学 生 实 验 报 告
课程名称： 计量经济学 专业班级： 经济1201班 姓 名： 学 号： 指导教师： 徐冬梅 职 称： 讲师 实验日期： 2014.12.11
农业大学经济贸易学院
.
.
学生实验报告
学生 学号 组员： 实验项目 EVIEWS的使用 √必修 □选修 √演示性实验 √验证性实验 □操作性实验 □综合性实验 实验地点 指导教师 管理模拟实验室 实验仪器台号 实验日期及节次 一、实验目的及要求
1、目的
会使用EVIEWS对计量经济模型进行分析
2、容及要求
（1）对经典线形回归模型进行参数估计、参数的检验与区间估计，对模型总体进行显著性检验；
（2）异方差的检验及其处理； （3）自相关的检验及其处理； （4）多重共线性检验及其处理； 二、仪器用具
仪器名称 计算机 Eviews 规格/型号 数量 1 1 备注 无网络环境 三、实验方法与步骤
(一)数据的输入、描述及其图形处理；
.
.
(二)方程的估计；
(三)参数的检验、违背经典假定的检验； (四)模型的处理与预测
250002000015000Y1000050006000800010000120001400016000X
四、实验结果与数据处理
实验一：中国城镇居民人均消费支出模型
数据散点图：
通过Eviews估计参数方程 回归方程：
Dependent Variable: Y Method: Least Squares Date: 11/27/14 Time: 15:02 Sample: 1 31
Included observations: 31
Variable
X
.
CoefficienStd. Error t-Statistic
t
1.359477
0.043302
31.39525
Prob. 0.0000
.
C
R-squared
Adjusted R-squared
-57.90655 377.7595 -0.153289 0.8792 11363.69 3294.469
0.971419 Mean dependent var 0.970433 S.D. dependent var
15000001000000E250000006000800010000120001400016000XS.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
566.4812 Akaike info
criterion
9306127. Schwarz criterion -239.4761 F-statistic 1.294974 Prob(F-statistic)
15.57911 15.67162 985.6616 0.000000
得出估计方程为：Y = 1.35947661442*X - 57.9065479515 异方差检验 1、图示检验法
图形呈现离散趋势，大致判断存在异方差性。 2、Park检验
Dependent Variable: LOG(E2) Method: Least Squares Date: 11/27/14 Time: 16:16 Sample: 1 31
Included observations: 31
Variable
.
t
CoefficienStd. Error t-Statistic
Prob.
.
C LOG(X)
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
19.82562 -0.956403
19.85359
0.998591
0.3263 0.6676
11.21371 2.894595 5.053338 5.145854 0.188290 0.667555
2.204080 -0.433924
0.006451 Mean dependent var -0.027809 S.D. dependent var 2.934568 Akaike info
criterion
249.7389 Schwarz criterion -76.32674 F-statistic 2.456500 Prob(F-statistic)
看到图中LOG(E2)中P值为0.6676 > 0.05，所以不存在异方差性 3、G-Q检验 e1检验：
Dependent Variable: X Method: Least Squares Date: 11/27/14 Time: 16:41 Sample: 1 12
Included observations: 12
Variable
C Y
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
t 4642.028 0.231046
2014.183 0.215824
2.304671 1.070530
0.0439 0.3095
6796.390 293.2762 14.33793 14.41875 1.146034 0.309538
CoefficienStd. Error t-Statistic
Prob.
0.102820 Mean dependent var 0.013102 S.D. dependent var 291.3486 Akaike info
criterion
848840.2 Schwarz criterion -84.02758 F-statistic 0.445146 Prob(F-statistic)
e2检验：
Dependent Variable: X Method: Least Squares Date: 11/27/14 Time: 16:42 Sample: 20 31
Included observations: 12
.
.
Variable
C Y
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
t 583.4526 0.697748
CoefficienStd. Error t-Statistic
593.4370 0.040196
0.983175 17.35870
Prob.
0.3487 0.0000
10586.89 2610.864 15.38082 15.46164 301.3245 0.000000
0.967879 Mean dependent var 0.964667 S.D. dependent var 490.7655 Akaike info
criterion
2408507. Schwarz criterion -90.28493 F-statistic 2.748144 Prob(F-statistic)
第一个图中的残差平方和为848840.2 第二个图中的残差平方和为2408507
所以F值为2408507/848840.2 = 2.8374 < 2.97，所以不存在异方差性 4、White检验
White Heteroskedasticity Test:
F-statistic Obs*R-squared
Test Equation:
Dependent Variable: RESID^2 Method: Least Squares Date: 11/27/14 Time: 16:50 Sample: 1 31
Included observations: 31
Variable
C X
.
2.240402 Probability 4.276524 Probability
0.125152 0.117860
t -2135113. 503.7331
CoefficienStd. Error t-Statistic
242.2078
2.079756
Prob.
1158576. -1.842876 0.0760 0.0468
.
X^2
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
-0.023609
0.011650 -2.026590
0.0523
300197.6 347663.4 28.36817 28.50694 2.240402 0.125152
0.137952 Mean dependent var 0.076378 S.D. dependent var 334122.9 Akaike info
criterion
3.13E+12 Schwarz criterion -436.7067 F-statistic 1.871252 Prob(F-statistic)
P值为0.11786 > 0.05，所以不存在异方差性
通过四种不同的检验得知除了图示检验法得出异方差的结论，其他的检验的结论都是不存在异方差的。
5、WLS（加权最小二乘法）修正
Dependent Variable: Y Method: Least Squares Date: 11/27/14 Time: 17:14 Sample: 1 31
Included observations: 31 Weighting series: E3
Variable
C X
Weighted Statistics R-squared
Adjusted R-squared S.E. of regression
CoefficienStd. Error t-Statistic
t
-85.69426 1.362221
24.15675 -3.547425 0.002307
590.5615
0.0013 0.0000 13474.53 61353.74 9.559810 Prob.
1.000000 Mean dependent var 1.000000 S.D. dependent var 27.93264 Akaike info
criterion
.
.
Sum squared resid Log likelihood Durbin-Watson stat
Unweighted Statistics R-squared
Adjusted R-squared S.E. of regression Durbin-Watson stat
22626.73 Schwarz criterion -146.1770 F-statistic 2.061818 Prob(F-statistic)
9.652325 348762.9 0.000000
11363.69 3294.469 9308110.
0.971413 Mean dependent var 0.970427 S.D. dependent var 566.5415 Sum squared resid 2.178992
实验二：中国粮食生产函数
1、回归方程
Dependent Variable: LOG(Y) Method: Least Squares Date: 12/11/14 Time: 15:06 Sample: 1983 2007
Included observations: 25
Variable LOG(X1) LOG(X2) LOG(X3) LOG(X4)
.
CoefficienStd. Error t-Statistic
t
0.381145 1.222289 -0.081110 -0.047229
0.050242 0.135179
7.586182 9.042030
Prob. 0.0000 0.0000 0.0000 0.3047
0.015304 -5.300024 0.044767 -1.054980
.
LOG(X5)
C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
-0.101174 -4.173174
0.057687 -1.753853 1.923624 -2.169434
0.0956 0.0429 10.70905 0.093396
0.981597 Mean dependent var 0.976753 S.D. dependent var 0.014240 Akaike info
criterion
0.003853 Schwarz criterion 74.24960 F-statistic 1.791427 Prob(F-statistic)
-5.459968 -5.167438 202.6826 0.000000
得出回归方程为：
LOG(Y) = 0.381144581612*LOG(X1) + 1.22228859801*LOG(X2) - 0.0811098881534*LOG(X3) - 0.04722870996*LOG(X4) - 0.101173736285*LOG(X5) - 4.17317444909
通过检验结果可知 R较大且接近于1，而且F=202.6826 > F0.05(5,19) = 2.74，故认为粮食产量与上述变量之间总体线性关系显著。但是由于其中X4、X5前的参数估计值未通过t检验，且符号的经济意义不合理，故认为解释变量之间存在多重共线。
2
2、相关系数表
LNX1 LNX2 LNX3 LNX4 LNX5 LNX1 1.000000 -0.568744 0.451700 0.964357 0.440205 LNX2 -0.568744 1.000000 -0.214097 -0.697625 -0.073270 LNX3 0.451700 -0.214097 1.000000 0.398780 0.411279 LNX4 0.964357 -0.697625 0.398780 1.000000 0.279528 LNX5 0.440205 -0.073270 0.411279 0.279528 1.000000 由表可知LnX1与LnX2之间存在高度的线性相关性
3、简单的回归形式
.
.
LnY与LnX1
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:15 Sample: 1983 2007
Included observations: 25
Variable
LNX1 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
t 0.224005 8.902008
0.025515 0.206034
8.779293 43.20657
0.0000 0.0000 10.70905 0.093396 -3.255189 -3.157679 77.07599 0.000000
Prob.
CoefficienStd. Error t-Statistic
0.770175 Mean dependent var 0.760182 S.D. dependent var 0.045737 Akaike info
criterion
0.048114 Schwarz criterion 42.68986 F-statistic 0.939435 Prob(F-statistic)
LnY与LnX2
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:16 Sample: 1983 2007
Included observations: 25
Variable
LNX2 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
.
t -0.383434 15.15748
Prob. 0.4595 0.0174 10.70905 0.093396 -1.809063 -1.711553 0.565986 0.459489
CoefficienStd. Error t-Statistic
5.912971
2.563429
0.509669 -0.752321
0.024017 Mean dependent var -0.018417 S.D. dependent var 0.094252 Akaike info
criterion
0.204321 Schwarz criterion 24.61329 F-statistic 0.335219 Prob(F-statistic)
.
LnY与LnX3
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:18 Sample: 1983 2007
Included observations: 25
Variable
LNX3 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
t 0.108067 9.619722
0.085271 0.859744
1.267335 11.18905
0.2177 0.0000 10.70905 0.093396 -1.852255 -1.754745 1.606139 0.217717
Prob.
CoefficienStd. Error t-Statistic
0.065274 Mean dependent var 0.024634 S.D. dependent var 0.092239 Akaike info
criterion
0.195684 Schwarz criterion 25.15319 F-statistic 0.597749 Prob(F-statistic)
LnY与LnX4
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:18 Sample: 1983 2007
Included observations: 25
Variable
LNX4 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood
.
t 0.166976 8.949090
CoefficienStd. Error t-Statistic
0.028274 0.298255
5.905670 30.00479
Prob.
0.0000 0.0000
10.70905 0.093396
0.602605 Mean dependent var 0.585327 S.D. dependent var 0.060143 Akaike info
criterion
0.083194 Schwarz criterion 35.84472 F-statistic
-2.707578 -2.610068 34.87693
.
Durbin-Watson stat
0.625528 Prob(F-statistic)
0.000005
LnY与LnX5
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:19 Sample: 1983 2007
Included observations: 25
Variable LNX5 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
CoefficienStd. Error t-Statistic
t
0.488731 5.600749
0.234606 2.452207
2.083199 2.283962
0.0485 0.0319 10.70905 0.093396 -1.957599 -1.860089 4.339718 0.048538 Prob.
0.158733 Mean dependent var 0.122156 S.D. dependent var 0.087506 Akaike info
criterion
0.176118 Schwarz criterion 26.46999 F-statistic 0.327932 Prob(F-statistic)
比较各个回归方程的R可知Y与X1的R最大，即粮食生产受农业化肥施用量最大，与经验相符，因此选为初始的回归方程。
且初始化回归方程为：
LOG(Y) = 0.224004867873*LOG(X1) + 8.90200821784
22
R2 = 0.770175 D.W.= 0.939435
4、逐步回归
.
.
LnY与LnX1
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:28 Sample: 1983 2007
Included observations: 25
Variable
LNX1 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
t 0.224005 8.902008
0.025515 0.206034
8.779293 43.20657
0.0000 0.0000 10.70905 0.093396 -3.255189 -3.157679 77.07599 0.000000
Prob.
CoefficienStd. Error t-Statistic
0.770175 Mean dependent var 0.760182 S.D. dependent var 0.045737 Akaike info
criterion
0.048114 Schwarz criterion 42.68986 F-statistic 0.939435 Prob(F-statistic)
LnY与LnX1、LnX2
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:29 Sample: 1983 2007
Included observations: 25
Variable
LNX1 LNX2 C
R-squared
Adjusted R-squared S.E. of regression
t 0.297854 1.258622 -6.295682
0.015482 0.150066
19.23929 8.387127
0.0000 0.0000 0.0022
10.70905 0.093396 -4.609666
CoefficienStd. Error t-Statistic
Prob.
1.814941 -3.468809
0.945246 Mean dependent var 0.940269 S.D. dependent var 0.022826 Akaike info
criterion
.
.
Sum squared resid Log likelihood Durbin-Watson stat
0.011463 Schwarz criterion 60.62083 F-statistic 1.595748 Prob(F-statistic)
-4.463401 189.9002 0.000000
由输出结果可知R有所提高，且各解释变量前得参数均通过t检验，符号也合理。
2
D.W.检验也表明不存在一阶自相关。可以考虑再此模型上继续引入X3。
LnY与LnX1、LnX2、LnX3
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:30 Sample: 1983 2007
Included observations: 25
Variable
LNX1 LNX2 LNX3 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
2
t 0.323385 1.290729 -0.086754 -5.999638
CoefficienStd. Error t-Statistic
0.010861 0.096153
29.77552 13.42365
Prob.
0.0000 0.0000 0.0000 0.0000
10.70905 0.093396
0.015155 -5.724484 1.162078 -5.162852
0.978616 Mean dependent var 0.975561 S.D. dependent var 0.014601 Akaike info
criterion
0.004477 Schwarz criterion 72.37318 F-statistic 1.412883 Prob(F-statistic)
-5.469854 -5.274834 320.3438 0.000000
由输出结果可知R再次提高且参数符号合理，变量通过t检验。但是D.W.=1.419（dL=1.12、dU=1.66）落入无法判断的区域，且X4的参数没有通过t检验。 LM检验
Breusch-Godfrey Serial Correlation LM Test:
.
.
F-statistic Obs*R-squared
Test Equation:
1.241319 Probability 1.460972 Probability
0.278428 0.226776
Dependent Variable: RESID Method: Least Squares Date: 12/11/14 Time: 15:43
Variable LNX1 LNX2 LNX3 C RESID(-1) R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
CoefficienStd. Error t-Statistic
t
0.002403 0.006952 -0.005478 -0.044729 0.257459
0.011012 0.095809
0.218225 0.072561
0.8295 0.9429 0.7333 0.9695 0.2784 1.07E-16 0.013658 -5.450070 -5.206295 0.310330 0.867655 Prob.
0.015850 -0.345589 1.156156 -0.038688 0.231082
1.114145
0.058439 Mean dependent var -0.129873 S.D. dependent var 0.014517 Akaike info
criterion
0.004215 Schwarz criterion 73.12588 F-statistic 1.794969 Prob(F-statistic)
LM检验显示不存在一阶自相关，继续引入X4。 LnY与LnX1、LnX2、LnX3、LnX4
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:32 Sample: 1983 2007
Included observations: 25
Variable
LNX1 LNX2 LNX3 LNX4 C
R-squared
Adjusted R-squared S.E. of regression
.
t 0.322061 1.294001 -0.086665 0.001303 -6.041554
CoefficienStd. Error t-Statistic
0.039161 0.135368 0.036972
8.223957 9.559117 0.035251
Prob.
0.0000 0.0000 0.0000 0.9722 0.0018
10.70905 0.093396
0.015730 -5.509509 1.682783 -3.590215
0.978617 Mean dependent var 0.974341 S.D. dependent var 0.014961 Akaike info
-5.389916
.
criterion
Sum squared resid Log likelihood Durbin-Watson stat
0.004476 Schwarz criterion 72.37395 F-statistic 1.413284 Prob(F-statistic)
-5.146141 228.8316 0.000000
由输出结果可知R2有所下降，且X4的参数未能通过t检验。去掉X4引入X5。
LnY与LnX1、LnX2、LnX3、LnX5
Dependent Variable: LNY Method: Least Squares Date: 12/11/14 Time: 15:33 Sample: 1983 2007
Included observations: 25
Variable
LNX1 LNX2 LNX3 LNX5 C
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
2
t 0.329539 1.322227 -0.081194 -0.063556 -5.806454
CoefficienStd. Error t-Statistic
0.011499 0.096705
28.65841 13.67276
Prob.
0.0000 0.0000 0.0000 0.1775 0.0001
10.70905 0.093396
0.015347 -5.290700 0.045474 -1.397624 1.144940 -5.071405
0.980518 Mean dependent var 0.976622 S.D. dependent var 0.014280 Akaike info
criterion
0.004078 Schwarz criterion 73.53802 F-statistic 1.635288 Prob(F-statistic)
-5.483042 -5.239267 251.6534 0.000000
由输出结果可知R虽然有所提高但是X5的参数未能通过t检验，且符号与经济意义不符。
经过逐步回归可知，X4与X5是多余的。同时还可以继续验证，如果用与X1高度相关的X4替代X1，则X4与X2、X3、X5之间的任意线性组合，均达不到以X1、X2、X3为解释变量的回归效果。因此，最终的粮食生产方程应以Y=f(X1,X2,X3)为最优，
.
.
拟合效果如下：
LOG(Y) = 0.323384862318*LOG(X1) + 1.2907291245*LOG(X2) - 0.086753884442*LOG(X3) - 5.99963819217
五、讨论与结论
首次接触计量经济学，通过利用Eviews软件将所学到的计量知识进行实践，让我加深了对理论的理解和掌握，直观而充分地体会到老师课堂讲授容的精华之所在。在实验过程中我们提高了手动操作软件、数量化分析与解决问题的能力，还可以培养我在处理实验经济问题的严谨的科学的态度，并且避免了课堂知识与实际应用的脱节。虽然在实验过程中出现了很多错误，但这些经验却锤炼了我们发现问题的眼光，丰富了我们分析问题的思路。通过这次实验教学使我受益匪浅。
通过此次实验，让我对Eviews软件有了进一步的了解，对于相关实验步骤也比较熟悉了，但是由于是全英文的软件操作，所以经常会遗忘一些英文字母的含义。虽然在做实操过程中还存在一定的难度，但是我坚信只要多加练习、操作，多加熟悉Eviews软件，以后的实验会慢慢熟悉并更好的操控。
计量经济学是一门比较难的课程，其中涉及大量的公式，不容易理解且需要大量的运算，所以在学习的过程中我遇到了很多困难。但通过这次的实验，我对课上所学的最小二乘法有了进一步的理解，在掌握理论知识的同时，将其与实际的经济问题联系起来。
六、指导教师评语及成绩：
评语：
.
.
.
指导教师签名：徐冬梅
批阅日期：2014.12.11
成绩：优秀
因篇幅问题不能全部显示，请点此查看更多更全内容