The Real Problem With The Romer/Bernstein Chart


It’s become conventional to note that the Romer/Bernstein chart forecasting the impact of the American Recovery and Reinvestment Act was flawed by failure to recognize how deep the downturn was. But looking back, I think it’s possible to see another, deeper flaw, that relates back to the main mistake about the Fed and expectations. Look at the Washington Post’s helpful updated version of the chart:

What’s noteworthy about this, in retrospect, isn’t really the mistake about the level. It’s the assumption that the with-ARRA and without-ARRA scenarios converge. The point of ARRA in this case is to cushion the blow. Note that they barely even claim that ARRA would speed the pace at which the economy regains its full employment level. What they claim, in essence, is that there’s some underlying force of convergence that will take a little while to work its magic. ARRA’s role is, during the interim period, to keep some people in work who would otherwise be out of work. But fundamentally, the X-Force is driving us back to full employment on its own schedule. Was that a reasonable assumption? I would say “yes.” After all, consider the long-term trajectory of total nominal spending in the economy since World War II:

Lots of stuff has happened in this time, but deviations in the total nominal spending trend always even out. And they do so relatively quickly. It’s not really a question of “in the long-term we’re all dead,” the total spending figure evens out in the medium-term. That right there is your X-Force. But of course even the medium-term matters to people’s lives. The ARRA scenario with its consistently lower unemployment is a much better world than the no-ARRA scenario even though the two worlds converge within four years.

Now here look at total nominal spending rates during the Great Moderation:

Recessions, you can see, happen when total nominal spending growth dips. But it normally bounces back. During 2009, however, we had an unprecedented collapse in total nominal spending. What’s more, the 2010 “recovery” year was just as bad as normal recession years. So now look at the large and growing gap between the actual path of total nominal spending and the 5 percent trend growth rate:

This is what’s not accounted for in the Bernstein/Romer projection. There is no X-Force driving convergence to the long-term trend. The failure of the X-Force to materialize has nothing to do with the fact that the Commerce Department initially underestimated the depth of the recession. The existence of the X-Force was a modeling assumption, not an empirical calculation. And I think it’s an assumption that’s best understood as an assumption about the stance of the Federal Reserve—a view that the Fed, with its words and deeds, would push us back up to the trend leaving Congress with the responsibility for safeguarding human welfare during the transition. It’s an assumption that I think anyone familiar with Ben Bernanke’s academic work would have shared, so I understand why Romer especially (who shares my view of the situation) espoused it. But of course she’s gone, and I’m not sure that references to “headwinds” from Europe fully accounts for the depth of the problem here.