Neighborhood-Based Subsidy Caps Can Make Housing Vouchers More Efficient and Effective
On June 2, the Department of Housing and Urban Development (HUD) announced plans to expand a policy designed to help families with Housing Choice Vouchers move to higher-opportunity areas by tying voucher subsidy amounts to rents in a given neighborhood rather than in the entire metropolitan area. Early evidence suggests that the policy, known as “small-area fair market rents” (SAFMRs), enables voucher holders to live in neighborhoods with lower poverty and crime rates. A growing body of research shows that living in lower-poverty neighborhoods has substantial benefits for families, including raising children’s college attendance and adult earnings considerably.
Expanding use of SAFMRs would be an important step to broaden housing opportunities for voucher holders — and, importantly, requires neither congressional action nor additional funding. In fact, because HUD’s plan would expand SAFMRs to replace a costly but apparently ineffective policy of raising voucher subsidies throughout certain metropolitan areas, it would likely reduce voucher costs, freeing up funds that could be used to assist additional families on voucher waiting lists.
Current Caps Limit Use of Vouchers in Low-Poverty Neighborhoods
Housing Choice Vouchers, the nation’s largest form of low-income rental assistance, help more than 2 million families rent modest units of their choice in the private market. Families pay 30 percent of their income toward rent and utilities and the voucher covers the rest, up to a cap called the “payment standard.” A family can choose to rent a unit for more than the payment standard but must pay the excess rent itself, in addition to the 30 percent of income it would otherwise pay.
State and local housing agencies must generally set the payment standard within 10 percent of the “fair market rent” (FMR) that HUD has established for the area. In most of the country, HUD sets FMRs at the 40th percentile of market rents for standard-quality, recently rented units in an entire metropolitan area or rural county. In certain metropolitan areas where families have difficulty using vouchers in a wide range of neighborhoods, HUD sets the FMR at the 50th percentile of market rents.
Rents can vary greatly from one neighborhood to another, so payment standards based on metro-wide FMRs are often too low to cover market rents in some neighborhoods and higher than needed in others. When the payment standard in a neighborhood is too high, families can afford units that are larger or have more amenities than they need, and owners can potentially charge above-market rents (unless housing agencies strictly enforce rules requiring that rents be reasonable in the local market). This reduces the cost-effectiveness of the voucher program and encourages families to use vouchers — and owners to accept them — in the lowest-rent neighborhoods. When the payment standard is too low (as is often the case in neighborhoods with low poverty, low crime, and high-performing schools), many families will struggle to find units they can afford to rent with their voucher.
Metro-level FMRs are one reason that vouchers have not been as effective as they could be in enabling families to move to neighborhoods with low poverty rates. Just 20 percent of families with children in the voucher program live in low-poverty neighborhoods (those where fewer than 10 percent of residents are poor). This is less than half of the share of all children that live in those neighborhoods.
The voucher program’s performance in broadening housing choices for participants can profoundly affect children’s long-term well-being. Research finds that children whose families used vouchers to move to low-poverty neighborhoods when they were young were more likely to attend college, less likely to become single parents, and earned significantly more as adults. (See Figure 1.) In addition, adults who moved to lower-poverty neighborhoods experienced substantial gains in mental and physical health.
50th Percentile FMRs Have Not Reduced Concentration of Voucher Holders in High-Poverty Neighborhoods
Since 2001, HUD has set FMRs at the 50th percentile of market rents in large metropolitan areas where voucher holders and affordable rental units are concentrated in a limited number of census tracts. In the 16 areas with 50th percentile FMRs in 2015, this policy raises FMRs by an average of 7 percent, according to HUD. The 50th percentile FMR policy was intended to help voucher holders rent in a wider range of neighborhoods, but evidence suggests that it has not been effective.
In many of the areas where 50th percentile FMRs have been instituted, there has not been a measurable reduction in the concentration of voucher holders, according to HUD. More broadly, between 2000 and 2010, voucher holders in the nation’s 50 largest metropolitan areas became more concentrated and more likely to live in neighborhoods with poverty rates at or above 30 percent, even though 50th percentile FMRs were used in most of those areas at some point during that period. A 2015 study by researchers Robert Collinson and Peter Ganong of the initial implementation of 50th percentile FMRs in 2001 found that the policy’s main effect had been to raise costs without improving the quality of the housing where voucher holders lived.
Two major flaws have likely undermined the policy’s effectiveness. First, HUD has implemented it in a manner that frequently shifts areas back and forth between 40th and 50th percentile FMRs. These shifts disrupt agencies, families, and owners and could be expected to reduce the policy’s impact on where voucher holders live.
Second, and more fundamentally, raising the FMR throughout a metropolitan area is unlikely to be the best way to encourage voucher use in low-poverty areas. While the higher payment standards help voucher holders afford housing in high-rent neighborhoods, they also encourage use of vouchers in low-rent areas by pushing payment standards there even further above market rents.
Small-Area Fair Market Rents Offer Promising Alternative
In recent years, HUD has tested an alternative “small-area fair market rent” (SAFMR) policy. SAFMRs are set at the 40th percentile of market rents for a particular zip code, so they reflect neighborhood rents more accurately than metro-level FMRs. SAFMRs were first implemented in the Dallas metropolitan area in December 2010 and early 2011 pursuant to a settlement of litigation claiming that metro-wide FMRs prevented minority voucher holders from moving to predominantly white neighborhoods with higher rents. Two years later, HUD initiated an SAFMR demonstration that allows five agencies in other parts of the country to test the approach.
HUD selected zip codes as SAFMR areas because they are small enough to reflect neighborhood rent differences and because voucher holders, landlords, and housing agencies can identify a unit’s zip code more easily than they can match the unit to other small areas for which data are available, such as census tracts.
Under HUD’s SAFMR policy, agencies can still set payment standards anywhere within 10 percent of the SAFMRs, giving them substantial flexibility for adjustments to reflect additional data on local rents, to simplify administration, or for other purposes. For example, the Cook County (IL) Housing Authority used this flexibility to consolidate 122 zip codes into ten payment standard zones because all zip codes in each zone have FMRs within 10 percent of the zone’s payment standard.
SAFMRs Can Help Families Move to Higher-Opportunity Areas
Early evidence suggests that SAFMRs help families move to higher-opportunity neighborhoods. Three years after the shift to SAFMRs in Dallas, people who used vouchers to move lived in neighborhoods with 17 percent less violent crime, and poverty rates 2 percentage points lower, than they would have lived in under metro-level FMRs, the Collinson and Ganong study estimated. The larger reduction in violent crime is consistent with other research suggesting that when low-income families’ housing choices are expanded, they place a priority on finding homes in neighborhoods with less crime.
These results are particularly impressive because SAFMRs, unlike some initiatives to improve locations for voucher holders, do not require families to use vouchers in particular types of neighborhoods. SAFMRs eliminate the incentives created by metro-level FMRs for families to live in neighborhoods with rents below the metro average: under SAFMRs, living in a very low-rent neighborhood will no longer enable families to rent units that are larger or have more amenities, because payment standards in that neighborhood will be lower. But families can still choose to live in low-rent neighborhoods if they wish to, for example because of ties to friends and family or proximity to a household member’s workplace.
Average Voucher Costs Have Fallen In Areas Using SAFMRs
Importantly, SAFMRs can help voucher holders move to higher-opportunity areas without raising overall voucher subsidy costs. On average across all rental units, SAFMRs roughly equal the 40th percentile FMR for the metropolitan area in which they are located. Yet the average voucher holder lives today in a zip code with an SAFMR well below the metro average, because vouchers tend to be used in lower-rent, higher-poverty neighborhoods. Voucher costs would drop by about 6 percent if SAFMRs were implemented nationally (after the change is fully phased in) and voucher holders continued to live in the same neighborhoods. Of course, the actual savings would be lower because SAFMRs would encourage some families to move to higher-rent, lower-poverty neighborhoods. Those moves likely would not be so widespread as to eliminate the savings entirely, particularly since negative factors such as racial discrimination and the unwillingness of some landlords to accept vouchers would still restrict choices for a substantial number of voucher holders.
In practice, voucher subsidy costs have fallen in areas where SAFMRs have been implemented, although it is possible that factors other than SAFMRs may have played a role in those declines. The average annual cost of a voucher in the Dallas area fell by $375 or about 5 percent from 2010 (just before SAFMRs went into effect) through 2014, without adjusting for inflation, HUD administrative data show. Average voucher costs rose by 2 percent both in the neighboring Fort Worth area and nationally during that period. At the five agencies in the SAFMR demonstration, average voucher costs fell by 5 percent from 2012 through 2014. (See Figure 2.)
SAFMRs would create some added administrative costs for activities such as training staff, educating landlords, reprogramming computers, and determining zip code payment standards. But these costs would likely be modest and consist largely of one-time transition expenditures. For example, one and a half years after the Dallas Housing Authority implemented SAFMRs, housing authority staff estimated the added cost at $10 per voucher.
HUD Plan Would Extend SAFMRs to Areas Where They Are Needed Most
HUD has broad discretion to determine how FMRs are set, so it could expand SAFMRs to additional areas without congressional action. On June 2, HUD issued an “Advance Notice of Proposed Rulemaking” announcing that it plans to propose regulations replacing 50th percentile FMRs with scaled-up use of SAFMRs and requesting public comments on the details of the policy. HUD intends to require use of SAFMRs in areas where they are needed most to expand housing opportunities for voucher holders, such as areas where voucher holders are disproportionately concentrated in high-poverty areas and where a large share of rental units are in zip codes where SAFMRs would be more than 10 percent above or below the metropolitan FMR.
The available evidence suggests that SAFMRs will be more effective than metro-wide FMRs at enabling families to move to lower-poverty areas. In addition, replacing 50th percentile FMRs with 40th percentile SAFMRs will likely reduce voucher costs significantly. Converting the 16 areas that now have 50th percentile FMRs to SAFMRs would, by itself, reduce costs by approximately 10 percent (or about $130 million annually) once the change is fully phased in — although the actual reduction would be lower since part of the savings would be offset by voucher holders moving to higher-rent neighborhoods. (Since HUD is considering using different criteria to select SAFMR areas than it uses to select 50th percentile areas today, some current 50th percentile FMR areas may switch to 40th percentile metro-wide FMRs and some current 40th percentile FMR areas may switch to SAFMRs, but this would also produce substantial net savings.)
HUD doesn’t currently plan to require use of SAFMRs throughout all metropolitan areas. This is reasonable. HUD will prioritize implementation of SAFMRs in areas where the benefits are clearest; findings from those areas and the ongoing SAFMR demonstration can inform future expansion decisions. HUD is, however, considering taking the important and beneficial step of allowing agencies in all metropolitan areas to implement SAFMRs.
HUD Could Improve SAFMR Policy Further
SAFMRs have generated promising results so far, and the basic design is sound. But HUD, in addition to extending SAFMRs to additional areas, should consider certain improvements.
Protection Against Sharp FMR Declines
As noted above, SAFMRs would raise FMRs in high-rent zip codes but reduce them in low-rent zip codes. If HUD expands SAFMRs to additional areas, it should phase in the sharpest FMR declines to avoid unnecessary disruption for low-income families. HUD’s existing regulations provide some protection, since payment standard reductions (and the resulting rent increases) are not applied to voucher holders who remain in place until a one- to two-year hold-harmless period has expired. Lower SAFMRs, however, could result in substantial rent increases for families that wish to remain beyond that period. Moreover, large, abrupt FMR reductions could narrow housing choices for voucher holders who wish to move by closing out opportunities in low-rent areas before new opportunities emerge in higher-rent areas. It could take time for owners in higher-rent areas to become accustomed to renting to voucher holders and for agencies to take steps to support voucher use there.
In 2011, HUD said that in zip codes where the SAFMR is more than 10 percent below the FMR, it would phase in the decline so that the FMR does not drop by more than 10 percent per year. HUD appears to have abandoned this approach due to concerns that it lacks the necessary legal authority. These concerns seem ill-placed, however, and in any case HUD has unambiguous authority for a policy that would provide very similar protections.
Improved Method for Setting SAFMRs
HUD should closely review the methodology it uses for setting SAFMRs for possible improvements to better reflect market rents in each zip code. SAFMRs come much closer to neighborhood rent levels than metro-level FMRs do, but evidence suggests they could be made even more accurate.
For example, the Inclusive Communities Project, a Dallas-based non-profit that operates a large housing mobility program, supports continued use of SAFMRs in Dallas but has argued — based on analysis of commercial rent survey data — that the SAFMRs HUD has published since 2012 tend to be lower than market rents in some neighborhoods and higher in others. Without these inaccuracies, SAFMRs would likely be even more effective in enabling voucher holders to move to higher-opportunity areas. Moreover, some aspects of HUD’s methodology likely distort SAFMRs in at least some neighborhoods. For example, in one phase of its calculations HUD lumps together rents for units of various sizes in a manner that will tend to set SAFMRs too high in zip codes where efficiencies and one-bedroom units make up an above-average share of rental units and set SAFMRs too low in areas with an above-average share of larger units.
HUD’s proposed SAFMR expansion is unlikely to occur before October 2016, since HUD is required by statute to update FMRs each October 1 and is unlikely to complete the needed regulatory change by October 2015. HUD should therefore have ample time to carefully review its methodology and implement any needed changes before the SAFMR expansion occurs.
HUD’s plan to extend SAFMRs to additional areas is an important step to expand opportunities for low-income families. Early evidence suggests that SAFMRs enable voucher holders to move to lower-poverty, lower-crime neighborhoods, and HUD’s plan would likely also reduce voucher costs, particularly since it would replace the costly and apparently ineffective 50th percentile FMR policy. It will be important, however, for HUD to conduct a careful review to ensure that SAFMRs are as accurate as possible, and for HUD to include in the regulation that expands use of SAFMRs strong measures to ease the effects of the sharpest FMR declines on low-income families.
 Families are not permitted to newly rent a unit with a voucher if their total payment would initially exceed 40 percent of their income.
 Housing agencies can request HUD approval to set payment standards above 110 percent of the local fair market rent, based on data showing that the increase is needed to cover market rents and avoid hardship for families, or below 90 percent of the FMR, if they can show this will not result in excessive rent burdens for families. Agencies can also set payment standards up to 120 percent of the FMR to enable individual households where a member has a disability to afford accessible units, and may request HUD approval of standards above 120 percent for this purpose. The 39 agencies participating in HUD’s Moving to Work demonstration have additional flexibility regarding payment standards.
 Analysis of 2012 HUD administrative data shows that families in higher-poverty census tracts are more likely to rent larger units — that is, units with more bedrooms than their subsidy is based on — compared to voucher holders in lower-poverty tracts.
 For additional discussion of the voucher program’s performance in enabling families to live in lower-poverty areas and other policies that could improve that performance, see Barbara Sard and Douglas Rice, “Creating Opportunity for Children: How Housing Location Can Make a Difference,” Center on Budget and Policy Priorities, October 15, 2014, https://www.cbpp.org/research/creating-opportunity-for-children.
 Raj Chetty, Nathaniel Hendren, and Lawrence F. Katz, “The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment,” May 2015, http://scholar.harvard.edu/files/lkatz/files/mto_manuscript_may2015.pdf.
 Lisa Sanbonmatsu et al., “Moving to Opportunity for Fair Housing Demonstration Program: Final Impacts Evaluation,” National Bureau of Economic Research, November 2011, http://www.huduser.org/portal/publications/pubasst/MTOFHD.html.
 “Establishing a More Effective Fair Market Rent (FMR) System; Using Small Area Fair Market Rents (SAFMRs) in Housing Choice Voucher Program Instead of the Current 50th Percentile FMRs; Advance Notice of Proposed Rulemaking,” Federal Register, vol. 80, no. 105, June 2, 2015, pp. 31333, http://www.gpo.gov/fdsys/pkg/FR-2015-06-02/pdf/2015-13430.pdf.
 Kirk McClure, Alex F. Schwartz, and Lydia B. Taghavi, “Housing Choice Voucher Location Patterns a Decade Later,” paper presented at the Annual Conference of the Association of Collegiate Schools of Planning, Cincinnati, Ohio, November 2012.
 Robert A. Collinson and Peter Ganong, “The Incidence of Housing Voucher Generosity,” May 2015, p. 16, http://papers.ssrn.com/sol3/Papers.cfm?abstract_id=2255799.
 Metropolitan areas qualify for 50th percentile FMRs if they have at least 100 census tracts, no more than 70 percent of the tracts have rents at or below the 40th percentile FMR in 30 percent or more of their two-bedroom units, and at least 25 percent of voucher holders live in 5 percent of the area’s census tracts. After three years, if the share of voucher holders in the 5 percent of census tracts with the most voucher holders has not declined, FMRs are reduced to the 40th percentile until the area is reevaluated after another three years. But if this share falls below 25 percent, the area “graduates” to the 40th percentile and is reevaluated each year to determine whether it again qualifies for 50th percentile FMRs. Thus, areas can only retain 50th percentile FMRs continuously so long as the voucher concentration indicator declines each three-year period but not so much that it drops below 25 percent.
 Collinson and Ganong, p. 38.
 Michael C. Lens, Ingrid Gould Ellen, and Katherine O’Regan, “Do Vouchers Help Low-Income Households Live in Safer Neighborhoods?” Cityscape, 13:3, 2011, http://www.huduser.org/portal/periodicals/cityscpe/vol13num3/Cityscape_Nov2011_dovouchers_help.pdf.
 HUD’s Moving to Opportunity (MTO) demonstration, for example, tested vouchers that families could only use in neighborhoods with poverty rates of 10 percent or less.
 This cost estimate is based on CBPP analysis of 2014 HUD voucher administrative data and 2015 FMRs and SAFMRs, with adjustments to the administrative data to reflect inflation from 2014 to 2015. We assumed that under SAFMRs, housing agencies would set payment standards as close as they can to the payment standards they set under metro-level FMRs while remaining between 90 and 110 percent of the SAFMR. Our analysis excluded vouchers administered by agencies participating in the Moving to Work demonstration (because they are not subject to the same restrictions regarding payment standards as other agencies) and agencies that already use SAFMRs.
 Collinson and Ganong (p. 37) estimated that through 2013, the switch to SAFMRs in Dallas was linked to a 3 percent reduction in rents (which would reduce voucher costs by a somewhat larger percentage), but this effect was not statistically significant.
 Collinson and Ganong, p. 19.
Federal Register, vol. 80, no. 105, June 2, 2015, pp. 31332-31336.
 HUD’s notice requests comments on whether it should only require individual agencies whose vouchers are highly concentrated in high-poverty areas to use SAFMRs, rather than applying the requirement to all agencies in metropolitan areas where vouchers are highly concentrated. An agency-by-agency SAFMR requirement of this type would seriously undermine the policy’s effectiveness, however, since it would often require use of SAFMRs that reduce payment standards in high-poverty central cities but leave in place inadequate metro-level FMRs in high-rent, high-opportunity suburbs.
 While overall voucher costs would likely drop under HUD’s proposal, individual agencies whose jurisdictions consist mainly of high-rent zip codes could experience increases in cost. The 2016 HUD appropriations bill that the House of Representatives approved on June 9 allowed HUD to provide funding to cover added voucher costs at SAFMR demonstration agencies. Congress could grant HUD similar authority to provide added funds to agencies whose costs rise under the planned SAFMR expansion. Those added funds would be offset at least in part by reduced funding eligibility at some of the agencies whose average voucher costs decline.
 Phasing in FMR declines as proposed here would increase budgetary costs — or reduce savings — during the first years of SAFMR implementation. If necessary, HUD could offset that effect by delaying payment standard increases so they take effect at the same time as payment standard reductions. Existing regulations delay application of payment standard reductions (and the resulting rent increases) until the family’s second annual recertification after the agency changes its payment standard, meaning the reductions would occur between one and two years after the agency change. Payment standard increases, however, apply at the family’s next annual recertification (that is, within one year). Revised payment standards apply immediately to new voucher holders and voucher holders who move.
These are sensible policies when payment standards change due to factors such as a rise (or fall) in market rents throughout a metropolitan area, since families should receive time to adjust to payment standard reductions but there is no need to delay payment standard increases. However, in the context of SAFMRs — which will raise payment standards in some neighborhoods but lower them in others — the existing rules mean that the costs of the transition are felt sooner than the savings. HUD could avoid this by revising its regulations to apply both reductions and increases in payment standards resulting from SAFMRs at the family’s second annual recertification for voucher holders remaining in place.
 Federal Register, volume 76, number 76, April 20, 2011, p. 22124.
 Federal Register, volume 78, number 192, October 3, 2013, p. 61671.
 HUD’s voucher program counsel has interpreted language in the U.S. Housing Act requiring HUD to set FMRs using “the most recent available data” to prohibit any policy limiting FMR changes compared to FMRs in place in a previous year, including a limitation on the year-to-year percentage change. It would be worthwhile for HUD to reconsider this interpretation, since it is unclear that this statutory language prohibits HUD from making a one-time, policy-based decision to phase in SAFMR implementation to prevent sudden, harmful payment standard declines. But even without a change in this interpretation, HUD could limit how far SAFMRs can fall below the current metro FMR. For example, HUD could set SAFMRs no lower than 90 percent of the metro FMR in the first year of implementation, no lower than 80 percent in the second year, and so on. This would have the effect of preventing nearly all of the sharpest declines and would unambiguously be within HUD’s authority since it makes no use of less-recent data.