《利率及收入对货币供应量的影响.doc》由会员分享，可在线阅读，更多相关《利率及收入对货币供应量的影响.doc（17页珍藏版）》请在三一办公上搜索。
1、利率及收入对货币供应量的影响（修订版） 内容摘要：本文以宏观货币需求理论为基础,引入存款利率和央行再贷款利率两个解释变量,利用计量经济学的方法,分析货币供应量与这两者的关系.从中国的实际情况出发,利用年度数据着重分析利率对货币需求的影响,从而将经济理论和中国现实情况结合进行分析.关键字：货币供应量 利率一理论模型的设定由于我们正在摸索阶段，所以先把模型设定为线性模型： 模型设定如下： Y=0+1X1+2X2+ui， Y货币供应量 X1央行再贷款利率（一年期） X2存款基准利率（一年期） ui-随机扰动项 0、1、2 -参数二、 数据来源及搜集处理方法 1 、货币供应量Y数据的搜集：M用广义货币
2、供应量M2代替，因为货币的供给主要是由中央银行来进行，而货币的需求则取决于流动性偏好，尤其是投机动机。由于流动性偏好是一种心理活动，难以操纵和控制，货币需求也就难以预测和控制，需要变动的是货币供应量。这种替代具有一定的合理性.M= M2= M1+M0.M0=现金流通量，M1= M0+银行活期存款，M2= M1+储蓄存款+定期存款。广义货币的供给量可以从中国人民银行网站（）中查得，但是由于统计项目的调整，只能直接得到广义货币供给量1999-2004年的数据。 2、利率数据的搜集 在目前中国的利率体系下存在这多种利率,按借贷主体可以分为:银行利率,非银行金融机构利率,有价证券的利率和市场利率.从数
3、据的代表性和可获得性两方面考虑,选用了中央银行的一年期再贷款利率. 央行的再贷款利率是中国人民银行向金融机构进行信用放贷时所使用的利率.从1984年起,再贷款利率成为中国中央银行的基准利率之一,起着宏观调控的作用.有关资料表明，1984-1993年，中央银行基础货币投放主要渠道是再贷款，95以上的基础货币是通过再贷款投放出去的。由于该时段较长，占样本长度的一半，因此，用再贷款利率数据是合理的，且考虑到数据的可获得性，于是统一使用再贷款利率数据。对于利率有变动的年度，按天数进行加权平均。数据来源：中国人民银行网站（） 这样,模型所需变量的数据都搜集齐了.下面就利用Eviews进行模拟.Depen
4、dent Variable: YMethod: Least SquaresDate: 06/08/05 Time: 00:35Sample: 1999 2004Included observations: 6VariableCoefficientStd. Errort-StatisticProb. C230280.8560854.20.4105890.7089X1-62576.68123000.7-0.5087500.6460X271406.33326460.80.2187290.8409R-squared0.084133 Mean dependent var161517.0Adjusted
5、R-squared-0.526446 S.D. dependent var50685.59S.E. of regression62621.75 Akaike info criterion25.23447Sum squared resid1.18E+10 Schwarz criterion25.13035Log likelihood-72.70340 F-statistic0.137792Durbin-Watson stat0.611858 Prob(F-statistic)0.876494从表中数据来看，拟合优度明显偏低，而且从d-w检验来看，存在明显的自相关，但是我们还是做了一下white检
6、验和ARCH检验，虽然并没有什么实际意义，但是我们想找出问题出在哪里，检验结果如下：ARCH Test:F-statistic3.695173 Probability0.345231Obs*R-squared3.523262 Probability0.171765Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 00:35Sample(adjusted): 2001 2004Included observations: 4 after adjusting endpointsVari
7、ableCoefficientStd. Errort-StatisticProb. C7.93E+092.36E+093.3544960.1844RESID2(-1)-1.7203232.068470-0.8316890.5583RESID2(-2)-4.9400012.028766-2.4349790.2481R-squared0.880815 Mean dependent var2.38E+09Adjusted R-squared0.642446 S.D. dependent var3.74E+09S.E. of regression2.24E+09 Akaike info criteri
8、on46.00886Sum squared resid5.01E+18 Schwarz criterion45.54858Log likelihood-89.01772 F-statistic3.695173Durbin-Watson stat1.900689 Prob(F-statistic)0.345231White Heteroskedasticity Test:F-statistic197.7849 Probability0.005035Obs*R-squared5.979844 Probability0.112595Test Equation:Dependent Variable:
9、RESID2Method: Least SquaresDate: 06/08/05 Time: 00:36Sample: 1999 2004Included observations: 6VariableCoefficientStd. Errort-StatisticProb. C1.41E+127.22E+1019.486850.0026X1-8.09E+114.13E+10-19.575910.0026X121.15E+115.85E+0919.731950.0026X28.15E+081.43E+090.5707250.6258R-squared0.996641 Mean depende
10、nt var1.96E+09Adjusted R-squared0.991602 S.D. dependent var2.97E+09S.E. of regression2.73E+08 Akaike info criterion41.91907Sum squared resid1.49E+17 Schwarz criterion41.78024Log likelihood-121.7572 F-statistic197.7849Durbin-Watson stat3.191815 Prob(F-statistic)0.005035从检验结果来看，各项数据都显示出模型设定的不合理，所以决定将模
11、型进行修改：在设立模型时将利率作为决定货币需求总量的解释变量.由于三个变量之间数量级存在差异,若直接回归会存在一些潜在问题,为了回避这一 问题,本文在设定模型时采用了双对数模型,此外,双对数模型中,各解释变量的参数即为弹性,具有良好的经济解释意义.故模型修改如下： logY=c+Y货币供应量 X1央行再贷款利率（一年期） X2存款基准利率（一年期） ui-随机扰动项 0、1、2 -参数注: 利率采用百分比,一方面可以避免对数取负,另一方面,可以用数学推导证明这种代入并不影响参数的意义, 2则表示利率对货币供应量的弹性.Dependent Variable: LOG(Y)Method: Leas
12、t SquaresDate: 06/08/05 Time: 09:47Sample: 1999 2004Included observations: 6VariableCoefficientStd. Errort-StatisticProb. C13.366392.7790274.8097370.0171LOG(X1)-1.6040952.534671-0.6328610.5718LOG(X2)0.8146784.0380570.2017500.8530R-squared0.133972 Mean dependent var11.95281Adjusted R-squared-0.443379
13、 S.D. dependent var0.305031S.E. of regression0.366466 Akaike info criterion1.137033Sum squared resid0.402893 Schwarz criterion1.032912Log likelihood-0.411098 F-statistic0.232046Durbin-Watson stat0.641835 Prob(F-statistic)0.805931F-statistic3.343312 Probability0.360688Obs*R-squared3.479615 Probabilit
14、y0.175554Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 09:48Sample(adjusted): 2001 2004Included observations: 4 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. C0.2746740.0852323.2226530.1915RESID2(-1)-2.4250822.099849-1.1548840.4543RESID2(-
15、2)-4.6349392.019299-2.2953210.2616R-squared0.869904 Mean dependent var0.082210Adjusted R-squared0.609712 S.D. dependent var0.129219S.E. of regression0.080727 Akaike info criterion-2.081780Sum squared resid0.006517 Schwarz criterion-2.542059Log likelihood7.163559 F-statistic3.343312Durbin-Watson stat
16、1.797040 Prob(F-statistic)0.360688White Heteroskedasticity Test:F-statistic105.8244 Probability0.009376Obs*R-squared5.962438 Probability0.113452Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 09:49Sample: 1999 2004Included observations: 6VariableCoefficientStd. Erro
17、rt-StatisticProb. C84.517205.86943214.399550.0048LOG(X1)-135.70819.387113-14.456850.0048(LOG(X1)254.258383.73194314.538910.0047LOG(X2)0.0498060.1425560.3493750.7602R-squared0.993740 Mean dependent var0.067149Adjusted R-squared0.984349 S.D. dependent var0.103003S.E. of regression0.012886 Akaike info
18、criterion-5.630639Sum squared resid0.000332 Schwarz criterion-5.769466Log likelihood20.89192 F-statistic105.8244Durbin-Watson stat3.193308 Prob(F-statistic)0.009376从表中结果看出，拟合优度还是偏小，所以我们觉得可能是数据收集和选题上出了问题，所以，我们借鉴了一下学长们文章里的数据，将模型数据修改了一下，得出以下结果：线性模型：Dependent Variable: YMethod: Least SquaresDate: 06/08/
19、05 Time: 10:50Sample: 1990 2004Included observations: 15VariableCoefficientStd. Errort-StatisticProb. C188624.928029.326.7295560.0000X1243594.9801019.20.3041060.7663X2-1948879.759125.2-2.5672690.0247R-squared0.691036 Mean dependent var96219.35Adjusted R-squared0.639542 S.D. dependent var67534.31S.E.
20、 of regression40546.36 Akaike info criterion24.23514Sum squared resid1.97E+10 Schwarz criterion24.37675Log likelihood-178.7635 F-statistic13.41973Durbin-Watson stat0.533591 Prob(F-statistic)0.000870ARCH Test:F-statistic1.034925 Probability0.390383Obs*R-squared2.229361 Probability0.328020Test Equatio
21、n:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 10:51Sample(adjusted): 1992 2004Included observations: 13 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. C8.58E+081.14E+090.7517500.4695RESID2(-1)0.7966130.6686561.1913640.2610RESID2(-2)-0.3261120.811268-0.
22、4019770.6962R-squared0.171489 Mean dependent var1.34E+09Adjusted R-squared0.005787 S.D. dependent var1.89E+09S.E. of regression1.88E+09 Akaike info criterion45.74792Sum squared resid3.54E+19 Schwarz criterion45.87829Log likelihood-294.3615 F-statistic1.034925Durbin-Watson stat1.436975 Prob(F-statist
23、ic)0.390383White Heteroskedasticity Test:F-statistic0.703472 Probability0.607292Obs*R-squared3.293952 Probability0.509891Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 10:52Sample: 1990 2004Included observations: 15VariableCoefficientStd. Errort-StatisticProb. C3.6
24、8E+094.86E+090.7563920.4669X19.40E+092.59E+110.0362410.9718X12-1.84E+111.49E+12-0.1236370.9041X2-8.97E+101.96E+11-0.4570630.6574X227.49E+111.28E+120.5865010.5705R-squared0.219597 Mean dependent var1.32E+09Adjusted R-squared-0.092565 S.D. dependent var1.76E+09S.E. of regression1.84E+09 Akaike info cr
25、iterion45.76069Sum squared resid3.37E+19 Schwarz criterion45.99671Log likelihood-338.2052 F-statistic0.703472Durbin-Watson stat1.040008 Prob(F-statistic)0.607292双对数模型：Dependent Variable: LOG(Y)Method: Least SquaresDate: 06/08/05 Time: 10:52Sample: 1990 2004Included observations: 15VariableCoefficien
26、tStd. Errort-StatisticProb. C9.1040530.75129112.117870.0000LOG(X1)1.7241520.7591352.2712060.0423LOG(X2)-2.2316080.550211-4.0559140.0016R-squared0.767403 Mean dependent var11.18738Adjusted R-squared0.728637 S.D. dependent var0.851629S.E. of regression0.443635 Akaike info criterion1.389227Sum squared
27、resid2.361743 Schwarz criterion1.530837Log likelihood-7.419202 F-statistic19.79570Durbin-Watson stat0.780945 Prob(F-statistic)0.000158ARCH Test:F-statistic1.371904 Probability0.297512Obs*R-squared2.798968 Probability0.246724Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05
28、Time: 10:53Sample(adjusted): 1992 2004Included observations: 13 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. C0.1730280.0519513.3306010.0076RESID2(-1)-0.1083940.268104-0.4043000.6945RESID2(-2)-0.3007850.193299-1.5560620.1507R-squared0.215305 Mean dependent var0.116330Adjust
29、ed R-squared0.058366 S.D. dependent var0.118729S.E. of regression0.115212 Akaike info criterion-1.284912Sum squared resid0.132738 Schwarz criterion-1.154539Log likelihood11.35193 F-statistic1.371904Durbin-Watson stat2.418769 Prob(F-statistic)0.297512White Heteroskedasticity Test:F-statistic0.587970
30、Probability0.678925Obs*R-squared2.856099 Probability0.582188Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/08/05 Time: 10:54Sample: 1990 2004Included observations: 15VariableCoefficientStd. Errort-StatisticProb. C1.2655293.0194110.4191310.6840LOG(X1)-1.1532002.574276-0.4479710
31、.6637(LOG(X1)2-0.1812040.493874-0.3669040.7213LOG(X2)1.7710561.9044940.9299360.3743(LOG(X2)20.2607970.3322390.7849690.4507R-squared0.190407 Mean dependent var0.157450Adjusted R-squared-0.133431 S.D. dependent var0.166267S.E. of regression0.177013 Akaike info criterion-0.363991Sum squared resid0.3133
32、34 Schwarz criterion-0.127974Log likelihood7.729932 F-statistic0.587970Durbin-Watson stat1.432611 Prob(F-statistic)0.678925从利率与货币需求的散点图可以看出，利率和货币需求明显成反方向变动关系。Estimation Command:=LS LOG(Y) C LOG(X1) LOG(X2)Estimation Equation:=LOG(Y) = C(1) + C(2)*LOG(X1) + C(3)*LOG(X2)Substituted Coefficients:=LOG(Y
33、) = 9.104053025 + 1.724151652*LOG(X1) - 2.231607684*LOG(X2)现在的拟合优度明显比前两次高得多了，现在我们终于找到了问题的关键所在，原来我们一直把利率写成了*.*的格式，而不是0.*的格式，所以造成了数据上的严重错误，但是从WHITE检验和ARCH检验的拟合优度来看，又比较低，而用最小二乘法做出来的数据来看，D-W值为0.780945，其结果表明存在自相关，而WHITE检验和ARCH检验的犯错误的概率都比较大，所以还需要对模型进行进一步修正，但是log（x1）和log（x2）都为负值，在EVIEWS上无法进行进一步修正（我只知道用ar（x
34、）这种方法，试过了，不行）。所以我们猜断这个问题的关键可能出在选择解释变量时两个利率本来就存在一定的联系，基本上都是同时浮动的，所以我们可能是一开始选题就出了问题，我们阅读了学长的一篇文章，上面选择的解释变量是收入和央行再贷款利率，可能这样选择解释变量更为科学一点，避免了两个利率之间的自相关，但是由于篇幅有限，而且学长们已经做了这个课题了，所以我们希望下次能够吸取教训，争取下次做得更好！三、经济现象浅析从以上对年度数据和季度数据的分析。我们认为，虽然再贷款利率是央行根据货币市场需求进行调整的，但是由于中国货币市场的市场化程度较低，其作用性仍然很低。随着，金融体制的改革，再贷款利率的市场代表性提
35、高。注：由于金融理论知识的欠缺，因此不能作出很好的经济解释。本论文的着重点并非在结论，而在于利用计量经济学这种定量的分析方法，解决现实中的问题。补充：由于模型设定有误，我们重新设定了模型，把x2设为收入，既GDP，模型重新设定如下：Y=Y-货币需求总量 X2-收入X1-利率（%） ui-随机扰动项0、1、2 -参数数据平稳性检验：ADF Test Statistic-2.090711 1% Critical Value*-4.3260 5% Critical Value-3.2195 10% Critical Value-2.7557*MacKinnon critical values for
36、 rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(X1)Method: Least SquaresDate: 07/01/05 Time: 12:54Sample(adjusted): 1992 2001Included observations: 10 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. X1(-1)-0.3806980.182090-2.0
37、907110.0749D(X1(-1)0.9191580.2931953.1349680.0165C0.0322640.0162741.9825300.0879R-squared0.586641 Mean dependent var-0.003640Adjusted R-squared0.468538 S.D. dependent var0.016227S.E. of regression0.011829 Akaike info criterion-5.793128Sum squared resid0.000980 Schwarz criterion-5.702353Log likelihoo
38、d31.96564 F-statistic4.967214Durbin-Watson stat1.727237 Prob(F-statistic)0.045409ADF Test Statistic-1.774646 1% Critical Value*-4.4613 5% Critical Value-3.2695 10% Critical Value-2.7822*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependen
39、t Variable: D(X2)Method: Least SquaresDate: 07/01/05 Time: 12:44Sample(adjusted): 1993 2001Included observations: 9 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. X2(-1)-0.0527380.029717-1.7746460.1361D(X2(-1)0.8331790.2734053.0474220.0285D(X2(-2)-0.5052070.238098-2.1218390.0
40、873C8260.1132639.9683.1288680.0260R-squared0.811418 Mean dependent var7699.467Adjusted R-squared0.698269 S.D. dependent var2998.985S.E. of regression1647.343 Akaike info criterion17.95282Sum squared resid13568700 Schwarz criterion18.04047Log likelihood-76.78768 F-statistic7.171234Durbin-Watson stat2
41、.594088 Prob(F-statistic)0.029258ADF Test Statistic-7.584454 1% Critical Value*-4.3260 5% Critical Value-3.2195 10% Critical Value-2.7557*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(Y)Method: Least SquaresDate: 07/01/05 Time: 12:58Sample(adjusted): 1992 2001Included observations: 10 after adjusti
