The effect of national industry shocks on local employment: impacts on geographical inequality and inefficiency

We analyse the effect of national industry shocks on local employment. By providing a novel structural view on the Bartik framework, we show that the difference in national and regional employment growth trends can be attributed to within-region spillovers. These spillovers can be quantified in a simple regression of regional employment change predictions versus actual regional employment changes, where regional employment change predictions are based on national shocks. We find consistent evidence that a predicted change in employment by 1% is associated with a 1.3% change in actual employment in a region. We hypothesize that agglomeration plays a key role in explaining the difference between the predicted and the actual employment growth. When we allow for non-linearities in a variety of setups, we find that the main driver of agglomeration effects are regions with particularly strong growth in employment which outperform their predictions. Taking the employment weighted mean as inflection point, regions with below mean predicted employment growth show a roughly 1:1 translation of predicted job creation to actual job creation. For regions with above mean predictions this ratio increases to 1:1.7.