Learning-by-Doing and Productivity Dynamics in Manufacturing Industries

This paper estimates a structural model of learning-by-doing. Treating production experience as a state variable, this paper provides estimates of the structural parameters obtained from the first order conditions arising from the plant’s maximization problem. Estimates are provided using data on 4-digit manufacturing industries and plant-level observations. Using aggregate industry data, the results indicate that estimated learning rates might be considerably lower than previous estimates. The results also reveal considerable variation in estimated learning rates, across broad industry groups, at both the plant-level and the 4-digit industry level. This implies that using results from existing studies that focus upon specific, narrowly defined industries or firms, may lead to misleading conclusions concerning the widespread importance of learning-by-doing for generating productivity dynamics within the manufacturing industry.