Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Structural Break in Panel Data

    Hello dear,
    Can you please guide me if there is a structural break test in Stata for panel data? Also, are there any unit root tests for panel data in Stata which can possibly incorporate known or unknown structural break?

  • #2
    Sidra:
    see
    Code:
    xtunitroot
    Kind regards,
    Carlo
    (Stata 18.0 SE)

    Comment


    • #3
      Dear Carlo,

      Thanks for reply. But I have actually been through xtunitroot and I do not remember finding any unit root test there which can take into account any known or unknown structural break in data. Can you please assure me the existence of any specific unit root test (for example with its name) within the xtunitroot command that can help me account for structural break in data?

      Thank you.

      Comment


      • #4
        As of Stata 14 there are no structural breaks test for panel data.
        In case you want to construct the Chow test statistics have a look at the Stata FAQ:

        https://www.stata.com/support/faqs/s...how-statistic/

        Otherwise, follow Carlo's suggestion.

        Comment


        • #5
          Dear Stefano,

          Thanks for your reply. While I am trying to understand Chow test and whether it can help me or not, can you please share your input about my response to Carlo's suggestion above?

          As I said above that I have actually been through xtunitroot and I do not remember finding any unit root test there which can take into account any known or unknown structural break in data. Can you please assure/confirm me about the existence of any specific unit root test (for example with its name) within the xtunitroot command that can help me account for any known or unknown structural break while checking the stationarity of my data?

          Thank you.

          Comment


          • #6
            Sydra:
            I do not know whether what follows might be useful, but typing -search xtunitroot- returns a set of user-written commands potentially consistent with your research goal (unfortunately, I'm not an expert in time-series analysis).
            Kind regards,
            Carlo
            (Stata 18.0 SE)

            Comment


            • #7
              Dear Carlo,

              Thanks for getting back to me. We are talking here in the context of panel data not the time-series data.

              Can you please confirm me do you get the following tests by typing -search xtunitroot- in stata?

              Levin-Lin-Chu test
              xtunitroot llc varname [if] [in] [, LLC_options]


              Harris-Tzavalis test

              xtunitroot ht varname [if] [in] [, HT_options]


              Breitung test

              xtunitroot breitung varname [if] [in] [, Breitung_options]


              Im-Pesaran-Shin test

              xtunitroot ips varname [if] [in] [, IPS_options]


              Fisher-type tests (combining p-values)

              xtunitroot fisher varname [if] [in], {dfuller | pperron} lags(#)
              [Fisher_options]


              Hadri Lagrange multiplier stationarity test

              xtunitroot hadri varname [if] [in] [, Hadri_options]

              Thank you.

              Comment


              • #8
                Sidra:
                I confirm that I find your very same tests under -xtunitroot-.
                Moreover, by typing -search xtunitroot- I get the following total entries:
                Code:
                Search of official help files, FAQs, Examples, SJs, and STBs
                
                [XT]    xtunitroot  . . . . . . . . . . . . . . . . Panel-data unit-root tests
                        (help xtunitroot)
                
                SJ-18-1 st0519  . . . . . . . Panel unit-root tests for heteroskedastic panels
                        . . . . . . . H. Herwartz, S. Maxand, F. H. C. Raters, and Y. M. Walle
                        (help xtpurt if installed)
                        Q1/18   SJ 18(1):184--196
                        implements the heteroskedasticity-robust panel unit-root tests
                        suggested in Herwartz and Siedenburg (2008, Computational
                        Statistics and Data Analysis 53: 137-150), Demetrescu and Hanck
                        (2012a, Economics Letters 117: 10-13), and, recently, Herwartz,
                        Maxand, and Walle (2017, Center for European, Governance and
                        Economic Development Research Discussion Papers 314)
                
                Web resources from Stata and other users
                
                (contacting http://www.stata.com)
                
                4 packages found (Stata Journal and STB listed first)
                -----------------------------------------------------
                
                hadrilm from http://fmwww.bc.edu/RePEc/bocode/h
                    'HADRILM': module to perform Hadri panel unit root test / hadrilm performs
                    a test for stationarity in heterogeneous panel / data (Hadri, 2000). This
                    Lagrange Multiplier (LM) test has a null / of stationarity, and its test
                    statistic is distributed as / standard normal under the null. The series
                
                ipshin from http://fmwww.bc.edu/RePEc/bocode/i
                    'IPSHIN': module to perform Im-Pesaran-Shin panel unit root test / ipshin
                    estimates the t-test for unit roots in heterogeneous / panels developed by
                    Im, Pesaran and Shin (IPS, 1997).  It allows / for individual effects,
                    time trends, and common time effects. / Based on the mean of the
                
                levinlin from http://fmwww.bc.edu/RePEc/bocode/l
                    'LEVINLIN': module to perform Levin-Lin-Chu panel unit root test /
                    levinlin estimates the panel unit root test developed by Levin, / Lin and
                    Chu (2002).  The test assumes that each individual unit / in the panel
                    shares the same AR(1) coefficient, but allows for / individual effects,
                
                xtfisher from http://fmwww.bc.edu/RePEc/bocode/x
                    'XTFISHER': module to compute Fisher type unit root test for panel data /
                    xtfisher combines the p-values from N independent unit root / tests, as
                    developed by Maddala and Wu (1999).  Based on the / p-values of individual
                    unit root tests, Fisher's test assumes / that all series are
                
                (click here to return to the previous screen)
                
                (end of search)
                My previous, possibly wrong, mention of time-series came from Chapter 8.10.7 (page 278) of https://www.stata.com/bookstore/micr...etrics-stata/:
                Panel methods for unit roots and cointegration are based on methods developed for a single time-series...
                Kind regards,
                Carlo
                (Stata 18.0 SE)

                Comment


                • #9
                  Sidra, all the available tests for stationarity in the panel context are those that you and Carlo reported. I'm aware of the Chow test and the supF test to detect statistical significance of structural changes. Whether they can be applied to panels in Stata I don't know, but I found a post some time ago that said it is not (currently) possible.

                  I just want to add that, in my personal, limited, experience structural changes are tested when the null of non-stationarity is marginally rejected, also considering the limits of standard unit root tests, for example the augmented Dickey-Fuller test. In addition, sometimes breaks can be seen from the plots. This doesn't mean that you can assert something based on visual diagnosis, but that you have a reason more to perform a specific test. I don't know exactly why you need to test structural changes but if your data show a unit root you might be able to proceed to the next phase of your analysis without too much concerns.

                  Said this, the only test I'm aware of in Stata is the estat sbsingle (supremum Wald test) but for time-series.

                  Hope this helps

                  Stefano

                  Comment


                  • #10
                    Maybe Stafano was referring to the following post:

                    https://www.statalist.org/forums/for...ith-panel-data

                    Comment


                    • #11
                      Indeed Dario, thanks

                      Comment


                      • #12
                        Stefano Grillini

                        # 9
                        but if your data show a unit root you might be able to proceed to the next phase of your analysis without too much concerns.
                        Hi, I may be wrong, but I think that you mean "if your data DOESN'T show a unit root" because I understand that a unit root (or non-stationarity) isn't good, is it?
                        --------------------
                        (Stata 15.1 MP)

                        Comment


                        • #13
                          Linh Nguyen and Sidra Ilyas regarding your question, I noted that Laurant Pauwels from University of Sidney presented a new routine (xtbreak) at the 2017 Oceania Stata meeting. This is the link:

                          https://www.stata.com/meeting/oceania17

                          I did not find the routine anyway. Possibly it is a work in progress . But you can try to contact him to have more feedback.
                          Last edited by Dario Maimone Ansaldo Patti; 13 Dec 2018, 13:52.

                          Comment


                          • #14
                            Thanks Dario Maimone Ansaldo Patti
                            --------------------
                            (Stata 15.1 MP)

                            Comment

                            Working...
                            X