David Anderson

This is a post on Medicaid work reporting requirements. Soon individuals who qualify for Medicaid through the ACA expansion will need to report 80 hours per month of work, community engagement or approved other activities. That will be, given what we know from Arkansas, Georgia and New Hampshire, an omni-shambles.

My elder teen (as I now have a pair of teenagers), recently started their first job at a fast food restaurant. They like it. They love having disposable income where they don’t need to get Mom or Dad to quasi-approve/consent to their spending. If they were on their own and not on their parents’ insurance, she would be making enough to qualify for Medicaid Expansion.

Last week, they were scheduled to work 23 hours. However, the restaurant was slow so she got cut twice two hours before the end of her scheduled shift. No big deal for her, as it merely delayed the purchase of an ultra cool pair of Doc Martens by a pay period. But in 18 months, this could be a huge deal for folks who need to hit 80 hours of work per month.

In that scenario, they prospectively would have believed that they were hitting their work hour pace on Sunday when the schedule was released but by the following Saturday, they were in the hole through no fault of their own. This will be a coercive element of low income labor control.

Now a call for all academic and wonky Jackals — the American Time Use Survey has a couple of questions that sort of get to this in hours actually worked and hours available for work but that is not quite the question I want to answer.

Does anyone have a data set that has scheduled hours versus actual hours where there is substantial variability?

Prospective and Retrospective Work Hour VolatilityPost + Comments (20)

I am not a lawyer, but this lawsuit in Illinois is fascinating:

A rapidly growing line of ERISA cases seeks to impose fiduciary standards of conduct—developed by courts largely in the retirement plan context—to health plan design choices.  The most recent, Barbich et al. v. Northwestern University et al., No. 1:25-cv-06849 (N.D. Ill.) filed on June 20, 2025, involves a new and potentially disruptive twist: plaintiffs allege that the university violated ERISA fiduciary duties by offering an allegedly imprudent health plan option…

laintiffs claim that Northwestern University improperly included a higher-premium, lower-deductible PPO alongside a lower-premium, higher-deductible PPO, arguing that the higher-premium option provided no meaningful financial advantage for participants.

I have a few publications on choice domination in the ACA marketplaces, and there is a rich literature on dominated choice in the employer sponsored insurance space.  Dominated choices occur when on all relevant factors of decision making, Plan A is never worse than Plan B and on some of those factors, Plan A is better.  My work in the ACA showed that avoiding dominated choices in the ACA markets during the era of enhanced subsidies would lead to substantial savings:

5.8% of 748 087 California marketplace enrollees currently default to dominated health care plans with higher premiums and cost sharing; more than 98.0% of enrollees have low incomes. By comparison, an alternative smart default system would default many enrollees to more generous plans with approximately $100 lower monthly premiums and almost $2000 lower deductibles.

These types of choice errors are expensive.  CMS under the Biden administration improved defaults for low income enrollees in 2024 and 2025.  The recently passed Republican reconciliation bill forbids keeping people out of bad plan choices.

Now this lawsuit is pushing the argument that employers have to avoid offering dominated plans.  I am again NOT A LAWYER, but I think this is a reasonable argument to make as the employers are acting as an agent for the employees and offering obviously bad things is probably a bad thing.

 

ERISA and DominationPost + Comments (15)

The Affordable Care Act individual health insurance markets will be more expensive and more burdensome to use in 2026 but the fundamental structure of these markets will  remain.

Let’s talk about more expensive first.

The enhanced premium tax credits that were passed in March 2021 and renewed in August 2022 are scheduled to expire on December 31, 2025.  These credits did two things.  First, for individuals with incomes over 400% Federal Poverty Level (~$62,000 for a single individual and $128K for a family of four in the contigous 48 states) premium tax credits WILL NOT BE AVAILABLE.  Under current policy in 2025, an individual with an income over 400% FPL will “only” pay 8.5% of their Modified Adjusted Gross Income for the premium of the 2nd least expensive Silver plan.  Without subsidies, and depending on age/county the cheapest plans could be 10% to 25% of gross income for this group.

The other big change is for individuals with incomes between 100-400% FPL ($15,500-$62000-ish).  The personal contribution amount for the benchmark plan will increase by several percentage points of income.  Under current policy, individuals with incomes under 150% FPL, the benchmark plan (a Silver plan with extra deductible help that reduces cost-sharing substantially) has a zero dollar expected personal contribution.  Zero premium gold and bronze plans are often available to 250% to 300% FPL.  Next year, someone making 150% FPL (~$23,000ish) will be paying about $60 per month (~4% of income) for that same benchmark plan.

There are a couple of other changes, we can assume the cheapest plans will have higher deductibles as the calculation used to determine the maximum allowable cost-sharing changed and we can also anticipate that the Silver plans will be slightly less generous in their coverage due to a regulatory change.

Burdensome

The big change in a recent rule that was finalized is a lot more administrative burden.  For people who are currently enrolled and want to maintain their coverage in 2026, you need to go in and validate/update your estimated income for 2025 after August in order to continue to receive premium tax credits in 2026.  You will also need to make sure to file your taxes and reconcile any previously received tax credits.  During the Open Enrollment Period (November through December,) please go back into your Healthcare.gov account and MAKE AN ACTIVE CHOICE even if the choice is to keep exactly what you have now.  Passive choice is being strongly discouraged and frictional.

There will be a lot more double checking and validation of income throughout the entire process.  A lot more paperwork will be required.

New enrollees will need to have their incomes validated before they can receive premium tax credits.

There are other things happening, but the fundamental outline of the ACA markets will look like it was in 2020 before COVID in 2026 rather than fundamentally transformed.

The ACA in 2026Post + Comments (13)

Yesterday at 1433 local time, my co-author S and I killed a project.  It was a great question.   We just can’t answer the question with rigor.  I’m not sure if we’re killing the question entirely, or making it a zombie idea that may be resurrected if/when we ever see a good enough data set.

This is part of the scientific process. A lot of good ideas are just not feasible with available resources.  The big achievement we had is that we killed the idea after only 30 to 40 hours worth of work.  Five years ago, I would have burned 400 hours on this project.

I want to talk through this process in the hope that a Jackal can give us a saving throw.

Question:  Does being a member of a sexual minority lead to differential uptake of Medicaid after a state expands Medicaid compared to similarly situated individuals who are not sexual minorities?

Theoretical justification:  We know that marginalized folks are often less likely to have positive interactions with complex governmental bureaucracies as the system is often designed to screw them over.  Medicaid is a complex system.  However sexual minorities tend to be younger/poorer so they are in the sweet spot for Medicaid Expansion.  This could go either way and either answer is inherently interesting.

Key Data elements:  State-year information, Medicaid Expansion dates, insurance status indicators (granular Medicaid is best), sexual minority question(s)

Want to have data elements:  Rich suite of demographic covariates, local geography, interview date to get detailed mid-year expansion identification right, time varying local geography covariates

Approach:  Triple difference (Sexual Minority X Medicaid Expansion State X Post-Expansion Time) with individual (if possible), time and geography fixed effects

We will drop the states that expanded Medicaid on 1/1/14 as they are an always treated group.  The staggered comparison group is the states that never expanded Medicaid or at least never expanded Medicaid during the study period.

Data Sets (and their problems):  

• AD Health survey (power)
• BRFFS (power, insurance definition, missingness, 2015 questions lacking)
• CENSUS PULSE (power and granularity)

We thought BRFFS would have been useful but there was just not a lot we could do with the data that was available both with and without weights.

This is the research process.  We love the idea but when we asked each other that given what we had discovered about the data set and without knowing the answer of the analytical runs, would we actually trust anything from the data?  And we could not get to an enthusiastic yes.  So this idea is going to take a long ass nap unless someone has a better approach in mind.

 

 

 

Sending a good idea to the big farm upstatePost + Comments (24)

Individual Choice Health Reimbursement Accounts (ICHRA) are basically a defined contribution model for employer sponsored insurance. Employers give groups of employees a pre-defined lump sum and the individuals then buy ACA plans. The new legislation signed yesterday widens the possibility space as it is now a creature of law rather than regulation. The big argument for ICHRA is for employers it gives cost predictability and for individuals it breaks the principal agent problem of HR buying way too much network/coverage than most people actually want.

It is predicated on the ACA individual markets being cost-competitive to cost-dominant over small group fully insured markets.

These estimates also assume average unsubsidized premiums only go up 4.3% in 2026.

So far the average preliminary rate hikes are averaging nearly 19%. 👀

— Charles Gaba ✡️ (@charlesgaba.com) July 5, 2025 at 9:56 AM

How does this scheme work in an ACA market that everyone is predicting will be much smaller, much more morbid and much more expensive?

I’m stuck at the moment.

How does ICHRA work going forward?Post + Comments (50)