This paper presents evidence that the quality of survey data on household incomes systematically improves across waves of a panel. Our estimates indicate that the effect of being interviewed for a second time is to increase the mean of reported monthly income by £142 (8 percent). Dependent...