A number of CBO-watchers were asking that question when the agency started scoring health care bills a few months back. As Washington and Lee law professor Timothy Stoltzfus Jost put it in Politico:
It is much easier to score costs than cost-savings. Legislation pending in both the House and Senate in fact includes state-of-the art proposals that many health policy experts do believe will result in real savings, as the CBO recognizes. It is easy, however, to figure out how many people are under a particular multiple of the poverty level and how much it will cost to cover them through Medicaid or to provide them with insurance subsidies, i.e the cost of reform. It is much harder to figure out how much public plan choice or accountable care organizatons will save the federal government. The CBO guesses conservatively with respect to savings, and the media reports this as a "blow to reform."
Now, in an interview with The Washington Post's Ezra Klein, CBO director Doug Elmendorf responds to the criticism:
I understand many people have that concern. But we try very hard to offer numbers in the middle of the distribution on outcomes, both on the spending and cost side. It's true that if you want to spend $10 billion a year of preventive medicine, we can be pretty certain that it will cost $10 billion a year. So the uncertainty is low. But when we turn to the benefits of, say, preventive health benefits of health spending, we'll try to be in the middle of the range of possibilities. That's a wide distribution of possibilities. The uncertainty will be much wider. But we don't try to be conservative in the sense of being close to zero. We try to be in the middle.
I think people's frustration arises partly because one's intuition about the effects of certain policies on future federal spending is not always confirmed by the hard evidence. On the preventive side, most people's intuition is that discovering a disease earlier lets you treat them earlier and saves a lot of money. The difficulty with prevention is that you end up doing the test for a lot of people who wouldn't have the disease. Even if you're saving a lot of money on those who would have had the disease you don't save that much money ultimately. This isn't our research to start with. A paper in the New England Journal of Medicine last year found that only 20 percent of preventive spending saved money. A large chunk improved health at a low cost, so it may be a good deal. And then some was just extremely expensive and didn't do much to improve health. But that's the sort of evidence we're reading.
It's a reasonable point. The only thing I'd add is that the criticism has actually been two-fold: both of CBO's methodology and of its track record (that is, the way its projections line up with what subsequently happens). Elmendorf addresses the first point but not the second. Yet, as Bruce Vladeck, a former administrator of Medicare and Medicaid, wrote this summer:
[T]he CBO’s track record in predicting the effects of health legislation is abysmal. Over the last two decades, the CBO has routinely overestimated the costs of expanded government health care benefits and underestimated the savings from program changes designed to reduce expenditures. Most recently, it overestimated the five-year cost of Medicare Part D — the prescription drug benefit -— by more than 35%. Even more dramatically, the CBO’s estimates of the Medicare savings from the Balanced Budget Act of 1997 underestimated the impact, on average, by a full 100%. That’s right: In the BBA’s first three years, Medicare spending fell fully twice as fast as the CBO had projected.
Elmendorf's point about uncertainty may be part of the answer, but it doesn't explain why the estimates would systematically lowball savings.