Commuting and Panel Spatial Interaction Models: Evidence of Variation of the Distance-Effect over Time and Space

We apply spatial interaction models using panel data to explain commuting behaviour in the Netherlands. Our main conclusion is that the distance-decay effect is not constant over time and that changes in this effect are region specific. In more densely populated regions the change in the distance-decay parameter is small suggesting that regional increases in congestion have a large negative effect on the increases in average commuting distance. The panel spatial interaction model we derive is well-suited for testing significance of the centrality index (an often used variable in spatial interaction models). Although evidence is found for competition effects in a pooled cross section framework, controlling for time invariant unobserved heterogeneity renders this relation spurious.