搜索
计量经济学论文影响粮食产量的因素分析：易士桢班级：金融1502学号：20153035影响粮食产量的因素分析我国土地资源稀缺，人口多而粮食需求量大，因此粮食产量的稳定增长，直接影响着人民生活和社会的稳定与发展。本文严格按照计量经济分析方法,以1996-2015年中国粮食产量及其重要因素的时间序列数据为样本,对影响中国粮食生产的多种因素进行了分析。一、模型的建立以Yi=粮食产量、X1=粮食播种面积、X2=农用化肥施用量、X3=农用机械总动力、X4=农、林、牧、渔业劳动力、X5=耕地灌溉面积，设定Yi=c+β1X1i+β2X2i+β3X3i+β4X4i+β5X5i+ui理论模型。由经济规律知β1、β2、β3、β4、β5都应大于零。二、数据的收集（资料来源于中国各年统计年鉴）年份粮食产量（万吨）粮食播种面积（千公顷）农用化肥施用量（万吨）农用机械总动力（万千瓦）农、林、牧、渔业劳动力（万人）耕地灌溉面积（千公顷）199650453.51125483827.938546.93291050381.4199749417.11129123980.742015.63309551238.5199851229.51137874083.745207.73323252295.6199950838.61131614124.348996.13349353158.4200046217.51084634146.452573.63335553820.3200145263.71060804253.855172.13297454249.4200245705.81038914339.457929.93248754354.9200343069.5994104411.657929.9484.554014.2200446946.91016064636.664027.9466.154478.4200548402.21042784766.268397.8446.355029.3200649804.21049584927.772522.1435.255750.5200750160.31056385107.876589.6426.356518.3200852870.9106793523982190.4410.158471.7200953082.11089865404.487496.1373.759261.4201054647.71098765561.792780.5375.760347.7201157120.81105735704.297734.7359.561681.62012589581112055838.8102559338.962490.5201360193.81119565911.9103906.8294.863473.3201460702.61127235995.9108056.6284.664539.5201562143.91133436022.6111728.127065872.6三、模型的参数估计利用Eviews8得到结果如下：DependentVariable:YMethod:LeastSquaresDate:06/01/17Time:20:10Sample:19962015Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C-66773.8737106.01-1.7995430.0935X10.7900680.1191396.6314990.0000X21.7688438.0599230.2194620.8295X3-0.0286920.338671-0.0847200.9337X4-0.0870170.051349-1.6946140.1123X50.4777650.6637450.7198020.4835R-squared0.976250Meandependentvar51861.43AdjustedR-squared0.967768S.D.dependentvar5548.066S.E.ofregression996.0571Akaikeinfocriterion16.88881Sumsquaredresid13889816Schwarzcriterion17.18753Loglikelihood-162.8881Hannan-Quinncriter.16.94712F-statistic115.0958Durbin-Watsonstat1.811852Prob(F-statistic)0.000000由此数据看出，可决系数和修正可决系数为0.976250和0.967768，F的检验值为115.0958，明显显著，拟合效果还可以。但当a=0.05时，ta/2(n-k-1)=2.1448，说明X2与X5的t检验不显著，而且X3与X4系数的符号与经济解释相反，可能存在多重共线性。四、模型的检验（一）Ⅰ、检验多重共线性（利用相关系数矩阵法）CovarianceAnalysis:OrdinaryDate:06/01/17Time:20:27Sample:19962015Includedobservations:20CovarianceCorrelationYX1X2X3X4X5Y292419881.000000X115462398175499180.6825521.000000X23301563.614506.7537424.10.8328320.2000921.000000X31.03E+0818979416168699155.33E+080.8281200.1963000.9970801.000000X4-3920129717367126-9252927.-2.83E+082.43E+08-0.4648140.265811-0.809287-0.7873271.000000X5210258985370007.3232375.1.03E+08-49230933203715990.8614670.2840040.9769010.987377-0.6993711.000000由相关系数矩阵可以看出，有些解释变量之间的相关系数很高，证实确实存在多重共线性。Ⅱ、修正多重共线性采用逐步回归的方法，去解决多重共线性的问题。分别做Y对X1，X2，X3,X4,X5的一元回归，结果如下：变量X1X2X3X4X5系数估计值0.8810526.1433110.194031-0.1611631.032118t统计量3.9623396.3834656.267782-2.2272657.197309R20.4658770.6936100.6857820.2160520.742125修正的R20.4362040.6765880.6683260.1724990.727799F统计量15.7001340.74862378.60304.96070851.80126按照各解释变量医院回归模型的拟合优度大小进行排序：X5、X2、X3、X1、X4。以Y对X5的一元回归模型为最优基本模型，将其他解释变量引入，寻找最优回归模型。1、加入X2，重新估计方程得到回归结果为：YΛ=-12886.76+1.256017X5-1.411091X2t=（-0.662047）（1.824837）（-0.332989）R2=0.743796，F=24.67670可以发现X2的系数估计值为负，参数经济意义不合理，予以剔除。2、加入X3，重新估计方程得到回归结果为：YΛ=-52135.10+2.091869X5-0.209898X3t=（-1.356001）（2.338466）（-1.199823）R2=0.762257，F=27.25294可以发现X3的系数估计值为负，参数经济意义不合理，予以剔除。3、加入X1，重新估计方程得到回归结果为：YΛ=-64569.78+0.870047X5+0.614831X4t=（-8.359677）（12.92946）（8.480448）R2=0.950698，F=163.9052可以发现X1的系数估计值高度显著，保留X1。4、加入X4，重新估计方程得到回归结果为：YΛ=-67319.59+0.564664X5+0.812037X1-0.104856X4t=（-12.02244）（6.356567）（11.44358）(-4.100895)R2=0.985263，F=233.9970可以发现X4的系数估计值为负，参数经济意义不合理，予以剔除。综上保留X1、X5两个解释变量，最终得：DependentVariable:YMethod:LeastSquaresDate:06/01/17Time:20:55Sample:19962015Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C-64569.787723.957-8.3596770.0000X50.8700470.06729212.929460.0000X10.6148310.0725008.4804480.0000R-squared0.950698Meandependentvar51861.43AdjustedR-squared0.944897S.D.dependentvar5548.066S.E.ofregression1302.352Akaikeinfocriterion17.31921Sumsquaredresid28834031Schwarzcriterion17.46857Loglikelihood-170.1921Hannan-Quinncriter.17.34837F-statistic163.9052Durbin-Watsonstat0.842832Prob(F-statistic)0.000000YΛ=-64569.78+0.870047X5+0.614831X1t=（-8.359677）（12.92946）（8.480448）R2=0.950698，修正R2=0.944897，F=163.9052，D-W值=0.842832（二）自相关性检验（利用拉格朗日乘数检验法）结果如下：Breusch-GodfreySerialCorrelationLMTest:F-statistic3.000670Prob.F(2,15)0.0801Obs*R-squared5.715197Prob.Chi-Square(2)0.0574TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/01/17Time:21:00Sample:19962015Includedobservations:20Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.C3405.8487926.9100.4296560.6736X50.0030610.0605690.0505430.9604X1-0.0330240.073925-0.4467170.6615RESID(-1)0.5904170.2539592.3248530.0345RESID(-2)-0.0982700.280190-0.3507250.7307R-squared0.285760Meandependentvar-8.73E-12AdjustedR-squared0.095296S.D.dependentvar1231.901S.E.ofregression1171.734Akaikeinfocriterion17.18268Sumsquaredresid20594423Schwarzcriterion17.43161Loglikelihood-166.8268Hannan-Quinncriter.17.23127