Throughout the recent healthcare fight, the estimates by the Congressional Budget Office (CBO) has been taken as holy writ among Trumpcare/McConnellcare/Ryancare’s opponents, as the CBO estimates of the American Carnage (to use a phrase) resulting from these ‘healthcare’ bills were horrific. As a result, Republicans proposed slashing the CBO budget, along with a ludicrous proposal to replace CBO estimates with reports from several thinktanks (coincidentally, none of which happened to be liberal. They did include the Heritage Foundation however).
So when the CBO was attacked, Democrats, progressives, and others howled with dismay (there was probably also some schadenfreude regarding past use of CBO estimates by Republicans). But it’s worth noting that the CBO doesn’t always provide good estimates; sometimes, their assumptions underlying estimates defy logical sense (boldface mine):
The forecasts of the Congressional Budget Office are holy writ in Washington, and they fuel scary headlines about an impending federal debt disaster. This is a shame, because the CBO’s projections are indefensible, internally inconsistent and economically impossible.
The CBO predicts that unemployment will fall to near 5 percent by 2014 and stay there. It also expects a rapid recovery in the next few years, followed by a steady 2.4 percent GDP growth rate thereafter. Inflation is expected to stay below 2 percent indefinitely.
But alongside these rosy numbers, the CBO also projects that short-term interest rates will increase from less than 0.2 percent now to 4 percent in 2014 (and higher later), while rising health-care costs will drive Medicare expenditures ever higher. These figures imply that interest payments on the federal debt will by 2020 “rival the defense budget,” as Clinton-era Treasury official Roger Altman recently wrote in the Financial Times.
These things cannot happen together. If the CBO’s happy growth scenario is right, with low inflation and low unemployment, why would short-term interest rates rise? Conversely, if the CBO’s assumptions about health-care costs and interest rates are correct, how can inflation stay low? Ballooning interest payments and health-care spending would spur the economy to full employment and drive up prices — but also slow the rise in debt as a proportion of the nation’s gross domestic product…
If we’d had a CBO in the 1930s, Franklin Roosevelt could never have gotten the New Deal off the ground.
One might wonder if the last sentence is partly the point. But it gets worse once you drill down even further (boldface mine):
Consider the baseline economic forecast of the Congressional Budget Office, the officially nonpartisan agency lawmakers rely on to evaluate the economy and their budget plans. In its early January forecast in 2009, CBO measured and projected the departure of actual from “normal” economic performance – the “GDP gap.” The forecast had two astonishing features. First, the CBO did not expect the recession to be any worse than that of 1981-1982, our then-deepest postwar recession. Second, CBO expected a strong turnaround beginning late in 2009, with the economy returning fully to the pre-crisis growth track by around 2015, even if Congress had taken no action at all.
Why did Congress’s budget experts reach this conclusion? On the depth and duration of the slump, CBO’s model was based on the postwar experience, which is also the run of continuous statistical history available to those who program computer models. But a computer model based on experience cannot predict outcomes more serious than anything already seen. CBO – and every other modeler using this approach – was stuck in the gilded cage of statistical history. Two quarters of GDP loss at annual rates of 8.9 and 5.3 percent were beyond the pale of that history. A long, slow recovery thereafter – a failure to recover in any full sense of that word – was even more so.
Further – and partly for the same reason that past recessions had been followed by quick expansions – there was baked into the CBO model a “natural rate of unemployment” of 4.8 percent. This meant that the model moved the forecast economy back toward that value over a planning horizon of five or six years, no matter what [Mad Biologist: U3 unemployment didn’t return to 4.8% until 2017]. And the presence of this feature meant that the model would become more optimistic when the news got worse. That is, if the news brought word of a ten percent unemployment rate, instead of eight percent, then the model would project a more rapid rebound, so as to bring the economy back to the natural rate. A twelve percent unemployment rate would bring a prediction of even faster recovery. In other words, whatever the current conditions, the natural rate of unemployment would reassert itself over the forecast horizon. The worse, the better….
Looking out over ten years or so, official economic forecasts tend to show minor losses from stimulus programs. This is thanks to what they project to be the financial consequences – higher interest rates – of increasing the government’s debt. This effect is supposed to “crowd out” private capital formation that would otherwise have occurred. Once the shortfall in total production is made up, the extra interest burden associated with recovering the lost ground more quickly than otherwise is projected to weigh as a burden on private economic activity going forward. With less private investment, there will be (it is projected) a smaller capital stock, slightly less output, and eventually the gains associated with stimulus will be outweighed by these offsetting losses.
And so the Obama Team found itself working from predictions that foresaw a top jobless rate of around nine percent, with no stimulus, and a fast recovery beginning in the summer of 2009 with stimulus or without it. Those forecasts helped to place an effective ceiling on what could be proposed or enacted, as a practical matter, in the form of new public spending.
So the CBO isn’t all that. But then, why should it be taken seriously regarding healthcare? Because quantitative differences don’t really matter. Whether the actual figure was ‘only’ five million losing healthcare, or thirty million losing healthcare (after all, the CBO could have underestimated the effects as well), it would have been a humanitarian crisis either way (American Carnage!). And no one thought that massively cutting Medicare and Medicaid would not have a harmful effect*. No one. The question was how awful it would have been–that is, how many millions would lose healthcare insurance, and how many thousands would die needlessly every year.
The point isn’t to ‘ban’ the CBO–or outsource it to the Heritage Foundation**–but to recognize that the estimates they produce have assumptions built in. Some of those assumptions aren’t very good, even if they’re ‘industry standard.’ Unfortunately, too many of our pundit class (including some economists) are just smart enough to read the reports, but not understand their methodological assumptions. If we give away our ability to critique CBO estimates when there is good cause to do so, we will be hamstrung on policy in the future.
So keep the CBO, but remember that it’s not infallible by a long shot.
*It’s worth noting that the effects on long-term assisted care weren’t even really brought into the discussion, but they would have been horrific as well.
**Though maybe they could hire a couple of people from EPI or CEPR?