: Measuring spatial effects in presence of institutional constraints: the case of Italian Local Health Authority expenditure

Spatial econometric models are now an established tool for measuring spillover effects between geographical entities. Unfortunately, however, when entities share common borders but are subject to different institutional frameworks, unless this is taken into account the conclusions may be misleading. In fact, under these circumstances, where institutional arrangements play a role, we should expect to find spatial effects mainly in entities within the same institutional setting, while the effect across different institutional settings should be small or nil even where the entities share a common border. In this case, factoring in only geographical proximity will produce biased estimates, due to the combination of two distinct effects. To avoid these problems, we derive a methodology that partitions the standard contiguity matrix into within-contiguity and between-contiguity matrices, allowing separate estimation of these spatial correlation coefficients and simple tests for the existence of institutional constraints. We then apply this methodology to Italian Local Health Authority expenditures, using spatial panel techniques. We find a high and significant spatial coefficient only for the within-contiguity effect, confirming the validity of our approach.