Comment on "Bottlenecks: Sectoral Imbalances and the US Productivity Slowdown"

You may be able to download this chapter for free via the Document Object Identifier.

Acemoglu, Autor, and Patterson’s paper is thought-provoking and full of rich insights. They propose that TFP growth among industry suppliers has become less balanced over time which, in turn, has reduced innovation spillovers. These reduced spillovers became an increasing drag on growth in the 1977-2007 period. In this comment, we first argue that they are silent on the important changes in trend TFP growth that are apparent since World War II—the slowdown after the early 1970s, the speedup in the mid-1990s, and the subsequent slowdown in the mid-2000s. The issue is that the NBER-CES productivity dataset at the core of their analysis looks very different from other datasets. Second, we explore whether factor-share anomalies in their data affect their results; if anything, correcting those anomalies strengthens their apparent results. Nevertheless, given the unusual properties of their data, we hesitate to give it too much weight on its own. Third, we repeat their analysis with higher quality but, unfortunately, less granular, industry data. These results look very different: Unbalanced upstream innovation is, if anything, good for growth. We conclude that the detrimental effects of unbalanced innovation are an intriguing conjecture but fall short of proven fact.