This study mainly used annual time series data for the entire period of 1980-2009 and all of the data were driven from Central Bank of Malaysia. The data used in this study are the fiscal adjustment and gross domestic product. Basically, equation indicates the Engle-Granger bivariate co-integration equation and it is used to test whether the residuals, mainly the error term are stationary in level using ADF or PP stationary tests. The Augmented Dickey-Fuller (ADF) and Phillip-Perron (PP) stationary statistic result is equal to -1.87 and -1.85.

It is important to determine the characteristics of the individual series before conducting the cointegration analysis. Many studies have shown that majority of macroeconomics variable time series are not stationary, rather stationary with a deterministic trend. This creates a problem for econometricians since in the conditions of non-stationary data the normal properties t-statistics and Durbin Watson statistics and measures such as R-squares break results. To test the order of integrations, we used ADF and PP stationary tests. It is widely acknowledged that ADF and PP tests are command stationary tests applied in macroeconomics variable studies recently.

Table 1 summarizes the outcome of the ADF and PP tests with trend on both variables in this study. The null hypothesis tested is that the variable under investigation has a unit root against the alternative that it does not. In the first half of Table 1, the null hypothesis that each variable has a unit root cannot be rejected by both ADF and PP tests. However, after applying the first difference, both ADF and PP tests reject the null hypothesis. Since the data appear to be stationary by applying the ADF and PP tests in first differences, no further tests are performed (Loganathan et. al., 2010). The null hypothesis that each variable is integrated of order in the same order, which is in /: Once both variables seems to be stationary in /, we employ the Gregory-Hansen framework which able to capture endogenous determined break with three alternative forms for a structural break; level shift, level shift with trend and regime shift.

The p1 slope represents the cointegrating equation, and a1 indicate the changes in the intercept coefficient. With this specification, cointegration is present only with the break in year 1997 for three of the Gregory-Hansen models. Basically, the Gregory Hansen estimation technique starts with a simple regression equation where a constant term (ц1) is included similar to the one in Engle-Granger approach. Table 2 shows the Gregory-Hansen estimation results: Model I in Table 2 reports the results of Gregory-Hansen long-run cointegration procedure in level shift effects. Since the ADF result is equal to -4.21, therefore the null hypothesis of no cointegration at 1% level with an endogenous break of 1997 is rejected. The dummy coefficient is equal to -0.09 and statistically significant at 1% level. Meanwhile, the long-run coefficients is equal 0.38 and also rejects the null hypothesis, which indicates that EP coefficient is positively cointegrated with FA in the long-run. Using the same endogenous break year in Model II, the ADF statistic is -4.10 and significant at 1% level. While the intercept coefficient is -0.10 and significant at 1% level; and the long-run coefficient is equal to 0.35 with 1% significant level. However, the time trend coefficient is insignificant level through this study. Finally, Model III presents the results for regime shift which also indicates ADF with 1% significant level (-4.88). While the intercept coefficient is -0.04; p is equal to 0.56; and p1 is equal to 0.21. All of these coefficients are significant at 1% level with different level of p1 slopes.

This study investigates whether there are structural breaks in cointegrating vectors of the Malaysia fiscal adjustment and economic performance function between 1980 and 2009. Several interesting finding has been achieved through this study. First, the Engle-Granger cointegration test failed to reject the null hypothesis of no integration between fiscal adjustment and economic performance. Secondly, once endogenous structural break used via Gregory-Hansen cointegration procedure, the long-run relationship between fiscal adjustment and economic performance appears in this study. The empirical Gregory-Hansen long-run cointegration findings from the ADF integration tests suggest that all three model became / after the stationary conducted. The Gregory-Hansen tests reveal that there exist structural breaks in the cointegrating vectors of the Malaysian long-run fiscal adjustment and economic performance function. Thus, the Malaysian evidence seems to suggest that economic performance viable sustainable tools for fiscal adjustment stability. The choice of Gregory-Hansen through this study is propelled by its superiority to other long-run cointegration techniques, especially the residual based Engle-Granger approach, Johansen and Juselius, Pesaran and Shin, Pesaran and Smith and Pesaran et al.. All these alternative models fail to capture the endogenous structural breaks in long-run equilibrium relationship and Gregory-Hansen cointegration estimation approach has fulfilled the gap with three different specifications.

**Table 1: Result for Stationary Tests**

Variables | ADF (t) | ^{PP }(ZJ) |
||

Level | First Different | Level | First Different | |

FA | -1.63 | -3.77* | -1.62 | -3.73* |

EP | -1.68 | -4.78* | -1.64 | -4.79* |

**Table 2: Gregory-Hansen Cointegration Tests**

Model Specification | ADF | H_{o} of no Cointegration |

Model /: Level shift | ||

FA_{t} = 0.18 – 0.09DUM + 0.38EP_{t} (0.08)* (0.02)* (0.12)* |
-4.21 * | Reject |

Model //: Level shift with trend | ||

FA_{t} = 0.18 – 0.10DUM + 0.01y + 0.35EP_{t} (0.01)* (0.02)* (0.18) (0.84)* |
-4.10 * | Reject |

Model ///: Regime shift | ||

FA_{t} = 0.19 – 0.04DUM + 0.56EP_{t} + 0.21(DUM x EP_{t}) (0.01)* (0.01)* (0.04)* (0.07)* |
-4.88 * | Reject |

This study mainly used annual time series data for the entire period of 1980-2009 and all of the data were driven from Central Bank of Malaysia. The data used in this study are the fiscal adjustment and gross domestic product. Basically, equation indicates the Engle-Granger bivariate co-integration equation and it is used to test whether the residuals, mainly the error term are stationary in level using ADF or PP stationary tests. The Augmented Dickey-Fuller (ADF) and Phillip-Perron (PP) stationary statistic result is equal to -1.87 and -1.85. It is important to determine the characteristics of the individual series before conducting the cointegration analysis. Many studies have shown that majority of macroeconomics variable time series are not stationary, rather stationary with