《北京市城镇居民消费模型.doc》由会员分享,可在线阅读,更多相关《北京市城镇居民消费模型.doc(5页珍藏版)》请在三一办公上搜索。
1、北京市城镇居民消费模型北京市城镇居民消费模型一切经济活动的目的都是为了满足人们不断增长的消费需求,消费活动是经济活动的起点,也是经济活动的终点,是推动经济增长的真正和持久的拉动力,这里仅从一个侧面来说明我国居民的消费行为。1 模型变量的选择经济社会中,影响消费的因素有很多,如:收入水平、收入分配情况、家庭财产情况、商品价格等等。在我国,国内生产总值是居民消费的重要影响因素,而且居民消费的多少很大程度上取决于居民收入情况,居民的储蓄也直接影响到消费支出。2 样本数据的及理论模型以t代表年份,c代表北京市城镇居民年人均消费额,y代表市年人均国内生产总值,I代表市城镇人均可支配收入,s代表市城镇居民
2、年人均储蓄余额。(数据来源:北京统计年鉴)消费模型样本数据 元年份t居民消费c国内生产总值y可支配收入I储蓄余额s1978359.861290365.4185.81979408.661391414.95204.711980490.441582501.36255.851981511.431558514.14295.311982534.821704561.05352.761983574.061977590.47450.811984666.752308693.7563.861985923.322704907.72720.8219861067.3829551067.52895.6519871147.6
3、33381181.871180.3819881455.5541251436.971393.0819891520.4144991787.082014.3119901646.0548811787.082793.9119911860.1757812040.433658.5719922134.6668052363.684742.9219932939.682403296.046824.3319944134.12102654731.241028819955019.76130735868.3613638.0419965729.45150446885.4818436.7919976531.8116735781
4、3.1121439.47利用以上观测值,由此可得该模型的理论方程:C=0+1Y+2I+3S+ (1)其中,i 为待估参数,i=0,1,2,3;为随机变量。3.计量经济模型的参数估计与和检验:A 首先,对(1)由OLS法估计得:Dependent Variable:cMethod: Least SquaresDate: 5/3/04 Time: 19:27Sample: 1978 1997Included observations: 20VariableCoefficientStd. Errort-StatisticProb. C0-6.32739933.77952-0.1873150.8538
5、Y0.0539610.0312841.7248970.1038I0.8579750.0919149.3345240.0000S-0.0510410.015858-3.2185600.0054R-squared0.999433 Mean dependent var1982.795Adjusted R-squared0.999327 S.D. dependent var1893.425S.E. of regression49.12271 Akaike info criterion10.80338Sum squared resid38608.65 Schwarz criterion11.00252L
6、og likelihood-104.0338 F-statistic9404.109Durbin-Watson stat2.375901 Prob(F-statistic)0.000000查表知t0.01(16)=2.583,F0.01(3,16)=5.29,常数项与变量y 的系数不能通过t检验。去掉变量Y ,修正模型如下:C=0+2I+3S+ (2)再次用OLS估计得:Dependent Variable: CMethod: Least SquaresDate: 5/3/04 Time: 19:31Sample: 1978 1997Included observations: 20Varia
7、bleCoefficientStd. Errort-StatisticProb. C021.5087031.352010.6860390.5019I0.9998670.04332123.080570.0000S-0.0617250.015424-4.0019460.0009R-squared0.999328 Mean dependent var1982.795Adjusted R-squared0.999249 S.D. dependent var1893.425S.E. of regression51.89814 Akaike info criterion10.87392Sum square
8、d resid45788.10 Schwarz criterion11.02328Log likelihood-105.7392 F-statistic12636.42Durbin-Watson stat2.321029 Prob(F-statistic)0.000000查表得:t0.01(17)=2.567,F0.01(2,17)=6.11,对模型进行统计检验,只有常数项不能通过t检验,暂时先不做处理,从经济意义上检验,参数2在经济理论中表示边际消费倾向,预期值在0,1之间,2的估计值=0.9999是符合理论预期的。3的估计值=-0.0617,符号为负,表明居民储蓄增加,会使居民消费额下降,
9、这符合实际生活情况。B 随机误差项的序列相关性检验:利用DW检验,样本容量n=20,解释变量个数k=2,可查表得dl=1.2,du=1.41,从而(du=1.41)(DW=2.32)(4-du=2.59),说明模型不存在一阶自相关性。C 异方差性检验:做三期滞后的ARCH检验:ARCH Test:F-statistic0.149239 Probability0.928330Obs*R-squared0.565983 Probability0.904174Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 5/3/
10、04 Time: 19:40Sample(adjusted): 1981 1997Included observations: 17 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. C2594.9292062.3701.2582270.2304RESID2(-1)0.1416980.2772480.5110900.6179RESID2(-2)-0.1367310.276914-0.4937650.6297RESID2(-3)0.0124170.2760580.0449800.9648R-squared
11、0.033293 Mean dependent var2649.652Adjusted R-squared-0.189793 S.D. dependent var6421.104S.E. of regression7003.989 Akaike info criterion20.74867Sum squared resid6.38E+08 Schwarz criterion20.94472Log likelihood-172.3637 F-statistic0.149239Durbin-Watson stat2.000002 Prob(F-statistic)0.928330由X0.05(3)
12、=7.81,(n-p)*r=0.565983,故不存在异方差性。D 模型的共线性检验:居民的可支配收入I与居民储蓄余额S间存在相关关系的可能性不能被排除。下面采用一阶差分法消除共线性。将模型化为差分模型:Ct=2It+3St (3)进行OLS估计得到Dependent Variable: DCMethod: Least SquaresDate: 5/3/04 Time: 20:36Sample(adjusted): 1979 1997Included observations: 19 after adjusting endpointsVariableCoefficientStd. Errort
13、-StatisticProb. DI0.8846330.0968219.1367920.0000DS-0.0274690.031363-0.8758250.3933R-squared0.958312 Mean dependent var324.8395Adjusted R-squared0.955860 S.D. dependent var361.3966S.E. of regression75.92787 Akaike info criterion11.59675Sum squared resid98005.70 Schwarz criterion11.69616Log likelihood
14、-108.1691 Durbin-Watson stat3.042976St的系数不能通过t检验,所以将模型(2)中的储蓄变量s删除,得到:C=0+2I (4)进行OLS估计得到:Dependent Variable: YMethod: Least SquaresDate: 5/3/04 Time: 20:38Sample: 1978 1997Included observations: 20VariableCoefficientStd. Errort-StatisticProb. C128.302322.288345.7564730.0000I0.8277570.007054117.3454
15、0.0000R-squared0.998695 Mean dependent var1982.795Adjusted R-squared0.998622 S.D. dependent var1893.425S.E. of regression70.28700 Akaike info criterion11.43769Sum squared resid88924.72 Schwarz criterion11.53726Log likelihood-112.3769 F-statistic13769.95Durbin-Watson stat1.193999 Prob(F-statistic)0.0
16、00000这便是西方经济学中绝对收入假说下的消费函数模型。该模型的各项检测指标都显著,故该模型也可行。4联系实际但若考虑到1989年前后,我国的经济发生了很大的变化,会影响居民消费行为,仍可将模型进一步改进。1989年以前,我国经济处于低产出、低消费的状态,市场上商品极为短缺,居民消费普遍不足,1989年以后,人们的可支配收入大幅度提高,消费水平也有所提高。所以考虑将19781997年分为两个时期:19781988,19891997,分阶段建模:(!)19781988年: C=2I 做OLS估计得到:Dependent Variable: CMethod: Least SquaresDate:
17、 5/3/04 Time: 20:41Sample: 1978 1988Included observations: 11VariableCoefficientStd. Errort-StatisticProb. I0.9926400.006624149.85440.0000R-squared0.997432 Mean dependent var739.9882Adjusted R-squared0.997432 S.D. dependent var355.3521S.E. of regression18.00874 Akaike info criterion8.706099Sum squar
18、ed resid3243.146 Schwarz criterion8.742272Log likelihood-46.88355 Durbin-Watson stat2.183329(2)19891997年: C=0+2I 做OLS估计得到:Method: Least SquaresCDate: 5/3/04 Time: 20:48Sample: 1989 1997Included observations: 9VariableCoefficientStd. Errort-StatisticProb. C186.527750.252463.7118120.0075I0.8158390.010
19、87275.039720.0000R-squared0.998758 Mean dependent var3501.781Adjusted R-squared0.998581 S.D. dependent var1907.132S.E. of regression71.83974 Akaike info criterion11.57988Sum squared resid36126.64 Schwarz criterion11.62371Log likelihood-50.10947 F-statistic5630.960Durbin-Watson stat1.017241 Prob(F-statistic)0.0000005总结从以上的分析可以看到,用计量经济模型,结合特定的历史时期,以特殊的国情为背景,引用经济理论模型作指导和检验,综合运用经济学、统计学、计算机等各种工具和方法,对模型进行不断的修正和改进,最终得到与经济理论相一致的的模型,这对经济政策的制定都有一定的意义。