《计量经济学论文最终版.doc》由会员分享，可在线阅读，更多相关《计量经济学论文最终版.doc（5页珍藏版）》请在三一办公上搜索。
1、计量经济学计量经济学论文盐城师范学院数学科学学院统计122班12213238刘佳2014年12月28日星期日对2003年的LEGL这项服务的需求进行预测美国人口普查局需要你通过分析消费者支出的相关数据来帮助他们预测对某些商品和服务的总需求。而我本人刘佳，被分到LEGL这项服务。1 模型建立: log(ljlegl) =c +log(ljtpe(-1)+ log(ljpop(-1) +log(ljrplegl(-1) +log(t(-1)建立结果如下图所示：Dependent Variable: LOG(LJLEGL)Method: Least SquaresDate: 01/05/15 Tim
2、e: 12:08Sample (adjusted): 1960 2002Included observations: 43 after adjustmentsVariableCoefficientStd. Errort-StatisticProb.C23.084015.8323173.9579490.0003LOG(LJTPE(-1)0.6806570.1967393.4596910.0014LOG(LJPOP(-1)-1.8967700.526326-3.6037950.0009LOG(LJRPLEGL(-1)0.3886700.1132663.4314720.0015LOG(T(-1)0.
3、1193880.0186886.3886370.0000R-squared0.986416Mean dependent var4.288345Adjusted R-squared0.984986S.D. dependent var0.264928S.E. of regression0.032462Akaike info criterion-3.908553Sum squared resid0.040044Schwarz criterion-3.703762Log likelihood89.03389Hannan-Quinn criter.-3.833032F-statistic689.8528
4、Durbin-Watson stat0.821206Prob(F-statistic)0.000000LOG(LJPOP(-1)为负的，说明虽然人口增长率变小，但是人口还是在增加的，所以，对LEGL这项服务需求的增加是正常的符合经济意义的！LOG(LJTPE(-1)是正的，说明随着个人收如的不断增加，对LEGL这项服务的需求也是在不断增加的，这也是符合经济意义的。LOG(LJRPLEGL(-1)是正的，说明LEGL的价格指数的增长率是在不断增加的，而LEGL也是在不断增加的，符合经济意义。2 方程检验2.1多重共线检验发现这4个变量是共线程度非常严重。LJDPILJPOPLJRPLEGLLJT
5、PELJDPI1.0000000.9964640.9913960.998335LJPOP0.9964641.0000000.9847260.992538LJRPLEGL0.9913960.9847261.0000000.990217LJTPE0.9983350.9925380.9902171.0000002.2方差检验进行怀特检验，通过检验，检验结果如图所示。Heteroskedasticity Test: WhiteF-statistic1.416820Prob. F(11,31)0.2147Obs*R-squared14.38565Prob. Chi-Square(11)0.2124Sca
6、led explained SS7.343762Prob. Chi-Square(11)0.7706Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 01/05/15 Time: 12:06Sample: 1960 2002Included observations: 43Collinear test regressors dropped from specificationVariableCoefficientStd. Errort-StatisticProb.C3.8837175.2161880.74455
7、10.4622LOG(LJTPE(-1)20.1512840.0976581.5491190.1315LOG(LJTPE(-1)*LOG(LJPOP(-1)0.1296760.2166320.5985980.5538LOG(LJTPE(-1)*LOG(LJRPLEGL(-1)-0.2795170.210063-1.3306350.1930LOG(LJTPE(-1)*LOG(T(-1)-0.0601450.038927-1.5450650.1325LOG(LJTPE(-1)-1.4771882.248751-0.6568930.5161LOG(LJPOP(-1)2-0.0305810.04000
8、6-0.7644000.4504LOG(LJPOP(-1)*LOG(LJRPLEGL(-1)0.0275170.0279810.9834230.3330LOG(LJPOP(-1)*LOG(T(-1)-0.0115300.009608-1.1999470.2393LOG(LJRPLEGL(-1)20.0598190.1023280.5845770.5631LOG(LJRPLEGL(-1)*LOG(T(-1)0.0903710.0593421.5228890.1379LOG(T(-1)20.0088480.0039602.2343250.0328R-squared0.334550Mean depe
9、ndent var0.000931Adjusted R-squared0.098423S.D. dependent var0.001077S.E. of regression0.001023Akaike info criterion-10.70126Sum squared resid3.24E-05Schwarz criterion-10.20976Log likelihood242.0770Hannan-Quinn criter.-10.52001F-statistic1.416820Durbin-Watson stat2.147269Prob(F-statistic)0.2147172.3
10、相关性检验进行拉格朗日检验，在方程后面加上ar（1）模型。检验通过！Breusch-Godfrey Serial Correlation LM Test:F-statistic1.308001Prob. F(1,35)0.2605Obs*R-squared1.513056Prob. Chi-Square(1)0.2187Test Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 01/05/15 Time: 12:06Sample: 1961 2002Included observations: 42Presample m
11、issing value lagged residuals set to zero.VariableCoefficientStd. Errort-StatisticProb.C-2.3687618.595524-0.2755810.7845LOG(LJTPE(-1)-0.0668890.280311-0.2386250.8128LOG(LJPOP(-1)0.2112960.7772840.2718390.7873LOG(LJRPLEGL(-1)-0.0001180.172219-0.0006870.9995LOG(T(-1)0.0080540.0786540.1024010.9190AR(1)
12、-0.2226040.243166-0.9154420.3662RESID(-1)0.3295830.2881781.1436790.2605R-squared0.036025Mean dependent var-4.16E-10Adjusted R-squared-0.129228S.D. dependent var0.025242S.E. of regression0.026823Akaike info criterion-4.248103Sum squared resid0.025182Schwarz criterion-3.958492Log likelihood96.21017Hannan-Quinn criter.-4.141949F-statistic0.218000Durbin-Watson stat1.965305Prob(F-statistic)0.9684813 模型进行预测预测区间为（100.5357， 115.2743）真实值为：103.3754 预测好坏判断：真实值在预测区间内，所以方程预测的还不错，对于决策者有一定的帮助作用。第 5 页 共 5 页
