Hello everyone,
I am trying to estimate a Cross-Sectional Augmented Distributed Lag (CS-ARDL) model using Stata's xtdcce2 command, following the approach by Chudik and Pesaran (2015). I found a model in a paper that looks like this:
CO2_jt = γ0 + Σ (ρj * CO2_{t-j}) + Σ (σj * R_{t-j}) + Σ (τj * Z_{t-j}) + εjt
where:
R represents financial development (FD), green innovation (GI), renewable energy (RE), and economic growth (EG).
Z = (ΔCO2, avg_R), where ΔCO2 is the first difference of CO2 emissions, and avg_R represents the cross-sectional averages of R variables.
I have written the following xtdcce2 command in Stata:
I would like to confirm:
1. Does this command correctly implement the CS-ARDL model?
2. Are there any improvements I should make to correctly capture cross-sectional dependence and heterogeneity?
3. How should I properly specify the lag structure?
Thank you.
I am trying to estimate a Cross-Sectional Augmented Distributed Lag (CS-ARDL) model using Stata's xtdcce2 command, following the approach by Chudik and Pesaran (2015). I found a model in a paper that looks like this:
CO2_jt = γ0 + Σ (ρj * CO2_{t-j}) + Σ (σj * R_{t-j}) + Σ (τj * Z_{t-j}) + εjt
where:
R represents financial development (FD), green innovation (GI), renewable energy (RE), and economic growth (EG).
Z = (ΔCO2, avg_R), where ΔCO2 is the first difference of CO2 emissions, and avg_R represents the cross-sectional averages of R variables.
I have written the following xtdcce2 command in Stata:
Code:
xtdcce2 CO2, lr(l.CO2 l.FD l.GN l.RE l.EG) lr_options(ardl) cr(d.CO2 FD GI RE EG) cr_lags(1)
1. Does this command correctly implement the CS-ARDL model?
2. Are there any improvements I should make to correctly capture cross-sectional dependence and heterogeneity?
3. How should I properly specify the lag structure?
Thank you.
Comment