유가충격이 거시경제변수에 미치는 영향

The paper examines how the macroeconomic effects of oil price shocks have been changed over the past 40 years using multivariate ECMs and bivariate VARs. It first estimates 5-variable ECMs 80, 60, and 40 times, respectively,using a moving window of 20, 25, and 30 years. The estimation results show that in case of estimating the 5-variable ECM based on a moving window of 30 years, statistical significance of impulse responses and variance decompositions is the highest. In this case, an 1% positive oil price shock results in 0.104% increase in CPI, but 0.042% and 0.034% decreases in GDP and M2 on average, respectively. Won/dollar exchange rates decrease in the initial stage, but begin to increase from 1 year later. The paper then estimates bivariate VARs 120 times using a moving window of 10 years in order to investigate whether the effects of oil price shocks on macroeconomic variables are time varying or not. The estimation results show that the responses of CPI and M2 to oil price shocks gradually become weaker until 2007, but that those of won/dollar exchange rates as well as CPI and M2 are bigger after the global financial crisis. On the other hand, the negative effects of oil price shocks on GDP become weaker from the end of the 1990s. Such responses of macroeconomic variables seem to be closely associated with either changes of the domestic industrial structure or undervalued won and low interest rate policies as well as outside factors such as global demand and supply changes. Particularly, the negative effect of oil price shocks on inflation is more deteriorated in case of including the post-crisis period. This phenomenon results from demand side factors like low interest rate and undervalued won policies rather than supply side factors like hike of wage and raw material price. It implies that recent inflation has a feature of demand pull inflation as well as cost push inflation. There exists a possibility that responses of macroeconomic variables in case of oil price increase are different from those in case of oil price decrease because of downward rigidity of price and asymmetry of monetary policy. In order to test asymmetric effects of oil price shocks, the paper estimates 6-variable ECMs 40 times using a moving window of 30 years. The oil price variable is decomposed into two new variables : OP+ includes positive changes in oil prices and zero otherwise, while OP- consists of negative changes in oil prices and zero otherwise. According to the empirical results, oil price shocks have more impact on macroeconomics variables when the oil price rises rather than falls. Furthermore, the asymmetric responses of won/dollar exchange rates and CPI to oil price shocks are statistically significant at the conventional level. In the near future, this argument needs to be developed in the direction to theoretically and empirically study what make the responses of macroeconomic variables to oil price shocks change over times. One of interesting topics may be a microeconomic approach based on causal relationships between parameter estimates or impulse response functions of ECMs and the industrial structure change.