由弹性价格货币模型论中国汇率和利率的联动性大学论文.doc

计 量 经 济 课 程 论 文-由弹性价格货币模型论中国汇率和利率的联动性 内容摘要本文的初衷是想通过计量经济学知识检验弹性价格货币模型，运用本学期所学到的计量经济学知识。但随着探讨的深入，我们不断地发现问题并试图找出修正问题的方法，对模型进行了不断的调整。最后我们得出结论，由于我国目前国情特殊等种种原因该模型并不适用于中国这种汇率并分析了背后的原因。在整个论文的完成中，我们发现我们学到的东西远比我们开始时预想得多。除了学会了运用计量经济学知识解决实际问题我们还获得了很多启示，也总结了许多经验，于是我们在写最后的报告时附加了我们的体会，希望可以和大家分享。关键词：弹性价格货币模型，汇率，实际国民收入水平，利率水平，货币供给水平导论汇率决定理论是西方外汇理论的核心，也一直是国际经济学中最为活跃的领域之一。随着世界经济的变化和国际货币体制的变迁，汇率决定理论也在不断地发展货币模型是西方汇率决定理论中资产市场分析法的一个重要的分支。其中资产市场分析法是从20世纪七十年代中期开始迅速成长起来的汇率决定理论。货币法（Monetary Approach）和资产组合平衡法（Portfolio Balance Approach）是资产市场法的两个主要的分支。货币法中也有两种分析模型，一是弹性价格货币模型（Flexible-Price Monetary Model），另一个是粘性价格货币模型（Sticky-Price Monetary Model）。我们检验的重点就是弹性价格货币模型。经济解释一、弹性价格货币模型1弹性价格货币模型的基本思想弹性价格货币模型是现代汇率理论中最早建立、也是最基础的汇率决定模型。其主要代表人物有弗兰克尔（JFrenkel）、穆莎（MMussa）、考霍（PKouri）、比尔森（JBilson）等人。它是在1975年瑞典斯德哥尔摩附近召开的关于“浮动汇率与稳定政策”的国际研讨会上被提出来的。弹性货币法的一个基本思想：汇率是两国货币的相对价格，而不是两国商品的相对价格，因此汇率水平应主要由货币市场的供求状况决定。2弹性货币法的论述重要假设：（1） 稳定的货币需求方程，即货币需求同某些经济变量存在着稳定的关系；（2） 购买力平价持续有效。S=(y*-y)+(i-i*)+(Ms-Ms*) 从模型中我们可以看出，本国与外国之间实际国民收入水平、利率水平以及货币供给水平通过对各自物价水平的影响而决定了汇率水平。本国利率上升会降低货币需求，在原有的价格水平与货币供给水平上，这会造成支出的增加、物价的上升，从而通过购买力平价关系造成本国货币的贬值相关数据收集在中经网中我们找到了1985年到2002年美国，中国各自的官方汇率，实际国民收入，实际利率，货币供给M1，M2。现在的问题是M1,M2都是衡量货币供给的指标，应当选哪个？我们选择了M2.因为在Frederic S. Mishkin(米什金)的The Economics of Money, Banking, and Financial Market书我们找到了m1,m2的定义，而且书中明确指出，M2由于其速率远比M1稳定，因而在衡量货币供给方面比M1更好。在P57给出了M1,M2的定义：M1=Currency +Travelers checks +Demand deposits + Other checkable depositsM2= M1 + Small denomination time deposits + savings deposits and money market deposit accounts + Money market mutual fund shares作者在p560写道:”The relative stability of M2 velocity suggests that money demand functions in which the money supply is defined as M2 might performed substantially better than those in which the money supply is defined as M1.”原始数据如下：年 度 中国汇率S美国国民收入Y*中国国民收入Y中国实际利率I美国实际利率I*美国M2*中国M219852.9455985.6512133-2.026.5228010.484874.919863.4557264.0213413.653.175.9730980.836348.619873.7259578.4415097.472.725.0131939.37957.419883.7262681.0116958.06-2.795.733946.549602.119893.7764121.7217739.942.336.7935906.3811393.119904.7864915.7318598.373.495.8737674.3614681.919915.3265061.6520405.991.794.6538900.3618598.919925.5166602.9923502.670.683.7239547.6424327.319935.7668257.0126798.67-3.123.5240137.5835680.819948.6271056.9830525.25-7.444.9640153.5146920.319958.357332233496.5-0.996.5142418.0360743.519968.3175932.5236830.483.936.245010.0276095.319978.2979473.8840400.257.766.3647968.6491867.8119988.2883736.6543205.397.0252809.77105560.1119998.2887407.5846178.058.226.4557122.22121042.0620008.2891182.2149249.84.866.9861088.86135960.2320018.2891502.3752826.994.614.4469677.66156411.9320028.2893332.1157512.115.623.4772688.63186790.54由于模型中s , y, I Ms是指汇率，实际国民收入，货币供给量的自然对数值，于是作如下数据处理：( s=lnS, y=lnY, i=I, m=lnM)年 度sy*-yi-i*m-m*19851.0784095812.607576-0.0854-2.8268919861.2383742312.689774-0.028-2.8235119871.3137236682.68649-0.0229-2.7034619881.3137236682.621039-0.0849-2.5765319891.3270750012.61204-0.0446-2.4749819901.5644405472.814457-0.0238-2.506819911.6714733032.83098-0.0286-2.4093719921.7065646232.7482-0.0304-2.1924719931.7509374752.685865-0.0664-1.8686419942.1540850852.999013-0.124-1.9983419952.1222615392.905681-0.075-1.7631719962.1174596092.840979-0.0227-1.5923619972.1150499692.7916420.014-1.4652519982.1138429682.7755570.0198-1.4212619992.1138429682.751920.0177-1.362920002.1138429682.729797-0.0212-1.3138120012.1138429682.6631860.0017-1.3052320022.1138429682.5980120.0215-1.17004平稳性检验：单位根检验：y一阶差分，滞后0 期 ADF Test Statistic-4.485774 1% Critical Value*-4.6712 5% Critical Value-3.7347 10% Critical Value-3.3086*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(Y,2)Method: Least SquaresDate: 06/14/05 Time: 09:35Sample(adjusted): 1987 2002Included observations: 16 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. D(Y(-1)-1.2116820.270117-4.4857740.0006C0.0568010.0658300.8628410.4039TREND(1985)-0.0065050.006251-1.0407610.3170R-squared0.607580 Mean dependent var-0.009211Adjusted R-squared0.547208 S.D. dependent var0.165921S.E. of regression0.111648 Akaike info criterion-1.379568Sum squared resid0.162049 Schwarz criterion-1.234707Log likelihood14.03654 F-statistic10.06389M的单位根检验：滞后期为0，2阶差分ADF Test Statistic-6.0988751% Critical Value*-4.73155% Critical Value-3.761110% Critical Value-3.3228*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(M,3)Method: Least SquaresDate: 06/14/05 Time: 09:36Sample(adjusted): 1988 2002Included observations: 15 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb.I的单位根检验滞后期为1，2阶差分ADF Test Statistic-4.409217 1% Critical Value*-4.8025 5% Critical Value-3.7921 10% Critical Value-3.3393*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(I,3)Method: Least SquaresDate: 06/14/05 Time: 09:29Sample(adjusted): 1989 2002Included observations: 14 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. E一阶差分滞后1期ADF Test Statistic-3.415388 1% Critical Value*-4.7315 5% Critical Value-3.7611 10% Critical Value-3.3228*MacKinnon critical values for rejection of hypothesis of a unit root.Augmented Dickey-Fuller Test EquationDependent Variable: D(E,2)Method: Least SquaresDate: 06/14/05 Time: 09:13Sample(adjusted): 1988 2002Included observations: 15 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. D(E(-1)-1.4804320.433459-3.4153880.0058D(E(-1),2)0.2971920.2855761.0406770.3204C0.2043620.0983912.0770420.0620TREND(1985)-0.0119810.007830-1.5301470.1542R-squared0.610438 Mean dependent var-0.005023Adjusted R-squared0.504194 S.D. dependent var0.168304S.E. of regression0.118509 Akaike info criterion-1.204485Sum squared resid0.154487 Schwarz criterion-1.015671Log likelihood13.03364 F-statistic5.745612Durbin-Watson stat2.071058 Prob(F-statistic)0.012930因果关系检验：E 与m2 互为因果Pairwise Granger Causality TestsDate: 06/14/05 Time: 09:23Sample: 1985 2002Lags: 2 Null Hypothesis:ObsF-StatisticProbability M does not Granger Cause E16 14.0385 0.00094 E does not Granger Cause M 6.99392 0.01097E与 y：互不为因果Pairwise Granger Causality TestsDate: 06/14/05 Time: 09:28Sample: 1985 2002Lags: 1 Null Hypothesis:ObsF-StatisticProbability Y does not Granger Cause E17 0.93813 0.34920 E does not Granger Cause Y 0.09072 0.76769E与 I 互不为因果Pairwise Granger Causality TestsDate: 06/15/05 Time: 11:50Sample: 1985 2002Lags: 3 Null Hypothesis:ObsF-StatisticProbability I does not Granger Cause E15 0.44195 0.72943 E does not Granger Cause I 0.83104 0.51325参数的估计最小二乘回归得Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 21:02Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C-0.6177650.246003-2.5112040.0249Y1.2900260.08533415.117310.0000I-0.1422620.263086-0.5407440.5972M0.5754300.01836631.330850.0000R-squared0.993250 Mean dependent var1.780155Adjusted R-squared0.991804 S.D. dependent var0.385816S.E. of regression0.034928 Akaike info criterion-3.677909Sum squared resid0.017080 Schwarz criterion-3.480048Log likelihood37.10118 F-statistic686.7379Durbin-Watson stat1.371224 Prob(F-statistic)0.000000线性关系显著（由F统计量得知）, R2=0.993250说明拟合优度很好，但是由I的T检验中t=-0.540744，其绝对值小于2，可以看出，I 作为解释变量不是很合理经济意义检验： 回归所得的I的系数符号与经济意义不符 ，其他变量经济意义符合计量经济学检验1 重共线性检验相关系数矩阵：的确存在多重线性，并且I的t统计量不显著IYMI 1.000000-0.238763 0.526896Y-0.238763 1.000000 0.176456M 0.526896 0.176456 1.000000逐步回归得：Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:08Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb.C-3.2696842.051116-1.5941000.1305Y1.8418040.7475272.4638620.0255R-squared0.275054Mean dependent var1.780155Adjusted R-squared0.229745S.D. dependent var0.385816S.E. of regression0.338608Akaike info criterion0.776491Sum squared resid1.834484Schwarz criterion0.875421Log likelihood-4.988419F-statistic6.070615Durbin-Watson stat0.115207Prob(F-statistic)0.025456Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:08Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C1.8878030.11263616.760170.0000I3.3224532.1846431.5208220.1478R-squared0.126299 Mean dependent var1.780155Adjusted R-squared0.071693 S.D. dependent var0.385816S.E. of regression0.371728 Akaike info criterion0.963132Sum squared resid2.210912 Schwarz criterion1.062063Log likelihood-6.668192 F-statistic2.312898Durbin-Watson stat0.271065 Prob(F-statistic)0.147820Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:08Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C2.9985070.12851323.332280.0000M0.6130070.0621829.8583000.0000R-squared0.858640 Mean dependent var1.780155Adjusted R-squared0.849805 S.D. dependent var0.385816S.E. of regression0.149523 Akaike info criterion-0.858295Sum squared resid0.357714 Schwarz criterion-0.759365Log likelihood9.724656 F-statistic97.18608Durbin-Watson stat1.194219 Prob(F-statistic)0.000000选M为第一个解释变量Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:20Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C-0.6750750.216703-3.1152040.0071M0.5695180.01440539.536540.0000Y1.3083240.07646917.109280.0000R-squared0.993109 Mean dependent var1.780155Adjusted R-squared0.992191 S.D. dependent var0.385816S.E. of regression0.034095 Akaike info criterion-3.768349Sum squared resid0.017437 Schwarz criterion-3.619953Log likelihood36.91514 F-statistic1080.952Durbin-Watson stat1.286969 Prob(F-statistic)0.000000Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:20Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C3.0702070.12732724.112840.0000M0.6771110.0687219.8530340.0000I-1.7193330.971156-1.7703980.0970R-squared0.883072 Mean dependent var1.780155Adjusted R-squared0.867482 S.D. dependent var0.385816S.E. of regression0.140449 Akaike info criterion-0.936939Sum squared resid0.295887 Schwarz criterion-0.788544Log likelihood11.43246 F-statistic56.64223Durbin-Watson stat1.499001 Prob(F-statistic)0.000000由于调整后可决系数最大的为Y,逐步回归可以终止，最后的回归模型为：Dependent Variable: EMethod: Least SquaresDate: 05/18/04 Time: 13:20Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb.C-0.6750750.216703-3.1152040.0071M0.5695180.01440539.536540.0000Y1.3083240.07646917.109280.0000R-squared0.993109Mean dependent var1.780155Adjusted R-squared0.992191S.D. dependent var0.385816S.E. of regression0.034095Akaike info criterion-3.768349Sum squared resid0.017437Schwarz criterion-3.619953Log likelihood36.91514F-statistic1080.952Durbin-Watson stat1.286969Prob(F-statistic)0.000000可以看出，只有Y和M对模型的影响显著，而I对模型没有什么影响。2异方差检验做ARCH（P=3）检验：ARCH Test:F-statistic5.461801 Probability0.015183Obs*R-squared8.974892 Probability0.029627Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/13/05 Time: 20:26Sample(adjusted): 1988 2002Included observations: 15 after adjusting endpointsVariableCoefficientStd. Errort-StatisticProb. C0.0001540.0003180.4821940.6391RESID2(-1)1.0389410.3298393.1498420.0092RESID2(-2)0.1003790.4230710.2372630.8168RESID2(-3)-0.0743750.227784-0.3265170.7502R-squared0.598326 Mean dependent var0.000925Adjusted R-squared0.488779 S.D. dependent var0.000874S.E. of regression0.000625 Akaike info criterion-11.69368Sum squared resid4.30E-06 Schwarz criterion-11.50487Log likelihood91.70261 F-statistic5.461801Durbin-Watson stat1.912115 Prob(F-statistic)0.015183Obs*R-squared 对应的P值是0.029627 0.03初步判定没有异方差存在用WHITE检验：White Heteroskedasticity Test:F-statistic2.788935 Probability0.066926Obs*R-squared10.86065 Probability0.092780Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/13/05 Time: 20:46Sample: 1985 2002Included observations: 18VariableCoefficientStd. Errort-StatisticProb. C0.0936890.1994970.4696270.6478M0.0068670.0051871.3239640.2124M20.0015060.0012491.2055700.2533I0.0021660.0149630.1447480.8875I20.1060440.1676490.6325360.5400Y-0.0586970.145379-0.4037520.6941Y20.0099880.0263640.3788650.7120R-squared0.603369 Mean dependent var0.000949Adjusted R-squared0.387025 S.D. dependent var0.000952S.E. of regression0.000746 Akaike info criterion-11.27944Sum squared resid6.12E-06 Schwarz criterion-10.93318Log likelihood108.5149 F-statistic2.788935Durbin-Watson stat1.248839