The rise in European unemployment is often blamed on increased mismatch between labour supply and demand- either by age, skill or region. To investigate this, we first develop models to explain differences in unemployment rates - both where labour supply is given and where it responds through labour mobility. Evidence supporting the model is presented using regional data for Britain and the US. We then ask how the intersectoral dispersion in unemployment rates is related to the overall average unemployment rate. We conclude from our model that average unemployment increases with the variance of relative unemployment rates across groups. Since this variance is substantial, it explains a good part of total unemployment. But, since the variance has not risen for skill groups or regions, it cannot explain the overall rise in European unemployment. We then turn to policy. If labour supply is given, there is a strong case for taxes and subsidies to redirect demand to high unemployment groups. But if there is a perfect labour mobility (infinitely elastic supply across groups), then even with job rationing there is no efficiency case for intervention except where there are externalities. With partial labour mobility the conclusion lies in between. Given the positive externalities of training and possible congestion externalities of migration, there may be a case for subsidies to training and to unemployment in high-unemployment areas.
