Danielle H Ferry wrote:
Can someone help me understand the error message, below (r430)? It
appears to be due to the "1" in the "ar()" condition. If I change the
"ar" option to "ar(2 3), I do not get this message. I also do not get
this message if I specify "ar(1)" but eliminate the regressors
"msa2-msa19" (and I get it if I specify "ar(1)" but keep "msa2-msa19".
[...]
-----------------------------------------
There must be some gremlins lurking in your data, because specifying
the -, ar(1 2 3) cond robust- option in -arima- works for me. Take a look:
. use http://www.stata-press.com/data/r9/friedman2.dta
. tsset time
time variable: time, 1946q1 to 1998q3
. arima consump m1 m2, ar(1 2 3) cond robust
(setting optimization to BHHH)
Iteration 0: log pseudolikelihood = -737.35182
Iteration 1: log pseudolikelihood = -721.77363
[...]
Iteration 40: log pseudolikelihood = -646.37582
Time-series regression -- AR disturbances
Sample: 1959q1 to 1998q3 Number of obs = 159
Wald chi2(5) = 1.71e+06
Log pseudolikelihood = -646.3758 Prob > chi2 = 0.0000
------------------------------------------------------------------------------
| Semi-robust
consump | Coef. Std. Err. z P>|z| [95% Conf. Interval]
-------------+----------------------------------------------------------------
consump |
m1 | .683976 .3032892 2.26 0.024 .0895402 1.278412
m2 | .043226 .1733303 0.25 0.803 -.2964951 .3829472
_cons | 197.8273 44.34107 4.46 0.000 110.9204 284.7342
-------------+----------------------------------------------------------------
ARMA |
ar |
L1. | 1.194732 .1194655 10.00 0.000 .9605834 1.42888
L2. | .1719741 .1677275 1.03 0.305 -.1567658 .500714
L3. | -.3584871 .1343686 -2.67 0.008 -.6218447 -.0951295
-------------+----------------------------------------------------------------
SIGMA2 |
_cons | 198.8691 31.52518 6.31 0.000 137.0809 260.6573
------------------------------------------------------------------------------
It also works my own (non-economic) datasets, too (although in my world,
specifying AR lags > 1 is unheard of!).
If I'm reading the error you get right, it's saying your data contains a
non-trivally long window of missing values (in your response variable?), and
essentially it's that which is preventing -arima- from calculating sensible
estimates. But others may be able to give you better advice.
Hope this helps,
Clive Nicholas
___________________________________________________________
Yahoo! Mail is the world's favourite email. Don't settle for less, sign up for
your free account today http://uk.rd.yahoo.com/evt=44106/*http://uk.docs.yahoo.com/mail/winter07.html
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
