还剩4页未读，继续阅读
文本内容:
第八章思考与练习答案
一、简述题略
二、单项选择题
1.D
2.C
3.D
4.A
5.B
6.A
7.D
8.D
9.B
10.A
11.A
12.C
13.B
14.D
15.D
1.AB
2.CD
3.ABCD
4.AC
5.ABCD
6.ABC
7.ABCD
8.ABC
9.ABC
10.AC
11.BCD
12.ABCD
13.ABCD
14.AC
15.AB
三、多项选择题
1.V
2.x
3.V4,x
5.x
四、判断题
11.x
12.V
13.x
14.V
15.x
8.V
9.V
10.x
五、填空题,加法截距.乘法
45.
12.
33.12虚拟变量交互.斜率
9.
10.
67.K-
18.
314.虚拟变量
15.完全多重共线性
11.
012.m-
113.
21.以第四季度为基础性变量，设置3个虚拟变量,如下所示:第一季度第二季度，第三季度1fl0其他季度2=jo其他季度3[0其他季度
六、计算题试建立城镇居民人均消费支出（）为被解释变量，人均可支配收入（）、虚拟变量Y X为解释变量的计量模型，设定模型形式为DI,D2,D3工=济%,.+分凤％+肛J4+A+模型估计LS YC XDI D2D3虚拟变量模型的估计结果如所示8-1Dependent Variable:Y Method:Least SquaresDate:07/19/20Time:08:02Sample:2013Q12019Q4Includedobservations:28Variable CoefficientStd.Error t-Statistic Prob.C
1263.
492133.
26389.
4811340.0000X
0.
5944560.
01514139.
261810.0000D1-
852.
765258.34191-
14.
616680.0000D2-
659.
169457.32637-
11.
498540.0000D3-
686.
248056.60532-
12.
123380.0000R-squared
0.988053Mean dependent var
5769.536Adjusted R-squared
0.985976S.D.dependent var
893.8838S.E.of regression
105.8573Akaike infocriterion
12.32249Sum squaredresid
257732.8Schwarz criterion
12.56039Log likelihood-
167.5149Hannan-Quinn criter.
12.39522F-statistic
475.5592Durbin-Watson stat
2.839627ProbF-statistic
0.000000图模型估计结果8-1由图可知当时，〃8-1DW=
2.8396,k=4,n=28=
1.104,%=
1.747,4-d DW4-d不能确定模型是否存在自相关因此采用LM检验法来进一步检验是否存在u L自相关，结果如所示8-2Breusch-Godfrey SerialCorrelation LMTest:F-statistic
6.993014Prob.F2,
210.0047Obs*R-squared
11.19329Prob.Chi-Square
20.0037Test Equation:Dependent Variable:RESIDMethod:Least SquaresDate:07/19/20Time:08:03Sample:2013Q12019Q4Included observations:28Presample missingvalue laggedresiduals setto zero.Variable CoefficientStd Error t-Statistic Prob.C
0.
881064111.
602100078950.9938X-
0.
0001050.012719-
0.
0082510.9935D1-
18.
4810747.58586-
0.
3883730.7016D
210.
1371347.
549860.
2131900.8332D
30.
01098945.
897220.
0002390.9998RESID-1-
0.
4873350.246524-
1.
9768250.0613RESID-
20.
3155640.
2507381.
2585440.2220R-squared
0.399760Mean dependent var
4.76E-13Adjusted R-squared
0.228263S.D.dependentvar
97.70188S.E.of regression
85.82968Akaike infocriterion
11.95492Sum squaredresid
154701.4Schwarz criterion
12.28798Log likelihood-
160.3689Hannan-Quinn criter.
12.05674F-statistic
2.331005Durbin-Watson stat
1.631514ProbF-statistic
0.069826图模型检验结果8-2LM由图8-2可知，nR2=
11.1933其对应〃=0037,在显著性水平
0.05下，模型存在自相关使用科克伦-奥克特迭代法作广义差分回归，估计结果如下Dependent Variable:YMethod:ARMA GeneralizedLeast SquaresBFGSDate:07/19/20Time:08:05Sample:2013Q12019Q4Included observations:28Convergence achievedafter6iterationsCoefficient covariancecomputed usingouter productof gradientsd.f.adjustment forstandard errorscovarianceVariable CoefficientStd.Errort-Statistic Prob.C
1274.
30087.
1907014.
615090.0000X
0.
593419000778276.
255920.0000D1-
880.
1003102.2407-
8.
6081220.0000D2-
645.
692239.27734-
16.
439310.0000D3-
696.
7881103.4557-
6.
7351320.0000AR⑴-
0.
6861920.195601-
3.
5081300.0020R-squared
0.992314Mean dependentvar
5769.536Adjusted R-squared
0.990567S.D dependentvar
893.8838S.E.of regression86,81610Akaike infocriterion1197560Sum squaredresid
165814.8Schwarz criterion
12.26108Log likelihood-
161.6584Hannan-Quinn criter.
12.06287F-statistic
568.0735Durbin-Watson stat
1.440713ProbF-statistic
0.000000Inverted ARRoots-.69图科克伦-奥克特迭代法估计结果8-3此时，通过检验，该模型已经无自相关由/检验值判断虚拟变量的引入方式，并写LM出模型的估计结果如下所示Y=
1274.3+
0.5934X—
880.1003D-
645.69222,一
696.7881Dt IZ3/t=
14.
615176.2560-
8.8081-
16.4393-
6.7351R2=
0.9923R2=
0.9906F=5689735DW=
1.4407所有参数均通过显著性检验，这说明居民人均消费支出确实存在季节性波动，对于每个季度的城镇居民人均可支配收入每增加元，则居民人均消费支出增加元，但是每个季度
10.5934的固定支出是不一样的，可以发现，第一季度的固定支出最小，第四季度的固定支出最大，每个季度的具体模型如下所示第一季度W=
367.1997+
0.5934X,第二季度£=60L6078+
0.5934X,第三季度£=
550.5119+
0.5934%.第四季度R=
1274.3+
0.5934X,男性已婚女性.首先定义%,%两个类别变量对应的的虚拟变量上;P2‘2~1o单身定义为延迟支付次数当时，取值；当取其它值X,X3=2-1,0X,0X3时，『按照教材例中的方法，可以得到相应的模型和模型X X
38.3Probit LogitDependent Variable:YMethod:ML-Binary ProbitNewton-Raphson/Marquardt stepsDate:08/24/21Time:08:52Sample:128Included observations:28Convergence achievedafter3iterationsCoefficient covariancecomputed usingobserved HessianCoefficientStd.Error z-Statistic Prob.C1-
1.
0772250.420036-
2.
5646000.0103C2-
0.
1236270.681623-
0.
1813720.8561C
30.
4183060.
5727320.
7303680.4652C
40.
9329960.
4737071.
9695640.0489McFadden R-squared
0.177947Mean dependentvar
0.285714S.D.dependentvar
0.460044S.E.of regression
0.430408Akaike infocriterion
1.269333Sum squaredresid
4.446034Schwarz criterion
1.459648Log likelihood-
13.77067Hannan-Quinn criter.
1.327514Deviance
27.54133Restr.deviance
33.50310Restr.log likelihood-
16.75155LR statistic
5.961765Avg.log likelihood-
0.491810ProbLR statistic
0.113485Obs withDep=020Total obs28Obs withDep=18图模型估计结果8-4Probit估计结果如图所示，最终得到的估计回归方程为Probit8-4+p=0-
1.0772—
0.
12360.4183D+
0.9330X24z=-
2.5646-
0.
18140.
73041.9696McFadden-R2=
0.1779,LT=
5.9618从结果来看，三个变量在的显著水平下均不显著，总体的显著性也不高,究其原因可D X35%2能是该数据集的样本量较小，另外评估客户的信息较少，缺少客户相应的资产和薪酬等信息支撑将和剔除掉，得到的模型估计如图所示D1D2Probit8-5DependentVariable:YMethod:ML-Binary ProbitNewton-Raphson/Marquardt stepsDate:08/24/21Time:08:57Sample:128Included observations:28Convergence achievedafter3iterationsCoefficient covariancecomputed usingobserved HessianCoefficientStd.Error z-Statistic Prob.C1-
0.
9283250.313170-
2.
9642810.0030C
21.
0034620.
4745172.
1146990.0345McFadden R-squared
0.159721Mean dependentvar
0.285714S.D.dependentvar
0.460044S.E.of regression
0.419064Akaike infocriterion
1.148284Sum squaredresid
4.565974Schwarz criterion
1.243442Log likelihood-
14.07598Hannan-Quinn criter.
1.177375Deviance
28.15196Restr deviance
33.50310Restr.log likelihood-
16.75155LR statistic
5.351136Avg.log likelihood-
0.502714ProbLR statistic
0.020709Obs withDep=020Total obs28Obs withDep=18图剔除变量后模型估计结果8-5Probit最终得到的估计回归方程为^=O-
0.9283+
1.0035X4z=-
2.
96432.1147McFadden-R2=
0.1597,LR=
5.3511此时解释变量在的显著水平下是显著的另外的估计系数是正的，这说明延迟支付X45%X’次数越多，该客户被判逾期的概率越大类似地，可以按照教材例中的方法针对模型进行估计和分析
8.3Logit。
个人认证
优秀文档
获得点赞 0
