5.1.2.2.2.2.6.2 Cusum square test
This test can detect structural instabilities. In this case
the curve is not outside the corridor (Figure11, Appendix 5); model
coefficients are stable. We deduce that the ECM is regularly stable.
5.1.2.2.2.2.7
Evaluation of model predictive power
According to the criterion of Thiel, if U is zero, the
prediction is perfect.
If U = 1 the prediction method is considered good (Figure12,
Appendix5). It should be 0? U? 1
The result of evaluation of the predictive power of the model
(Figure12, Appendix5) shows that:
MAPE = 0.570525% (Average Absolute Errors in percentage) and
Criterion U= 0.003937 (close to zero) (See figure 12, Appendix7)
This error correction model can be used for the forecasting
purposes.
Conclusion:
The different tests allow us to come to the conclusion
of a correct specification of the model, its validity and predictive power.
Let's turn now to the interpretation.
| 
