Question

Q.3.2 What are the conditions for Granger causality testing?

Q.3.3 Explain in detail an alternative procedure to Granger causality if the conditions in Q.3.2 are not

met. (10)

Answer 1

The rough idea of the Granger Causality test is that the events of the past can cause events in the future but events in the future cannot cause past events. In the test, we have to make sure that the equations in the system are identified either exactly or over. This can be achieved by assuming that some of the predetermined variables are present in only some equations. The error terms entering the causality test have to be uncorrelated. There should not be any spurious causality.

When the conditions of the Granger Causality test are not met that is there is a spurious causality between the variables, then to consider such multi equation system we have to make use of vector auto regression (VAR). This system does not have exogenous variables. Hence, one need not worry to determine which variables are endogenous and which are exogenous. Also the usual OLS method can be applied to each equation separately.