我国粮食产量的波动与预测方法选择  

On Fluctuation and Prediction Methods of Grain Output in China

作　　者：叶浩[1] 庄大昌[2] 张慧霞[2] 

机构地区：[1]广东财经大学地理与旅游学院,广东广州510320 [2]广东财经大学公共管理学院,广东广州510320

出　　处：《衡阳师范学院学报》2014年第3期89-93,共5页Journal of Hengyang Normal University

基　　金：教育部人文社会科学研究基金(10YJA790275);广东省高等院校学科建设专项资金人文社科重大攻关项目(2012ZGXM-0009)

摘　　要：运用R/S分形理论对我国及各省区1952—2012年期间粮食产量的波动规律及预测方法的选择进行了研究,结果表明:自1952年以来,我国粮食产量不断增长,但是增速逐渐变缓。粮食产量存在着明显的波动特征,正负波动基本相当。在1952—2012年期间,我国粮食产量的时间序列具有较明显的持续性,但这种持续性影响经过一段时期之后即会消失。各个省区的Hurst指数存在较大的差别,从区域分布来看,Hurst指数较高的省区主要为我国水土资源及气候条件均相对较为优越的粮食主产区。Hurst指数高的粮食产量时间序列具有自相关性,使用ARMA模型进行短期预测的精度会相对较高。Hurst指数在0.5附近的粮食产量时间序列波动基本处于随机状态,适合使用马尔科夫链模型进行短期预测。Hurst指数低的时间序列是反持久性的时间序列,具有均值回复的特征,适合使用EMD提取趋势线进行预测。The paper uses R/S fractal theory to research the fluctuation and prediction methods of grain output of all the provinces in China from 1952 to 2012. The results showed that： Since 1952, China＇s grain output continues to grow, but the growth rate slows down gradually. There was an obvious fluctuation characteristics in grain output. Positive and negative flucluation was almost equal. From 1952 to 2012, China＇s grain output time series is persistence, but this continuous effect will disappear after a sustained period. Hurst exponent of each province had a big difference. From the perspective of regional dis tribution, the provinces with higher Hurst exponent mostly were main grain production area. Its land and water resources and climatic conditions were relatively favorable. Grain output time series with higher Hurst exponent has a statistically significant autocorrelation, rising the ARMA model for short term prediction will has relatively high accuracy. Grain output time series with HursI exponent near to 0.5 basically is in random fluctuation state, Markov chain model is suitable for its short-term forecast. Time series with lower Hurst exponent has anti-persistence and can predict by EMD.

关 键 词：粮食产量 波动 预测方法 重标极差法 中国 

分 类 号：F307.11[经济管理—产业经济]