| The finance market has been excessively volatile in 2015. Not only the stock market, but also the real estate market and the futures market are immersed in great fluctuation. Some investors have gained a lot while others suffered from the fluctuation. Since corns are treated as food crop, feed crop and energy resources, the corn futures has become the one of the most active agricultural futures. The price of corn futures is sensitive to corn farmers, corn factories and clean energy industry. Under these circumstances, the study of corn volatility is practical and realistic meaningfully.This paper is based on the corn futures which are traded in DEC from 2011.1.1 to 2015.12.31. The study firstly examines the corn futures market efficiency, then the characteristics of corn futures and the corn price forecasting. The empirical results are as follow:First, according to the market efficiency hypothesis, the paper analyzes the market efficiency of Chinese corn futures with variance ratio test, unit root test and co-integration test. The results show that Chinese corn futures market has met the weak-form efficient market, but there are still room for improvement. Thus it can use the historical data to forecast the future price of corn futures.Second, the empirical study of volatility characteristics shows that the volatility of both near-futures and main-futures contracts have clustering effect and volatilities are not uniform distributed. It means the volatilities of corn futures are time-varying and big wave after big wave, small wave after small wave. Volatilities are mutually affected.But in ARCH effect test, there are differences between near-month contract and the main contract. The ARCH test shows the main contract has ARCH effect while near-month contract don’t. In leverage effect test, the difference is more obvious in near-month contract and main contract. TARCH test shows that the volatility of main contract has significant leverage effect and bad news has greater shock impact to volatility of main contract than the good news, while the volatility of near-month contract doesn’t approve the leverage effect in 5% significance level. However, in 10% significance level the empirical result indicates near-month contract has leverage effect and good news has bigger shock to volatility than bad news. But EGARCH test shows totally opposite results. Near-month contract shows obvious leverage effect in EGARCH test and bad news gives greater shock to volatility. The main contract doesn’t show any leverage effect in EGARCH test at 5% or 10% significance level.Third, the GARCH-type models are better model in forecasting corn futures price after considering the volatility as an independent variable. As for near-month contract, GARCH-σ and GARCH-GED models are more accurate in inside-sample fitting while EGARCH are more accurate in out-of-sample forecasting. As for main contract, ARIMA and GARCH-t models have more accuracy in inside-sample fitting and the three kinds of GARCH-M models are better than other models in out-of-sample forecasting.Compared with near-month contract, the main contract is more predictive and of higher accuracy. For inside-sample fitting, the evaluation percentage of main contract is about 96% while only 94% to near-month contract. Besides, the AIC and AC are both lower than near-month contract. For out-of-sample forecasting, all the lose function in main contract are smaller than near-month contract. Therefore, it indicates that the models are better to main contract than near-month contract. |
PDF Full Text Request