1. I have used the paper titled "Estimation of Dynamic Panel Threshold Model Using Stata" by Seo, Kim, and Kim (2019), and I wanted to clarify a few aspects regarding the FDGMM threshold model. Specifically, I would like to confirm if the FDGMM model is effective in handling a mix of I(1) and I(0) variables, as mentioned in the paper.
In our current analysis, we are using the following variables:
xthenreg emppop internet ai inflation l_pr gdp_growth fdi_gdp school_enrollment, grid_num(100) trim_rate(0.1) boost(200)
2. I have been working on an error correction model (ECM) using the xtpmg command in Stata and encountered a situation that I need some clarification on. Below is the command I used for the ECM:
xtpmg d.emppop d.ai d.internet d.school_enrollment d.l_pr d.inflation d.fdi_gdp d.gdp_growth, lr(l.emppop l.ai l.internet l.school_enrollment l.l_pr l.inflation l.fdi_gdp l.gdp_growth) ec(ec) replace pmg
The results returned a negative significant coefficient for the error correction term. I am wondering if this result justifies the previous xthenreg model that I had estimated.
In our current analysis, we are using the following variables:
- Non-stationary variables (I(1)): empop, internet, ai, school_enrollment, l_pr
- Stationary variables (I(0)): inflation, gdp_growth, fdi_gdp
xthenreg emppop internet ai inflation l_pr gdp_growth fdi_gdp school_enrollment, grid_num(100) trim_rate(0.1) boost(200)
2. I have been working on an error correction model (ECM) using the xtpmg command in Stata and encountered a situation that I need some clarification on. Below is the command I used for the ECM:
xtpmg d.emppop d.ai d.internet d.school_enrollment d.l_pr d.inflation d.fdi_gdp d.gdp_growth, lr(l.emppop l.ai l.internet l.school_enrollment l.l_pr l.inflation l.fdi_gdp l.gdp_growth) ec(ec) replace pmg
The results returned a negative significant coefficient for the error correction term. I am wondering if this result justifies the previous xthenreg model that I had estimated.
