Claims about the “hollowing out” of American manufacturing are as common today as is sand on a beach. But as readers of this blog know, the data as conventionally gathered and reported contradict both the assertion that Americans “don’t make things anymore” and that America’s capacity to produce industrial outputs has been declining for decades. But are these data the results of manufacturing appropriately defined and of economic activities appropriately categorized? Such questions should be asked from time to time. What, exactly, is meant by a term such as “manufacturing”?
According to the U.S. government, manufacturing is “the mechanical, physical, or chemical transformation of materials, substances, or components into new products.” This definition is reasonable; I’ve no objection to it. But it is nevertheless a human contrivance. And so, if we contrive carelessly – or, worse, tendentiously – the data that we gather and report will give us highly misleading impressions of the underlying reality that we’re interested in better understanding.
If, for example, manufacturing were defined to involve only the manual production of handcrafts, then indeed America’s manufacturing sector would be small. If it were instead defined to include all commercial activities that involve physical transformations of inputs into outputs, the growing of rye and roses, and the raising of pigs and chickens, would be classified – along with steel production, automobile manufacturing, and watchmaking – as manufacturing, resulting in America’s reported manufacturing output today being even larger than is shown in the conventional report. The point is that manufacturing could be defined otherwise, and there surely are possible definitions narrower or broader than the definition now officially used that some serious, thoughtful, and informed people would find reasonable.
The above thoughts were front and center in my mind as I read, on the advice of Doug Irwin, Dartmouth economist Teresa Fort’s splendid paper in the Summer 2023 issue of the Journal of Economic Perspectives – a paper titled “The Changing Firm and Country Boundaries of US Manufacturers in Global Value Chains.” This paper is remarkably informed and informative. My purpose here isn’t to summarize it, but to point out a couple of things in it that especially caught my eye. Fort’s paper only further strengthens the conclusion that American manufacturing is thriving.
The first is this passage on page 35 (which follows this note on page 34: “The Census Bureau defines an ‘establishment’ as a physical location at which employment and payroll records are kept. A firm can thus have multiple establishments—and these establishments need not be classified in the same industry”):
An establishment’s industry is the primary means that government agencies and researchers use to identify manufacturing activity. US statistical agencies use the North American Industry Classification System, commonly referred to as NAICS (and described at https://www.census.gov/naics) to classify establishments. The guiding principle of NAICS is to assign an industry code to an establishment based on the main activities performed by its employees. By contrast, the earlier Standard Industrial Classification System (SIC) classified establishments that provided support services for other establishments of their firm to those establishments’ industry. For example, an R&D lab is always in Services under NAICS, but would have been classified in manufacturing under SIC if its R&D supported the firm’s manufacturing plants. US Census data transitioned from NAICS to SIC between 1997 to 2002 [DBx: Obviously she meant to write “from SIC to NAICS”], a period that coincides with China’s entry to the World Trade Organization, making this issue particularly relevant for research on globalization.
Note that a reported change in U.S. manufacturing output or employment might occur absent any change in actual economic activity. The cause of the reported change might well be due exclusively to a change in how data-gatherers and processors classify the data they gather and process.
The second thing in Fort’s paper that caught my eye for purposes of this post is this passage on page 43:
In 2010, the US Office of Management [and Budget]’s Economic Classification Policy Committee recommended classifying a factoryless goods producer as a firm that “outsources all transformation steps that traditionally have been considered manufacturing, but undertakes all of the entrepreneurial steps and arranges for all required capital, labor, and material inputs required to make a good” (Office of Management and Budget 2011). Moreover, the committee recommended reclassifying establishments that performed those related tasks into manufacturing for the 2012 Economic Census (Doherty 2015) to facilitate collection of additional information about use of inputs and sales by product, which are already part of the Census of Manufactures survey questions. However, this proposal was met with strong opposition from the US manufacturing lobby, and the reclassification effort was abandoned.6
6 For example, the director of industry research and technology at the Precision Machined Products Association stated, “We think it would be bad for policy makers to say, ‘Look at these numbers, we have great manufacturing.’” See https://www.wsj.com/articles/SB10001424052702303546204579439170777269630 .
Firms involved in activities conventionally classified as manufacturing have, on at least the one occasion mentioned here, succeeded in ensuring that the official data show that manufacturing in the U.S. is weaker than it would appear to be under a more reasonable classification of economic activities. Here we have a clear demonstration of how insidious rent-seeking can be: Work to ensure that the official data on manufacturing create the impression that American manufacturing is weaker than it really is in order to have better prospects of persuading the public and policymakers that government should bestow special privileges on conventionally defined manufacturing firms.
