Without transparency about peer wages in a real effort experiment, a change of wages does not affect performance. With transparency, however, higher paid workers tend to work more accurately, and lower paid workers shirk more under piece rates.