The impact of subsidized transit passes on health and well-being for people with low incomes

The FFPT program is a collaboration between Metro, Sound Transit, and social service agencies who assist with enrollment. As the region’s largest public transit agency, Metro leads the program’s design, implementation, and evaluation of FFPT. Sound Transit, the region’s second largest transit agency, has partnered with Metro on FFPT as a two-year pilot. This allows people who hold the FFPT annual pass to, at no cost, ride on and transfer between all of the two agencies’ services across an extensive regional transit network, primarily Metro’s bus services and Sound Transit’s light rail services.

The enrollment process for FFPT operates through partner social service agencies. The initial roll-out of FFPT is designed to meet capacity constraints of the enrollment agencies, with most people learning about and enrolling in FFPT when they seek other State benefits. Due to COVID-19, most eligible individuals enroll over the phone by calling DSHS or Public Health - Seattle & King County. Eligible individuals can also enroll in person at Public Health offices. The enrollment agencies verify clients' identity, income, and participation in one of the qualifying cash-benefit programs; enter clients’ information into Metro’s FFPT program registry; and provide them with the FFPT pass immediately in-person or through mail. As of mid-March 2021, about 4,000 people (less than 4% of total eligible participants) have been enrolled in FFPT.

The FFPT pass is valid for one year from the date of issuance, regardless of whether the holder’s income or eligibility change. After one year, the pass holder must re-enroll and meet eligibility criteria again to continue to access FFPT. The FFPT pass is visually indistinguishable from the region’s other transit passes (known as ORCA passes) that are commonly used by residents who ride public transit.

2021-10-01

2023-10-01

Number of boardings on public transit at agencies participating in the ORCA LIFT program
Total trips by any mode
Trips disaggregated by travel mode
Payment method
Trip purpose
Travel time

Hours worked per quarter
Trips for employment
Earnings
wage rate
being employed
job exits
job starts
job-to-job moves
Presence of a credit score
credit score
credit delinquencies

Psychological distress
Self-reported physical health

Annual preventive care visit
Trips for healthcare purposes
Substance use disorder treatment
Mental health treatment
Prescription fills
Vaccinations
Hospitalizations
Emergency Department
Outpatient visits
Dual eligibility (Medicare AND Medicaid)
Costs of care

Enrolled in SNAP
Trips for social services
Enrolled in different public benefit programs operated by ESA(TANF, etc.)

Housing moves
having a formal address
travel time from home to downtown
neighborhood characteristics
number of recent days spent unhoused
eviction filings
homelessness program use

Loneliness scale
Trips to see family and friends

How FFPT is being implemented in the program’s first year creates an opportunity to leverage an RCT and difference-in-differences analysis to generate measures of the impacts of FFPT with strong internal and external validity. Our study sample will include three groups of people, all of whom have incomes ≤80% FPL:

Group A: Eligible – Treated
Group B1: Ineligible – Treated
Group B2: Ineligible – Control

In our study sample, Group A is comprised of people in the LIFT registry who are eligible for FFPT and who, after the baseline survey, are immediately connected for enrollment in FFPT.

Group B1 is comprised of people in the LIFT registry who are ineligible for FFPT because they are not enrolled in one of the six State cash benefit programs, who are screened for comparability with Group A, and, after the baseline survey, are randomly assigned to immediate referral for enrollment in FFPT.

Group B2 is the same as Group B1, except that they are randomly assigned to not receive FFPT and remain enrolled in the ORCA LIFT discounted fare
program.

An RCT (B1 vs. B2) allows us to make an internally valid comparison of the impacts of FFPT on people with incomes ≤80% FPL but who are not receiving one of the six State cash benefit programs that make one eligible for FFPT. This directly informs decisions policymakers face about potential expansion of the FFPT eligibility criteria. A difference-in-differences analysis comparing people who are currently eligible for FFPT (Group A) with the control group from the RCT (B2) provides policymakers with relevant information regarding the external validity of the RCT results to people currently receiving FFPT, thereby informing decisions regarding both continuation and expansion of FFPT as well as decisions by other transit agencies in the region about joining FFPT.

Not available

Computer-based randomization to determine placement in group B1 vs. group B2.

individual

No

3,225 households initially; 2,250 households expected in final survey

3,225 households initially; 2,250 households expected in final survey

Group A: Eligible – Treated (1,075 households initially; 750 households expected in final survey)
Group B1: Ineligible – Treated (1,075 households initially; 750 households expected in final survey)
Group B2: Ineligible – Control (1,075 households initially; 750 households expected in final survey)

For the RCT, a straightforward power calculation indicates that we will have sufficient power to detect a reasonable improvement in psychological distress. We expect to have 1500 people in the follow-on survey who are not eligible for FFPT under current eligibility rules and who are randomly assigned at the end of the baseline survey into equal groups who are and are not referred for enrollment in FFPT (Groups B1 & B2). For a two-sided test with a level of 5% and 80% power, we will be able to detect any ITT effect larger than 0.14σ in both the survey and administrative data. Because referral occurs immediately, we expect take-up of FFPT to be high, but it may be imperfect to the extent that some people may self-report income that then is not verified at ≤80% FPL. In a separate project, only 3% of applicants for emergency financial assistance reported income to King County that made them appear eligible when they were not. Assuming 90% take-up we can measure a 0.16σ TOT effect. A similar 0.15σ ITT effect (0.21 TOT effect) on the K6 index of psychological distress is demonstrated in Gubits, et al.32 , measuring the impact of long-term vs. short-term housing assistance. The difference-in-differences analysis will have similar power for survey measures and better power for administrative records. Both the treatment group (A) and the control group (B2) will include 750 people, or 1500 total. Power calculations for the difference-in-differences analysis are therefore similar to those calculated above for the RCT. For outcomes from administrative records, however, we gain higher power if take-up of the FFPT program is high among people outside those who we include in our survey sample. We estimate that 18% of those in the LIFT registry – about 5,800 people – are eligible for FFPT, far more than we will survey. Although take-up is difficult to estimate, if more than 25% of these people enroll in FFPT, then using this larger sample will improve power for administrative records.

Pre-Analysis+Plan+-+King+County+Metro+Fully-Subsidized+Annual+Pass.pdf