A többskálás földrajzilag súlyozott regresszió alkalmazása a településszintű vállalati felszámolások hazai vizsgálatához, 2019
The multiscale geographically weighted regression method applied in the study of settlement-level enterprise liquidation, 2019

Summary/Abstract: Regression is one of the most common statistical tools used in exploring relationships between socioeconomic phenomena. Traditional regression models, however, have limited reliability in case of spatial data. One solution to this problem is the geographically weighted regression (GWR), which considers spatial processes as variables. Traditional GWR assumes that all modelled processes operate on the same spatial bandwidth. A novel extension allows each relationship to vary according to different spatial bandwidths, leading to the multiscale geographically weighted (MGWR) method. This paper is comparing standard OLS, GWR and MGWR by investigating corporate liquidation. The author estimated the spatial variability of settlement level corporate liquidations using meso-level economic, municipal, and transportation data. The findings of the three models indicated that MGWR out performed the other two methods. Among the economic factors affecting liquidation, the global effect of the unemployment rate and the local effect of per capita income may be observed. The presence of bank branches in the settlement and the weight of the industrial sector significantly impact business liq-uidations, as well as transportation indicators.

• Details
• Contents