How many people would lose health insurance if King wins?

4/07/2015 03:23:00 PM
You've probably heard of King v Burwell, the lawsuit before the Supreme Court that could take ACA subsidies away from 34 states if the plaintiff, King, wins. You're probably also aware that in such an event there is no viable plan to pass a Congressional fix and, realistically, little chance that states will be able to establish their own exchange in time to keep the subsidies.

It would be unusual if the Court stayed the ruling--though Justice Alito hinted at it--so if King wins that probably means 9,346,000 people lose subsidies for their health insurance immediately. How many will lose insurance completely?

That is a question with several dimensions. Without the subsidies, many people won't want to pay for coverage anymore at the new higher price. Then there is an additional effect caused by adverse selection: those who do still want insurance at the higher price will be the higher-cost, sicker patients, meaning that the risk pool will become higher-cost, further driving up premiums for everyone above and beyond the current pre-tax prices. RAND says premiums would rise 47 percent above the current unsubsidized rates due to adverse selection. It doesn't stop there. Those who do continue to buy insurance will downgrade to the cheapest plans offering the least amount of coverage.

Rand estimates that 9.6 million people would lose insurance entirely in this scenario, about 70 percent of the non-group market for the affected states. The Urban Institute's estimates don't look much better: 8.2 million lose insurance. And I'd add that much of the loss is likely to be permanent, as those who get burned by a sudden, massive, and totally unexpected (more than half of the public is totally unaware of King v Burwell, and even fewer know it affects their state) rate hikes will never return to the exchanges again.

What got me thinking about this was this paper from 2009 that estimated the elasticities of demand for health insurance among the self-employed. The elasticities aren't necessarily generalizeable to the ACA--theory tells us that insurance equilibria aren't very robust and depend heavily on market structure. But out of curiosity I did the calculation. They estimated an extensive (the decision whether or not to buy insurance) elasticity of -0.3 and an intensive (decision of how much to buy) elasticity of -0.7, which is pretty moderate as far as published estimates go. Just counting the loss of the subsidies, that works out to 7,205,766 people losing insurance due directly to the loss of the subsidy. If we borrow RAND's estimate of the price effect from adverse selection, that's more than a million additional people loosing insurance, for a total of 8,813,166 loosing insurance (see update). So, RAND's and Urban Institute's estimates look pretty reasonable to me.

And what about the intensive margin? The ACA metal tiers didn't exist in the study sample, of course, so we have no idea what will happen. But if we do apply the -0.7 estimate literally, that means a King victory will totally wipe out all but the very cheapest plan in the Federal Exchanges.

Update: The 8,813,166 figure has been revised up because I originally forgot to include exchange plans that experience a price hike from adverse selection but were not receiving subsidies. I'm still probably missing people because the ACA also requires insurers to pool risks across all conforming plans, including both those sold on and off the exchanges. Thus most of the off-exchange non-group market would experience a similar 47 percent rate hike.