The studies using Mincer equations are generally applied to cross-sectional data at the micro-level. There are however limited studies conducted with macro or panel data for wage equations. Pseudo panel data methods can be applied to empirical studies by creating cohorts from repeated cross-sectional data in the absence of genuine panel data. Difference in both the human and labour resources according to the spatial positions may also affect the prediction of the wage equations. We aim to introduce the application of spatial pseudo panel models by creating cohorts according to the birth years of employees and regions in which they live from the Turkish household labour survey for the period 2010-2015. As a result, we find that the spatial autocorrelation model is appropriate for wage equations of Turkey. We also find that return of education on wages is 11% while return of experience on wages is 4%.

We propose a method of constructing multisector-multiregion input-output tables, based on the standard multisector tables and the tools of spatial econometrics. Voivodship-level (NUTS-2) and subregion-level data (NUTS-3) on sectoral value added is used to fit a spatial model, based on a modification of the Durbin model. The structural coefficients are calibrated, based on I-O multipliers, while the spatial weight matrices are estimated as parsimoniously parametrised functions of physical distance and limited supply in certain regions. We incorporate additional restrictions to derive proportions in which every cross-sectoral flow should be interpolated into cross-regional flow matrix. All calculations are based on publicly available data. The method is illustrated with an example of regional economic impact assessment for a generic construction company located in Eastern Poland.