7

How do I test if two time series' coefficients differ significantly from one another? I feel like this should be pretty simple... should I just use the estimates/standard errors and calculate a Z-score?

Your help is much appreciated.

Pseudo_Scientist
  • 504
  • 3
  • 11

1 Answers1

1

The Chow Test can be used to test the equivalence of two (or more ) time series models.

IrishStat
  • 27,906
  • 5
  • 29
  • 55
  • Awesome - this looks like exactly what I need. Do you have any recommendations for packages to use in R? – Pseudo_Scientist Feb 12 '13 at 22:03
  • 1
    Actually this does not appear to be what I am looking for - is the Chow Test only used for looking for structural breaks in time series models? That is all I can seem to find references on. I am wanting to compare the time series of two separate groups, starting from the same time point. – Pseudo_Scientist Feb 12 '13 at 22:08
  • The Choe test can be used to test for structural breals IF you pre-specify the break point. AUTOBOX , a piece of software that I have helpewd write actually searches for the breakpoint. The null hypothesis you are specifying is that all the coeffiecients are equal . Thid id the null hypothesis of the Chow Test thus it is exactly what you are lookin for at least in my opinion. – IrishStat Aug 12 '13 at 12:15