对当前我国财政政策和货币政策的效果的实证分析金融11-2贺嘉091142472对当前我国财政政策和货币政策的效果的实证分析自改革开放以来,对我国经济起着主要推动力量的是:消费,财政政策和货币政策。而对于政府的宏观调控来说,财政政策与货币政策更能为政府所控制,从而使之成为国家干预经济的最主要手段。当选择这两种手段时,我们的认真考虑它们的效果如何,以确定两者用量及两者间的比例。基于此,我以下对1978—1999年间的财政政策和货币政策的效果做实证分析。一、实证分析下表资料中,zc代表居民消费,m代表货币与准货币,g代表财政支出,y代表国民生产总值。以y为被解释变量,其余为解释变量,进行回归分析。年份居民消费zc国民生产总值y货币与准货币m财政支出g19781759.13605.61159.11122.0919792005.44073.91458.11281.7919802317.74551.31842.91228.8319812604.14901.42234.51138.4119822867.95489.22589.81229.9819833182.56076.33075.01409.5219843674.57164.44146.31701.0219854589.08797.15198.92004.2519865175.010132.86720.92204.9119875961.211784.08330.92262.1819887633.114704.010990.82491.2119898523.516466.011949.62823.7819909113.218319.515293.73083.59199110315.921280.419349.93386.62199212459.825863.625402.23742.20199315682.434500.634879.84642.30199421230.047110.946923.55792.62199527838.959404.960750.56823.72199633187.969366.076094.97937.55199736117.876077.290995.39233.56199833682.978345.0104499.010798.18199935533.881911.0119898.013187.673DependentVariable:YMethod:LeastSquaresDate:10/23/13Time:15:40Sample:19781999Includedobservations:22VariableCoefficientStd.Errort-StatisticProb.C12.589131054.4230.0119390.9906ZC1.6694080.08367019.952380.0000M0.1678360.0771982.1740930.0433G0.2611290.6960970.3751330.7119R-squared0.998933Meandependentvar27723.87AdjustedR-squared0.998755S.D.dependentvar27544.15S.E.ofregression971.7583Akaikeinfocriterion16.75906Sumsquaredresid16997655Schwarzcriterion16.95743Loglikelihood-180.3496F-statistic5617.935Durbin-Watsonstat1.791097Prob(F-statistic)0.00000041、样本回归方程Yi=12.58913+1.669408*X1i+0.167836*X2i+0.261129*X3iR^2=0.998933F=5617.935DW=1.7910972.统计检验R-squared0.998933Meandependentvar27723.87AdjustedR-squared0.998755S.D.dependentvar27544.15S.E.ofregression971.7583Akaikeinfocriterion16.75906Sumsquaredresid16997655Schwarzcriterion16.95743Loglikelihood-180.3496F-statistic5617.935Durbin-Watsonstat1.791097Prob(F-statistic)0.000000方差分析与F检验与SSI相对应,自由度I-1也被分解为两部分,5(I-1)=(k-1)+(I-k)回归均方定义为MSR=1kSSR,误差均方定义为MSE=kTSSE表1.1方差分析表方差来源平方和自由度均方回归SSR=Yˆ'Yˆ-Iy2k-1MSR=SSR/(k-1)误差SSE=uˆ'uˆI-kMSE=SSE/(I-k)总和SSI=Y'Y-Iy2I-1H0:1=2=…=k-1=0;H1:j不全为零F=MSEMSR=)/()1/(kTSSEkSSRF(k-1,I-k)设检验水平为,则检验规则是,若FF(k-1,I-k),接受H0;若FF(k-1,I-k),拒绝H0。i检验H0:j=0,(j=1,2,…,k-1),H1:j0i=)ˆ(ˆjjs=1121)'(ˆ)ˆ(ˆjjjjsVarXXi(I-k)判别规则:若iik接受H0;若iik拒绝H0。3.散点图6怀特检验7①首先对上式进行OLS回归,求残差tuˆ。(reside)②做如下辅助回归式,2ˆtu=0+1xt1+2xt2+3xt12+4xt22+5xt1xt2+vtWhite检验的零假设和备择假设是H0:ut不存在异方差,H1:ut存在异方差④在不存在异方差假设条件下统计量TR22(5)其中T表示样本容量,R2是辅助回归式的OLS估计式的可决系数。自由度5表示辅助回归式中解释变量项数。TR2属于LM统计量。⑤判别规则是若TR22(5),接受H0(ut具有同方差)若TR22(5),拒绝H0(ut具有异方差)Goldfeld-Quandt检验H0:ut具有同方差,H1:ut具有递增型异方差。8用两个子样本分别估计回归直线,并计算残差平方和。相对于n2和n1分别用SSE2和SSE1表式。③F统计量是F=)/()/(1122knSSEknSSE=12SSESSE,(k为模型中被估参数个数)在H0成立条件下,FF(n2-k,n1-k)判别规则如下,若FF(n2-k,n1-k),接受H0(ut具有同方差)若FF(n2-k,n1-k),拒绝H0(递增型异方差)4.自相关95.虚拟变量10