SPATIAL ECONOMETRIC ANALYSIS OF REGIONAL CAPITAL INVESTMENTS IN UKRAINE

Анотація

The aim of this paper is to identify and quantify spatial dependence in the distribution of capital investments across Ukrainian regions, to determine the type of spatial relationships (spatial lag versus spatial error), and to decompose the total effect of gross regional product on capital investments into direct and indirect (spillover) components. Spatial dependence is a fundamental characteristic of regional economic processes; however, in the context of Ukrainian regional studies, the spatial dimension of capital investment distribution remains insufficiently explored. Classical econometric models that treat regions as isolated units produce biased and inefficient parameter estimates when interregional interactions are present. This gap is particularly significant for Ukraine, where institutional heterogeneity, uneven spatial development, and the impact of armed conflict create complex patterns of regional interdependence. The empirical basis consists of a balanced panel of 25 regional units (24 oblasts and the city of Kyiv) over the period 2016–2021, yielding 150 observations. The spatial interaction structure is defined by a first-order queen contiguity weight matrix based on GADM Level-1 shapefiles. Estimation is performed via maximum likelihood using the R packages splm, plm, and spdep. The study compares spatial autoregressive (SAR) and spatial error (SEM) models with fixed and random effects. Model selection relies on Baltagi–Song–Koh LM tests and the spatial Hausman test. The analysis confirms statistically significant spatial dependence in regional capital investments. The elasticity of capital investments with respect to gross regional product is consistently above unity (β ≈ 1.19–1.27), indicating procyclical investment behaviour consistent with the accelerator theory. The effect decomposition reveals an indirect spatial effect of 0.196, approximately 14 percent of the total effect, confirming the existence of interregional investment spillovers. The spatial Hausman test favours fixed-effects specifications (χ² = 18.71, p < 0.001). Ignoring spatial interactions underestimates the impact of economic growth on investments by 14–16 percent, which has direct implications for regional investment policy design and post-war recovery strategies.

Класифікація

Ідентифікатори

Схожі роботи