A model selection test for an AR(1) versus and MA(1) model


Article
pp 281-284.
Related Files
asset icon
(eur_franses_ET4.pdf, 0.3MB)

(publisher's version.url.txt, 48 bytes)

This paper proposes a model selection test statistic for the choice between an AR(1) and an MA(1) model. It is a function of the first two sample autocorrelations of a time series. This establishes that it can be compared directly with a statistic given in Burke, Godfrey and Tremayne (1990). From Monte Carlo evidence it appears that the new test meets its purpose more.



Keywords


Automatically Extracted Terms
  • model
  • statistic
  • 0.10
  • 0.05
  • rejection rate
  • selection
  • test statistic
  • rejection
  • hypothesis
  • series
  • result
  • parameter
  • testing
  • test statistics
  • sample
  • qan test
  • autocorrelation
  • value
  • process
  • burke