Spurious regression under broken-trend stationarity. (English) Zbl 1112.62096

Noriega, Antonio E.; Ventosa-Santaulària, Daniel

J. Time Ser. Anal. 27, No. 5, 671-684 (2006).

Two spurious linear regression models are considered of the form \(y_t=\alpha+\delta x_t+u_y\) and \(y_t=\alpha+\beta t+\delta x_t+u_y\), where \(u_t\) is an error term, and \(x_t\) and \(u_t\) are the observed time series generated independently as trend-stationary processes. The trends of \(x_t\) and \(y_t\) are linear with multiple breaks. The asymptotic behaviour of OLS estimates for \(\alpha\), \(\beta\), \(\delta\) and the t-statistics for hypotheses like \(H_0:\;\delta=0\) is investigated. It is shown that the phenomenon of spurious regression occurs asymptotically independent of the error structure in the data generating process but the rate of divergence of the t-statistics is lower (\(T^{1/2}\)) than in the case of linear trends without breaks (\(T^{3/2}\)). Results of simulations are presented.

Reviewer: R. E. Maiboroda (Kyïv)

MSC:

62M10	Time series, auto-correlation, regression, etc. in statistics (GARCH)
62J05	Linear regression; mixed models
62F12	Asymptotic properties of parametric estimators
62F03	Parametric hypothesis testing
62G10	Nonparametric hypothesis testing

Keywords:

ordinary least squares estimate; t-statistics; asymptotic behaviour

References:

[1]	DOI: 10.1162/003465397557132
[2]	DOI: 10.2307/2998540 · Zbl 1056.62523
[3]	DOI: 10.1016/S0304-4076(97)00041-9 · Zbl 0887.62088
[4]	DOI: 10.1016/0304-4076(74)90034-7 · Zbl 0319.62072
[5]	C. W. J. Granger, N. Hyung, and Y. Jeon(1998 ) Spurious regressions with stationary series . University of California at San Diego, Mimeo.
[6]	Hansen B., Journal of Economic Perspectives 15 pp 117– (2001)
[7]	DOI: 10.1016/S0165-1889(01)00071-9 · Zbl 1023.91044
[8]	T.H. Kim, Y.S. Lee, and P. Newbold(2003 ) Spurious regressions with processes around linear trends or drifts . Discussion Paper in Economics, No. 03-07, University of Nottingham.
[9]	DOI: 10.1016/j.econlet.2003.10.020 · Zbl 1255.62257
[10]	DOI: 10.1016/S0165-1889(97)00021-3 · Zbl 0897.90049
[11]	DOI: 10.1162/003465397556791
[12]	Marmol F., Journal of Time Series Analysis 16 pp 313– (1995)
[13]	DOI: 10.1016/S0304-4076(97)00085-7 · Zbl 1041.62529
[14]	DOI: 10.1016/0304-3932(82)90012-5
[15]	DOI: 10.1016/S0264-9993(00)00050-X
[16]	DOI: 10.2307/1913712 · Zbl 0683.62066
[17]	DOI: 10.1016/0304-4076(86)90001-1 · Zbl 0602.62098
[18]	DOI: 10.1023/A:1009726004023 · Zbl 0927.91020
[19]	Stock J. H., Handbook of Econometrics pp 2740– (1994)
[20]	W.J. Tsay, and C.F. Chung(1999 ) The spurious regression of fractionally integrated processes . Institute of Economic Academia Sinica, Mimeo.