Dynamic Forecast for Tax Revenue

This study forecasted tax revenues by estimating Korea's national tax revenue functions using time series techniques to enhance methodological accuracy. Analysis results demonstrated that cointegration methods had superior predictive power, providing statistically significant estimates for tax revenue projections. Tax revenue projections estimated using the cointegration method of DOLS (Dynamic Ordinary Least Squares) and FMLS (Fully Modified Least Squares) demonstrated superior predictive accuracy, particularly outperforming government forecasts in tax categories with significant revenue shortfalls. However, changes in assumptions of explanatory variables were found to substantially alter tax revenue projections. This underscores the importance of accurate forecasting through appropriate model selection tailored to economic conditions and continuous updates of explanatory data.