This paper presents a new method to correct for measurement error in wage data and applies this method to address an old question. How much downward wage flexibility is there in the U.S? We apply standard methods developed by Bai and Perron (1998b) to identify structural breaks in time series data. Applying these methods to wage histories allows us to identify when each person experienced a change in nominal wages. The length of the period of constant nominal wages is left unrestricted and is allowed to differ across individuals, as is the size and direction of the nominal wage change. We apply these methods to data from the Survey of Income and Program Participation. The evidence we provide indicates that the probability of a cut in nominal wages is substantially overstated in data that is not corrected for measurement error.