計(jì)量經(jīng)濟(jì)學(xué)實(shí)驗(yàn)報(bào)告

中國(guó)海洋大學(xué)《計(jì)量經(jīng)濟(jì)學(xué)》實(shí)驗(yàn)報(bào)告實(shí)驗(yàn)項(xiàng)目名稱(chēng)：能源消費(fèi)總量的影響因素分析指導(dǎo)教師：姓名：學(xué)號(hào)：年級(jí)專(zhuān)業(yè)：2011級(jí)金融【實(shí)驗(yàn)?zāi)康摹?．通過(guò)實(shí)驗(yàn)課使自己能夠了解并深入認(rèn)識(shí)什么是計(jì)量經(jīng)濟(jì)學(xué)，掌握計(jì)量經(jīng)濟(jì)學(xué)的理論與方法。了解和掌握計(jì)量經(jīng)濟(jì)分析的步驟和方法，同時(shí)知道如何在實(shí)踐中運(yùn)用計(jì)量經(jīng)濟(jì)學(xué)。2．通過(guò)課程實(shí)驗(yàn)，利用計(jì)量經(jīng)濟(jì)學(xué)模型定量分析研究經(jīng)濟(jì)問(wèn)題。培養(yǎng)自我的分析問(wèn)題和解決問(wèn)題的能力，在學(xué)完該課程后，不僅學(xué)習(xí)了理論知識(shí)和計(jì)算方法，還能接觸到社會(huì)實(shí)際中有待解決的計(jì)量經(jīng)濟(jì)問(wèn)題，并能建立數(shù)學(xué)模型求解。3．通過(guò)實(shí)驗(yàn)教學(xué)培養(yǎng)自己發(fā)現(xiàn)問(wèn)題、分析問(wèn)題、解決問(wèn)題的能力，能夠自己分析數(shù)據(jù)結(jié)果。4．通過(guò)世間使自己能夠由理論認(rèn)知上升到實(shí)踐認(rèn)知，通過(guò)自己的理解真切的感受經(jīng)濟(jì)現(xiàn)象的變動(dòng)。【實(shí)驗(yàn)要求】1．要求我們了解和熟悉計(jì)量經(jīng)濟(jì)學(xué)的基本知識(shí)，為具體操作做好知識(shí)準(zhǔn)備。2．要求我們熟悉什么是計(jì)量經(jīng)濟(jì)，如何利用計(jì)量分析，他們各自的分類(lèi)又有什么，為具體操作做好知識(shí)準(zhǔn)備。3．要求我們利用計(jì)量經(jīng)濟(jì)學(xué)軟件，按實(shí)際操作規(guī)范和流程進(jìn)行操作處理，培養(yǎng)自己的動(dòng)手操作能力，培養(yǎng)自己EViews熟悉程度。4．能夠正確運(yùn)用軟件，能夠看懂軟件中給出的數(shù)據(jù)所代表的意義。能夠了解理論、數(shù)據(jù)與實(shí)際之間的某些相關(guān)性。5.對(duì)于外界條件的變化，具有一定的分析解決問(wèn)題的能力。【實(shí)驗(yàn)原理】EViews6.0軟件普通最小二乘法、擬合優(yōu)度的判定、t檢驗(yàn)等利用教材《計(jì)量經(jīng)濟(jì)學(xué)實(shí)驗(yàn)教程》以及搜集的數(shù)據(jù)【實(shí)驗(yàn)內(nèi)容】創(chuàng)建工作文件；利用并建立“能源消費(fèi)總量的影響因素分析”的計(jì)量經(jīng)濟(jì)學(xué)模型輸入（編輯）數(shù)據(jù)回歸分析進(jìn)行經(jīng)濟(jì)意義檢驗(yàn)和統(tǒng)計(jì)推斷檢驗(yàn)進(jìn)行進(jìn)行多重共線(xiàn)性檢驗(yàn)、異方差檢驗(yàn)和序列相關(guān)性檢驗(yàn)及其修正【實(shí)驗(yàn)步驟】模型設(shè)定理論上認(rèn)為影響能源消費(fèi)需求總量的因素主要有經(jīng)濟(jì)發(fā)展水平、收入水平、產(chǎn)業(yè)發(fā)展、人民生活水平提高、能源轉(zhuǎn)換技術(shù)等因素。為此，收集了1980——2011年中國(guó)能源消費(fèi)標(biāo)準(zhǔn)煤總量、國(guó)民生產(chǎn)總值、國(guó)內(nèi)生產(chǎn)總值GDP、工業(yè)增加值、交通運(yùn)輸郵電業(yè)增加值、城鎮(zhèn)居民人均消費(fèi)性支出、能源加工轉(zhuǎn)換總效率的統(tǒng)計(jì)數(shù)據(jù)。本文旨在建立多元先行回顧模型，以探究不同因素對(duì)能源消費(fèi)總量的影響。其中：Y為能源消費(fèi)總量；X1為國(guó)民生產(chǎn)總值；X2為國(guó)內(nèi)生產(chǎn)總值；X3為工業(yè)增加值X4為交通運(yùn)輸業(yè)增加值；X5為城鎮(zhèn)居民人均消費(fèi)性支出；X能源加工轉(zhuǎn)換總效率表1：年份YX1X2X3X4X5X61980602754545.6244545.6241996.5213.3619412.4469.541981594474889.46114891.56112048.4220.698456.8469.281982620675330.4515323.3512162.3246.855347169.21983660405985.55165962.65162375.6274.9457505.9269.931984709047243.75177208.05172789338.5157559.4469.161985766829040.73669016.03663448.7421.7173673.268.2919868085010274.3792102758358798.9668.3219878663212050.615112058.61514585.8568.3195884.467.4819889299715036.82315042.8235777.2685.7351103.9866.5419899693417000.919116992.31916484812.7411210.9566.5119909870318718.322418667.822468581167.03271278.8967.2199110378321826.199421781.49948087.11420.30021453.8165.9199210917026937.276526923.476510284.51688.97811671.7366199311599335260.024735333.924714187.9712173.99472110.8167.32199412273748108.456448197.856419480.712787.88852851.3465.2199513117659810.529260793.729224950.6113244.29713537.5771.05199613894870142.491771176.591729447.6073782.15913919.4771.5199713817378060.852878973.03532921.3894148.56644185.6469.23199813221483024.279884402.279834018.4314660.90094331.669.44199913011988479.154889677.054835861.4795175.17154615.969.19200013855398000.454399214.554340033.5936160.9517499869.042001143199.3108068.2206109655.170643580.6166870.2508530969.032002151796.59119095.6893120332.689347431.3087492.94666029.8869.042003174990135173.9761135822.756154945.5277913.18986510.9469.42004203227159586.7736159878.337965210.0299304.3937182.170.712005224682183618.5053184937.36977230.77910666.16317942.8871.162006246270.15215883.9487216314.425991310.93612182.98468696.671.242007265583266411.0218265810.3058110534.8814601.0394999770.772008285000315274.7098314045.4271130260.2416362.503211242.8571.552009306647341401.4756340902.8126135239.9516727.109812264.5572.012010324939403259.9564401202.0284160867.0118968.4813471.4572.862011348002472115471563.7188470.221931.91516173.52參數(shù)估計(jì)和模型檢驗(yàn)1.直接對(duì)模型進(jìn)行OLS估計(jì)，結(jié)果如下表2：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:10:20Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C301791.897597.723.0922020.0048X17.1522614.0769791.7543040.0916X2-8.9844144.043508-2.2219360.0356X35.1912941.7009443.0520090.0053X4-9.3751115.429815-1.7265990.0966X525.984976.1445454.2289500.0003X6-3528.8591432.604-2.4632480.0210R-squared0.990118

Meandependentvar149585.4AdjustedR-squared0.987747

S.D.dependentvar82617.95S.E.ofregression9145.279

Akaikeinfocriterion21.27050Sumsquaredresid2.09E+09

Schwarzcriterion21.59113Loglikelihood-333.3281

Hannan-Quinncriter.21.37678F-statistic417.4962

Durbin-Watsonstat1.501467Prob(F-statistic)0.000000得到的模型為：Y=301791.847537+7X1-8.98441426858*X2+5.19129414597*X3-9.37511099997*X4+25.9849710382*X5-3528.85937586*X62模型檢驗(yàn)2.1經(jīng)濟(jì)意義檢驗(yàn)從各變量的參數(shù)看，國(guó)民生產(chǎn)總值（x1）、工業(yè)增加值(x3)、城鎮(zhèn)居民人均消費(fèi)性支出(x5)的參數(shù)符號(hào)均為正，與能源消費(fèi)需求總量呈正相關(guān)，與現(xiàn)實(shí)經(jīng)濟(jì)意義相符，通過(guò)經(jīng)濟(jì)意義檢驗(yàn)，國(guó)內(nèi)生產(chǎn)總值GDP（x2）、交通運(yùn)輸郵電業(yè)增加值（x4）的符號(hào)為負(fù)，與能源消費(fèi)總量成負(fù)相關(guān)，不符合經(jīng)濟(jì)意義，未通過(guò)經(jīng)濟(jì)意義檢驗(yàn)；能源加工轉(zhuǎn)換總效率（x6）符號(hào)為負(fù)，與能源消費(fèi)總量成負(fù)相關(guān)，符合經(jīng)濟(jì)意義，通過(guò)經(jīng)濟(jì)意義檢驗(yàn)。2.2統(tǒng)計(jì)推斷檢驗(yàn)（1）擬合優(yōu)度檢驗(yàn)從表中可以看出，可決系數(shù)為0.990118，調(diào)整后的可決系數(shù)為0.987747，均很高，說(shuō)明模型的擬合優(yōu)度較好。（2）t統(tǒng)計(jì)量檢驗(yàn)從表中可以看出，各解釋變量的t統(tǒng)計(jì)值均大于5%的顯著性水平下的臨界值，對(duì)能源消費(fèi)總量均有顯著影響。（3）F統(tǒng)計(jì)量檢驗(yàn)上表中，F(xiàn)統(tǒng)計(jì)量為417.4962，大于臨界值，說(shuō)明方程總體顯著。2.3計(jì)量經(jīng)濟(jì)學(xué)檢驗(yàn)2.3.1多重共線(xiàn)性檢驗(yàn)在經(jīng)濟(jì)意義檢驗(yàn)中，國(guó)內(nèi)生產(chǎn)總值GDP（x2）、交通運(yùn)輸郵電業(yè)增加值（x4）的符號(hào)為負(fù)，與能源消費(fèi)總量成負(fù)相關(guān)，不符合經(jīng)濟(jì)意義，未通過(guò)經(jīng)濟(jì)意義檢驗(yàn)，因此懷疑存在多重共線(xiàn)性。表3各解釋變量間的相關(guān)關(guān)系系數(shù)：從相關(guān)系數(shù)矩陣可以判斷出，各解釋變量之間存在嚴(yán)重的多重共線(xiàn)性。2.3.2使用逐步差分法對(duì)多重共線(xiàn)性進(jìn)行修正（1）用OLS法逐一Y對(duì)各個(gè)解釋變量的回歸，可以發(fā)現(xiàn)Y對(duì)城鎮(zhèn)居民人均消費(fèi)性支出(x5)的線(xiàn)性關(guān)系最好，結(jié)果如下：表4：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:10:23Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C61600.853167.35119.448700.0000X519.305430.51283537.644550.0000R-squared0.979269

Meandependentvar149585.4AdjustedR-squared0.978578

S.D.dependentvar82617.95S.E.ofregression12092.17

Akaikeinfocriterion21.69897Sumsquaredresid4.39E+09

Schwarzcriterion21.79057Loglikelihood-345.1834

Hannan-Quinncriter.21.72933F-statistic1417.112

Durbin-Watsonstat0.241752Prob(F-statistic)0.000000（2）逐步進(jìn)行回歸，將其他解釋變量逐一代入：①將變量X1代入，結(jié)果如下：表5：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:10:19Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

X514.516663.1531204.6039030.0001X10.1621230.1053891.5383250.1348C65999.254215.47015.656440.0000R-squared0.980833

Meandependentvar149585.4AdjustedR-squared0.979511

S.D.dependentvar82617.95S.E.ofregression11825.85

Akaikeinfocriterion21.68302Sumsquaredresid4.06E+09

Schwarzcriterion21.82044Loglikelihood-343.9284

Hannan-Quinncriter.21.72857F-statistic742.0125

Durbin-Watsonstat0.282255Prob(F-statistic)0.000000從上表可以看出，R^2統(tǒng)計(jì)量提高到0.980833，模型擬合優(yōu)度略有提高。但是解釋變量X1的t統(tǒng)計(jì)量為1.53，小于5%顯著性水平下的臨界值1.697，不能拒絕系數(shù)為零的原假設(shè)，決定舍去變量X1。②將變量X2代入，結(jié)果如下：表6：DependentVariable:YMethod:LeastSquaresDate:06/26/13Time:19:38Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

X514.576773.2529624.4810770.0001X20.1603760.1089991.4713530.1520C65876.544254.61515.483550.0000R-squared0.980709

Meandependentvar149585.4AdjustedR-squared0.979379

S.D.dependentvar82617.95S.E.ofregression11864.03

Akaikeinfocriterion21.68947Sumsquaredresid4.08E+09

Schwarzcriterion21.82688Loglikelihood-344.0315

Hannan-Quinncriter.21.73502F-statistic737.1511

Durbin-Watsonstat0.280819Prob(F-statistic)0.000000從上表可以看出，與表4中的結(jié)果相比，R^2統(tǒng)計(jì)量提高到0.980709，模型擬合優(yōu)度略有提高。但是解釋變量X2的t統(tǒng)計(jì)量為1.47，小于5%顯著性水平下的臨界值1.697，不能拒絕系數(shù)為零的原假設(shè)，決定舍去變量X2。

③將變量X3代入，結(jié)果如下：表7：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:10:34Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

X513.172423.3271333.9590900.0004X30.5131720.2753221.8638980.0725C67151.314258.58715.768450.0000R-squared0.981487

Meandependentvar149585.4AdjustedR-squared0.980210

S.D.dependentvar82617.95S.E.ofregression11622.41

Akaikeinfocriterion21.64832Sumsquaredresid3.92E+09

Schwarzcriterion21.78573Loglikelihood-343.3731

Hannan-Quinncriter.21.69387F-statistic768.7280

Durbin-Watsonstat0.295303Prob(F-statistic)0.000000從上表可以看出，與表4中結(jié)果相比，R^2統(tǒng)計(jì)量提高到0.981487，模型擬合優(yōu)度略有提高。解釋變量的t檢驗(yàn)的值均大于5%顯著性水平下的臨界值1.697，通過(guò)了顯著性檢驗(yàn)，決定保留變量X3。④將變量X4代入，結(jié)果如下：表8：DependentVariable:YMethod:LeastSquaresDate:06/27/13Time:14:23Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C67594.214917.45613.745770.0000X30.4515610.4281761.0546150.3006X41.1547926.0713950.1902020.8505X512.210676.0842632.0069270.0545R-squared0.981511

Meandependentvar149585.4AdjustedR-squared0.979530

S.D.dependentvar82617.95S.E.ofregression11820.50

Akaikeinfocriterion21.70953Sumsquaredresid3.91E+09

Schwarzcriterion21.89274Loglikelihood-343.3524

Hannan-Quinncriter.21.77026F-statistic495.4648

Durbin-Watsonstat0.298883Prob(F-statistic)0.000000從上表可以看出，與表7中的結(jié)果相比，R^2統(tǒng)計(jì)量提高到0.981511，模型擬合優(yōu)度略有提高。但是各解釋變量不再顯著，決定舍去變量X4。

⑤將變量X6代入，結(jié)果如下：表9：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:11:34Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C281950.5100834.92.7961610.0092X30.6045360.2634052.2950860.0294X513.242223.1408744.2160930.0002X6-3163.1081483.702-2.1319030.0419R-squared0.984072

Meandependentvar149585.4AdjustedR-squared0.982366

S.D.dependentvar82617.95S.E.ofregression10971.17

Akaikeinfocriterion21.56040Sumsquaredresid3.37E+09

Schwarzcriterion21.74362Loglikelihood-340.9664

Hannan-Quinncriter.21.62113F-statistic576.6474

Durbin-Watsonstat0.538128Prob(F-statistic)0.000000從上表可以看出，與表7中結(jié)果相比，R^2統(tǒng)計(jì)量提高到0.984072，模型擬合優(yōu)度略有提高。解釋變量的t檢驗(yàn)的值均大于5%顯著性水平下的臨界值1.697，通過(guò)了顯著性檢驗(yàn)，決定保留變量X6。2.3.3異方差檢驗(yàn)使用White方法檢驗(yàn)異方差是否存在表10：HeteroskedasticityTest:WhiteF-statistic2.540638

Prob.F(9,22)0.0359Obs*R-squared16.30874

Prob.Chi-Square(9)0.0607ScaledexplainedSS7.302465

Prob.Chi-Square(9)0.6057TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:06/25/13Time:10:56Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C-1.63E+103.94E+10-0.4123160.6841X3-385322.0356328.4-1.0813680.2912X3^20.6267310.6223561.0070300.3249X3*X5-17.3425114.71616-1.1784670.2512X3*X65917.9795081.0581.1647140.2566X53011515.3576284.0.8420800.4088X5^2105.193087.056281.2083330.2397X5*X6-46483.8350634.75-0.9180220.3686X64.59E+081.15E+090.3992960.6935X6^2-3213696.8380279.-0.3834830.7050R-squared0.509648

Meandependentvar1.05E+08AdjustedR-squared0.309050

S.D.dependentvar1.16E+08S.E.ofregression96197333

Akaikeinfocriterion39.85201Sumsquaredresid2.04E+17

Schwarzcriterion40.31005Loglikelihood-627.6321

Hannan-Quinncriter.40.00384F-statistic2.540638

Durbin-Watsonstat1.789067Prob(F-statistic)0.035937上圖中檢驗(yàn)結(jié)果是White統(tǒng)計(jì)量Obs*R-square=16.30874，小于5%顯著性水平下的臨界值16.92，不能拒絕原模型同方差性的原假設(shè)，則原模型不存在異方差性。2.3.4序列相關(guān)性檢驗(yàn)(1)使用殘差散點(diǎn)圖檢驗(yàn)序列相關(guān)性。圖2：圖3：從上面兩張圖可以基本看出模型存在正自相關(guān)。(2)使用D.W.統(tǒng)計(jì)量檢驗(yàn)序列相關(guān)性：表11：DependentVariable:YMethod:LeastSquaresDate:06/25/13Time:11:34Sample:19802011Includedobservations:32VariableCoefficientStd.Errort-StatisticProb.

C281950.5100834.92.7961610.0092X30.6045360.2634052.2950860.0294X513.242223.1408744.2160930.0002X6-3163.1081483.702-2.1319030.0419R-squared0.984072

Meandependentvar149585.4AdjustedR-squared0.982366

S.D.dependentvar82617.95S.E.ofregression10971.17

Akaikeinfocriterion21.56040Sumsquaredresid3.37E+09

Schwarzcriterion21.74362Loglikelihood-340.9664

Hannan-Quinncriter.21.62113F-statistic576.6474

Durbin-Watsonstat0.538128Prob(F-statistic)0.000000從表中可以看出，D.W.統(tǒng)計(jì)量的數(shù)值為0.538，小于統(tǒng)計(jì)量下限的dL，模型存在正自相關(guān)。(3)拉格朗日乘數(shù)檢驗(yàn)首先從1階開(kāi)始檢驗(yàn)，結(jié)果如下：表12：Breusch-GodfreySerialCorrelationLMTest:F-statistic38.31136

Prob.F(1,27)0.0000Obs*R-squared18.77106

Prob.Chi-Square(1)0.0000TestEquation:DependentVariable:RESIDMethod:LeastSquaresDate:06/25/13Time:11:51Sample:19802011Includedobservations:32Presamplemissingvaluelaggedresidualssettozero.VariableCoefficientStd.Errort-StatisticProb.

C-101191.868017.03-1.4877420.1484X3-0.3222340.180154-1.7886610.0849X52.9813572.1121851.4115040.1695X61461.512999.75801.4618650.1553RESID(-1)0.8600330.1389486.1896170.0000R-squared0.586596

Meandependentvar2.18E-11AdjustedR-squared0.525351

S.D.dependentvar10426.80S.E.ofregression7183.525

Akaikeinfocriterion20.73957Sumsquaredresid1.39E+09

Schwarzcriterion20.96