誤差修正模型

1、單位根檢驗(yàn)、協(xié)整與誤差修正模型實(shí)驗(yàn)與案例分析單位根檢驗(yàn)、協(xié)整與誤差修正模型實(shí)驗(yàn)與案例分析為了深入分析研究中國城鎮(zhèn)居民的生活費(fèi)支出與可支配收入的具體數(shù)量關(guān)系，收集了中國城鎮(zhèn)居民月人均可支配收入（SR）和生活費(fèi)支出（ZC）1992年至1998年各月度數(shù)據(jù)序列（見表10.3）。表10.3城鎮(zhèn)居民月人均生活費(fèi)支出和可支配收入序列序列月份19921993199419951996199719981151.83265.93273.98370.0043

2、8.37521.01643.402159.86196.96318.81385.21561.29721.01778.623124.00200.19236.45308.62396.82482.38537.164124.88199.48248.00320.33405.27492.96545.795127.75200.75261.16327.94410.06499.90567.996134.48208.50273.45338.53415.385

3、08.81555.797145.05218.82278.10361.09434.70516.24570.238138.31209.07277.45356.30418.21509.98564.389144.25223.17292.71371.32442.30538.46576.3610143.86226.51289.36378.72440.81537.09599.4011149.12226.62296.50383.58449.03534.

4、12577.40可支配收入Sr12139.93210.32277.60427.78449.17511.22606.141139.47221.74234.28307.10373.58419.39585.702168.07186.49272.09353.55471.77528.09598.823110.47185.92202.88263.37350.36390.04417.274113.22185.26227.89281.22352.154

5、05.63455.605115.82187.62235.70299.73369.57426.81466.206118.2012.11237.89308.18370.41422.00455.197118.03186.75239.71315.87376.90428.70458.578124.45187.07252.52331.88387.44459.29475.409147.70219.23286.75385.99454.93517.065

6、91.4110135.14212.80270.00355.92403.77463.98494.5711135.20205.22274.37355.11410.10422.96496.69生活費(fèi)支出Zc12128.03192.64250.01386.08400.48460.92516.16數(shù)據(jù)來源：轉(zhuǎn)摘自易丹輝《數(shù)據(jù)分析與Eviews的應(yīng)用》，中國統(tǒng)計(jì)出版社2002，P141。由于所用數(shù)據(jù)為時(shí)間序列數(shù)據(jù)，需要檢驗(yàn)其平穩(wěn)性，并用EG兩步法考

7、察它們之間是否存在協(xié)整關(guān)系。根據(jù)協(xié)整關(guān)系的檢驗(yàn)方法，首先回答人均可支配收入（SR）和生活費(fèi)支出（ZC）序列是否為非平穩(wěn)序列，即考察其單整階數(shù)。在Eviews中具體操作過程如下：在Eviews中建立文檔，錄入人均可支配收入（SR）和生活費(fèi)支出（ZC）序列的數(shù)D(SR(1))0.4093800.1089053.7590600.0003D(SR(2))0.3369980.1072733.1415020.0024C22.6360115.7591

8、91.4363690.1549Rsquared0.221103Meandependentvar5.952346AdjustedRsquared0.190756S.D.dependentvar60.73081S.E.ofregression54.63220Akaikeinfocriterion10.88725Sumsquaredresid229820.1Schwarzcriterion11.00549Loglikelihood436.93

9、34Fstatistic7.285920DurbinWatsonstat2.151282Prob(Fstatistic)0.000230為了得到人均可支配收入（SR）序列的單整階數(shù)，在單位根檢驗(yàn)（UnitRootTest）對話框（圖10.3）中，指定對一階差分序列作單位根檢驗(yàn)，選擇帶截距項(xiàng)（intercept），滯后差分項(xiàng)（Laggeddifferences）選2階，點(diǎn)擊OK，得到估計(jì)結(jié)果，見表10.5。圖10.3單位根檢驗(yàn)回歸方程設(shè)定

10、（一階差分序列）表10.5SR差分序列的ADF檢驗(yàn)結(jié)果ADFTestStatistic8.3743391%CriticalValue3.51325%CriticalValue2.897610%CriticalValue2.5858MacKinnoncriticalvaluesfrejectionofhypothesisofaunitroot.AugmentedDickeyFullerTestEquationDependentVariab