Yuval Levin warns that it could be extremely high for the federal government:
[I]n the Obamacare exchanges, the subsidy system is intended to prevent people from feeling the effect of annual premium increases after the first year. The subsidies are designed to make sure that each recipient pays only a certain percentage of his income in premium costs. That percentage stays essentially the same year after year, so if premiums get more expensive the government covers the difference.
In other words, if premiums for coverage purchased in the exchanges were to double or triple in 2015 because of severe adverse selection, people eligible for subsides would still pay the same amount they did in 2014 (assuming their incomes didn’t change) and the federal government would pay for the entirety of the increase. Subsidized beneficiaries would therefore not feel the effect and the healthy among them would not necessarily have much reason to flee the exchanges.
The Congressional Budget Office estimates that about 86 percent of the people who buy coverage in the exchanges in 2014 will receive subsidies. The technical problems limiting enrollment may mean that figure is even higher (since the incentive to enroll is much greater if you’re eligible for subsidies than if you’re not). Those individuals would not feel the effect of second-year premium spikes, which means the result of such spikes, if they were to happen, would likely not fall into the usual pattern of an adverse-selection spiral.
Instead, the sort of severe adverse selection the exchanges may experience would dramatically increase federal spending and would drive unsubsidized exchange participants (other than those in very poor health) and many insurers out of the exchanges.