A Comparative Note about Estimation of the Fractional Parameter under Additive Outliers
Files
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Pontificia Universidad Católica del Perú. Departamento de Economía
Abstract
En un artículo reciente, Fajardo et al. (2009) proponen un estimador semiparamétrico alternativo del parámetro fraccional en modelos ARFIMA que es robusto a la presencia de valores atípicos aditivos. Los resultados son muy interesantes, sin embargo, utilizan muestras de 300 ó 800 observaciones que rara vez se encuentran en la macroeconomía o la economía. Para realizar una comparación, yo uso el procedimiento para la detección de valores atípicos aditivos basados en el estimador Td propuesto por Perron y Rodríguez (2003). Además, utilizo variables Ficticias asociadas a la ubicación de los valores atípicos seleccionados para estimar el parámetro fraccional. Los resultados son mejores para la media y el sesgo de este parámetro cuando T = 100 y los resultados en términos de la desviación estándar y el MSE son muy similares. Sin embargo, para tamaños de muestra más altos como 300 ó 800, el procedimiento robusto tiene un mejor rendimiento, especialmente sobre la base de la desviación estándar y el MSE. Aplicaciones empíricas para siete series de inflación de América Latina, con muy pequeños tamaños de muestras contaminadas por los valores atípicos aditivos es discutida. Lo que encontramos es que cuando no se realiza ninguna corrección para los valores atípicos aditivos, se subestima el parámetro fraccional.
In a recent paper, Fajardo et al. (2009) propose an alternative semiparametric estimator of the fractional parameter in ARFIMA models which is robust to the presence of additive outliers. The results are very interesting; however, they use samples of 300 or 800 observations which are rarely found in macroeconomics or economics. In order to perform a comparison, I use the procedure to detect for additive outliers based on the estimator Td suggested by Perron and Rodríguez (2003). Further, I use dummy variables associated to the location of the selected outliers to estimate the fractional parameter. I found better results for the mean and bias of this parameter when T = 100 and the results in terms of the standard deviation and the MSE are very similar. However, for higher sample sizes as 300 or 800, the robust procedure performs better, specially based on the standard deviation and MSE measures. Empirical applications for seven Latin American inflation series with very small sample sizes contaminated by additive outliers are discussed. What we find is that when no correction for additive outliers is performed, the fractional parameter is underestimated.
In a recent paper, Fajardo et al. (2009) propose an alternative semiparametric estimator of the fractional parameter in ARFIMA models which is robust to the presence of additive outliers. The results are very interesting; however, they use samples of 300 or 800 observations which are rarely found in macroeconomics or economics. In order to perform a comparison, I use the procedure to detect for additive outliers based on the estimator Td suggested by Perron and Rodríguez (2003). Further, I use dummy variables associated to the location of the selected outliers to estimate the fractional parameter. I found better results for the mean and bias of this parameter when T = 100 and the results in terms of the standard deviation and the MSE are very similar. However, for higher sample sizes as 300 or 800, the robust procedure performs better, specially based on the standard deviation and MSE measures. Empirical applications for seven Latin American inflation series with very small sample sizes contaminated by additive outliers are discussed. What we find is that when no correction for additive outliers is performed, the fractional parameter is underestimated.
Description
Keywords
Outliers aditivos, Errores ARFIMA, Estimación semiparamétrica
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess