Yin (2026) uses a Callaway & Sant'Anna staggered difference-in-differences estimator on state-year
ABLE program launches from 2016 through 2026, with four never-adopter control states (North Dakota,
South Dakota, Wisconsin, Nevada). Each panel below shows the event-study coefficient for a different
outcome, five years before and five years after each state's ABLE program launch. Pre-trends are
statistically indistinguishable from zero across all outcomes.
+1.5 to 3.3%
Disposable income for adults with disabilities in states that launched an ABLE program
Yin (2026) draft 2026-07-07
+7.7%
Wage income for adults with disabilities after ABLE program launch
Yin (2026) draft 2026-07-07
+0.7 pp
Employment rate for adults with disabilities after ABLE program launch
Yin (2026) draft 2026-07-07
3×
Effect magnitude among young adults who satisfy the pre-age-26 onset rule automatically
Yin (2026) draft 2026-07-07
Event study — Log disposable income
Callaway & Sant'Anna event-time coefficients, adults with disabilities. Event time 0 is the year of the state's ABLE program launch. Year t = -1 is the reference period.
Bars are 95% confidence intervals. Pre-treatment coefficients (t = -5 through t = -2) are within one standard error of zero for the disposable-income outcome. Data: Yin (2026), 2026-07-07 draft, figures/final/fig1_event_data.csv and fig3_event_data.csv.
How to read this dashboard
Event time. The horizontal axis is event time — the number of years relative to each state's ABLE program launch. A state that launched in 2016 has t = 0 in 2016, t = +1 in 2017, and so on. A state that launched in 2020 has t = 0 in 2020, t = +1 in 2021.
Coefficients. Each dot shows the average treatment effect on the treated (ATT) at that event time, estimated using the Callaway & Sant'Anna (2021) doubly-robust estimator with never-adopter control states.
Interpretation. Pre-trend coefficients that hover around zero support the parallel-trends assumption. Post-launch coefficients that rise (for disposable income) or fall (for transfer share) show the causal effect of ABLE program availability on financial outcomes.
Triple difference (Disability × ABLE × Post)
Comparing adults with disabilities to adults without disabilities in the same state, before and after ABLE launch. Isolates the ABLE-specific effect on the target population.
Log disposable income
+1.05%
Disability-specific effect on log disposable income for adults with disabilities relative to adults without disabilities in ABLE states, post launch.
SE 0.0049 · p < 0.05 **
Transfer share of income
-0.52 pp
Reduction in transfer-share dependence among adults with disabilities relative to adults without disabilities in the same states.
SE 0.0029 · p < 0.10 *
Aggregate transfer receipts
-$0.13B
Per-state-year reduction in aggregate transfer receipts for adults with disabilities in ABLE states.
SE $0.06B · p < 0.05 **
The triple-difference specification isolates the disability-specific response to ABLE by netting out any coincident state-year shocks that affected adults without disabilities in the same states. Consistent with substitution from safety-net transfers toward accumulated assets and higher labor-market income. N = 2,448 state-year × disability cells. Yin (2026) Table 2.
ABLE and Medicaid expansion are substitutes
ABLE program effect on log disposable income for adults with disabilities, by Medicaid expansion status. Interaction test.
The ABLE effect is largest in non-Medicaid-expansion states (+2.95%, p < 0.01), and is offset by roughly two-thirds in Medicaid-expansion states (-2.36%, p < 0.10 on the interaction). ABLE and Medicaid expansion both relax asset-related constraints for adults with disabilities; where one is present, the other's marginal effect is smaller. Data: Yin (2026), N = 1,632.
REACHABLE is built by the RISEI Lab at Northwestern University with support from the U.S. Department of Health and Human Services