Post by Sapphire Capital on Jul 11, 2008 6:15:14 GMT 4
Modeling the Volatility of the London Gold Market Fixing as an Asymmetric Power ARCH Process
GIORGIO CANARELLA
California State University, Los Angeles - Department of Economics & Statistics
STEPHEN K. POLLARD
California State University, Los Angeles
--------------------------------------------------------------------------------
The Icfai Journal of Applied Finance, Vol. 14, No. 5, pp. 17-43, May 2008
Abstract:
This paper considers the applicability of the asymmetric power ARCH (APARCH) model of Ding et al. (1993) to the London Gold Market Fixing. The study investigates the long memory features and conditional volatility behavior of daytime and overnight returns under three alternative conditional error distribution assumptions normal distribution, t distribution, and Generalized Error Distribution (GED). The findings provide strong evidence that (a) the APARCH models estimated using leptokurtic distributions are superior to their counterparts estimated under normality and outperform the parsimonious GARCH and Taylor/Schwert models; (b) the optimal power transformation is remarkably similar across the two returns, irrespective of the conditional error distribution assumptions; (c) the APARCH models adequately capture the long memory property of the returns; and (d) shocks have strong symmetric effects on volatility; however, unlike the results reported in the stock market literature, conditional volatility in the London Gold Market Fixing is impacted more by positive shocks ("good news") than negative shocks ("bad news").
papers.ssrn.com/sol3/papers.cfm?abstract_id=1126142
GIORGIO CANARELLA
California State University, Los Angeles - Department of Economics & Statistics
STEPHEN K. POLLARD
California State University, Los Angeles
--------------------------------------------------------------------------------
The Icfai Journal of Applied Finance, Vol. 14, No. 5, pp. 17-43, May 2008
Abstract:
This paper considers the applicability of the asymmetric power ARCH (APARCH) model of Ding et al. (1993) to the London Gold Market Fixing. The study investigates the long memory features and conditional volatility behavior of daytime and overnight returns under three alternative conditional error distribution assumptions normal distribution, t distribution, and Generalized Error Distribution (GED). The findings provide strong evidence that (a) the APARCH models estimated using leptokurtic distributions are superior to their counterparts estimated under normality and outperform the parsimonious GARCH and Taylor/Schwert models; (b) the optimal power transformation is remarkably similar across the two returns, irrespective of the conditional error distribution assumptions; (c) the APARCH models adequately capture the long memory property of the returns; and (d) shocks have strong symmetric effects on volatility; however, unlike the results reported in the stock market literature, conditional volatility in the London Gold Market Fixing is impacted more by positive shocks ("good news") than negative shocks ("bad news").
papers.ssrn.com/sol3/papers.cfm?abstract_id=1126142