Design of cointegrating vectors and its empirical implications

• Autores: Olorunsola E. Olowofeso, A.B. Fagbohun
• Localización: International journal of the computer, the internet and management, ISSN 0858-7027, Vol. 10, Nº. 1, 2002, págs. 1-19
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • This work examines the design of cointegrating vectors by evaluating the performance of standard tests for unit roots/ stationary. It shows the behaviour of the estimated coefficients of the models with little variation in the number of replications. The maximum likelihood estimates were obtained by using Johansen¿s algorithm. The system design indicates from the Monte Carlo experiments that for the cointegrating system to be efficient and consistent, the method must incorporate all prior knowledge about the presence of unit roots.

    This eliminates the median bias, the nonsymmentry; part of the nuisance parameter dependencies and increase efficiency.

    Empirically, the vector error correction estimates shows that the higher the lag interval the better the error correction model and the cointegrating equation provided the sample size of the series is not too short.