计量经济学实验报告.docx
计量经济学实验报告中国农业总产值的影响因素分析一、背景资料我国是一个农业大国,农业是整个国民经济的基础产业,农业是否能够稳定发展关系到整个国民经济能否健康发展。本文采用计量经济学的相关分析方法,构建农业总产值及其影响因素的回归模型,借助Eviews3.1软件,分析各影响因素(农作物播种面积、农用机械总动力、有效灌溉面积、化肥施用量)农业总产值的影响大小和特征,并在此基础上,对我国的农业经济发展战略提出政策建议。(注:分析中所指的农业是指狭义的农业,即农林牧副渔中的农。)二、数据选取通过在经济学学科的学习和对生活的观察可以知道,影响农业总产值的因素有很多,如农作物播种面积、农业机械总动力、有效灌溉面积、农业化肥施用量、粮食产量、农业劳动力、受灾面积、农业政策制度、风俗习惯等等。但考虑到样本数据的可收集性和中国农业经济的实际情况,选择了农作物的播种面积、农业机械总动力、有效灌溉面积、化肥施用量作为影响农业总产值的主要变量。根据日常观察和统计研究都表明,农作物播种面积、农业机械总动力、有效灌溉面积都是会直接影响农业中产值的核心因素,取决于一个国家的资源禀赋和经济及技术实力,对研究一个国家的农业经济发展异常重要,因此这三个因素的入选都是毫无疑问的。化肥施用量是指粮食生产中施加的化肥量,农业总产值与化肥施用量之间存在此消彼长的关系,过度施肥会直接减少人均粮食产量,并影响粮食市场上价格的变动,可见农业总产值与化肥施用量息息相关。以下数据来自中国统计年鉴年份19971998199920002001农业总产值(亿元)Y13852.514241.914106.213873.614462.8农作物播种面积(千公顷)Xl153969155706156373156300155708农用机械总动力(万千瓦)X242015.645207.748996.152573.655172.1有效灌溉面积(千公顷)X351238.552295.653158.453820.354249.4化肥施用量(万吨)X43980.74083.74124.34146.44253.8200220032004200520062007200820092010201114931.514870.118138.419613.421522.324658.128044.230777.536941.141988.615463615241515355315548815214915346415626615861416067516228357929.960386.564027.968397.872522.176589.682190.487496.192780.597734.754354.954014.254478.455029.355750.556518.358471.759261.460347.761681.64339.44411.64636.64766.24927.75107.852395404.45561.75704.2四、模型设计Yt=+1X1t+2X2t+3X2t+4X4t+UtYt农业总产值(亿元)Xit农作物播种面积(千公顷)X2t农业机械总动力(万千瓦)X3t有效灌溉面积(千公顷)X4t化肥施用量(万吨)Ut随机随机项B1、B2、B3、B4、一一待估参数五、模型检验假设模型中随机误差项Ut满足古典假设,运用C)LS方法估计模型的参数得结果:DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX2X3X4Coefficient-235837.10.806506-0.5608950.92590024.96366Std.Error42692.560.3546180.3959201.4631517.434267t-Statistic-5.5240792.274297-1.4166860.6328133.357918Prob.0.00030.04620.18700.54110.0073R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodDurbin-Watsonstat0.981033Meandependentvar21468.150.973446S.D.dependentvar1491.307Akaikeinfocriterion22239972Schwarzcriterion-127.8542F-statistic1.486864Prob(F-statistic)9151.67817.7139017.94991129.30600.0000001、经济意义检验由回归估计结果可以看出:农作物播种面积、化肥有效灌溉面积、化肥施用量与农业总产值呈线性正相关,与现实经济理论相符。而农业机械总动力与农业生产总值呈线性负相关,这两点上,不符合经济意义。2、统计意义检验从估计的结果可知,可决系数R2=0.981033,F=129.306(),表明模型在整体上拟合地比较理想。系数显著性检验:给定(X=O.05,XI、X4的t的P值小于给定的显著性水平,拒绝原假设,接受备择假设,表明农作物播种面积、化肥施用量与农业总产值有显著性影响;仅有X2、X3的t的P值大于给定的显著性水平,接受原假设,表明农业机械总动力、有效灌溉面积与农业总产值影响不显著。六、计量经济学检验1、多重共线性检验模型整体上线性回归拟合较好,R2与F值较显著,而解释变量X2、X3的t检验不显著,并且X2的系数的符号与经济意义相悖,则说明该模型存在多重共线性。在Eviews中计算解释变量之间的简单相关系数,得到如下结果,也可以看出解释变量之间存在多重共线性。X1X2X4X110.595686366208X21X3X40.5956863662080.7179106164010.5802567634480.9809837973590.99456986018310.96859320165810.7179106164010.9809837973590.5802567634480.9945698601830.968593201658由回归分析得农业总产值与化肥施用量线性关系好,拟合程度强。对丫和X4进行回归分析DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCX4R-squaredCoefficient-49783.6915.11975Std.Error5941.8771.252057t-Statistic12.07593Prob.0.000()0.00000.91815IMeandependentvar21468.15AdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodDurbin-Watsonstat0.911854S.D.dependentvar2717.071Akaikeinfocriterion95972166Schwarzcriterion-138.8205F-statistic0.320444Prob(F-statistic)9151.67818.7760618.87047145.82820.()()0()()()对Y、XbX4进行回归分析DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX4R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodDurbin-WatsonstatCoefficient-183711.20.940828Std.Error25469.510.1773370.874818t-Statistic-7.2129835.30530314.20488Prob.0.00000.00020.00000.975535Meandependentvar21468.150.971457S.D.dependentvar1546.145Akaikeinfocriterion28686768Schwarzcriterion-129.7633F-statistic0.946275Prob(F-statistic)9151.67817.84338239.24440.000000对Y、X1、X2、X4进行回归分析DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX2X4R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodCoefficient-218773.81.001572-0.35440222.89205Std.Error32185.010.1704670.2180246.490246t-Statistic-6.7973835.875453-1.6255183.527147Prob.0.00000.00010.13230.00479151.67817.80864182.20550.980273Meandependentvar21468.150.974893S.D.dependentvar1450.096Akaikeinfocriterion23130575Schwarzcriterion-128.1487F-statisticDurbin-Watsonstatl.632382Prob(F-statistic)0.000000对Y、X1、X3、X4进行回归分析DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX3R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresidLoglikelihoodDurbin-WatsonstatCoefficient-186555.91.135590-0.78248415.82914Std.Error25858.270.2799410.8657183.866231t-Statistic-7.2145554.056534-0.9038554.094204Prob.0.00000.00190.38540.00180.977226Meandependentvar21468.150.971015S.D.dependentvar1558.074Akaikeinfocriterion26703535Schwarzcriterion-129.2260F-statistic1.553676Prob(F-statistic)9151.67817.7634717.95228157.33560.000000因为X2、X3对Y的影响不显著,所以舍去X2、X3,最后所得回归模型为:Yt=0+1X1t+2X4tDependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX4R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresidLoglikelihoodDurbin-WatsonstatCoefficient-183711.20.94082812.42668Std.Error25469.510.1773370.874818t-Statistic-7.2129835.30530314.20488Prob.0.00000.00020.00009151.67817.7017717.84338239.24440.0000000.975535Meandependentvar21468.150.971457S.D.dependentvar1546.145Akaikeinfocriterion28686768Schwarzcriterion-129.7633F-statistic0.946275Prob(F-statistic)Y=-183711.2+0.940828X1+12.42668X4(-7.212983)(5.305303)(14.20488)R2=0.9755352、异方差检验运用怀特检验法检验模型是否存在异方差。DependentVariable:YMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX4R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresidLoglikelihoodDurbin-WatsonstatCoefficient-183711.20.94082812.42668Std.Error25469.510.1773370.874818t-Statistic-7.2129835.30530314.20488Prob.0.00000.00020.00009151.67817.84338239.24440.0000000.975535Meandependentvar21468.150.971457S.D.dependentvar1546.145Akaikeinfocriterion28686768Schwarzcriterion-129.7633F-statistic0.946275Prob(F-statistic)计算残差平方和e1,将X1、X4、X2-2、X4-2、X2*X4做回归分析e1=resid2x5=×12x6=x22x7=x1*×4x6=×42DependentVariable:ElMethod:LeastSquaresSample:19972011Includedobservations:15VariableCXlX4X5X6X7R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresidLoglikelihoodDurbin-WatsonstatCoefficient-4.17E+0965762.96-369094.3-0.2428348.2275181.874767Std.Error3.41E+0948209.93183968.40.1689196.5823351.235531t-Statistic-1.2223081.364096-2.006291-1.4375771.2499391.517378Prob.0.25260.20570.07580.18440.24280.16350.425512Meandependentvarl912451.0.106352S.D.dependentvar2723624.Akaikeinfocriterion6.68E+13Schwarzcriterion-239.7150F-statistic1.96066IProb(F-Statistic)2881136.32.7620033.045221.3332230.332831由上表可知nR2=15*0.425512=6.38268,在=0.05显著程度下,X9=19.02,因为nR2=638268V19.02,所以接受原假设,不存在异方差。3、自相关检验(1)运用杜宾沃森检验法进行自相关检验d=0.946275在显著水平a=0.05下,查DW表得dL=O.95du=1.54因为dVdL,所以存在正自相关。(2)用杜宾两步法解决自相关问题DependentVariable:YMethod:LeastSquaresSample(adjusted):19982011Includedobservations:14afteradjustingendpointsVariableCY(-l)XlXl(-1)X4X4(-l)R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodDurbin-WatsonstatCoefficient-58868.240.8937860.2683740.03679016.70522-13.82475Std.Error43794.930.2380470.1874990.1806885.9499686.979401t-Statistic-1.3441793.7546601.4313362.807615-1.980792Prob.0.21580.00560.19020.84370.02290.08299242.06116.4552216.72910362.97770.0000000.99561IMeandependentvar22012.120.992868S.D.dependentvar780.4776Akaikeinfocriterion4873163.Schwarzcriterion-109.1865F-statistic3.052742Prob(F-statistic)genry1=y-eqO1.coefs(2)*ygenr×8=x1-eq01.coefs*×1genr×9=x4-eq01.coefs*x4Isy1c×8x9DependentVariable:YlMethod:LeastSquaresSample:19972011Includedobservations:15VariableCX8Coefficient-19512.700.940828Std.Error2705.2190.177337t-Statistic-7.2129835.305303Prob.0.00000.0002X9R-squaredAdjustedR-squaredS.E.ofregressionSumsquaredresid1.oglikelihoodDurbin-Watsonstat12.426680.87481814.204880.0000972.036213.2171713.35878239.24440.975535Meandependentvar2280.2170.971457S.D.dependentvar164.2222Akaikeinfocriterion323627.2Schwarzcriterion-96.1288IF-statistic1.206275Prob(F-statistic)B0=-19512.70/(1+58868.24)=-0.331458331=0.9408282=12.42668