Medicaid Expansion Opt-Out: Researchers, Prepare Your Difference-in-Difference Models

July 4th, 2012 at 1:47 pm

I don’t have anything good to say about the opt-out provision that the SCOTUS added to the Medicaid expansion part of the health care law.  There’s lots of hypothesizing about which governors will opt out, despite the strong monetary incentives not too.  But you can’t figure this one out based on a rational balance sheet with costs on one side and benefits on the other.  Ideology is in this house, big time, and that can mess up the math.  The bottom line is that a lot of poor people who would have gained coverage under the law may well end up without it now, at least for a while.

Anyway, here are a few thoughts and observations, including some interesting questions that might be answered by the natural experiment created by the opt-out.

–Opting out isn’t forever. A state that opts out starting in 2014 could easily come back in later.  Especially if the electorate decides to “opt out” of the governor who made the call.  As Merrill Goozner pointed out right after the court’s decision:

Every time Medicaid expanded in the ensuing decades – such as the extension of coverage to all uninsured children in the 1990s – the federal government allowed states to opt out. Eventually, none did.

–The savings to opting out, slight already, are smaller than they seem. The feds pay 100% of the costs of the Medicaid expansion for the first three years that it is in place, but after that, fed support starts to decline, ending up at 90% by 2020.  That’s still a very high federal match rate compared to the regular Medicaid program.  Even so, some governors are arguing that their state can’t afford the expansion.

But, as my CBPP colleague Judy Solomon wrote the other day:

[According to CBO,] the additional cost to the states represents only a 2.8 percent increase in what states would have spent on Medicaid from 2014 to 2022 in the absence of health reform.

Moreover, this 2.8 percent figure overstates the net impact on state budgets because it does not reflect the savings that state and local governments will realize in health-care costs for the uninsured. [my bold]

That is, for states that opt-out, they have to net the cost of uncompensated care out of whatever savings they achieve by opting out.  And they will find these costs have gone up.  Right now, hospitals that treat a lot of uninsured poor people get a chunk of dough from the feds to offset those costs.  But under the assumption that the new law would provide these folks with coverage, the ACA reduces those offsets, and that piece of the law remains in place.  All of which means that there are hospitals out there that agreed to lose funding they currently get to treat the uninsured poor only because they believed those poor patients would no longer be uninsured.  In states that opt out, these hospitals, their patients, and importantly, the state’s citizens—because someone’s got to pay for this—are now big losers.

–Natural experiment: OK, this part is where we employ: “if life sends you lemons, make lemonade.”  Opting out states create a natural experiment of the type that’s rare in economics.   Come 2014 (barring “repeal and replace”, of course), there will be states where poor, non-elderly adults are uncovered by Medicaid next to states where they are covered.  This difference exists already, btw, because states vary a great deal on this parameter.  But they’ll be a lot more coverage than there is today.  Moreover, there’s a strong incentive for these poor uninsured in opt-out states to increase their earnings, so at least they can hit the subsidy range that starts at 100% of poverty (the WaPo has a helpful graphic on this).

So, researchers can compare various outcomes, controlling for economic conditions that are presumably similar across neighboring states.  What is the impact on state budgets?  On the health of the poor?  On health care expenditures?  On the labor supply/work effort of the affected population?  On the educational outcomes of kids in their families?  On economic growth in the state?

This is the much vaunted “difference-in-difference” model where you back out the impact of a policy intervention by comparing places that are as alike as they can be in every way except that in one place the intervention is “turned on” and in the other, it isn’t.

Forgive me, fellow academics, but I sincerely hope none of this research can move forward because the governors all come to their senses.  But I fear that such questions will, in fact, be answerable.

Print Friendly, PDF & Email

2 comments in reply to "Medicaid Expansion Opt-Out: Researchers, Prepare Your Difference-in-Difference Models"

  1. fausto412 says:

    I hope they opt out. Experiments like these are the kind that drive home the point of what works and what doesn’t. Voters elected idiotic officials…vote them out. But in the mean time I want the experiment to play out.

  2. Th says:

    Has anyone tried to determine if the 2.8% increase in Medicaid expenses is covered by increased state taxes due to the increased economic activity?