Hello everybody, my name is Khosrul and I am a PhD candidate.
I read around the discussions but I was unable to find an answer to my problem. I will try to explain myself briefly. In my thesis, I want to identify the effects of CO2 emissions on financial markets. Since it would be important to understand if these effects are symmetric or (more likely) not, I thought of using the two-step non-linear ARDL. In my understanding (but please correct me if I am wrong) the current frontier in this dimension is the methodology of Cho et al. (2020) (https://scholar.google.com.au/schola...Qu&oi=scholarr ).
With this method, one deploys a first step FM-TOLS (Fully modified transformed OLS) for long-run relationship estimation, and then, a second step OLS for short-run dynamics.
In the paper I reference above the authors used this approach at the country level. However, in my research I want to expand the method for the panel case. I am wondering if it makes any sense to apply the FM-TOLS framework into the panel data case? For example, Mean Group (MG) or Pooled Mean Group (PMG) type extension are defensible here? I thought hard about it but I couldn't come to a conclusive answer. Any recommendation (also in terms of reading) would be greatly, greatly appreciated!
I also considered shifting to a dynamic panel threshold model as per the methodology of Seo and Shin (2016) (https://doi.org/10.1016/j.jeconom.2016.03.005). This seems to me a more general approach, but I don't understand whether I can make such claim.
Thank you all
I read around the discussions but I was unable to find an answer to my problem. I will try to explain myself briefly. In my thesis, I want to identify the effects of CO2 emissions on financial markets. Since it would be important to understand if these effects are symmetric or (more likely) not, I thought of using the two-step non-linear ARDL. In my understanding (but please correct me if I am wrong) the current frontier in this dimension is the methodology of Cho et al. (2020) (https://scholar.google.com.au/schola...Qu&oi=scholarr ).
With this method, one deploys a first step FM-TOLS (Fully modified transformed OLS) for long-run relationship estimation, and then, a second step OLS for short-run dynamics.
In the paper I reference above the authors used this approach at the country level. However, in my research I want to expand the method for the panel case. I am wondering if it makes any sense to apply the FM-TOLS framework into the panel data case? For example, Mean Group (MG) or Pooled Mean Group (PMG) type extension are defensible here? I thought hard about it but I couldn't come to a conclusive answer. Any recommendation (also in terms of reading) would be greatly, greatly appreciated!
I also considered shifting to a dynamic panel threshold model as per the methodology of Seo and Shin (2016) (https://doi.org/10.1016/j.jeconom.2016.03.005). This seems to me a more general approach, but I don't understand whether I can make such claim.
Thank you all
Comment