Finally, we emphasize that the degree of smoothness for trend estimation should be chosen by the user at the outset. In fact, we recommend to choose the percentage of smoothness on the basis of expert judgment. As the method is mainly designed to enhance comparability of trends we suggest carrying out a pilot study of the series to be smoothed (e.g. a group of interest rates). Thus, by visual inspection of the trends in the study group, we can fix an appropriate percentage and apply that value to all the series in that category. Even for just one series, say an intraday series, the idea is to calculate the trend for that series each and every day, with the same percentage of smoothness, no matter what that percentage is. Once the percentage of smoothness has been decided, the resulting trends for different time series of the same length and frequency of observation or for the same series with different lengths can be compared appropriately.
ACKNOWLEDGEMENTS
The authors gratefully acknowledge the constructive comments and suggestions provided by two anonymous referees. V. M. Guerrero thanks Asociacion Mexicana de Cultura, A. C. for providing him support´ through a Professorship on Time Series Analysis and Forecasting in Econometrics.
REFERENCES
1. Findley D, Monsell BC, Bell WR, Otto MC, Chen BCh. New capabilities and methods of the X-12-ARIMAseasonal adjustment program. Journal of Business and Economic Statistics 1998; 16:127–152.
2. Gomez V, Maravall A. Seasonal adjustment and signal extraction of time series. In´ A Course in Time Series Analysis, Chapter 8, Pena D, Tiao GC, Tsay RS (eds). Wiley: New York, 2001.˜
3. Hodrick RJ, Prescott E. Post-war U.S. business cycles: an empirical investigation. Journal of Money, Credit and Banking 1997; 29:1–16.
4. Maravall A. Stochastic linear trends. Models and estimators. Journal of Econometrics 1993; 56:5–37.
5. Baillie RT, Bollerslev T. Common stochastic trends in a system of exchange rates. The Journal of Finance 1989; 44:167–181.
6. Narayan PK, Smyth R. Are OECD stock prices characterized by a random walk? Evidence from sequential trendbreak and panel data models. Applied Financial Economics 2005; 15:547–556.
7. Tsay RS. Analysis of Financial Time Series. Wiley: New York, 2002.
8. King RG, Rebelo ST. Low frequency filtering and real business cycles. Journal of Economic Dynamics and Control 1993; 17:207–231.
9. Guerrero VM. Time series smoothing by penalized least squares. Statistics and Probability Letters 2007; 77: 1225–1234.
10. Guerrero VM. Estimating trends with percentage of smoothness chosen by the user. International Statistical Review 2008; 76:187–202.
11. Gomez V. Three equivalent methods for filtering nonstationary time series.´ Journal of Business and Economic Statistics 1999; 17:109–116.
12. Ruppert D, Wand MP, Carroll RJ. Semiparametric Regression. Cambridge University Press: New York, 2003.
13. Bell WR. Signal extraction for nonstationary time series. The Annals of Statistics 1984; 12:646–664.
14. Kaiser R, Maravall A. Measuring Business Cycles in Economic Time Series. Lecture Notes in Statistics, vol. 154. Springer: New York, 2001.
15. McElroy T. Matrix formulas for nonstationary ARIMA signal extraction. Econometric Theory 2008; 24:988–1009.
16. Maravall A, del R´ıo A. Temporal aggregation, systematic sampling, and the Hodrick–Prescott filter. Computational Statistics and Data Analysis 2007; 52:975–998.
17. Brewer KRW. Some consequences of temporal aggregation and systematic sampling for ARMA and ARMAXmodels. Journal of Econometrics 1973; 1:133–154.
Уважаемый посетитель!
Чтобы распечатать файл, скачайте его (в формате Word).
Ссылка на скачивание - внизу страницы.