Macroeconomic Implications of Demographic Change
Age and Labor Productivity in Manufacturing
Demographic change will bring in its wake a massive aging of manpower over the next 20 years. Against this background, we have set out to explore the relation between worker age and labor productivity. For this, we have compiled a unique data from a truck assembly plant owned by a large German car manufacturer with plants in Asia, Europe and the U.S. At this plant, trucks are assembled by work teams on an assembly line. We have selected this plant because it features a taylorized production process typical for the manufacturing industry, and because it stacks our cards against finding flat or increasing productivity with age. Compared to many service-sector jobs, productivity in this plant requires more physical strength, dexterity, agility etc. (which tend to decline with age) than experience and knowledge of the human nature (which tend to increase with age). These data permit us to overcome the above-mentioned methodological problems in an unprecedented way. The data have three innovative elements. First, we measure productivity in an assembly line environment in which the time to produce a unit of output is as standardized as the quality of the final product. As the assembly line has the same speed for all work teams and the design of the trucks is pre-defined, more productive work teams are not able to produce more or better output than less productive work teams. Workers, however, make errors which are detected at end control. More productive work teams differ from less productive work teams only in the errors they make. We therefore use the number and severity of production errors during the assembly process as a precise and well-observed measure of productivity. We exploit the daily variation in the team composition of work teams over four years to identify the age-productivity profiles. Second, we have merged the daily production error data (almost 1000 days) with longitudinal personnel data (3,800 workers in 100 work teams). This permits us to hold a broad range of workers’ characteristics constant. In addition, and most importantly, by differencing out worker-workplace fixed effects we are able to correct for the selection effects marring so many earlier studies due to the endogeneity of early retirement and team composition. Third, we measure the joint productivity of workers in a work team. This takes into account the individual workers’ contribution to their co-workers’ productivity. Particularly the contribution of older workers may be underestimated if productivity is measured at an individual level. Examples for such potential contributions to a team’s productivity are the instruction of younger workers, being relaxed in tense or hectic situations, and contributing positively to the work climate. We think that our approach solves the major aggregation problems in earlier studies. Our results are striking. Due to the very large number of observations and our identification strategy, we are able to estimate rather precise age-productivity profiles at the individual level and at the level of a work team. These profiles do not show a decline in the relevant age range between 25 and 65 years of age. On the individual workers’ level, our average productivity measure actually increases monotonically up to age 65. This project has been successfully completed with a paper published in the Journal of the Economics of Aging.