«Moving to opportunity voluMe 14, nuMber 2 • 2012 U.S. Department of Housing and Urban Development | Office of Policy Development and Research ...»
The Section 8 and control groups no longer exhibited any statistical difference, however. Both the experimental and Section 8 groups were significantly less likely than the control group to report a variety of housing problems, including peeling paint, broken plumbing, vermin, and broken windows. Households in the Section 8 group experienced fewer problems than the control group with broken heat, and the households in the experimental group experienced fewer problems with broken locks.
The consequences of living in substandard housing have been well documented. Poor housing quality has been linked with poor physical and mental health (Krieger and Higgins, 2002; Matte and Jacobs, 2000). Children and adults living in damp, moldy housing, often the result of plumbing problems, tend to have more medical problems such as respiratory problems, headaches, nausea, and vomiting compared with residents living in drier houses (Hunt, 1993; Platt et al., 1989).
Increased incidents of asthma have been linked to exposure to vermin such as cockroaches, mice, and rats (Rosenstreich et al., 1997). In addition, in a review of literature connecting mental health and housing quality, Evans, Wells, and Moch (2003) found a positive correlation between housing quality and psychological well-being. This same result has also been seen in children; children living in poor-quality housing are more likely to display behavioral problems and have difficulties concentrating (Evans, Saltzman, and Cooperman, 2001).
The MTO final impacts evaluation found important positive mental and physical health effects for adults and female youth (Sanbonmatsu et al., 2011). Most previous research studies relied primarily on correlations that omitted individual or family characteristics that could have affected the types and quality of housing those families selected. The experimental and longitudinal design of the MTO demonstration removed the selection bias that may have challenged previous studies.
The sustained housing improvements documented for the families in the experimental group might have contributed to these important physical and mental health gains.
No Effect on Housing Assistance Receipt Part of the rationale for the MTO demonstration was that moving to low-poverty, resource-rich neighborhoods could potentially improve families’ economic well-being. It follows that if families’ economic status improved, they might earn their way off housing assistance. The specific form of housing assistance a family receives (that is, public housing or tenant-based housing voucher) may also be an important mediator for other key MTO demonstration outcomes. Families who receive vouchers may have more and better options about where to live than those living in public housing. Alternatively, voucher holders may experience more residential instability and have trouble making payments and operating in the private market, which could undermine other outcomes.
Ultimately, MTO had no effect on housing assistance receipt at the time of the final impacts evaluation. No statistical significance emerged between the experimental and control groups in the proportion of households still on assistance, and the Section 8 group was only slightly more likely to be receiving any sort of assistance than the control group (5 percentage points more, where p.10). This finding is consistent with other findings from the final impacts evaluation, showing
no effect on families’ employment levels and earnings. Therefore, our finding that families in the experimental group were no more likely than the families in the control group to “income out” of housing subsidies is unsurprising.
Although no ITT effect existed, exhibit 3 shows that almost two-thirds (62 percent) of the households in the control group were still receiving housing assistance 10 to 15 years after the start of the demonstration. The proportion still on assistance was slightly lower than at the time of the interim survey, which was 71 percent (determined exclusively from the interim survey responses) or 66 percent (determined exclusively from the administrative data) of the control group (Orr et al., 2003). Analysts did not use the previously described new triangulation method for the interim survey results.
The fact that so many MTO participants were still receiving housing subsidies after 10 to 15 years is unusual. The median length of time that households use housing assistance is 4.7 years for those living in public housing and 3.1 years for voucher holders, and families with children receive assistance for even less time—3.2 years for those living in public housing and 2.6 years for those participating in the voucher program (Turner and Kingsley, 2008). MTO families started in severely distressed public housing developments, however; by the 1990s, when the MTO demonstration began, families with more resources or better options largely had fled the original public housing and project-based housing because of crime and disorder, leaving behind a population dominated by the most vulnerable households. The HOPE VI Panel Study, which tracked families relocated as the result of HOPE VI redevelopment initiatives, found a similar pattern; at baseline, those respondents reported having lived in public housing for 10 or more years, on average (Popkin, Levy, and Buron, 2009). These long-term public housing residents, in general, have tenuous connections to the labor market and are unlikely to earn their way off assistance (Briggs, Popkin, and Goering, 2010; Theodos et al., 2012).
Finally, compared with households in the control group, households in both the experimental and Section 8 groups were less likely to be living in public housing and more likely to be using a
voucher (16 and 19 percentage points more, respectively). This finding is not surprising, given that both groups were originally offered vouchers. More surprising is that 25 percent of the households in the control group were receiving vouchers at the time of the final impacts evaluation. This finding likely reflects the fact that 42 percent of MTO families started off in public housing developments that were later “HOPE VI’ed” and could have been offered a voucher during the relocation process (Sanbonmatsu et al., 2011).
Mixed Effect on Housing Instability The ITT analysis found mixed results regarding whether being offered a voucher resulted in more instances of homelessness or doubling up. Being offered a voucher had no discernible additional effect on the amount of time spent homeless for the experimental or Section 8 groups (exhibit 4), a finding that did not support the hypothesis that the treatment would improve families’ social and economic circumstances, enabling families to earn their way off housing assistance and possibly leaving them vulnerable to economic reversals. The Section 8 group, however, was 5 percentage points more likely than the control group to experience doubling up with friends and family. Looking at the groups who actually leased up with the voucher (TOT), as opposed to just being offered the voucher (ITT), the instances of doubling up for the Section 8 group increased to 7 percentage points more than the control group.
Although results were mixed regarding whether being offered a voucher influenced homelessness or doubling up, 21 percent of control group participants reported experiencing being either doubled up or homeless at some point during the demonstration. Nearly one in five (19 percent) of the control group experienced at least one instance of doubling up and 5 percent experienced a spell of homelessness at some point during the demonstration.
Assessing whether these rates of homelessness and being doubled up are higher than usual for a population that has received housing assistance is difficult. The best benchmarks are the HOPE VI Panel Study, which found similar rates of homelessness among original residents of HOPE VI developments, and MTO participants at the time of the interim evaluation. The HOPE VI Panel Study found that at the time of the survey or during the previous 12 months, 1.7 percent of HOPE VI participants lived in a homeless shelter or on the streets and 3.9 percent lived in a doubled-up situation (McInnis, Buron, and Popkin, 2007). These shares were similar to those found in an analysis of all three groups of MTO participants at the time of the interim survey who reported that they did not live in the same housing unit for the past 12 months.
Another often-referenced random assignment research study, the Welfare-to-Work voucher study, found that housing vouchers dramatically reduced the risk of homelessness (Khadduri, 2008;
Mills et al., 2006; Patterson et al., 2004). Participants in the Welfare-to-Work voucher study, however, were likely to start off in the unsubsidized private market, not in public housing like the MTO families. Only 13 percent of Welfare-to-Work participants reported receiving some type of housing assistance at baseline, before any treatment was applied (Mills et al., 2006). In addition, the effects were tracked for a much shorter period of time. Note that at baseline, 25.8 percent of Welfare-to-Work participants reported living with family or friends (comparable to the MTO doubled-up category) and 1.9 percent reported living in a homeless shelter or transitional housing;
the percentage of MTO participants who reported experiencing homelessness at the time of the final impacts evaluation was twice as great. Because the populations differ so markedly, however, a clear comparison is difficult to make.
No Effect on Housing Affordability Exhibit 5 shows that average housing costs for households in the control group at the time of the final impacts evaluation were relatively low, at $493 for monthly rent or mortgage and $679 total, even when compared with just the average rents in the five metropolitan areas. Because almost two-thirds of MTO heads of household were receiving federal housing subsidies, we would expect their housing costs to be relatively low. Furthermore, no statistically significant differences emerged between either the experimental and control groups or the Section 8 and control groups, except that low-poverty voucher group households paid approximately $10 per month more for electricity than control group households.
Whether the low-poverty voucher treatment would result in higher cost burdens for families in the experimental group was another question.8 Economic improvements for households in the Housing cost burdens are the ratio of the amount that families pay for their rent or mortgage plus utilities over their household income. HUD recommends that families pay no more than 30 percent of their income on housing costs;
otherwise, they are considered to have high housing cost burdens.
experimental group could have resulted in families’ “incoming out” of their housing assistance, removing an important safety net. As exhibit 6 shows, however, neither low-poverty nor traditional voucher offers had an effect, either positive or negative, on housing cost burden.
Exhibit 6 also shows that many MTO families had high housing cost burdens. More than two-thirds of the households in the control group paid more than 30 percent of their incomes on housing and 43 percent were severely burdened with housing costs, paying more than 50 percent of their
incomes on housing. This finding is surprising considering that two-thirds of MTO participants received housing subsidies that should have buffered households from spending so much of their incomes on housing.