实验地点: 实验楼 时间:
课程名称 | 计量经济学模拟实验 | 实验项目名称 | 多元线性回归模型线性与非线性估计检验 | |||||||||||
班级 | 姓名 | 学号 | 学时 | |||||||||||
小组成员 | ||||||||||||||
实验目的: 掌握生产函数的估计、检验以及多参数的线性约束检验等内容 | ||||||||||||||
实验说明: 数据来源于教材p65页表4.1.1,工作文件夹是sy3.WF1,实验目的是让学生掌握生产函数的估计、检验以及多参数的线性约束检验等内容。注意:实际GDP是以1978年为100计算、资本存量是以1952年的不变价格计算。 实验内容: 1.估计双对数模型,以及说明各回归系数的经济含义;
(1)由于实验数据中已给出变量的对数形式,所以new objects“eq01”“gdp1 c k1 l1 Dependent Variable: GDP1 | ||||||||||||||
Method: Least Squares | ||||||||||||||
Date: 02/27/13 Time: 08:41 | ||||||||||||||
Sample: 1978 2006 | ||||||||||||||
Included observations: 29 | ||||||||||||||
Coefficient | Std. Error | t-Statistic | Prob. | |||||||||||
C | -9.047486 | 0.787252 | -11.49250 | 0.0000 | ||||||||||
K1 | 0.747596 | 0.021997 | 33.986 | 0.0000 | ||||||||||
L1 | 0.678880 | 0.090147 | 7.530822 | 0.0000 | ||||||||||
R-squared | 0.998348 | Mean dependent var | 5.887599 | |||||||||||
Adjusted R-squared | 0.998220 | S.D. dependent var | 0.800400 | |||||||||||
S.E. of regression | 0.033765 | Akaike info criterion | -3.841113 | |||||||||||
Sum squared resid | 0.0291 | Schwarz criterion | -3.699669 | |||||||||||
Log likelihood | 58.69614 | Hannan-Quinn criter. | -3.796814 | |||||||||||
F-statistic | 7854.199 | Durbin-Watson stat | 0.799482 | |||||||||||
Prob(F-statistic) | 0.000000 | Second-Stage SSR | 0.0291 |
所以该模型函数形式为Ln GDP= -9.0474855847 + 0.747595717651LnK + 0.678880192961LnL
回归系数的经济含义:资本每增加1%,GDP平均增加0.74759571765%,劳动每增加1%,GDP平均增加0.67888019296%
2.对模型做t检验和F检验;
T(β0)=-11.49250,T(β1)=33.986,T(β2)=7.530822,P值均为0,所以T检验说明回归模型中系数不为0,在一定显著性水平下这个模型是有意义的,模型中解释变量对于被解释变量有一定解释力度。
F=7854.199,P=0.000000,F检验说明拒绝原假设,模型总体存在。
3.在5%的显著性水平下对随机干扰项的方差做如下检验:和
输入scalar deltasqrhat1=0.027/(29-3)
4.利用F统计量来检验:
打开eq1ViewCoefficient TestsWald Coefficient Restrictions输入c(2)+c(3)=1ok
Wald Test: | |||
Equation: EQ1 | |||
Test Statistic | Value | df | Probability |
F-statistic | 37.318 | (1, 26) | 0.0000 |
Chi-square | 37.318 | 1 | 0.0000 |
Null Hypothesis Summary: | |||
Normalized Restriction (= 0) | Value | Std. Err. | |
-1 + C(2) + C(3) | 0.4276 | 0.069746 | |
Restrictions are linear in coefficients. |
5*.对模型进行非线性OLS估计:
a.设定初始值(双击序列C,在c(1)、c(2)和c(3)所对应的单元格中分别输入0,option中的收敛精度设为0.001,迭代次数100次),保存模型;
objecteq2GDP=C(1)*(K^C(2))*(L^C(3))optionok
Dependent Variable: GDP | ||||
Method: Least Squares | ||||
Sample: 1978 2006 | ||||
Included observations: 29 | ||||
Convergence achieved after 1 iteration | ||||
GDP=C(1)*(K^C(2))*(L^C(3)) | ||||
Variable | Coefficient | Std. Error | t-Statistic | Prob. |
C(1) | 4.014973 | 3.34E+17 | 1.20E-17 | 1.0000 |
C(2) | 1.942001 | 1.44E+15 | 1.35E-15 | 1.0000 |
C(3) | -4.547718 | 8.71E+15 | -5.22E-16 | 1.0000 |
R-squared | -1.858693 | Mean dependent var | 481.4144 | |
Adjusted R-squared | -2.078593 | S.D. dependent var | 359.35 | |
S.E. of regression | 630.5381 | Akaike info criterion | 15.82872 | |
Sum squared resid | 10337035 | Schwarz criterion | 15.97017 | |
Log likelihood | -226.5165 | Hannan-Quinn criter. | 15.87302 | |
Durbin-Watson stat | 0.008228 |
objecteq3GDP=C(1)*(K^C(2))*(L^C(3))optionok
Dependent Variable: GDP | ||||
Method: Least Squares | ||||
Sample: 1978 2006 | ||||
Included observations: 29 | ||||
Convergence achieved after 1 iteration | ||||
GDP=C(1)*(K^C(2))*(L^C(3)) | ||||
Variable | Coefficient | Std. Error | t-Statistic | Prob. |
C(1) | 4.014973 | 3.34E+17 | 1.20E-17 | 1.0000 |
C(2) | 1.942001 | 1.44E+15 | 1.35E-15 | 1.0000 |
C(3) | -4.547718 | 8.71E+15 | -5.22E-16 | 1.0000 |
R-squared | -1.858693 | Mean dependent var | 481.4144 | |
Adjusted R-squared | -2.078593 | S.D. dependent var | 359.35 | |
S.E. of regression | 630.5381 | Akaike info criterion | 15.82872 | |
Sum squared resid | 10337035 | Schwarz criterion | 15.97017 | |
Log likelihood | -226.5165 | Hannan-Quinn criter. | 15.87302 | |
Durbin-Watson stat | 0.008228 | |||
①回归系数的符号及数值是否合理;
分析以上两个模型的参数发现符号及数值是合理的
②模型的更改是否提高了拟合优度;
③模型中各个解释变量是否显著;
④残差分布情况 | ||||||||||||||
实验结果与分析: | ||||||||||||||
讨论与心得: | ||||||||||||||
成绩评定 | 评阅教师 | 评阅时间 |