Some Current Issues in the Statistical Analysis of Spillovers

Spillover phenomena are usually statistically estimated on the basis of regional and temporal panel data. In this paper we review and investigate exploratory and confirmatory statistical panel data techniques. We illustrate the methods by calculations in the stetting of the well known Research and Development Spillover study by Coe and Helpman (1995). It will be demonstrated that alternative estimation techniques that are well compatible with the data can lead to opposite conclusions.