Spatial econometric approach to the EU regional employment process

Analyzing the Factors that Affect and Predict Employment Density Using Spatial Machine Learning: The Case Study of Seoul, South Korea

“…Among nonspatial factors, per capita GDP, employment incentives, and business tax were the most important, while population density and the proportion of elderly people were relatively less important. This result is consistent with previous studies showing that economic factors of the region, such as GDP and self‐reliance, are important factors influencing employment density (Kim, 2019; Furková and Chocholatá, 2021). Notably, the fact that the college entrance rate was not derived as an important influencing factor is consistent with the result of most previous studies.…”

“…Therefore, it was selected as the best model for predicting employment density in Seoul in 2019 among the four models. This result is consistent with previous studies that analyzed that employment distribution has spatial correlation and is suitable for analysis with spatial regression analysis (Li et al, 2009;Nam and Lim, 2009;Gutiérrez Posada, Rubiera Morollón, and Viñuela, 2018;Furková and Chocholatá, 2021;Kim and Jeon, 2021). This result differs from other studies in that the prediction error of RMSE and MAE, which represent the difference between the predicted value and the actual value in the RF model and GWRF model, has greatly decreased, significantly improving the predictive power, and the main predictors affecting employment density have been identified according to the importance of the variables.…”

“…Many previous studies of employment distribution considered spatial autocorrelation and spatial heterogeneity of employment distribution, and spatial regression analyses were conducted based on this assumption. Furková and Chocholatá (2021) analyzed that regional employment distribution in Europe is influenced by both spatial autocorrelation and spatial heterogeneity, and also by employee wages and GDP, which are regional economic variables. Li et al (2009) used the OLS model and the GWR model as predictors of retail employment in southeast Queensland and found that both the explanatory power and AIC of the GWR model were significantly improved.…”

Spatial Heterogeneity and Spillovers of Employment in the EU Regions

“…Ertur and LeGallo (2008) distinguish two possible ways of differences: space-varying parameters and/or space-varying variances. In general, parameters can vary across group of regions (spatial regimes) or in a more general case these can vary even across individual regions (for more information see e.g., Fotheringham et al, 2002;Furková and Chocholatá, 2021).…”

“…Also, Pavlyuk (2011) investigated the differences in employment rates in Latvian regions based on instruments of spatial analysis and spatial econometrics. Furková and Chocholatá (2019) and Furková and Chocholatá (2021) used the Spatial Durbin Model (SDM) and geographically weighted regression (GWR) in order to verify territorial interconnections within the EU regions in the context of employment rates. Majchrowska and Strawiński (2021) analysed the spatial dependencies in the relationship between employment and minimum wage for local Polish labour markets revealing significant heterogeneities in the model.…”