3. MOVIMIENTO DE SUELOS Y DEMOLICIONES 0 GENERALIDADES
3.9. RETIRO DE EXCEDENTES
Using the threshold levels estimated in Section 6.5, a dummy variable and an interaction variable was then created for both the US and Japan in order to analyse the threshold effect of real exchange rate volatility on Australian export volumes. The dummy variable (DUM) is an indicator that triggers the threshold effect when bilateral real exchange rate exceeds or is equivalent to the threshold value estimated for the US and Japan in the preceding section. The interaction term captures the effect of higher real exchange rate volatility on Australian real export volumes. The results from the OLS estimation of the threshold trade models are summarised in Table 6.9 below.43
The threshold trade models estimated in Table 6.9 (please refer overleaf) follows equation (5.7) in Chapter 5. The dependent variable in both models is LEXP – real export volumes to either US or Japan.
t t t
t t
t LINCOME LPRICE LV DUM INTERACTION LEXP =α2 +β1 +β2 +β32 +γ1 +γ2 +ε2
(5.7)
43 A common problem in time series models is that significant serial correlation and heteroskedasticity are often present. Hence, as part of the diagnostic testing, this study has also conducted the Breusch-Godfrey LM and the White heteroskedasticity (no cross terms) tests to check for the existence of serial correlation and heteroskedasticity. Due to a small sample size, the white heteroskedasticity (no cross terms) test was chosen over the test with cross terms. Both the basic and threshold trade models for US and Japan appear to exhibit significant serial correlation at the 5% level. Heteroskedasticity is also present in the models for US but not in the models for Japan. Hence, autocorrelation/heteroskedasticity consistent t-statistics are generated using the White heteroskedasticity consistent covariance matrix option in EViews (White, 1980). It should also be noted that the use of the heteroskedasticity consistent t-statistics does not appear to have affected the significance of the variables greatly as all variables retain their significance before and after the corrections were made.
Table 6.9: Threshold Trade Models
Variables US Japan
C -4.917231*
R-square 0.833117 0.795898
Notes: Heteroskedasticity-Consistent Standard errors in ( ) Heteroskedasticity-Consistent T-statistics in [ ]
* Significant at 5 per cent level. The T-test critical value at 5 per cent level of significance is 1.6694, based on T = 68, K=5.
** Jointly significant using Wald Test44
As seen in Table 6.9, all variables are individually statistically significant at the 5 per cent level45, except for LV, DUM and INTERACTION in the US model. However, the Wald Test conducted shows that LV, DUM and INTERACTION are jointly significant in the US model.
From Table 6.9, the results for US when there is a threshold effect and when there is no threshold effect can be summarised in equation (6.1) and (6.2) respectively:
44 The p-values of 0.000 (for the F-test and Chi-sq test statistics) are < 0.05, rejecting the null hypothesis that LV, DUM and interaction are not jointly significant.
45 The t-distribution critical value at the 5 per cent level of significance is 1.6694.
LEXPt =−4.92+1.67LINCOMEt −1.42LPRICEt +0.11LVt (6.1)
when τ <−7.1402
LEXPt =−5.49+1.67LINCOMEt −1.42LPRICEt +0.086 LVt (6.2)
when τ ≥−7.1402
Similarly, the results for Japan when there is a threshold effect and when there is no threshold effect can be summarised in equation (6.3) and (6.4) respectively:
LEXPt =−7.49+2.28LINCOMEt −1.07LPRICEt +0.04LVt (6.3)
when τ <−5.6624
LEXPt =−8.26+2.28LINCOMEt −1.07LPRICEt −0.076LVt (6.4)
when τ ≥−5.6624
The estimated coefficients for the importing country’s income and relative price of imports have their expected signs and the elasticities of these two variables are similar to those in the basic trade models estimated in Section 6.4. As we can see, the presence of a threshold effect has dramatically affected the elasticity of real export volumes to the US with respect to the bilateral real exchange rate volatility but not for Japan. Specifically, for the US, when there is no threshold effect, there is a 0.11 per cent increase in real export volumes to the US as the USD/AUD real exchange rate volatility increases by 1 per cent. This is different from the basic US trade model estimated in Section 6.4 when the impact of a 1 per cent increase in USD/AUD real exchange rate volatility resulted in a statistically significant 0.071 per cent decrease in real export volumes to US. For Japan, the existence
of a threshold effect has not dramatically affected the elasticity of real export volumes to Japan. Specifically, a 1 per cent increase in YEN/AUD real exchange rate volatility in the threshold and basic models result in a 0.04 and 0.031 per cent increase in real export volumes to Japan respectively.
Table 6.9 also shows us that both interaction terms are negative and are significant/jointly significant. For the US trade threshold model, we see that as the USD/AUD real exchange rate volatility exceeds the threshold value of -7.1402, the individual effect of USD/AUD real exchange rate volatility on real export volumes to US declines slightly from 0.11 per cent to 0.086 per cent but remains positive.
In addition, when the USD/AUD real exchange rate enters this excessively volatile period, the significant coefficient of -0.57 on DUM affects the exogenous real export volumes to US. As a result, the exogenous real export volumes changes from negative 4.92 per cent to negative 5.49 per cent. Therefore, for the US threshold model, we see that there is clearly an overall significant negative impact on real export volumes as a result of the USD/AUD real exchange rate volatility exceeding the threshold value.
For the Japan trade threshold model, when the YEN/AUD real exchange rate volatility exceeds the threshold value of -5.6624, the individual effect of the YEN/AUD real exchange rate volatility on real export volumes changes from a positive 0.04 per cent to a negative 0.076 per cent. This clearly indicates that Australian firms exporting to Japan do respond negatively when the YEN/AUD real exchange rate is excessively volatile.
The significant and negative coefficient of -0.776203 on the DUM variable also indicates that the exogenous real export volumes to Japan worsens, changing from a negative 7.49 per cent to a negative 8.26 per cent. As such, for Japan, it can also be observed when the YEN/AUD real exchange rate enters an excessively volatile period, there is an overall significant negative impact on real export volumes to Japan.
Therefore, the estimation of the US and Japan threshold trade models in this section has clearly shown that accounting for a threshold effect of real exchange rate volatility when it exists is important. In both the US and Japan cases, we see that when real exchange rate volatility exceeds a certain threshold level, there is an unambiguous significant overall negative impact on real export volumes to the US and Japan.