1、计量经济学报告 固定资产投资的计量经济学模型 一、解释模型 固定资产对一个企业来说是其主要的劳动手段,它的价值是逐渐地转移到所生产的产品上去.企业同时又是重要的市场主体,因此对固定资产的投资间接得影响到了一个经济体的产出.这里主要对gdp及国有经济固定资产投资额(x1),集体经济固定资产投资额(x2),个体经济固定资产投资额(x3),进行计量经济学多元线性回归模型分析. 原始数据如下:单位(亿元) obsgdpx1x2 19804517.8745.946 19814860.3667.5115.2 19825301.8845.3174.3 19835957.4952156.3 19847206.
2、71185.2238.7 19858989.11680.5327.5 19862023201.4220239.4391.8 198711954.42448.8547 198814922.3302202311.7 198916917.8282023.2570 199018598.42986.3529.5 199121662.53713.8697.8 199226651.95498.71359.4 199334560.57925.92317.3 1994466720236152758.9 199557494.91202398.23289.4 199666850.512023.23651.5 199
3、773142.71320231.73850.9 199876967.215369.34192. 2由以上数据得到如下ls估计结果, dependentvariable:gdp method:leastsquares date:12/30/2023time:2023:52 sample:19801998 includedobservations:19variablecoefficientstd.errort-statisticprob. c632.0385787.85220.8022300.4349 x2023.420234461.202383520.3718720.7152 x26.99351
4、22.9834202.3441260.0333 x311.194781.8313866.1127360.0000r-squared0.996478meandependentvar27022.51x3119178.322023.8321.842023535.2649.4759.9202322.1202332.2202301.21182.912221476.21970.62560.23211.23429.43744.4 adjustedr-squareds.e.ofregressionsumsquaredresidloglikelihooddurbin-watsonstat 0.995774s.d
5、.dependentvar1612.032akaikeinfocriterion38979701schwarzcriterion-165.0339f-statistic1.219467prob(f-statistic) 24797.6217.7930417.991871414.7900.000000 显然x1的t检验为非显著性检验,故将x1与x2合并为一个解释变量。也就是将国有经济与集体经济固定资产投资额的和看作为公有经济固定资产投资额(x1+x2).令x1+x2=x1得到如下检验结果: dependentvariable:gdpmethod:leastsquaresdate:12/30/20
6、23time:2023:53sample:19801998includedobservations:19 variablecx1+x2x3 r-squaredadjustedr-squareds.e.ofregressionsumsquaredresidloglikelihooddurbin-watsonstat coefficient-200.2023932.133202392023.14031 std.error633.13990.3291901.802497 t-statistic-0.3160116.479820235.625704 prob.0.7562023.00000.00002
7、7022.5124797.6217.8523718.001491918.9160.000000 0.995848meandependentvar0.995329s.d.dependentvar1694.728akaikeinfocriterion45953627schwarzcriterion-166.5975f-statistic1.13802023prob(f-statistic) ,从而得到多元线性回归方程:gdp=-200.202393+2.13320239x1+2023.14031x3 二、模型检验1.统计学检验 tstatistic检验,显著水平0.05,其临界值为t/2=2.11
8、,显然6.472744及5.625704远远大于它,其解释变量的prob均为0.0000,即从统计学检验的角度上讲解释变量的选取是有意义的. fstatistic检验及拟合优度r-squared检验,r-squared值越接近于1,那么f值越大,这里的r-squared值为0.995329,大于0.9拟合优度比较高,因此fstatistic检验亦通过.2.计量经济学检验 a.异方差性检验: whiteheteroskedasticitytest:f-statisticobsxr-squared testequation: dependentvariable:resid2method:least
9、squaresdate:12/30/2023time:2023:55sample:19801998includedobservations:19 variablecx1+x2(x1+x2)2(x1+x2)xx3 x3x32 r-squaredadjustedr-squareds.e.ofregressionsumsquaredresidloglikelihooddurbin-watsonstat coefficient247520237.-505.6392023.444067-4.429261-289.824412.02527 std.error1758485.1727.662023.5569
10、20235.849492023934.68616.41130 0.553341probability3.33420236probability t-statistic1.42023517-0.2926730.797376-0.757205-0.0324380.732744 0.7336340.648629 prob.0.18270.77440.43960.46240.97460.47672418612.2725196.32.8585433.156780.553342023.73363 40.175478meandependentvar-0.141646s.d.dependentvar2911813.akaikeinfocriterion1.2023e+14schwarzcriterion-306.1561f-statistic1.938280prob(f-statistic) 由表中数据可知没有哪个参数的t检验是显著的,且可决系数的值也比较小。nr2=3.33462023 第6页 共6页