Dear all,
I am currently writing my thesis and my data is an unbalanced panel data set with 82 countries over a time period of 20 years. I'm trying to deal with the issue of cointegration between the variables and therefore I first of all conduct a panel unit root test, using the Fisher-ADF method to see if my variables have a unit root. The Fisher-ADF test rejects the null hypothesis (unit root is present) at level for some variables, at first difference for some variables and not at all for some variables. In other words the data consists of;
- stationary variables at level
- stationary variables at first difference
- non-stationary variables at both level and first difference (and if differentiated further)
How do I deal with a data set with variables like this and which test are appropriate? Panel cointegration test do not seem appealing since only a few of my variables are non-stationary.
Best regards,
Christopher Marton
I am currently writing my thesis and my data is an unbalanced panel data set with 82 countries over a time period of 20 years. I'm trying to deal with the issue of cointegration between the variables and therefore I first of all conduct a panel unit root test, using the Fisher-ADF method to see if my variables have a unit root. The Fisher-ADF test rejects the null hypothesis (unit root is present) at level for some variables, at first difference for some variables and not at all for some variables. In other words the data consists of;
- stationary variables at level
- stationary variables at first difference
- non-stationary variables at both level and first difference (and if differentiated further)
How do I deal with a data set with variables like this and which test are appropriate? Panel cointegration test do not seem appealing since only a few of my variables are non-stationary.
Best regards,
Christopher Marton
Comment