We conduct a field experiment among 189 stores of a retail chain to study dynamic incentive effects of relative performance pay. Employees in the randomly selected treatment stores could win a bonus by outperforming three comparable stores from the control group over the course of four weeks. Treatment stores received weekly feedback on relative performance. Control stores were kept unaware of their involvement, so that their performance generates exogenous variation in the relative performance of the treatment stores. As predicted by theory, we find that treatment stores that lag far behind do not respond to the incentives, while the responsiveness of treatment stores close to winning a bonus increases in relative performance. On average, the introduction of the relative performance pay scheme does not lead to higher performance.