Cost of adding insurance (marginal vs inframarginal)

How much does it cost to add one more person to an insurance pool?  That is a damn good policy analysis question that implicitly partially motivates my dissertation plus several other papers I have under review or in development.

TLDR: “WELL IT DEPENDS!!!!”

The biggest thing that it depends on is how much of the new policy is getting eaten up by folks who already would have bought the policy anyways under the old policy regime.  This is known as “inframarginal” consumption.  The cost of a new enrollment is the combination of the cost of actually getting someone who has not bought to change their behavior  which is the cost on the margin plus the costs consumed by people who already bought.  Well targeted policies are by an efficiency definition policies that have most of their efforts happening on the margin with very little, if any effort, being consumed by the inframarginal buyers.  Exceedingly well targeted policies should be just enough to change an individual’s behavior, no more nor less.

In the context of the ACA, there are two big thoughts on how to increase ACA enrollment.  Increasing subsidies is the federal policy response since 2018.  It was implicit in the second half of the Trump Administration with Silverloading.  It is the explicit policy of the Biden Administration with the passage of the American Rescue Plan Act (ARPA) enhanced subsidies.  Low or no cost plans are more likely to be purchased by people who were flipping a coin on whether or not they are going to buy health insurances than medium or high cost plans.  Dropping the net of subsidy premium to zero or near zero should increase enrollment, and that new enrollment should be comparatively healthier than average.  However,  these subsidies apply to everyone.  The premium drop applies to someone who would have signed up on the first minute of Open Enrollment under the old subsidy regime.  The premium drop applies to someone who signed up with a minute left in the Open Enrollment Period because they figure paying $1.03 to quiet their nagging mother is worth it even if they don’t ever intend to use the insurance.  Most of the effect is likely a transfer to folks who were inframarginal anyways.  Now inframarginal transfers can be valid policy objectives but it is a different policy objective than targeting efficiency.

The other big thought is that money is important but there are lots of non-monetary things that are barriers to enrollment to.  Most of these things get lumped (in my conceptualization), as “administrative burdens” or “targeting ordeals.”  Here is where providing information or improving defaults or linking state SNAP and MEDICIAD  and tax records to Exchange enrollment files would matter.  The research shows that these effects are almost entirely on the margin and pull in pretty healthy populations that may drive down average premium at fairly low costs to the state.

I’ve been thinking about this a lot when I read a new paper in Health Economics by Jessica Brown on the effect of advertising  on the purchase of long-term care insurance.  LTC is a private insurance that pays for nursing home and home care for individuals whose health declines substantially at some point in the future.  This market competes with and is supplemented by Medicaid.  LTC coverage rates are fairly low even among people who can afford the premiums.  Over most of a decade, there was an advertising campaign that Brown analyzed and found some interesting results:

Using data from the Health and Retirement Study and a difference-in-differences framework, I find that OYF increased long-term care insurance coverage in participating states by one percentage point. The increase occurred primarily for higher-asset individuals, with individuals in the top 20% of the asset distribution increasing coverage rates by four percentage points, or 17%. This increase is only slightly attenuated when controlling for the roll-out of another LTCI program with similar timing, the Partnership for Long-Term Care (PLTC)….

Using a back-of-the-envelope calculation, I estimate that the increase in LTCI will save Medicaid $483 million in present value, or $27 per letter sent. Compared to the impact of other government interventions, the OYF campaign was at least as effective and less costly than the PLTC program and raised coverage by about one-third of an average tax subsidy at a fraction of the cost

The provision of information acts almost entirely on the margin so it is an efficient targeting mechanism to prompt a moment of attention from people, who if they’re thinking about LTC are likely to want LTC, to buy LTC insurance. The outreach was mostly by letters by the state which meant the state had a pretty good idea that they were hitting up people who did not have LTC. Television advertising would have more inframarginal hits as someone watching a golf tournament on TV could plausibly already have LTC so in that case, the ad would not be relevant.

We should be thinking about how targeting policies work and if we are concerned about efficiency as a primary goal, targeting policies that are mostly landing on the marginal buyer instead of the inframarginal should be more likely to be adopted.