Arent and Nagl (2013) use the BA Employment panel 1998-2007 to identify effects of the German Hartz reform and find that it caused a considerable reduction of wages. Our replication study suggests that their clear and strong conclusions are based on implausible assumptions regarding the error structure of their regression models and on a too coarse modelling of the time effects. They become blurred and weak once better estimates of the standard errors are obtained and the development of wages is investigated at a finer time grid. Furthermore, Arent and Nagl’s reform effects shrink considerably when a more appropriate price index is used to deflate the wages and when the censoring of wages is treated correctly. Methodological considerations suggest that their conclusions depend on several further daring and untested assumptions.
© 2014 by Lucius & Lucius, Stuttgart