4.3 Panel Data Analysis
4.3.1 Basic Results
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ness increases the sacrifice ratio in the model of Danels and VanHoose (2006), the trade openness also has a negative effect on the inflation rate arising from discretionary monetary policy.
4.3 Panel Data Analysis
In addition to using the cross-sectional analysis to capture the long-run effect of trade openness on inflation, we also apply the panel data to analyze the relationship between openness and inflation in the short-run. Our panel data set includes 127 countries in 1973-2008, and we rely on the instrumental vari-able quantile regression for panel data proposed by Galvao (2008), Galvao and Montes-Rojas (2009) and Harding and Lamarche (2008).
4.3.1 Basic Results
Given the following model:
log(1 + πit) = Zit0η(θ) + D0itα(θ) + Xit0β(θ) + eit,θ,
where πit is the inflation rate, Zit denotes the dummy variable which identifies the N distant individuals, Dit is the trade openness with endogeneity, Xit is the control variable without endogeneity including GDP per capita, and eit,θ is the error term. Because there exist some deflation in some countries, it can not transform the inflation by taking log. Following Gruben and McLeod (2004), we divide the inflation rate by 100 and plus 1 to take log. In the basic empirical results, we control the growth of GDP per capita to the empirical model.
As suggested by Bowdler and Malik (2006), we address the potential endo-geneity of trade openness by using the lagged value of trade openness and lagged value of population as instrument variables. Besides, Cavallo and Frankel (2008) use the data of Frankel and Rose (2002) to compute the grav-ity estimates and deal with the endogenegrav-ity of trade openness by using the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
predicted ratio of trade to GDP based on an instrument. Based on Cavallo and Frankel (2008), we also take the gravity estimates into account. In con-clusion, we use three kinds of instrument variables: lagged term of openness, population, and gravity estimates to solve the potential endogeneity of trade openness.
We apply the instrumental variable quantile regression for panel data pro-posed by Galvao (2008), Galvao and Montes-Rojas (2009) and Harding and Lamarche (2008) to our empirical work, and we also consider the 2SLS within estimate to compare with QR estimates. Table 8 shows the main empirical results of our empirical work which just control the growth of GDP per capita in the basic empirical results. Figure 5 shows the QR estimates of our main results, and they stand for the empirical results by using three kinds of instru-ment variables (lagged term of trade openness, lagged term of trade openness and population, lagged term of trade openness and gravity estimates) sepa-rately. In order to compare the differences between the mean and quantile effects, we draw the 2SLS within estimate and QR estimates in the same fig-ure. In Figure 5, we use estimation values as the vertical axis and quantiles as the horizontal axis to show the empirical results. Where the horizontal solid line is the 2SLS within estimate, the solid curve is the QR estimates for panel data with endogeneity, and the two dotted curves represent the 95% confidence interval of QR estimates.
According to Table 8, when using the lagged term of trade openness and using the lagged term of trade openness plus population as instruments, the 2SLS within estimate is negative and significant at level 10%. The QR esti-mates all have negative effects. Deserve to be mentioned, the negative effect of trade openness on inflation becomes stronger when the inflation is higher.
For example, the 0.1 quantile estimate is −5.36 × 10−5 with a significance level 10% and the 0.9 quantile estimate is −4.91 × 10−4 and is significant at level
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
1%. It reveals that if a country has higher inflation, the dynamic inconsistency problem would be more serious.
Our empirical results are similar to Romer (1993), Lane (1997), Gruben and McLeod (2004), and Badinger (2009), but different from the positive effect of Alfaro (2005). The QR estimates show that the effects of trade openness on inflation are stronger when the inflation is higher. When a country increases the share of imports in GDP, it will cause the inflation to fall. When using the lagged term of openness and lagged term of population as instrument variables, the empirical result is similar to using the lagged term of openness as instruments. But when using the lagged term of openness and gravity estimate as instruments, the 2SLS within coefficient and the 0.1-0.7 quantile coefficients are insignificantly positive but the 0.8-0.9 quantile coefficient are insignificantly negative. It reveals that when using the lagged term of openness and gravity estimates as instruments, the empirical results are not as good as using the lagged term of openness as an instrument variable. We guess that the difference when using a lagged term of trade openness and gravity estimates as instruments may be caused by the sample period. Because the 5-year data set of Frankel and Rose (2002) which we used to compute the gravity estimates is up to 1995, we estimate and fill the gravity estimates up to 1999. In short, the period of using a lagged term of trade openness and gravity estimates as instruments is in 1973-1999, and the period of using the other two instruments is in 1973-2008. For this reason, we make a conjecture that the difference between using different instruments is caused by the sample period. According to this reason, our following empirical work just shows the empirical results of using a lagged term of openness as an instrument variable. Furthermore, we also measure trade openness as the exports share of GDP and the imports plus exports share of GDP to deal with the empirical work, the empirical results are similar to measuring the trade openness as the imports share of GDP, see
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Table 9 and Figure 6.
In addition to trade openness, we also apply financial openness to the empirical work. Following Badinder (2009), we measure the financial openness in terms of total foreign assets plus liabilities as a share of GDP, and the data set is calculated by Lane and Milesi-Ferretti (2006). In Table 9, we show the empirical results of financial openness and inflation. When we use the lagged term of financial openness as an instrument, the 2SLS within estimate is significantly positive but the QR estimates are significantly negative. The negative effect of financial openness is stronger when the quantile is higher, see Figure 6.
According to the empirical results, it show that whether we measure open-ness as trade openopen-ness or financial penopen-ness, the negative effect of openopen-ness is stronger when the inflation is higher. It reveals that if a country has higher inflation, the dynamic inconsistency problem would be more serious.