To listen to most discussions of U.S. manufacturing these days one would not be faulted for believing that we have turned the ship of decline around, that production jobs are coming back and that the on-shoring good times are finally here.

We see this optimistic theme almost daily. In a recent article in The Atlantic, entitled “Comeback: Why the Future of Industry is In America,” Charles Fishman argues that the outsourcing wave is largely over and now U.S. companies, exemplified by GE’s appliance division doing more work in the U.S., are seeing the light and moving work back home.

Ryan Galloway writes in Forbes that “For the first time in decades, the future of American manufacturing looks promising.”  And of course the Boston Consulting Group’s Hal Sirkin, who started all this talk about manufacturing reshoring, writes that when it comes to manufacturing locating in America, “the math these days is starting to work in America's favor again.”

But to paraphrase Mark Twain, the reports of manufacturing’s renewal are premature, at best.  

Let’s start with data on inflation-adjusted trends in manufacturing value added. Bureau of Economic Analysis (BEA) data at the three-digit NAICS level are available only through 2011. According to BEA, from 2007 (the year before the Great Recession) to 2011, real GDP increased by 1%, but with the exception of the computers and electronics industry, real value added declined in every major industrial sector; from a 31% decline in furniture to a 1.5% decline in machinery.

Surely motor vehicles must be up, though, given the Big Three’s comeback. In fact, real output in motor vehicles was 28% below 2007 levels.

Surely chemicals must be up given the fracking revolution and decline of natural gas prices. In fact, real output in chemicals was 15% below 2007 levels. 

Only NAICS 334, computers and electronic equipment, was up, by 44%.

But as ITIF showed in its report Worse Than the Great Depression, this is due to the particular and misleading way BEA measures output, attributing “Moore’s law” improvements in computer processing speed and storage to actual production output. In other words, when a computer doubles in processing power and storage in two years, BEA data indicate that industry output doubled.

2012 data are only available for non-durables (this includes industries like chemicals, plastics, foods, printing, etc.) and durables (e.g., transportation equipment, machinery, primary metals) as major groups.

The trends here are better, but hardly indicative of a reshoring tsunami. Non-durable goods’ real output in 2012 was still 10% below 2007 levels.

Durable goods are up 8%, but all of this increase and more is due to the NAICS 334 overstatement. Taking out NAICS 334 (by using the presumably lower 2011 value added levels) the real value added of durables was actually 3.5% lower in 2012 than it was 2007.