分析影响我国财政收入的因素

《分析影响我国财政收入的因素》由会员上传分享，免费在线阅读，更多相关内容在应用文档-天天文库。

分析影响我国财政收入的因素一、研究的目的要求研究财政收入的影响因素离不开一些基本的经济变量。大多数相关的研究文献中都把总税收、国内生产总值这两个指标作为影响财政收入的基本因素,还有一些文献中也提出了其他一些变量,比如其他收入、经济发展水平等。影响财政收入的因素众多复杂,但是通过研究经济理论对财政收入的解释以及对实践的观察,对财政收入影响的因素主要有总税收、国内生产总值、就业人数等。本文针对我国财政收入影响因素建立了计量经济模型，并利用Eviews软件对收集到的数据进行相关回归以及多重共线性分析，建立了财政收入影响因素的模型，分析了影响财政收入主要因素及其影响程度，并提出了相关政策建议。二、模型设定及其估计经分析，影响我国财政收入影响财政收入的因素众多复杂，本文从财政支出、国内生产总值、年末从业人员数、税收总额四方面进行分析。以国家财政收入为被解释变量，财政支出、国内生产总值、年末从业人员数、税收总额作为解释变量建立线性回归模型：Y=β0+β1X1+β2X2+β3X3+β4X4+ui其中，Y——财政收入X1——财政支出X2——国内生产总值X3——年末从业人员数X4——税收总额为估计模型参数，收集了1991-2010年的统计数据，如下表所示1991-2010影响财政收入的因素的数据表1年份财政收入财政支出国内生产总值年末从业人员数税收总额(亿元)(亿元)(亿元)（万人）（亿元）19913149.483386.6221826.2654912990.1719923483.373742.226937.3661523296.9119934348.954642.335260668084255.319945218.15792.6248108.5674555126.8819956242.26823.7259810.5680656038.0419967407.997937.5570142.5689506909.8219978651.149233.5678060.9698208234.0419989875.9510798.1883024.3706379262.8199911444.0813187.6788479.27139410682.58200013395.2315886.598000.57208512581.5112 200116386.0418902.58108068.27279715301.38200218903.6422053.15119095.77328017636.45200321715.2524649.951351747373620017.31200426396.4728486.89159586.87426424165.68200531649.2933930.28183618.57464728778.54200638760.240422.73215883.97497834804.35200751321.7849781.352664117532145621.97200861330.3562592.66315274.77556454223.79200968518.376299.93341401.57582859521.59201083101.5189874.164032607610573210.79DependentVariable:YMethod:LeastSquaresDate:05/30/13Time:20:20Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C13451.662790.3614.8207590.0002X10.0757260.0295902.5591570.0218X20.0177200.0101271.7497710.1006X3-0.2140210.043506-4.9193540.0002X40.9903520.06185116.011950.0000R-squared0.999924Meandependentvar24564.97AdjustedR-squared0.999904S.D.dependentvar23918.97S.E.ofregression234.1815Akaikeinfocriterion13.96239Sumsquaredresid822614.8Schwarzcriterion14.21132Loglikelihood-134.6239F-statistic49549.63Durbin-Watsonstat2.120774Prob(F-statistic)0.000000Yt=13451.66+0.075726X1+0.017720X2-0.214021X3+0.990352X4t=（4.820759）（2.559157）（1.749771）（-4.919354）（16.01195）R2=0.999924R2=0.999904F=49549.63DW=2.120774由此可见，该模型R2=0.999924，R2=0.999904可决系数很高，F检验值49549.63，明显显著，但是当α=0.05时，tα/2(n–k)=t0.025(20－4)=2.120,不仅X2、X3的系数t检验不显著，而且X3的系数的符号与预期相反，这表明很可能存在严重的多重共线性。12 三、多重共线性的检验和修正多重共线性的检验:计算各解释变量的相关系数，得相关系数矩阵如下：X1X2X3X4X110.9941238260420.8253147288930.997410960034X20.99412382604210.8610306165840.997132988649X30.8253147288930.86103061658410.830723342128X40.9974109600340.9971329886490.8307233421281由相关系数矩阵可以看出，各解释变量相互之间的相关系数较高，证实确实存在严重多重共线性。修正多重共线性分别采用逐步回归的办法，去检验和解决多重共线性问题。分别作Y对X1、X2、X3、X4的一元回归，结果如下表所示。DependentVariable:YMethod:LeastSquaresDate:05/30/13Time:21:45Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C-318.8016540.0402-0.5903290.5623X10.9418100.01493263.074310.0000R-squared0.995496Meandependentvar24564.97AdjustedR-squared0.995246S.D.dependentvar23918.97S.E.ofregression1649.250Akaikeinfocriterion17.74867Sumsquaredresid48960463Schwarzcriterion17.84824Loglikelihood-175.4867F-statistic3978.368Durbin-Watsonstat1.178424Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:05/30/13Time:21:50Sample:19992010Includedobservations:12VariableCoefficientStd.Errort-StatisticProb.C-1782.2691019.555-1.7480850.1110X20.1552240.01339311.589650.000012 R-squared0.930710Meandependentvar9042.181AdjustedR-squared0.923780S.D.dependentvar5130.406S.E.ofregression1416.396Akaikeinfocriterion17.50063Sumsquaredresid20061790Schwarzcriterion17.58145Loglikelihood-103.0038F-statistic134.3200Durbin-Watsonstat0.304979Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:05/30/13Time:21:51Sample:19992010Includedobservations:12VariableCoefficientStd.Errort-StatisticProb.C-120964.610145.61-11.922850.0000X31.8729950.14606912.822710.0000R-squared0.942668Meandependentvar9042.181AdjustedR-squared0.936935S.D.dependentvar5130.406S.E.ofregression1288.391Akaikeinfocriterion17.31119Sumsquaredresid16599504Schwarzcriterion17.39200Loglikelihood-101.8671F-statistic164.4219Durbin-Watsonstat0.443831Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:05/30/13Time:21:52Sample:19992010Includedobservations:12VariableCoefficientStd.Errort-StatisticProb.C-183.722156.20664-3.2686900.0084X41.0820490.005819185.95130.0000R-squared0.999711Meandependentvar9042.181AdjustedR-squared0.999682S.D.dependentvar5130.406S.E.ofregression91.49262Akaikeinfocriterion12.02141Sumsquaredresid83709.00Schwarzcriterion12.10222Loglikelihood-70.12843F-statistic34577.88Durbin-Watsonstat1.772665Prob(F-statistic)0.000000变量　X1　X2　X3　X4t统计量63.0743111.5896512.82271185.9513R20.9954960.930710.9422680.999711R2　0.9952460.9237800.9369350.99968212 其中，加入X4的方程R2最大，以X4为基础，顺次加入其他变量逐步回归。结果如下表：加入新变量的回归结果（一）DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:17:27Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C-770.9700133.9451-5.7558670.0000X10.0880210.0504701.7440400.0992X41.0396380.06129716.960750.0000R-squared0.999749Meandependentvar24564.97AdjustedR-squared0.999719S.D.dependentvar23918.97S.E.ofregression400.8759Akaikeinfocriterion14.96266Sumsquaredresid2731925.Schwarzcriterion15.11202Loglikelihood-146.6266F-statistic33812.68Durbin-Watsonstat0.566775Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:17:29Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C-243.7025296.1522-0.8228960.4220X2-0.0222380.010572-2.1034780.0506X41.2644400.05634222.442220.0000R-squared0.999765Meandependentvar24564.97AdjustedR-squared0.999737S.D.dependentvar23918.97S.E.ofregression387.7219Akaikeinfocriterion14.89593Sumsquaredresid2555581.Schwarzcriterion15.04529Loglikelihood-145.9593F-statistic36146.45Durbin-Watsonstat1.347835Prob(F-statistic)0.00000012 DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:17:30Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C10212.972254.6494.5297390.0003X3-0.1606920.032854-4.8910580.0001X41.1687810.005541210.91940.0000R-squared0.999877Meandependentvar24564.97AdjustedR-squared0.999862S.D.dependentvar23918.97S.E.ofregression280.5409Akaikeinfocriterion14.24880Sumsquaredresid1337954.Schwarzcriterion14.39816Loglikelihood-139.4880F-statistic69049.86Durbin-Watsonstat2.404257Prob(F-statistic)0.000000　X1X2X3X4　R2X4,X10.088211.0396380.999719（-1.74404）（16.96075）X4,X2-0.0222381.264440.999737（-2.103478）（22.44222）X4,X3-0.160691.1687810.999877　　（-4.89106）（210.9194）经比较，新加入X3的方程R2=0.999877最大，而且各参数的t检验显著，选择保留X3，再加入其他新变量逐步回归，结果如下DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:18:19Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C9855.5632005.3374.9146670.000212 X10.0744560.0314292.3689910.0308X3-0.1550560.029235-5.3036850.0001X41.0777960.03871927.836050.0000R-squared0.999909Meandependentvar24564.97AdjustedR-squared0.999892S.D.dependentvar23918.97S.E.ofregression248.8124Akaikeinfocriterion14.04813Sumsquaredresid990521.4Schwarzcriterion14.24728Loglikelihood-136.4813F-statistic58523.98Durbin-Watsonstat1.926348Prob(F-statistic)0.000000DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:18:20Sample:19912010Includedobservations:20VariableCoefficientStd.Errort-StatisticProb.C13685.943236.5554.2285520.0006X20.0170850.0117491.4540840.1653X3-0.2176350.050463-4.3127290.0005X41.0859700.05720318.984460.0000R-squared0.999891Meandependentvar24564.97AdjustedR-squared0.999871S.D.dependentvar23918.97S.E.ofregression271.7747Akaikeinfocriterion14.22468Sumsquaredresid1181784.Schwarzcriterion14.42383Loglikelihood-138.2468F-statistic49051.46Durbin-Watsonstat2.528001Prob(F-statistic)0.000000　X1X2X3X4　R2X4，X3，X10.074456-0.0292351.0777960.999892（2.368991）（-5.303685）（27.83605）X4,X3，X20.017085-0.2176351.085970.999871　（1.454084）（-4.312729）（18.98446）当加入X2时，R2有所增加，但其参数的t检验不显著。加入X1后，R2有所增加，且其参数的t检验显著。这说明主要是X2引起了多重共线性，予以剔除。X1应予以保留。最后修正严重多重共线性影响后的回归结果为：Yt=9855.536+0.074456X1-0.029235X3+1.077796X412 t=（4.914667）（2.368991）（-5.303685）（27.83605）R2=0.999909R2=0.999892F=58523.98DW=1.926348四、异方差的检验和修正Goldfeld-Quanadt检验:由于n=20删除五分之一的观测值，也就是大约4个观测值，余下部分平分得到两个样本区间：1991~1998和2003~2010，它们的样本个数均为8个，即n1=n2=8。采用OLS进行估计DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:19:26Sample:19911998Includedobservations:8VariableCoefficientStd.Errort-StatisticProb.C-5083.25218547.51-0.2740670.7976X10.5653180.3180561.7774190.1501X30.0794750.2926520.2715680.7994X40.3526060.4453350.7917770.4728R-squared0.999001Meandependentvar6047.148AdjustedR-squared0.998252S.D.dependentvar2445.727S.E.ofregression102.2605Akaikeinfocriterion12.39978Sumsquaredresid41828.86Schwarzcriterion12.43950Loglikelihood-45.59911F-statistic1333.346Durbin-Watsonstat1.432862Prob(F-statistic)0.000002DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:19:27Sample:20032010Includedobservations:8VariableCoefficientStd.Errort-StatisticProb.C-52904.9354297.14-0.9743590.3851X10.1012950.0553241.8309190.1411X30.7016640.7414800.9463010.3976X41.0095490.08557211.797670.0003R-squared0.999825Meandependentvar47849.14AdjustedR-squared0.999693S.D.dependentvar21882.76S.E.ofregression383.1202Akaikeinfocriterion15.04143Sumsquaredresid587124.3Schwarzcriterion15.0811512 Loglikelihood-56.16571F-statistic7610.881Durbin-Watsonstat1.848470Prob(F-statistic)0.000000由结果计算F统计量F=e2i2e1i2=587124.341828.86=14.036345判断在α=0.05下，分子分母的自由度都是（20-4）/2-4=4，查F分布表得到临界值F0.05（4，4）=6.39，因为F=14.036345>F0.05（4，4）=6.39，所以拒绝原假设，表明模型存在异方差。异方差的修正修正后得：DependentVariable:YMethod:LeastSquaresDate:05/31/13Time:21:54Sample:19912010Includedobservations:20Weightingseries:W1VariableCoefficientStd.Errort-StatisticProb.C9651.242714.026513.516640.0000X10.0438620.0284071.5440640.1421X3-0.1522180.010820-14.067790.0000X41.1133570.03081736.127740.0000WeightedStatisticsR-squared0.999999Meandependentvar22907.82AdjustedR-squared0.999999S.D.dependentvar58084.06S.E.ofregression58.85521Akaikeinfocriterion11.16489Sumsquaredresid55422.97Schwarzcriterion11.36404Loglikelihood-107.6489F-statistic1116847.Durbin-Watsonstat1.241857Prob(F-statistic)0.000000UnweightedStatisticsR-squared0.999901Meandependentvar24564.97AdjustedR-squared0.999883S.D.dependentvar23918.97S.E.ofregression258.7312Sumsquaredresid1071069.Durbin-Watsonstat2.135772y=9651.242+0.043862X1-0.152218X3+1.113357X4再次检验12 WhiteHeteroskedasticityTest:F-statistic2.017430Probability0.168011Obs*R-squared12.49475Probability0.186832TestEquation:DependentVariable:RESID^2Method:LeastSquaresDate:06/06/13Time:22:24Sample:19912008Includedobservations:18VariableCoefficientStd.Errort-StatisticProb.C5.55E-095.30E-091.0479960.3253X1-2.95E-144.31E-13-0.0684380.9471X1^23.02E-185.28E-180.5714160.5834X1*X32.15E-196.31E-180.0340760.9737X1*X4-6.52E-181.09E-17-0.5967050.5672X3-1.66E-131.55E-13-1.0665360.3173X3^21.24E-181.13E-181.0891680.3078X3*X4-7.67E-197.06E-18-0.1087020.9161X46.97E-144.82E-130.1447580.8885X4^23.54E-185.59E-180.6332870.5442R-squared0.694153Meandependentvar2.57E-12AdjustedR-squared0.350075S.D.dependentvar4.59E-12S.E.ofregression3.70E-12Akaikeinfocriterion-49.50901Sumsquaredresid1.09E-22Schwarzcriterion-49.01436Loglikelihood455.5811F-statistic2.017430Durbin-Watsonstat2.698321Prob(F-statistic)0.168011从表可以看出，nR2=9.754782，由white检验知，在α=0.05下，查χ2分布表，得临界值χ0.052（14）=22.3621，同时t检验量也显著。比较计算的χ2统计量与临界值，因为nR2=9,。754782<χ0.052（14）=22.3621,所以接受原假设，表明模型不存在异方差。五、自相关采用DW检验法，检验结果如下：DependentVariable:YMethod:LeastSquaresDate:06/06/13Time:22:08Sample:19912008Includedobservations:1812 VariableCoefficientStd.Errort-StatisticProb.C9651.2422.23E-054.33E+080.0000X10.0438625.94E-10739031660.0000X3-0.1522183.31E-10-4.59E+080.0000X41.1133576.36E-101.75E+090.0000R-squared1.000000Meandependentvar18867.36AdjustedR-squared1.000000S.D.dependentvar17021.49S.E.ofregression1.82E-06Akaikeinfocriterion-23.40411Sumsquaredresid4.63E-11Schwarzcriterion-23.20625LoglikelihoodF-statistic4.97E+20Durbin-Watsonstat1.601498Prob(F-statistic)0.000000由上表可知，DW=1.601498，查表可知当n=18k=4时DL=0.820DU=1.872DL