1、3.2(1)用Eviews分析如下Dependent Variable: YMethod: Least SquaresDate: 12/01/14 Time: 20:25Sample: 1994 2011Included observations: 18VariableCoefficientStd. Errort-StatisticProb.X20.1354740.01279910.584540.0000X318.853489.7761811.9285120.0729C-18231.588638.216-2.1105730.0520R-squared0.985838Mean dependent
2、 var6619.191Adjusted R-squared0.983950S.D. dependent var5767.152S.E. of regression730.6306Akaike info criterion16.17670Sum squared resid8007316.Schwarz criterion16.32510Log likelihood-142.5903Hannan-Quinn criter.16.19717F-statistic522.0976Durbin-Watson stat1.173432Prob(F-statistic)0.000000由表可知模型为:Y
3、= 0.135474X2 + 18.85348X3 - 18231.58检验:可决系数是0.985838,修正的可决系数为0.983950,说明模型对样本拟合较好。 F检验,F=522.0976F(2,15)=4.77,回归方程显著。 t检验,t统计量分别为X2的系数对应t值为10.58454,大于t(15)=2.131,系数是显著的,X3的系数对应t值为1.928512,小于t(15)=2.131,说明此系数是不显著的。(2)(2)表内数据ln后重新输入数据:Dependent Variable: LNYMethod: Least SquaresDate: 10/25/15 Time: 22
4、:18Sample: 1994 2011Included observations: 18VariableCoefficientStd. Errort-StatisticProb.C-10.810901.698653-6.3643970.0000LNX21.5737840.09154717.191060.0000X30.0024380.0009362.6053210.0199R-squared0.986373Mean dependent var8.400112Adjusted R-squared0.984556S.D. dependent var0.941530S.E. of regressi
5、on0.117006Akaike info criterion-1.302176Sum squared resid0.205355Schwarz criterion-1.153780Log likelihood14.71958Hannan-Quinn criter.-1.281714F-statistic542.8930Durbin-Watson stat0.684080Prob(F-statistic)0.000000模型为 lny=-10.81090+1.573784lnx2+0.002438x3检验:经济意义为其他条件不变的情况下,工业增加值每增加一个单位百分比出口货物总和增加1.57单
6、位百分比,汇率每增加一单位百分比,出口总额增加0.0024个单位百分比。拟合优度检验,R2=0.986373 修正可决系数为0.984556,拟合很好。F检验对于H0:X2=X3=0,给定显著性水平a=0.05 F(2,15)=4.77 F=542.8930F(2,15) 显著t检验对于H0:Xj =0(j=2,3),给定显著性水平a=0.05 t(15)=2.131 当j=2时tt(15)显著,当j=3时 tt(15)显著。(3)两个模型表现出的汇率对Y的印象存在巨大差异 3.3(1)用Eviews分析如下Dependent Variable: YMethod: Least SquaresD
7、ate: 12/01/14 Time: 20:30Sample: 1 18Included observations: 18VariableCoefficientStd. Errort-StatisticProb.X0.0864500.0293632.9441860.0101T52.370315.20216710.067020.0000C-50.0163849.46026-1.0112440.3279R-squared0.951235Mean dependent var755.1222Adjusted R-squared0.944732S.D. dependent var258.7206S.E
8、. of regression60.82273Akaike info criterion11.20482Sum squared resid55491.07Schwarz criterion11.35321Log likelihood-97.84334Hannan-Quinn criter.11.22528F-statistic146.2974Durbin-Watson stat2.605783Prob(F-statistic)0.000000由表可知模型为:Y = 0.086450X + 52.37031T-50.01638检验:可决系数是0.951235,修正的可决系数为0.944732,说
9、明模型对样本拟合较好。 F检验,F=539.7364 F(2,15)=4.77,回归方程显著。 t检验,t统计量分别为2.944186,10.06702,均大于t(15)=2.131,所以这些系数都是显著的。经济意义:家庭月平均收入增加1元,家庭书刊年消费支出增加0.086450元,户主受教育年数增加1年,家庭书刊年消费支出增加52.37031元。(2)用Eviews分析如下Y与T的一元回归Dependent Variable: YMethod: Least SquaresDate: 12/01/14 Time: 22:30Sample: 1 18Included observations:
10、18VariableCoefficientStd. Errort-StatisticProb.T63.016764.54858113.854160.0000C-11.5817158.02290-0.1996060.8443R-squared0.923054Mean dependent var755.1222Adjusted R-squared0.918245S.D. dependent var258.7206S.E. of regression73.97565Akaike info criterion11.54979Sum squared resid87558.36Schwarz criter
11、ion11.64872Log likelihood-101.9481Hannan-Quinn criter.11.56343F-statistic191.9377Durbin-Watson stat2.134043Prob(F-statistic)0.000000模型:Y = 63.01676T - 11.58171X与T的一元回归Dependent Variable: XMethod: Least SquaresDate: 12/01/14 Time: 22:34Sample: 1 18Included observations: 18VariableCoefficientStd. Erro
12、rt-StatisticProb.T123.151631.841503.8676440.0014C444.5888406.17861.0945650.2899R-squared0.483182Mean dependent var1942.933Adjusted R-squared0.450881S.D. dependent var698.8325S.E. of regression517.8529Akaike info criterion15.44170Sum squared resid4290746.Schwarz criterion15.54063Log likelihood-136.97
13、53Hannan-Quinn criter.15.45534F-statistic14.95867Durbin-Watson stat1.052251Prob(F-statistic)0.001364模型:X = 123.1516T + 444.5888(3)对残差模型进行分析,用Eviews分析如下Dependent Variable: E1Method: Least SquaresDate: 12/03/14 Time: 20:39Sample: 1 18Included observations: 18VariableCoefficientStd. Errort-StatisticPro
14、b.E20.0864500.0284313.0407420.0078C3.96E-1413.880832.85E-151.0000R-squared0.366239Mean dependent var2.30E-14Adjusted R-squared0.326629S.D. dependent var71.76693S.E. of regression58.89136Akaike info criterion11.09370Sum squared resid55491.07Schwarz criterion11.19264Log likelihood-97.84334Hannan-Quinn
15、 criter.11.10735F-statistic9.246111Durbin-Watson stat2.605783Prob(F-statistic)0.007788模型:E1 = 0.086450E2 + 3.96e-14参数:斜率系数为0.086450,截距为3.96e-14(4)由上可知,2与2的系数是一样的。回归系数与被解释变量的残差系数是一样的,它们的变化规律是一致的。3.4为了分析中国税收收入(Y)与国内生产总值(X2)、财政支出(X3)、商品零售价格指数(X4)的关系,利用19782007年的数据,用EViews作回归,部分结果如下:表3 回归结果Dependent Var
16、iable: LNYMethod: Least SquaresDate: 06/30/13 Time: 19:39Sample: 1978 2007Included observations: 30VariableCoefficientStd. Errort-StatisticProb. C-2.7553670.640080(1)0.0002LNX20.451234(2)3.1748310.0038LNX30.6271330.161566(3)0.0006X4(4) 0.0056451.7955670.0842R-squared0.987591 Mean dependent var8.3413
17、76Adjusted R-squared(5) S.D. dependent var1.357225S.E. of regression(6) Akaike info criterion-0.707778Sum squared resid0.662904 Schwarz criterion-0.520952Log likelihood14.61668 F-statistic(7)Durbin-Watson stat0.616136 Prob(F-statistic)0.000000填补表中空缺数据:(1)t c=4.304723(2)=0.130789(3)=3.881590(4)=0.010
18、136(5)=0.986159(6)S.E of regression回归标准差=0.154783(7)=689.751148分析回归结果:根据图中数据,模型估计的结果写为:=-2.755367+0.451234+0.627133+0.0101361) 拟合优度:由上图数据可以得到,可决系数=0.987591,修正的可决系数=0.986159,这说明模型对样本的拟合很好。2)F检验:针对,给定显著性水平,在F分布表中查出自由度为k-1=3和n-k=26的临界值=8.63。由上图得到F=689.751148,由于F=689.751148,应拒绝原假设,说明回归方程显著,即国内生产总值、财政支出、商品零售价格指数等变量联合起来对中国税收收入有显著影响。3)t检验:由上图中的P值可以判断,在的显著性水平下,与、估计值对应的P值小于,表明对应解释变量对被解释变量影响显著。在的显著性水平下,与估计值对应的P值小于,表明对应解释变量对被解释变量影响显著。评估参数的经济意义:当其他变量不改变时,国内生产总值每增加1%,中国税收收入增加0.451234%。当其他变量不改变时,财政支出每增加1%,中国税收收入增加0.627133%。当其他变量不改变时,商品零售价格指数每增加1%,中国税收收入增加0.010136%。