Implications of a Medicaid Work Requirement: National Estimates of Potential Coverage Losses

Methods

Classifying Medicaid Enrollees

Our analysis of which enrollees are already working, likely exempt, or not working but subject to new work requirements is based on analysis of the 2017 Annual Social and Economic Supplement (ASEC) to the Current Population Survey. We restricted the analysis to non-elderly adult (age 19-64), non-dual eligible Medicaid enrollees who did not receive Supplemental Security Income. Within this population, we grouped people into the three groups based the following hierarchy:

1. Individuals working full time (defined as ­>35 hours/week) (people working full-time may work multiple jobs)
2. Individuals working part time (defined as <35 hours/week)
3. People in self-reported fair/poor health or those who say the reason they are not working is due to illness or disability
4. People age 60 or older
5. People who are full-time students
6. Parents
7. People who do not fall into one of the previous categories

While specific exemption policies vary across states with proposed or approved work requirements, this hierarchy allows us to group people into general categories of already working (groups 1 and 2); likely exempt due to medical frailty (group 3), age (group 4), school attendance (group 5) or parent caretaking duties (group 6); or not working and subject to work requirements (group 7).

Developing Disenrollment Rates for People Exempt or Already Working

For people already working or likely exempt from work requirements, we developed our “low” and “high” disenrollment rate based on studies examining the effect of administrative requirements in Medicaid on enrollment. Studies focus on a range of settings (e.g., nationwide or a particular state), populations (e.g., children or all enrollees), and time periods and are therefore difficult to summarize in one single estimate. This research generally groups into two categories.

First, some studies examine how many and why Medicaid and or CHIP enrollees lost coverage despite evidence that they remain eligible for the program. Studies that examine the share of beneficiaries who lose coverage despite continued eligibility (implying an administrative issue) or explicitly lose coverage due to not completing necessary paperwork/forms report between 5% (children in Utah¹) and 36% (Commonwealth Care enrollees in Massachusetts²) disenrollment due to administrative barriers. The case study of Utah appears to be a low-end estimate, as other estimates are higher: 13% (children nationwide³), 18% (MassHealth enrollees⁴), and 29% (nonelderly adults nationwide⁵). Examples of studies that look at churn (people losing coverage and then re-enrolling shortly thereafter, implying loss of coverage is related to administrative re-enrollment barriers) find that between 10% (children in California⁶) and 18% (children in Massachusetts⁷ and children in South Carolina⁸) lose coverage and re-enroll. The average administrative burden disenrollment rate across these studies is 18%. Together, this research shows that paperwork or administrative requirements in Medicaid or CHIP lead to between 5% and 30% of people losing coverage.

A second category of studies examine how certain enrollment policy changes (such as adding or dropping reporting requirements) affected enrollment, generally using a pre-post research design. All of the studies we reviewed in this category focus on children and many on the effect of simplification measures implemented following the establishment of CHIP. Again, there is a broad range of estimates of the effect of administrative burden. A study of imposition of new reporting requirements for children’s coverage in Washington found that enrollment dropped 5% following the change, with about half due to incomplete paperwork.⁹ Several reports on the effect of enrollment and recertification simplification in Louisiana’s children’s coverage program report a 16 percentage point drop in loss of coverage due to not returning forms¹⁰ and a 12% drop in rates of churn.¹¹ Similarly, Texas showed an 11% drop in cases closed for incomplete paperwork for children’s coverage.¹² An analysis of the effect of a change in renewal procedures in Florida found that implementation of an active renewal process (from passive renewal) increased risk of disenrollment from the Florida Healthy Kids Program from 1.3% to 22%.¹³ Altogether, these studies indicate that changes in paperwork/reporting requirements in Medicaid/CHIP are associated with change in enrollment between 3 and 20%. Though not directly comparable to the studies outlined above, it is logical that these estimates are lower, as simplification measures in most states do not entirely eliminate reporting requirements. Notably, one study of SCHIP disenrollment found that passive re-enrollment (which requires no action/reporting for most enrollees) nearly eliminated the loss of enrollment that occurred at recertification in other states.¹⁴

Based on this review of literature, we use 5% as a “low” assumption of disenrollment due to paperwork or reporting requirements and 15% as a “high” assumption. These assumptions conservatively use the low-end estimate and mid-point estimate, rather than higher estimates that some studies find.

Developing Disenrollment Rates for People Subject to Work Requirements

For people likely subject to work requirements, we developed our “low” and “high” disenrollment rate based on research examining disenrollment due to the imposition of work requirements in TANF and SNAP as well as state estimates of enrollment effects of imposing work or similar requirements in Medicaid.  The studies we reviewed estimated a broad range of disenrollment rates.

For example, evaluations of work requirements in SNAP showed that enrollment dropped 50-85% among the population subject to the requirement,¹⁵ and CBO’s estimate of the Agriculture and Nutrition Act of 2018 (which would require certain SNAP recipients to be employed or in a state-government-sponsored training program unless they qualify for certain waivers) assumes that 24% of people potentially subject to the requirement would no longer receive benefits under the proposal.¹⁶ Preliminary evaluations of Medicaid Section 1115 waivers in Indiana, Iowa, and Michigan showed a range of participation in complex new programs such as payment of premiums or participation in healthy behavior programs. About half of HIP 2.0 enrollees in Indiana required to pay premiums did not do so, between 83 and 95% of enrollees in Iowa failed to complete all required healthy behavior actions, and between 60-70% of enrollees in Michigan were unaware of or misinformed about most cost-sharing and healthy behavior provisions¹⁷ (though retaining coverage in Iowa and Michigan was not contingent on completing these activities). Last, of states that have submitted Section 1115 waivers to use work requirements in Medicaid, Indiana’s provides the most detail on expected participation/enrollment changes. It assumes that 25% of people subject to the work requirement lose enrollment. Based on this information, we assumed a “low” disenrollment rate of 25% and a “high” rate of 50% among people likely not exempt from work requirements.

Estimating Disenrollment

To estimate disenrollment, we applied our “low” and “high” disenrollment rates for each group, for a total of four possible scenarios. We estimate disenrollment in the aggregate and do not model a particular individual’s likelihood of remaining enrolled in Medicaid. Estimates are intended to provide illustrative examples of the potential scope of work requirements nationwide; actual experiences of a particular state will likely differ due to unique nature of each waiver, the characteristics of the state’s Medicaid population, and the local/state economy.