Almost 24 percent of voucher families offered mobility services moved to an opportunity area against 4 percent who were not, and HUD has cancelled the arm of the experiment that was meant to find a cheaper version
Twenty percentage points is a very large effect for a housing programme.
An analysis of twelve months of data from the Community Choice Demonstration found that almost 24 percent of voucher families with children who were offered Comprehensive Mobility-Related Services moved to an opportunity area, against 4 percent of families in the control group. The sample was about 600 families and the result was preliminary. It reached HUD in 2024, and a notice filed for public inspection on Friday morning explains what the department did next.
It cancelled the second treatment arm.
The arm that will not be tested
The demonstration was designed in two phases. Phase 1 ran from 2022 and delivered the comprehensive services package, which addresses the obstacles that stop voucher holders reaching lower poverty areas: the money, the knowledge gaps, the reluctance of families to move and of landlords in those areas to take part, and the difficulty of staying once there. Phase 2 was to run from 2025 to 2028 and add Selected Mobility-Related Services, a second arm testing up to three smaller bundles drawn from the same suite, to find out whether a cheaper version would work nearly as well.
That is the question that will now go unanswered. HUD writes that in early 2025 it became concerned that adding the second arm could undermine the impact of the first and put too much pressure on sites that were already struggling, and it chose instead to improve the comprehensive package, add staff at sites that needed it and expand training.
The enrollment numbers explain the pressure. Between August 2022 and August 2024 the sites signed up about 3,200 families, which the notice puts at 67 percent of the target for that point.
Enrollment ended ten months early, and could have run to 2028
With one arm instead of two, HUD and its evaluation contractor recalculated the sample size the research questions actually require, and found it smaller. Study enrollment closed on 30 June 2026.
That date replaces two earlier ones. The 2020 implementation notice had agencies enrolling families through 30 April 2027, and guidance published in July 2024 allowed an agency to extend to 31 December 2027, or to 30 April 2028, at its own discretion to meet its enrollment commitments. Measured against the first of those the study closed ten months early. Measured against the furthest extension it closed twenty-two months early.
The money
Congress provided $50m for services and new vouchers across the 2019 and 2020 appropriations, with a further $3m under a separate heading for the evaluation itself. HUD awarded $45.7m to nine lead and four partner public housing agencies, split as $35.9m in service funding and $9.8m in payments funding for the Mobility Demonstration Vouchers.
Agencies have since withdrawn. HUD has recaptured roughly $7.4m of service funding and $3m of voucher payment funding, and it will reallocate about $7.4m on demonstrated need, taking year 6 budgets first. Each remaining site may still spend up to $80,000 a year on recruitment and enrollment staff time, even though the study-specific work of informed consent and baseline questionnaires has ended, because recruitment continues and the sites now have to monitor which families are embargoed.
The control group is the reason for the embargo. Until 1 October 2028, a participating agency is strictly prohibited from giving comprehensive mobility services to a family that was randomly assigned to the control group, so that the comparison holds to the end of the demonstration. The notice is signed by Benjamin Hobbs, assistant secretary for Public and Indian Housing.
