The view that high unemployment in West Germany and other European countries is caused by a path dependence effect - or hysteresis effect - is quite popular among economists. However, because of an identification problem, much of the empirical evidence for this hypothesis is not fu lly convincing. This paper suggests a testing procedure that overcomes this problem. It is argued that in a cointegration framework it is reasonable to define hysteresis as the absence of weak exogeneity of the explanatory variables. Building on a cointeg ration model of the employment rate in West Germany, I find only weak evidence for hysteresis.