计量经济学报告

实 验 报 告

课程名称： 计量经济学 任课教师： 班 级： 姓 名： 实验日期 学 号：

实验项目名称: 专门问题模型 一、实验目的及要求 1.使学生了解常用计量经济学软件的基本应用，熟悉在信息化条件下的经济学中数据处理的一般流程和方法，增强对经济数据计量化的感性认识。 2.使学生熟练开发工具和平台的使用，增强实践动手能力，提高信息技术的应用能力。 3.通过统计检验加深对放宽基本假定的模型的了解。 二、实验环境 1．系统软件：Eviews 3.0 2．工具：Eviews 3.0 三、实验内容与步骤 （1）虚拟变量模型 1、统计第一产业当月固定资产投资额，并作出折线图，分析折线图找出转折点，与题目所给的进行比较分析。 ①、1）对Y DD C 回归： Variable DD C Coefficient -26.91095 79.11095 Std. Error 12.98339 7.111294 t-Statistic -2.072722 11.12469 Prob. 0.0475 0.0000 第 1 页 共 11 页

R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.133024 Mean dependent var 0.102061 S.D. dependent var 32.58804 Akaike info criterion 29735.45 Schwarz criterion -146.0516 F-statistic 1.774765 Prob(F-statistic) 71.03767 34.39021 9.870108 9.963521 4.296177 0.047518 由p=0.0475，通过显著性检验，接受此点是转折点的检验。但R20.102061，统计结果的解释力度不大。加变量时间T，2） 对Y T DD C 进行回归。结果如下： Dependent Variable: Y Method: Least Squares Date: 11/30/13 Time: 20:14 Sample: 1 30 Included observations: 30 Variable Coefficient T 2.634096 DD 12.60049 C 26.42902 R-squared 0.300933 Adjusted R-squared 0.249151 S.E. of regression 29.79966 Sum squared resid 23976.53 Log likelihood -142.8227 Durbin-Watson stat 2.153268 Std. Error t-Statistic 1.034361 2.546592 19.53673 0.49 21.68520 1.218759 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 2Prob. 0.0169 0.5244 0.2335 71.03767 34.39021 9.721510 9.861630 5.811461 0.007962 加入时间变量T后，虽然整体通过检验，R=0.249也有增大，但是T和DD两变量都没有通过检验。 3）乘法方式引入虚拟变量，回归结果如下： Y DD*T T C Variable DD*T T C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 2.900845 2.636700 25.81755 Std. Error 2.459209 0.765024 15.31059 t-Statistic 1.179585 3.446558 1.686254 Prob. 0.2485 0.0019 0.1033 71.03767 34.39021 9.686549 9.826669 6.498549 0.004966 0.324951 Mean dependent var 0.274947 S.D. dependent var 29.28327 Akaike info criterion 23152.77 Schwarz criterion -142.2982 F-statistic 2.242692 Prob(F-statistic) 2从回归结果来看，整体通过检验，但DD*T未能通过检验，R=0.2749，比加法引入时得解释力度要大。 4）乘法加法方式引入，Y DD*T DD T C 回归结果如下： Variable DD*T DD T C R-squared Adjusted R-squared S.E. of regression Coefficient 4.184929 -12.86202 2.331571 32.47952 Std. Error 3.986677 31.12259 1.071884 22.39869 t-Statistic 1.049728 -0.413270 2.175209 1.450063 Prob. 0.3035 0.6828 0.03 0.1590 0.329356 Mean dependent var 71.03767 0.251974 S.D. dependent var 34.39021 29.74357 Akaike info criterion 9.746669 第 2 页 共 11 页

Sum squared resid Log likelihood Durbin-Watson stat 23001.67 Schwarz criterion -142.2000 F-statistic 2.255636 Prob(F-statistic) 9.933495 4.256243 0.014245 由回归结果看来，乘法加法引入，整体未能通过检验。因此，判断得出加法引入变量最合适。 ②、对加法引入虚拟变量的回归结果进行序列相关性检验： Breusch-Godfrey Serial Correlation LM Test: F-statistic Obs*R-squared 0.073869 Probability 0.169504 Probability 0.9287 0.918740 Test Equation: Dependent Variable: RESID Method: Least Squares Date: 12/03/13 Time: 20:01 Variable DD C RESID(-1) RESID(-2) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 0.115019 0.122639 0.059943 0.0419 Std. Error 13.45007 7.365911 0.200884 0.201954 t-Statistic 0.008552 0.016650 0.298397 0.229850 Prob. 0.9932 0.9868 0.7678 0.8200 1.E-15 32.02125 9.997775 10.18460 0.049246 0.985211 0.005650 Mean dependent var -0.109083 S.D. dependent var 33.72253 Akaike info criterion 29567.44 Schwarz criterion -145.9666 F-statistic 1.9406 Prob(F-statistic) 由检验结果可以看出，一阶p=0.7678，接受原假设，因此加法引入虚拟变量不存在序列相关性。 2、统计第二产业当月固定资产投资额，并作出折线图， 从折线图看出，整体呈现上升趋势。 ①、1） 作Y DD C回归分析： 第 3 页 共 11 页

Dependent Variable: Y Method: Least Squares Date: 11/30/13 Time: 20:26 Sample: 1 30 Included observations: 30 Variable Coefficient DD -947.1378 C 2705.617 R-squared 0.313960 Adjusted R-squared 0.2458 S.E. of regression 6.1137 Sum squared resid 12349317 Log likelihood -236.4869 Durbin-Watson stat 1.883507 Std. Error t-Statistic 2.52 -3.579653 144.9215 18.66953 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) 2Prob. 0.0013 0.0000 2421.475 787.8574 15.912 15.99254 12.81392 0.001281 从上述回归结果看，整体结果较显著地通过检验，R=0.2458，解释力度较小，受时间变量的影响，加入时间T，再进行回归。 2）加入时间T，加法引入虚拟变量回归分析： Variable DD T C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 90.73748 73.83680 1270.879 Std. Error 406.7183 20.14632 502.1785 t-Statistic 0.223097 3.665026 2.530731 Prob. 0.8249 0.0009 0.0167 2597.716 922.0611 15.78658 15.92126 20.80570 0.000002 0.573070 Mean dependent var 0.545526 S.D. dependent var 621.6047 Akaike info criterion 11978163 Schwarz criterion -265.3719 F-statistic 2.296929 Prob(F-statistic) 2由结果得出，整体通过检验，但DD未能通过检验，R=0.545526，有增大。 3）乘法引入虚拟变量，回归如下： Variable DD*T T C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 2.531857 70.65046 1354.557 Std. Error 31.39310 13.121 305.5052 t-Statistic 0.080650 5.366858 4.433825 Prob. 0.9362 0.0000 0.0001 2597.716 922.0611 15.78798 15.92266 20.75511 0.000002 0.572474 Mean dependent var 0.5442 S.D. dependent var 622.0382 Akaike info criterion 11994878 Schwarz criterion -265.3956 F-statistic 2.293487 Prob(F-statistic) 2 从结果看来，整体通过检验，DD*T仍为能通过检验，且R=0.5442，较上述结果下降。 4）加法乘法同时引入虚拟变量，回归如下： Variable DD*T DD T Coefficient -8.005230 172.7911 75.36721 Std. Error 52.05960 674.9369 22.76240 t-Statistic -0.153770 0.256011 3.311040 Prob. 0.8788 0.7997 0.0024 第 4 页 共 11 页

C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 1234.149 563.4173 2.190470 0.03 2597.716 922.0611 15.84462 16.02419 13.44149 0.000010 0.573406 Mean dependent var 0.530747 S.D. dependent var 631.6310 Akaike info criterion 11968730 Schwarz criterion -265.3585 F-statistic 2.298528 Prob(F-statistic) 整体通过检验，但变量DD*T，DD均没有通过检验，R2=0.530747，解释力度下降，因此得出，加法引入虚拟变量最合适. ②、对加法引入虚拟变量的回归结果进行序列相关性检验： Breusch-Godfrey Serial Correlation LM Test: F-statistic Obs*R-squared 0.366609 Probability 0.811156 Probability 0.696142 0.666591 Test Equation: Dependent Variable: RESID Method: Least Squares Date: 12/03/13 Time: 19:41 Variable DD C RESID(-1) RESID(-2) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 5.635720 2.2147 0.127263 0.077125 Std. Error 2.0081 163.1412 0.185416 0.185615 t-Statistic 0.021347 0.016195 0.686365 0.415510 Prob. 0.9831 0.9872 0.4978 0.6807 2.88E-13 721.2856 16.18124 16.36081 0.244406 0.8638 0.023858 Mean dependent var -0.073757 S.D. dependent var 747.4122 Akaike info criterion 16758752 Schwarz criterion -271.0811 F-statistic 1.965824 Prob(F-statistic) 由结果看，不存在序列相关。 3、统计第三产业当月固定资产投资额，并作出折线图， 第 5 页 共 11 页

1） 作Y DD C 回归分析，结果如下： Variable DD C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient -1332.413 4117.958 Std. Error 396.3316 245.0704 t-Statistic -3.361866 16.80316 Prob. 0.0020 0.0000 3608.506 1286.466 16.94251 17.03230 11.30214 0.002018 0.261007 Mean dependent var 0.237913 S.D. dependent var 1123.054 Akaike info criterion 40359988 Schwarz criterion -286.0227 F-statistic 1.936929 Prob(F-statistic) 2从结果看，整体通过检验，且DD也通过检验，R=0.237913，解释力度较小，加入时间变量T，再回归。 2）Y DD T C Variable DD T C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 4.197056 78.62415 2230.979 Std. Error 690.0022 34.17846 851.9515 t-Statistic 0.006083 2.300400 2.618669 Prob. 0.9952 0.0283 0.0135 3608.506 1286.466 16.84373 16.97841 9.054913 0.000800 0.368762 Mean dependent var 0.328037 S.D. dependent var 1054.559 Akaike info criterion 34474955 Schwarz criterion -283.3434 F-statistic 2.221291 Prob(F-statistic) 2从结果看,整体通过检验,但DD没有通过检验, R=0.328037,有所增大。 2） 乘法引入虚拟变量，回归结果如下： Variable DD*T T C R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient 13.76628 81.70293 2141.860 Std. Error 53.120 22.29359 517.3729 t-Statistic 0.2539 3.6863 4.139875 Prob. 0.7974 0.0009 0.0002 3608.506 1286.466 16.84157 16.97625 9.107993 0.000773 0.370123 Mean dependent var 0.329486 S.D. dependent var 1053.421 Akaike info criterion 34400592 Schwarz criterion -283.3067 F-statistic 2.2280 Prob(F-statistic) 2整体通过检验，但DD*T没有通过检验，R=0.329486，增大。 3） 乘法加法引入虚拟变量，回归如下： Variable DD*T DD T C R-squared Coefficient 36.03351 -365.14 71.73539 2396.309 Std. Error 88.10923 1142.309 38.524 953.5660 t-Statistic 0.40 -0.319656 1.862065 2.512997 Prob. 0.6855 0.7514 0.0724 0.0176 0.372261 Mean dependent var 3608.506 第 6 页 共 11 页

Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat 0.309488 S.D. dependent var 1069.015 Akaike info criterion 34283821 Schwarz criterion -283.24 F-statistic 2.235563 Prob(F-statistic) 21286.466 16.699 17.07657 5.930199 0.002658 整体通过检验，但DD，DD*T没有通过检验，且R=0.309488减小。因此，加法引入虚拟变量最合适。 ② 对加法引入虚拟变量进行序列相关性检验： Breusch-Godfrey Serial Correlation LM Test: F-statistic Obs*R-squared 0.379369 Probability 0.838690 Probability 0.687529 0.657477 Test Equation: Dependent Variable: RESID Method: Least Squares Date: 12/03/13 Time: 19:55 Variable DD C RESID(-1) RESID(-2) R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Coefficient -8.502311 -4.535801 0.012278 -0.160340 Std. Error 404.3674 250.1269 0.1837 0.184690 t-Statistic -0.021026 -0.018134 0.0667 -0.868156 Prob. 0.9834 0.9857 0.9472 0.3922 3.34E-13 1105.907 17.03518 17.21476 0.252912 0.858630 0.024667 Mean dependent var -0.072866 S.D. dependent var 1145.490 Akaike info criterion 393413 Schwarz criterion -285.5981 F-statistic 1.946608 Prob(F-statistic) 接受原假设，不存在相关性。 （2）滞后变量模型 统计某地区1970-1991年固定资产投资（y）与销售额（x）的数据，建立分布滞后模型来考察两者的关系。 阿尔蒙变换 Lny pdl（lnx，2,2）c Dependent Variable: LNY Method: Least Squares Date: 11/30/13 Time: 20:57 Sample(adjusted): 1972 1991 Included observations: 20 after adjusting endpoints Variable Coefficient Std. Error t-Statistic C -1.527878 0.190656 -8.013797 PDL01 0.940123 0.342860 2.742000 PDL02 -0.558403 0.174528 -3.199507 PDL03 -0.794286 0.517184 -1.5357 R-squared 0.993491 Mean dependent var Adjusted R-squared 0.992270 S.D. dependent var S.E. of regression 0.045955 Akaike info criterion Sum squared resid 0.0337 Schwarz criterion Log likelihood 35.45471 F-statistic Durbin-Watson stat 1.209248 Prob(F-statistic) Lag Distribution i CoefficienStd. Error 第 7 页 共 11 页

Prob. 0.0000 0.0145 0.0056 0.1441 4.612088 0.522698 -3.145471 -2.946324 814.0290 0.000000 T-Statistic

of LNX . * | . *| * . | t 0 0.70424 1 0.94012 2 -0.41257 Sum of 1.23180 Lags 0.26227 0.34286 0.23087 0.03478 2.68515 2.74200 -1.78697 35.4213 Pdl03 没有通过检验。 Lny pdl（lnx，9,2）c Dependent Variable: LOG(Y) Method: Least Squares Date: 12/03/13 Time: 20:13 Sample(adjusted): 1979 1991 Included observations: 13 after adjusting endpoints Variable C PDL01 PDL02 PDL03 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Lag Distribution of LOG(X) . *| . * | . * | . * | *. | * . | * . | * . | * . | * | Coefficient Std. Error -4.332286 -0.048096 -0.165252 0.035660 0.961115 0.948154 0.047763 0.020532 23.48359 1.732188 1.503392 0.070412 0.037429 0.009537 t-Statistic -2.881673 -0.683063 -4.415038 3.739097 Prob. 0.0181 0.5118 0.0017 0.0046 4.949667 0.209765 -2.997475 -2.8235 74.15129 0.000001 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) i Coefficient 0 1 2 3 4 5 6 7 8 9 1.18348 0.76860 0.42505 0.15282 -0.04810 -0.17769 -0.23596 -0.22291 -0.13854 0.01715 Std. Error T-Statistic 0.24311 0.15938 0.10087 0.07301 0.07041 0.07299 0.06995 0.06551 0.07965 0.12671 0.26519 4.86798 4.82252 4.21381 2.09300 -0.68306 -2.43457 -3.37349 -3.40271 -1.73929 0.13536 6.50055 Sum of 1.72391 Lags Lny pdl（lnx，10,2）c Dependent Variable: LOG(Y) Method: Least Squares Date: 12/03/13 Time: 20:14 Sample(adjusted): 1980 1991 Included observations: 12 after adjusting endpoints Variable C PDL01 PDL02 PDL03 R-squared Adjusted R-squared Coefficient Std. Error -4.583732 -0.140233 -0.108181 0.030103 2.701115 0.072980 0.043423 0.007995 t-Statistic -1.696978 -1.921539 -2.491312 3.765347 Prob. 0.1281 0.0909 0.0374 0.0055 4.982535 0.180776 0.938713 Mean dependent var 0.915730 S.D. dependent var 第 8 页 共 11 页

S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Lag Distribution of LOG(X) . *| . * | . * | . * | * | * . | * . | * . | * . | *. | .* | 0.052478 0.022032 20.77384 1.685437 Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) -2.7950 -2.634005 40.84431 0.000034 i Coefficient 0 1 2 3 4 5 6 7 8 9 10 1.15324 0.77413 0.45523 0.19654 -0.00195 -0.14023 -0.21831 -0.23618 -0.19385 -0.09131 0.07143 Std. Error T-Statistic 0.33121 0.24181 0.16994 0.11704 0.08506 0.07298 0.07313 0.07880 0.09139 0.11726 0.16099 0.476 3.48191 3.20143 2.67875 1.67918 -0.02293 -1.92154 -2.98517 -2.99726 -2.12120 -0.77868 0.44371 3.71225 Sum of 1.76874 Lags 2 各参数都通过了检验，且R=0.915解释力度较大。 Lny pdl（lnx，11,2）c Dependent Variable: LOG(Y) Method: Least Squares Date: 12/03/13 Time: 20:15 Sample(adjusted): 1981 1991 Included observations: 11 after adjusting endpoints Variable C PDL01 PDL02 PDL03 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelihood Durbin-Watson stat Lag Distribution of LOG(X) . *| . * | . * | . * | * | *. | * . | * . | * . | * . | * | . * | Coefficient Std. Error -4.472823 -0.112009 -0.111220 0.025792 0.930121 0.900173 0.053477 0.020019 19.09114 1.674927 6.046963 0.104376 0.080075 0.007361 t-Statistic -0.739681 -1.073128 -1.3846 3.503918 Prob. 0.4836 0.3188 0.2074 0.0099 5.006053 0.169256 -2.743844 -2.599154 31.05785 0.000204 Mean dependent var S.D. dependent var Akaike info criterion Schwarz criterion F-statistic Prob(F-statistic) i Coefficient 0 1 2 3 4 5 6 7 8 9 10 11 1.08888 0.74554 0.45378 0.21360 0.02500 -0.11201 -0.19744 -0.23128 -0.21354 -0.14422 -0.02332 0.14917 Std. Error T-Statistic 0.59288 0.466 0.35442 0.25553 0.17095 0.10438 0.07145 0.020 0.12979 0.17456 0.22098 0.27060 1.071 1.83660 1.59682 1.28032 0.835 0.14626 -1.07313 -2.76314 -2.59290 -1.534 -0.82623 -0.10553 0.55124 1.754 Sum of 1.75415 第 9 页 共 11 页

Lags 因此，在滞后10的时候，结果通过检验，回归方程LOG(Y) = -4.583731944 + 1.1532385*LOG(X) + 0.7741331369*LOG(X(-1)) + 0.4552332292*LOG(X(-2)) + 0.196538841*LOG(X(-3)) - 0.0019500277*LOG(X(-4)) - 0.1402333769*LOG(X(-5)) - 0.2183112067*LOG(X(-6)) - 0.2361835169*LOG(X(-7)) - 0.1938503077*LOG(X(-8)) - 0.091311575*LOG(X(-9)) + 0.07143266925*LOG(X(-10)) 从表中可以看出，去年的影响对今年的投资最大，而且周期为十年，十年前的消费对今年的投资也有影响。 四、总结 这次上机实验做了两个模型的实验，这俩模型，与实际联系紧密，在操作中，也发现了不少问题，但最终还是了解了它出现的原因。实践让知识掌握的更好。 五、附录 评语： 评分： 评阅人： 日期： 第 10 页 共 11 页

第 11 页 共 11 页