Approximating long-memory processes with low-order autoregressions: Implications for modeling realized volatility

Richard T. Baillie, Dooyeon Cho, Seunghwa Rho

Research output: Chapter in Book/Report/Conference proceedingChapterpeer-review

Abstract

Several articles have attempted to approximate long-memory, fractionally integrated time series by fitting a low-order autoregressive AR(p) model and making subsequent inference. We show that for realistic ranges of the long-memory parameter, the OLS estimates of an AR(p) model will have non-standard rates of convergence to non-standard distributions. This gives rise to very poorly estimated AR parameters and impulse response functions. We consider the implications of this in some AR type models used to represent realized volatility (RV) in financial markets.

Original languageEnglish
Title of host publicationAdvanced Studies in Theoretical and Applied Econometrics
PublisherSpringer Science and Business Media Deutschland GmbH
Pages455-481
Number of pages27
DOIs
StatePublished - 2024

Publication series

NameAdvanced Studies in Theoretical and Applied Econometrics
Volume55
ISSN (Print)1570-5811
ISSN (Electronic)2214-7977

Keywords

  • ARFIMA
  • HAR models
  • Long-memory
  • Realized volatility

Fingerprint

Dive into the research topics of 'Approximating long-memory processes with low-order autoregressions: Implications for modeling realized volatility'. Together they form a unique fingerprint.

Cite this