From a model by Hopenhayn, three hypotheses can be derived: (H1) Firms that exit in year t were less productive in t- 1 than firms that continue to produce in t. (H2) Firms that enter in year t are less productive than incumbent firms in year t. (H3) Surviving firms from an entry cohort were more productive than non-surviving firms from this cohort in the start year. This paper uses unique newly available panel datasets for all manufacturing plants from Germany (1995-2002) to test these hypotheses. All three hypotheses are supported empirically for West and East Germany. © 2009 The Author. Journal Compilation © Verein für Socialpolitik and Blackwell Publishing Ltd.