1.Multicollinearity The classical linear regression model assumes the explanatory variables are not correlated with one another. However, this assumption is hard to hold in practice. Multicollinearity, is used to describe the problem that the explanatory variables are very highly correlated with each other. When applying Econometrics, the main purpose is to separate each explanatory’s biased influence on the explained variable. The exist of multicollinearity would destroy the system, so that we must test it. Perfect Multicollinearity and Imperfect Multicollinearity There are two types of multicollinearity: perfect multicollinearity and imperfect multicollinearity. Perfect multicollinearity means the exact relationship between variables. In this situation, it is not possible to estimate all of the coefficients in the model. While in practice, imperfect multicollinearity is more common to observe. It occurs when there is a non-negligible, but not perfect relationship between variables. Consequences of Multicollinearity The two classes of multicollinearity could …show more content…
One problem may owe to the inappropriate way of splitting sub-period. The Chow test assumes there is a known break point in the series. If this is not known, the Chow test is not appropriate. We could use the predictive failure test to do the test again to verify the result. Other reasonable approaches include splitting the data according to any obvious structural change in the series showing in the graph or any known important historical events. We could also adopt the forwards predictive failure test or backwards predictive failure test. A more widely used way to deal with the sub-set problem is Quandt likelihood ratio test. It can be seen as a modified version of Chow test. Beyond the splitting problem, another reason for the unsatisfied result may be the volatility of the time series