The Hidden Decline of Manufacturing Output
Imagine a Fortune 500 company trying to plot its strategy based on significantly inflated sales figures. Do you think its CEO would be able to make sound decisions about what is working well and what needs to change? Of course not. Yet that is exactly the position President Obama and other federal policymakers are in when they look at U.S. manufacturing: They are seeing a picture that is far rosier than reality. That is because the Department of Commerce is using a badly flawed methodology to report manufacturing output. As a result, policymakers, pundits, and the media are falsely concluding that U.S. manufacturing is fundamentally healthy.
First, some background. In order to assess the health of U.S. manufacturing, the most important number you need to know is how much production is growing relative to the economy. To understand changes in manufacturing output, statisticians have to control for inflation, and they have to measure "value added," not gross production. As the term implies, value added is the value that manufacturers add to the inputs they buy. For example, in a car, the value added is the difference between the price of the vehicle and the amount that the manufacturer had to pay its suppliers for parts such as spark plugs and window glass. This way, we focus specifically on the output of U.S. manufacturers — counting the value of work that was done here by them, not by foreign manufacturers exporting parts to the U.S. or by domestic companies in other industries, like mining and electric utilities, providing inputs.
According to official government data, U.S. manufacturers produced the same amount of real value added in 2013 that they produced in 2006, even though manufacturing employment was down 15 percent. The lesson we are supposed to draw is that while times might be tough on manufacturing workers who have lost their jobs due to automation, U.S. manufacturing remains healthy.
But something is terribly wrong with that picture. Because of the way the Department of Commerce measures manufacturing output, especially in the computer and electronics sector, where rapid changes in technology regularly make computers more powerful, the official data significantly overstate output growth. This in turn makes it look like all the lost jobs and shuttered factories across America were simply the result of firms becoming more efficient — just like 100 years ago, when farmers got more productive and shed agricultural labor. But if you remove computer and electronics output (known in official tables as NAICS 334) from the picture, it turns out that real U.S. manufacturing output fell by 6 percent between 2006 and 2013, while the overall economy grew 8 percent. This is hardly a sign of health.
You might ask, so what? If the 92 percent of U.S. manufacturing (by employment) is producing less, but a boom in computers and electronics is making up the difference, that’s a wash, right? But we actually aren’t booming in computers. According to the Department of Commerce, real NAICS 334 output increased 52 percent from 2006 to 2013. But this is not measuring what you and I would think of as increases in output. Companies in America are not producing 52 percent more computers and electronics.
From 1990 to 2010, the real gross output for computers and electronic products grew by 570 percent, 20 times faster than the growth rate for the rest of manufacturing. But nominal shipments of computer and electronic products from U.S. factories — not just the "value added," but the full price of the products sold — grew just 24 percent between 1992 and 2011, while between 2000 and 2010 they actually fell by about 70 percent.
So what’s going on? In a word — well, two — Moore’s Law: the prediction by Intel co-founder Gordon Moore that the number of transistors on a chip would double every 24 months while the price would fall by half. This rapid growth in computing power (including processing and storage), which has been a boon to the digital economy and consumers, makes it seem as if the industry is producing much more than it really is. In other words, if a company sells a computer worth $500 one year and then two years later sells another for $500 but it has twice the processing power, the government reports the company output as having doubled, even though it is still selling only one computer. This is why the government reports that output for computers and electronics grew by 570 percent in 20 years, because computing power got exponentially better during this time. While the rapid quality improvement may indeed accurately represent the increased computing value that consumers experience, it falsely implies rapid expansion of industry output.
This dynamic creates the misleading impression that overall U.S. manufacturing output is growing when in fact, when NAICS 334 output is properly measured, it is clear that manufacturing output is falling, by more than 10 percent over the last 15 years.
So when policymakers look at the top-line numbers on manufacturing output change, they are not seeing the real picture. This is why so many commentators in the media have parroted the false claim that manufacturing output has grown at the same rate as GDP. Just this week the New York Times stated that manufacturing output has not gone down, and therefore trade has not had an impact on manufacturing job loss.
So, what’s wrong with all this? If policymakers are lulled into a sense of complacency that all is well with U.S. manufacturing, they are much less likely to push for the steps needed to restore U.S. manufacturing competitiveness, including a lower corporate tax rate and increased support for manufacturing-related government R&D. Thus, it's critical for Federal Reserve economists, congressional support agencies, pundits, and journalists not to accept federal manufacturing data at face value; we need to do the critical analysis necessary to get a real picture. Otherwise, Washington will continue believing a myth.
Robert Atkinson is president of the Information Technology and Innovation Foundation.