We examine the effect of plant managers on productivity using unique matched manager-plant panel data on U.S. auto-assembly plants during 1993–2007. Our econometric approach is two-pronged. Our first approach relies on using the panel nature of our data to measure variation in productivity due to managerial influence. We estimate the interquartile range of the effect of individual plant managers on average hours-per-vehicle to be about 30%. Further, we find that plant managers’ experience with the models that are in production ameliorates the negative impact of new model introductions on productivity. We also observe evidence that managers’ plant-specific tenure has a positive impact on productivity. In our second approach, we use high-frequency time-series data, along with structural-break tests and machine-learning methodologies, to predict variation in production using plant-manager switches. We find that a plant manager’s identity is predictive of changes in both the mean and variance of production, further highlighting their channels of managerial influence. These findings are robust to narrowing the sample to focus on retirements as an exogenous source of managerial switches. This paper was accepted by Serguei Netessine, operations management. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2022.4427 .
Read full abstract