Diagnostic Tools for Nonlinearity in Spatial Models

There is an increasing awareness of the potential of nonlinear modeling in regional science, which can partly be explained by the recognition of the limitations of conventional equilibrium models in complex situations and partly by the easy availability and accessibility of sophisticated computational techniques. Among the class of nonlinear models chaos theory stands out as an interesting approach, in which notable progress has been made in economic research. However, the operational significance – in terms of solid empirical tests on chaos – has still been rather limited and a rigorous statistical-econometric treatment of chaos phenomena is still lacking. This paper is concerned with a methodological and empirical analysis of chaos in spatial systems. It seeks to break new research ground by linking the classical diagnostic tools developed in spatial econometrics to nonlinearity tests for empirical data series, in particular the so-called <I>BDS</I> (Brock, Dechert, Scheinkman) test. A spatial variant of the <I>BDS</I> test is developed, and subsequently applied in the context of a shift-share model for Dutch regional labor markets.