Análisis del efecto del COVID-19 en la mortalidad en España
Loading...
Official URL
Full text at PDC
Publication date
2025
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Citation
Abstract
El presente trabajo analiza la evolución temporal de la mortalidad mensual en España con el objetivo de modelar su comportamiento y estudiar su cambio tras la pandemia de COVID-19. Se realiza una descomposición de las series temporales para identificar patrones estacionales y de tendencia, observándose una marcada estacionalidad anual sin una tendencia clara antes de la pandemia. Para detectar cambios estructurales en la serie, se aplica el test de Bai-Perron, identificando puntos de ruptura que coinciden aproximadamente con las fases de impacto del COVID-19. Posteriormente, se evalúa la estacionariedad de la serie pre-COVID mediante los tests ADF y KPSS, concluyendo una estacionariedad débil, y se ajustan dos modelos: Holt-Winters y SARIMA(0, 0, 1) × (1, 1, 0)12. El análisis de residuos, las métricas de error y los criterios de información (AIC, BIC) indican que el modelo SARIMA ofrece un mejor ajuste. A partir de este modelo se generan predicciones para el periodo post-COVID, permitiendo estimar la mortalidad esperada y comparar con la observada. Se identifican periodos de exceso de mortalidad, especialmente al inicio de la pandemia, seguidos de una convergencia progresiva hacia los valores esperados.
Abstract: This paper analyzes the temporal evolution of monthly mortality in Spain with the aim of modeling its behavior and studying its changes after the COVID-19 pandemic. A decomposition of the time series is performed to identify seasonal and trend patterns, revealing a marked annual seasonality without a clear trend prior to the pandemic. To detect structural changes in the series, the Bai-Perron test is applied, identifying breakpoints that approxi- mately coincide with the phases of COVID-19 impact. Subsequently, the stationarity of the pre-COVID series is evaluated using the ADF and KPSS tests, concluding weak stationarity, and two models are fitted: Holt-Winters and SARIMA(0, 0, 1) × (1, 1, 0)12. Residual analysis, error metrics, and information criteria (AIC, BIC) indicate that the SARIMA model provides a better fit. Based on this model, forecasts are generated for the post-COVID period, allowing the estimation of expected mortality and comparison with observed values. Periods of excess mortality are identified, especially at the beginning of the pandemic, followed by a progressive convergence toward expected values.
Abstract: This paper analyzes the temporal evolution of monthly mortality in Spain with the aim of modeling its behavior and studying its changes after the COVID-19 pandemic. A decomposition of the time series is performed to identify seasonal and trend patterns, revealing a marked annual seasonality without a clear trend prior to the pandemic. To detect structural changes in the series, the Bai-Perron test is applied, identifying breakpoints that approxi- mately coincide with the phases of COVID-19 impact. Subsequently, the stationarity of the pre-COVID series is evaluated using the ADF and KPSS tests, concluding weak stationarity, and two models are fitted: Holt-Winters and SARIMA(0, 0, 1) × (1, 1, 0)12. Residual analysis, error metrics, and information criteria (AIC, BIC) indicate that the SARIMA model provides a better fit. Based on this model, forecasts are generated for the post-COVID period, allowing the estimation of expected mortality and comparison with observed values. Periods of excess mortality are identified, especially at the beginning of the pandemic, followed by a progressive convergence toward expected values.