Capacity Utilization
One thing there is almost universal agreement on is that the US economy has very large excess capacity and that capacity constraints are unlikely to be a significant factor for the foreseeable future.
But is it. In November capacity utilization in manufacturing rose to 75.2%. But this does not mean that the manufacturing sector has 25% of its capacity sitting idle waiting for demand to appear. If you look at the chart of capacity utilization it shows a downward trend of generally lower highs and lower lows across time. For example in the last cycle
capacity utilization peaked at 81.7% in April, 2007 before plunging to a low of 68.2% in May, 2009. The average from the last peak to trough was 75.0% so with capacity utilization now at 75.2% the system is already over half way back to its previous peak.
If you do a trend of the lower lows and lower highs the current trend is at 77.5%, so capacity utilization is only slightly below the long term trend at 97.0% of trend. This 97% figure is clearly a much more realistic measure of how much excess manufacturing capacity exist than the 75% observation being commonly cited.
It is interesting to look at one of the most important components of capacity utilization, that of semiconductors. Semis are often though of as industrial production on steroids as semi output moves in lock step with overall industrial production. This is logical since semis may be the ultimate intermediate product. Each of you probably owns dozens of semis in your cars, computers, phones, TVs and other appliances. But I bet virtually none of you have ever gone into your local Radio Shack and actually bought one. Over the past year total industrial production rose 5.4% while semiconductor output rose 11%. But by historic norms this is very weak as the historic average growth rate for semis output growth is about ten times the growth rate of total industrial production.
As the chart show, with semi output expandingonly some 11% it is so weak that capacity utilization in the semiconductor industry is actually falling. It is now at 65.4% as compared to its recent peak of 70.7% in April, 2010.
Moreover, the growth in semiconductor capacity is slowing significantly. I do not know if this is due to foreign competition or industry executives expectations of weak demand growth. But neither alternative is bullish. Note that back in the 1990s boom the capacity to build semis was doubling almost every year as compared to only a 10% growth rate this year.
How sensitive is the trend you’ve derived to the period covered? If we limited the period to just the Great Moderation, for instance, wou;dn’t the trend be less steeply downward, suggesting a larger output gap now? Is it reasonable to accept that there have been no changes in the structure of the economy over the period you’ve chosen significant enough to run an F-test?
One way to look at your point is to look at capacity growth.
From 1975 to 1995 it fluctuated around 2%.
1997 was the peak year when it was 7.6%.
Obviously the IT boom in the 1990s was
responsible for this surge.
In recent years it has been:
2003…-0.2%
2004…-0.4%
2005…+0.4%
2006…+1.5%
2007…+2.0%
2008…+0.9%
2009…+0.9%
2010…-0.4%.
In the late 1990s, early 2000s auto and light truck output
was around 12 million.
Currently auto and light truck output is around 8 million
but capacity probably has been reduced to some 10 million.
Ask your auto analyst what current capacity is as this is
just a wag– that is an old enginerering term for
wild ass guess.
By stage of processing crude output is now 88.6%
versus the 2007 peak of 90.0%. Much of this is energy.
Finished output is at 73.3% of capacity versus
the 2007 peak of 78%.
The weaker capacity growth in recent years
implies that it is more realistic to look at
output compared to trend now than
it was in earlier years. I calculated the trend
over the entire period since 1967.
Spencer,
If capacity utliization is almost half way back to the previous peak without any meaningful growth in job creation, what is the likelyhood of capacity utilization gaining much more ground. It would seem that the labor force, domestically, will be most likely left behind in the current growth cycle. It might be good for buisness. Not so sure about main street.
Can not say that I disagree with you.
Hi Spencer:
I now work at the PU Mgr. at a Tier One Electronics Manufacturer. Interesting graph on Mosfets metal–oxide–semiconductor field-effect transistor. Our experience over the last year was, and is, contrived capacity amongst the manufacturers. NXP, OnSemi, ST, Infineon, Freescale, NEC, Vishay, Fairchild, etc. have purposely not brought more capacity on line so there is no risk to them if the market collapses again. We are on allocation for these mosfets with lead times as long as 99 weeks and most around 26 weeks. Seeing capacity drop is interesting as we are being told, they are bringing more capacity on line.
I would not look to Labor input in this process as it is miniscule.
spencer – “Over the past year total industrial production rose 5.4% while semiconductor output rose 11%. But by historic norms this is very weak as the historic average growth rate for semis output growth is about ten times the growth rate of total industrial production.”
“As the chart show, with semi output expandingonly some 11% it is so weak that capacity utilization in the semiconductor industry is actually falling. It is now at 65.4% as compared to its recent peak of 70.7% in April, 2010.”
“Moreover, the growth in semiconductor capacity is slowing significantly. I do not know if this is due to foreign competition or industry executives expectations of weak demand growth. But neither alternative is bullish. Note that back in the 1990s boom the capacity to build semis was doubling almost every year as compared to only a 10% growth rate this year.”
I see no reason to expect that U.S.-based production of semiconductors will increase substantially in light of projected sales growth and global sourcing decisions including the ramp up of semiconductor production in China, a market that also represents over 46% of global semiconductor sales.
Semiconductor sales are up substantially over 2009 levels, but sales growth is projected to slow dramatically in 2011 and 2012.
A few years ago, the Semiconductor Industry Association (SIA) stated, “The industry has a 17% compound annual growth rate for the past 40 years and we expect that to continue for the foreseeable future despite periodic cycles.” The SIA is now “projecting record sales of $300.5 billion in 2010, an increase of 32.8%. The forecast calls for sales to grow by 6.0% in 2011, to $318.7 billion, followed by an increase of 3.4% to $329.7 billion in 2012. The projected compound annual growth rate is estimated to be 13.4% for the period 2009 through 2012.”
Here is semiconductor global sales data provided by the SIA.
Semiconductors
Total Global Sales, End of Year
Year to Year sales comparison
Billions
1999 $149.0 18.9%
2000 $204.0 36.9&
2001 $138.9 -31.9%
2002 $140.7 1.3%
2003 $166.4 18.3%
2004 $213.0 28.0%
2005 $227.5 l6.8%
2006 $247.7 8.9%
2007 $255.6 3.2%
2008 $248.6 -2.8%
2009 $226.3 -9.0%
2010 $300.5 32.8% projected*
*Projections provided in Nov 2010
—-
Semiconductors
Total Global Sales
Month to Month, 2010 to 2009
Billions
Jan $22.5 $15.3 47.2%
Feb $22.0 $14.1 56.2%
Mar $23.1 $14.6 58.3%
Apr $23.6 $15.7 50.4%
May $24.7 $16.7 47.6%
Jun $24.9 $17.2 43.6%
Jul $25.2 $18.4 37.0%
Aug $25.7 $19.4 32.6%
Sep $26.5 $20.1 31.8%
Oct $26.3 $22.0 19.8%
All monthly sales numbers represent a three-month moving average.
—–
October 2010
In $ billions of sales
Month-to-Month Sales
Market…Last Month…Current Month…%Change
Americas 4.81 4.81 0.0%
Europe 3.25 3.36 3.3%
Japan 4.20 4.22 0.3%
Asia Pacific 14.06 13.93 -0.9%
Total 26.33 26.32 0.0%
Year-to-Year Sales
Market…Last Year…Current Month…%Change
Americas 3.68 4.81 30.7%
Europe 2.81 3.36 19.6%
Japan 3.75 4.22 12.5%
Asia Pacific 11.73 13.93 18.8%
Total 21.96 26.32 19.8%
—–
SIA
spencer – “Over the past year total industrial production rose 5.4% while semiconductor output rose 11%. But by historic norms this is very weak as the historic average growth rate for semis output growth is about ten times the growth rate of total industrial production.”
“As the chart show, with semi output expandingonly some 11% it is so weak that capacity utilization in the semiconductor industry is actually falling. It is now at 65.4% as compared to its recent peak of 70.7% in April, 2010.”
“Moreover, the growth in semiconductor capacity is slowing significantly. I do not know if this is due to foreign competition or industry executives expectations of weak demand growth. But neither alternative is bullish. Note that back in the 1990s boom the capacity to build semis was doubling almost every year as compared to only a 10% growth rate this year.”
I see no reason to expect that U.S.-based production of semiconductors will increase substantially in light of projected sales growth and global sourcing decisions including the ramp up of semiconductor production in China, a market that also represents over 46% of global semiconductor sales.
Semiconductor sales are up substantially over 2009 levels, but sales growth is projected to slow dramatically in 2011 and 2012.
A few years ago, the Semiconductor Industry Association (SIA) stated, “The industry has a 17% compound annual growth rate for the past 40 years and we expect that to continue for the foreseeable future despite periodic cycles.” The SIA is now “projecting record sales of $300.5 billion in 2010, an increase of 32.8%. The forecast calls for sales to grow by 6.0% in 2011, to $318.7 billion, followed by an increase of 3.4% to $329.7 billion in 2012. The projected compound annual growth rate is estimated to be 13.4% for the period 2009 through 2012.”
Here is semiconductor global sales data provided by the SIA.
Semiconductors
Total Global Sales, End of Year
Year to Year sales comparison
Billions
1999 $149.0 18.9%
2000 $204.0 36.9&
2001 $138.9 -31.9%
2002 $140.7 1.3%
2003 $166.4 18.3%
2004 $213.0 28.0%
2005 $227.5 l6.8%
2006 $247.7 8.9%
2007 $255.6 3.2%
2008 $248.6 -2.8%
2009 $226.3 -9.0%
2010 $300.5 32.8% projected*
*Projections provided in Nov 2010
—-
Semiconductors
Total Global Sales
Month to Month, 2010 to 2009
Billions
Jan $22.5 $15.3 47.2%
Feb $22.0 $14.1 56.2%
Mar $23.1 $14.6 58.3%
Apr $23.6 $15.7 50.4%
May $24.7 $16.7 47.6%
Jun $24.9 $17.2 43.6%
Jul $25.2 $18.4 37.0%
Aug $25.7 $19.4 32.6%
Sep $26.5 $20.1 31.8%
Oct $26.3 $22.0 19.8%
All monthly sales numbers represent a three-month moving average.
—–
October 2010
In $ billions of sales
Month-to-Month Sales
Market…Last Month…Current Month…%Change
Americas 4.81 4.81 0.0%
Europe 3.25 3.36 3.3%
Japan 4.20 4.22 0.3%
Asia Pacific 14.06 13.93 -0.9%
Total 26.33 26.32 0.0%
Year-to-Year Sales
Market…Last Year…Current Month…%Change
Americas 3.68 4.81 30.7%
Europe 2.81 3.36 19.6%
Japan 3.75 4.22 12.5%
Asia Pacific 11.73 13.93 18.8%
Total 21.96 26.32 19.8%
—–
SIA
MG:
Spencer is talking capou are discussing sales dollars.acity of units and you are discussing Sales dollars.
Spencer is talking capcityand you are discussing sales dollars.
run,
Why do think we need capacity if we don’t have sales? The two measurements are very related.
run,
Why do we need capacity if we don’t have sales? The two measurements are very related.
MG — the Fed does the output and capacity calculation and they work closely with the semiconductor
association and base most of their estimates on data collected by the SIA.
Much of the government data is actually data they get from the various trade associations.
Interestingly the Census orders and shipments numbers used to break out the detailed numbers on
semi shipments, orders, backlogs and inventories and it was also based on the semiconductor
association data. But a few years ago the association required the government to quit breaking out the
detailed semiconductor data although it is still included in the aggregate data provided by the government. The association did this so that they could sell the data to subscribers and make money to
pay for the association operations rather than depending on member dues.
There is no conflict between the data you present and my data.
run,
The 12+ mln figure is roughly equal to sales – domestic and imported. Sales of domestic autos – which I’m gonna guess is pretty close to output – has run at just under 8.8 mln (monthly average over 11 months) so far this year. Gotta hope domestic capacity is way below 14 mln.
apencer,
I am familiar with the methodology used by the Federal Reserve. SIA occasionally publishes a breakdown of semiconductor sales by type in the public domain. Some other outfits want $1,500 or more for the breakdown and projected sales data.
spencer,
I am familiar with the methodology used by the Federal Reserve. SIA occasionally publishes a breakdown of semiconductor sales by type in the public domain. Some other outfits want $1,500 or more for the breakdown and projected sales data.
kharris:
Makes sense. Thanks
MG:
The demand is there; but, the Mosfet industry is controlling capacity purposely. We have see prices increase 25 to 50% in some cases because there is no where else to go. The entire industry is on allocation with lead times from 26 to 99 weeks. The capacity increases you are talking about are not that big in comparison to demand. 2011 has much of the same “allocation” as 2010.
Frankly, the entire semiconductor industry is merrily engaged in price-fixing. Demand is there, sales are there, profits are there. Supply is not.The fact is, most classes of parts have only a few manufacturers now from the mergers-and-acquisitions craze in the last 10 years or so. Competition no longer exists.
It’s seriously hindering development of new gadgets now.
MG:
Example: ST increased the price of the STB200 and the STB80 by 50%. Sales dollars went up, capacity remained the same, and we still ordered the same quantity to meet our demand. We have seen the same on the Infineon BCR135WE6327XT is in the same mode of increased pricing and no capacity additons.
They are not bringing on capacity to meet demand and are just lengthening lead times. Take a ticket and wai for your number to be called for your allocation.
Yepppers!!!