Research Article

Panel Unit Root Tests by Combining Dependent 𝑃 Values: A Comparative Study

Table 1

Size and power of panel unit root tests: cross-section independence.

NTP S

20200.0590.0560.0520.063
500.0540.0460.0530.050
1000.0470.0470.0530.052
50200.0400.0440.0500.045
500.0480.0450.0480.045
1000.0570.0540.0470.058
100200.0510.0500.0500.047
500.0470.0510.0480.057
1000.0520.0450.0490.049

20200.0660.0720.0520.056
500.0870.0800.0620.083
1000.1720.1440.1120.0174
50200.0740.0810.0540.065
500.1230.1280.0640.102
1000.3030.2510.1210.276
100200.0800.0850.0480.066
500.1670.1650.0600.126
1000.4640.3660.1300.435

20200.1200.1440.0660.083
500.4170.4890.1060.253
1000.9510.9310.3600.860
50200.1810.2610.0580.108
500.7490.8380.1190.454
1001.0001.0000.4170.998
100200.2920.4220.0590.142
500.9500.9790.1080.683
1001.0001.0000.4471.000

20200.1820.2830.0580.105
500.8160.9330.1270.495
1001.0001.0000.5800.994
50200.3580.5620.0540.154
500.9941.0000.1510.817
1001.0001.0000.6491.000
100200.5910.8340.0690.257
501.0001.0000.1560.969
1001.0001.0000.6761.000

Note. Rejection rates of panel unit root tests at nominal level , using 2000 simulations. is Maddala and Wu’s [5] original Fisher test, is Demetrescu et al.’s [7] modified inverse normal method, is Hanck’s [8] Simes test, and is Sheng and Yang’s [9] modified TPM.