unit root test - Fisher-type for unbalanced panel

[rozita saadan]

Hello Statalist,

I had run a test for units roots for panel data. Since the data is unbalanced, so I used Fisher-type test as the null hypothesis that all panels contain a unit root.
As the result below, there are many panels could not be computed. However, by looking at P-value, the null hypothesis is rejected and the data contain no unit root.
My question is, as some of panel could not be computed, does the P-value valid to reject the null hypothesis?

 xtunitroot fisher return, dfuller lags(0)
(4,525,121 missing values generated)

could not compute test for panel 7
could not compute test for panel 11
could not compute test for panel 17
could not compute test for panel 18
could not compute test for panel 19
could not compute test for panel 20
could not compute test for panel 113
could not compute test for panel 125
could not compute test for panel 171
could not compute test for panel 274
could not compute test for panel 277
could not compute test for panel 316
could not compute test for panel 341
could not compute test for panel 346
could not compute test for panel 359
could not compute test for panel 400
could not compute test for panel 463
could not compute test for panel 509
could not compute test for panel 622
could not compute test for panel 697
could not compute test for panel 719
could not compute test for panel 744
could not compute test for panel 821
could not compute test for panel 859
could not compute test for panel 873
could not compute test for panel 986
could not compute test for panel 987
could not compute test for panel 1042
could not compute test for panel 1049
could not compute test for panel 1077
could not compute test for panel 1081

Fisher-type unit-root test for return
Based on augmented Dickey-Fuller tests
--------------------------------------
Ho: All panels contain unit roots           Number of panels       =   1095
Ha: At least one panel is stationary        Avg. number of periods =1167.49

AR parameter: Panel-specific                Asymptotics: T -> Infinity
Panel means:  Included
Time trend:   Not included
Drift term:   Not included                  ADF regressions: 0 lags
------------------------------------------------------------------------------
                                  Statistic      p-value
------------------------------------------------------------------------------
 Inverse chi-squared(2128) P       7.15e+04       0.0000
 Inverse normal            Z      -251.1396       0.0000
 Inverse logit t(5279)     L*     -606.2025       0.0000
 Modified inv. chi-squared Pm     1062.8323       0.0000
------------------------------------------------------------------------------
 P statistic requires number of panels to be finite.
 Other statistics are suitable for finite or infinite number of panels.
------------------------------------------------------------------------------

Thanks in advanced for your help.

Regards,
Rozita

[mohina saxena]

Hello All,

I have exactly the same question, please help in resolving this issue.
Should I treat the data as stationary even when it is not able to compute for large number of panels in the data set.
I also have unbalanced data that spans for 18 years and contains 3000 panel Id's. I get same error message as the above question does followed by the results table. Can someone please clarify why does'nt it compute for some of the panels? How to proceed further for such cases?
Please help!!

[mohina saxena]

Can someone please help in resolving this question or at-least give me a lead from where I can settle it. I face exactly the same problem as #1.
Should I even go for unit root test in such cases? Or should I just report the last table results even when the test is not able to compute stationarity for certain panels.

Please help!!

Many Thanks and regards,
Mohina

[Katie Zhang]

Can anyone helps with this issue???Thanks!