Geographic concentration and establishment scale: An extension using panel data

In a recent study, Holmes and Stevens (2002) identify for the first time the positive relationship that exists between establishment scale and local industry concentration using a large cross-sectional plant-level data set for the United States. Using an exhaustive plant-level panel data set for Irish manufacturing covering nearly three decades, we are able to extend their analysis in two ways. First, we show that failing to control for fixed effects biases the relationship upward, although the essence of the result still remains after fixed effects are included. Second, the link is substantially weaker when plants locate for the first time in an area, but strengthens with age for those that survive in the long run. We link our results to recent contributions on the dynamics of geographic concentration.