Spatial econometrics. Statistical foundations and applications to regional convergence. (Q2494647)

The structure of spatial econometric models can be created, e.g., by a set of countries or by a set of regions. An important concept of spatial theory is the \(\beta\)-convergence, which is formulated for explaining the economic convergence of the per-capita income. The neoclassical models predict that the economy converges towards its steady-state, which is defined as the situation where all variables grow at constant rates. This is the idea of the conditional convergence. The hypothesis of the convergence can be validated using the usual statistical testing. It is possible to distinguish four major typologies of economic data: cross sectional, time series, spatial series and spatial panel data. The main types of random field models, which can be used in spatial modelling, are as follows: spatial white noise, Markov and non-Markov fields. The specificity of the spatial likelihood for estimation and testing procedures is discussed. The standard linear regression model and its more common notations are studied in the spatial context. The theory is illustrated by a spatial econometric analysis of the Italian province \(\beta\)-convergence model and by an analysis of the European regions \(\beta\)-convergence model. The overview of alternative models and available software used in spatial analysis are given.