The Unassuming Economist

From a new working paper by Susan N. Houseman:

“Two stylized facts underlie the prevailing view that automation largely caused the relative decline and, in the 2000s, the large absolute decline in U.S. manufacturing employment: first, manufacturing real output growth has largely kept pace with that of the aggregate economy for decades, and second, manufacturing labor productivity growth has been considerably higher. These statistics appear to provide a compelling case that domestic manufacturing is strong, and that, as in agriculture, productivity growth, assumed to reflect automation, is largely responsible for the relative and absolute decline in manufacturing employment. Although the size and scope of the decline in employment manufacturing industries in the 2000s was unprecedented, many
see it as part of a long-term trend and deem the role of trade small.”

“That view, I have argued, reflects a misinterpretation of the numbers. First, aggregate manufacturing output and productivity statistics are dominated by the computer industry and mask considerable weakness in most manufacturing industries, where real output growth has been much slower than average private sector growth since the 1980s and has been anemic or declining since 2000. Second, labor productivity growth is not synonymous with, and is often a poor indicator of, automation. Measures of labor productivity growth may capture many forces besides automation—including improvements in product quality, outsourcing and offshoring, and a changing industry composition owing to international competition. Indeed, the rapid productivity growth in the computer and electronics products industry, and by extension in the manufacturing sector, largely reflects improvements in product quality, not automation. In short, the stylized facts, when properly interpreted, do not provide prima facie evidence that automation drove the relative and absolute decline in manufacturing employment.”

“It is difficult to parse out the effects of various factors on manufacturing employment, and research does not provide simple decompositions of the total contribution that trade and the broader forces of globalization make to manufacturing’s recent employment decline. Nevertheless, the research evidence points to trade and globalization as the major factor behind the large and swift decline of manufacturing employment in the 2000s. Although manufacturing processes continue to be automated, there is no evidence that the pace of automation in the sector accelerated in the 2000s; if anything, research comes to the opposite conclusion.”

“Manufacturing still matters, and its decline has serious economic consequences. Reflecting the sector’s deep supply chains, manufacturing’s plight contributed to the weak employment growth and poor labor market outcomes prevailing during much of the 2000s. Research shows that such large-scale shocks have persistent adverse effects on affected communities and their residents, though these costs rarely are fully considered in policy making (Klein, Schuh, and Triest 2003). In addition, because manufacturing accounts for a disproportionate share of R&D, the health of manufacturing industries has important implications for innovation in the economy. The widespread denial of domestic manufacturing’s weakness and globalization’s role in its employment collapse has inhibited much-needed, informed debate over trade policies.”