分布滯后模型與自相關(guān)模型

分布滯后模型與自相關(guān)模型實(shí)驗(yàn)?zāi)康牧私庠摰毓潭ㄙY產(chǎn)與銷售額的關(guān)系,并分析其經(jīng)濟(jì)學(xué)意義，熟練掌握Eviews的用法。實(shí)驗(yàn)要求表中給出了某地區(qū)1980-2001年固定資產(chǎn)投資Y與銷售額X的資料（單位：億元）。年份YX年份YX198036.9952.8051991128.68168.129198133.6055.9061992123.97163.351198235.4263.0271993117.35172.547198342.3572.9311994139.61190.682198452.4884.7901995152.88194.538198553.6686.5891996137.95194.657198658.5398.7971997141.06206.326198767.48113.2011998163.45223.541198878.13126.9051999183.80232.724198995.13143.9362000192.61239.4591990112.60154.3912001182.81235.142試就下列模型，按照一定的處理方法估計(jì)模型參數(shù)，并解釋模型的經(jīng)濟(jì)意義，探測(cè)模型擾動(dòng)項(xiàng)的一階自相關(guān)性。（1） 設(shè)定模型Y*=a+0X+u運(yùn)用局部調(diào)整假定。（2） 設(shè)定模型Y*=aX^eut運(yùn)用局部調(diào)整假定。（3） 設(shè)定模型=a+0X*+u運(yùn)用自適應(yīng)預(yù)期假定。（4） 運(yùn)用阿爾蒙多項(xiàng)式變換法，估計(jì)分布滯后模型：=a+。X+pX++pX+u0t1t-1 4t-4t

實(shí)驗(yàn)原理最小二乘法，德賓h檢驗(yàn)實(shí)驗(yàn)步驟1.設(shè)定的模型為Y*=a+^Xt+ut令Y—Y=8(Y*—Y)，運(yùn)用局部調(diào)整模型，得新模型為Y=a*+阡X+B*Y +u*，t t—1 tt-1 t 0t1 t—1 t其中a*=8a,P*=8P,P*=1—8,u*=8u。在Eviews的命令窗口輸入Isycxy(-1),回車0 1 t t得如下回歸結(jié)果DependentVariable:YMethod:LeastSquares□ate:11/18/10Time:18:04Sample(adjusted):19812001Includedobservations:21afteradjustmentsVariableCoefficientStd.Errort-St.atisticProb.C-15.104034.729460-3.1936130.0050X0.6292730.0978196.4330310.0000Y同0.2716760.1148682.3653150.0294R-squared0.987126Meandependentvar109.2167AdjustedR-squared0.985695S.D.dependentvar51.78550S.E.ofregression6.193728Akaikeinfocriterion6.616615Sumsquaredresid690.5208Schwarzcriterion6.765733Loglikelihood-66.47341F-statistic6900561Durbin-Watsonstat1.518595Prob(F-statistic)0.000000對(duì)該模型進(jìn)行德賓h檢驗(yàn)h=h=(1一勺n一"/1—nVarp*1k7=(1-￥)Wk^=1..29取顯著性水平a=0.05，查標(biāo)準(zhǔn)正太分布表得臨界值h=1.96，由于a2hl=1.29<h=1.96，因此接受原假設(shè)p=0，說(shuō)明自回歸模型不存在一階自相a關(guān)。2關(guān)。由上表可知a*=8a=—15.10403,P：=8P=0.629273,P;=1—8=0.271676TOC\o"1-5"\h\z\o"CurrentDocument"則最后模型為Y*=-20.7381+0.8640Xt t該模型說(shuō)明了銷售額每增加1億元，固定資產(chǎn)投資增加0.8640億元。2.設(shè)定模型為Y*=aXPeut令lnY-InY1=5(InY*-InY)，對(duì)Y*=aXPe?t左右取對(duì)數(shù)得\o"CurrentDocument"InY*=a+PInX+u，得新模型為InY=a*+P*lnX+P*lnY+u*，其中0t1 t-1\o"CurrentDocument"a*=5a,P*=5p,P*=1-5,u*=5u。在Eviews的命令窗口輸入Islog(y)clog(x)0 1 t tlog(y(-1)),回車得如下回歸結(jié)果DependentVariable:LOG(Y)Method:LeastSquares□ate:11/26/10Time:22:50Sample[adjusted):19812001Includedobservations:21afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C-1.0780460.184144-5.8543660.0000LOG[X)0.9046220.1112438.1310390.0000LOG[Y(-1))0.2600330.0877992.9616840.0084R-squared0.993725Meandependentvar4.559823AdjustedR-squared0.993028S.D.dependentvar口一562953S.E.ofregression0.047007Akaikeinfocriterion-3.U5469Sumsquaredresid0.039774Schwarzcriterion-2996251Loglikelihood36.02742F-statistic1426.219Durbin-Watsonstat1.479333Prob(F-statistic)0.000000對(duì)該模型進(jìn)行德賓h檢驗(yàn)=1.40h=(1-d)：―- =a-*)「―21—=1.402d1-〃s"〔 2 V】-21X0.08779921 "1J取顯著性水平a=0.05，查標(biāo)準(zhǔn)正太分布表得臨界值h=1.96，由于a2hl=1.40<h=1.96，因此接受原假設(shè)P=0，說(shuō)明自回歸模型不存在一階自相a2關(guān)。由上表可知a*=5lna=-1.078046,Pj=5P=0.904522,P;=1-5=0.260033則最后模型為W*=0.2330X1.2223該模型說(shuō)明了當(dāng)銷售額增大'到原來(lái)的兩倍時(shí)固定資產(chǎn)投資增大到原來(lái)的2.333倍3.設(shè)定模型為Y=a+0X*+u令X廣x_『(X*-xti)，運(yùn)用自適應(yīng)預(yù)期模型，得新模型為Y=a*+P*X+P*Y+u*，其中a*=8a,。*=y。,P*=1-y,u*=yu。在Eviews0t1t-1 t 0 1的命令窗口輸入Isycxy(-1),回車得如下回歸結(jié)果DependentVariable:YMethod:LeastSquares□ate:11/26/10Time:11:23Sample[adjusted):19812001Includedobservations:21afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C-15.104034.729450-3.19361J0.0050X0.6292730.0978196.4330310.0000Y(-1)0.27167601148582.3653160.0294R-squared0.987125Meandependentvar109.2167AdjustedR-squared0.985695S..D.dependentvar51.78550S.E.ofregression6.193728Akaikeinfocriterion6.616515Sumsquaredresid690.5208Schwarzcriterion6.765733Loglikelihood-66.47341F-statistic690.0561Durbin-Watsonstat1.518595Prob(F-statistic)0.000000同第一種模型，則最后回歸模型為Y*=-20.7381+0.8640Xt t該模型說(shuō)明了銷售額每增加1億元，固定資產(chǎn)投資增加0.8640億元.。4.運(yùn)用阿爾蒙多項(xiàng)式變換法，估計(jì)分布滯后模型：Y=a+pX+PX++PX+u0t1t-1 4t-4 t取Z=X+X+X+X+X0ttt-1t-2t-3 t-4Z=X+2X+3X+4X1tt-1t-2t-3t-4Z=X+4X+9X+16X2tt-1t-2t-3t-4則P=aP=a+a+aP=a+2a+4ap=a+3a+9ap=a+4a+16a將Y對(duì)Zo,z『z2回歸，得如下回歸結(jié)果DependentVariable:YMethod:LeastSquares□ate:11/28/10Time:20:46Sample(adjusted):19842001Includedobservations:18afteradjustmentsVariableCoefficientStd.Errort-StatisticProb.C-35-492348.192884-4.3320930.0007zo0.8910120.1745635.1042480.0002Z1-0.&699040.264447-2.6327830.0197Z20.1043320.0623111.6763380.1160R-squared0.984670Meandependentvar121.2322AdjustedR-squared0.981386S.D.dependentvar4563343S.E.ofregression6.226131Akaikeinfocriterion6.688517Sumsquaredresid542.7059Schwarzcriterion6.886378Loglikelihood-56.19666F-statistic299.7429Durbin-Watsonstat1.130400Prob[F-statistic)0.000000由上表可知，a0=0.891012a1=-0.6