Resumen de Network analysis of the mexican stock market
- español
This study investigates the dynamics of equity networks in Mexico from 2018 to 2023, focusing on the impact of the COVID-19 pandemic. Methodological steps include calculating stock returns, estimating annual GARCH models, constructing lower-tailed dependency matrices, and forming networks based on these matrices. The characteristics of the resulting networks are described. In addition, 10,000 Erdos-Reyni simulations are performed to estimate GNAR models up to order two, selecting the best estimates according to AIC, BIC, and llk criteria. The predictive performance of GNAR models compared to univariate AR and VAR models is evaluated. These stages help to better understand the interconnection between Mexican financial markets, offering valuable insights for risk management and decision-making.
ANÁLISIS DE REDES DEL MERCADO MEXICANO ACCIONARIO RESUMEN Este estudio investiga la dinámica de las redes accionarias en México de 2018 a 2023, focalizándose en el impacto de la pandemia de COVID-19. Los pasos metodológicos incluyen el cálculo de rendimientos accionarios, la estimación de modelos GARCH anuales, la construcción de matrices de dependencia en colas inferiores y la formación de redes basadas en estas matrices. Se describen las características de las redes resultantes. Adicionalmente, se realizan 10,000 simulaciones Erdos-Reyni para estimar modelos GNAR hasta orden dos, seleccionando las mejores estimaciones según criterios AIC, BIC y llk. Se evalúa el desempeño predictivo de los modelos GNAR en comparación con modelos AR y VAR univariados. Estas etapas ayudan a comprender mejor la interconexión entre los mercados financieros mexicanos, ofreciendo información valiosa para la gestión de riesgos y la toma de decisiones.
- English
This study investigates the dynamics of equity networks in Mexico from 2018 to 2023, focusing on the impact of the COVID-19 pandemic. Methodological steps include calculating stock returns, estimating annual GARCH models, constructing lower-tailed dependency matrices, and forming networks based on these matrices. The characteristics of the resulting networks are described. In addition, 10,000 Erdos-Reyni simulations are performed to estimate GNAR models up to order two, selecting the best estimates according to AIC, BIC, and llk criteria. The predictive performance of GNAR models compared to univariate AR and VAR models is evaluated. These stages help to better understand the interconnection between Mexican financial markets, offering valuable insights for risk management and decision-making.
