Additive Outliers in Open-Loop Threshold Autoregressive Models: A Simulation Study

• Autores: Sergio Calderon, Daniel Ordoñez Callamad
• Localización: Revista Colombiana de Estadística, ISSN-e 2389-8976, ISSN 0120-1751, Vol. 45, Nº. 1, 2022, págs. 1-40
• Idioma: inglés
• DOI: 10.15446/rce.v45n1.92965
• Títulos paralelos:
  • Datos atípicos aditivos en modelos autorregresivos de umbrales: un estudio de simulación
• Enlaces
  • Texto completo
• Resumen
  • español

    Resumen Se investiga el efecto de observaciones atípicas aditivas en la adaptación de una prueba de no linealidad y un método de estimación robusto para los coeficientes autorregresivos de modelos SETAR(self-exciting threshold autoregressive) a modelos open-loop TAR(threshold autoregressive). A través de un experimento Monte Carlo se estudia la potencia y el tamaño de la prueba de no linealidad. Respecto a la estimación, se compara el sesgo y la razón de error cuadrático medio entre el estimador robusto y el de mínimos cuadrados. Adicionalmente, se evalúa la aproximación de la distribución empírica de los estimadores GM de los coeficientes a la distribución normal univariada junto a los niveles de cobertura de los intervalos de confianza asintóticos. Los resultados indican que la prueba de no linealidad adaptada presenta una potencia superior a la basada en mínimos cuadrados y no presenta distorsiones en el tamaño bajo la presencia de datos atípicos aditivos. Por otro lado, el método de estimación robusto para los coeficientes autorregresivos supera al de mínimos cuadrados en términos de error cuadrático medio bajo la presencia de este tipo de observaciones. Estos resultados fueron análogos a los obtenidos para modelos SETAR. Finalmente, se ilustra a través de dos ejemplos reales el uso de la prueba de no linealidad y el método de estimación.

  • English

    Abstract The effect of additive outlier observations is investigated in adapting a non-linearity test and a robust estimation method for the autoregressive coefficients from SETAR(self-exciting threshold autoregressive) models to open-loop models. TAR (threshold autoregressive). Through a Monte Carlo experiment, the power and size of the non-linearity test are studied. Regarding the estimation, the bias and the mean square error ratio between the robust estimator and the least-squares estimator are compared. Additionally, the approximation of the GM estimators' empirical distribution to the univariate normal distribution is evaluated together with the coverage levels of the asymptotic confidence intervals. The results indicate that the adapted non-linearity test has higher power than that based on least squares and does not present distortions in size under the presence of additive outliers. On the other hand, the robust estimation method for autoregressive coefficients exceeds the least-squares one in terms of the mean square error in the presence of this type of observations. These results were analogous to those obtained for SETAR models. Finally, the use of the non-linearity test and the estimation method are illustrated through two real examples.

• Referencias bibliográficas
  • Battaglia, F.,Orfei, L. (2005). 'Outlier detection and estimation in nonlinear time series'. Journal of Time Series Analysis. 26....
  • Calderón, S.,Nieto, F. (2017). 'Bayesian analysis of Multivariate Threshold Autoregressive Models with Missing Data'. Communications...
  • Chan, K.,Tong, H. (1990). 'On likelihood ratio tests for threshold autoregression'. Journal of the Royal Statistical Society. Series...
  • Chan, W.-S.,Cheung, S.-H. (1994). 'On robust estimation of threshold autoregressions'. Journal of Forecasting. 13. 37-49
  • Chan, W.-S.,Ng, M.-W. (2004). 'Robustness of alternative non-linearity tests for setar models'. Journal of Forecasting. 23. 215
  • Chen, C.,Liu, L.-M. (1993). 'Joint estimation of model parameters and outlier effects in time series'. Journal of the American Statistical...
  • Denby, L.,Martin, R. D. (1979). 'Robust estimation of the first-order autoregressive parameter'. Journal of the American Statistical...
  • Franses, P. H.,Van Dijk, D.. (2000). Non-linear time series models in empirical finance. Cambridge University Press.
  • Giordani, P. (2006). 'A cautionary note on outlier robust estimation of threshold models'. Journal of Forecasting. 25. 37-47
  • Gonzalez, J.,Nieto, F. (2020). 'Bayesian analysis of multiplicative seasonal threshold autoregressive processes'. Revista Colombiana...
  • Granger, C.,Terásvirta, T. (1993). Modelling Non-Linear Economic Relationships. Oxford University Press.
  • Hampel, F. R.,Ronchetti, E. M.,Rousseeuw, P. J.,Stahel, W. A. (1986). Robust statistics: the approach based on influence functions. John Wiley...
  • Hansen, B. E. (2011). 'Threshold autoregression in economics'. Statistics and its Interface. 4. 123
  • Hung, K. C.,Cheung, S. H.,Chan, W.-S.,Zhang, L.-X. (2009). 'On a robust test for setar-type nonlinearity in time series analysis'....
  • Knotters, M.,De Gooijer, J. (1999). 'Tarso modeling of water table depths'. Water Resources Research. 35. 695-705
  • LeBaron, B. (1992). 'Some relations between volatility and serial correlations in stock market returns'. Journal of Business. 199-219
  • Luukkonen, R.,Saikkonen, P.,Terásvirta, T. (1988). 'Testing linearity against smooth transition autoregressive models'. Biometrika....
  • Maronna, R.,Martin, R.,Yohai, V.,Salibián-Barrera, M. (2019). Robust Statistics Theory and Methods (with R). John Wiley and Sons.
  • Mohammadi, H.,Jahan-Parvar, M. R. (2012). 'Oil prices and exchange rates in oil-exporting countries: evidence from tar and m-tar models'....
  • Petruccelli, J. (1990). 'A comparison of tests for setar-type non-linearity in time series'. Journal of Forecasting. 9. 25-36
  • Petruccelli, J.,Davies, N. (1986). 'A portmanteau test for self-exciting threshold autoregressive-type nonlinearity in time series'....
  • Romero, L.,Calderón, S. (2021). 'Bayesian estimation of a multivariate TAR model when the noise process follows a Student-t distribution'....
  • Rousseeuw, P. J. (1984). 'Least median of squares regression'. Journal of the American statistical association. 79. 871
  • Rousseeuw, P. J.,Van Zomeren, B. C. (1990). 'Unmasking multivariate outliers and leverage points'. Journal of the American Statistical...
  • Sorour, A.,Tong, H. (1993). 'A note on tests for threshold-type non-linearity in open loop systems)'. Journal of the Royal Statistical...
  • Tiao, G. C.,Tsay, R. S. (1994). 'Some advances in non-linear and adaptive modelling in time-series'. Journal of Forecasting. 13. 109
  • Tong, H. (1978). 'On a threshold model'. Pattern Recognition and Signal Processing.
  • Tong, H. (1990). Non-linear Time Series: A Dynamical System Approach. Dynamical System Approach, Clarendon Press.
  • Tsay, R.,Chen, R. (2019). Nonlinear Time Series Analysis. Wiley Interscience.
  • Tsay, R. S. (1988). 'Outliers, level shifts, and variance changes in time series'. Journal of forecasting. 7. 1-20
  • Tsay, R. S. (1989). 'Testing and modeling threshold autoregressive processes'. Journal of the American statistical association. 84....
  • Tsay, R. S. (1998). 'Testing and Modeling Multivariate Threshold Models'. Journal of the American Statistical Association. 93. 467
  • Vargas, H. (2011). B. for International Settlements, ed., 'Capital flows, commodity price movements and foreign exchange intervention',...
  • Zhang, H.,Nieto, F. (2015). 'Tar modeling with missing data when the white noise process follows a students t-distribution'. Revista...
  • Zhang, L.-X.,Chan, W.-S.,Cheung, S.-H.,Hung, K.-C. (2009). 'A note on the consistency of a robust estimator for threshold autoregressive...