Thursday, October 13, 2011

Manufacturing Metrics

I am a bit fascinated that a Republican presidential candidate could, or would want to, visit a steel plant in Western PA as Governor Perry is doing tomorrow.  Again, it has been a long time since the early 1980's when I suspect such a visit would not have been a positive photo op.  Also...  maybe manufacturing is coming back is a theme you hear a lot. On that NPR today has this: Gas Drilling Boom Brings New Life To Steel Industry.

Which together got me thinking.  A basic metric we teach for how concentrated an industry is in a region is a Location Quotient.  Check the link if you want the long explanation, but the LQ is a measure that is more than 1.0 for an industry if it has a higher concentration than the national average.  Less concentration than average you get a number less than 1.0. You compute a LQ for distinct industries or occupations typically. LQ analysis has been around a long time, but it is not far removed from the more in vogue of late 'cluster analysis' the EDA director mentioned in his oped the other day. Old is new again....  except for Pittsburgh it seems.

Suffice it to say that for most of its history, Pittsburgh has had a LQ for manufacturing industries higher than 1.0.  Far higher than 1.0 actually.  For specific industies such as steel far far higher than 1.0.  So I just calculated the current LQ for manufacturing in Pittsburgh and it works out like this with employment numbers in thousands:

So not only are we not the manufacturing region we once were... the most recent data shows we have a LQ less than 1.0!  I know that's an esoteric stat for most, but it really is deeply telling factoid. When I calculated this for 2007, not that long ago, we were at least at parity with the US with a manufacturing LQ exactly 1.0.  It's not that manufacturing has declined nationally, but Pittsburgh's relative concentration of manufacturing employment has continued to decline to the point where we are now well below average. So if it is coming back for Pittsburgh, it has quite a ways to go.  While the NPR article points out some of positive points in local manufacturing industries, there are at the same time continuing hits that net out a lot of the impact.


Blogger Mike Madison said...

Is there related data on manufacturing output? In other words, is Pittsburgh producing less stuff, or is Pittsburgh producing more or less the same amount of stuff (or more or less the value of stuff) but with fewer people?

I ask because I was talking with someone recently who is in a position to know details about electricity consumption in the region. I asked whether power usage had dropped since mid-2008, and I asked specifically about commercial customers -- the sorts of users where manufacturing would be clustered. The answer: there hadn't been much of a decline. There's a lot of noise in that comment, but his implication was that Pittsburgh's manufacturing base, such as it is these days, is using about as much power as it did before the economy went in the tank.

Thursday, October 13, 2011 2:19:00 PM  
Blogger C. Briem said...

The short answer is that over both the long and short term labor force productivity continues to go up. So yes, we make a lot more in output per worker than ever in the past. We could parse the manufacturing GDP data to narrow in on that a bit... but sure it's possible overall electicity use has not declined much at all in recent years.

That is a bit irrelevant to the LQ calculation though since labor force productivity is going up nationally and I just used the employment numbers in the calculation.

There may be better electricity data, I'll see. It's hard to break down below the state level just because of the way electricity is produced and sold.

Thursday, October 13, 2011 2:37:00 PM  
Anonymous BrianTH said...

Of course manufacturing's LQ can go down even if that sector is expanding if other sectors are expanding even more rapidly.

Not to be cute, but for example, if the only two sectors were energy and manufacturing, then a primary boom in energy could indeed be causing a secondary boom in manufacturing, but the manufacturing LQ could also be going down at the same time if the primary boom in energy was more rapid.

And in fact I am personally convinced that if a primary energy boom is creating secondary booms, there is no particular reason to expect those secondary booms to be limited to manufacturing (there are lots of candidates both upstream and downstream). Meanwhile, energy is not the only candidate for a primary boom in the Pittsburgh region.

So I guess I am not sure what LQ tells us in this particular situation.

Thursday, October 13, 2011 3:43:00 PM  
Blogger C. Briem said...

It is by definition a relative measure, so yes if (big if) everything else is spontaneously going up you could see a particular LQ down. Not sure that is the case here. It's the big chunks of manufacturing that have gone away in recent years. Take the final closing of Sony. Electronics manufactuing was the only growth sector in local mfg employment for many years. That turned off dramatically. That one zinc plant in the news recently moving south is a decent sized place by current standards.

Thursday, October 13, 2011 4:17:00 PM  
Anonymous The Wiz said...

Mike Madison; Good point. I know that Lordstown, where the Chevy Cruze is manufactured, has less than half the workers it used to have but still makes roughly the same number of vehicles. Much is out sourced but they also have a huge number of robotics. Whole sections of the line have no one is sight but robots. Bet the same is true for many manufacturers.

As for electrical use, many mills use electric arc now and computer/server rooms use large volumes of juice so I am not sure how closely that correlates.

Thursday, October 13, 2011 4:33:00 PM  
Blogger C. Briem said...

I'm pretty sure we have not opened up a new electric arc minimill in the region in the last couple years. Have some great maps we've of all the steel mfg in the region.

So nationally sure.. more electricity for electric arc plants than integrated steel plants. Not our story though. Same for the data server answer.. not sure many are being built here.

Thursday, October 13, 2011 4:45:00 PM  
Blogger C. Briem said...

But here is the time series for Pittsburgh MSA GDP.. GDP being a measure value added it should represent what is going on here. It also has a certain bias (wrong word) toward manufacturing. But the bottom line is while there was a drop in 2009, there was a big rebound in 2010. While these are nominal #'s, inflation has been relatively low:


2000 75,081,990
2001 76,823,437
2002 78,130,434
2003 79,690,488
2004 82,855,306
2005 84,955,826
2006 91,744,589
2007 96,492,165
2008 100,276,257
2009 99,610,767
2010 103,039,425

Thursday, October 13, 2011 4:54:00 PM  
Anonymous BrianTH said...

I wasn't trying to deny that manufacturing employment has been contracting in recent years--we know that straight from the manufacturing employment series. On the other hand, as I recall manufacturing employment is tracking its biggest year-over-year gains in a long time (not much of a statement when it has mostly been declining year-over-year, but still).

So I guess I am still unclear on what we are trying to get out of LQ that we aren't getting out of that series. I certainly think you could have a definition of a manufacturing recovery in the region that didn't require an LQ recovery, again in circumstances where other sectors were also growing.

In other words, in my mind a local manufacturing recovery is not inconsistent with the local economy also continuing to diversify, as opposed to becoming more concentrated in manufacturing once again. And I would even dare say that is the preferable outcome.

Thursday, October 13, 2011 5:46:00 PM  
Blogger C. Briem said...

If you just look at the time series, the how can you tell how much of the local trend is just based on overall industry trends?

What you want to avoid in measuring regional competitiveness is to confuse exogenous trends that would be impacting a specific industry in a region no matter what is going on in the region.

So generally speaking, a high LQ will tell you what a region is good at producing.. what it is competitive at. What has a low LQ are more likely things it is not competitive at producing.

Thursday, October 13, 2011 7:40:00 PM  
Blogger C. Briem said...

or to expand on that. LQ analysis is just a first step is called shift share analysis.

What if Pittsburgh manufacturing employment expanded by 10%.. but national mfg emoployment went by 20% over the same period. Would we be gaining in competitiveness or declining?

Thursday, October 13, 2011 7:44:00 PM  
Blogger Jim Russell said...

Somewhat off topic from the LQ discussion, I honed in on what Vallourec's rep had to say in the NPR piece. Mastervich mentions infrastructure and experienced workforce before getting to the proximity issue (i.e. Marcellus Shale).

Vallourec, if I remember correctly, was looking at Oklahoma and Brazil as well as Youngstown as possible sites for the steel pipe plant. My point being that looking at just the Pittsburgh MSA is myopic. There's a bigger talent geography (and LQ) geography to consider.

Thursday, October 13, 2011 9:32:00 PM  
Anonymous BrianTH said...

I can certainly understand why you would want to compare manufacturing employment trends in this region to manufacturing employment trends elsewhere (national, peer regions, or so forth). But you don't need LQ for that.

I can also understand why you would use LQ to answer questions like, "what are our most competitive sectors as compared to each other at this time?"

But these are two different questions, and it seems to me it would be easy to get confused if you tried to use LQ for both purposes.

To give a variation on my prior example, suppose nationally there was a huge relative slump in some sector--say construction--but there was some locality which managed to dodge that slump. Assuming every other sector in that locality was exactly tracking the national trends, if looked at LQs, you should see the construction LQ in that locality rising, and the LQs of everything else falling.

But those falling LQs don't mean those other sectors are growing less than average, and similarly they don't mean those sectors in that locality are getting less competitive in comparison to other places. They just mean construction is gaining in relative competitiveness in that locality, all because construction elsewhere is slumping.

In short, all these are different ways of saying I don't think we should be automatically discounting a local manufacturing recovery if other local sectors are outperforming national employment averages--something which we know is likely happening because we are outperforming national employment averages when you total up all non-farm sectors. And that is what LQ will do.

Friday, October 14, 2011 1:27:00 PM  

Post a Comment

<< Home