Despite the apparent stability of the wage bargaining institutions in West Germany, aggregate union membership has been declining dramatically since the early 90's. However, aggregate gross membership numbers do not distinguish by employment status and it is impossible to disaggregate these sufficiently. This paper uses four waves of the German Socioeconomic Panel in 1985, 1989, 1993, and 1998 to perform a panel analysis of net union membership among employees. We estimate a correlated random effects probit model suggested in Chamberlain (1984) to take proper account of individual specfic effects. Our results suggest that at the individual level the propensity to be a union member has not changed considerably over time. Thus, the aggregate decline in membership is due to composition effects. We also use the estimates to predict net union density at the industry level based on the IAB employment subsample for the time period 1985 to 1997.
union membership gross and net union density correlated random effects probit model