We are still the world's workshop

Increasing productivity is less useful if it is not distributed outside of the managerial group:

http://www.census.gov/hhes/www/income/histinc/IE-1.pdf
http://www.demos.org/inequality/images/charts/top1percent.tif

This is all fine and dandy, but Canada beat the US in men's hockey earlier today.

I suggest that what creates the view of declining manufacturing is that the US produces fewer consumer goods relative to the past and more industrial goods and tools and the like. So most people see things that are made outside the US and therefore believe that US manufacturing is declining.

Another example of how common wisdom can be reliably be counted on to be wrong. Most of the jobs lost in manufacturing has been due to automation (this has actually occurred in China as well).

This is because the primary role of people from the end of the industrial revolution to the 1980s in production was computation. The industrial revolution replaced people and animals as the chief means of energy and work (the actual physical handling of material). The remaining critical component they could not replace was computation. Farming requires a significant amount of computation of course, but the demand for food is ultimately finite. This worked out well because most human activity was tied up in food production. Industry no longer needed people for energy or a lot of the work, but it did need something to direct its tools. With the vastly wider array of activities now possible due to the lifting of the energy and work constraints, industry was relatively easily able to absorb those displaced in agriculture. This was essentially the economic order for most of the 19th and 20th centuries.

At some point in the 80's computational technology became powerful enough to begin replacing humans in this role. This is evident in how the trend reverses itself with the early 80's recession in the second graph, where manufacturing employment never recovers to its pre-80's peak. It has continued to decline ever since.

The problem is that computation is not simply restricted to production, it is critical in all human activity. Indeed, most people fill either what are essentially computational or human service rolls. The other major role, creativity/problems solving, needs and involves significantly fewer people and the percentage of the population that can even do economically useful work in this area is restricted (and many of those able to do it tend are often in highly skilled, but what are essentially computational jobs). So, more and more people will be forced into human services (which many are not well suited for) and the remaining (but shrinking) parts of production that are not easily automated.

Human service roles are likely to be limited due to it being unnecessary for anything like a 1:1 worker/customer ratio. So, we'll have a growing part of the population unemployed and unemployable. This will likely be the single greatest challenge of the 21st century by far - what happens when you've rendered a significant fraction of the population economically useless?

Sorry for the long post, just needed to get that out of me and this seemed like a relevant post to do it. :)

*As a side note: I know many economists refer to this as the Luddite fallacy. What is missed is that, while demand may be infinite (debatable), the roles that humans can fill are not and we have developed much more efficient means of filling much of them. This is likely why both monetary and fiscal policy is not affecting jobs as most economists expect: economic supply (and a fair amount of demand) has decoupled itself from the total population.

Those of us who use absolute terms instead of relative ones are admittedly at fault when we bemoan the loss of manufacturing. Still, the total industrial output from 1986 thru the peak in 2008 grew at only a 2.7% annual rate - not a terribly impressive number (overall GDP growth was 3% per year over that time). (Data here).

Looking at durable goods manufacturing we get a brighter picture - a 4.5% annual rate. But when you look at the details, it gets fuzzier. Here are the annual rates of growth over that period for the components of this index:

Durable Goods total4.5
Nonmetallic mineral product1.2
Primary Metal1.2
Fabricated Metal1.8
Machinery1.8
Computer and electronic14.7
Electrical equipment and appliance1.0
Motor Vehicles2.4
Aerospace and Misc Transp0.9
Furniture and related1.4
Miscellaneous3.7

So when you look at the pieces, it's a pretty anemic growth rate, except for "Computer and electronic product" which grew at 15% annual rate! I couldn't for the life of me figure out the weighting methodology, but the simple average of these components yields a 3% growth rate. By holding each of the other components at a weight of 1.0, you can get to the 4.5% overall growth rate by giving the "Computer and electronic" component a weight of 2.1. So while the "U.S. manufacturing is dying" meme may be greatly exaggerated, the fact that the non-computer components are growing significantly more slowly than the economy overall isn't all that rosy, either.

Yes, look at the graph here. The graph that shows manufacturing as a percent of GDP vs. manufacturing employment. It shows that its percent of GDP stayed the same and employment dropped.

http://mpettis.com/2009/12/the-pace-of-change/