Problems using -xtdpd- for dynamic panel

[jensb]

Hi all

I am running an unbalanced dynamic panel data regression using the xtdpd. I am using stata/se 12.0

I have the following data:

. summarize d_udlaan d_overdaekning solvpct

Variable | Obs Mean Std. Dev. Min Max
-------------+--------------------------------------------------------
d_udlaan | 1403 -.0104751 .0341303 -.1158373 .0574174
d_overdaek~g | 1401 -.1675946 1.020612 -3.7 1.9
solvpct | 1464 16.08552 2.684648 10.3 21.5

when running the command:
xtdpd L(0/2).d_udlaan L(0/1).d_overdaekning L(0/2).solvpct, noconstant div(L(0/1).d_overdaekning L(0/2).solvpct) dgmmiv(d_udlaan)
I get the error: No observations, and this is where i am stuck. I have tried to run the xtdpd example from the documentation, which works fine.

Anyone know what I am doing wrong ?

Thanks i advance

[E. David Aja]

It's hard to know based on the data summary you've posted here, but one thing you might consider checking is the -delta- option of -xtset-, which is relevant if your data is spaced by more than one time unit.

[jensb]

Thanks for replying. Could it be a problem that the date variable is specified as eg. 201006. 201009 as a quarter reference. Although i have no probles running an ordinary fixed effect panel data regression....

xtdescribe

regnr: 400, 522, ..., 13460 n = 76
refper: 2.0e+05, 2.0e+05, ..., 2.0e+05 T = 27
Delta(refper) = 1 quarter
Span(refper) = 607 periods
(regnr*refper uniquely identifies each observation)

Distribution of T_i: min 5% 25% 50% 75% 95% max
12 13 24 26 27 27 27

Freq. Percent Cum. | Pattern*
---------------------------+--------------------------------------------------------------------------------------------------------
31 40.79 40.79 | 21..............22..............121..............1 21..............22..............121..............1 21
3 3.95 44.74 | 21..............22..............121..............1 21..............12..............121..............1 21
3 3.95 48.68 | 21..............22..............121..............1 21..............22................................ ..
3 3.95 52.63 | 21..............22..............121..............1 21..............22..............111..............1 21
3 3.95 56.58 | 21..............22..............121..............1 21..............22..............12................ ..
2 2.63 59.21 | 21..............22..............121..............1 .................................................. ..
2 2.63 61.84 | 21..............22..............121..............1 21..............22..............12...............1 21
2 2.63 64.47 | 21..............22..............121..............1 21..............22..............121..............1 2.
1 1.32 65.79 | .1..............22..............121..............1 21..............22..............121..............1 21
26 34.21 100.00 | (other patterns)
---------------------------+--------------------------------------------------------------------------------------------------------
76 100.00 | XX..............XX..............XXX..............X XX..............XX..............XXX..............X XX

[Nick Cox]

201006 means to Stata the integer 201,006.

However, you have declared your date as a quarterly date, so to Stata they are all quarterly dates of the order of 200,000 i.e. this is some kind of economic science fiction set nearly 50,000 years in the future, except that Stata keeps a straight face.

Read help dates and times repeatedly until you understand this!

Tell us how quarters 1 to 4 in 2006 would be represented: 200103, 201006, 201009, 201012?

NB Please register with your full real name.

[jensb]

Thanks. Properly managing dates did the job.