings may have resulted from inclusion of households that did not receive Section 8 subsidies, and paid rents set somewhat below market due to federal subsidies to reduce owners’ mortgage payments.
 For elderly and disabled households receiving assistance in 2010, the median lengths of stay were 63 months in the PBRA program, 77 months in the public housing program, and 72 months in the HCV program.
 This calculation is based on the actual rent and applicable utility allowance for the units families occupied in 2010.
 HUD, “Worst Case Housing Needs 2011: Report to Congress – Summary,” February 2013, http://www.huduser.org/portal/publications/affhsg/wc_HsgNeeds11.html. HUD defines “worst case” housing needs as renter households without housing assistance who have incomes below 50 percent of the median income in their local area and who paid more than half of their income for housing or lived in severely substandard housing.
 Since the recession began in late 2007, the number of homeless families with children living in temporary shelters or transitional housing has risen by 32 percent, according to the most recent HUD estimate. HUD Office of Community Planning and Development, “The 2011 Annual Homeless Assessment Report to Congress,” Dec. 2012, https://www.onecpd.info/resources/documents/2011AHAR_FinalReport.pdf.
 For a recent review of the literature, see Christopher T. King and Carolyn J. Heinrich, “How Effective are Workforce Development Programs: Implications for U.S. Workforce Policies, November 2011,http://www.utexas.edu/research/cshr/pubs/pdf/King-Heinrich-APPAM-10222011.pdf.
 For studies showing success from a mixed approach, see Gayle Hamilton, “Moving People from Welfare to Work: Lessons from the National Evaluation of Welfare-to-Work Strategies, USHHA/ASPE, 2002, http://www.mdrc.org/sites/default/files/full_453.pdf (Portland OR); Randall W Eberts, "Design, Implementation, and Evaluation of the Work First Profiling Pilot Project." ETA Occasional Paper 2002-07. Washington DC: U.S. Dept. of Labor, Employment and Training Administration, Office of Policy and Research, 2002, http://research.upjohn.org/externalpapers/26/ (Kalamazoo-St. Joseph MI); V. Joseph Hotz et al., “The Long-Term Gains from GAIN: A Re-Analysis of the Impacts of the California GAIN Program, RAND, 2000 (Riverside CA).
 Gayle Hamilton, “Moving People from Welfare to Work: Lessons from the National Evaluation of Welfare-to-Work Strategies,” id.
 James A. Riccio, “Sustained Earnings Gains for Residents in a Public Housing Jobs-Program, Seven-Year Findings from the Jobs-Plus Demonstration,” MDRC, 2010, http://www.mdrc.org/sites/default/files/policybrief_33.pdf. The three sites that implemented the partnership between the housing agency and the welfare and workforce agencies well were Los Angeles, St. Paul, and Dayton.HUD requested up to $15 million in the 2014 budget for public housing agencies that partner with local workforce investment boards and other agencies and organizations providing work supports, based on the Jobs-Plus demonstration. The fiscal year 2014 appropriations that bills the House and Senate Appropriations Committees have approved include these funds.
 Sheila Maguire et al., “Tuning In to Local Labor Markets: Findings from the Sectoral Employment Impact Study,” Public/Private Ventures, 2010, http://ppv.org/ppv/publications/assets/325_publication.pdf.
 Alicia Meckstroth et al., “Teaching Self-Sufficiency Through Home Visitation and Life Skills Education,” Mathematica Policy Research Inc., 2009. The cost-benefit analysis found that if the average earnings impacts observed in the last six months of the 30-month follow-up were to persist for 1.7 years longer and the program was targeted only to the hard-to-employ, the program would pay for itself.
 U.S. Department of Housing and Urban Development, Office of Public and Indian Housing, Office of Program, Policy, and Legislative Initiatives, “Tenth Annual Report to Congress on the Impact of the Quality Housing and Work Responsibility Act of 1998 on the Public Housing and Section 8 Voucher Program: 2000 to 2010.” Italics in original.
 James Riccio, “Subsidized Housing and Employment,” MDRC, 2007, http://www.mdrc.org/sites/default/files/full_521.pdf. Table 1 summarizes the results of 10 different analyses, 8 of which show larger and statistically significant earnings effects for welfare recipients who were receiving some form of housing assistance.
 These two policies are at pages 107-109 of the AHSSIA draft, available at http://www.cbpp.org/files/4-12-12-AHSSIA-text.pdf. Since 1998, federal law has required public housing agencies to make their “best efforts” to enter into cooperation agreements with public assistance agencies to promote the economic self-sufficiency of their tenants, but HUD has largely ignored the requirement, and other federal agencies have done little to make such efforts productive.
 U.S. Government Accountability Office, “Moving to Work Demonstration: Opportunities Exist to Improve Information and Monitoring,” GAO-12-490, April 2012 at 16.
 GAO, p 21.
 See note 20. The point-in-time comparison of work rates at MTW agencies compared to other PHAs does not illuminate whether efforts by MTW agencies impacted tenants’ earnings compared to what they would have been absent the flexibilities MTW provides. But the comparison does underscore the lack of evidence of positive results.
 Joe Parilla and Brett Theodos, “Moving “Hard to House” Residents to Work,” The Urban Institute, 2010, http://www.urban.org/UploadedPDF/412256-Moving-Hard-to-House-Residents.pdf.
 Susan J. Popkin and Elizabeth Davies, “Improving the Lives of Public Housing’s Most Vulnerable Families,” Urban Institute, January 2013, http://www.urban.org/UploadedPDF/412763-Improving-the-Lives-of-Public-Housings-Most-Vulnerable-Families.pdf; Susan J. Popkin et al., “CHA Residents and the Plan for Transformation,” Urban Institute, January 2013, http://www.urban.org/UploadedPDF/412761-CHA-Residents-and-the-Plan-for-Transformation.pdf. Reported health improvements also could have been due to improved and more stable housing. The follow-up study found an 18 percentage point increase in households reporting some employment compared to 2007 for the subgroup of families who relocated to traditional public housing, but no increase in average household incomes. The authors note that this may have been due in part to the work requirement that the Chicago Housing Authority initiated in 2009, but without a controlled evaluation of the work requirement, its role could not be definitively determined.
 The average annual cost of services was estimated to be $2,900, which also included assistance with relocation, as all of the families in the Demonstration had to move due to the demolition of the public housing properties where they lived. Participants who received substance abuse services or Transitional Jobs stipends and services were significantly more costly than the average. Susan J. Popkin et al., “A New Model for Integrating Housing and Services,” Urban Institute, December 2010, http://www.urban.org/UploadedPDF/412255-New-Model-for-Integrating-Housing.pdf.
 Will Fischer, “‘Moving to Work’ Compromise Would Reduce Risk from Expanding Demonstration,” Center on Budget and Policy Priorities, June 25, 2013, http://www.cbpp.org/cms/index.cfm?fa=view&id=3981.
 The imputed welfare income rule is 24 C.F.R. §5.615. The HUD data show a total of 5,656 families subject to the special “imputed welfare income” rent rules due to a work-related sanction. Most of these families still received some amount of TANF benefits or were working.
 HUD rules count as “income” for rent-determination purposes only TANF benefits that qualify as “assistance” under federal rules. (24 C.F.R. §5.609(b)(6).) Typically, benefits count as “assistance” only if they are received for more than four months.
 For information on state GA programs, see Liz Schott and Clare Cho, “General Assistance Programs: Safety Net Weakening Despite Increased Need,” Center on Budget and Policy Priorities, December 19, 2011, http://www.cbpp.org/cms/?fa=view&id=3603.
 HUD rules require the exclusion of “temporary, nonrecurring or sporadic income.” 24 C.F.R. §5.609(c)(9).