GAUSS Codes

Tests with Structural Breaks

    With Given Break Points

• Perron's (augmented type) Test (Econometrica, 1989) (perron.txt)

   

• Amsler and Lee's LM (augmented type) Test (Econometric Theory, 1995) (AL.txt)

   

• Stationarity Test with Multiple Breaks (Editor's book, Lee) (stat.txt)

   With Data-dependent Break Points (Gauss)

       (data: gdp.txt  or  gdp.xls)

• Zivot and Andrews Test (One break) (JBES, 1992) (ZA.txt)
   

• Lumsdaine and Papell Test (Two breaks) (Restat, 1997) (LPtwo.txt)

• Lee and Strazicich LM Test (One break) (LS_One.g)
   

• Lee and Strazicich LM Test (Two breaks, Restat, 2003) (LS_two.g)
   

• LM Test with Multiple Breaks (multi.txt) (not ready)

With Data-dependent Break Points (RATS)

       (data: gdp.txt  or  gdp.xls)

Note:  The following RATS codes are extended (cloned) from LSunit.src and other codes, which are posted on the Estima web site

     http://www.estima.com/forum/viewtopic.php?t=126

These RATS codes may produce a little different results.  The difference can be minor and negligible in large samples.  In the following codes, the number of discarded initial observations will be smaller as it varies depending on the selected lag value.

• Zivot and Andrews Test (One break) (JBES, 1992) (not ready)
   

• Lumsdaine and Papell Test (Two breaks) (Restat, 1997) (not ready)
   

• Schmidt-Phillips Augmented Test (LS_UROOT.SRC, and LS_UROOT_run.prg)
   

• Lee and Strazicich LM Test (One break) (LS_UROOT.SRC, and  LS_UROOT_run.prg)
   

• Lee and Strazicich LM Test (Two breaks, Restat, 2003) (LS_UROOT.SRC, and LS_UROOT_run.prg)
   

Tests with a Fourier Non-linear Trend Function (Enders and Lee; not ready)

• DF Type Test  (Gauss Code | RATS Code)

• LM type Test (Gauss Code | RATS Code)

Panel Unit Root Tests

• Panel LM Unit root Test with Breaks

   (data: gdp.txt  or  gdp.xls)

Note:  The Panel LM tests are free of the nuisance parameters indicating the break points.  Critical values (Table 1): panelcri.dat

Note:  In the following codes, both the optimal (heterogeneous) break points and optimal lags are endogenously determined from the data, for each of the cross-section units.  One break (two break) test assumes one break (two breaks) in all cross section units.  

To mitigate the effects of cross-correlations, time fixed effects are allowed as an option.

• No break Panel LM test : panel_sp.g

• One break Panel LM test: panel_one.g

• Two break Panel LM test: panel_two.g

Note: It may be desirable to allow for different numbers of breaks for different cross-section units.  This may need additional modification of the above codes.