Summary DISCUSSION AND CONCLUSION

111

CHAPTER V DISCUSSION AND CONCLUSION

A. Summary

1. At level, after testing with Dickey-Fuller test, Augmented Dickety-Fuller, and Phillips-Perron test, all variables are not stationary. Nevertheless, series are integrated at first difference noted as I1 therefore stationary. 2. For bivariate model, result from Granger causality test explains the short- run causal relationship between the variables. Only a single pairwise series represents a bidirectional causality that is between ISEQ and RTSI. In other words, ISEQ has impacts upon RTSI and reversely; the actual value of the former is resulted from the previous change upon the latter and vice versa. In a unidirectional short run causal relationship, IBOVESPA Granger causes RTSI, CNXNIFTY and JKSE. RTSI has impact on JKSE; SSE over ISEQ. PSIG influences JKSE, SSE and ATHEX. FTSEMIB Granger causes JKSE and ATHEX. ISEQ has influence over JKSE, PSIG and ATHEX; IBEX35 over JKSE, ISEQ and ATHEX. 3. In multivariate perspective, by the means of Johansen co-integration test, either with default lag length 4 or with the optimum lag 1, the presence of co-integrating equation always verified; five for the former and one for the latter. Effectively, the first hypothesis H 1 is supported that means series are 112 co-integrated. Since, the series derived from BRIIC and PIIGS are co- integrated, that implies shift persists during eleven years and co-movement could be drawn as conclusion. 4. In short term, using ECT, there is some disturbance within the series nevertheless; this latter is corrected between one to four months alongside eleven years. That is a trace of disequilibrium within the series. That trace tends to be more significant for Brazil in light of Russia, India and Spain for less than two months. Similarly to Russia that tends to decline with Brazil, Indonesia and Greece. Indian needs four months to sort out the disequilibrium with Brazil, Indonesia and Portugal; about three months for Indonesia in light of Russia and India. For China within about two months, that imbalance collapses in tie with Brazil, Portugal, Ireland for a negative impact and Greece. Considering the PIIGS countries, Portugal needs also about four months to correct the imbalance with both Ireland and Spain; whereas a couple of months for Italy in light with Portugal, Ireland, Greece and Spain. Only less than a month, this trace persists as well as for both Ireland with Brazil, Portugal, Italy and Spain as Greece with Russia, Italy and Spain. Thus, Spain requires three months to connect with Brazil, Portugal, Italy and Greece. VECM enhances these results for the short run disequilibrium between Russia and Ireland, China and Greece, Italy with Ireland and Spain, Ireland 113 and Portugal; Italy and Greece. Since the approach is different and the lag length used is optimu lag=1, results may not accurately the same. For instance, Brazil, India, Indonesia, and Portugal represent no trace of disequilibrium in short term, and Italy apart from Ireland has also Spain. 5. In the long term, in light of the VECM, for Brazil, only Italy and Ireland have no impact over the long-run equilibrium relationship between both groups. Besides, only Brazil, Italy and Ireland do not have influence over the long-run relationship for no significant value towards Russia, India, Indonesia, China, Portugal, Italy, Ireland, Greece and Spain. Brazil is mostly6 influenced positively by Indonesia and Greece; negatively by India, Portugal and Spain similar to the case of Russia. For India, only Portugal has the most positive impact and Indonesia the negative. Both Indonesia and China over influenced positively and negatively by Greece and Portugal respectively. PIIGS countries on the other side, negatively also positively influenced by Greece, Spain, Portugal, India and Indonesia. 6. Almost all variables, that is to say RTSI, CNXNIFTY, JKSE, SSE, PSIG, ATHEX and IBEX35 have impact on the others as well as positively as negatively. Only IBOVESPA, FTSEMIB and ISEQ have not. The disequilibrium takes place only for a very short period of time less than four months during eleven years. Hence, the second hypothesis H 2 of strong co-movement is supported. 114 7. Briefly, at the first period, the variable itself because of shock responds that shock greatly, while considering IRF. This fact changes more and more pursuing the following countries. Before and after the crisis, the variables tend to look for the average value of response, which has a trace of decrease at the heart of the crisis. 8. FEVD highlights the importance of every variable because of shock, not far from IRF, at the first period then almost every variable has great impact on itself though more and more increases, starting from 100 to 17.5 on average.

B. Discussion and conclusion