Can futures price be a powerful predictor? Frequency domain analysis on Chinese commodity market

a r t i c l e i n f o JEL classification: C13 C32 G14 Keywords: Futures price Spot price Chinese commodity market Frequency domain approach Garbade–Silber Model This paper presents the causal relationships between futures and spot prices of six metal and agriculture commodities in Chinese commodity market, using GC test, frequency domain approach proposed by Brietung and Candelon (2006) and Garbade–Silber (G–S) Model. Frequency domain approach indicates that futures price of each commodity is really a powerful predictor for spot price in both long and short terms, but not vice versa. From the results of G–S model, futures price of each commodity decides more than 70% of the price movements, which plays a dominant role in price discovering process. There are bi-directional casual relationships between futures and spot prices of all the six commodities excluding aluminum (Al) from the conclusions of time domain GC test. Futures markets serve several functions, among which price discovery function is usually regarded as the leading indicator of judging the efficiency of a futures market. The existence of causal relationship between futures and spot prices is usually used as a best description of price discovery function in empirical studies. If close causality exists , either price can provide signals for the other in price movements to avoid risks. Obviously, it is very important for market participants to know whether there is a bidirectional or unidirectional causal relationship between the two prices, for financial risk management is really vital in arbitraging and hedging. Moreover, the causal relationship can be useful to judge if the market has a good information transformation and if price movements adjust to the information volatility well. Numerous insights have yielded about our topic, but they generally ignore the possibility that the strengths and/or directions of the causal relationships could vary over different frequencies. If causality exists, when can futures price be a powerful predictor? When in reverse? When can't? These questions are rarely fully studied but rather practical in investing. The frequency domain approach gives a complete inter-frequency characterization of causality, instead of a one-shot measure which is supposed to apply across all periods only with one result. This paper will first apply frequency domain approach (or called " a spectral-density approach ") to study the causality between futures and spot prices, which is never used before in previous papers; furthermore, we have paid great attention to the booming …