Causality in the aluminum market

Wavelets based on undifferenced cash and futures log are used to determine causality in the aluminum markets. Wavelets are used versus I(d) processes as daily I(d)s adjusted for conditional heteroskedasticity, are not second-order stationary. Weekly I(d)s are second-order stationary after accounting for conditional heteroskedasticity but are found to be less informative than weekly wavelets in terms of causality. Wavelets exhibit causality at most daily time scales up to 2 years and weekly time scales up to 5 years. As causality exists between cash and futures in both the short and long term, the aluminum market meets the first condition of market efficiency. The daily cost-of-carry-model is supported, and as causality is determined without reference to I(d)s, there is increased confidence in the setting of initial hedge ratios. The wavelets exhibit causality at most daily time scales up to 2 years and weekly time scales out to 5 years. As causality exists between cash and futures in both the short and long term, the aluminum market meets the first condition of market efficiency. The results give support to the cost-of-carry-model on a daily basis and increases confidence in setting of initial hedge ratios 1 year or more out.