Capacity Utilization & Economic Growth: A Broken Relationship
By Garett Jones
The new Econ Journal Watch piece by David Cushman checks this claim that Krugman made in March 2009:
…it is right to expect high growth in future if the economy is depressed now.
I’m not writing about that quote, since Cushman and Greg Mankiw have done a good job on that one. I’m writing about this claim, that when we started using unemployed capital again we’d start building GDP as well:
Krugman (March 2009):
And yes, we can expect fast growth if and when that capacity comes back into use.
Even in context, Krugman doesn’t define “that capacity” clearly but he does use the expression “capacity utilization.” So I pulled down the widely-used capacity utilization index to see if, post-crisis, the old relationship between capacity utilization and GDP growth still holds true. Spoiler alert: It doesn’t.
In the past, there was quite a strong relationship between changes in capacity utilization and short-term GDP growth, especially if you leave out the service sector (which is usually stable anyway). So Krugman’s prediction really was based on long-term experience. A good 1996 paper on capacity utilization by two Fed economists notes:
The correlation between annualized changes in the real output of goods and structures and the Federal Reserve’s index of capacity utilization for manufacturing, the main goods-producing sector, is about 0.9. In short, capacity utilization in manufacturing is indicative of the cyclical state of the overall product market…
I estimated the post-1967 relationship between capacity utilization and GDP growth on a quarterly basis, and it’s usually quite strong (correlation = +0.7). Then I checked to see if post-crisis the usual relationship has held: It hasn’t.
Just looking from the 3rd quarter of 2009 onward–as soon as growth returned–it turns out that in 13 out of the 14 quarters, GDP growth is below the predicted level (Q3 of 2012 is the big success). On average, GDP growth is half of the predicted growth rate.
[OLS estimate: Predicted GDP Growth = 2*(Change in Capacity Utilization) + 2.8. Estimated on quarterly data 1967-2012. Results reported at annualized rates.]
It’s not just that the relationship between capacity utilization and growth is noisier than it used to be before the crisis: It’s that growth has consistently been less than you would’ve expected based on how many unused machines got turned back on in 2009 and 2010. Part of the story is that the service sector took a hit after the crisis, so it wasn’t the usual stabilizing force.
So during the crisis we stopped using some machines, then we went back to using them but we didn’t produce the usual amount of GDP growth overall. The recovery has been good for capital, mediocre for GDP and worse for workers.
Apparently, one doesn’t often apologize for forecast errors these days–nice people don’t need to read the apologies and mean people will just gloat–and in any case Krugman might be embracing a lesson of his inaccurate forecast all the same. Now he’s writing about the rise of the robots. Perhaps the workers are ZMP but the capital is not.