panel VAR estimation
Dear Statalisters,
this is about the estimation of a panel VAR. I have read the previous posts about this topic, then this should not be a repetition.
-----------
I have a macro panel (N=8/9, T=65) and I want to estimate a VAR including three endogenous variables. Given the dimension of my panel (long T, short N), Nickell's bias should be negligible. Then, as stated in Canova+Ciccarelli-2013 and Canova-2007, the Least Squares Dummy Variables estimator should be a sound estimator to apply. For an application: Lof+Malinen-2014.
As we know, there are no built-in commands to estimate panel-VAR in Stata. Although, there are two user-written routines available on the internet: 1) pvar, 2) xtvar.
1) pvar (by I.Love) applies a GMM estimator à-la Arellano+Bond to deal with Nickell's bias. Then, this is appropriate for large-N/short-T panels, which is not my case.
2) xtvar [Cagala, T. and Glogowsky, U. 2014. Panel Vector Autoregressions for Stata (xtvar)], which instead applies the Least Squares Dummy Variables Corrected estimator à-la Bun and Kiviet (2006).
"xtvar" is closer to what I need. However, I would like to enjoy the flexibility of the "var" built-in command available in Stata, and I come to my question now.
-----------
The "var" command in Stata does not work because of the panel structure. Nonetheless, I made it work by disregarding the panel structure and by considering the data as long time series.
Then, to perform the VAR estimation with the Least Squares Dummy Variables estimator, I use the "var" command and insert dummies as exogenous variables for each longitudinal unit:
xi: var y1 y2 y3, exog(i.id_group) lag(1/3)
Do you think this makes the trick? Or, do you believe that this possibly generates some kind of error in the estimation?
Any suggestion, warning, advice is very much appreciated.
Gianluca
PS
The other option is, first, to apply the within-transformation to the data, then, to use the "var" command as above but without exog().
---------------
References:
Bun, M. J. G. and J. F. Kiviet. 2006. The effects of dynamic feedbacks on LS and MM estimator accuracy in panel data models. Journal of
Econometrics. 132(2). 409-444.
Lof, M. & Malinen, T.
Does sovereign debt weaken economic growth? A panel VAR analysis
Economics Letters, Elsevier, 2014, 122, 403-407
Canova, F. & Ciccarelli, M.
Panel Vector Autoregressive Models: A Survey
Canova, F.
Methods for applied macroeconomic research
Princeton University Press, 2007, 13
this is about the estimation of a panel VAR. I have read the previous posts about this topic, then this should not be a repetition.
-----------
I have a macro panel (N=8/9, T=65) and I want to estimate a VAR including three endogenous variables. Given the dimension of my panel (long T, short N), Nickell's bias should be negligible. Then, as stated in Canova+Ciccarelli-2013 and Canova-2007, the Least Squares Dummy Variables estimator should be a sound estimator to apply. For an application: Lof+Malinen-2014.
As we know, there are no built-in commands to estimate panel-VAR in Stata. Although, there are two user-written routines available on the internet: 1) pvar, 2) xtvar.
1) pvar (by I.Love) applies a GMM estimator à-la Arellano+Bond to deal with Nickell's bias. Then, this is appropriate for large-N/short-T panels, which is not my case.
2) xtvar [Cagala, T. and Glogowsky, U. 2014. Panel Vector Autoregressions for Stata (xtvar)], which instead applies the Least Squares Dummy Variables Corrected estimator à-la Bun and Kiviet (2006).
"xtvar" is closer to what I need. However, I would like to enjoy the flexibility of the "var" built-in command available in Stata, and I come to my question now.
-----------
The "var" command in Stata does not work because of the panel structure. Nonetheless, I made it work by disregarding the panel structure and by considering the data as long time series.
Then, to perform the VAR estimation with the Least Squares Dummy Variables estimator, I use the "var" command and insert dummies as exogenous variables for each longitudinal unit:
xi: var y1 y2 y3, exog(i.id_group) lag(1/3)
Do you think this makes the trick? Or, do you believe that this possibly generates some kind of error in the estimation?
Any suggestion, warning, advice is very much appreciated.
Gianluca
PS
The other option is, first, to apply the within-transformation to the data, then, to use the "var" command as above but without exog().
---------------
References:
Bun, M. J. G. and J. F. Kiviet. 2006. The effects of dynamic feedbacks on LS and MM estimator accuracy in panel data models. Journal of
Econometrics. 132(2). 409-444.
Lof, M. & Malinen, T.
Does sovereign debt weaken economic growth? A panel VAR analysis
Economics Letters, Elsevier, 2014, 122, 403-407
Canova, F. & Ciccarelli, M.
Panel Vector Autoregressive Models: A Survey
Canova, F.
Methods for applied macroeconomic research
Princeton University Press, 2007, 13
Comment