Cyclical Unemployment: Sectoral Shifts or Aggregate Disturbances?

Recent work by David Lilien has argued that the existence of a strong positive correlation between the dispersion of employment growth rates across sectors (G) and the unemployment rate implies that shifts in demand from some sectors to others are responsible for a substantial fraction of cyclical variation in unemployment. This paper demonstrates that, under certain empirically satisfied conditions, aggregate demand movements alone can produce a positive correlation between G and the unemployment rate. Two tests are developed which permit one to distinquish between a pure sectoral shift interpretation and a pure aggregate demand interpretation of this positive correlation. The finding that G and the volume of help wanted advertising are negatively related and the finding that G is directly associated with the change in unemployment rather than with the level of unemployment both support an aggregate demand interpretation. A proxy for sectoral shifts that is purged of the influence of aggregate demand is then developed. Models which allow sectoral shifts in the composition of demand and fluctuations in the aggregate level of demand to affect the unemployment rate independently are estimated using this proxy. The results support the view that pure sectoral shifts have not been an important source of cyclical fluctuations in unemployment.