Moving to Opportunity Experiment (MTO)

The cost of a voucher relative to public housing is roughly equivalent, so that the main additional cost is the cost of counseling and move services for families in the experimental group, which totals $3,783.

To estimate the additional impacts on government revenue, Hendren and Sprung-Keyser (2020) sum the impacts across parents, young, and old children. Hendren and Sprung-Keyser (2020) use an estimate of 2.5 children per household (Goering et al. (1999), pg. 32), 33% between the ages of 13-18, and 67% between the ages of 0-12 (Chetty et al. (2016), Table 1); the HUD evaluations focused on one adult per household to measure impacts on earnings. Hendren and Sprung-Keyser (2020) therefore assume one adult per household.

For children: Chetty et al. (2016) find that MTO increases the earnings of children by 31% in their mid-20s for those children who received the experimental voucher between ages 0 and 12 (These children were average 8.2 years old at the time their parents obtained the voucher). Chetty et al. (2016) also find that MTO lowers earnings by 17% for those children who received the experimental voucher between ages 13 and 18 (Those children were on average 15.1 years old at the time their parents obtained the voucher). Extrapolating the observed earnings effects over the lifecycle and applying a 3% discount rate implies an increase in earnings of $89,733 for the 0-12 year old children and a decrease in earnings of $77,028 for the 13-18 year old children. Hendren and Sprung-Keyser (2020) then estimate the fiscal externality during the observed earnings period using the tax rate calculated in Chetty et al. (2016). Hendren and Sprung-Keyser (2020) estimate the fiscal externality during the period of projected earnings using estimates from the Congressional Budget Office. When analyzing the combined sample of all treated children, Hendren and Sprung-Keyser (2020) find that after-tax earnings rise by $27,866 and that tax revenue rises by $7,248 per child. Hendren and Sprung-Keyser (2020) then combine these estimates with the estimated average of 2.5 children per family. Finally, Hendren and Sprung-Keyser (2020) also consider the public cost of changes in college enrollment for these children, which Hendren and Sprung-Keyser (2020) estimate to be -$342 in total due to the fall in college attendance amongst the older children.

For adults: Sanbonmatsu et al. (2011) finds a yearly decrease in earnings over the 10 years after the program of $672.54 (s.e. 716.54), to which Hendren and Sprung-Keyser (2020) apply a tax rate of 30.7%. Sanbonmatsu et al. (2011) also find impacts of $120.29 (s.e. 245.44) and $664.54 (s.e. 335.54) on yearly TANF and SNAP receipt respectively. Hendren and Sprung-Keyser (2020) project these impacts over the lifecycle and then discount back to estimate the fiscal externality for adults.

Combining the fiscal externality from adults with the upfront cost of the counseling of $3,783 and the fiscal externality from children gives a net cost to the government of -$9,215 for the MTO services, with a fairly wide 95% confidence interval of [-42,941,25,773].

Combining the benefits and cost to the government, the estimates imply an infinite MVPF with a 95% CI of [-2.8, \infty]. So, the point estimate implies the policy pays for itself, but the sampling uncertainty means we also cannot reject the hypothesis that the policy does not provide any benefit. Hendren and Sprung-Keyser (2020) also discuss how a modified MTO policy that provided vouchers to families with young children (as opposed to teenagers) has higher returns, with an infinite MVPF and a 95% confidence interval that lies above an MVPF of 1.