还剩10页未读，继续阅读
本资源只提供10页预览，全部文档请下载后查看！喜欢就下载吧，查找使用更方便
文本内容:
回Equation:UNTITLEDWorkfile:UNTITLED::Untitled\_□XBreusch-GodfreySerialCorrelationLMTest:滞后期为2得以下结果Breusch-GodfreySerialCorrelationLMTest:TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/23/18Time:17:34Sample:19801997Includedobservations:18Presamplemissingvaluelaggedresidualssettozero.从上表可知，当滞后期为1时，〃R2=
0.404321probnR2=
0.524866当滞后期为2时h/2=
0.466047probnR2=
0.792135伴随概率均大于给定的显著性水平巾.05并且残差滞后期的回归系数的t统计量值绝对值均小于2这表明广义差分法估计的回归模型已消除高阶自相关性通过在LS命令中直接加上AR1AR2AR3项来检测模型的是否三阶自相关性@Equation:UNTITLEDWorkfile:UNnTLED::Untitled\-nXDependentVariable:YMethod:LeastSquaresDate:06/23/18Time:17:36Sampleadjusted:19811997Includedobservations:17afteradjustmentsConvergenceachievedafter7iterationsAR3回归系数的t检验不显著，表明模型确实不存在三阶自相关；上述检验表明，广义差分法估计的回归模型已消除自相关性，并且，经济意义合理，可决系数R2提高，t和F检验均显著，我们得到理想模型:Yt=
837.+
0.^+[AR1=
1.AR2=-
0.]边际分析模型表明国内生产总值x每增加一亿元，财政收入增加
0.亿元将其与OLS相比，OLS估计的常数项估计偏高，斜率估计偏低，且高估系数估计值的标准差试验结果:我国1978—1997年财政收入Y和国民生产总值GNPX的统计资料如表1所示单位亿元表格3⑴利用DW统计量，偏相关系数和BG检验，检测模型的自相关性；⑵通过在LS命令中直接加上AR1AR2项来检测模型的自相关性，并与1中的检验结果进行比较；⑶分析调整自相关性之后，模型估计结果的变化情况；答D利用DW统计量，偏相关系数和BG检验，检测模型的自相关性DW统计量一元线性回归模型估计IsycxDATAYXLSYCX[=]Equation:UNTITLEDWorkfile:UNTHLED::Untitled\_HXDependentVariable:YMethod:LeastSquaresDate:06/23/18Time:17:16Sample:19781997Includedobservations:20y=
858.4836+
0.100016%
67.
155770.02173t二
46.01678R=
0.991571F=
2117.544S.E=
208.675DW=
0.861307此模型的可决系数为
0.991571接近于1表明模型对样本拟合优度高；F统计量为
2117.544其伴随概率为
0.00000接近于零，表明模型整体线性关系显著，且回归系数均显著；对样本数n为20解释变量个数k为1若给定的显著性水平a=
0.05查DW统计表得，为=
1.201du=
1.411而0DW=
0.861307dL=
1.201这表明模型存在一阶正自相关偏相关系数检验方程窗口点击view\residualtest\correlogram-Q-statistics=Equation:UNTITLEDWorkfile:UNTnLED::Untitled\_nXViewProcObjectPrintNameFreezeEstimateForecastStatsResidsCorrelogramofResidualsDate:06/23/18Time:17:20Sample:19781997Includedobservations:20AutocorrelationPartialCorrelationACPACQ-StatProb从上图可知，所有滞后期的偏自相关系数PAC的直方图均在虚线内，表明回归模型不存在高阶自相关性BG检验方程窗口点击view\residualtest\serialCorrelationLMTest滞后期为1得以下结果=]Equation:UNTITLEDWorkfile:UNHTLED::Untitled\Breusch-GodfreySerialCorrelationLMTest:ViewProcObjectPrintNameFreezeEstimateForecastStatsResidsTestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/23/18Time:17:23Sample:19781997Includedobservations:20Presamplemissingvaluelaggedresidualssettozero.由上表可以看出，A*.132746〉/1=
3.84146probnR2=
0.013270小于给定的显著性水平a=
0.05并且e-回归系数的T统计量值绝对值均大于2回归系数显著不为零，表明模型存在一阶自相关性滞后期为2得以下结果FileEditObjectViewProcQuickOptionsAdd-insWindowHelpDATAYXLSYCXBreusch-GodfreySerialCorrelationLMTest:从上表可以看出，=
10.84049〉/2=
5.99147probnR=
0.004426小于给定的显著性水平a=
0.05并且ei和以一2回归系数的t统计量值绝对值均大于2回归系数显著不为零，表明模型存在一阶、二阶自相关性滞后期为3得以下结果DATAYXLSYCX国Equation:UNTITLEDWorkfile:UNTHLED::Untitled\-nXBreusch-GodfreySerialCorrelationLMTest:从上表可以看出，nR2=U.03133/口3=
7.81473probnR=
0.011558小于给定的显著性水平a=
0.05et-i回归系数的t统计量值绝对值均大于2回归系数显著不为零，但et-
2、e-回归系数的t统计量其P值均大于
0.05也大于
0.10回归系数显著地为零，表明模型可能存在三阶自相关上述检验表明模型存在一阶、二阶自相关，可能存在三阶自相关，0LS估计模型中的t统计量和F统计量的结论不可信，需应用广义差分法修正模型⑵通过在LS命令中直接加上AR1AR2项来检测模型的自相关性，并与⑴中的检验结果进行比较广义差分法估计模型LSYCXAR1AR2DATAYXLSYCXAR1AR2=]Equation:UNTITLEDWorkfile:UNUTLEDuUntitledX_nXDependentVariable:YMethod:LeastSquaresDate:06/23/18Time:17:31Sampleadjusted:19801997Includedobservations:18afteradjustmentsConvergenceachievedafter7iterationsR=
837.+
0.^+[AR1=
1.AR2=-
0.]R2=
0.995782F=
1101.6900probF=
0.000000DW=
2.013725输出结果显示AR1为
1.AR2为-
0.且回归系数的t检验显著，表明模型确实存在一阶、二阶自相关；调整后模型DW为
2.013725样本容量n为18个，解释变量个数k为1查5%显著水平DW统计表可得158du=
1.391而du=L391〈DW=2平137254-击这表明广义差分法估计的回归模型不存在一阶自相关偏相关系数检验广义差分法估计的模型DATAYXLSYCXAR1AR2=]Equation:UNTITLEDWorkfile:UNnTLED::Untitled\-nXViewProcObjectPrintNameFreezeEstimateForecastStatsResidsCorrelogramofResidualsDate:06/23/18Time:17:33Sample:19801997Includedobservations:18Q-statisticprobabilitiesadjustedfor2ARMAtermsAutocorrelationPartialCorrelationACPACQ-StatProb从上图可知，所有滞后期的偏自相关系数直方图均在虚线内，且Q统计量P值大于
0.05也表明了调整后回归模型不存在高阶自相关性BG检验广义差分法估计的模型滞后期为1得以下结果《计量经济学》上机实验报告六题目自相关实验日期和时间:
2018.531班级16金融学类5班学号姓名马雨柔实验室202实验环境WindowsXP;EViews
3.1实验目的掌握自相关检验及修正方法-广义差分法估计，熟悉EViews软件的相关应用实验内容利用实例数据和EViews软件，采用有关方法对建立的回归模型进行自相关的检验及处理第六章习题
6.
16.4实验步骤
一、建立工作义件.菜单方式.命令方式CREATEA起始期终止期
二、输入数据
三、自相关检验
①估计回归模型进行DW检验Isycx
②高阶自相关检验-偏相关系数检验方程窗口点击view\residualtest\correlogram-Q-statistiesPAC绝对值大于
0.5表明存在几阶自相关
③高阶自相关检验-BG检验布罗斯-戈弗雷检验或拉格朗日乘数检验方程窗口点击view\residualtest\serialCorrelationLMTest得到nR2给定a若nR2/P表明模型存在几阶自相关
四、采用广义差分法估计回归模型根据上述检验，若模型存在
一、二阶自相关则广义差分法估计回归模型命令LSYCXAR1AR2ViewProcObjectPrintNameFreezeEstimateForecastStatsResidsF-statisticObs*R-squared
0.
2987200.404321Prob.F113Prob.Chi-Squarel
0.
59390.5249TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/23/18Time:17:34Sample:19801997Includedobservations:18Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.C-
4.
379656102.5468-
0.
0427090.9666X
0.
0003610.
0041440.
0870900.9319AR
10.
2079420.
4927550.
4219990.6799AR2-
0.
1320090.388707-
0.
3396120.7396RESID-1-
0.
2624740.480235-
0.
5465520.5939R-squared
0.022462Meandependentvar
1.82E-08AdjustedR-squared-
0.278319S.D.dependentvar
144.8925S.E.ofregression
163.8194Akaikeinfocriterion
13.26554Sumsquaredresid
348878.4Schwarzcriterion
13.51286Loglikelihood-
114.3899Hannan-Quinncriter.
13.29964F-statistic
0.074680Durbin-Watsonstat
1.885849ProbF-statistic
0.988677@Equation:UNTITLEDWorkfile:UNTTFLED::Untitled\*_nxViewProcObjectPrintNameFreezeEstimateForecastStatsResidsF-statistic
0.159478Prob.F2f
120.8544Obs*R-squared
0.466047Prob.Chi-Square
20.7921VariableCoefficientStd.Errort-StatisticProb.C-
6.
538895107.0634-
0.
0610750.9523X
0.
0005910.
0044490.
1328080.8965AR⑴
0.
2942860.
6622670.
4443620.6647AR2-
0.
1430650.407434-
0.
3511360.7316RESID-1-
0.
3457110.642632-
0.
5379610.6004RESID-2-
0.
1064240.517789-
0.
2055350.8406ViewProcObject1PrintNameFreezeEstimateForecastStatsResidsVariableCoefficientStd.Errort-StatisticProb.C
846.
926699.
265728.
5319140.0000X
0.
1024650.
00417124.
564020.0000AR
11.
1501200.
3418503.
3643970.0056AR2-
0.
6626090.478437-
1.
3849440.1913AR3-
0.
1913740.415276•
0.
4608360.6532R-squared
0.995686Meandependentvar
3422.446AdjustedR-squared
0.994248S.D.dependentvar
2232.947S.E.ofregression
169.3462Akaikeinfocriterion
13.34170Sumsquaredresid
344137.6Schwarzcriterion
13.58676Loglikelihood-
108.4044Hannan-Quinncriter.
13.36606F-statistic
692.4502Durbin-Watsonstat
1.899493ProbF-statiStic
0.000000InvertedARRoots.68-.69I.68+69i-.20年份FINANCEGDP年份FINANCEGDP
19781132.
263624.
119882357.
2414922.
319791146.
384038.
219892664.
916917.
819801159.
934517.
819902937.
118598.
419811175.
794860.
319913149.
4821662.
519821212.
335301.
819923483.
3726651.
919831366.
955957.
419934348.
9534560.
519841642.
867206.
719945218.
14667019852004.
828989.
119956242.
257494.
919862122.
0110201.
419967404.
9966850.
519872199.
3511954.
519978651.
1473452.5ViewProcObjectIPrintNameFreezeIEstimateForecastStatsResidsVariableCoefficientStd.Errort-StatisticProb.C
858.
483667.
1557712.
783470.0000X
0.
1000160.
00217346.
016780.0000R-squared
0.991571Meandependentvar
3081.008AdjustedR-squared
0.991103S.D.dependentvar
2212.282S.E.ofregression
208.6715Akaikeinfocriterion
13.61404Sumsquaredresid
783788.0Schwarzcriterion
13.71361Loglikelihood-
134.1404Hannan-Quinnenter.
13.63348F-statistic
2117.544Durbin-Watsonstat
0.861307ProbF-statistic
0.000000F-statistic
7.518193Prob.F
1170.0139Obs*R-squared
6.132746Prob.Chi-Squarel
0.0133VariableCoefficientStd.Errort-StatisticProb.C-
31.
9210958.70662-
0.
5437390.5937X
0.
0021400.
0020191.
0600190.3040RESID-
10.
7011850.
2557272.
7419320.0139r目Equation:UNTITLEDWorkfile:UNTnLED::Untitled\-nxViewProcObjectPrintNameFreezeEstimateForecastStatsResidsF-statisticObs*R-squared
9.
46817810.84049Prob.F216Prob.Chi-Square
20.
00190.0044TestEquation:DependentVariable:RESIDFvlethod:LeastSquaresDate:06/23/18Time:17:27Sample:19781997Includedobservations:20Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.C-
23.
9874649.25821-
0.
4869740.6329X
0.
0017700.
0016971.
0432710.3123RESID-
11.
2425910.
2855504.
3515740.0005RESID-2-
0.
8075480.281604-
2.
8676760.0112R-squared
0.542024Meandependentvar
6.54E-13AdjustedR-squared
0.456154S.D.dependentvar
203.1059S.E.ofregression
149.7823Akaikeinfocriterion
13.03310Sumsquaredresid
358955.8Schwarzcriterion1323225Loglikelihood-
126.3310Hannan-Quinncriter.1307197F-statistic
6.312119Durbin-Watsonstat
2.036319ProbF-statistic
0.004971ViewProcObjectPrintNameFreezeEstimateForecastStatsResidsF-statisticObs*R-squared
6.
14993011.03133Prob.F315Prob.Chi-Square
30.
00610.0116TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/23/18Time:17:29Sample:19781997Includedobservations:20Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.C-
27.
3127150.68377-
0.
5388850.5979X
0.
0020200.
0017901.
1289220.2767RESID-
11.
1909400.
3058113.
8943610.0014RESID-2-
0.
6276370.429223-
1.
4622630.1643RESID-3-
0.
2099550.371623-
0.
5649690.5804R-squared
0.551567Meandependentvar
6.54E-13AdjustedR-squared
0.431984S.D.dependentvar
203.1059S.E.ofregression
153.0744Akaikeinfocriterion
13.11204Sumsquaredresid
351476.6Schwarzcriterion
13.36098Loglikelihood-
126.1204Hannan-Quinncriter.
13.16064F-statistic
4.612448Durbin-Watsonstat
1.913584ProbF-statistic
0.012544ViewProcObjectPrintNameFreezeEstimateForecastStatsResidsVariableCoefficientStd.Errort-StatisticProb.C
837.
367099.
645638.
4034490.0000X
0.
1025790.
00398825.
723340.0000AR
11.
2163740.
3052043.
9854460.0014AR2-
0.
8261830.296827-
2.
7833840.0147R-squared
0.995782Meandependentvar
3296.751AdjustedR-squared
0.994878S.D.dependentvar
2230.951S.E.ofregression
159.6637Akaikeinfocriterion
13.17715Sumsquaredresid
356895.0Schwarzcriterion
13.37501Loglikelihood-
114.5943Hannan-Quinncriter.
13.20443F-statistic
1101.690Durbin-Watsonstat
2.013725ProbF-statistic
0.000000InvertedARRoots.61+
681.61-.68i111।।1-
0.086-
0.
0860.1551111।1।
20.
0350.
0280.18221匚1।匚।3-
0.212-
0.
2081.
25960.2621□1।]।
40.
1260.
0961.
66700.4351匚1।匚।5-
0.285-
0.
2773.
91200.271111।匚।6-
0.101-
0.
1994.
21890.3771匚1।匚।7-
0.122-
0.
1224.
70670.45313।।[।
80.109-
0.
0555.
13390.5271□।匚।9-
0.143-
0.
1945.
94740.5461।।匚।10-
0.002-
0.
1785.
94770.6531।।匚।11-
0.017-
0.
1585.
96260.7441□।।[।
120.148-
0.
0907.
27840.699。
个人认证
优秀文档
获得点赞 0
