NBER** has a really interesting (to me) paper on the economic and policy ramifications of linking subsidies to a non-fixed price versus a subsidy that is independent of price. This is very important to my thinking because my Silver Gap and Silver Spam ideas are effectively hacks that exploit the problems that this paper identifies with price linked subsidies. Let’s look at the paper and then go through a phalanx of indecipherable text:
price-linked subsidies as involving a basic tradeoff. On the one hand, they create a competitive distortion, increasing consumers’ or government’s cost by leading to higher prices compared to “fixed subsidies” set independently of prices. On the other hand, price-linked subsidies create an indirect link between subsidies and cost shocks, which can be desirable in the face of uncertainty about health care costs…
If a higher price yields a higher subsidy for the firm relative to other options in the market, the firm has an incentive to raise its price. Even in markets where subsidies are constant across plans, as in the ACA, the subsidy usually does not apply to the “outside option” of not purchasing insurance. A higher subsidy decreases the cost of buying a market plan relative to not buying insurance. (My Emphasis) Each firm gains some of the consumers brought into the market by the higher subsidy, so each firm has an incentive to raise the price of any plan that may affect the subsidy.
The second paragraph is the key insight behind the Silver Gap strategy where a carrier in a region that owns the least expensive Silver should attempt to maximize the spread between in order to make the subsidized buy/no buy decision more favorable and less expensive compared to running naked. We have seen examples where aggressive Silver Gapping leads to families of three being able to pay $0 per month in premiums for a Silver plan up to $38,500 in income and $0 in premiums for a Bronze plan while earning over 300% FPL per year. Signing up is painless for quite a few people in aggressively gapped counties even before we account for the mandate penalty.
And now the big meaty policy questions below the fold — what does this do to enrollment?
We use standard demand estimation methods for micro-data, similar to Berry et al. (2004). Our model allows for adverse selection by letting both demand and cost parameters vary with demographics. An important strength of our method is that we use quasiexperimental variation to identify a key statistic – the responsiveness of insurance demand to the price of the outside option.8 We draw on two natural experiments: the state’s introduction of a mandate penalty and a subsidy increase that lowered all plans’ premiums (but
left the penalty for uninsurance unchanged). In both cases, we use income groups excluded from the change as control groups. The results are similar: each $1 increase in the relative price of uninsurance raised insurance demand by about 1%….
When costs are as expected, fixed subsidies result in higher welfare. But as costs diverge from the regulator’s expectations, the gap narrows, and with a large enough shock, price-linked subsidies may do better
A Silver Gap strategy lowers the price of insurance for subsidized individuals compared to uninsurance. It may increase the price of insurance for non-subsidized individuals as some plausible products that could have been offered between the actual lowest Silver and the actual second lowest Silver were not offered. However the QHP market is has an odd feature. The on-Exchange market and product profile is only a subset of what is offered by the Off-Exchange market. Products can be offered off-exchange that are not on-Exchange. Thus carriers could offer the forty three isomorphs of their lowest priced Silver plan. Buyers who are not subsidy qualified can buy on or off-Exchange so they lose very little utility from the option value if the carrier offers the full array of products off-Exchange that they would have offered on-Exchange if they were not trying to Silver Gap.
The most important finding of the paper is the obvious one. Cost-linked subsidie shift risk from individuals of large cost spikes to the Federal government. Subsidized individuals this year are fairly well protected from 20% rate increases. They are protected if they live in Rhode Island which has very low base rates and they are protected if they live in Alaska which has the highest national rates. Unsubsidized individuals got kicked in the shorts this year on the rate hikes. If the subsidy was a formula that was not tied to rate hikes but rather to expected rate information, most of the subsidized population would be paying more which means lower enrollment uptake.
This is a very useful paper that puts some meat to the bones of the subsidy manipulation games that I like to think about.
** Jaffe, S., & Shepard, M. (2017). Price-Linked Subsidies and Health Insurance Markups. NBER Working Paper Series. doi:10.3386/w23104