計量經(jīng)濟學Eviews多重共線性實驗報告

1、頁尉內(nèi)容14-SQroHH 山V*凰東目打至大學CUANCDONC UNIVERSITV OF FINANCE & ECONOMICS實驗報告課程名稱實驗項目名稱多重共線性班級與班級代碼專 業(yè)任課教師學 #:姓名：實驗日期： 2014年05月 11日姓名一評語:廣東商學院教務處制實驗報告成績指導教師（簽名）年 月曰說明：抬導教師評分后.實驗報告交院（系）辦公室保存。計量經(jīng)濟學實驗報告一、實驗目的：掌握多元線性回歸模型的估計方法、掌握多重共線性模型的識別和修正。二、實驗要求：應用教材第127頁案例做多元線性回歸模型，并識別和修正多重共線性。三、實驗原理：普通最小二乘法、簡單相關(guān)系數(shù)檢驗法、綜合判

2、斷法、逐步回歸法。四、預備知識：最小二乘法估計的原理、t檢驗、F檢驗、A?值。五、實驗步驟1、選擇數(shù)據(jù)理論上認為影響能源消費需求總量的因素主要有經(jīng)濟發(fā)展水平、收入水平、產(chǎn)業(yè)發(fā)展、 人民生活水平提髙、能源轉(zhuǎn)換技術(shù)等因素。為此，收集了中國能源消費標準煤總量、國民總 收入、國內(nèi)生產(chǎn)總值GDP、工業(yè)增加值、建筑業(yè)增加值、交通運輸郵電業(yè)增加值、人均生 活電力消費、能源加工轉(zhuǎn)換效率等19S52007年的統(tǒng)計數(shù)據(jù)。本題旨在通過建立這些經(jīng) 濟變量的線性模型來說明影響能源消費需求總量的原因。主要數(shù)據(jù)如下：19852007年統(tǒng)計數(shù)據(jù)年 份能源消費一民收國總?cè)雵鴥?nèi) 生產(chǎn)總III工業(yè) 增加 值筑業(yè)増加值交通運輸 郵電

3、增加值人均生電力消 費能源 加工轉(zhuǎn)換 效率yX1X2XsX4X5X6X719S5766829040.79016S*8.7117.9106.921.S6S.2919S68OS5O10274.10275.2S967525.7475.62S.26S.521987S665212050.61205S.645S5.8665.85址926.4-67.士S19SS9299715056.S15042.S5777.2S1066131.266.5419S996954-17000.916992.S64S479土7S635.566.5119909S70S1S71S.S1S667.S6S58S39V1147.542.4-

4、67.219911OS78521826.2217S1.080S7.11015.11409.746.965.9199210917026957.S2692S.5102S土 5141516S1.85土 666.00199511599S55260S5SSS.9141SS2266.52205.661.267.521994-12275I4S108.54S197.9194S0.72964.72898.372.765.2199515117659S10.j6079S.724950.6572S.85424.18S.571.05199615S9+S7014*2.571176.6294*4-7.615S7.4106S

5、.59S.171.5199715779S77655.17897S52921.14-621.64*595101.869.25199S15221S5024.5S4402.S5-tO1S.41985.85178 理106.669.141999155S51881S9S9677.1S5S61.55172.15821.Sns 269.1920001585559S000.j99214.640055.65522.S7S55.41S2.4-69.0土200114319910S06S.2109655.2455S0.65951.7S406.1144.669.05200215179119095.7120S32.747

6、451.S6165.59595.4156.569.042005174990155171S5S22.S54-945.574-90S1009S.417S.769.42004-20522I159586.7159S7S.S65210S69S12147.6190.270.7120052255191S5956.11SS0S4.S76912.910155.S10526.1216.771.08200624627021S1S1.721192S.591S10.911S51.1124S1.1249.471.242007Q655SS2514SS.2249529.9107S67.214014.114604.1274.9

7、71.25資料來源：中國統(tǒng)計年鑒，中國統(tǒng)計出版社2000、200S年版。為分析Y與XI、X2. X5、X4. Xo. X6、X7之間的關(guān)系，做如下折線圖:丫X1 -X2X3 X4 X6-X6 X7能源消費Y在19S6到1996年間緩慢增長，在96至9S年有短暫的下跌，但是9S 至02年開始緩慢回升，02年到06年開始快速增長。國民總收入XI和國內(nèi)生產(chǎn)總值X2以相同的趨勢逐年緩慢增長。工業(yè)增加值XS在1985年-1999年期間一直是緩慢增長，但在2000年出現(xiàn)了急劇 下降的現(xiàn)象，2001年又急劇增長，達到下降前的水平，2001年以后開始緩慢增長。建 筑業(yè)增長值盼、交通運輸郵電業(yè)增加值X5、人均生

8、活電力消費X6、能源加工轉(zhuǎn)換效率 曲數(shù)值較低，但都以較平緩的方式增長。2、設眾并估il多元線性回歸模型X =0 + PlXX + 尸3X3? +04*4,七卩+ PfyXz +l,i （2-1）2.1錄入數(shù)據(jù)，得到圖。File Eiit Object Viev Proc guick Options 世 irdcw Helpdata yHi x2 x3x4 x5x6 x7 Tuxkfilc; UiniTLED Group: UITITLED Torkfile: UHTITLED:Untitled IT)YiEwXPmdQbje戈 |Print5dveRanae: 1985 2007 - 23od

9、sSample: 1G85 2007 - 23obsg cgrupoga Ecit47Sm日+/|fiSHI 0 x2EZJ x3 0 vA 055 0 ：6 0 z7 0VobsY冶、門1985NA9040.7009013.0003448.700417.9014%198680050.0010274.4010275.203967.000525.7011987866320012050 6012058.604585.800665.801198892997.0015036.8015042.805777.200310.001198996934.0017000.0016992.3064B4.000794

10、.001199098703.0018718.3018667.806858.000859.4011991103783.021 826.2021731.508087.1001015111992109170 026937.3026923.5010294 50U150I19931159S3035260.0035333.9014188 002266.5I1994122737.048108.5048197.9019480702964 7I1995131176059810.5060793.7024950 603728 8I1998138948.070142.5071176.6029447.604387 4I

11、1997137708.077653.1078973.0032921 404621.611998132214083024.3084402.3034018.404985.811999133831 088189.0089677.1035B61.505172112000138553 09800050992U.604003.6005522 312001143199.0108068.2109355.243580605931.712002yam1M7Q7 0HAOAh TIIUi1909 747431 snPU佔 Al YProdpbjxt Default2.2.1）采用OLS估計參數(shù)在主界而命令框欄中輸入

12、Is yc xl X2 X5 x4 x5 x6 x7回車，即可得到參數(shù)的估讓結(jié)果。1? oil n-t x on z ITHTTITI-ED Wo rki-le： z TIWTT TI.RI = =一 X |Dependent variable: V Metnod: ueast Squares DQt： 0 5/1 8/1 4 Timo： 08:09 Sample: 1 9Q5 2007 Included obsBrvtions 23xxxxxxx123 4 56 7R-squaredAdju =le?d R- = quare d S.E. orraresion 9umrdLoa likeli

13、hood F-statlstic尸 rob/ar139364.6Adjusted R-squared0968441S.D. dependentvar51705 05SE of regression9185 318Aka ike info criterion21 20971Sum squared resid1.69E*D9Schwatz criterion21 35782Loo likelihood-240.9116H加nan-Quinn criter21 24696F-statistic338.5541Durbin-Wats on stat0.370310Prob(F-s1a1istic)oo

14、ooooo經(jīng)比較，可以發(fā)現(xiàn)加入X2、X5、X6. X7后參數(shù)的符號與預期相反，不符合經(jīng)濟意義, 且t檢驗部顯著。而加入X+后變化并不顯著，只有加入X5后修正的可決系數(shù)有所提高，而 且參數(shù)符號的經(jīng)濟意義合理，而且參數(shù)的t檢驗，任。二0.1, t(O.O5,15)=1.755時顯著，所以 保留X5。再加入其他新變量逐步回歸。 Equation:UWTirLEDTorirfile: IIKTITLED:.匸問闔lcv l；Froc | Object| Prirt Home Freeze Esbm威|?也出収8討sD9p9rid9ntVariabie: YM9thod: Least Squaros D

15、ate: 05/19/14 Time: 06:56 Sample: 1985 2007IncludEd obseivlions. 23coemcienistd. Errort-StatsticProoc83887.252725.96130.777240.0000K15.6170572 0097372.8944360.0093K30 2520350 2201381.1448980.2665X2-5 2106231.986646-2.6228240.0167R-squared0981702Mean dependemvar1 39364 6Adjusted R-squared0.978813S.D.

16、 d9p9nd9ntvar61 705.06S E. GfrogrGGGion7526.088jAkaiko irifo critorion20.84691Sum squared resid1.08E4 09Schwarz criterion21.04439Log likelihood-235.7395Hannan-Quinn criter.20.89657F-stalisfic3397883Durbin-Watson stal0.585884Prot)(F-$tatisiic)0000000I Equation: IJHTirLED orirfile: IJHTITLED:Mcvs 卜od|

17、Objcct| Pri上 | Nome I Freeze Esbmot cForewst iMtats |05麗D9p9nd9nt Vari able: Y Mothod: Least Squaros Date: 05/19/14 Time: 06:56 Sample 1985 2007 Included obseivations. 23coetuciemstd. Errort-statsticPronC80B77 082934.05927.496750.0000K10.3370630.4939460.6823880.5032K30.4330770.2424781.7660400.0901X4

18、38737908 7311180.4436760.6623R-squared0975333Mean dependemvar1 39364.6Adjusted R-squafed0.971438S.D. d9p9nd9ntvar51 706.06S E. cfrogrGGGion8738.358Zkaike irifo critGrion21.14560Sum squared resid1.45E409Schwarz criterion21.34308Log likelihood-239.1744Hannan-Quinn criter21.19527F-stalistic2504150Durbi

19、n-Watson stal0.697040PfOb(F-$tatislic)0000000Equation: UBTIILED forirfiLe: OTTIILED: :. JQJNwi訓ProcobE- Prnd|xa卜no 住北戲司農(nóng)“苗號父戲卜亦djDependent variable: YMethod: Leasl SquarQGDate: 05/1 8f14 Time: 08:57Sample 1935 2007included o&Ger/ationG： 23Coeffl cientStd Errort-SiatjsticProbc81 000.322592.22231.2474

20、40.0000X10 89530801849454 8409530 0001X30.2677660.2330951.1058390.2826X54.262556I 969195-2.1B46190.0434R-squarcd0.S80007Moan doponderitvr139364.6Adjusted R-sauared0.976851S.D. de pen dent var51705.05S.E or regression7868.886Akaike info criterion20.93548Sum cquarod rosid1.18E09Schwarz criterion21.132

21、96Log likelihood-236.7581Hannan-Quinn c liter20.98515F-siatjstic3104495DurbIn-Watson stat0761915Prob(F-statiGlic)0.000000Ediiation: UlTTTn.ED TorWile: DWTITLED:. 旦區(qū)JFyoJ Object Prirt Harris) Feezg EsbmmgjFo陽2或彌回門8討5Dependent vanaDie：YMathod： Least squaresDate: 06/18/14 Time: 08:68Sample: 1985 2007In

22、cluded obsstvtions: 23CoefficientStd. Error卜 StatisticProb82340.87血 3.24811.641170.00000 63716104065781 5671290.13360 4152480 2593611.6010450.1259 72 52703331 3391-0 2188910 8291Rsquared0.976140Moan d8p9nd9ntv3r1 39364.6Adjusted R-squared0 971214G.D. dependentvar51705.05S E. of regression9772454Akai

23、ke irifo criterion21.15339Sum squared resid1.46E-09Schwarz criteiion21.35087Log likelihood-2392640Hannan-Quinn criter.21.20306F-StatlSBC2484231Durbin-Watson stai0.677386PfOD(F-$tatiSliC)0000000 Equation: UMriTLED lorkfile: UBTITLED:.回區(qū)|vieN|prod-Obieo:.Freeze凸亦氏“巾畋対-而總他旳由DependGnt Variable: Method:

24、Least Squares Dat9： 06/18M4 Time: 08:58Sample: 1986 2007Included obspnations: 23coemcierStd. Errort-SiatlsilcProb.138962.596708.101.4369270.000.587BT00.1080845.2521240.00000.4377540.2414661.8129010.0857-865.56571441731 0.0003660.5554R-squared0.975541Mean depenclenMar1 39364.6Adjusted R-squared0.971

25、679S.D dependent var51 705.05S. of regression9701.361Akaike info criterion21.13712Gum squared residi 44E*09Gcharz criterion21.33459Log likelihood-239.0769Hannan-Quinn criter.21.19678F-statisiic252.6D30Durtoin-WatGon stat0.765793Prob(F-statistic)0.000000當加入X2時，雖然R，2有所增加，但其系數(shù)的符號與預期相反且參數(shù)的t檢驗不顯著: 加入:K士后，各參數(shù)的t檢驗不顯著：加入X5后，雖然R2有所增加，但是但其系數(shù)的符 號與預期相反且參數(shù)的t檢驗不顯著：加入X6、X7后，其系數(shù)的符號與預期相反且參數(shù)的 t檢驗不顯箸,這說明主要是X2、Xo. X6、X7引起了多重共線性，應予以剔除。i Equation: UN