Followers of Congress hear lots about Congressional Budget Office (CBO) cost estimates. But what, exactly, is a CBO cost estimate? Is it:

1. An estimate of the comprehensive economic and social cost of legislation?
2. A benefit-cost evaluation of legislation?
3. An estimate of the federal budgetary impact of legislation?
4. All of the above?

The correct answer is c) – an estimate of the federal budgetary impact of legislation. Specifically, a cost estimate shows the year-by-year change in revenues and spending over the budget window (currently 10 years) that would occur if the legislation were to become law. (This change is measured relative to the CBO “baseline” which shows how the federal budget would evolve over the budget window if left on autopilot under current law.)

That’s the CBO’s job: to measure the federal budget cost of legislation. It is a quite limited assignment, and those limitations lead to endless confusion. In particular:

1. It is just costs, no benefits. This is important. CBO never says, and its cost estimate cannot be interpreted as shedding light on, whether a piece of legislation is desirable. Identifying the benefits is the job of those proposing the legislation; CBO just focuses on the costs. Voting yes or no is strictly up to Congress.
2. It is a narrow definition of cost. This is the federal budget cost, not the federal, state, and local cost – or any other governmental entity for that matter. It is also not the economic cost, such as slower growth or distortions of the private sector.
3. It is a short-run measure of cost. The cost estimate contains the budget impacts each year in the budget window, which is currently 10 years. Observant readers will note that the world has not ended at the conclusion of the budget window, and one hopes this empirical regularity will continue. As a result, budgetary impacts continue in years 11, 12, 13, and so forth – indeed, they may even go from negative to positive, or vice versa.

This final feature is a matter of perpetual confusion and frustration in scoring health policy legislation. In many cases, advocates intend for legislation to enable a new therapy (drug, device, etc.) that may be better and/or cheaper than the existing approaches. Notice that if it is better, that is a benefit that is not part of the cost estimate landscape and, indeed, it may even be more expensive.

Alternatively, it could be cheaper than the existing therapy or therapies but only comes into widespread use after a delay for regulatory approval (say 4-5 years) and adoption by medical provides. The result? Few savings show up in the first 10 years and the advocates go on a tirade about how CBO does not appreciate the new therapy. It does, but it also appreciates the empirical hurdles to widespread usage.

Finally, it could be the case that a new therapy prevents the onset of diseases and diminishes the need (and cost) to treat them. This is the kind of benefit one might expect from anti-cholesterol drugs (fewer heart attacks) or preventing the onset of diabetes (and the ensuing heart disease, chronic kidney disease, nerve damage, and other problems). These cases look like an “investment” – one spends the money now and gets a return (lower costs of other health care) later. The bad news in a cost estimate is that the spending occurs in the budget window and the savings after the budget window. Once again, this result is due to the nature of the exercise and not (as advocates will claim) because CBO does not grasp the wonder of the innovation.

CBO costs estimates are a useful tool in evaluating proposed federal legislation. But they are only a tool, not more, and a chronically misunderstood one.