Learning by Doing

See allHide authors and affiliations

Science  13 Dec 2013:
Vol. 342, Issue 6164, pp. 1292
DOI: 10.1126/science.342.6164.1292-c

Many have championed the importance of formal education for industrial innovation and technical change in a modern economy. On-the-job training, however, or learning by doing, is critical as well. Despite this importance, theorists and managers still lack fine-grained details about how such knowledge is acquired, aggregated, transmitted, and embodied in a production process. Levitt et al. studied an auto assembly plant that was, in many respects, starting over, having just completed major changes in assembly line layout, in types of vehicle assembled, and into new production teams. Over the course of a year, they tracked hundreds of line workers and the 189,000 vehicles they made. They measured hundreds of features on each vehicle, from the torque applied to a bolt to defects in the fit of parts, and the timing of each production step. Defect rates and production times both improved dramatically over the first 7 weeks, at which point a second shift began working on the same line. Despite minimal overlap between first- and second-shift workers, second-shift defect rates began, and remained, slightly better than for the first shift, who had a 7-week head start. Features that were more defect-prone in shift one were similarly more defect-prone in shift two. Defect rates and production time spiked up again when new styles of vehicle were introduced. The authors discuss the importance of institutional systems to harness and apply this knowledge, and speculate as to why workers might voluntarily “overshare” their hard-earned and valuable knowledge with management.


J. Politi. Econ. 121, 643 (2013).

Navigate This Article