Study on the Transmission-Mechanism of International Factors Influencing Chinese Price Level

4.1 Preliminary Analysis

Table 1 reports the correlation coefficients of the sample data. Chinese CPI is negatively associated with USDX and DJI, and positively associated with PF, CRB and BDI. The absolute value of correlation coefficient between CPI and DJI is the maximum, and that between CPI and BDI is the minimum. Chinese PPI is negatively associated with USDX, and positively associated with other variables. The absolute value of coefficients is between 0.41 and 0.52 and very close to each other. The absolute value of correlation coefficient between PPI and BDI is the maximum while that between PPI and DJI is the minimum. According to these results, the correlation between PPI and international factors is much higher than that between CPI and international factors. Moreover, CPI is positively related to PPI and the correlation coefficient between them is 0.7506; USDX is negatively associated with other variables; PF is positively associated with CRB and the correlation coefficient is 0.9374, so they exhibit relative strong substitutability for each other.

The Granger test is employed to study the causality and test the direction of causality between two variables, in order to judge the predicting effect of different factors on CPI and PPI. The Akaike and Schwarz information criterion are used to determine the appropriate lag length. The results suggest that PF, CRB and DJI Granger cause CPI; USDX and BDI don’t Granger cause CPI. Especially, CPI Granger causes PF and BDI. On the contrary, PF, USDX, CRB and BDI Granger cause PPI, but DJI doesn’t Granger cause PPI. Especially, PPI Granger causes only USDX.

4.2 Analysis on Transmission-Effect

According to impulse response function, when exerting a unit standard variance shock to a variable in VAR at period 1, all variables in the VAR system will respond in the subsequent periods. Figure 1 shows the accumulated response in the first 12 periods.

Figure 1

Analysis on impulse response of CPI and PPI

The analysis on the accumulated response of CPI found the following characteristics. First, from the perspective of reaction time, VAR system doesn’t react immediately to innovations. The reaction of CPI to PF shocks lagged about 3 months, while that to USDX shocks lagged about 2 months. The reaction of CPI to CRB shock lagged about 1 month, which is small at first and then turns to significant reaction. Additionally, the reaction of CPI to BDI shocks lagged 3 months. Chinese CPI has a quick reaction to DJI shocks and responds at the subsequent periods, which shows that the influence of international financial market transmits quickly. Second, from the perspective of influence persistence, the reaction of VAR system to innovations shows no immediate convergence. Generally, the reaction will last 12 months at least and the shocks are digested gradually. Third, from the perspective of influence magnitude, a shock to variables in VAR system leads to gradual reaction of CPI, and the reaction usually reaches the maximum value at period 6th∼9th. However, the reaction of CPI to different shocks varies greatly and exhibits no obvious regularity. Fourth, from the perspective of reaction direction, CPI positively reacts to shocks from PF and DJI and negatively reacts to the shocks from USDX. However, CPI has a positive reaction to shocks from CRB and BDI at first and then the reaction turns negative. Fifth, from the perspective of transmission pattern, the mechanism of shock transmission varies obviously for different variables. The transmission of shocks from PF, USDX and DJI are similar and exhibit a mode of enhancing on the same direction. The transmission mechanism of shocks from CRB and BDI is similar and shows a common feature of repeated correction mode, i.e., the positive reaction of CPI begins small and turns larger, but then the impact becomes smaller too until it exhibits a reversal correction at last. In other words, the influence of oil price and international financial market on Chinese CPI exhibits a positively enhancing mode, but the effect of U.S dollar exchange rate exhibits a negatively enhancing mode. This conclusion is consistent with economic theory. However, the effect of international commodity price and international shipping price on Chinese CPI exhibits a positively enhancing mode at first and then shows a negatively enhancing mode. In fact, the rise of these two factors directly lead to a rise of consumption goods price in China to some extent, however, any too fast rise of price is harmful to economic growth and may result in government intervention. Therefore, it constrains the growth of final consumption demand and so the influence of these two factors on Chinese CPI becomes negative finally.

The analysis on the accumulated response of PPI found the following characteristics. First, from the perspective of reaction time, VAR system reacts immediately to innovations of all variables except for DJI. That is, PPI reacts to shocks from PF, USDX, CRB and BDI in the next period, but the reaction to shocks from DJI is insignificant until period 4th. Therefore, the reaction of Chinese PPI to international factors is more sensitive than that of CPI. Second, from the perspective of influence persistence, the reaction of PPI to innovations also shows no immediate convergence. Generally, the reaction will last 12 months at least and the shocks are digested gradually too. Third, from the perspective of influence magnitude, the reaction of PPI to shocks of VAR system exhibits a process of gradual evolution. The reaction to shocks from USDX reaches the maximum value at period 8th, and the reaction to shocks from CRB, BDI and DJI reaches the maximum values at period 5th. And compared to CPI, the influence of international factors on PPI is much stronger. Fourth, from the perspective of reaction direction, PPI shows a continuously positive reaction to shocks from PF, CRB and BDI, while a continuously negative reaction to shocks from USDX. However, PPI shows a negative reaction to shocks from DJI in the first two periods, and then turns positive. Fifth, from the perspective of transmission mechanism, the modes of transmission of shocks vary greatly for different variables. The patterns of PF, USDX, CRB and BDI are very similar and the reaction converges after the expansion at first. But PPI exhibits a negative reaction to shocks from DJI at first and then the reaction increases positively before coming to be stable gradually. This means that the influence of international financial market on Chinese upstream price is different from other international factors and the magnitude of influence is much smaller.

The variance decomposition can obtain relative importance of stochastic errors that influence variables in VAR system. Therefore, the importance of different factors in transmission mechanism of price shocks can be assessed. Table 2 reports the results of variance decomposition and main conclusions are as follows.

First, the largest share of CPI and PPI fluctuation is explained by its owner variance. According to the algorithm, the forecasting error at first step all comes from its owner innovation, that is, it accounts for all of the variance by itself. In the subsequent forecasts, the error will be affected by shocks from all variables in VAR system. The share that CPI itself accounts for is more than 92% in the first 12 periods. However, the share that PPI itself accounts for falls quickly from 93.23% at the 2nd period to 50% at the 7th period and 34.86% at the 12th period. Obviously, the influence of international factors on Chinese PPI is more significant than that on Chinese CPI. Comparably speaking, the exogeneity of PPI is more significant while the endogeneity of CPI is more significant. This result is consistent with economic intuition that China has been an important processing trade economy in the world which imports a large number of raw materials and energy and exports industrial goods.

Second, the results of RVC indicator exhibit relatively great changes in the early stage for both CPI and PPI, and almost keep stable from the 10th period. During the whole periods, the forecasting error decomposition of CPI is relatively stable and the shares of different variables show very small changes over time. This reveals that the shares are not sensitive to prediction time and the relative importance of variables almost doesn’t change over time. However, the prediction error decomposition of PPI changes greatly over time and the relative importance of different variables varies along with forecasting periods.

Third, the transmission-mechanism of shocks in VAR system is very different. According to the maximum contribution share to variance, the contribution shares of PF, USD, DJI and BDI to variance of CPI all reach their peaks at the 12th period, and account for 3.42%, 2.51%, 1.55% and 0.22%, respectively. However, the contribution share of CRB to variance of CPI reaches its peak at the 8th period and accounts for 0.35%. The contribution shares of PF, USD and DJI to variance of PPI reach their peaks at the 12th period and account for 26.48%, 22.71% and 2.89%, respectively. However, the contribution share of CRB to variance of PPI reaches its peak at the 7th period accounting for 13.06%, and the contribution share of BDI to variance of PPI reaches its peak at the 6th period accounting for 2.74%. Therefore, there are differences in the transmission-mechanism of international factors influencing Chinese CPI and PPI, and both the magnitude and relative importance are different.

Fourth, the relative importance of different factors influencing Chinese price level is different over time. The shares of PF, USDX, BDI and DJI to variance of CPI show an upward trend in the first 12 periods. However, the share of CRB reaches the maximum at the 8th period and then declines. The shares of PF, USDX and DJI to variance of PPI show an upward trend too, while the shares of CRB and BDI reach the maximum at the 7th and 6th period respectively and then decline. It is worth to point out that the results of variance decomposition show that the influence of international commodity price on Chinese price level disappears very fast. This is inconsistent with the intuition inference to some extent. The reason might be that as an export-oriented economy China’s price level is sensitive to international commodity price and make a fully reaction immediately to the fluctuation of commodity price. So, it only takes relatively shorter time to digest the shock from international commodity price.

4.3 Analysis on Mechanism-Switch

Equation (6) is estimated in order to comprehensively investigate the influence of international factors on Chinese general price and Table 3 shows the results. The regression of CPI on PF, USDX, BDI, CRB and DJI obtains the following findings. The coefficient of PF is insignificantly positive while that of USDX is insignificantly negative. The coefficients of CRB and BDI are significantly positive at 5% and 1% level respectively and that of DJI is significantly negative at 1% level. Therefore, the international trade factors represented by CRB and BDI and the international financial market factor represented by DJI have significant impact on Chinese CPI. Intuitively, this is consistent with theoretical inference. However, the empirical results indicate that the impact of international oil price and U.S. dollar index on Chinese CPI isn’t statistically significant and the reason might be as follows. On the one hand, international oil price and commodity price are good substitutions for each other, and the influence of oil prices on Chinese CPI might have been internalized into the effect of international commodity price. On the other hand, because the financial system especially capital and financial account of China has not yet been fully liberalized and there are some blocks hindering the price transmission, it is difficult for shocks of international monetary and exchange rate factors to form a direct and immediate influence on Chinese price.

Meanwhile, the regression of PPI on PF, USDX, BDI, CRB and DJI obtains the following findings. The coefficients of PF and DJI are significantly negative at 5% level, while the coefficients of USDX, CRB and BDI are significantly positive at 1% level. Obviously, the comprehensive analysis shows that international commodity price, international trade and exchange rate have a significantly positive influence on Chinese PPI, while international oil price and financial market have a significantly negative effect on PPI. The rise in international commodity price and shipping price will promote Chinese industrial product price. Because the rise in U.S. dollar index is usually resulted from the improvement of U.S. economy, so it’s easy to understand that driven by the increase of U.S. market demands and Chinese PPI will go up too. About the estimated coefficients of PF and DJI, the reasons are as follows. On the one hand, international oil price and international commodity prices are substitutions for each other to some extent. On the other hand, international financial markets vary ceaselessly. PPI mainly depends on the real economy and whose behavior isn’t so vulnerable to timely fluctuations in international financial markets. Therefore, the direct impact of international financial factors on Chinese PPI isn’t significant.

Considering the changes in economic environment, the transmission mechanism of the influence of international factors on domestic price level may changes too. Since joining the WTO in 2001, the linkage between China and the world economy has been much closer and China has played an increasingly important role in international finance and trade. Moreover, 2008 international financial crisis has a profound effect on the world economy as well as Chinese economy. This study conducted Chow breakpoint test referring to January 2001 and January 2008 and the results obtain the following findings. For CPI, the values of F-statistic are 95.44 and 41.45, and the corresponding p-values are both 0. For PPI, the values of F-statistic are 6.25 and 4.44, and the corresponding p-values are 0 and 0.0003 respectively. Therefore, the results reveal that there do exist structural changes, which is consistent with the conclusion of Joshi and Acharya[9].

So, the sample is divided into three subsamples with the above breakpoints and Equation (6) is estimated by the subsamples. The results are showed in Table 5 and there are significant changes in transmission mechanisms of shocks from international factors to Chinese CPI and PPI. For CPI, the main conclusions are found as follows. First, from the perspective of influence direction, the influence of USDX, CRB and DJI changes directions among the three subsamples. The coefficients of USDX are positive in subsample 1 and 3 while that is negative in subsample 2. The coefficients of CRB and DJI are negative in subsample 1, while those are positive in subsample 2 and 3. Meanwhile, the signs of coefficients of PF and BDI keep stable. This means that the direction of impact of international commodity price and Dow Jones industrial average index changes from negative to positive after China joining the WTO. The international exchange rate factor represented by USDX has a positive effect on Chinese CPI and a negative effect after China joining the WTO, while the influence returns to be positive after the outbreak of 2008 international financial crisis. The international oil price has a negative effect on Chinese CPI in the three subsamples, but the international trade has a positive effect on Chinese CPI among three stages. Second, from the perspective of significance, the coefficients of BDI are statistically significant in all three subsamples at 1% level, while the significance for the coefficients of other variables change through three subsamples. In subsample 1, the coefficient of PF is significant at 5% level, while the coefficients of USDX, CRB, BDI and DJI are significant at 1% level. In subsample 2, only the coefficient of BDI is significant at 1% level, while others are all statistically insignificant. In subsample 3, the coefficients of CRB and BDI are significant at 1% level, but others are insignificant.

For PPI, the main conclusions about the subsample regression are found as follows. First, from the perspective of influence direction, the influence of PF, USDX and BDI changes direction among the three subsamples. The coefficients of PF and USDX are positive in subsample 1 and 3 while those are negative in subsample 2. The coefficient of BDI is negative in subsample 1 while those are positive in subsample 2 and 3. The coefficients of CRB and DJI keep positive and negative respectively in the three stages. This means that the influence direction of international shipping price on Chinese PPI changes from negative to positive after China joining the WTO. International oil price and exchange rate factors represented by U.S. dollar index have a positive effect on Chinese PPI in the first stage and have a negative effect after China joining the WTO, but the influence returns to be positive after the outbreak of 2008 international financial crisis. CRB has a positive influence on Chinese PPI in the three stages, while DJI has a negative influence on PPI at the same time. Second, from the perspective of significance, all coefficients of international factors change through the three stages. In subsample 1, the coefficient of CRB is significant at 1% level, and those of PF and DJI are significant at 5% level; the coefficient of USDX is significant at 10% level and that of BDI is insignificant at all. In subsample 2, the coefficient of USDX is significant at 1% level, while those of PF and BDI are significant at 5% level. However, the coefficient of DJI is significant at 10% level but that of CRB is insignificant at all. In subsample 3, the coefficients of BDI and CRB are significant at 1% level and that of USDX is significant at 5% level. However, the coefficients of PF and DJI are statistically insignificant. In all, the impacts of international factors on Chinese CPI and PPI exhibit obvious structural changes.