28 Pages Posted: 13 Feb 2003
The existence of daily seasonal patterns in the returns to 5 base metals traded on the London Metal Exchange (Aluminium, Copper, Zinc, Lead and Nickel) is examined, using robust methods, over the 1989-2002 period. The paper begins by examining the extent of daily seasonality in asset returns, the majority of papers on this area dealing with equities. However, there is some evidence of daily seasonality in areas other than corporate liabilities, particularly in gold. No papers have to date been published that have examined the issue in base metals. The paper then describes the operations of the London Metal Exchange, the exchange on which the metal contracts analysed here are traded. The data are cash market data on a daily frequency from January 1989 to the end of August 2002. The paper then proceeds to discuss methodological issues, pointing out the need to adjust for large sample sizes and for the distributional characteristics of the data prior to making any inferences regarding the existence or otherwise of seasonality. The roles of resampling methods, robust regressions (Least Trimmed of Squares, M-Class Estimators and Least Absolute Deviation Regression) are also discussed, as are non-parametric methods. The results indicate that daily seasonality does appear to exist in the metal markets, particularly important days being Monday and Thursday. Monday is the lowest return of the week and also negative, with Thursday being the highest (Friday being the second highest). Thus the metal market appears to show the stereotypical pattern of daily seasonality that was commonly described in the literature on the equity markets.
Keywords: Metals, Robust Regression, Resampling, Seasonality
JEL Classification: G12, C15, L79
Suggested Citation: Suggested Citation