The Productivity Effect of Non-Union Representation

Declining union density in many industrialized countries raises the interest in alternative ways of labor relations and worker representation. Provided that workplace representation has desirable social consequences; for a policy maker its effect on firm productivity is crucial for evaluating its usefulness. German works councils are the most powerful form of institutionalized work place representation in industrialized countries and are thus a natural starting point for productivity evaluation. In Germany, the existence of a works council is widely observed to be positively correlated with productivity. However, the estimated productivity effects vary between zero and up to 30 percent. Existing studies suffer from various shortcomings due to the fact that a firm’s capital stock is not controlled for in the estimated production functions, and often no convincing strategy to avoid endogeneity problems has been adopted. To fill these gaps, a recently developed method for capital stock approximation is applied in this study and panel estimators are used to deal with the potential biases that may arise from unobserved firm-specific effects and the simultaneity of the input decision. An Oaxaca-Blinder decomposition answers the question to which extent different productivity or different endowments can explain the output gap between firms having no council and those who have. I find that differences in productivity amount to 4 percent in favor of the latter. Because of considerable self-selection into the observed works council regime, I claim this estimate to be the lower bound for the average productivity effect.