Modelización financiera mediante modelos híbridos Arima–Garch: evidencia para Argentina
The aim of this work is to model the volatility pattern during the historical stock return of the most important index of the Buenos Aires Stock Exchange (MERVAL) from January 1 of 2013 to June 6 of 2016, using the family of hybrid Arima-Garch models. The study is based on econometrics bibliography...
Guardado en:
| Autores principales: | , |
|---|---|
| Formato: | Artículo revista |
| Lenguaje: | Español |
| Publicado: |
Ediciones UNL
2020
|
| Materias: | |
| Acceso en línea: | https://bibliotecavirtual.unl.edu.ar/publicaciones/index.php/CE/article/view/9268 |
| Aporte de: |
| Sumario: | The aim of this work is to model the volatility pattern during the historical stock return of the most important index of the Buenos Aires Stock Exchange (MERVAL) from January 1 of 2013 to June 6 of 2016, using the family of hybrid Arima-Garch models. The study is based on econometrics bibliography with a focus on stock index modeling for other emerging economies. The conditions to employ this family of models are verified. The analysis confirms the existence of asymmetry and a leverage effect, which is the reason why the asymmetric E-Garch and GJR-Garch models are used, with both normal and student’s t distributions. For different orders of the aforementioned specifications, the models are repeatedly estimated. For the selection of models to use insample, the Schwarz information criterion is opted for. The estimated models are subject to hypothesis testing in order to guarantee compliance with the following properties: each systematic component of the process is taken into account, there are no bias of indicators or magnitude, and there is parameter stability. Then, the out-of-sample forecast performance is tested. Finally, it is observed that the E-Garch ~ t (1, 1), with ARMA (2,0) and ARMA (2,1) models, is superior in-sample and its forecasting performance is not significantly inferior to the other estimated models. |
|---|