We organize a real-effort field experiment with varying piece rates to assess the impact of wages and social comparisons on productivity. In addition to analyzing how piece rates and social comparisons affect productivity during the ‘paid stage’ of the experiment, we also consider how they affect effort supply during a voluntary and unpaid follow-up task. Our main results are that effort supply is relatively unresponsive to variation in own earnings, but responds strongly to pay inequality. While we obtain weak support for the hypothesis that positive social comparisons invite extra effort during paid stages of the experiment, our most important finding is that social comparisons matter for voluntary tasks when shirking is cheap. Specifically, positive social comparisons positively affect productivity during unpaid tasks, and negative comparisons have the opposite impact.