Hi Everyone,
I am running panel unit root test to test for nonstationarity of variables in Stata. I am using a dataset of N = 44 and T = 26. I am using LLC test in Stata. WhenI am suppressing the individual panel effects i.e., panel specific constants in LLC test, I am getting this result:
xtunitroot llc x, noconstant lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: root(N)/T -> 0
Panel means: Not included
Time trend: Not included
ADF regressions: 0.16 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t 1.3262 0.9076
Adjusted t* 1.2451 0.8934
------------------------------------------------------------------------------
However, when I am adding back the constant the test result is:
xtunitroot llc x, lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: N/T -> 0
Panel means: Included
Time trend: Not included
ADF regressions: 0.16 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t -11.6883
Adjusted t* -6.2082 0.0000
------------------------------------------------------------------------------
Test result with constant + trend is:
xtunitroot llc x, trend lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: N/T -> 0
Panel means: Included
Time trend: Included
ADF regressions: 0.23 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t -20.5140
Adjusted t* -9.2485 0.0000
------------------------------------------------------------------------------
My questions are: Why the LLC test is only accepting the null hypothesis of "Panels contain unit roots" when I am removing the panel specific constants/Individual-specific effects? What do the above results signify?
Kindly please help! Thank you very much!
Regards,
Sudatta
I am running panel unit root test to test for nonstationarity of variables in Stata. I am using a dataset of N = 44 and T = 26. I am using LLC test in Stata. WhenI am suppressing the individual panel effects i.e., panel specific constants in LLC test, I am getting this result:
xtunitroot llc x, noconstant lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: root(N)/T -> 0
Panel means: Not included
Time trend: Not included
ADF regressions: 0.16 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t 1.3262 0.9076
Adjusted t* 1.2451 0.8934
------------------------------------------------------------------------------
However, when I am adding back the constant the test result is:
xtunitroot llc x, lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: N/T -> 0
Panel means: Included
Time trend: Not included
ADF regressions: 0.16 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t -11.6883
Adjusted t* -6.2082 0.0000
------------------------------------------------------------------------------
Test result with constant + trend is:
xtunitroot llc x, trend lags(aic)
Levin-Lin-Chu unit-root test for x
----------------------------------
Ho: Panels contain unit roots Number of panels = 44
Ha: Panels are stationary Number of periods = 26
AR parameter: Common Asymptotics: N/T -> 0
Panel means: Included
Time trend: Included
ADF regressions: 0.23 lags average (chosen by AIC)
LR variance: Bartlett kernel, 9.00 lags average (chosen by LLC)
------------------------------------------------------------------------------
Statistic p-value
------------------------------------------------------------------------------
Unadjusted t -20.5140
Adjusted t* -9.2485 0.0000
------------------------------------------------------------------------------
My questions are: Why the LLC test is only accepting the null hypothesis of "Panels contain unit roots" when I am removing the panel specific constants/Individual-specific effects? What do the above results signify?
Kindly please help! Thank you very much!
Regards,
Sudatta
