I think that these value proposition strategies are limited in two ways. First they are expensive. Increasing premium subsidies either directly like in the ARA or indirectly by increasing the actual AV of plans means most of the benefits go to people who already would have signed up. We see this in Silverloading. Secondly, they fundamentally require significant attention. The work that Coleman Drake, Sih Ting Cai, Dan Sacks and I have done on zero-premium plans shows no notable differences in plan selection for people who just were exposed to zero premium plans and people who were not exposed to zero premiums plans by only a few bucks. If we want to discount that work, we can also look at Medicaid where the coverage is by default zero premium but there are large swarthes of people who are eligible but not enrolled in all states. Zero premium and coverage dominates zero premium and no coverage. Yet we have observed massive empirical evidence of millions of people in a dominated choice state. This is weird.

The other approach that can plausibly increase enrollment is a behavioral economics and public administration concept of ordeals and administrative burdens. Coleman, Sih-Ting, Dan and I found that the act of not having to pay a premium to initiate coverage led to a 30-60 day gain in days covered the following year compared to folks who just had to pay a nominal premium. These gains were concentrated in new buyers and individuals earning under 200% Federal Poverty Level. A new paper using Covered California data by Domurat, Monashe and Yin found that sending customized letters to individuals who were likely to be eligible for low cost plans but who had not signed up was worth the equivilent bumping up subsidies by $25-$53 per member per month. The people who signed up in response to these letters tended to be both young and relatively low cost to cover which lowered overall premiums. Goldin, Lurie and McCubbin in a soon to be seminal health economics paper found that an IRS notification letter to families that had paid the mandate penalty outlining affordable options led to both significant enrollment and significant detectable lower death rates for people who received the letters. Cheap letters overwhelmingly performed better than massive theoretical outlays.

In a just released working paper, Shepard and Wagner looked at the impact of automatic enrollment policies in Massachusetts. The set up was that to enroll in a policy, people had to complete a two-step process. The first was to fill out a demographic and financial information form. After that was accepted, the individual would need to pick a plan. There was significant drop-off between the number of people who filled out the information form and the much smaller number of people who picked a plan. Massachusetts had a variety of policies that at times placed people who had filled out the form but did not choose a plan into plans and at other times, the Commonwealth left these people adrift.

We find a 32.6% decline in new enrollment when auto-enrollment ends in 2010….
we see no evidence of an uptick in active enrollment after the policy change – something we would expect if passive enrollees were deliberately choosing passivity because they know they will be auto-enrolled. Based on this evidence, auto-enrollment does not appear to crowd out active choice; instead, under a non-enrollment default, passive individuals simply fail to take up health insurance. Using a simple model of the dynamics of participation, we estimate that the 48% persistent increase in the flow of new enrollees translates to 32% higher enrollment in steady state…

On the one hand, we find that passive enrollees do have characteristics consistent with lower demand for (i.e., expected value of) health insurance. Demographically, they are younger (by 4 years on average) and healthier (e.g., 34% less likely to be chronically ill), with a particularly large share of young men age 19-34 (a group sometimes called “young invincibles”)… Consistent with their better health, passive enrollees incur 44% lower average medical spending per month enrolled. Similarly, their shorter durations imply fewer months of state-funded subsidies.

So Shepherd/Wagner find that automatic enrollment of people who have indicated interest in getting covered because they had filled out the first stage form but failed to complete the enrollment process picks up a lot of very cheap to insure individuals. These folks get real insurance value as this population has a disproroprtionate share of healthcare spending that is predicated on the phrase “Hold my beer and watch this….”

More importantly, we have a growing body of evidence in healthcare and other fields where seemingly small frictions accumulate non-cash costs that keep people from being able to access benefits and programs that they objectively qualify for.

Finding ways to lower administrative frictions and burdens by smoothing out rough edges and applying lubrication to the squeaky parts of application processes will be a significant policy pathway for state and federal entities that are seeking to enroll more people in health coverage. I think that this pathway, at this time, is likely to be far more productive and efficient use of taxpayer dollars than throwing money at the problem to improve the value proposition as improving the administrative experience by allowing individuals to pay premiums through their tax refunds or request automatic enrollment in a zero premium plan if they qualify by linkages to state or federal tax information spends subsidy only on the marginal enrollees who otherwise would not have enrolled while making the process less painful for people who had already shown a willingness to jump through hoops to get and stay enrolled.