93计量经济学论文

中国粮食总产量多因素分析专业年级：13金融（2）班学号：201312030140姓名：谢昊摘要：本文选取1990年到2013年的相关数据，应用计量经济学所学知识对根据经济理论选取的影响我国粮食产量的各因素进行分析、检验，并对其影响程度的大小进行定量分析，进一步明确和完善相关的经济学知识。关键词：粮食产量粮食播种面积农用机械总动力有效灌溉面积农业化肥使用量一、文献综述农业作为我国最基础的产业，农产品的每年的产量直接关系着我们的民生，故而粮食的产量一直是我们最关心的。影响因素的分析首先，粮食作为农作物，其产量肯定会受到农用化肥施用量条件的影响其次，我认为粮食的播种面积对于粮食产量也有一些影响最后，农业机械总动力也是影响粮食产量的一大重要因素二、数据收集与模型的建立（一）数据收集1983年—2009年中国粮食生产与相关投入的资料（表1）年份粮食总产量Y粮食耕种面积(x1)农用化肥施用量(x2)农业机械总动力(x3)1990446241134662590.3287081991435291123142805.1293891992442641105602930.2303081993456491105093151.9318171994445101095443317.9338021995466621100603593.7361181996504541125483827.9385471997494171129123980.7420161998512301137874083.7452081999508391131614124.3489962000462181084634146.4525742001452641060804253.8551722002457061038914339.457930200343070994104411.6603872004469471016064636.6640282005484021042784766.2683982006498041049584927.7725222007501601056385107.8765902008528711067935239821902009530821089865404.4874962010546481098765561.7927802011571211105735704.2977352012589581112055838.81025602013601941119565911.9103907（二）模型设计为了具体分析各要素对我国粮食产量影响的大小，我们可以用粮食总产量（y）作为衡量，代表粮食发展；用粮食耕种面积（x1）、农用化肥施用量(x2)以及农业机械总动力（x3）。运用这些数据进行回归分析。采用的模型如下：y=β1+β2x1+β3x2+β4x3+ui其中，y代表粮食总产量，x1代表粮食耕种面积，x2代表农用化肥施用量，x3代表农业机械总动力，ui代表随机扰动项。我们通过对该模型的回归分析，得出各个变量与我国粮食产量的变动关系。三、模型估计和检验（一）模型初始估计（表二）DependentVariable:YMethod:LeastSquaresDate:12/21/15Time:16:27Sample:19902013Includedobservations:24VariableCoefficientStd.Errort-StatisticProb.C-44644.146601.867-6.7623500.0000X10.6841160.05311312.880430.0000X24.0429710.9747514.1476970.0005X30.0310320.0383520.8091310.4280R-squared0.966281Meandependentvar49317.62AdjustedR-squared0.961223S.D.dependentvar4867.060S.E.ofregression958.4155Akaikeinfocriterion16.71945Sumsquaredresid18371206Schwarzcriterion16.91579Loglikelihood-196.6334F-statistic191.0450Durbin-Watsonstat1.534928Prob(F-statistic)0.000000回归函数为：12344644.140.684116X4.0429710.031032iYXX（6601.867）（0.053113）（0.974751）（0.038352）T=（-6.762350）（12.88043）（4.147697）（0.809131）20.966281R20.961223RF=191.0450（二）多重共线性检验相关系数矩阵（表三）X1X2X3X11-0.267566314901-0.23239867238X2-0.26756631490110.977074961235X3-0.232398672380.9770749612351根据多重共线性检验，解释变量之间可能存在着线性相关。为了进一步了解多重共线性的性质，我们可以做辅助回归。（表四）被解释变量可决系数R2的值方差扩大因子X10.090191.09913X20.95640922.9405X30.95558322.6398由上表可以得知，辅助回归的可决系数很高，经验表明，方差扩大因子jVIF=10时，通常说明该解释变量与其余解释变量之间有严重的多重共线性，这里的x2、x3的方差扩大因子远大于10，表明存在严重的多重共线性问题。为了进一步筛选并剔除引起多重共线性分变量，需要采用逐步回归的方法。分别作Y对X1、X2、X3的一元回归，意愿回归结果如下表（表五）变量X1X2X3参数估计值0.3696284.0710710.162556t统计量1.4728006.7542466.867695R20.0897480.6746520.6819212R0.0483730.6598630.667463(表六)X1X2X32RX1、X30.641034（9.246298）0.186325（16.84505）0.937277X2、X31.587586(0.558181)0.100949（0.893659）0.686571通过采用剔除变量法，多重共线性的修正结果如下：剔除X2。（表七）DependentVariable:YMethod:LeastSquaresDate:12/25/15Time:10:06Sample:19902013Includedobservations:24VariableCoefficientStd.Errort-StatisticProb.C-31636.647732.436-4.0914190.0005X10.6410340.0693299.2462980.0000X30.1863250.01106116.845050.0000R-squared0.937277Meandependentvar49317.62AdjustedR-squared0.931303S.D.dependentvar4867.060S.E.ofregression1275.661Akaikeinfocriterion17.25679Sumsquaredresid34173555Schwarzcriterion17.40404Loglikelihood-204.0814F-statistic156.9019Durbin-Watsonstat1.001388Prob(F-statistic)0.000000修正后方程为1231636.640.6410340.186325iYXX(7732.436)(0.069329)(0.011061)T=(-4.091419)(9.246298)(16.84505)20.937277R20.931303R156.9019F（三）异方差检验（表八）ARCHTest:F-statistic0.037667Probability0.847978Obs*R-squared0.041181Probability0.839189TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:12/24/15Time:18:58Sample(adjusted):19912013Includedobservations:23afteradjustingendpointsVariableCoefficientStd.Errort-StatisticProb.C1280357.504218.42.5392910.0191RESID^2(-1)0.0415310.2139870.1940810.8480R-squared0.001790Meandependentvar1341173.AdjustedR-squared-0.045743S.D.dependentvar1852594.S.E.ofregression1894492.Akaikeinfocriterion31.82974Sumsquaredresid7.54E+13Schwarzcriterion31.92848Loglikelihood-364.0420F-statistic0.037667Durbin-Watsonstat1.986528Prob(F-statistic)0.847978由上表可以得知，（n-p）2R=0.041181，给定显著性水平为0.05，查2分布表得临界值2（p）=5.9915（n-p）2R,则接受原假设，表明模型中的随机误差项不存在异方差。（四）自相关检验（表九）DependentVariable:YMethod:LeastSquaresDate:12/25/15Time:10:06Sample:19902013Includedobservations:24VariableCoefficientStd.Errort-StatisticProb.C-31636.647732.436-4.0914190.0005X10.6410340.0693299.2462980.0000X30.1863250.01106116.845050.0000R-squared0.937277Meandependentvar49317.62AdjustedR-squared0.931303S.D.dependentvar4867.060S.E.ofregression1275.661Akaikeinfocriterion17.25679Sumsquaredresid34173555Schwarzcriterion17.40404Loglikelihood-204.0814F-statistic156.9019Durbin-Watsonstat1.001388Prob(F-statistic)0.0000001231636.640.6410340.186325iYXX(7732.436)(0.069329)(0.011061)T=(-4.091419)(9.246298)(16.84505)20.937277R20.931303R156.9019F查DW表可知，dl=1.188，du=1.546,模型中DWdl，显然有自相关。（表十）残差的变动有系统模式，连续为正和连续为负，表明残差项存在一阶自相关。对模型进行BG检验，用Eviews分析结果如下：（表十一）Breusch-GodfreySerialCorrelationLMTest:F-statistic2.642994Probability0.097113Obs*R-squared5.223742Probability0.073397TestEquation:DependentVariable:RESIDMethod:LeastS