Spend enough time reading Congressional Budget Office reports, and you start to notice something: Although the nonpartisan congressional scorekeeper puts a lot of effort into carefully relaying their estimates, they are more cautious when it comes to explaining how those projections were obtained, noting the various forces in play but not describing the precise methodology at great length. Similarly, the CBO often highlights uncertainty around its estimates, but less often attempts to quantify that uncertainty. The office tends to release point estimates rather than ranges, noting that its estimates are designed to fall in the center of the likely range, however large or small it might be.
There’s a reason for this: In addition to providing estimates about the likely effects of legislation, the CBO’s goal is to corral discussion of a bill around a single number. That number may not be exactly correct; in fact, it almost never is, and CBO does not dispute this. But it represents a sort of extremely well-educated best guess, one that attempts to identify the most likely outcome, even if that specific outcome is itself not all that likely. And the CBO’s opaque characterizations of its methodology limit, to some extent, the arguments that can be had about its calculations and assumptions.
What this means is that the CBO is designed to push people—the media, the public, and especially legislators in Congress—to discuss the potential results of legislation rather than to argue about how the CBO came up with its estimate. The CBO wants lawmakers to debate what a bill would do, not what the CBO has done.
There is some genuine utility in this approach. Focusing the discussion on point estimates rather than on ranges means that partisans cannot cherry pick a convenient high or low figure, and it nudges the conversation toward one that is more focused on policy outcomes than on procedural arcana.
But there is also some organizational self-interest involved as well. It’s a process that is designed to preserve the CBO’s authority, because it implicitly encourages legislators to act as if any given bill will do what the CBO says it will do, regardless of how likely it actually is. It’s a process, in other words, that is designed to preserve the CBO’s authority.
And there are limits to that authority, as well as to the process as a whole. The CBO’s latest score of the Republican health care plan tested those limits.
Released yesterday evening, the score is CBO’s third report on the American Health Care Act (AHCA), which would partially repeal and replace on Obamacare. But it is the first to CBO report to assess the version of the bill that actually passed in the House earlier this month. House Republicans made a series of amendments to the earlier versions of the legislation, adding funding for high-risk pools and a provision that was supposed to give states more flexibility to opt out of Obamacare’s major regulations. Republicans argued that the adjustments would make coverage more affordable while preserving preexisting conditions protections.
The House GOP then took the unusual step of voting on the bill without the CBO’s score. The results show why.
The CBO estimated that the bill, as passed, would result in 14 million fewer Americans with health coverage in the space of a year, and 23 million fewer with coverage a decade from now. That represents a slight decrease in projected coverage losses versus a previous version of the bill, which was estimated to result in 24 million fewer people with coverage. But it is still a rather large number, especially for a bill that purports to replace Obamacare and preserve some of its regulatory protections for preexisting conditions.
Notably, 23 million is the exact same number that CBO projected for a simple repeal of Obamacare with no replacement at all. That’s worth dwelling on for a moment. CBO is saying that, in terms of coverage, the GOP bill to replace Obamacare would not be any different than repealing the existing health care law outright and replacing it with nothing.
The CBO also warned that the state flexibility amendments added to the version of the bill that passed in the House would undermine insurance markets rather than improve them.
Despite additional funds, less health people, including those with preexisting conditions, “would ultimately be unable to purchase comprehensive nongroup health insurance at premiums comparable to those under current law, if they could purchase it at all,” the report states. “As a result, the nongroup markets in those states would become unstable for people with higher-than-average expected health care costs.”
Like I said, you can see why Republicans chose to vote on the bill before the score came through.
But how, exactly, did the CBO arrive at those headline numbers? That’s where things get tricky. The final House amendments to the AHCA gave states the option of applying for waivers to opt out of Obamacare’s community rating, which limits the ways in which insurers can price insurance based on health history, and essential health benefits rules. So one of the questions the CBO had to answer was: How many states would pursue a waiver? That’s not a question that can be answered with an economic model or a historic comparison. It is a political question, and it’s one that the CBO simply isn’t well-suited to answer.
Yet in order to score this version of the AHCA, the CBO had no choice but to make assumptions about which states would opt out. The report assumes that about a third of the population would end up in states that obtain waivers to opt out of some of Obamacare’s regulations. About a sixth of the nation’s population would end up living in states that opted out of both the essential health benefits rules and community.
And, tellingly, it simply declines to estimate the effects that opting out would have on premiums would have on those states. Yet it still provides an overall estimate on premiums nationally. (CBO officials told The Washington Examiner‘s Philip Klein that the range was so large that it would be misleading to provide an average—information that wasn’t in the original report.)
The report offers a general explanation as to why it settled on these estimates including market conditions and the past behavior of states. The report stresses the broad uncertainty involved in making an estimate like this.
But what it doesn’t quite say—what it can’t say—is that ultimately, the decision to apply for a waiver is one that cannot really be estimated by the economists and wonks at the CBO. It’s a political decision, not an economic projection.
That is not the CBO’s fault. It’s not because the office lacks expertise or because it is somehow institutionally biased. On the contrary, the deeply technical and apolitical nature of the nonpartisan CBO, which has always tried to keep politics out of its economics, is part of the reason this is such a challenge. Instead, it’s because the question is one that cannot really be answered by anyone, because it is a prediction about highly contingent human whims.
At this point, no one knows how states would respond to the opt-out provisions, including, I think it’s fair to say, most state lawmakers. After Wisconsin Gov. Scott Walker said he would consider applying for a waiver, he ended up in a tense confrontation with a local Democrat. It’s clear from that scene alone that the politics surrounding the waivers would be tumultuous.
A non-partisan economic scorekeeper like the CBO doesn’t really have a way to deal with a bill like the AHCA. It is in some sense unscoreable. But the CBO has to score it anyway.
That is not to say that the CBO’s scores have no value, or should be dismissed entirely. Even on a project like this, the CBO provides directionally useful analysis that can and should help shape the debate. Even an imperfect score is more useful than no analysis at all, which is what many of the CBO’s critics would seem to prefer. In this particular case, it is almost certainly the case that the GOP legislation would lead to millions fewer with coverage, and would result in pockets of insurance market instability.
But it’s hard to go much further than that, and harder still to predict how the political debate around waivers would turn out. Even the CBO’s founding director, Alice Rivlin, seems to agree. “It’s just too complex and too political,” she told Politico prior to the analysis being published. “I just don’t see how they do this. It will cause complications but the idea that they could come up with a number seems to me, just a bridge too far.” But come up with a number is what the CBO did.
Which is why the AHCA, perhaps more than any other legislation, reveals the structural limitations of the CBO’s process and self-conception, and the problems that eventually arise from the opacity with which it conducts its estimates. It is hard for economists to make political predictions, and harder still to convincingly project authority on a subject about which no one can truly be authoritative.
