资源描述
居民消费水平的影响因素实验报告第十二组组长:牧春枝组员:雷志豪 郑皓谦 李远致 姜婧怡 刘三科 李晗静 雷金祎一、摘要:居民消费水平是指居民在物质产品和劳务的消费过程中,对满足人们生存、发展和享受需要方面所达到的程度。通过消费的物质产品和劳务的数量和质量反映出来。居民消费作为社会再生产的终点和起点,对于实现社会再生产的良性循环,促进国民经济的持续发展具有决定性作用,要刺激消费,扩大内需,拉动经济发展,那么研究居民消费水平对于我国经济发展以及宏观调控起到很大的作用,因此居民消费水平具有研究性。通过文献参考发现居民消费水平的影响因素主要有五点:国内生产总值GDP、城镇居民可支配收入、农村居民可支配收入、人口自然增长率和居民消费价格指数,本文将分别总结以上因素的影响情况。本文运用计量分析的方法,研究了居民消费水平的影响因素。二、关键词:国内生产总值GDP、城镇居民可支配收入、农村居民可支配收入、人口自然增长率、居民消费价格指数。 在研究影响居民消费水平影响之前,首选我们要明白居民消费水平的意义。居民消费水平指按常住人口平均计算的居民消费支出。即是居民在物质产品和劳务的消费过程中,对满足人们生存、发展和享受需要方面所达到的程度。通过消费的物质产品和劳务的数量和质量反映出来。 在现实生活中,居民消费水平是受多方面因素影响因此来确定的一个量,所以要想了解居民消费水平就必须从多个影响因素进行分析。国内生产总值GDP 常被公认为是衡量国家经济状况的最佳指标,它不但可以反映一个国家的经济表现,更可以反映一国的国力与财富。随着国民经济的发展,人民收入水平不断提高。国内生产总值增加,意味着国民经济水平提高,居民收入增加,居民的消费能力提升,消费水平随之提高。由此选择了国内生产总值GDP 作为居民消费水平计量分析的因素之一。 国内生产总值GDP是从整个宏观方面来对居民消费水平的整体水平做出个大致的分析,从微观层面上分析,主要的影响因素就是居民可支配收入,在我国居民由城镇居民与农村居民两部分组成,所以城镇居民可支配收入与农村居民可支配收入也是影响居民消费水平的因素。收入是消费的基础和前提;其他条件不变,则人们当前可支配收入(当前收入)越多,对各种商品和服务的消费量就越大。因此要提高居民的生活水平,必须保持经济的稳定增长,增加居民收入。一般地,未来预期收入(未来收入)越高,预期支出的可能性就会越大。社会收入差距与社会总体消费水平有密切的联系。人们的收入差距过大,总体消费水平会降低;反之,收入差距缩小,会使总体消费水平提高。 农村居民人均可支配收入对居民消费水平的影响大大超过了城镇居民人均可支配收入对居民消费水平的影响。主要原因:第一是我国是农民人口占绝大多数的国家,而居民消费水平是以人口数为权数对农村居民消费水平和城镇居民消费水平进行加权平均计算而得到的;第二是农村居民的消费动力远远大于城镇居民。 人口自然增长率,是反映人口发展速度和制定人口计划的重要指标,也是计划生育统计中的一个重要指标,它表明人口自然增长的程度和趋势。在人口数量一定的情况下,经济发展水平越高,消费品数量越多,那么居民消费水平就会越高;反之,在经济发展水平稳定的条件下,人口数量的多少就决定着消费水平的高低。 居民消费价格指数,是一个反映居民家庭一般所购买的消费商品和服务价格水平变动情况的宏观经济指标。它是度量一组代表性消费商品及服务项目的价格水平随时间而变动的相对数,是用来反映居民家庭购买消费商品及服务的价格水平的变动情况。消费物价指数对居民根据经济理论分析,物价越高,越会抑制人们的消费,消费水平会越低。以经济建设为中心,大力发展生产力,落实科学发展观,使国民经济又好又快发展。增加居民收入,农民收入,完善社保,缩小城乡差距。国家加强宏观调控,稳定物价。首先,要保持GDP稳速增长,同时,政府应当加大资金和物质投入,扩大社会保障的资金来源,增加融资来源和渠道,有效地促进社会保障制度的运行和社会保障体系的建设。 其次,应合理制定税收政策,积极推进税制改革,减轻中低收入者的税收负担,充分发挥税收对收入和消费的调节作用,通过提高个人所得税的起征点和对高收入群体征收较高的个人所得税,促进收入的相对公平。 三、基于计量经济模型的居民消费影响因素分析1、变量的选择与设计Y-居民消费水平X1-国内生产总值X2-人口自然增长率X3-居民消费价格指数(上年=100)X4-城镇居民家庭人均可支配收入X5-农村居民家庭人均纯收入2、模型的设立Y=0+1X1+2X2+3X3+4X4+5X5回归结果:Dependent Variable: YMethod: Least SquaresDate: 12/01/14 Time: 15:08Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C66.06162215.65380.3063320.7615X10.0072830.0013485.4028660.0000X2-7.5834609.544774-0.7945140.4334X3-0.3277672.011271-0.1629650.8717X40.1033700.0436012.3708020.0246X50.9749710.06820414.294810.0000R-squared0.999806Mean dependent var3896.143Adjusted R-squared0.999772S.D. dependent var4226.517S.E. of regression63.80517Akaike info criterion11.30435Sum squared resid118061.9Schwarz criterion11.57098Log likelihood-191.8261Hannan-Quinn criter.11.39639F-statistic29831.70Durbin-Watson stat1.158136Prob(F-statistic)0.0000003、模型检验一,模型检验由以上回归结果可看出:1, 经济意义检验X2与理论分析和经验判断不一2, 拟合优度检验可决系数和修正的可决系数,拟合很好3, F检验显著性水平0.05下,F(6,29)=2.545 F=29831.72.545 回归方程显著4, t检验显著性水平0.05下,t(29)=2.045 2和3不能通过检验二,多重共线性检验:可决系数R2较高,经F检验的 参数联合显著性也很高,但X2和X3的系数不显著,且X2的符号与预期相反,这表明可能存在严重的多重共线性。1, 相关系数检验法X1X2X3X4X5X11.000000-0.800638-0.2644240.9960660.993517X2-0.8006381.0000000.409727-0.842008-0.829955X3-0.2644240.4097271.000000-0.280410-0.279674X40.996066-0.842008-0.2804101.0000000.997214X50.993517-0.829955-0.2796740.9972141.000000由上述相关系数矩阵可以看出确实存在一定的多重共线性2,方差扩大因子检验法分别以X1,X2,X3,X4,X5为被解释变量,做关于其他解释变量的回归分析,结果Dependent Variable: X1Method: Least SquaresDate: 12/01/14 Time: 16:20Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C-61641.1626953.42-2.2869520.0294X25762.032751.35597.6688440.0000X3-288.1224267.2832-1.0779660.2896X428.350852.8428259.9727750.0000X5-14.579188.845958-1.6481180.1098R-squared0.997432Mean dependent var129747.2Adjusted R-squared0.997089S.D. dependent var160186.0S.E. of regression8641.897Akaike info criterion21.09820Sum squared resid2.24E+09Schwarz criterion21.32039Log likelihood-364.2184Hannan-Quinn criter.21.17490F-statistic2912.951Durbin-Watson stat0.871799Prob(F-statistic)0.000000Dependent Variable: X2Method: Least SquaresDate: 12/01/14 Time: 16:22Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C6.8846823.9288941.7523200.0899X10.0001151.50E-057.6688440.0000X30.0746470.0359772.0748610.0467X4-0.0037170.000485-7.6642340.0000X50.0026950.0012082.2302390.0334R-squared0.918477Mean dependent var9.882000Adjusted R-squared0.907607S.D. dependent var4.015221S.E. of regression1.220477Akaike info criterion3.367924Sum squared resid44.68692Schwarz criterion3.590117Log likelihood-53.93867Hannan-Quinn criter.3.444625F-statistic84.49805Durbin-Watson stat1.032575Prob(F-statistic)0.000000Dependent Variable: X3Method: Least SquaresDate: 12/01/14 Time: 16:23Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C83.2894012.328256.7559780.0000X1-0.0001290.000120-1.0779660.2896X21.6811450.8102442.0748610.0467X40.0047490.0038621.2296240.2284X5-0.0047270.006131-0.7710960.4467R-squared0.218961Mean dependent var105.3943Adjusted R-squared0.114822S.D. dependent var6.156152S.E. of regression5.791948Akaike info criterion6.482378Sum squared resid1006.400Schwarz criterion6.704570Log likelihood-108.4416Hannan-Quinn criter.6.559079F-statistic2.102588Durbin-Watson stat1.019268Prob(F-statistic)0.105193Dependent Variable: X4Method: Least SquaresDate: 12/01/14 Time: 16:24Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C1605.419854.13011.8795960.0699X10.0270980.0027179.9727750.0000X2-178.104923.23844-7.6642340.0000X310.104338.2174121.2296240.2284X51.1100330.2012305.5162330.0000R-squared0.998877Mean dependent var7010.643Adjusted R-squared0.998727S.D. dependent var7489.172S.E. of regression267.1760Akaike info criterion14.14526Sum squared resid2141491.Schwarz criterion14.36745Log likelihood-242.5420Hannan-Quinn criter.14.22196F-statistic6671.194Durbin-Watson stat0.889682Prob(F-statistic)0.000000Dependent Variable: X5Method: Least SquaresDate: 12/01/14 Time: 16:25Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C-201.4293576.1034-0.3496410.7291X1-0.0056950.003455-1.6481180.1098X252.7754023.663572.2302390.0334X3-4.1109635.331326-0.7710960.4467X40.4536340.0822365.5162330.0000R-squared0.995234Mean dependent var2328.203Adjusted R-squared0.994598S.D. dependent var2323.844S.E. of regression170.7978Akaike info criterion13.25040Sum squared resid875156.4Schwarz criterion13.47259Log likelihood-226.8820Hannan-Quinn criter.13.32710F-statistic1566.006Durbin-Watson stat0.388839Prob(F-statistic)0.000000由方差扩大因子VIF=10判断,该模型存在严重多重共线性问题。3,对多重共线性的处理。变换模型形式:对所有变量取对数得:Dependent Variable: LNYMethod: Least SquaresDate: 12/01/14 Time: 16:34Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C-1.1789680.357753-3.2954820.0026LNX10.2545160.1110382.2921670.0293LNX20.0831980.0375152.2177420.0346LNX3-0.0325300.099234-0.3278150.7454LNX40.4835020.1099294.3983220.0001LNX50.2795360.0627194.4569360.0001R-squared0.999752Mean dependent var7.567862Adjusted R-squared0.999709S.D. dependent var1.324190S.E. of regression0.022575Akaike info criterion-4.589165Sum squared resid0.014779Schwarz criterion-4.322534Log likelihood86.31038Hannan-Quinn criter.-4.497124F-statistic23391.41Durbin-Watson stat1.339966Prob(F-statistic)0.000000有上述结果可看出该模型中LNX3对应的系数不能通过t检验,LNX3不显著,又由于对数模型误差相对较大,故综合而言,仍采用原有模型。,对原模型逐步回归一元回归分别做Y关于X1,X2 ,X3 ,X4 ,X5的一元回归,根据回归结果比较可决系数,引入X5二元回归分别做Y关于X5和X1,X5和X2,X5和X3,X5和X4的二元回归,根据回归结果比较修正的可决系数,再引入X1三元回归分别做Y关于X1,X5和X2;X1,X5和X3;X1,X5和X4的三元回归,根据回归结果比较修正的可决系数,再引入X4四元回归分别做Y关于X1,X4,X5和X2以及X1,X4,X5和X3的四元回归,结果如下:Dependent Variable: YMethod: Least SquaresDate: 12/01/14 Time: 17:23Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C38.76206133.58880.2901600.7737X10.0073250.0013015.6306870.0000X2-8.1344858.779795-0.9265010.3616X40.1018130.0418462.4330250.0211X50.9765200.06643414.699170.0000R-squared0.999805Mean dependent var3896.143Adjusted R-squared0.999779S.D. dependent var4226.517S.E. of regression62.76145Akaike info criterion11.24812Sum squared resid118170.0Schwarz criterion11.47032Log likelihood-191.8422Hannan-Quinn criter.11.32482F-statistic38540.17Durbin-Watson stat1.154076Prob(F-statistic)0.000000Dependent Variable: YMethod: Least SquaresDate: 12/01/14 Time: 17:24Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C13.85191204.13200.0678580.9463X10.0064110.0007798.2343640.0000X3-0.8938531.869247-0.4781890.6360X40.1315540.0251955.2214610.0000X50.9545350.06277915.204640.0000R-squared0.999801Mean dependent var3896.143Adjusted R-squared0.999775S.D. dependent var4226.517S.E. of regression63.41182Akaike info criterion11.26874Sum squared resid120631.8Schwarz criterion11.49093Log likelihood-192.2030Hannan-Quinn criter.11.34544F-statistic37753.52Durbin-Watson stat1.141169Prob(F-statistic)0.000000由于引入X2或X3,都使模型存在不能通过t检验的问题故逐步回归的结果是最终保留X1, X4, X5。模型改为Y=0+1X1+4X4+5X5三,异方差性检验:1,White检验Heteroskedasticity Test: WhiteF-statistic4.557144Prob. F(9,25)0.0013Obs*R-squared21.74530Prob. Chi-Square(9)0.0097Scaled explained SS14.92853Prob. Chi-Square(9)0.0929Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 12/01/14 Time: 20:12Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C5605.1813026.3151.8521470.0758X12-6.33E-074.79E-06-0.1320920.8960X1*X40.0005000.0002482.0192480.0543X1*X5-0.0015440.000510-3.0270190.0057X10.4155350.3132561.3265030.1967X42-0.0123720.005331-2.3205790.0288X4*X50.0441620.0227121.9444040.0632X44.54858810.958180.4150860.6816X52-0.0163990.029419-0.5574220.5822X5-36.5037824.10651-1.5142710.1425R-squared0.621294Mean dependent var3472.893Adjusted R-squared0.484960S.D. dependent var4661.580S.E. of regression3345.443Akaike info criterion19.30354Sum squared resid2.80E+08Schwarz criterion19.74793Log likelihood-327.8120Hannan-Quinn criter.19.45694F-statistic4.557144Durbin-Watson stat2.171776Prob(F-statistic)0.001251NR2=21.74532(9)=16.9190,故存在异方差(二),由于上列t检验结果表明上述模型3个变量均可能存在异方差,变换模型形式:LNY =0 + 1LNX1 +2 LNX2 +3LNX3 +4 LNX4 +5 LNX5一,多重共线性由上文可知该模型存在多重共线性问题,经逐步回归剔除变量LNX3,回归结果如下Dependent Variable: LNYMethod: Least SquaresDate: 12/01/14 Time: 22:42Sample: 1979 2013Included observations: 35VariableCoefficientStd. Errort-StatisticProb.C-1.2764400.195955-6.5139420.0000LNX10.2401650.1005132.3893960.0234LNX20.0746820.0266572.8015740.0088LNX40.4885080.1072314.5556480.0001LNX50.2905750.0521215.5749750.0000R-squared0.999751Mean dependent var7.567862Adjusted R-squared0.999718S.D. dependent var1.324190S.E. of regression0.022236Akaike info criterion-4.642609Sum squared resid0.014834Schwarz criterion-4.420416Log likelihood86.24566Hannan-Quinn criter.-4.565908F-statistic30135.81Durbin-Watson stat1.301994Prob(F-statistic)0.000000由以上结果可看该模型LNY =0 + 1LNX1 +2 LNX2 +4 LNX4 +5 LNX5不再存在多重共线问题。二,异方差性检验1,White检验Heteroskedasticity Test: WhiteF-statistic0.504366Prob. F(13,21)0.8972Obs*R-squared8.327780Prob. Chi-Square(13)0.8216Scaled explained SS4.825566Prob. Chi-Square(13)0.9788Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 12/01/14 Time: 22:27Sample: 1979 2013Included observations: 35Collinear test regressors dropped from specificationVariableCoefficientStd. Errort-StatisticProb.C0.2807320.3170120.8855550.3859LNX120.0385680.0451340.8545300.4025LNX1*LNX20.0140300.0290450.4830560.6341LNX1*LNX4-0.0706270.058710-1.2029700.2424LNX1*LNX5-0.0214750.058737-0.3656160.7183LNX1-0.1307900.227483-0.5749430.5714LNX220.0066260.0091780.7219150.4783LNX2*LNX4-0.0066220.034156-0.1938720.8481LNX2*LNX5-0.0056800.014646-0.3878390.7020LNX2-0.0880960.094697-0.9302930.3628LNX4*LNX50.0955510.0754361.2666530.2191LNX40.0844430.1941730.4348830.6681LNX52-0.0405720.036667-1.1064990.2810LNX50.0507800.1074850.4724430.6415R-squared0.237937Mean dependent var0.000424Adjusted R-squared-0.233817S.D. dependent var0.000540S.E. of regression0.000600Akaike info criterion-11.71047Sum squared resid7.56E-06Schwarz criterion-11.08833Log likelihood218.9332Hannan-Quinn criter.-11.49571F-statistic0.504366Durbin-Watson stat2.370349Prob(F-statistic)0.897163nR2=8.3278X(13)=22.3621,不存在异方差三,自相关检验1, DW检验对模型估计结果的DW统计量为1.3020,5%的显著性水平下,查DW统计表可知,dl=1.222,du=1.728,模型中dlDWdu,显然不能判断模型是否有自相关。2, 残差3,LM检验 Breusch-Godfrey Serial Correlation LM Test:F-statistic2.333364Prob. F(2,28)0.1155Obs*R-squared5.000055Prob. Chi-Square(2)0.0821Test Equation:Dependent Variable: RESIDMethod: Least SquaresDate: 12/01/14 Time: 23:01Sample: 1979 2013Included observations: 35Presample missing value lagged residuals set to zero.VariableCoefficientStd. Errort-StatisticProb.C-0.1620250.211571-0.7658200.4502LNX10.1221610.1184811.0310550.3113LNX20.0034610.0257810.1342370.8942LNX4-0.1405220.129178-1.0878100.2860LNX5-0.0031290.050028-0.0625450.9506RESID(-1)0.4351940.2051132.1217280.0428RESID(-2)0.0376490.2014830.1868590.8531R-squared0.142859Mean dependent var-1.19E-15Adjusted R-squared-0.040814S.D. dependent var0.020887S.E. of regression0.021309Akaike info criterion-4.682476Sum squared resid0.012715Schwarz criterion-4.371406Log likelihood88.94333Hannan-Quinn criter.-4.575095F-statistic0.777788Durbin-Watson stat1.835787Prob(F-statistic)0.594216上图显示LM=TR2=5.000055,其P值为0.0821,表明不存在自相关。(三),综上:我国居民消费水平的影响因素模型是:LNY=-1.2764+0.2402LNX1+0.0747LNX2+0.4885LNX4+0.2906LNX5四、模型应用1、整理与分析 由模型分析可知,居民消费水平与国内生产总值X1,人口自然增长率X2,城镇居民家庭人均可支配收入X4,农村家庭人均纯收入X5都有很大关系。值得注意的是,城镇和农村居民家庭人均收入,对于居民消费水平的影响尤其大。而居民消费价格指数就影响比较微小了,故剔除。 2、思考与建议 在模型分析中,我们可以看出城镇和农村居民的收入以及可支配收入对居民消费的影响很大,所以显而易见,在扩大内需、促进消费的进程中,努力发展经济产业、带动就业、提高居民收入水平是不可忽视的。但是从现实中来看,居民收入水平和居民可支配收入水平是两个有区别又有联系的概念,其影响因素不仅包括经济层面的还包括社会层面的。经济层面的因素比较容易估量,主要包括储蓄,商品价格,通货膨胀系数等等。社会层面的因素则比较不容易估量,而且在消费支出影响中的比重也占比较大,比如居住地区、医疗社会保障程度、家庭人口状况、受教育程度等等,这些因素常难以用数学方法准确计量,但对消费水平的影响又比较重大。 所以,针对此种情况,提出建议如下:(一) 切实发展国家经济,努力提高居民收入水平。这是最根本的发展之道。(二) 我国当前城镇与农村发展较不平衡,应该共同发展,以城带村,共同进步。(三) 发展农业科技,提高劳动生产率,提高农产品价值附加值,改善农村经济结构,发展第二第三产业。(四) 进一步改善整体的消费环境,完善基础设施建设,拉动社会特别是农村消费观念进步。(五) 加大政策鼓励力度,提高社会保障水平,建立健全全社会的医疗保障体系,使人民后顾之忧减少,可支配收入自然增多。
展开阅读全文