Working Papers, Department of Economics, Lund University
Fredrik N. G. Andersson
Abstract: Economic theory commonly distinguishes between different
time horizons such as the short run and the long run, each with its own
relationships and its own dynamics. Engle (1974) proposed a bandspectrum
regression to estimate such models. This paper proposes a new estimator for
non-stationary panel data models, a bandspectrum cointegration estimator.
The bandspectrum cointegration estimator uses first differenced data to
avoid spurious results. Such estimates are, however, less efficient than
estimates from a model with non-stationary data. Still, simulation results
in the paper show that the bandspectrum cointegration estimator is more
efficient than common time domain estimators, for example VECM and OLS
levels estimators, if the data generating process contains more than one
time horizon. The BSCE furthermore identifies all horizons in the data
generating process and estimates an individual parameter vector for each, a
property that neither time domain estimator possesses.
Keywords: Cointegration; Bandspectrum Regression; Simulations; Wavelets; Frequency domain; (follow links to similar papers)
JEL-Codes: C14; C15; C23; (follow links to similar papers)
34 pages, December 2, 2008
Before downloading any of the electronic versions below
you should read our statement on
for viewing Postscript files and the
Acrobat Reader for viewing and printing pdf files.
Full text versions of the paper:
Questions (including download problems) about the papers in this series should be directed to David Edgerton ()
Report other problems with accessing this service to Sune Karlsson ()
or Helena Lundin ().
Design by Joachim Ekebom