Automatic Identification of Spatial Econometric Model Using Parallel Optimization for Regional Economic Disparity

The spatial dependence degree of economic data among regions becomes increasingly higher due to the rapid development of economic integration and globalization. In consideration of the fact that both the inhibition mechanisms of capital, labor and other factors on economic growth and spatial interaction are still unclear. Based on 140 county economic data of Shandong province in china, a spatial econometric identification model simultaneously solving selection of spatial weight matrix (SWM) and parameter identification is established, in which 15 possible explanatory variables(including the intercept),20 SWMs and 327680 possible spatial econometric models are involved. A parallel trans-dimensional simulated annealing algorithm(PTD-SA) is constructed to find the optimum SWM and parameters, which allows for moves that not only change parameter values and SWMs but also moves between competing models. Numerical results show that the optimum spatial econometric model can describe the regional economic disparity reasonably.