Medicaid Work Requirement Policy: Politics Aside
The statistical type I and type II errors clash again with Medicaid brought under close scrutiny as the Trump Administration looks to trim spending. Here is the issue:
The Trump Administration is looking to eliminate possible freeriding on the very generous Medicaid by able bodied non-elderly members of the population who are unwilling to work, by allowing states to attach work requirements as a condition of receiving aid. The current law prohibits this but states can apply for and be granted a waiver to do this by the Administration.
The opposition to this largely on the Democrats side of the aisle argues that this goes against the intent of Medicaid as the objective of Medicaid is to provide full healthcare coverage to the poor, disabled, and elderly populations. The Obama administration in the past had rejected waiver applications from states seeking to attach a work requirement.
So where do the type I and type II errors fit in here and how might we resolve this conflict in an apolitical sense?
A type I error is committed when we reject a true or correct proposition. A type II error is committed when we fail to reject a wrong proposition. These are possible errors that might be committed in an environment of uncertainty.
The provision of Medicaid can be justified as some form of social contract in which because of uncertainty about future states of the world society agrees to provide healthcare for any individuals who draw from the lottery of life the short end of the stick. Well, so far so good. A potential problem however arises when in the real world such draws involve adverse selection and moral hazard:
Work involves effort which may be distasteful to some able bodied, non disabled, non-elderly adults. If healthcare can be obtained while choosing not to work then it may be attractive for these individuals to rather be unemployed and benefitting than choosing to work. This is an example of adverse selection.
With moral hazard, we may have the case for some who genuinely qualified for Medicaid but have found it more convenient to remain on Medicaid even when the main reason for having qualified has either been addressed or if they are able to get off it through gainful employment.
So the analogue of the type I error is when the design of Medicaid is punitive for those who would genuinely qualify because it tries to eliminate adverse selection and moral hazard. The analogue of the type II error, on the other hand, is when the design of Medicaid allows for adverse selection and moral hazard to exist because it is careful in making sure that those who genuinely qualify are not inconvenienced. In statistical theory reducing the one error increases the likelihood of committing the other error.
We can now see that the Trump administration policy seeks to reduce, in this case, the probability of committing the type II error. The opposition is seen as coming to the defense of those who genuinely qualify as such a policy affects them punitively by increasing the probability of committing the type I error.
What is the appropriate policy? A study carried out by the Kaiser Family Foundation finds that for non-SSI non-elderly Medicaid enrollees in 2016, reasons for enrolling are: 36% report illness or being disabled, 30% report taking care of home or family, 15% report going to school, 9% report retirement, 6% report not being able to find work, and 3% provide some other reason. This totals 9.8 million people.
The question then is how do we treat these explanation categories and weight the magnitudes within each category? Is it OK for people to choose not to work and still get free and fairly generous healthcare? Should society be responsible for supporting private personal choices to do something else other than work? How have we defined "disability" such that a large percent (36%) are stating this as a reason when we do have in place supplemental social insurance (SSI) that should address this?
This is what might be fair. First, the Administration might want to revisit the criteria being used to qualify for SSI and to see whether it is unduly restrictive and to relax it. If the finding however is that it is not restrictive, then there might be no clear reason for the disability claims being made by some of the respondents, especially if long term.
Second, the work requirement appears to be an incentive driven policy, but it should come with a few more time based preclusions in addition to the standard ones (the waivers already preclude the elderly, pregnant women, those with disability and children from being subject to the work requirement). Anyone who is not working, not covered by SSI, and who does not meet the other specified preclusions should be covered by Medicaid for a specified period of time, say three to six months, after which a work requirement or community engagement of some sort should be required.
So I would think that the states that feel the need to, should apply for such waivers but should incorporate such time-based provisions so that there is both concern for those who genuinely need the assistance but at the same time a nudge to those who can to find work or ensure that they qualify formally for SSI and be covered, if they fall into that category. It is also suggested that the definition of work be expanded to include taking care of home or family, as this is work that economists consider part of output that is not reflected in GDP because it is usually unpaid.
As much as there is the general partisan outcry that such a requirement is unnecessary, will increase joblessness, raise healthcare costs etc. it is not at all clear in a big picture sense that society is made worse off by a policy measure that comes at little administrative cost, is flexible enough to interpret work in a broad sense, provides an adjustment period for all potential recipients and incentivizes people to choose productive work effort. This is perhaps a happy medium that should appeal to all policymakers.