EVIEWS操作各种模型学习.docx

1、EVIEWS操作各种模型学习章、图形基础与回归原始数据URURIRURURIR19780.152.9719970.312.2219790.162.3219980.322.2219800.172.4319990.342.3519810.172.0920000.402.4319820.181.6820010.412.4819830.181.7020020.412.7719840.181.5020030.422.8519850.231.6720040.432.7719860.241.9120050.432.8319870.252.0620060.442.8619880.272.2620070.442

2、.8719890.282.2020080.452.8219900.292.1220090.462.8519910.262.5420100.502.8919920.292.7620110.522.8419930.313.1120120.542.9119940.282.8520130.552.9119950.313.4820140.552.8319960.342.33，、散点图R Rl u-eEONo:、分布图：JB 3判断为正太分布Series: URSample 1 37Observati ons 37Mea n0.336145Media n0.312400Maximum0.547000Min

3、imum0.15087-Std. Dev.0.120626Skew ness0.143256Kurtosis1.887565Jarque-Bera2.034384Probability0.361609三、UR的单因素联表Tabulation of URDate: 09/05/15 Time: 21:25Sample: 1 37Included observations: 37Number of categories: 5ValueCountPercentCumulativeCountCumulativePercent0.1,0.2)718.92718.920.2, 0.3)924.321643

4、.240.3, 0.4)616.222259.460.4, 0.5)1129.733389.190.5, 0.6)410.8137100.00Total37100.0037100.00四、协方差与相关性Covariance Analysis: OrdinaryDate: 09/05/15 Time: 21:40Sample: 1 37Included observations: 37CovarianceCorrelationURURIRUR0.0141571.000000URIR0.0331700.2048520.6159341.000000Date: 09/05/15 Time: 21:44

5、Sample: 1 37Included observations: 37Correlations are asymptotically consistent approximationsURrURlR(-i)UR,URIR(+ilagleadO i00.61590.&15910.57670.654920.49970.651530.42820.638440.37950.575550.36210.499060.32670.415970.32920.291880.29830.173290.25850.0795100.19970.0124110.1420-0.055612 0809-0.099813

6、0.0097-0.153814-0.02310.174015-0.033B-0.179016-0.0565*0.140117-0.0433-0.123818-0.0316-0.053219-0.0395-0.093920-0.1629-0.1381五、CDF经验分布图六、Q - Q图URIRQua ntiles of URIR七、回归散点图UR邻近拟合散点图：（分布回归的结果）八、实际值、拟合值、残差值折线图Residual Actual Fitted9 Stan dardized Residuals九、回归模型预测URF 2 S.EForecast: URFActual: URForecas

7、t sample: 1 37In cluded observati ons: 37Root Mea n Squared Error 0.093736Mea n Absolute Error 0.072711Mea n Abs. Perce nt Error 25.93140Theil In equality Coefficie nt0.133790Bias Proportio n 0.000000Varia nee Proportio n 0.237674Covaria nee Proportion 0.762326十、两回归系数的联合检验置信区间是一个椭圆区域C(1)!一、Wald系数约束条

8、件检验Wald Test:Equation: UntitledTest StatisticValuedfProbabilityF-statistic Chi-square272.1503272.1503(1. 35)10.00000.0000Null Hypothesis Summary:Normalized Restriction = Q)ValueStd. Err,-1 + C(1) + C(2)-0,9075270.055012Restrictions are linear in cgeAI匚ientsChow分割点检验结果Chow Breakpoint Test 1991Null Hy

9、pothesis: No breaks at specified breakpointsVarying regressors: All equation variablesEquation Sample: 1978 2014FStatistic11.00551Prob. F(2r33)0.0002Log likelihood ratio18.90797Prob,Chi-Square(2)0.0001Wald Statistic22.01103Prob.Chi-Squa(e(2)0.0000F、LR的P值显著，表示：模型无显著的结构变化十二、Chow稳定性检验（p75）Chow预测结果:Chow

10、 Forecast Test: Forecast from 1991 to 2014F-statisticLog likelihood ratio4.74478289.83840PfOb. F(24,11)Prob. Chi-Square(24)0.00500.0000Test Equation:Dependent Variable: UR Method: Least SquaresDate: 09/07/15 Time: 10:31Sample: 197S 1990Included! observations: 13CoefficientStd. Error t-StatisticProb.

11、URIRC-0 0147160.2412890.037436 -03930940.078775 3.063016070180.0100R-squaredAdjusted R-squaredS.E of regressionSum squared resid Log likelihood F-statisticProb(F-statistic)0.013853*0.0757970.0510240.02863721.320730.1545230.701762Lie an dependent varS.D dependent var .Akaike info criterionSchwarz cri

12、terion Hannan-Ouinn criter.Durbin-Watson stat0.2108270.0491932972421 -2.885505 -29902860.127458十三、零均值附近的递归残差曲线图22递归OLS的CUSUM检验曲线图（p78）CUSUM 5% Significance*注：红线为5%的临界值线，在1991年后的CUSUM 曲线变得十分陡峭，说明：回归方程系数并不是稳定 的。3.步预测检验:N-Step Probability Recursiv e ResidualsOne-StepProbabilityRecursive Residuals*注：上部分

13、是递归残差，下部分是检验显著性的概率.3 Recursiv e C Estimates 士 2 S.E.Recursiv e C(2) Estimates 士 2 S.E.十四、White异方差检验Obs*R-squared=10.4,其P值=0.0055表示残差存在异 方差性。F统计量表示：检验辅助方程的整体显著性， 下图中整体显著。Heteroskedasticity Test: WhiteF-statistic6.658663Prob. F(2.34)0.0036Obs*R-squared10.41354Prob Chi-Square(2)0.0055ScaieC explained S

14、S10,36651Prob. ChkSquare2)0,0056Test Equation:Dependent VariaDle: RESIDA2 Method: Least SquaresDate: 09/07/15 Time: 11:02Sample: 1978 2014Irt匚luded observations: 37Coefficient Std. Error t-Statistic Prob.C0.0398190.0458S6 0.8677790.3916URIR-0.0428590 038582 -1 1108630.2744URIRA20.0117800.007941 1.48

15、35080.1472R-squared0.281447Mean dependentvar0.008786Adjusted R-squared0 239179S.D. dependent var0 013287S.E of regression0.011590Akaike info criterion-5 999002Sum squared resid0.004567匚hwam criterion-5.869187Log likelihood113 9963Hannan*Quinn enter.-5 953754F-statistic6.658663Durbin-Watson stat0.374

16、222Prob(F-statistic)0.003629十五、WLS加权最小二乘法Dependent Variable: URMethod: Least Squares Date: 09/07/15 Time: 11:29Sample: 1978 2014 included observations: 37Weighting series: WCoefficientStd. Errort-StatisticProb.URIR0.1685570.0311395,4130330.0000C-0.0860690.076542-1.1244650.2685Weighted StatisticsR-sq

17、uared0.455684(Jean dependentvar0.328603Adjusted R-squared0 440132S.D dependent var0.100794S E, of regression0.090040Akaike inla criterion*1.924588Sum squared resid0.283752Schwarz criterion-1,837512Log liKelitiaod37.50488Hannan-Quinn criter-1 893890F-statistic29.30092Durbin-Watson stat0.279531Prob(F-

18、statistic)0.000005Unweighted Statisti匚sR-squared 378738Mean dependent var0.33614SAdjusted R-squared0.350987S D. dependent var0.120526S.E. of regression0.096427Sum squared resid0.325433Durbin-Watson stat0.310271十六、残差自相关图及其 Q检验统计量Correlogram of ResidualsDate: 09/07/15 Time: 15:50Sample: 1978 2014Inclu

19、ded observations: 37Autocorrelation Partial Correlation AC PAC Q-Stat Prob11 111 110.712 0.712 20.333 0 00020 573 0 133 33 850 0 0001二1匚II3 0.380 -0 142 39.963 0.00011 14 0.239 -0 059 42.452 0.000II11 (15 0.111 -0.055 43.006 0 000II】H16 0.060 0.052 43.176 0.0001H17 0.053 0.078 43.310 0.000IIH1& 0.06

20、8 0.042 43.539 0.000II1 (II9 0.056 -0.051 43.698 0.00011 1110 0.034 0.050 44.075 0 00011 11111 0.11S 0 082 44.343 0 0001 111 112 0.157 0072 46.273 0 0001n1:113 0.217 0 114 49.100 0.000IIZI |1 114 0.203 -0.067 51.634 0.0001 1匚115 0 148 -0.119 53.114 0.000111匚115 0 024 -0.171 53.152 0.0001-16阶的p值都小于0.

21、01,说明拒绝原假设，残差序 列存在自相关性。十七、残差自相关LM检验结果Breuscn-Gocrrrey serial CorrslatJon LM Tst:F-statisticOtl*R-. 口1_1自耳N石4 640尹Rrot F2,33)Pro. Chi-Scilj岂D OOOO0.0000Test E q li ation:Depenitjent Varia&le RS SIOMetnodi i_east SquaresDate: 09/07/1 5 Time: 1Sam pile: 1 9ZB 201lindudied olbservations. 37PreS a innp0

22、& ll u e 1 日 口 些旦日孕导iCli nH导 set to ziroC o effi ci e ntStd. E rrort-Statisti 匚Pro ti.URIR oRESIDC-1 )RESIDC-2)-OO49356 O -1 2B67O O.63&8SS OJ284L8O90.02:33 1 8 O O日口日64 Q.1 6S1 Z3Ck 4374-2.07242122 40IS日3.92OB261 633 32J20.046 1 O 04-2 0.0004 O-1 1 il 9R-SQuareclAdju sted R-squ arecdl S.E. of reore

23、sslion S Li m s cjuore cl resi dl Loo 1 11h(O6 1 S958O S810450.061 509O. 121852S2.7926064271O.DDOOO1Mean dependent var S. D dependent var AKalKe inTo criterion Schwar-z. criterionMninn-Ouinn： crirtier Durbi n -Wats o nO 0-95029 -2.637-453 -24I333OO -2.S76OS61 2O4229F与Obs两个的P值显示：存在自相关十八、Newey-West致协方

24、差估计Dependentvariable: URMethod: Least SquaresDate: 09/07/15 Time: 1635Sample: 1978 2014Included observations: 37NeweyAVestHAC Standard Errors & Covariance (lag truncation=3)CoefficientStdL Error t-StatisticProb.URIR0.1619220.050813 3,1866200.0030C-0.0694490.109907 -0.6318370.5316R-squared0.379375Mea

25、n dependentvar0.336145Adjusted R-squared0.361643S.D dependent var0.120626S.E. of regression0.096377Aka ike info criterion-1.788557Sum squared resid0.325099Schwarz criterion-1,701481Log likelihood35.08831Hannan-Quinn criter.*1.757959F-statistic2139474Durbin-Watson stat0.289676Pro b(F-stati stic)0.000

26、049十九、两阶段TSLS估计检验结果Dep endent Variable: UR lethod: Two-Stage Least SquaresDate: 09/07/15 Time: 16:43Sample: 1978 2014Induded observations: 37Instrument list CURCoefficientStd. Error t-StatisticProb.URIR0.4268130.092275 4.6254450.0000C*07329650.232564 -3.1516750.0033R-squared-0.635916Mean dependent v

27、ar0.336145Adjusted R-squared-0.682657S.D. dependent var0.120626S.E, of regression0.156473Sum squared redid0 856934F-statistic2139474Durbin-Watson stat0.632913Prob(F-statistic)0.000049SecontTStage SSR4.14E*30二十、广义矩估计GMM检验结果toependentVariable: URMethod: Generalized Method of MomentsDate: 09/07/15 Time: 16:51Sample: 1978 2014Included observations: 37