This study empirically analyzed the dynamic relationship among regional education fiscal expenditure, regional economic growth, and regional human capital by 16 cities and provinces in Korea from 1998 to 2021 to find “stylized facts” and how to use them to establish local fiscal policies. In addition to analyzing the 16 cities and provinces, the capital, non-capital, metropolitan-city, and provincial regions were analyzed and compared.
The main contents of the study are as follows. We used panel data such as regional education fiscal expenditure, human capital, and economic growth to perform the panel unit root test, panel cointegration test, panel Granger causality test, and estimated the Panel Vector Error Correction (PVEC) model for impulse response and forecasting error variance decomposition based on the standard research methodology.
The analysis results of this study are summarized as follows. First, as a result of the first and second-generation panel unit root tests, it was found that unit roots exist in all three variables. Second, as a result of the panel cointegration test, it was found that there was one independent cointegration relationship among the three variables. Third, we used the PVEC model according to the results of the panel unit root test and the panel cointegration test, and the PVEC model with a lag length of 4 was estimated. As a result of the panel Granger causality test, we found that the long-term causal relationship between regional education fiscal expenditure and regional economic growth to regional human capital exists. Fourth, according to the impulse response, the growth rate of the three variables is most affected by their shock. Fifth, regional education fiscal expenditure and regional human capital are limited but mutually explanatory. The mutual explanatory power of all three variables was more significant in the capital regions than in the non-capital regions and more significant in the provincial regions than in the metropolitan-city region.
In this study, we obtained robust research results by applying all of the standard research methodologies used in the analysis using the VAR model. However, we need to utilize the characteristics of the panel data when the PVAR or PVEC model is used. We applied a methodology to analyze the entire region, capital region, non-capital region, metropolitan-city region, and provincial region. We compared the results and confirmed its usefulness. Therefore, we propose that this methodology be used as one of the standard research methodologies in future studies.