The aggregate average wage is often used as an indicator of economic performance and welfare, and as such often serves as a benchmark for changes in the generosity of public transfers and for wage negotiations. Yet if economies experience a high degree of (nonrandom) fluctuation in employment the composition of the employed population will have a considerable effect on the computed average. In this paper we demonstrate the extent of this problem using data for Poland for the period 1996-2003. During these years unemployment in Poland almost doubled. We show that about a quarter of the growth in the average wage during this period could be contributed purely to changes in employment.