Leave the Average Productivity Behind

Official measures of U.S. manufacturing productivity are overstated. Productivity at the extremes is the real story.


Facebook Share Icon LinkedIn Share Icon Twitter Share Icon Share by EMail icon Print Icon



American manufacturing is not as productive as we have imagined it to be. Official statistics for overall U.S. manufacturing productivity are inflated, and arguably these statistics do not measure productivity at all. That’s OK, though. The discrediting of these numbers need not be disheartening, because the important measures of productivity relate to individual manufacturing firms.

The fallacies of the U.S. government’s productivity metrics were summarized in a recent report on manufacturing from the Brookings Institution. One of the biggest blind spots in these statistics relates to supply chains, the report says. Suppose a production plant realizes more net income per manufacturing employee than it once did. That gain might have come from the plant implementing automation and/or lean techniques that make every employee’s efforts more efficient. On the other hand, that gain might instead have come because components coming into the plant from another link in the supply chain have gotten cheaper as a result of that production being offshored. Either development could show up as “productivity” in the official government numbers.

According to those numbers, U.S. manufacturing saw a 5.4 percent annualized growth rate in productivity from 1997 to 2007. If one corrects for the offshoring error, Brookings says, and corrects for other false contributors that the report describes, then the actual rate of manufacturing productivity growth is found to be 2.3 percent.

Oh well. As interesting as this is, I found a seemingly unrelated finding within the same report to be more interesting still, and more directly relevant to manufacturing’s success. The report's authors measured a phenomenon that I have observed—namely, the way that manufacturing efficiency varies dramatically from plant to plant, even among manufacturers in similar fields. The Brookings analysis compared automotive metal stampers, finding that the leading 10 percent of them (that is, a lot of stampers) had productivity that was better than twice that of the median in the sample. In other words, the spread is that great among stampers—and I believe the spread would be found to be just as great among mold shops, contract shops, assembly operations and captive machining facilities. 

That spread tells the story. Rather than focusing on the mean of manufacturing productivity, what if we focused on the standard deviation?

The size of the performance spread indicates that large numbers of innovative and determined U.S. manufacturers are achieving levels of productivity well ahead of what their peer companies routinely accept. To put the matter bluntly: Copy those leading shops! If more U.S. manufacturers followed the models of leading manufacturers in their fields, then our real productivity would be even greater than what we once imagined.