计量经济学实验报告
姓名
学号
班级
任课老师
日期
一、基本数据
某地区消费总额Y和货币收入总额X的年度资料(单位:亿元)
年份
1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
X
103.169 115.07 132.21 156.574 166.091 155.099 138.175 146.936 157.7 179.797 195.779 194.858 189.179 199.963 205.717 215.539 220.391 235.483 280.975 292.339 278.116 292.654 341.442 401.141 458.567 500.915 450.939 626.709 783.953 890.637 Y
91.158 109.1 119.187 143.908 155.192 148.673 151.288 148.1 156.777 168.475 174.737 182.802 180.13 190.444 196.9 204.75 218.666 227.425 229.86 244.23 258.363 275.248 299.277 345.47 406.119 462.223 492.662 539.046 617.568 727.397
(计量经济学课本第七章课后习题7.8中的数据)
二、散点图
建立消费总额Y和货币收入总额X的回归模型,利用以上的消费总额Y和货币收入总额X的数据表,作散点图
1000800600X40020000200400Y600800
三、均衡模型的构建
从以上散点图可以看出消费总额Y和货币收入总额X大体呈现为线性关系,为分析某地区消费总额随货币收入总额变动撒数量规律性,可以建立如下简单线性回归模型:
Yt12Xtut
Eviews的回归结果为
Dependent Variable: Y Method: Least Squares Date: 06/20/12 Time: 14:22 Sample: 1975 2004 Included observations: 30 Variable C X R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 27.76594 0.807731 Std. Error 7.945083 0.022840 t-Statistic 3.494733 35.36542 Prob. 0.0016 0.0000 159.3349 9.257921 9.351334 1250.713 0.000000 0.978103 Mean dependent var 262.1725 0.977321 S.D. dependent var 23.99515 Akaike info criterion 16121.49 Schwarz criterion -136.8688 F-statistic 1.280986 Prob(F-statistic) 用规范的形式将参数估计和检验结果写为 ˆ27.765940.807731X Ytt (7.945083) (0.022840)
t=(3.945083) (35.36542) R= 0.978103 F=1250.713 n=30
2对回归系数的t检验:针对H0:1=0和H0:2=0,由上表可以看出,估计的回
ˆ的标准误ˆ的标准误差和t值分别为:SE(ˆ)=7.945083,t(ˆ)=3.945083;归系数2111ˆ)=0.022840,t(ˆ)=35.36542。取=0.05,查t分布表得自差和t值分别为:SE(22ˆ)=3.945083>t(28)=2.048,由度为n-2=30-2=28的临界值t0.025(28)=2.048。因为t(0.0251ˆ)=35.36542>t(28)=2.048,所以应拒绝H。这表明,所以应接受H0;因为t(0.02502货币收入总额对消费总额有显著影响。
四、 ADF检验
首先,对货币收入总额X进行ADF检验,检验如下: 1、 对水平差分序列作单位根检验
ADF Test Statistic
Dependent Variable: D(X) Method: Least Squares Date: 06/20/12 Time: 14:38 Sample(adjusted): 1977 2004
Included observations: 28 after adjusting endpoints
Variable X(-1) D(X(-1)) C R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 0.208932 -0.087765 -27.71535 Std. Error 0.071378 0.247909 17.33647 t-Statistic 2.927103 -0.354020 -1.598673 Prob. 0.0072 0.7263 0.1225 2.927103 1% Critical Value*
5% Critical Value 10% Critical Value
-3.6852 -2.9705 -2.6242
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
0.395717 Mean dependent var 27.69882 0.347374 S.D. dependent var 39.49003 Akaike info criterion 38986.56 Schwarz criterion -141.0730 F-statistic 2.060252 Prob(F-statistic) 48.88271 10.29093 10.43367 8.185658 0.001843 从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.7204、-2.9850、-2.6318,t检验统计量值为2.927103,大于相应临界值,从而不能拒绝H0,表明货币收入总额X的查分序列存在单位根,是非平稳序列。 2、对一阶差分序列作单位根检验:
ADF Test Statistic
-2.113198 1% Critical Value*
5% Critical Value 10% Critical Value
-3.6959 -2.9750 -2.6265
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(X,2) Method: Least Squares Date: 06/20/12 Time: 14:46 Sample(adjusted): 1978 2004
Included observations: 27 after adjusting endpoints
Variable D(X(-1)) D(X(-1),2)
C R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient -0.559375 0.021299 17.05944 Std. Error 0.264705 0.247353 10.65828 t-Statistic -2.113198 0.086108 1.600580 Prob. 0.0452 0.9321 0.1226 0.247292 Mean dependent var 3.316444 0.184566 S.D. dependent var 46.68466 Akaike info criterion 52306.97 Schwarz criterion -140.4935 F-statistic 1.988314 Prob(F-statistic) 51.69872 10.62915 10.77313 3.942435 0.033076 从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.6959、-2.9750、-2.6265,t检验统计量值为-2.113198,大于相应临界值,从而不能拒绝H0,表明货币收入总额X的查分序列存在单位根,是非平稳序列。 3、对二阶差分序列作单位根检验:
ADF Test Statistic
Dependent Variable: D(X,3) Method: Least Squares Date: 06/20/12 Time: 14:49 Sample(adjusted): 1979 2004
Included observations: 26 after adjusting endpoints Variable D(X(-1),2) D(X(-1),3)
Coefficient -1.925268 0.435787
Std. Error 0.323033 0.196558
t-Statistic -5.959968 2.217095
Prob. 0.0000 0.0368
-5.959968 1% Critical Value*
5% Critical Value 10% Critical Value
-3.7076 -2.9798 -2.6290
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
C R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
8.550394 9.435772 0.906168 0.3742 84.96286 10.65214 10.79730 29.11190 0.000000
0.716832 Mean dependent var -2.222462 0.692208 S.D. dependent var 47.13652 Akaike info criterion 51102.57 Schwarz criterion -135.4778 F-statistic 2.126304 Prob(F-statistic)
从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.7076、-2.9798、-2.6290,t检验统计量值为-5.959968,小于相应临界值,从而拒绝H0,表明货币收入总额X的查分序列不存在单位根,是平稳序列。即X序列是二单整的,X~I(2)。
其次,再对消费总额Y进行ADF检验,检验如下: 1、对水平差分序列作单位根检验
ADF Test Statistic
Dependent Variable: D(Y) Method: Least Squares Date: 06/20/12 Time: 14:53 Sample(adjusted): 1977 2004
Included observations: 28 after adjusting endpoints Variable Y(-1) D(Y(-1)) C R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
Coefficient 0.105921 0.536051 -14.65243 Std. Error 0.033040 0.214124 5.705804 t-Statistic 3.205793 2.503461 -2.567987 Prob. 0.0037 0.0192 0.0166 26.93653 7.859215 8.001951 58.96984 0.000000
3.205793 1% Critical Value*
5% Critical Value 10% Critical Value
-3.6852 -2.9705 -2.6242
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
0.825101 Mean dependent var 22.08204 0.811109 S.D. dependent var 11.70704 Akaike info criterion 3426.370 Schwarz criterion -107.0290 F-statistic 1.610061 Prob(F-statistic)
从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.6852、-2.9705、-2.6242,t检验统计量值为3.205793,大于相应临界值,从而不能拒绝H0,表明消费总额Y的查分序列存在单位根,是非平稳序列。 2、对一阶差分序列作单位根检验:
ADF Test Statistic
0.576398 1% Critical Value*
5% Critical Value 10% Critical Value
-3.6959 -2.9750 -2.6265
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation Dependent Variable: D(Y,2) Method: Least Squares Date: 06/20/12 Time: 14:55 Sample(adjusted): 1978 2004
Included observations: 27 after adjusting endpoints
Variable D(Y(-1)) D(Y(-1),2)
C R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 0.090569 0.106004 1.750697 Std. Error 0.157129 0.260915 3.788948 t-Statistic 0.576398 0.406276 0.462054 Prob. 0.5697 0.6881 0.6482 0.044619 Mean dependent var 3.694148 -0.034996 S.D. dependent var 13.95920 Akaike info criterion 4676.623 Schwarz criterion -107.8970 F-statistic 1.776474 Prob(F-statistic) 13.72117 8.214594 8.358576 0.560433 0.578255 从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.6959、-2.9750、-2.6265,t检验统计量值为0.576398,大于相应临界值,从而不能拒绝H0,表明消费总额Y的查分序列存在单位根,是非平稳序列。 3、对二阶差分序列作单位根检验:
ADF Test Statistic
Dependent Variable: D(Y,3) Method: Least Squares Date: 06/20/12 Time: 14:57 Sample(adjusted): 1978 2004
Included observations: 27 after adjusting endpoints Variable D(Y(-1),2)
C
Coefficient -0.808926 3.265436
Std. Error 0.212268 2.692781
t-Statistic -3.810872 1.212663
Prob. 0.0008 0.2366
-3.810872 1% Critical Value*
5% Critical Value 10% Critical Value
-3.6959 -2.9750 -2.6265
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.367453 Mean dependent var 1.450444 0.342151 S.D. dependent var 13.77151 Akaike info criterion 4741.362 Schwarz criterion -108.0826 F-statistic 1.730854 Prob(F-statistic) 16.97925 8.154268 8.250256 14.52275 0.000804 从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值分别为-3.6959、-2.9750、-2.6265,t检验统计量值为-3.810872,小于相应临界值,从而拒绝H0,表明消费总额Y的查分序列不存在单位根,是平稳序列。即Y序列是二阶单整的,Y~I(2)。
五、模型检验 1、GQ检验
本题中样本容量n=30,删除中间1/4的观测值,即大约6个观测值,余下部分平分得两个样本区间:1975~1986和1993~2004,它们的样本个数均是12个,即n1n2=12
将区间定义为1975~1986,用OSL方法求的结果如下:
Dependent Variable: Y Method: Least Squares Date: 06/20/12 Time: 22:06 Sample: 1975 1986 Included observations: 12 Variable C X R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Dependent Variable: Y Method: Least Squares Date: 06/20/12 Time: 22:08 Sample: 1993 2004 Included observations: 12
Variable C X
R-squared
Coefficient 47.60486 0.772759
Std. Error 27.85910 0.055079
t-Statistic 1.708773 14.03000
Prob. 0.1183 0.0000
Coefficient 1.409389 0.947145 Std. Error 13.40484 0.086354 t-Statistic 0.105140 10.96813 Prob. 0.9183 0.0000 0.923254 Mean dependent var 146.2325 0.915579 S.D. dependent var 8.008268 Akaike info criterion 641.3236 Schwarz criterion -40.89903 F-statistic 1.946507 Prob(F-statistic) 27.56223 7.149838 7.230656 120.3000 0.000001 将区间定义为1993~2004,用OSL方法求的结果如下:
0.951654 Mean dependent var 408.1219
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
0.946819 S.D. dependent var 37.28323 Akaike info criterion 13900.39 Schwarz criterion -59.35586 F-statistic 1.872715 Prob(F-statistic)
161.6722 10.22598 10.30679 196.8408 0.000000
由两表得残差平方和分别为
e21t=641.3236、
e22t=13900.39,根据GQ检验F统计
e量为Fe21t2t213900.39=21.6745。在=0.05下,分子、分母的自由度均为12,查
641.3236F分布表得临界值F0.05(12,12)=2.69,因为F=21.6745>F0.05(12,12)=2.69,所以拒绝原假设,表明模型存在异方差。 2、White检验
White Heteroskedasticity Test: F-statistic Obs*R-squared
Test Equation:
Dependent Variable: RESID^2 Method: Least Squares Date: 06/20/12 Time: 22:24 Sample: 1975 2004 Included observations: 30
Variable C X X^2 R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
Coefficient -1613.959 11.59854 -0.010037 Std. Error 1229.604 7.097164 0.007622 t-Statistic -1.312585 1.634250 -1.316856 Prob. 0.2004 0.1138 0.1990 1851.110 17.91526 18.05538 1.963587 0.159889
21.963587 Probability 3.809440 Probability
0.159889 0.148864
0.126981 Mean dependent var 537.3829 0.062313 S.D. dependent var 1792.508 Akaike info criterion 86753306 Schwarz criterion -265.7289 F-statistic 2.409913 Prob(F-statistic)
2从上表可以看出,nR=3.80943,由White检验知,在=0.05时,查分布表,得临界值20.05(2)=5.9915,同时X和X的t检验值也显著。比较计算的统计量与临界值,因为nR=3.80943<20.05(2)=5.9915,所以接受原假设,拒绝备择假设,表明模型不存在异方差。则估计结果为
222ˆ181748.067.52214X0.004835X2 Yttt (1229.604) (7.097164) (0.007622)
R=0.126981 DW=2.409913 F=1.963587
六、协整检验
由(三、均衡模型的构建)中以消费总额Y为被解释变量和货币收入总额X为解释变
2ˆ27.765940.807731Xe。 量的OSL回归模型可再估计新的回归模型:Yttt在新的模型中我们可以令et=resid,然后对et序列进行单位根检验:
ADF Test Statistic
Dependent Variable: D(E) Method: Least Squares Date: 06/20/12 Time: 22:51 Sample(adjusted): 1976 2004
Included observations: 29 after adjusting endpoints Variable E(-1)
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Coefficient -0.883970
Std. Error 0.187663
t-Statistic -4.710406
Prob. 0.0001
-4.710406 1% Critical Value*
5% Critical Value 10% Critical Value
-2.6453 -1.9530 -1.6218
*MacKinnon critical values for rejection of hypothesis of a unit root.
Augmented Dickey-Fuller Test Equation
0.442097 Mean dependent var 0.006104 0.442097 S.D. dependent var 23.53726 Akaike info criterion 15512.07 Schwarz criterion -132.2393 Durbin-Watson stat 31.51204 9.188921 9.236069 2.014805 从检验结果看,在1%、5%、10%三个显著性水平下,单位根检验的Mackinnon临界值
分别为-2.6453、-1.9530、-1.6218,t检验统计量值为-4.710406,小于相应临界值,从而拒绝H0,表明货币收入总额X的查分序列不存在单位根,是平稳序列,说明消费总额Y和货币收入总额X之间存在协整关系。
七、误差修正模型的检验
因为消费总额Y和货币收入总额X之间存在协整关系,则表明两者之间有长期均衡关系。但从短期来看,可能会出现失衡,为了增强模型的精度,可以把协整回归式中的误差项通过建立误差修正模型把食品消费的短期行为与长期变化联系起来。误差et看作均衡误差,
修正模型的结构如下:YtXtet1t
由此估计回归模型式结果为:
Dependent Variable: DY Method: Least Squares Date: 06/20/12 Time: 22:58 Sample(adjusted): 1976 2004
Included observations: 29 after adjusting endpoints Variable C DX E(-1)
R-squared
Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat
Coefficient 2.703547 0.730568 -0.883584
Std. Error 5.477547 0.115190 0.193112
t-Statistic 0.493569 6.342311 -4.575509
Prob. 0.6258 0.0000 0.0001 42.58763 9.309047 9.450491 30.32425 0.000000
0.699937 Mean dependent var 21.93928 0.676855 S.D. dependent var 24.20929 Akaike info criterion 15238.33 Schwarz criterion -131.9812 F-statistic 2.047582 Prob(F-statistic)
因此最后得到的误差修正模型的估计结果:
ˆ2.7035470.730568X-0.883584e Yttt1 (5.477547) (0.115190) (0.193112)
t=(0.493569) (6.342311) (-4.575509)
R=0.699937 DW=2.047582
上述估计结果表明,食品消费支出的变化不仅取决于可支配收入的变化,而且取决于上一期食品消费支出对均衡水平的偏离,误差项et1估计的系数-0.883584体现了对偏离的修正,上一期偏离越远,本期修正的量就越大,即系统存在误差修正机制。
2
因篇幅问题不能全部显示,请点此查看更多更全内容