Stationarity test for residual in panel data

[Khati Zolfaghari]

Hello everyone,
I want to test the statinarity (panel) for the residual in my regression.
I use

predict r

But through the test when I go to:
Longitudinal/panel data>unit root test
And select the test I got error.
my model is fixed effect and panel
.

Many thanks for your valuable time and advice.

Best regards

[William Lisowski]

It is hard to determine what caused the error when we don't know what the error was.

In the xtunitroot dialog you should select the dependent variable in your model. If you didn't select any variable, that would generate an error which you clicked the OK button in the dialog box.

If you selected your predicted value r, that was a mistake, and perhaps that would generate a "no observations" error, or some other error, in your Stata Results window.

Otherwise, copy from your Stata Results window the xtunitroot command that Stata generated and the error message it gave and post it here, again using CODE delimiters.

[Khati Zolfaghari]

@William Lisowski thank you so much for your reply. actually I select "r" , so it is my mistake and I recieve this error:

. asdoc reghdfe f(0).tot_birth l(0/2)dum_recession l(1/2)tot_birth , absorb( i.ifscode year) vce(cluster ifscode) nest reset dec(6)
(File Myfile.doc already exists, option append was assumed)
(MWFE estimator converged in 3 iterations)

HDFE Linear regression                            Number of obs   =      3,116
Absorbing 2 HDFE groups                           F(   5,    141) =    6530.91
Statistics robust to heteroskedasticity           Prob > F        =     0.0000
                                                  R-squared       =     0.9998
                                                  Adj R-squared   =     0.9998
                                                  Within R-sq.    =     0.9512
Number of clusters (ifscode) =        142         Root MSE        =     0.0218

                               (Std. Err. adjusted for 142 clusters in ifscode)
-------------------------------------------------------------------------------
              |               Robust
    tot_birth |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
dum_recession |
          --. |  -.0011832   .0015607    -0.76   0.450    -.0042685    .0019022
          L1. |  -.0018752   .0013405    -1.40   0.164    -.0045252    .0007748
          L2. |   .0007233   .0012545     0.58   0.565    -.0017567    .0032033
              |
    tot_birth |
          L1. |   1.323623   .0588983    22.47   0.000     1.207185     1.44006
          L2. |  -.3680806   .0538483    -6.84   0.000    -.4745351   -.2616261
              |
        _cons |   .5614883    .088636     6.33   0.000      .386261    .7367157
-------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
     ifscode |       142         142           0    *|
        year |        22           0          22     |
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

 predict r
(option xb assumed; fitted values)
(291 missing values generated)

. xtunitroot llc r
Levin-Lin-Chiu test requires strongly balanced data
r(498);

my adviser told me: after taking the residual of the regression do a test of stationarity (Panel) for residual . I am so thankful to recieve your advice.

many thanks for your valuable time and advice.

best regards,

[Khati Zolfaghari]

hello everyone, I hope I could receive your advice soon. many thanks,

best regards,

[Andrew Musau]

You have several issues in #3

predict r (option xb assumed; fitted values) (291 missing values generated)

-residual- is an option of predict, and therefore should be preceded by a comma. Your code is predicting the dependent variable, and Stata is explicitly telling you this. You want

predict, res

Secondly, if I recall correctly, only xtunitroot fisher handles unbalanced panels. See

help xtunitroot

[Khati Zolfaghari]

@Andrew Musau thank you so much for your reply, I did in this way. I have some problem in terms of the interpretation. so as I expect, it is non-stationary data and I can reject the null hypothesis. am I right? but it is based on the unit root of residual. I hope I could recieve your advice.

 reghdfe f(0).lnfertility l(0/2)dum_recession l(1/2)lnfertility , absorb( i.ifscode year) vce(cluster ifscode) residuals(res)
(MWFE estimator converged in 3 iterations)

HDFE Linear regression                            Number of obs   =      3,116
Absorbing 2 HDFE groups                           F(   5,    141) =    6592.37
Statistics robust to heteroskedasticity           Prob > F        =     0.0000
                                                  R-squared       =     0.9998
                                                  Adj R-squared   =     0.9998
                                                  Within R-sq.    =     0.9493
Number of clusters (ifscode) =        142         Root MSE        =     0.0190

                               (Std. Err. adjusted for 142 clusters in ifscode)
-------------------------------------------------------------------------------
              |               Robust
  lnfertility |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
--------------+----------------------------------------------------------------
dum_recession |
          --. |  -.0013254   .0015998    -0.83   0.409    -.0044881    .0018372
          L1. |  -.0015468   .0011184    -1.38   0.169    -.0037579    .0006642
          L2. |    .000273   .0010136     0.27   0.788    -.0017309    .0022769
              |
  lnfertility |
          L1. |   1.316072   .0545001    24.15   0.000     1.208329    1.423815
          L2. |    -.36521   .0512967    -7.12   0.000    -.4666201      -.2638
              |
        _cons |   .8267416   .1036365     7.98   0.000     .6218594    1.031624
-------------------------------------------------------------------------------

Absorbed degrees of freedom:
-----------------------------------------------------+
 Absorbed FE | Categories  - Redundant  = Num. Coefs |
-------------+---------------------------------------|
     ifscode |       142         142           0    *|
        year |        22           0          22     |
-----------------------------------------------------+
* = FE nested within cluster; treated as redundant for DoF computation

. predict r, resid
(292 missing values generated)

. xtunitroot fisher r, dfuller lags(0)
(292 missing values generated)

Fisher-type unit-root test for r
Based on augmented Dickey-Fuller tests
--------------------------------------
Ho: All panels contain unit roots           Number of panels       =    142
Ha: At least one panel is stationary        Avg. number of periods =  21.94

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(284)  P      2180.3843       0.0000
 Inverse normal            Z       -34.7407       0.0000
 Inverse logit t(714)      L*      -49.9196       0.0000
 Modified inv. chi-squared Pm       79.5705       0.0000
------------------------------------------------------------------------------
 P statistic requires number of panels to be finite.
 Other statistics are suitable for finite or infinite number of panels.
------------------------------------------------------------------------------

best regards,

[Andrew Musau]

The null hypothesis is that all panels have a unit root. So a significant test implies that at least one panel is stationary. The Hadri Lagrange multiplier test (implemented as xtunitroot hadri) has the null that all panels are stationary. It requires large \(T\) and moderate \(N\).

[Khati Zolfaghari]

@Andrew Musau thank you so much for your reply.

[Khati Zolfaghari]

Hello everyone, I get the first differences and run again the regression and I take the result, but when I test the unit root test of the resduals , it stay the same and non stations,
what should I do?
Best regards

[Khati Zolfaghari]

Hello and good day , I need your assistance. I hope I recieve your help .

best regards,