Table 3.
Gregory-Hansen cointegration test with structural break.
| Test Statistic | Breakpoint (Date) | |
|---|---|---|
| Model 1: break (level) | ||
| ADF | -5.203388 | (26/12/2014) |
| Zt | -4.989807 | (24/12/2014) |
| Za | -49.28401 | (24/12/2014) |
| Model 2: break (level and trend) | ||
| ADF | -5.189309 | (17/11/2014) |
| Zt | -5.080157 | (14/11/2014) |
| Za | -51.33842 | (14/11/2014) |
| Model 3: break (regime) | ||
| ADF | -5.507631 | (24/12/2014) |
| Zt | -5.468028 | (14/11/2014) |
| Za | -59.29250 | (14/11/2014) |
Note: The maximum lag used for the Gregory-Hansen cointegration test is 1 as suggested by the BIC information criterion for VAR lag-length selection. The AIC criterion suggested a VAR lag length of 3, however, the results are similar and the conclusions are the same6.
Source: Computed by the Authors