We investigate the consequences of an unexpected pay change in a personnel search firm. Top management�s decision to adjust the pay of some divisions� workers to the pay of other divisions resulted in a quasi-experiment increasing the fixed wage but decreasing the slope of the bonus function. Consultants� output in the affected divisions decreased by around 30%, and attrition increased. Based on data from the firm�s management information system, we document that efforts decreased by the same order of magnitude, and so did absenteeism. We provide evidence that workers� spot reaction to the flatter slope of the bonus function is unable to explain these effects. Rather, the findings support the fair wage hypothesis and relational contracting. Observing the effects over a period of more than three years, we show long-term negative reciprocity of those affected, but no negative selection effects. Consultants who enter after the pay change are equally productive as the ones who were in the firm when the pay change occurred. Nonetheless, the pay change would have been profitable for management only for a horizon of ten years or more. |