计量经济学论文对我国财政收入影响因素的分析（可编辑） .doc

对我国财政收入影响因素的分析 班级: 农经1102 姓名:荆文珊 学号:112505311 摘要:对我国财政收入影响因素进行了定量分析,建立数学模型,并提出了提高我国财政收入质量的政策建议。 关键词:财政收入 实证分析 影响因素引言 财政收入对于国民经济的运行及社会发展具有重要影响。首先,它是一个国家各项收入得以实现的物质保证。一个国家财政收入规模大小往往是衡量其经济实力的重要标志。其次,财政收入是国家对经济实行宏观调控的重要经济杠杆。宏观调控的首要问题是社会总需求与总供给的平衡问题,实现社会总需求与总供给的平衡,包括总量上的平衡和结构上的平衡两个层次的内容。财政收入的杠杆既可通过增收和减收来发挥总量调控作用,也可通过对不同财政资金缴纳者的财政负担大小的调整,来发挥结构调整的作用。此外,财政收入分配也是调整国民收入初次分配格局,实现社会财富公平合理分配的主要工具。在我国,财政收入的主体是税收收入。因此,在税收体制及政策不变的情况下,财政收入会随着经济繁荣而增加,随着经济衰退而下降。 我国的财政收入主要包括税收、国有经济收入、债务收入以及其他收入四种形式,因此,财政收入会受到不同因素的影响。从国民经济部门结构看,财政收入又表现为来自各经济部门的收入。财政收入的部门构成就是在财政收入中,由来自国民经济各部门的收入所占的不同比例来表现财政收入来源的结构,它体现国民经济各部门与财政收入的关系。我国财政收入主要来自于工业、农业、商业、交通运输和服务业等部门。 因此,本文认为财政收入主要受到总税收收入、国内生产总值、其他收入和就业人口总数的影响。二、预设模型 令财政收入Y(亿元)为被解释变量,总税收收入X1(亿元)、国内生产总值X2(亿元)、其他收入X3(亿元)、就业人口总数为X4(万人)为解释变量,据此建立回归模型。数据收集 从2010中国统计年鉴得到1990-2009年每年的财政收入、总税收收入、国内生产总值工、其他收入和就业人口总数的统计数据如下:obs财政收入Y总税收收入X1国内生产总值X2其他收入X3就业人口总数X419902937.12821.8618667.8299.536474919913149.482990.1721781.5240.16549119923483.373296.9126923.5265.156615219934348.954255.335333.9191.046680819945218.15126.8848197.9280.186745519956242.26038.0460793.7396.196806519967407.996909.8271176.6724.666895019978651.148234.0478973682.36982019989875.959262.884402.3833.370637199911444.0810682.5889677.1925.4371394200013395.2312581.5199214.6944.9872085200116386.0415301.38109655.21218.173025200218903.6417636.45120332.71328.7473740200321715.2520017.31135822.81691.9374432200426396.4724165.68159878.32148.3275200200531649.2928778.54184937.42707.8375825200638760.234804.35216314.43683.8576400200751321.7845621.97265810.34457.9676990200861330.3554223.79314045.45552.4677480200968518.359521.59340506.97215.7277995模型建立散点图分析单因素或多变量间关系分析YX1X2X3X4Y10.99890.9934790452908040.87700.983602719841508X10.998910.9937402677184690.8556377347447820.984935296593492X20.9934790452908040.99374026771846910.85610.986241165680459X30.87700.8556377347447820.856110.8381X40.9836027198415080.9849352965934920.9862411656804590.83811 由散点图分析和变量间关系分析可以看出被解释变量财政收入Y与解释变量总税收收入X1、国内生产总值X2、其他收入X3、就业人口总数X4呈线性关系,因此该回归模型设为: 模型预模拟由eviews做ols回归得到结果:Dependent Variable: YMethod: Least SquaresDate: 11/14/11Time: 17:51Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC7299.5231691.8144.3146140.0006X11.0628020.02110850.349720.0000X20.0017700.0045280.3910070.7013X30.8733690.1198067.2898520.0000X4-0.1159750.026580-4.3631600.0006R-squared0.999978?Mean dependent var20556.75Adjusted R-squared0.999972?S.D. dependent var19987.03S.E. of regression106.6264?Akaike info criterion12.38886Sum squared resid170537.9?Schwarz criterion12.63779Log likelihood-118.8886?F-statistic166897.9Durbin-Watson stat1.496517?ProbF-statistic0.000000 4.314614 50.34972 0.391007 7.289852 -4.363160模型检验1.计量经济学意义检验多重共线性检验与解决 求相关系数矩阵,得到:Correlation MatrixYX1X2X3X410.99890.9934790452908040.87700.9836027198415080.998910.9937402677184690.8556377347447820.9849352965934920.9934790452908040.99374026771846910.85610.9862411656804590.87700.8556377347447820.856110.83810.9836027198415080.9849352965934920.9862411656804590.83811 发现模型存在多重共线性。接下来运用逐步回归法对模型进行修正:将各个解释变量分别加入模型,进行一元回归: 作Y与X1的回归,结果如下:Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:02Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-755.6610145.2330-5.2030940.0001X11.1449940.005760198.79310.0000R-squared0.999545?Mean dependent var20556.75Adjusted R-squared0.999519?S.D. dependent var19987.03S.E. of regression438.1521?Akaike info criterion15.09765Sum squared resid3455590Schwarz criterion15.19722Log likelihood-148.9765?F-statistic39518.70Durbin-Watson stat0.475046?ProbF-statistic0.000000作Y与X2的回归,结果如下:Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:06Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-5222.077861.2067-6.0636740.0000X20.2076890.00554837.432670.0000R-squared0.987317?Mean dependent var20556.75Adjusted R-squared0.986612?S.D. dependent var19987.03S.E. of regression2312.610?Akaike info criterion18.42478Sum squared resid96267005?Schwarz criterion18.52435Log likelihood-182.2478?F-statistic1401.205Durbin-Watson stat0.188013?ProbF-statistic0.000000作Y与X3的回归,结果如下:Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:08Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC2607.879773.99883.3693580.0034X310.030730.29431134.082090.0000R-squared0.984740?Mean dependent var20556.75Adjusted R-squared0.983893?S.D. dependent var19987.03S.E. of regression2536.645?Akaike info criterion18.60971Sum squared resid1.16E+08?Schwarz criterion18.70929Log likelihood-184.0971?F-statistic1161.589Durbin-Watson stat1.194389?ProbF-statistic0.000000作Y与X4的回归,结果如下:Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:08Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-272959.337203.65-7.3368940.0000X44.0974030.5184677.9029180.0000R-squared0.776276?Mean dependent var20556.75Adjusted R-squared0.763846?S.D. dependent var19987.03S.E. of regression9712.824?Akaike info criterion21.29492Sum squared resid1.70E+09?Schwarz criterion21.39449Log likelihood-210.9492?F-statistic62.45611Durbin-Watson stat0.157356?ProbF-statistic0.000000依据可决系数最大的原则选取X1作为进入回归模型的第一个解释变量,再依次将其余变量分别代入回归得: 作Y与X1、X2的回归,结果如下Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:09Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-188.4285239.0743-0.7881590.4415X11.2815940.04947225.905680.0000X2-0.0250550.009029-2.7749080.0130R-squared0.999687?Mean dependent var20556.75Adjusted R-squared0.999650?S.D. dependent var19987.03S.E. of regression374.0345?Akaike info criterion14.82405Sum squared resid2378330Schwarz criterion14.97341Log likelihood-145.2405?F-statistic27118.20Durbin-Watson stat0.683510?ProbF-statistic0.000000 作Y与X1、X3的回归,结果如下Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:10Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-351.105483.15053-4.2225270.0006X10.9928130.01870753.071960.0000X31.3569360.1651098.2184100.0000R-squared0.999908?Mean dependent var20556.75Adjusted R-squared0.999898?S.D. dependent var19987.03S.E. of regression202.1735?Akaike info criterion13.59361Sum squared resid694859.9?Schwarz criterion13.74297Log likelihood-132.9361?F-statistic92839.33Durbin-Watson stat1.177765?ProbF-statistic0.000000 作Y与X1、X4的回归,结果如下Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:10Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC11853.461824.5226.4967480.0000X11.1858860.006645178.46080.0000X4-0.1866450.026984-6.9170030.0000R-squared0.999881?Mean dependent var20556.75Adjusted R-squared0.999867?S.D. dependent var19987.03S.E. of regression230.8464?Akaike info criterion13.85886Sum squared resid905931.0?Schwarz criterion14.00822Log likelihood-135.5886?F-statistic71206.90Durbin-Watson stat1.459938?ProbF-statistic0.000000在满足经济意义和可决系数的条件下选取X3作为进入模型的第二个解释变量,再次进行回归则: 作Y与X1、X3、X2的回归,结果如下Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:13Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-76.04458100.1724-0.7591370.4588X11.0859240.02980136.438810.0000X31.2108530.1334449.0738770.0000X2-0.0140730.003944-3.5679010.0026R-squared0.999949?Mean dependent var20556.75Adjusted R-squared0.999939?S.D. dependent var19987.03S.E. of regression155.5183?Akaike info criterion13.10826Sum squared resid386975.0?Schwarz criterion13.30741Log likelihood-127.0826?F-statistic104602.9Durbin-Watson stat1.196933?ProbF-statistic0.000000 作Y与X1、X3、X4的回归,结果如下Dependent Variable: YMethod: Least SquaresDate: 11/22/11Time: 23:13Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC6781.7641024.7456.6180030.0000X11.0686420.01451473.627640.0000X30.8910690.1079498.2545510.0000X4-0.1076390.015451-6.9666750.0000R-squared0.999977?Mean dependent var20556.75Adjusted R-squared0.999973?S.D. dependent var19987.03S.E. of regression103.7654?Akaike info criterion12.29900Sum squared resid172276.1?Schwarz criterion12.49814Log likelihood-118.9900?F-statistic234970.9Durbin-Watson stat1.451447?ProbF-statistic0.000000可见加入其余任何一个变量都会导致系数符号与经济意义不符,故最终修正后的回归模型为:Dependent Variable: YMethod: Least SquaresDate: 11/30/11Time: 12:18Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC-351.105483.15053-4.2225270.0006X10.9928130.01870753.071960.0000X31.3569360.1651098.2184100.0000R-squared0.999908?Mean dependent var20556.75Adjusted R-squared0.999898?S.D. dependent var19987.03S.E. of regression202.1735?Akaike info criterion13.59361Sum squared resid694859.9?Schwarz criterion13.74297Log likelihood-132.9361?F-statistic92839.33Durbin-Watson stat1.177765?ProbF-statistic0.000000 -4.222527 53.07196 8.218410 异方差检验与修正图示法 ee与X1的散点图如下: 说明ee与X1存在单调递增型异方差性。 ee与X3的散点图如下: 说明ee与X3存在单调递增型异方差性。G-Q检验 对20组数据剔除掉中间四组剩下的进行分组后,第一组(1990-1997)数据的回归结果:Dependent Variable: YMethod: Least SquaresDate: 11/30/11Time: 12:54Sample: 1990 1997Included observations: 8VariableCoefficientStd. Errort-StatisticProbX10.9841230.01625560.543200.0000X30.8515180.1566885.4344720.0029C-28.3427545.36993-0.6247030.5596R-squared0.999686?Mean dependent var5179.791Adjusted R-squared0.999560?S.D. dependent var2099.840S.E. of regression44.05899?Akaike info criterion10.68893Sum squared resid9705.972?Schwarz criterion10.71872Log likelihood-39.75573?F-statistic7947.575Durbin-Watson stat1.663630?ProbF-statistic0.000000 残差平方和RSS19705.972第二组(2002-2009)数据的回归结果:Dependent Variable: YMethod: Least SquaresDate: 11/30/11Time: 12:55Sample: 2002 2009Included observations: 8VariableCoefficientStd. Errort-StatisticProbX11.0664040.02774738.433210.0000X30.8472280.2151143.9385030.0110C-1184.159261.8258-4.5226980.0063R-squared0.999932?Mean dependent var39824.41Adjusted R-squared0.999905?S.D. dependent var18639.16S.E. of regression182.0047?Akaike info criterion13.52594Sum squared resid165628.5?Schwarz criterion13.55573Log likelihood-51.10375?F-statistic36705.08Durbin-Watson stat1.326122?ProbF-statistic0.000000 残差平方和RSS2 165628.5 所以F RSS2/RSS1 165628.5/9705.97217.0646在给定5%下查得临界值 ,因此否定两组子样方差相同的假设,从而该总体随机项存在递增异方差性。White 方法检验White Heteroskedasticity Test:F-statistic6.142010?Probability0.003919Obs*R-squared12.41812?Probability0.014498Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 11/30/11Time: 13:21Sample: 1990 2009Included observations: 20VariableCoefficientStd. Errort-StatisticProbC24856.5019211.301.2938480.2153X1-20.573277.549127-2.7252520.0156X120.0002128.04E-052.6399820.0186X3237.181378.613233.0170670.0087X32-0.0240730.006568-3.6652300.0023R-squared0.620906?Mean dependent var34743.00Adjusted R-squared0.519815?S.D. dependent var49156.00S.E. of regression34062.86?Akaike info criterion23.92212Sum squared resid1.74E+10?Schwarz criterion24.17105Log likelihood-234.2212?F-statistic6.142010Durbin-Watson stat1.560937?ProbF-statistic0.0039195%下,临界值拒绝同方差性 修正Dependent Variable: YMethod: Least SquaresDate: 11/30/11Time: 14:29Sample: 1990 2009Included observations: 20Weighting series: 1/E1VariableCoefficientStd. Errort-StatisticProbC-314.207443.68550-7.1924860.0000X10.9797580.008622113.63360.0000X31.4572910.06592222.106290.0000Weighted StatisticsR-squared0.999999?Mean dependent var27246.27Adjusted R-squared0.999999?S.D. dependent var74471.17S.E. of regression73.91795?Akaike info criterion11.58127Sum squared resid92885.67?Schwarz criterion11.73063Log likelihood-112.8127?F-statistic3138195.Durbin-Watson stat0.956075?ProbF-statistic0.000000Unweighted StatisticsR-squared0.999902?Mean dependent var20556.75Adjusted R-squared0.999891?S.D. dependent var19987.03S.E. of regression209.0283?Sum squared resid742778.2Durbin-Watson stat1.365483 -7.192486 113.6336 22.10629序列相关性检验从残差项e2与e2-1及e与时间t的关系图(如下)看,随机项呈现正序列相关性。Q统计量检验由图可以看出,存在一阶序列相关回归检验残差e2与e2(-1)做回归得:Dependent Variable: EMethod: Least SquaresDate: 12/04/11Time: 15:21Sample adjusted: 1991 2009Included observations: 19 after adjustmentsVariableCoefficientStd. Errort-StatisticProbC16.8152545.696110.3679800.7174E-10.3035700.2311141.3135080.2065R-squared0.092138?Mean dependent var25.28519Adjusted R-squared0.038734?S.D. dependent var201.1252S.E. of regression197.1916?Akaike info criterion13.50553Sum squared resid661036.6?Schwarz criterion13.60494Log likelihood-126.3025?F-statistic1.725303Durbin-Watson stat1.776498?ProbF-statistic0.206464e与e-1、e-2做回归得:Dependent Variable: EMethod: Least SquaresDate: 12/04/11Time: 15:24Sample adjusted: 1992 2009Included observations: 18 after adjustmentsVariableCoefficientStd. Errort-StatisticProbC7.44976046.209120.1612180.8741E-10.4195640.2444751.7161870.1067E-2-0.3798940.278641-1.3633800.1929R-squared0.192570?Mean dependent var16.45940Adjusted R-squared0.084912?S.D. dependent var203.1349S.E. of regression194.3193?Akaike info criterion13.52789Sum squared resid566399.7?Schwarz criterion13.67629Log likelihood-118.7510?F-statistic1.788727Durbin-Watson stat2.055382?ProbF-statistic0.201043由上表明不存在序列相关性。D.W检验由异方差检验修正后的结果:得D.W1.365483取5%,由于n20,k3包含常数项,查表得:dl1.10,du1.54由于dlDW1.365483 du ,故: 序列相关性不确定。拉格朗日检验Dependent Variable: EMethod: Least SquaresDate: 12/04/11Time: 15:05Sample adjusted: 1992 2009Included observations: 18 after adjustmentsVariableCoefficientStd. Errort-StatisticProbY0.0009840.0025480.3862170.7051C-14.1479273.42247-0.1926920.8500E-10.3920090.2616331.4983160.1563E-2-0.3477300.298739-1.1639920.2639R-squared0.201082?Mean dependent var16.45940Adjusted R-squared0.029885?S.D. dependen