SSE/EFI Working Paper Series in Economics and Finance
No 656:
A Note on the Accuracy of Markov-Chain Approximations to Highly Persistent AR(1)-Processes
Martin Floden ()
Abstract: This note examines the accuracy of methods that are
commonly used to approximate AR(1)-processes with discrete Markov chains.
The quadrature-based method suggested by Tauchen and Hussey (1991)
generates excellent approximations with a small number of nodes when the
autocorrelation is low or modest. This method however has problems when the
autocorrelation is high, as it typically is found to be in recent empirical
studies of income processes. I suggest an alternative weighting function
for the Tauchen-Hussey method, and I also note that the older method
suggested by Tauchen (1986) is relatively robust to high
autocorrelation.
Keywords: numerical methods; income processes; autoregressive process; (follow links to similar papers)
JEL-Codes: C60; (follow links to similar papers)
9 pages, March 12, 2007
Before downloading any of the electronic versions below
you should read our statement on
copyright.
Download GhostScript
for viewing Postscript files and the
Acrobat Reader for viewing and printing pdf files.
Full text versions of the paper:
hastef0656.pdf 
(114kB)
Files with additional material for the paper:
hastef0656.zip Matlab code, PKZip archive (106kB)
Download Statistics
Questions (including download problems) about the papers in this series should be directed to Helena Lundin ()
Report other problems with accessing this service to Sune Karlsson ()
or Helena Lundin ().
Programing by
Design by Joachim Ekebom
