Changes in the Organization of Production

This thesis tests for a shift in the organization of production from mass to flexible production using data on the skill composition of manufacturing employment. Mass production involves high volume output of standardized goods using highly specialized, dedicated capital equipment and a division of labor that seeks efficiency by reducing to the bare minimum the number of tasks performed by each worker. Flexible production, on the other hand, refers to the production of specialized, or customized products made in small quantities using flexible capital equipment and workers who can quickly be productive at a wide and varying range of tasks. The much more demanding nature of the task structure for production workers in flexible production increases the importance of training costs relative to wages. Since skilled--more highly educated-workers are likely to be less costly to train, the shift to flexible production precipitates a relative shift in demand toward skilled workers. The testable implication of a shift to flexible production is, therefore, an increase in the employment share of skilled production workers in total manufacturing production employment. The results support the hypothesis that there has been a shift since the early 1970s from mass to flexible production in both the U.S. and western Europe. The thesis is organized in four chapters. Chapter one begins with qualitative evidence which illustrates the implications of changes in the organization of production for the learning and skill requirements of workers. Next, a formal model is developed in which the firm varies its employment mix between skilled and unskilled workers as a function of relative wages, training costs and the length of product life-cycles. The model yields labor demand functions by skill groups which are used to test for a shift from mass to flexible production. Chapter two contains the empirical tests of the model for the U.S. Data on employment composition by detailed occupation for the period 1977 to 1986 are presented which show a significant increase in the share of skilled production workers. Relative wage data show no decline for skilled workers. These facts, together, imply a shift in relative labor demand. Nonetheless, formal labor demand estimation follows using annual Current Population Survey data on manufacturing production workers for 1973 to 1985. The estimates demonstrate a shift in demand favoring skilled at the expense of less skilled workers. While the shift to skilled workers is derived as a test of competing production modes, the fact that such an up-skilling has taken place, both in the U.S. and elsewhere, is an important finding in its own right. The unskilled have suffered significant relative wage and employment losses in recent years. Change in the organization of production is a promising explanation for these losses that transcend national boundaries. Chapter three explores our model in the context of two countries with very different institutional structures: Germany and the United Kingdom. The first part contrasts the vocational training systems in Ger-nany and the U.K. and makes predictions, on the basis of our model, of te implications for the skill composition of manufacturing production employment in the two countries. The second part presents employment share and relative wage data from Germany and the U.K. which confirm the central predictions that: flexible production has for decades been used more in Germany than in the U.K. or the US; the poor quality of vocational training in the U.K., coupled with the use of strict craft job demarcations, forestalled a shift to flexible production until recently; and, there has been a shift in relative labor demand favoring more highly skilled production workers in both Germany and the U.K. in recent years. Chapter four generates further testable implications of flexible production using non-labor data. The first section considers increases in intraindustry trade (IIT) as a proxy for shortened product life-cycles. A ranking of two-digit industries on the basis of change in IIT is shown to correspond closely to a ranking of industries based on the extent of the relative shift to skilled production workers. The next section draws on recent theoretical work in which firms respond to uncertain product demand by either mass producing and using finished goods inventories to smooth production, or building to order (produce flexibly) to meet demand as it is realized. The ratio of order backlogs to finished goods inventories should rise if firms move from mass to flexible production. This prediction is confirmed for the period 1970 to 1987 using firm-level data from Compustat. These results support the shift to flexible production as the cause of the observed changes in labor demand. The last section of Chapter examines competing explanations for the shift in relative labor demand. Secular technological change, and the outsourcing of low skilled jobs abroad are both considered and found not to be important causes of the observed changes. Recent shifts in relative labor demand arising from capital-skill comnlementaries associated with new, information-intensive technologies is a more likely alternative. However, the evidence suggests that new technologies have been important enablers to changes stimulated by product market conditions. Chapter five recapitulates the main points of the thesis, and offers lessons for methodology and some broad policy recommendations on the subjects of assistance for unskilled workers and improvements in the U.S. vocational training system. Thesis Supervisor: Michael J. Piore Title: Professor of Economics