To construct the linked employer-employee data set, we first select establish- ments from the establishment panel data. From the available waves, we use the years 1995 to 2005. Since information on a number of variables, such as investment expenditures and sales are gathered retrospectively for the preceding year, we lose information on the last wave. Moreover, we restrict our sample to western Ger- man establishments from the mining and manufacturing sector with at least two employees. From the establishment level data we gain information on a number of establishment characteristics, such as establishment size, collective bargaining cov- erage and the existence of a works council. .
