We estimate the effects of works councils on productivity, 1997-2000, using the IAB Establishment Panel, a nationally representative German data set. We recoup the works council effect by estimating translog production functions, stochastic frontier production functions, and a model in first differences. Once we focus on a core sample of establishments with 21 to 100 employees in which the powers of the works council are a datum, it emerges that the positive productivity differential found in recent studies is a chimera. By the same token, neither is the works council effect negative. This result is important in its own right given the sharply opposing findings of past empirical research and the partisan positions these have helped sustain.