Fiscal Multipliers of Korea: A Bayesian VAR Approach (in Korean)

The vast existing literature on fiscal multipliers for Korea is based on structural vector autoregressive models with a specific parametric assumption for identification. A caveat of this approach is that fiscal multipliers tend to be sensitive to the identification assumption. As an alternative approach, this paper estimates the effect of tax cuts and government spending expansions on GDP using a Bayesian methodology which imposes the interval value constraint on the parameters. The results show that both tax cuts and government spending expansions have a significant effect on economic growth in year one and two after the initial fiscal policy shocks; and the growth effect of expenditure is found to be larger than that of tax cuts. Compared to the conventional OLS estimates, the Bayesian procedure tends to be associated with larger GDP multipliers of taxes while it produces relatively smaller multipliers of government spending. However, the quantitative changes are not significant enough to reverse the conclusion that increasing government spending is more effective than tax cuts in fostering economic growth. In addition, we analyze the growth effect of sub-tax and government spending variables by dividing tax revenue into 'income and property taxes' and 'indirect tax', and expenditures into 'government consumption/investment' and 'transfers.'