Ace commenter John Thacker pointed out that Greg Mankiw (and others) have argued that macro aggregates have a unit root and thus reversion to a fixed trend is not to be expected.
I don't want to get into a big discussion about the power of unit root tests here, so let me show a picture (from the blog Calculated Risk) that illustrates what I was trying to say (clic the pic for a more glorious image):
The graph shows job losses in the recessions since WWII. All but the last three could be reasonably described as "sort of V shaped" and except for them, time to recovery seems almost independent of the severity of the recession. Our current situation is notable both for the severity of the job losses and the extreme slowness of the job market to recover.
2 comments:
Is it possible to explain the graph by comparing recession duration to year of onset, with the general trend that the more recent, the longer (and now by far the longest).
This in turn would imply something about the US historically changing in the way it deals with recessions. (I have my own pet idea - the more government interference to 'fix', the longer the recession).
Both this comparative trend as well as the duration of the recession itself argues that we won't be back to baseline (0.0%) for another couple years minimum, thus a 7-year recession in total.
Bill K.
Even the previous two recessions before this one at least do look symmetric-- they took a long time to recover, but the job loss was also slow and steady. What makes this recession particularly bad is that the job loss was quick, but the job recovery has been as slow as the the two previous recessions. (I assume that's what you're focusing on as the weirdness.) Oh, for the days when 1991-2 and 2002-3 could be described as "jobless recoveries."
However, it's actually quite difficult mathematically to distinguish whether it's trend-stationary or not, especially from that graph. You would need to look to see what the trend lines looked like after emerging from the recession. The scatterplot that Brad DeLong posted was a textbook case of a dubious correlation-- all the points that drove the correlation were from the boom following the 1982 recession. Aside from that outlier, unemployment and change in real GDP was uncorrelated. (Of course, there's a difference between change in GDP and change in unemployment. When was the last time we raised minimum wage sharply just as a recession was starting, anyway?)
Incidentally, Greg Mankiw linked to this report by the IMF's Oliver Blanchard noting that research suggests that banking crises are particularly likely to look like a unit root is in effect:
The historical evidence is worrisome, however. The IMF’s forthcoming World Economic Outlook presents evidence from 88 banking crises over the past four decades in a wide range of countries. While there is large variation across countries, the conclusion is that, on average, output does not go back to its old trend path, but remains permanently below it.
Post a Comment