计量经济学作业文档格式.docx

1、265178270191230137250189第一次作业第一题：下列数据中，X表示家庭收入，Y表示家庭支出，请对如下数据运用戈德菲尔德-匡特检验。第二题：X代表职工的工龄，Y代表薪水。要求：1. 通过散点图或残差图对样本进行初步观察。2. 对可能存在的问题进行检验。3. 采取措施消除问题。4. 写出最终表达式。0.5690002.5705004.5740506.5826008.59143910.58312712.58470014.58260116.59328618.59040020.598200231000002699662301160123485200步骤：（1）、将样本数据排序，分组，剔

2、除中间样本 ，然后做OLS回归 第一组： Dependent Variable: YMethod: Least SquaresDate: 04/18/11 Time: 08:52Sample: 1 11Included observations: 11Y=C（1）+C（2）*XCoefficientStd. Errort-StatisticProb.C（1）12.5369510.538701.1896100.2646C（2）0.6059110.0952716.3598670.0001R-squared0.817990Mean dependent var78.63636Adjusted R-sq

3、uared0.797767S.D. dependent var12.87069S.E. of regression5.787989Akaike info criterion6.512413Sum squared resid301.5074Schwarz criterion6.584757Log likelihood-33.81827Hannan-Quinn criter.6.466810F-statistic40.44791Durbin-Watson stat2.272765Prob（F-statistic）0.000131第二组：54-36.6018740.67131-0.8999430.3

4、9160.8296260.1700964.8774050.00090.725518161.00000.69502021.4802211.862447.9475981266.4588.019942-41.711797.90199523.789081.1786800.000875（2）： F=1266.458/301.5074=4.2004给定a=5%查表，得临界值： F0.05（9,9）=2.97判断F大于F0.05（9,9）否定两组子样本方差相同的假设，从而该总体随机项存在异方差性。做散点图： 09:04 1 24 24-84.2480311.17191-7.5410600.00007.783

5、6110.62113412.531290.87711851.274380.87153238.3032813.728818.1565264146.5678.254697-95.878318.182571157.03330.6021730.000000由散点图可看出，该样本可能存在异方差性。作G-Q检验：将大样本分为两个小样本， 第二次上机先对全样本作整体回归：将样本分组，分别作OLS回归，15.8710.817.7311.919.0413.121.6114.917.218.8513.716.7111.314.8110.419.0214.220.5515.224.2917.124.3816.942

6、 1 12 121.1498011.2470100.9220460.37821.3349530.09122414.633830.95538719.171670.9509263.0639240.6787442.2138664.6069332.294684-11.283202.183944214.14911.23767227.843.0118.4665.6519.967.0520.388.2521.578.1520.991.48522.11110.323.8105.3323.189.9919.2119.2425.1114.2724.344-147.585028.43609-5.1900610.00

7、0410.831851.3256228.1711470.86973783.377080.85671028.4556610.771497.7426951160.2507.823513-44.456177.71277366.767651.7984630.000010RSS（R）=4146.567RSS（U）=RSS1+RSS2=4.606933+1160.250 =1164.856933 （4146.567-1164.856933）/12 F= = 4.2662035 1164.856933/（24-2*2）给定显著性水平5%，F0.05（12,20）=2.28F所以，不显著，拒绝原假设。第三次作

8、业对整体样本作OLS回归： 04/25/11 Time:27 1 30 30Y=C（1）+C（2）*X1+C（3）*X2+C（4）*X3+C（5）*X4+C（6）*X515.3836914.017301.0974790.28330.0207520.0128651.6131100.1198C（3）-0.0021780.133207-0.0163490.9871C（4）0.0343380.0135352.5370080.0181C（5）-0.0037900.004617-0.8209530.4198C（6）0.0117740.3982550.0295650.97670.94920045.63967

9、0.93861621.743525.3871136.382753696.50376.662992-89.741296.47240489.687760.983723用软件得到参差序列：Last updated: 04/25/11 - 08:32Modified: 1 30 / makeresid-2.387916742786327-3.626222161418159-6.831392318290824-2.9562125828705524.0820951169858878.7055956662165242.933711819027032-1.3743818562942562.8555907084

10、027792.5120149260854350.9061041373137081.2181295483755831.130*2603.0753*-4.706987305005896.010*-2.727309400155605-3.452771889540848-2.6047627777539557.683285351384584.0875*5.786371769840066-0.181*704897.4683729606522042.206659904875522-2.243012885258935-6.286316222062766-10.063598007776852.478021029

11、1062593.474021064943272 然后以et作为解释变量，做回归检验： U49Sample （adjusted）: 2 30 29 after adjustmentsU=C（1）*U（-1）0.5041170.1642263.0696440.00470.2515780.0823424.9663334.2964435.787326516.86375.834474-82.916235.8020921.72782451 3 30 28 after adjustmentsU=C（1）*U（-1）+C（2）*U（-2）0.6287380.1887293.3314250.0026-0.2686160.188790-1.4228230.16670.2991360.2147910.2721805.0050364.2699155.809814474.03665.904972-79.337405.8389052.047637由上可知，存在着一阶滞后相关性。然后，用广义差分法重新估计模型： YT-0.628738*YT（-1）21 6 30 25 after adjustmentsYT-0.628738*YT（-