Firstly you should identify the two markets by some variables. Secondly, you should check the stationarity of the variables using some unit root tests such as ADF or PP tests. If you find that the variables are integrated of the same order. In this case the variables can be cointegrated. Thirdly, you should use the multivariate cointegration approach of Johansen (1990, 1991) to see if the variables are cointegrated. Finally, if there is at least one cointegration relationship, you should estimate the vector error correction model to determine the dynamic causal relationships between the variables representing the two markets.

I wish that my answer help you.

Thanks Mounir that was very helpful. by the way how do you estimate the Long-term multiplier and Speed of adjustment from the vector error correction model. thanks

Dear Francis, as you said you estimate the long-term multiplier and the peed of adjustment (coefficient of the error correction term) by estimating the VECM using the software Eviews for example.