1、2000元 表8.5 城镇居民收入与消费的有关数据城镇家庭平均每人可支配收入(元)城镇家庭平均每人全年消费性支出平均每户城镇家庭就业人口数(人)城镇家庭平均每一就业者负担人数(含本人)1985739.1673.22.151.811986900.97992.121.819871002.1884.42.091.7919881180.211042.0319891373.9121121.7819901510.161278.891.981.7719911700.61453.81.961.7519922026.61671.71.951.7319932577.42110.81.921.7219943496.
2、22851.31.881.7419954282.953537.571.8719964838.93919.51.8619975160.34185.61.8319985425.14331.6199958544615.920006279.9849981.6820016859.65309.011.6520027702.86029.921.5820038472.26510.941.9120049421.67182.11.562005104937942.881.51200611759.458696.551.531.93200713785.819997.471.541.89200815780.811242.
3、91.481.97若模型设定为:Consumet=Ct+1Incomet+2Consumet-1+3Employmentt+4Burdent+5d1t+6d2t+7d3t+8d4t+t其中Consumet表示t期城镇居民家庭人均消费支出,Incomet表示t期城镇居民家庭人均可支配收入,Employmentt表示t期城镇居民家庭平均每户就业人口, Burdent表示t期城镇居民家庭平均每一就业者负担人数,dit(i=1,2,3,4)相应的虚拟变量。1)构造用于描述个人所得税调整的虚拟变量,并简要说明其理由;2)用散点图描述两两变量之间的关系,并给出你对模型设定的结论;3)依据测算,选择你认为更
4、能描述客观实际的模型,并简要说明其理由;4)根据分析结果,你对提高个人所得税起征点影响居民消费的有效性能得出什么结论?练习题8.2参考解答:1) 构造用于描述个人所得税调整的虚拟变量,并简要说明其理由个人所得税起征点调整年份为1987年、1994年、2006年及2008年,为了分析这些个人所得税起征点调整对居民消费支出会产生重要的影响,可分别研究居民消费行为在1987年、1994年、2006年及2008年等不同时期的变化情况,为此引入四个虚拟变量D1、 D2 、D3和D4,具体如下:2) 用散点图描述两两变量之间的关系,并给出你对模型设定的结论 由相关图可知,收入INCOME与消费CONSUM
5、E之间呈现一种线性关系,消费CONSUME与就业人口EMPLOYMENT之间表现为一种负的对数关系,而与每一就业者负担人数BURDEN线性相关程度相较而言不高,相关系数仅为0.7859。为此,将模型设定调整为CONSUMEt=C+a1INCOMEt+ a2CONSUMEt-1+ a3lnEMPLOYMENTt+ a4 BURDENt+ a5D1t+ a6D2t+ a7 D3t + a8 D4t+ 录入如下数据obsCONSUMEINCOMEEMPLOYMENTD1D2D3D4673.2000739.10002.1500000.000000799.0000900.90002.120000884.
6、40001002.1002.0900001.0000001104.0001180.2002.0300001211.0001373.9002.0000001278.8901510.1601.9800001453.8001700.6001.9600001671.7002026.6001.9500002110.8002577.4001.9200002851.3003496.2001.8800003537.5704282.9501.8700003919.5004838.9001.8600004185.6005160.3001.8300004331.6005425.1001.8000004615.900
7、5854.0001.7700004998.0006279.9801.6800005309.0106859.6001.6500006029.9207702.8001.5800006510.9408472.2007182.1009421.6001.5600007942.88010493.001.5100008696.5501.5300009997.4701.54000011242.9015780.801.480000应用最小二乘法估计回归模型Ls CONSUME C INCOME CONSUME(-1) LOG(EMPLOYMENT) BURDEN D1 D2 D3 D4估计结果如下: = 150
8、4.5250 + 0.6292*INCOME + 0.0883*CONSUME(-1) - 1086.2756*LOG(EMPLOYMENT) - 286.4018*BURDEN + 6.8829*D1 + 197.6548*D2 - 120.0845*D3 - 167.3091*D4尽管模型拟合优度较高,且整体线性关系显著,但在显著性水平0.10下,每一就业者负担人数BURDEN、CONSUMEt-1 、D1、D3的回归系数T检验不显著,为此,先删除变量每一就业者负担人数BURDEN,重新估计回归模型:Ls CONSUME C INCOME CONSUME(-1) LOG(EMPLOYMEN
9、T) D1 D2 D3 D4Dependent Variable: CONSUMEMethod: Least SquaresDate: 08/24/09 Time: 13:14Sample (adjusted): 1986 2008Included observations: 23 after adjustmentsVariableCoefficientStd. Errort-StatisticProb.C744.7966378.06621.9700170.0676CONSUME(-1)0.0848730.0509071.6672210.11620.6331180.03519817.98729
10、0.0000LOG(EMPLOYMENT)-762.9720478.5280-1.5944140.131737.4346050.234450.7451980.4677221.076538.308405.770966-122.049373.81439-1.6534610.1190-178.868865.87071-2.7154520.0160R-squared0.999861Mean dependent var4428.906Adjusted R-squared0.999796S.D. dependent var3060.917S.E. of regression43.70477Akaike i
11、nfo criterion10.66100Sum squared resid28651.61Schwarz criterion11.05595Log likelihood-114.6015F-statistic15413.79Durbin-Watson stat2.977604Prob(F-statistic)由上表可知,尽管模型拟合优度较高,且整体线性关系显著,但在显著性水平0.10下,变量lnEMPLOYMENT、CONSUMEt-1 、D1、D3的回归系数T检验不显著,为此,先删除变量D1,将D1归入D2中重新估计回归模型:Ls CONSUME C INCOME CONSUME(-1)
12、LOG(EMPLOYMENT) D2 D3 D4871.9310332.66272.6210670.01850.0835760.0501651.6660170.11520.6299220.03444718.28676-889.4616441.1508-2.0162300.0609226.036137.197916.076579-110.888471.26752-1.5559460.1393-171.692464.25105-2.6722110.01670.9998560.99980243.0931610.6104029712.3310.95598-115.019618496.742.78747
13、9由上表可知,尽管模型拟合优度较高,且整体线性关系显著,但在显著性水平0.10下,变量CONSUMEt-1 、D3的回归系数T检验不显著,为此,先删除变量D3,将D3归入D2和D4中重新估计回归模型:Ls CONSUME C INCOME CONSUME(-1) LOG(EMPLOYMENT) D2 D4151204.936265.10544.5451220.00030.0993140.0511471.9417090.06890.5991650.02936620.40320-1325.942354.4143-3.7412220.0016251.367534.815737.219940-141.
14、771063.81647-2.2215430.04020.9998340.99978544.8580210.6643434208.1210.96056-116.639920483.462.638666= 1204.936 + 0.599*INCOME + 0.0993*CONSUME(-1) - 1325.94*LOG(EMPLOYMENT) + 251.37*D2 - 141.77*D4 (265.1054) ( 0.0511) (0.0294) (354.4143) (34.8157)(63.8165) (4.5451) (1.9417) (20.4032) (-3.7412) (7.21
15、99) (-2.2215)R2=0.9998 F=20483.46 DW=2.6387估计结果经检验模型拟合优度较高,整体线性关系显著,不存在自相关性,在显著性水平0.10下,所有变量的回归系数T检验均显著,但在显著性水平0.05下,变量CONSUMEt-1的回归系数T检验不显著,可重新估计模型,结果如下:表明1994年及2008年个税调整对居民消费在截距上有影响。16Sample: 1985 2008 241460.937233.29226.2622630.6531010.00913271.51539-1651.937314.1501-5.258431277.404833.627838.24
16、9261-154.274266.05969-2.3353770.03060.9998104272.4180.9997693090.23946.9259810.7180741838.9110.96350-123.616924931.152.292463= 1460.937 +0.6531*INCOME -1651.937*LOG(EMPLOYMENT) + 277.4048*D2 - -154.2742*D4 (233.2922) (0.0091) (314.1501) (33.6278) (66.0597) (6.2623) (71.5154) (-5.2584) (8.2493) (-2.3354) R2=0.9998 F=24931.15 DW=2.2925估计结果经检验模型拟合优度较高,整体线性关系显著,不存在自相关性,且在显著性水平0.05下,所有变量的回归系数T检验均显著,表明1994年及2008年个税调整对居民消费在截距上有显著影响,收入INCOME与对数就业人口EMPLOYMENT对居民消费也有显著影响。8.3 在统计学教材中,采用了