Chaos Before Order: Productivity Patterns in U.S. Manufacturing

Within-industry productivity dispersion is pervasive and exhibits substantial variation across countries, industries, and time. We build on prior research that explores the hypothesis that periods of innovation are initially associated with a surge in business start-ups, followed by increased experimentation that leads to rising dispersion potentially with declining aggregate productivity growth, and then a shakeout process that results in higher productivity growth and declining productivity dispersion. Using novel detailed industrylevel data on total factor productivity and labour productivity dispersion from the Dispersion Statistics on Productivity dataset along with novel measures of entry rates from the Business Dynamics Statistics and productivity growth data from the Bureau of Labor Statistics for U.S. manufacturing industries, we find support for this hypothesis, especially for the high-tech industries. An increase in entry rates in a two-year period t is associated with an increase in dispersion and decrease in aggregate productivity growth in two-year