Kybernetika 54 no. 6, 1138-1155, 2018

Robust recursive estimation of GARCH models

Tomáš Cipra and Radek HendrychDOI: 10.14736/kyb-2018-6-1138

Abstract:

The robust recursive algorithm for the parameter estimation and the volatility prediction in GARCH models is suggested. It seems to be useful for various financial time series, in particular for (high-frequency) log returns contaminated by additive outliers. The proposed procedure can be effective in the risk control and regulation when the prediction of volatility is the main concern since it is capable to distinguish and correct outlaid bursts of volatility. This conclusion is demonstrated by simulations and real data examples presented in the paper.

Keywords:

outlier, volatility, Kalman filter, GARCH model, robust recursive estimation

Classification:

62M10, 62F35, 91G70

References:

  1. A. Aknouche and H. Guerbyenne: Recursive estimation of GARCH models. Comm. Statist. Simul. Comput. 35 (2006), 925-938.   DOI:10.1080/03610910600880328
  2. N. S. Balke and T. B. Fomby: Large shocks, small shocks, and economic fluctuations: outliers in macroeconomics time series. J. Appl. Econometr. 31 (1994), 307-327.   DOI:10.1002/jae.3950090205
  3. T. Bernholt, R. Fried, U. Gather and I. Wegener: Modified repeated median filters. Statist. Comput. 16 (2006), 177-192.   DOI:10.1007/s11222-006-8449-1
  4. T. Bollerslev: Generalized autoregressive conditional heteroskedasticity. J. Econometr. 31 (1986), 307-327.   DOI:10.1016/0304-4076(86)90063-1
  5. A. Bose and K. Mukherjee: Estimating the ARCH parameters by solving linear equations. J. Time Series Anal. 24 (2003), 127-136.   DOI:10.1111/1467-9892.00296
  6. L. E. Calvet, V. Czellar and E. Ronchetti: Robust filtering. J. Amer. Statist. Assoc. 110 (2015), 1591-1606.   DOI:10.1080/01621459.2014.983520
  7. M. A. Carnero, D. Peña and E. Ruiz: Effects of outliers on the identification and estimation of GARCH models. J. Time Series Anal. 28 (2007), 471-497.   DOI:10.1111/j.1467-9892.2006.00519.x
  8. M. A. Carnero, D. Peña and E. Ruiz: Estimating GARCH volatility in the presence of outliers. Econom. Lett. 114 (2012), 86-90.   DOI:10.1016/j.econlet.2011.09.023
  9. A. Charles: Forecasting volatility with outliers in GARCH models. J. Forecast. 27 (2008), 551-565.   DOI:10.1002/for.1065
  10. A. Charles and O. Darné: Outliers and GARCH models in financial data. Econom. Lett. 86 (2005), 347-352.   DOI:10.1016/j.econlet.2004.07.019
  11. T. Cipra: Robust exponential smoothing. J. Forecast. 11 (1992), 57-69.   DOI:10.1002/for.3980110106
  12. T. Cipra: Robust recursive estimation in nonlinear time-series. Comm. Statist. Theory Methods 27 (1998), 1071-1082.   DOI:10.1080/03610929808832146
  13. T. Cipra and T. Hanzák: Exponential smoothing for time series with outliers. Kybernetika 47 (2011), 165-178.   CrossRef
  14. T. Cipra and R. Romera: Robust Kalman filter and its applications in time series analysis. Kybernetika 27 (1991), 481-494.   CrossRef
  15. R. Crevits and C. Croux: Forecasting using robust exponential smoothing with damped trend and seasonal components. Working paper KBI_1714, KU Leuven, Leuven 2016 (DOI:10.13140\/RG.2.2.11791.18080).   CrossRef
  16. C. Croux and S. Gelper: Computational aspects of robust Holt-Winters smoothing based on M-estimation. Appl. Math. 53 (2008), 163-176.   DOI:10.1007/s10492-008-0002-4
  17. C. Croux, S. E. C. Gelper and K. Mahieu: Robust exponential smoothing of multivariate time series. Comput. Statist. Data Anal. 54 (2010), 2999-3006.   DOI:10.1016/j.csda.2009.05.003
  18. R. Dalhaus and S. S. Rao: A recursive online algorithm for the estimation of time-varying ARCH parameters. Bernoulli 13 (2007), 389-422.   DOI:10.3150/07-bej5009
  19. R. F. Engle: Autoregressive conditional heteroskedasticity with estimates of the variance of United Kingdom inflation. Econometrica 50 (1982), 987-1007.   DOI:10.2307/1912773
  20. B. Eraker, M. Johannes and N. Polson: The impact of jumps in volatility and returns. J. Finance 58 (2003), 1269-1300.   DOI:10.1111/1540-6261.00566
  21. A. Fasso: Recursive least squares with ARCH errors and nonparametric modelling of environmental time series. Working Paper 6, University of Bergamo 2009.   CrossRef
  22. J. Franke, W. K. Härdle and C. M. Hafner: Statistics of Financial Markets: An Introduction. Springer, Berlin 2011.   DOI:10.1007/978-3-642-16521-4
  23. P. H. Franses and H. Ghijsels: Additive outliers, GARCH and forecasting volatility. Int. J. Forecast. 15 (1999), 1-9.   DOI:10.1016/s0169-2070(98)00053-3
  24. P. Galeano and D. Peña: Finding outliers in linear and nonlinear time series. In: Robustness and Complex Data Structures (C. Becker, R. Fried, S. Kuhnt, eds.), Springer, Berlin 2013, pp. 243-260.   DOI:10.1007/978-3-642-35494-6\_15
  25. S. Gelper, R. Fried and C. Croux: Robust forecasting with exponential and Holt-Winters smoothing. J. Forecast. 29 (2009), 285-300.   DOI:10.1002/for.1125
  26. L. Gerencsér, Z. Orlovits and B. Torma: Recursive estimation of GARCH processes. In: Proc. 19th International Symposium on Mathematical Theory and Systems - MTNS (A. Edelmayer, ed.), Eötvös Loránd University, Budapest 2010, pp. 2415-2422.   CrossRef
  27. A. Grané and H. Veiga: Wavelet based detection of outliers in financial time series. Comput. Statist. Data Anal. 54 (2010), 2580-2593.   DOI:10.1016/j.csda.2009.12.010
  28. A. V. Gregory and J. J. Reeves: Estimation and inference in ARCH model in the presence of outliers. J. Financ. Econometr. 8 (2010), 547-569.   DOI:10.1093/jjfinec/nbq028
  29. C. Grillenzoni: Optimized adaptive prediction. J. Ital. Statist. Soc. 6 (1997), 37-58.   DOI:10.1007/bf03178900
  30. C. Grillenzoni: Recursive generalized M-estimators of system parameters. Technometrics 39 (1997), 211-224.   DOI:10.2307/1270909
  31. R. Hendrych and T. Cipra: Robustified on-line estimation of the EWMA models: Simulations and applications. In: Proc. 33rd International Conference Mathematical Methods in Economics (D. Martinčák, J. Ircingová, and P. Janeček, eds.). University of West Bohemia, Pilsen 2014, pp. 237-242.   DOI:10.3311/ppee.9684
  32. R. Hendrych and T. Cipra: Self-weighted recursive estimation of GARCH models. Comm. Statist. Simul. Comput. 47 (2018), 315-328.   DOI:10.1080/03610918.2015.1053924
  33. J. B. Hill: Robust estimation and inference for heavy tailed GARCH. Bernoulli 21 (2015), 1629-1669.   DOI:10.3150/14-bej616
  34. L. K. Hotta and R. S. Tsay: Outliers in GARCH processes. In: Economic time series: Modeling and seasonality (W. R. Bell, S. H. Holan, and T. S. McElroy, eds.). CRC Press, Boca Raton 2012, pp. 337-358.   DOI:10.1201/b11823-20
  35. R. J. Hyndman, A. B. Koehler, J. K. Ord and R. D. Snyder: Forecasting with Exponential Smoothing. The State Space Approach. Springer, Berlin 2008.   DOI:10.1111/j.1751-5823.2009.00085\_17.x
  36. J. Jiang, Q. Zhao and Y. V. Hui: Robust modelling of ARCH models. J. Forecast. 20 (2001), 111-133.   DOI:10.1002/1099-131x(200103)20:2<111::aid-for786>3.0.co;2-n
  37. J. Kierkegaard, J. Nielsen, L. Jensen and H. Madsen: Estimating GARCH models using recursive methods.    CrossRef
  38. K. R. Koch and Y. Yang: Robust Kalman filter for rank deficient observation models. J. Geodesy 72 (1998), 436-441.   DOI:10.1007/s001900050183
  39. V. Lanius and U. Gather: Robust online signal extraction from multivariate time series. Comput. Statist. Data Anal. 54 (2010), 966-975.   DOI:10.1016/j.csda.2009.10.009
  40. S. Ling: Self-weighted and local quasi-maximum likelihood estimators for ARMA-GARCH/ IGARCH models. J. Econometr. 140 (2007), 849-873.   DOI:10.1016/j.jeconom.2006.07.016
  41. L. Ljung: System Identification: Theory for the User. Prentice Hall PTR, Upper Saddle River 1999.   CrossRef
  42. L. Ljung and T. S. Söderström: Theory and Practice of Recursive Identification. MIT Press, Cambridge 1983.   CrossRef
  43. J. Michálek: Robust methods in exponential smoothing. Kybernetika 32 (1996), 289-306.   CrossRef
  44. N. Muler and V. Yohai: Robust estimates for GARCH models. J. Statist. Planning Inference 138 (2008), 2918-2940.   DOI:10.1016/j.jspi.2007.11.003
  45. B.-J. Park: An outlier robust GARCH model and forecasting volatility of exchange rate returns. J. Forecast. 21 (2002), 381-393.   DOI:10.1002/for.827
  46. L. Peng and Q. Yao: Least absolute deviations estimation for ARCH and GARCH models. Biometrika 90 (2003), 967-975.   DOI:10.1093/biomet/90.4.967
  47. R. Romera and T. Cipra: On practical implementation of robust Kalman filtering. Comm. Statist. Simul. Comput. 24 (1995), 461-488.   DOI:10.1080/03610919508813252
  48. P. Ruckdeschel, B. Spangl and D. Pupashenko: Robust Kalman tracking and smoothing with propagating and non propagating outliers. Statist. Papers 55 (2014), 93-123.   DOI:10.1007/s00362-012-0496-4
  49. S. Sakata and H. White: High breakdown point conditional dispersion estimation with application to S\&P 500 daily returns volatility. Econometrica 66 (1998), 529-567.   DOI:10.2307/2998574
  50. H. U. Shaolin, K. Meinke, H. Ouyang and S. Guoji: Outlier-tolerant Kalman filter of state vectors in linear stochastic system. Int. J. Advanced Computer Sci. Appl. 2 (2011), 37-41.   DOI:10.14569/ijacsa.2011.021206
  51. T. S. Söderström and P. Stoica: System Identification. Prentice Hall, New York 1989.   CrossRef
  52. R. S. Tsay: Analysis of Financial Time Series. Wiley, Hoboken 2013.   CrossRef
  53. Y. Yang: Adaptively robust Kalman filters with applications in navigation. In: Sciences of Geodesy (G. Xu, ed.), Springer, Berlin 2010, pp. 49-82.   DOI:10.14569/ijacsa.2011.021206
  54. Y. Yang, W. Gao and X. Zhang: Robust Kalman filtering with constraints: a case study for integrated navigation. J. Geodesy 84 (2010), 373-381.   DOI:10.1007/s00190-010-0374-6
  55. R. Zhang and S. Ling: Asymptotic inference for AR models with heavy-tailed G-GARCH noises. Econometr. Theory 31 (2015), 880-890.   DOI:10.1017/s0266466614000632
  56. K. Zhu and W. K. Li: A new Pearson-type QMLE for conditionally heteroskedastic models. J. Business Econom. Statist. 33 (2015), 552-565.   DOI:10.1080/07350015.2014.977446
  57. K. Zhu and S. Ling: Global self-weighted and local quasi-maximum exponential likelihood estimators for ARMA-GARCH/IGARCH models. Ann. Statist. 39 (2011), 2131-2163.   DOI:10.1214/11-aos895
  58. K. Zhu and S. Ling: LADE-based inference for ARMA models with unspecified and heavy-tailed heteroscedastic noises. J. Amer. Statist. Assoc. 110 (2015), 784-794.   DOI:10.1080/01621459.2014.977386