Hello,
I'm trying to estimate a PVAR model using a strongly balanced dataset using the new xtvar command in Stata 19. I have no problem running the xtvar examples, however when estimating it with published datasets to replicate pvar results elsewhere the xtvar command gets hung up with a generic error message 'singular matrix'. (Note that I can replicate using the user developed pvar command for example). Could you point out what the issue is with xtvar?
Here are the relevant commands and output:
.xtdescribe
panel_id: 1, 2, ..., 21 n = 21
year: 1999, 2000, ..., 2016 T = 18
Delta(year) = 1 unit
Span(year) = 18 periods
(panel_id*year uniquely identifies each observation)
Distribution of T_i: min 5% 25% 50% 75% 95% max
18 18 18 18 18 18 18
Freq. Percent Cum. | Pattern
---------------------------+--------------------
21 100.00 100.00 | 111111111111111111
---------------------------+--------------------
21 100.00 | XXXXXXXXXXXXXXXXXX
.xtvar dtariff gY infl tby, lags(1) collapse
singular matrix encountered
The moment covariance matrix based on the one-step residuals is singular
and therefore not invertible. You may use option maxldep() or option
collapse to control the number of instruments used for estimation.
Please advise.
I'm trying to estimate a PVAR model using a strongly balanced dataset using the new xtvar command in Stata 19. I have no problem running the xtvar examples, however when estimating it with published datasets to replicate pvar results elsewhere the xtvar command gets hung up with a generic error message 'singular matrix'. (Note that I can replicate using the user developed pvar command for example). Could you point out what the issue is with xtvar?
Here are the relevant commands and output:
.xtdescribe
panel_id: 1, 2, ..., 21 n = 21
year: 1999, 2000, ..., 2016 T = 18
Delta(year) = 1 unit
Span(year) = 18 periods
(panel_id*year uniquely identifies each observation)
Distribution of T_i: min 5% 25% 50% 75% 95% max
18 18 18 18 18 18 18
Freq. Percent Cum. | Pattern
---------------------------+--------------------
21 100.00 100.00 | 111111111111111111
---------------------------+--------------------
21 100.00 | XXXXXXXXXXXXXXXXXX
.xtvar dtariff gY infl tby, lags(1) collapse
singular matrix encountered
The moment covariance matrix based on the one-step residuals is singular
and therefore not invertible. You may use option maxldep() or option
collapse to control the number of instruments used for estimation.
Please advise.
