The Empirical Research On Efficiency Of China's Commodity Futures Markets

China's futures markets have been greatly developed after experiencing two periods of stringent adjustment. Especially with the greatly increasing of futures transaction volumes and the new kinds of futures products listed continually in China, the futures market is going into a period of rapid development and prosperity, and it is urgently necessary to research on the efficiency of the futures markets.The price behavior and efficiency of commodity futures markets have been extensively investigated in Western countries, whereas the relatively fewer studies have systematically investigated the price behavior and efficiency of futures markets in China. Firstly, this study gives the definition of the efficiency of futures markets and considers that the information efficiency of futures markets should be researched and tested from two different aspects which are the test of the random walk hypothesis for futures prices series and research on the information transmission mechanisms between futures prices series and contemporaneous spot prices series. Starting with the research on volatility and liquidity of futures markets, this study investigates the information efficiency of China's main futures markets from above-mentioned two different aspects. The main empirical results are summarized as follows:Through investigating the volatility and liquidity of futures markets, we can fairly understand the whole situation of China's futures markets during the entire sample period. In the research on the volatility of futures markets, unconditional volatility estimation is adopted to measure the general volatility of China's futures markets during the entire sample period from January 4, 1999 to January 4, 2006. According to the results of unconditional volatility estimation, the entire sample is divided into two sub-samples and then GARCH and EGARCH model are applied to the two sub-samples respectively for investigating the volatility clustering and unsymmetry effects and it is found that there exists evident volatility clustering and unsymmetry effects in futures markets, moreover, the volatility unsymmetry effects are opposite in the two sub-samples. In the study on the liquidity of futures markets, the liquidity ratio index L is adopted to examine the liquidity of futures markets during the entire sample period from January 4, 1999 to January 4, 2006. The non-parameter test is applied to compare with the difference in liquidity of the same futures market in different year, and it is found there is increase trend in liquidity of every futures markets. In addition, according to the stage character of the volatility of futures markets, the change in liquidity of futures markets during the two sub-samples is analyzed and it is found that the liquidity of futures markets in the second sub-sample had increased significantly relative to that in the first sub-sample.In the tests of the random walk hypothesis (RWH) for futures prices, the continuous futures price series for soybean, wheat, copper and aluminum during the entire sample period from January 4, 1999 to January 4, 2006 are selected and then the entire sample is also divided into two sub-samples according to the research on volatility of futures markets, and variance ratio test combining with unit root test is adopted to the two sub-sample for the random walk hypothesis tests. Firstly, unit root test shows all the futures price series for soybean, wheat, copper and aluminum are nonstationary I(1) process which means that each futures price series has the characteristic of stochastic trend components. Then variance ratio test which aims to examine the autocorrelation of random walk's increments gives the results, and the results obtained from the heteroscedastic variance ratio test statistic are selected. The results show that the random walk hypothesis consistently"cannot be rejected"by all tests for copper futures in the two sub-samples; there are some changes in market efficiency during the two different sub-sample periods for the other futures, that is, for soybean and aluminum futures, the results that the random walk hypothesis"cannot be rejected"in the first sub-sample have changed into the results that the random walk hypothesis are"slightly rejected"in the second sub-sample; but for wheat futures, the result that the random walk hypothesis is"rejected"in the first sub-sample has changed into the result that the random walk hypothesis"cannot be rejected"in the second sub-sample. These results mean that relative to the first sub-sample, the market efficiency of wheat futures improves to a certain degree during the second sub-sample and the market efficiency of soybean and aluminum futures decreases to some degree during the second sub-sample.In the research on the information transmission efficiency between futures prices and contemporaneous spot prices, the continuous futures price series for soybean, wheat, copper and aluminum and the corresponding spot price series during the entire sample period from January 4, 1999 to January 4, 2006 are selected and at the same time, VAR model, Johansen cointegration test, Vector error correction model (VECM), Granger causality test and Hasbrouck common factor model etc are adopted to investigate the dominant-satellite (lead-lag) relationship and information transmission mechanisms between futures prices and contemporaneous spot prices. Firstly, Johansen cointegration test shows that there are long-term equilibrium relationships between the futures prices and contemporaneous spot prices for soybean, wheat, copper and aluminum products, and the VECM and the Granger causality test show that, on the whole, the futures prices of each products can lead their respective contemporaneous spot prices rather than the spot prices lead the futures prices. On this basis, Hasbrouck common factor model is adopted to estimate the information share of each futures market and their corresponding spot market, and the results show that for soybean, wheat, copper and aluminum products, the average information share of the futures markets are 84.9%, 78.5%, 57.7% and 75.9% respectively and the average information share of the spot markets are 15.1%, 21.5%, 42.3% and 24.1% respectively, obviously, the average information share of each futures market are all larger than that of spot market. In addition, Impulse response analysis shows that each futures market affects their spot market more greatly than the spot market affects their futures market. These results all further prove the above results obtained from VECM and Granger causality test, that is each futures market can incorporating or react to new information more rapidly relative to their spot market, thereby futures markets exhibit dominance over the contemporaneous price discovery and information transmission between futures markets and their corresponding spot markets, which also reflects the information efficiency of soybean, wheat, copper and aluminum futures markets from the other different aspect.