Good day I am trying to estimate a cointegrating relationship between RGDP and four other explanatory variables for a panel of countries using the Westerlund (2007) cointegration test (XTWEST), after establishing that each of the series are integrated. The test results from the xtwest command shows no cointegration for the group of variables. However, combining RGDP with each of the explanatory variables separately reports a cointegrating relationship. Please how can i reconcile these conflicting results.. Example of the two results are provided below: xtwest y ec co2, constant trend lags(1 3) leads(0 3) lrwindow(3) Calculating Westerlund ECM panel cointegration tests.......... Results for H0: no cointegration With 20 series and 2 covariates Average AIC selected lag length: 2.25 Average AIC selected lead length: 1.55 -----------------------------------------------+ Statistic | Value | Z-value | P-value | -----------+-----------+-----------+-----------| Gt | -2.473 | 0.294 | 0.616 | Ga | -5.187 | 5.152 | 1.000 | Pt | -8.660 | 1.741 | 0.959 | Pa | -5.238 | 3.473 | 1.000 | -----------------------------------------------+
and xtwest y ec, westerlund constant trend lags(1 3) leads(0 3) lrwindow(3) Calculating Westerlund ECM panel cointegration tests.......... Results for H0: no cointegration With 20 series and 1 covariate Average AIC selected lag length: 2.95 Average AIC selected lead length: 2.1 -----------------------------------------------+ Statistic | Value | Z-value | P-value | -----------+-----------+-----------+-----------| Gt | -3.193 | -4.662 | 0.000 | Ga | -14.293 | -1.610 | 0.054 | Pt | -5.938 | 4.089 | 1.000 | Pa | -11.036 | -1.559 | 0.060 | -----------------------------------------------+
Thank you Lateef Akanni
