Learning by Doing, Worker Turnover, and Productivity Dynamics

Recent evidence based on longitudinal firm-level data suggests that within-firm productivity growth explains about 50 percent of total factor productivity growth in the manufacturing sector while net entry effects account for about 30 percent of total factor productivity growth. These two forces may be connected via learning by doing of young businesses. That is, the recent evidence also shows that young businesses that survive exhibit more rapid productivity growth than older incumbents. The idea that learning-by-doing is important and, in particular, important for young businesses is not novel to this paper. What is novel about this paper is that newly developed longitudinal employer-employee matched data are used to characterize and measure the nature of learning. In contrast to the existing literature on learning, this paper shows that learning is not only affected by past output, but also by worker turnover within firms. The basic idea is that firms with high worker turnover will make learning-by-doing more difficult. Using this approach, I estimate that firms with historically lower rates of turnover "learn" faster than those with higher turnover given the same amount of past outp