Table 3.
Results for the KPSS-PUR Test (Real GDP per capita).
| Panel A: Structural Breaks | ||||||||||
| Countries | # of Brks | Break 1 | Break 2 | Break 3 | Break 4 | Break 5 | Break 6 | Break 7 | ||
| Belgium (BE) | 6 | 1979 | 1985 | 1992 | 1996 | 2006 | 2016 | |||
| Denmark (DK) | 7 | 1984 | 1988 | 1993 | 1997 | 2002 | 2006 | 2016 | ||
| France (FR) | 7 | 1981 | 1985 | 1990 | 1998 | 2006 | 2010 | 2016 | ||
| Ireland (IR) | 7 | 1980 | 1985 | 1993 | 1997 | 2005 | 2009 | 2014 | ||
| Italy (IT) | 6 | 1978 | 1982 | 1987 | 1994 | 1999 | 2010 | |||
| Luxembourg (LU) | Reference Country | |||||||||
| Netherlands (NL) | 6 | 1981 | 1985 | 1990 | 1998 | 2007 | 2016 | |||
| United Kingdom(UK) | 6 | 1982 | 1986 | 1990 | 1998 | 2006 | 2016 | |||
| Panel B: Optimal Frequency | ||||||||||
| Frequency Selection |
Bootstrap critical values |
|||||||||
| Countries |
Hypotheses |
Frequency |
F Statistics |
10% |
5% |
1% |
||||
| Belgium (BE) | H0:ρ1 = ρ2 = 0 | 1 | 20.095*** | 2.768 | 3.739 | 5.947 | ||||
| Denmark (DK) | H0:ρ1 = ρ2 = 0 | 14 | 5.361** | 2.679 | 3.600 | 5.910 | ||||
| France (FR) | H0:ρ1 = ρ2 = 0 | 10 | 13.877*** | 3.131 | 4.205 | 7.154 | ||||
| Ireland (IR) | H0:ρ1 = ρ2 = 0 | 10 | 4.165** | 2.516 | 3.318 | 5.394 | ||||
| Italy (IT) | H0:ρ1 = ρ2 = 0 | 2 | 19.640*** | 2.876 | 3.837 | 6.075 | ||||
| Luxembourg (LU) | Reference Country | |||||||||
| Netherlands (NL) | H0:ρ1 = ρ2 = 0 | 10 | 11.260*** | 2.610 | 3.547 | 5.833 | ||||
| United Kingdom(UK) | H0:ρ1 = ρ2 = 0 | 8 | 3.300* | 2.648 | 3.513 | 5.847 | ||||
| Panel C: Cross-sectional Dependence Test | ||||||||||
| Statistic | P value | |||||||||
| CD Test | 16.565*** | <0.01 | ||||||||
| Panel D: KPSS-PUR Test | ||||||||||
| Cross-section independent |
Bootstrap critical values |
|||||||||
| Panel Stationary Test |
Statistics |
P values |
10% |
5% |
1% |
|||||
| Z(λ) Homogeneous | 6.274 | <0.01 | 7.822 | 8.571 | 9.927 | |||||
| Z(λ) Heterogeneous | 17.425 | <0.01 | 17.945 | 20.704 | 25.962 | |||||
Note: “***”, ”**”, and “*” represent 1%, 5%, and 10% significance levels, respectively. The finite sample critical values are computed using the bootstrap method, replicating samples 10,000 times. The maximal number of breaks and frequencies are fixed at 7, and 20, respectively.