This paper reports the results of an individual real e¤ort laboratory experiment where subjects are paid for measured performance. Measured performance equals actual performance plus noise. We compare a stable environment where the noise is small with a volatile environment where the noise is large. Subjects exert signi cantly more e¤ort in the volatile environment than in the stable environment. This nding is in line with standard agency theory and contrasts a distinct element of expectancy theory; noisier performance measures do not lower work motivation.