MLR rebates, negative premiums and option values

Some people have been effectively paid to buy health insurance on the ACA marketplaces.

What do I mean by this?

The ACA requires individual market insurers to spend 80% of qualified premiums on qualified claims or quality improvement expenses.  If, over a 3 year period, an insurer spends less than this amount, they have to write a check to the policy holders of the last year that reduces premium so that the medical loss ratio (MLR) is updated to 80%.  The check is usually written twenty months after a policy is purchased (the 2021 checks will be written August/September 2022 for instance).  The size of the check is a function of the proportion of gross premiums a policy holder pays.  This means that people with higher value plans and who are older and from larger families get larger checks than younger families with skinnier policies.

Note the important phrase above is “GROSS PREMIUM”.

Many ACA buyers are subsidized. They pay NET PREMIUMS that are often much less than the gross premium.  Some people pay nothing in monthly premiums.  If they purchase an ACA plan that generates an MLR rebate for that policy year, they are effectively getting paid to buy insurance.

Some MLR rebates are fairly idiosyncratic.  An insurer might have a shock year like 2018 where every insurer massively overpriced due to mis-estimating how many people would be chased out of the marketplaces due to political uncertainty.  But before and after that one off shock year, normal-ish years resumed so the MLR rebate is not particularly predictable nor anticipatable when plans are purchased.  Some MLR rebates are somewhat predictable ahead of time. Insurers could have a big shock ear like 2018 and then in 2019 and 2020 have very profitable years as they still could have high premiums relative to claims and low competition.

Can we think of buying an ACA plan as buying both health insurance and also buying an option on an MLR rebate?

There is some probability that at the point of purchase that an MLR rebate will be generated for a particular plan.  The probability is seldom 100% but in many areas for certain plans the probability is well above 0%.  There is variance on the size of an MLR rebate ranging from zero to a lot.   Some people have an expected MLR rebate of effectively zero.  Some people have an expected MLR rebate that could be quite large ($1,000 +).

Does this drive decision-making?

Do people realize that when they purchase ACA health insurance plans, there is a possibility that their plan choices could lead to a potentially big MLR rebate check.  Do people who are are exposed to a higher probability of a big pay-off choose differently which plans they are buying to reflect the lagged effects of an MLR rebate on premiums.  Do they buy a plan that has a high probability of an MLR rebate that has higher initial premiums than they otherwise would have if they were not taking into account MLR rebate probabilities?

This is a messy problem.  It requires knowledge of the MLR mechanics and history.  It involves long lags.  It involves optimal rational maximization problem solving under many axis of uncertainty.  It is a nasty problem.

My hypothesis is we should not see effects due to informational and behavioral frictions.  But I’m curious about this.