VECM
Posted: Mon Dec 06, 2010 5:32 am
Hi everyone,
I have lots of problem with my empirical work. First of all I have 12 series, 9 of them are MENA stock indices while the 3 others are the stock index of 3 developped countries. My work consists of testing the linkages between these series. I have done the test of stationarity and found out that all the series ares I(1). while runnig the test of johansen, the first step consists of the determination of the number of lag of the VAR. I will run multivariate cointegration betwenn MENA markets, then multivariate cointegration between MENA and developped markets, then bivariate cointegration between the MENa markets and bivariate cointegration between MENA and developped markets. My question is: should i determine the lag length of the VAR for each group alone ( for exp VAr( MENA), VAR (mena, developped), VAR( Market1, market2)... ) or not ??
Next, i read that for the causality test, if the series are cointegrated the test is based on VECM and if the series are non stationary and are not cointegrated the test is based on VAR on first differences. How do I run these tests for these different cases ?
I have lots of problem with my empirical work. First of all I have 12 series, 9 of them are MENA stock indices while the 3 others are the stock index of 3 developped countries. My work consists of testing the linkages between these series. I have done the test of stationarity and found out that all the series ares I(1). while runnig the test of johansen, the first step consists of the determination of the number of lag of the VAR. I will run multivariate cointegration betwenn MENA markets, then multivariate cointegration between MENA and developped markets, then bivariate cointegration between the MENa markets and bivariate cointegration between MENA and developped markets. My question is: should i determine the lag length of the VAR for each group alone ( for exp VAr( MENA), VAR (mena, developped), VAR( Market1, market2)... ) or not ??
Next, i read that for the causality test, if the series are cointegrated the test is based on VECM and if the series are non stationary and are not cointegrated the test is based on VAR on first differences. How do I run these tests for these different cases ?