This paper examines changes in the distribution of wages using bounds to allow for the impact of non-random selection into work. We show that worst case bounds can be informative. However, since employment rates in the UK are often low they are not informative about changes in educational or...