This year’s health-care debate has launched, or at least reawakened, a related argument about the work of the Congressional Budget Office.
Republicans have found themselves very frustrated with CBO’s health-economy model, and especially its tendency to ascribe much greater effectiveness to the individual mandate than the evidence of the last few years would suggest is merited. This has led the CBO to score Republican proposals as dramatically reducing insurance coverage, even when they spend huge sums to stabilize markets and subsidize premiums. Democrats have tended to lay off the CBO and to defend it, in part because its eccentricities are politically useful for them this time (unlike in 2010). But they have had reasons to be frustrated too: The agency’s scores of the Republican bills have allowed Republicans to get away with claiming enormous budgetary savings (and at times with “spending” that savings on various ways of bridging intra-Republican policy differences) when such money is very unlikely to materialize in reality under the plans they propose.
When Republican health-care efforts resume next month, they seem likely to be focused on the kinds of massive intergovernmental transfers envisioned by senators Graham and Cassidy this summer (transforming Obamacare’s individual entitlements into block grants to the states), which will cause them to intersect with another common set of complaints about the CBO, related to the challenges it faces in assessing and projecting state policies.
In both cases, a lot of the frustration with the CBO has to do with a sense that its work is kept behind a veil, and it tends to release conclusions without being as clear as it might be about methods and uncertainties. I took up this general question around here a couple of months ago and suggested (as I and others have argued before in recent years) that while the CBO’s critics certainly sometimes go too far, the core critique is basically justified and the agency should begin to take steps toward radically greater transparency, and ultimately toward some form of open-source modeling.
A few days ago, Senator Mike Lee proposed legislation that would begin to point the way in that direction. It’s a modest first step, but it would be the right first step and an important one. Lee introduced what he calls the “CBO Show Your Work Act.” The bill would basically require CBO to follow the “data availability guidelines” that the American Economics Association recommends for academic economists. It would have to publish the data, the models, and the data-preparation routines used to arrive at its cost estimates and projections. Those could be used by other modelers, and through them by policymakers and analysts, to better understand CBO’s methods and assumptions, and to offer informed criticism, improvements, and alternatives. There would be protections for sensitive data used by the CBO in its modeling, which would be designed to make those constraints clear and to allow people who also have access to the data to replicate CBO’s work.
The bill doesn’t require CBO to change how it does its core work. It doesn’t really move the agency toward something more like tending and curating a set of open-source models. That should happen over time. But this first step toward transparency, which would only require CBO’s professional staff to follow the standards and practices their colleagues in academic institutions do, would make a big difference. It would strengthen, not undermine, CBO’s ability to offer the projections required of it by law, and would make it easier for the agency to answer critics and to pre-empt them.
There are some real obstacles and concerns to think through when it comes to a truly open-source CBO, some of which I took up here in June. But if there is a serious argument against this very modest opening reform proposed by Senator Lee, I haven’t heard it.