« Fed Watch: Bernanke Not Afraid of Bubbles | Main | Links for 02-23-2013 »

Friday, February 22, 2013

'Nature Abhors an Output Gap'

David Altig admits to "overly rosy projections" about the course of the economy (which can cause policy to be too timid), and he tries to explain how this can happen. I can't really claim the high ground here. My first post for CBS MoneyWatch in November 2009 explained why employment might not fully recover until 2013. I was very pessimistic relative to most, and wondered at the time if I was being too pessimistic in saying it could be four years, but of course it turned out I wasn't pessimistic enough (I suppose I could argue that I relied in part upon an SF Fed forecast for the recovery timing, but that's not much of a defense since I didn't question the forecast in the post). My main message was that it was going to be a long time before employment recovered and policy needed to do more to help, much more, (that's still true) so I was at least pushing in the right direction:

Nature Abhors an Output Gap, macroblog: In The Washington Post, Neil Irwin highlights a shortcoming that I know all too well... In fact I don't think Irwin's indictment is overly harsh, and he is on the right track when he offers up this explanation for the last several years' persistently overly rosy projections:

Economic forecasters tend to look at past experience and extrapolate; in the past, when there has been a recession, the very forces that caused the recession become unwound, sowing the seeds for expansion...

Here is a basic fact about macroeconomic forecasting. The truly powerful driver of forecasts is mean reversion, which is the tendency of models to predict that gross domestic product (GDP) will move toward an average trend over time. This fact holds true whether we are talking about formal statistical analysis or the intuitive judgmental adjustments that all forecasters apply to their formal statistical models.

Forecasters are not completely robotic, of course. Irwin is correct when he says "forecasters tend to look at past experience and extrapolate, but forecasters do leaven past experience with incoming details that alter judgments about what is the mean—the "normal state," if you will—to which the economy will converge. But whatever is that normal state, our models insist that we will converge to it.

Nothing illustrates this property of forecasting reality better than this chart, which supplements the latest economic projections from the Congressional Budget Office:

The potential GDP line in that chart is the level of production that represents the structural path of the economy. Forecasters, no matter where they think that potential GDP line might be, all believe actual GDP will eventually move back to it. "Output gaps"—the shaded area representing the cumulative miss of actual GDP relative to its potential—simply won't last forever. And if that means GDP growth has to accelerate in the future (as it does when GDP today is below its potential)—well, that's just the way it is.
Unfortunately, potential GDP is not so simple to divine. We have to guess (or, more generously, estimate) what it is. That guessing game has been harder than usual over the past several years. Here is the record of the CBO's potential GDP since 2009:

I think this picture is a fairly representative record of how views about the potential level of U.S. GDP has evolved over the past several years. ...

This much, in any event, is clear: Given any starting point where the level of GDP is below its potential level—that is, given an output gap—forecasts will include a bounce back in GDP growth above its long-run average, at least for a while. That's just the way it works.

If, contrary to conventional wisdom, you believe that the true output gaps are much smaller than suggested in the CBO picture above, you might want to take the under on a bet to whether GDP forecasts will prove too optimistic once again.

    Posted by on Friday, February 22, 2013 at 04:24 PM in Economics, Monetary Policy | Permalink  Comments (61)


    Feed You can follow this conversation by subscribing to the comment feed for this post.