Just Because The CBO Says Something Doesn't Mean It's True

One of the headlines out of yesterday's Congressional Budget Office report was that the risk corridor provision of the Affordable Care Act, which Republicans have assailed as a "bailout" of insurance companies, is actually expected to save taxpayers $8 billion through 2016.

Maybe. Maybe not. Don't put too much stock in this prediction from CBO.

Remember: The risk corridor is a program through which the government assumes, for the first three years of Obamacare, much of the risk of selling health insurance through the exchanges. If people who buy insurance in are sicker than expected, and therefore consume lots of health care, the government will pay insurers to cover much of the excess cost. If costs come in below expectations, insurers will make payments to the government.

This program exists because the cost of providing insurance to a new pool of customers is hard to predict. Insurers did their best to predict the costs when they set their premiums, and they have big incentives to get it right: If they price too low, they'll lose a bunch of money, even after the partial offset from the risk corridor. And if they price too high, they'll lose customers (and they'll end up sending the government much of the excess revenue from the signups they do get, through the risk corridor).

Still, insurers won't have a really clear view of costs until signup is complete in March and several months of claims experience are in place. If insurers don't know how much it's going to cost to cover their members (and therefore how much they will owe or be owed in risk corridor payments) how can the CBO know? The short answer is, it can't. It's guessing, just like the insurers are.

The CBO says it based its projection in significant part on the performance of risk corridors in Medicare Part D. That's the prescription drug benefit that started in 2006; it has risk corridors, and they have tilted toward payments from insurers to the government.

But there are two key differences between Medicare Part D and the ACA. One is that we had a pretty good idea in advance who would sign up for Medicare Part D: Most everybody who's on Medicare. The other is that over the last decade, prescription drug costs have grown slowly, as many key drugs have become available as generics and relatively few expensive new drugs have been released. In that environment, it's no surprise Part D costs have come in below expectation.

CBO's report does not say it considered the one piece of information that really would make them smarter than the insurers: up-to-date demographic information on the health status of the insurance enrollees. CBO can't have this information because it's not collected through the signup process; it will only be known as people start filing claims.

I don't mean to insult the economists at CBO, who are smart people doing careful work. But the value of their projections is driven by the quality of information available to formulate those projections, and that quality is lower here than when, say, CBO estimates how much income tax the federal government will collect last year.

So it's too early to say the risk corridor will save taxpayers money. We don't know. Which, again, is why the program exists to begin with.