Eviews計量經(jīng)濟學第三版龐皓課后習題3教學提綱

1、精品文檔6.3 1.分析比較股票市場繁榮程度與宏觀經(jīng)濟運行情況之間的關(guān)系，建立估計回歸模型：=+式中，表示股票價格指數(shù)；表示國內(nèi)生產(chǎn)總值(10億美元)(1)利用Eviews做回歸分析:7iObject- PrintFreezeFore b5tStatsDependent Variabfe: YMelliod: Least 白esData: 12/02/17 Time: 11 43Sample: 1981 2006Included observatlons: 25VariablsC 口 afiici 曰 ntStd. Errct-%atisficProb.| ,C-2123.664324 901

2、2-6.5399660 ooo oX07E41D60 04127519.99680Q.OODOR-squared0.9S7643Mean dependent var35G4.775Adjusted R-squared0,935044S.D. dependent var2414.QDQS E. oTregressian615.2413Afcailrtf info criterion1575571Sum squared resid8084%Sctiwarc criterion15.8 9 9Log likelihood-2028242Hannan-Quinn criter1578358F-stat

3、lstlc350.S734口岫 mWatsoo stat0440822Riri&F-siat 話tic)0,000400模型估計結(jié)果：=-2123.864+0.784106(324.8012) (0.041276)t=(-6.538966) (18.99680)=0.937643 F=360.8784 DW=0.440822(2)自相關(guān)的檢驗圖示檢驗法精品文檔精品文檔從上圖可以看出，大部分點都落在1,3象限，表明隨機誤差項 存在正自相關(guān)。bG僉驗/iew Proc Objecj： Print Jame Freeze j istimate Forecast I Stats! Resids

4、Breusch*Gadfre/ Serial Correlation LM TestF-statistic3741424 Prob Ft2,22)0 0000Otis'R-squared20 09265 Prob ChkSquar(2)0 0000Test Equation.Dependent VarlaiMe: RESIDMethod: Least SqnaresDate： 1202/17 Time. 11:52Sample: 61 2006Included obser<alians: 2BPresample missjng value Jagged residuals set

5、 to zero.VariableCoefficientStd Error1-StatisticProb.C-35.62611161.7673-0.2367760 8135X0.0067190 0205650 32672S0 7470RESIDM)1.296702O.1515S98.027-1520 ooooRE SID (-2；-0.6815590.152165-4202063Q.0D04R-squard0 772794Men dependentvar-1.B9E-13Adjusted R-squared0.741812S D depernJentuar6028109S E of regre

6、ssion305 301 eAkai ke info criterion14.42766Sum squared resid2054057Schwaiz criterion14,62121Loq likeliriood-183.5595Hannan-Quinn criter.14.48339F-s tali stic24,94283Durbin-Watson stat2224598Prab(F-statistic)OOQOOOQ=5.9915 且由上圖可知 LM= =26*0.772794=20.09265>P=0.0000< =0.05,表明存在自相關(guān)。通過LM檢驗的結(jié)果可知與的t

7、檢驗均顯著，所以判斷為二階自相關(guān)精品文檔精品文檔(3)修正自相關(guān)廣義差分法在EViews命令欄中輸入“ ls e e(-1) ”得到估計方程對原模型進行廣義差分回歸:-0.768816)+-0.768816=(1-0. 768816 ) +在Eviews中得到廣義差分回歸的輸出結(jié)果:/itwv ProcObjrdPrintNameFreezeEstimateFo recast5tat5ResidsDependentYariaoie: YC,77956Method: Least 3qu5e&Date: 12/02/17 Time: 12:13Sample (adjusted；: 1982

8、 2006Included Gbservatlons:2S alteradjustmeiYtsVatableCoeffiaentStd 一&tot-StatisticPrefarc-634.1843220 3597-2 9774290.00850,7795891)0.9169601052518 187054O.DODCR-squared0.744524Mean dependentvar1055.640Adjusted R-5quared.733416S.D. dependent var748.5019S.E ofregression3064649Akaike info criterio

9、n14 82358Sum squared resid3435166.Schwarz crilerion14.92609Log hkelihood*193 3572Hannan-Quinn criter.14.85562F-statlstlc87.02796Durbin-Watson stat0.908327ProD：P-statistic)0.000000模型估計結(jié)果：=-634.1843+0.861696(220.3997) (0.105251)t=(-2.877429) (8.187054)=0.744524, F=67.02786 DW=0.908327式中， =-0.779589=-0

10、.779589精品文檔精品文檔對廣義差分法做BG 僉驗由上圖可知 LM= =8.9746>=5.9915 且 P=0.02282V =0.05,表明存在自相關(guān)。由于使用了廣義差分數(shù)據(jù)，樣本容量減少了一個，為 25個。查5%a著水平的DW統(tǒng)計表可知 =1.288,=1.454,模型中DW=0.902421V1.288=,表明該模型依然存在自相關(guān)關(guān)系，也說明可能為高階自相關(guān)使用科克倫-奧科特迭代法進行廣義差分回歸：精品文檔精品文檔View Proc:J Object: j | Print bJamc Ff ee-2je J E.stniTLat:e J ForecasJ S1ta.-ts J

11、 ResidsDepandent Vsriobl&： VMetnoa: Least SauiaresL) ste:：皿口山1 7 I i rne: 13：2SSample (adjusted?: T333 2UU8incliJcdedi oossrvfll'ions： 24 網(wǎng)行/r Gdjuatrngct占Oonvergnc achisvi=：dl *Rfter 4 iti=ritionsVariableCoefTicientStd. ErrortStatiatic.Prot>.c-2354 2S?475 5444-4 95079O OOOiXO FllfipnFt0.

12、0567521 H CM門口o nooDARC"| >1 33 D 9 Cl IS 1 73Sa.2Q73S0.0000ARC>-O 7O6-tQ5O 160716一4 石白2今O.DOO3R-s q i* clO ASBBaMeri d9pnctvit “xr3342 195Acfjutfrl 曰-總7，r|Q 口掃4"。S O rlFp-nderil vrsanS 'E of reig re s s- i o nJ90 3fi5F!Akmik日 infa 4=ritsrion14 3T2OBSum wquia日d i晅與id1 756654.弓匚-工

13、 crisis r io n1 4.3G岔4之Lug IKII 1 ioocI-1 ©8.409Hajni lan-Oulnn crltr.1 d 424 17F-£lltlStlC479.471 SurLiIii-WatKun slut2211620F*口 0 CF- &taliStl C)0.0 0 000 011nl09rtecl qR RLoots.67-n.S 1 i=-2354.292+0.816288(475.5444) (0.058753)t=(-4.950729) (13.8938)=0.986286, F=479.4716 DW=2.21162在

14、用BG對上述方程檢驗精品文檔精品文檔VicvwProcObjectPrintNameFre ezeEstimaUForecast_E re us ch-Godfrey Serial Correlati on LM 7 est:F-aUlisiic0.350931Prcb- F(2,18)0.7067at>s*R-squared0 900597 Prob. Chi-Square(2>0.6374Test EqgtgrrDependent VariabEe: RESIDMethod: Least SquaresDote： 12J02/17 Time： 13:31ample- 1S83

15、20DCIncluded observations 24Presampl e missing yalue lagged residuals sei to zero.varia&iecoerrsGienista Errorvstatisticproo.cS1Q2139501.33G90 ,013400.0S71X0 0007S20.053159DX 123750.9903ARC)Q1647540 3437550.4724060.5423AR(2>-0 120068 244523-0.4910280,S293RESIDi-1 ：-0 3028690.441030-0.6367300 5010RCSIDC-2)-0 0405940.410557-.0974 520 923JR-squared0 0375?qMean rtepRndentvar-3 KdE-CSAdjusted R-squared-0 229824S.D. dependent var276.3626S.E. of 白缶sion306 4790Akaike info criterion14.50049Sum squared resid1690729Scriwarz cntenon14 795t1Log likelihood-15