课程名称:计量经济学
调查报告题目:随机解释变量分析
学生姓名学号:汪佳瑶 2013105020
所在院系:经济学院2013级经济统计专业
日期:2015年6月20日
一、实验目的:掌握用Eviews检验模型中随机解释变量的方法,学会豪斯曼检验和工具变量法 二、实验内容:
1.以1983-2013年中国城镇居民人均消费为被解释变量,人均可支配收入和前一年人均消费为解释变量,建立模型。数据来自2014统计年鉴。 2.选择工具变量
3.采用豪斯曼检验,判断城镇居民人均可支配收入是否为内生变量
4.采用工具变量估计 三、实验步骤:
1.建立工作表,导入EXCEL数据,在命令窗口输入 LS Y C X1 X2,点击enter,可得原模型最小二乘法估计
obs 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Y 15161 11265 9997 7943 6511 5309 4616 4186 3538 2111 1454 1211 884 673 506 457
X1 21810 17175 13786 10493 8472 6860 5854 5160 4283 2577 1700 1373 1002 739 565 491
X2 13471 11243 8697 7182 6030 4998 4332 3919 2851 1672 1279 1104 799 559 471 412
Z 19109 15781 11759 9421 7702 6280 5425 4838 3496 2026 1510 1181 899 651 526 477
Dependent Variable: Y
Method: Least Squares Date: 06/24/15 Time: 16:59 Sample: 1 16
Included observations: 16
Variable
C X1 X2
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
Coefficient
Std. Error
t-Statistic
Prob.
249.9378112139.648890661.78975865900.09680072
98103 6476148 0829256 0.9956787050.995013891
3352694 S.D. dependent var 311.6232288
551938 Akaike info criterion 1262417.477
907774 Schwarz criterion -112.910619
3450516 Hannan-Quinn criter. 1497.679010
989362 Durbin-Watson stat 4.280454747098539e-16
84478 329699 416775
32254 290437808 63113 470833046 56515 00661276
4738.875 4413.15268
827173 14.48882741813145 14.63368780355141 14.49624545833328 2.166335430262757
0.3978727320.17408210872.28554637540.039708890.4506686500.27700544551.62693065180.12773497
8239001 Mean dependent var
Y=249.937+0.39787X1+0.4506X2
(1.78) (2.28) (1.62) F=1497.67 DW=2.16633543
2.如果考虑居民人均消费支出(Y)由人均可支配收入(X1)决定的同时,人均消费支出(Y)又反过来影响着同期居民人均可支配收入。所以死那些在模型中独立列出而纳入随机干扰项的影响居民人均消费支出的因素,也影响着居民可支配收入。据此有理由怀疑居民可支配收入和随机干扰项u同期相关对前一年城镇
居民消费支出而言,它影响着当年的消费支出,而当年的消费支出不会反过来影响前一年的消费支出,因此可认为X2是同期外生变量。
3.选择前一年的可支配收入Z作为X1的工具变量,Z与原模型的随机干扰项不会存在同期相关性,将X1关于X2。Z进行普通最小二乘估计得到
Dependent Variable: X1 Method: Least Squares Date: 06/24/15 Time: 17:01 Sample: 1 16
Included observations: 16
Variable
C X2
Coefficient
Std. Error
t-Statistic
Prob.
70.58253532154.673245100.45633318980.65568081
737854 93010492
00168 854325
916166 30450225 8830496 29944285
0.00034065
1.2974957550.26974719164.81004360803381114732-0.245824060.3815450051-0.6442859050.53059440
Z
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
139393 0.9981638320.997881344
824275
42194
2
0306457 Mean dependent var 6507451 S.D. dependent var 297.7847984
126216 Akaike info criterion 1152785.220
153394 Schwarz criterion -112.183838
7754418 Hannan-Quinn criter. 3533.481150
137393 Durbin-Watson stat 1.642217389432566e-18
6396.25 6469.52457
29497 14.39797984693022 14.54284023235018 14.40539788713206 2.671838385990076
X1=70.58-0.245X2+1.297Z
记录残差V,并将其加入原模型进行最小二乘法估计得到
Dependent Variable: Y Method: Least Squares Date: 06/24/15 Time: 17:15 Sample: 1 16
Included observations: 16
Variable
C X1
Coefficient
Std. Error
t-Statistic
Prob.
57.7027648966.2843715300.87053348410.40108847
569188 90838129
10042 4522919
937581 58211431 3748743 95205639
7.67245245
1.1282403030.15131807807.45608402206570388e-00.0290332390.0951821989-0.3050280430.76557061
X2 V
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood F-statistic Prob(F-statistic)
88578 21132
0.9992365040.999045630
352256 913454
5013 6
1.1866843080.15869279827.47787121397.45132102
95969 279135e-06
4738.875 4413.15268
827173 12.88042475857535 13.07357193913529 12.89031547884446 2.023506287031011
2527139 Mean dependent var 3158924 S.D. dependent var 136.3349713
792029 Akaike info criterion 223046.6930
51617 Schwarz criterion -99.0433980
6860279 Hannan-Quinn criter. 5235.059960
999177 Durbin-Watson stat 5.807015110481363e-19
t检验表明,V之前的参数显著不为0,因此判断居民人均可支配收入确实是内生变量。从而原模型的普通最小二乘估计量有偏并且非一致。需采用工具变量法进行估计。