1、 計量經(jīng)濟學實驗報告異方差（五）一、實驗內(nèi)容為了了解某社區(qū)居民的消費情況，現(xiàn)從該社區(qū)隨機抽取了35個家庭，取得了如下的一組數(shù)據(jù)，其中X為可支配收入，Y為消費性支出，試建立兩者適當?shù)幕貧w模型，然后檢驗該模型是否存在異方差，如果存在異方差，請設(shè)法消除異方差?？芍涫杖隭消費性支出Y1386.10991.85958.87723.42803.65641.091143.23904.842148.232106.02961.54900.841114.11913.081542.581120.301200.28881.47963.62890.942089.991712.491061.86952.361187.2

2、0845.292499.902348.212194.692036.461819.521306.841457.141214.63944.28751.32912.00786.791432.821000.25922.04710.301063.761010.86893.70852.141193.401132.181684.751463.391931.581677.381993.191403.991548.981190.10968.36791.05894.15858.101931.021512.941035.20807.162260.952240.592065.011529.551084.21972.2

3、7二、實驗目的1、掌握異方差出現(xiàn)的來源、后果、檢驗及修正的原理，以及相關(guān)的EViews軟件操作方法。2、建立工作文件，輸入數(shù)據(jù)3、利用最小二乘法建立方程，進行殘差的懷特（white）異方差檢驗；4、利用加權(quán)最小二乘法修正異方差，列出方程及主要統(tǒng)計量，檢驗是否消除了異方差性，列出檢驗結(jié)果。三、實驗步驟（簡要寫明實驗步驟）1、首先建立一個工作文件、建立回歸模型；2、由于這是截面數(shù)據(jù)，所以要檢驗是否存在異方差1）、可以從圖形的角度來檢驗，為此首先生成該殘差序列;2）、WHITE檢驗3、消除異方差，WLS估計法1-165.02202-54.7614130.4933684-36.755475273.60

4、696120.29197-2.7039233-175.27369-110.693810108.548311-68.299961282.8397313-135.2799U204.081715162865516-234.209217-5.2107951S-13.929051950.15348點擊resid得到殘差項IzN12723-22522993.81230.24341241350.95557486076614470.1517.311202Q30720.339122&3.111011732.72114664.834126S70707n13300一64u41649341526525.1S16548

5、53961727.1529儲194.01M19251&.S72203921742211241.5642219517.65點擊e2和x右鍵asgrop得到散點圖DependentVariable:L0G(E2)Method:LeastSquaresDate:O5J13J16Time:19:25Sample:135Includedobservations:3-5VariableCaefTicientStd.Errort-StatisticProb.C-20.742619.445974-2.19594400S52LOG兇4.0550711.311756工.Dgi工艾0.0040R-squared0.

6、224557Meandependentvar8.426411AdjjstedR-squared0.201059S.D.dependentvar2.894483S.E.afregressionZ587195Akaikeinfocriterion4.7B4471Sumsquaredresid220.8B81Schwarzcriterion4.883348Loglikelihood-1.9032-4Hannan-Quinncrite匚+.825151F-statistic9-556331Durtin-Watsonsbt2.155571ProbfF-statistiO0.004033genrel=ab

7、s(resid)ViewProclObjectlPrintName1Freeze1Estimate1Fore-cast1Stats1ResidsiHeteroskedastieityTet:WhiteAF-statistic11.63448Prob.F2,2)0.0002bsR-squared14.77206Prob.Clii-Square(2)0.0006SealedexplainedSS9.39B701Prob.Clii-Square(2)0.0091TestEquafiuiLDependentVariable:RESIDEMethod:LeastSquaresate:0W13J16Tim

8、e:19:34Sample:135Includedobservations:35VariableCoefficientStd.Errort-StatisticProt.c-2850.59339750.10-0.0717130.9433X0.+0467455.516470.0072S90.9942XA20.0103010.0-176300.5842840.5631R-squared0422059Meandependentvar205-99.49Acyust&dR-squared0.335933S.D.dependentvar25005.S9S.E.ofregression19594.75Akai

9、keinfocriterion22.63573S-umsquaredresid1.23E+10Schwarzcriterion22.81904Loglikelihood-394.0002Hannan-Quinncriter.22.73175F-statistic116B448Lirbin-WatsonstatZ353OT5ProbF-5tatistic0.000155Viivil11*.rr11V1-ll-11I-*.廠丿1II11亠1在上方輸入lsycX,點擊view中的White檢驗得到數(shù)據(jù)Sample:135Includledobservations:35Weightingseries:

10、1/XWeighttype:Standarddleviation(averagescalingVariableCoefficientStri.Errort-StatisticProb.C-227.5470109.4062-2.0798370.0454X0.9793491.0603716.226060.0000VYi叮ht胡StatisticsFl-squared0.8S8621Meandependentvar1253.226AdjutedR-squared0.8S524-6SLD.dependentvar954.4&49S.E.ofregression176.014-6Akaikeinfocriterion13.24253Sumsquaredresid103169ZSchwarzcritericn1333240Loglikelihood-229.7617Hannan-Quinncriter.13.27421F-statistic263.2S49Durbin-Watsonstat2.281674Prot(F-statiStic)0.000000Weightedmeandep.14B0.162UnweightedStatisticsFl-squared0.-8S7832Meandependentvar1176.