« Some mixed economic news for your Friday | Main | The measure of a press release »
March 18, 2005
Capacity Utilization
I meant to respond to this earlier, but am just catching up. Macroblog reported on the increase in capacity utilization. In a comment over there, PGL speculates on the rate that is consistent with full employment. Here are some quick stats on capacity utilization. Since 1967, the mean has been 81.4% and the median 81.9%. Since 1988 (chosen somewhat arbitrarily to include the peak before the 90-91 recession), the mean has been 81.0% and the median 82.1%. The peak in the latter period was 85.1%. As the chart below shows, the speed of the recovery of capacity utilization has been about on pace with the last recovery, but the level at its nadir was lower this time (i.e. we have more ground to make up).
Looking at the 3 month moving average of the changes in the capacity utilization rate shows that the recovery has been, in fact, quite steady, with fewer fits and starts than after the last recession. But again, we did (and still do) have more ground to make up.
Posted by William Polley at March 18, 2005 4:27 PM
Trackback Pings
TrackBack URL for this entry:
http://www.williampolley.com/cgi-bin/mt-tb.cgi/145
Comments
It is interesting to see the time series graph. Lots of swings with an unclear message, but I think your interpretation on this is about right. The last time I plotted something like this was about 10 years ago and my starting point was circa 1960. An even more unclear message as the peak exceeded 90%.
Posted by: pgl at March 18, 2005 5:20 PM
It's been awhile since I last looked at this, but if I remember correctly capacity utilization does map fairly well with deviations of either GDP or the unemployment rate from the CBO estimates of the natural rate. But there is another way to measure business cycles that is even more correlated with capacity utilization.
I don't know if you are familiar with this or not, but Friedman has, at least twice in his life, proposed a "plucking model" of aggregate fluctuations. In this model, output does not deviate around a centrally located natural rate. Instead, it deviates from a ceiling. Friedman noted first that visually a "ceiling" trend fits the output data well (a quadratic trend with the constant shifted upward works fairly well on logged data). He also noted that measured in this way, downturns are highly correlated with subsequent upturns as you would expect, but upturns are not correlated with subsequent downturns (unlike in natural rate models).
The point I’m getting to is that these types of deviations (i.e. ceiling value minus actual value) are more correlated with capacity utilization than are deviations from the natural rate as traditionally measured; at least my recollection is that this is the case.
It shouldn’t be surprising that these are correlated since capacity is relative to a ceiling, but when such a measure is used to measure cycles it conforms more closely to the types of fluctuations Friedman proposed than to natural rate measures.
Posted by: Mark Thoma at March 19, 2005 12:15 AM
Mark,
Interesting. I will look into that.
Posted by: William Polley at March 19, 2005 12:23 AM
Wow, I just searched for "friedman plucking model" in Google and I was surprised how many hits came up. Wish I had time to look into them further. The last time I looked into this topic was after seeing Friedman present it at a SF Fed meeting. I wanted to see if such fluctuations measured as deviations from the ceiling were more correlated with money shocks than cycles as more typically measured. At that time, there were very few papers on this topic, but it appears that has changed. I suppose that is how it should be.
Posted by: Mark Thoma at March 19, 2005 4:21 AM