Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.638162
Title: Investigations of volatility in intra-day UK futures market data
Author: McMillan, D. C.
Awarding Body: University of Wales Swansea
Current Institution: Swansea University
Date of Award: 1998
Availability of Full Text:
Access from EThOS:
Abstract:
The aim of this study is to provide a systematic investigation of volatility in high-frequency intra-day financial futures data. There are several motivations for such a study. First, to examine whether the empirical results achieved with daily and lower frequency data, for example, the high persistence of volatility to shocks, continue to hold with intra-day data. Second, to examine hypotheses concerning the intra-day patterns in volatility and volume, the cause of ARCH volatility clustering in financial data, and the temporal aggregation properties of intra-day data. Third, to examine some of the issues raised by other researchers currently examining such data, particularly those pertaining to the degree of non-linearity within the data, whether a model from the GARCH class can adequately account for this non-linearity and whether that result is frequency dependent. In sum, graphical and simple regression evidence suggests a 'U'-shape intra-day pattern in both volatility and volume, interspersed by spikes resulting from the release of macroeconomic news. Non-linear dependence is reported in conditional variance, this taking GARCH form with high persistence to shocks. Evidence of an asymmetric response to these shocks is, however, limited. The model orders and coefficient estimates of temporally aggregated data suggest some support for established theoretical results. Residual diagnostics for GARCH models become insignificant at the one hour and lower frequencies, but indicate remaining non-linear structure in higher frequencies and five-minute returns in particular. The inclusion of volume in the GARCH equation suggests support for the information flow hypothesis as the cause of volatility clustering, but remaining significant GARCH parameters suggest that it is not the full explanation. Non-linear dependence in conditional mean at higher frequencies is also reported, and models of smooth transition threshold behaviour prove able to account for some but not all of this structure.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.638162  DOI: Not available
Share: