School of Economics and Finance

No. 408: Bivariate FIGARCH and Fractional Cointegration

Celso Brunetti , University of Edinburgh
Christopher L. Gilbert , Queen Mary and Westfield College, University of London

December 1, 1999

Download full paper


We consider the modelling of volatility on closely related markets. Univariate fractional volatility (FIGARCH) models are now standard, as are multivariate GARCH models. In this paper we adopt a combination of the two methodologies. There is as yet little consensus on the methodology for testing for fractional cointegration. The contribution of this paper is to demonstrate the feasibility of estimating and testing cointegrated bivariate FIGARCH models. We apply these methods to volatility on the NYMEX and IPE crude oil markets. We find a common order of fractional integration for the two volatility processes and confirm that they are fractionally cointegrated. An estimated error correction FIGARCH model indicates that the preponderant adjustment is of the IPE towards NYMEX.

J.E.L classification codes: G0, C2, C3

Keywords:FIGARCH, Fractional Cointegration, ECM