4. Empirical results
In this section we discuss the empirical results of the VAR model analysis described in the previous section. In subsection 4.1 we analyse significance of coefficients in the model and apply Granger Causality test. In subsection 4.2 we present the results of impulse response functions analysis and variance decomposition. Afterwards, we turn to subsection 4.3 to test reliability of the VAR model.
4.1 Testing for significance and Granger-causality
According to Wald test (Table 5) the hypothesis of zero coefficient before oil price variable in GDP equation is not rejected in all analysed countries except Russia. This means that Russia is the only country where oil prices effect real economic activity directly. The fact …show more content…
As Granger Causality does not show us a complete picture of interactions, it is of a great interest to know how an impulse in one variable could influence another variable in the system which contains several other variables as well. Figure 1 shows changes of endogenous variable GDP growth expected by a shock in oil price of 1%. The upper and lower lines show two standard errors bounds based on confidence interval of 95%. On the basis of impulse response functions it could be concluded that there is an immediate similar reaction of real GDP growth in all countries within the first two quarters. The positive changes in oil price cause real GDP to grow in all countries with the only difference in period when the impact reaches its peak. However, not all impulse response functions are statistically significant. In this paper statistical significance is tested by analysing confidence intervals. It has become standard in relevant papers to evaluate statistical significance using upper and lower limits of impulse response functions. “An easy way of assessing whether any of these responses is significantly different from zero is to search for impulse response functions in the joint confidence set that cross the zero line” (Inoue and Kilian, 2016). That is why …show more content…
The result for Russia is expected because oil export activity is the main component of Russia’s GDP. The test for UK shows that oil price variable is significant for the whole VAR system except GDP equation which signalises about possible indirect effect of oil price on GDP growth. Other countries have not demonstrate direct relationship on the basis of Wald test. However, Granger Causality/Block Exogeneity test identifies indirect effect of oil prices in previous periods on GDP of Canada via CPI index. As Wald and Granger Causality/Block Exogeneity test do not reflect the whole picture, impulse response analysis is used as our main method of investigation. A similar positive reaction of real GDP growth on oil price shock is revealed within the first two quarters in all the countries. However, statistical significance is detected only in Canada, Russia and Mexico. Among these countries, the smallest response is in Mexico and the biggest one is in Canada. Accumulated response functions show significance of the exchange rate response to oil price shocks in all countries except UK. These findings confirm the main source for Russian crisis in 2014 when decrease in oil price caused depreciation of the currency. The variance decomposition analysis suggests that oil price shocks are a significant source of volatility for CPI,
In this section we discuss the empirical results of the VAR model analysis described in the previous section. In subsection 4.1 we analyse significance of coefficients in the model and apply Granger Causality test. In subsection 4.2 we present the results of impulse response functions analysis and variance decomposition. Afterwards, we turn to subsection 4.3 to test reliability of the VAR model.
4.1 Testing for significance and Granger-causality
According to Wald test (Table 5) the hypothesis of zero coefficient before oil price variable in GDP equation is not rejected in all analysed countries except Russia. This means that Russia is the only country where oil prices effect real economic activity directly. The fact …show more content…
As Granger Causality does not show us a complete picture of interactions, it is of a great interest to know how an impulse in one variable could influence another variable in the system which contains several other variables as well. Figure 1 shows changes of endogenous variable GDP growth expected by a shock in oil price of 1%. The upper and lower lines show two standard errors bounds based on confidence interval of 95%. On the basis of impulse response functions it could be concluded that there is an immediate similar reaction of real GDP growth in all countries within the first two quarters. The positive changes in oil price cause real GDP to grow in all countries with the only difference in period when the impact reaches its peak. However, not all impulse response functions are statistically significant. In this paper statistical significance is tested by analysing confidence intervals. It has become standard in relevant papers to evaluate statistical significance using upper and lower limits of impulse response functions. “An easy way of assessing whether any of these responses is significantly different from zero is to search for impulse response functions in the joint confidence set that cross the zero line” (Inoue and Kilian, 2016). That is why …show more content…
The result for Russia is expected because oil export activity is the main component of Russia’s GDP. The test for UK shows that oil price variable is significant for the whole VAR system except GDP equation which signalises about possible indirect effect of oil price on GDP growth. Other countries have not demonstrate direct relationship on the basis of Wald test. However, Granger Causality/Block Exogeneity test identifies indirect effect of oil prices in previous periods on GDP of Canada via CPI index. As Wald and Granger Causality/Block Exogeneity test do not reflect the whole picture, impulse response analysis is used as our main method of investigation. A similar positive reaction of real GDP growth on oil price shock is revealed within the first two quarters in all the countries. However, statistical significance is detected only in Canada, Russia and Mexico. Among these countries, the smallest response is in Mexico and the biggest one is in Canada. Accumulated response functions show significance of the exchange rate response to oil price shocks in all countries except UK. These findings confirm the main source for Russian crisis in 2014 when decrease in oil price caused depreciation of the currency. The variance decomposition analysis suggests that oil price shocks are a significant source of volatility for CPI,