Abstract
This paper analyzes the MSE of the exponentially weighted least squares (EWLS) estimator in dynamic regression models with time-varying parameters. Under the assumption of differentiable parameter functions, it is derived an asymptotic expression which is the sum of a stationary and of an evolutionary component. The validity of the analytical expression is illustrated with simulation experiments, and its usefulness in designing the exponential discounting factor is illustrated on a real case-study. The practical finding is similar to the plug-in bandwidth selection in nonparametric smoothers.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Arvastsons L, Olsson H and Holst J (2000). Asymptotic bias in parameter estimation in AR-processes is recursive least squares with exponential forgetting. Scand J Stat 27: 177–192
Belitser E (2000). Recursive estimation of a drifted autoregressive parameter. Ann Stat 28: 860–870
Bittanti S and Campi M (1994). Bounded error identification of time-varying parameters by RLS techniques. IEEE Trans Autom Control 39: 1106–1110
Boutahar M and Deniau C (1996). Least squares estimator for regression models with some deterministic time-varying parameters. Metr 43: 47–57
Box GEP and Jenkins GM (1976). Time Series Analysis: Forecasting and Control. Holden Day, San Francisco
Chon KH, Zhao H, Zou R and Ju K (2005). Multiple time-varying dynamic analysis using multiple sets of basis functions. IEEE Trans Biomed Eng 52: 956–960
Dahlhaus R (1996). Asymptotic statistical inference for nonstationary processes with evolutionary spectra. In: Robinson, PM and Rosemblatt, M (eds) Lecture notes in statistics vol 115, pp 145–159. Springer, New York
Dahlhaus R (1997). Fitting time series models to nonstationary processes. Ann Stat 25: 1–37
Grillenzoni C (1990). Modeling time varying dynamical systems. J Am Stat Assoc 85: 499–507
Grillenzoni C (1993). ARIMA processes with ARIMA parameters. J Bus Econ Stat 11: 235–250
Grillenzoni C (1996). Testing or causality in real time. J Econ 73: 335–376
Grillenzoni C (1999). Adaptive tests for changing unit roots in nonstationary time series. J Comput Graph Stat 8: 763–778
Guo L and Ljung L (1995). Performance analysis of general tracking algorithms. IEEE Trans Autom Control 40: 1388–1402
Lindoff B and Holst J (1996). Bias and covariance of the recursive least squares estimator with exponential forgetting in vector autoregressions. J Time Ser Analysis 17: 553–570
Ljung L and Söderström T (1983). Theory and practice of recursive identification. MIT Press, New York
Lund R, Shao Q and Basawa I (2006). Parsimonious periodic time series modelling. Australian NZ J Stat 48: 33–47
Moulines E, Priouret P and Roueff F (2005). On recursive estimation of time varying autoregressive processes. Ann Stat 33: 2610–2654
Pagilla P and Zhu Y (2006). Adaptive estimation of time-varying parameters in linearly parametrized systems. J Dyna Syst Measur Control 128: 691–695
Robinson PM (1989). Nonparametric estimation of time-varying parameters. In: Hackl, P (eds) Statistical analysis and forecasting of economic structural change, pp 254–264. Springer, Berlin
Stoica P and Nehorai A (1988). On the asymptotic distribution of exponentially weighted prediction error estimators. IEEE Trans Signal Process 36: 136–139
White H (2001). Asymptotic theory for econometricians, revised edn. Academic Press, New York
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Grillenzoni, C. Performance of adaptive estimators in slowly varying parameter models. Stat Meth Appl 17, 471–482 (2008). https://doi.org/10.1007/s10260-007-0083-3
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10260-007-0083-3