计量经济学课程设计我国服务贸易竞争力影响因素实证分析.doc

计量经济学期末课程设计题目：我国服务贸易竞争力影响因素的实证分析我国服务贸易竞争力影响因素实证分析摘要：服务贸易对以郭经济增长的作用日益重要，一定程度上决定了一国国际贸易在国际贸易在国际市场的竞争力。本文分析了服务贸易竞争力的影响因素，并对这些影响因素与服务贸易的关系进行了实证检验，在此基础上提出了促进我国服务贸易发展的对策建议。关键词：服务贸易 竞争力 影响因素 一、 引言在经济全球化趋势加强的时代背景下，国际服务贸易异军突起，成为推动一国经济增长的重要一级。全球服务贸易出口总额从1970年得700多亿美元上升到2006年的26882亿美元。其平均增长速度超过了同期货物贸易的增长速度，在很大程度上决定了一国国际贸易的发展状况和在国际市场上的竞争能力。近几年来，我国国际服务贸易正在以平均10%左右的速度迅速增长，但明显落后于货物贸易。2008年，我国服务贸易出口总额1465亿美元，占世界贸易出口比重3.9%。在某种意义上说，积极发展国际服务贸易并实现国际贸易的自由化，将是21世纪国际经济合作最重要的内容之一。有必要对我国服务贸易竞争力的影响因素进行分析，以便更好的制定政策措施促进我国服务贸易发展。二、模型建立与分析根据理论和经验分析，影响我国服务贸易竞争力（）<服务贸易出口额-数据来自（中国服务贸易网） 单位：亿美元>的主要因素有：服务业产值()-用第三产业GDP代表 数据来自（中国统计局）单位：亿人民币元；第三产业就业人数()-数据来自中国统计年鉴单位：万人次；对外开放度()-用对外依存度代表 数据来自中国统计年鉴单位：% ；外商直接投资额()-数据来自（中国统计局）单位：亿美元；货物出口()-数据来自（中国统计局）单位：亿美元。下表列出了我国服务贸易竞争力相关数据，拟建立我国服务贸易竞争力函数。我国服务竞争力与相关影响资料年份 Y服务贸易出口额（亿美元）X1第三产业GDP(亿元)X2第三产业就业人数（万人次）X3对外依存度(%)X4外商直接投资额(亿美元)X5货物出口（亿美元）1987423574.0939527.923.14394.371988474890.3993325.431.94475.161989455448.41012924.433.92525.381990575888.41197929.7834.87602.91991697337.11237833.1743.68719.11992919357.11309833.87110.08849.4199311011915.71416331.9275.15917.44199416416179.81551542.29337.671210.06199518419978.51688038.66375.211487.8199620623326.21792733.91417.251510.48199724526988.11843234.2452.571827.92199823930580.51886031.8454.631837.12199926233873.41920533.3403.191949.31200030138714.01982339.6407.22492.03200132944361.62022838.47468.782660.98200239449898.92109042.70527.43255.96200346456004.72180951.89535.054382.28200462164561.32301159.76606.39533.26200573973432.92377163.88603.257619.53200691484721.42461465.17694.689689.7820071216103879.62491762.73747.7012200.620081465120486.62571757.29923.9514306.93(一)、初步模型估计1、用普通最小二乘法估计模型 假设拟建立如下多元回归模型：用Eviews软件对上表数据进行回归分析,回归结果如（1.399） (-0.253) (0.503) （-3.398） （2.442） （39.024）我国服务竞争力影响因素的回归：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 20:15Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C46.7221333.388571.3993450.1808X1-1.90E-057.48E-05-0.2532840.8033X20.0019720.0039230.5025750.6221X3-2.6083900.767719-3.3975830.0037X40.1940000.0794332.4423070.0266X50.0926790.00237539.024280.0000R-squared0.998592    Mean dependent var372.9091Adjusted R-squared0.998152    S.D. dependent var391.9083S.E. of regression16.84754    Akaike info criterion8.713287Sum squared resid4541.434    Schwarz criterion9.010844Log likelihood-89.84616    Hannan-Quinn criter.8.783383F-statistic2269.515    Durbin-Watson stat2.125136Prob(F-statistic)0.000000=0.998592 =2269.515 D.W.=2.125136 由于较大且接近于1，而且=2269.515>，故认为我国贸易服务竞争力与上述解释变量间总体线性关系显著。但在5%和1%的显著性水平下、，、的参数未能通过T检验，而且，、的参数符号的经济意义也不合理，故认为解释变量间存在多重共线。2、检验简单相关系数关系系数表CX1X2X3X4X5C1114.796288469424-0.000296422907595618-0.10861420823581261.5326597305536042.080849623855607-0.004489475309564226X1-0.0002964229075956185.600730636236473e-091.857060663994502e-084.876034127844083e-06-1.254027888788278e-06-4.821507583681227e-09X2-0.10861420823581261.857060663994502e-081.538884114019114e-05-0.001743513768716993-0.00028530866264310344.218347352956007e-06X31.5326597305536044.876034127844083e-06-0.0017435137687169930.5893930851074340.02440858911091001-0.001212197870760436X42.080849623855607-1.254027888788278e-06-0.00028530866264310340.024408589110910010.006309631224484014-0.0001050255393620551X5-0.004489475309564226-4.821507583681227e-094.218347352956007e-06-0.001212197870760436-0.00010502553936205515.640233935541623e-06表中数据发现解释变量之间存在高度相关性。3、找出最简单的回归形式分别作出与间的回归：（1）、与回归如下：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 19:29Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C220.723595.722302.3058730.0320X10.0031970.0012622.5322440.0198R-squared0.242776    Mean dependent var372.9091Adjusted R-squared0.204915    S.D. dependent var391.9083S.E. of regression349.4550    Akaike info criterion14.63713Sum squared resid2442376.    Schwarz criterion14.73632Log likelihood-159.0085    Hannan-Quinn criter.14.66050F-statistic6.412261    Durbin-Watson stat0.533244Prob(F-statistic)0.019818(2.306) (2.532) (2)、 与回归如下：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 19:31Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C-773.2488168.5972-4.5863670.0002X20.0641820.0090867.0638180.0000R-squared0.713867    Mean dependent var372.9091Adjusted R-squared0.699560    S.D. dependent var391.9083S.E. of regression214.8143    Akaike info criterion13.66393Sum squared resid922903.5    Schwarz criterion13.76312Log likelihood-148.3033    Hannan-Quinn criter.13.68730F-statistic49.89753    Durbin-Watson stat0.184083Prob(F-statistic)0.000001(-4.586) (7.064)(3)、与回归如下：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 19:32Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C-690.0178144.8250-4.7644930.0001X325.923043.3723497.6869370.0000R-squared0.747120    Mean dependent var372.9091Adjusted R-squared0.734476    S.D. dependent var391.9083S.E. of regression201.9464    Akaike info criterion13.54039Sum squared resid815646.7    Schwarz criterion13.63958Log likelihood-146.9443    Hannan-Quinn criter.13.56375F-statistic59.08899    Durbin-Watson stat0.581338Prob(F-statistic)0.000000(-4.764) (7.687) (4)、 与回归如下：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 19:33Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C-148.434973.87686-2.0092200.0582X41.3481570.1602888.4108430.0000R-squared0.779596    Mean dependent var372.9091Adjusted R-squared0.768575    S.D. dependent var391.9083S.E. of regression188.5338    Akaike info criterion13.40294Sum squared resid710899.9    Schwarz criterion13.50213Log likelihood-145.4323    Hannan-Quinn criter.13.42630F-statistic70.74228    Durbin-Watson stat0.197493Prob(F-statistic)0.000000 （-2.009） （8.411）(5)、与回归如下：Dependent Variable: YMethod: Least SquaresDate: 06/13/11 Time: 20:36Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C31.146149.4335383.3016400.0036X50.0978170.00180154.303510.0000R-squared0.993263    Mean dependent var372.9091Adjusted R-squared0.992927    S.D. dependent var391.9083S.E. of regression32.96085    Akaike info criterion9.915026Sum squared resid21728.35    Schwarz criterion10.01421Log likelihood-107.0653    Hannan-Quinn criter.9.938391F-statistic2948.871    Durbin-Watson stat0.536380Prob(F-statistic)0.000000（3.302） （54.304）比较可见，我国服务贸易竞争力受货物出口额影响最大，因此选（5）为初始的回归模型。4、逐步回归将其他解释变量分别导入上述初始回归模型，寻找最佳回归方程 逐步回归CX1X2X3X4X5W.D.31.1460.09780.993260.55T值（3.302）（54.304）26.9760.000180.09650.993890.731T值（2.784）（1.395）（48.327）-65.5500.006830.09060.995940.9862T值（-2.311）(3.535)（36.316）-17.6860.01088-3.39450.09600.998061.679T值（-0.774）（6.608）（-4.441）(44.694)53651-0.005760.30270.08730.997581.264T值（1.311）(-1.469)(-3.489)(39.817)讨论：第一步，在初始模型中引入,虽有拟合优度的提高，但变量未能通过T检验。第二步，去掉，引入，拟合优度再次提高，且参数符号合理，变量也通过了T检验值；在5%的显著性水平下D.W.(22,3), =1.15，=1.54 而D.W.=0.9862 ，所以存在自相关。第三步，引入，拟合优度虽有提高，变量参数也能通过T检验，但是，参数的符号与经济意义不符。第四步，去掉，引入，拟合优度有所提高，但变量参数未能通过T检验，且参数符号与经济意义不符。由此表明：是多余的。同样也可以继续验证，如果用分别来替代，则它们之间的任意线性组合，都军不高达以，为解释变量的回归效果。因此，最终的服务贸易因素函数应以为最优，拟合结果如下： (-2,.311) (3.535) (36.316) (二)、模型检验与优化1、异方差检验怀特（White）检验如下：Heteroskedasticity Test: WhiteF-statistic19.45448    Prob. F(5,16)0.0000Obs*R-squared18.89245    Prob. Chi-Square(5)0.0020Scaled explained SS26.76766    Prob. Chi-Square(5)0.0001Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 06/20/11 Time: 14:34Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C-7497.4343337.554-2.2463860.0391X21.8414440.5464153.3700470.0039X22-5.87E-051.94E-05-3.0224570.0081X2*X50.0008610.0001386.2443400.0000X5-19.020173.102961-6.1296840.0000X52-0.0002083.73E-05-5.5817410.0000R-squared0.858748    Mean dependent var595.7866Adjusted R-squared0.814606    S.D. dependent var1188.605S.E. of regression511.7823    Akaike info criterion15.54068Sum squared resid4190738.    Schwarz criterion15.83823Log likelihood-164.9474    Hannan-Quinn criter.15.61077F-statistic19.45448    Durbin-Watson stat2.865975Prob(F-statistic)0.000003可得，Obs*R-squared=18.89 在5%的显著性水平下、自由度为5的分布的相应临界值=11.07 显然，18.89>11.07，因此，拒绝同方差的原假设。存在异方差。对其进行修正，以为权重系数才，修正结果如下：Dependent Variable: YMethod: Least SquaresDate: 06/20/11 Time: 15:11Sample: 1987 2008Included observations: 22Weighting series: X2(-2)VariableCoefficientStd. Errort-StatisticProb.  C-68.4106812.76835-5.3578320.0000X20.0070060.0011915.8803210.0000X50.0902750.00276932.602950.0000Weighted StatisticsR-squared0.994693    Mean dependent var205.1462Adjusted R-squared0.994135    S.D. dependent var116.6005S.E. of regression15.42066    Akaike info criterion8.435418Sum squared resid4518.141    Schwarz criterion8.584196Log likelihood-89.78959    Hannan-Quinn criter.8.470465F-statistic1780.735    Durbin-Watson stat0.820703Prob(F-statistic)0.000000Unweighted StatisticsR-squared0.995928    Mean dependent var372.9091Adjusted R-squared0.995500    S.D. dependent var391.9083S.E. of regression26.29138    Sum squared resid13133.50Durbin-Watson stat0.992108此时，在对其进行怀特（White）检验，结果如下：Heteroskedasticity Test: WhiteF-statistic3.197257    Prob. F(6,15)0.0316Obs*R-squared12.34623    Prob. Chi-Square(6)0.0547Scaled explained SS4.893942    Prob. Chi-Square(6)0.5575Test Equation:Dependent Variable: WGT_RESID2Method: Least SquaresDate: 06/20/11 Time: 15:16Sample: 1987 2008Included observations: 22VariableCoefficientStd. Errort-StatisticProb.  C732.40801774.3220.4127820.6856WGT2-3234.4072479.833-1.3042840.2118X22*WGT2-2.99E-052.92E-05-1.0246790.3218X2*WGT20.7032410.5224061.3461580.1982X2*X5*WGT20.0001636.66E-052.4522670.0269X52*WGT2-4.18E-054.22E-05-0.9894160.3382X5*WGT2-3.4438321.065963-3.2307240.0056R-squared0.561192    Mean dependent var205.3700Adjusted R-squared0.385669    S.D. dependent var216.7130S.E. of regression169.8581    Akaike info criterion13.36117Sum squared resid432776.4    Schwarz criterion13.70832Log likelihood-139.9729    Hannan-Quinn criter.13.44295F-statistic3.197257    Durbin-Watson stat2.886893Prob(F-statistic)0.031641此时，查表可得5%显著性水平下，自由度为6的分布相应临界值为=12.59> Obs*R-squared=12.34623，所以，接受同方差假设。修正后的结果如下：(-5.358) (5.88) （32.603）=0.994693 =0.994135 =1780.735 =0.820703 2、序列相关性检验在5%的显著性水平下D.W.(22,3), =1.15，=1.54 而D.W.=0.9862 ，所以存在正自相关。用Cochrane-Orcutt迭代法对模型修正，的结果如下：Dependent Variable: YMethod: Least SquaresDate: 06/20/11 Time: 23:02Sample (adjusted): 1989 2008Included observations: 20 after adjustmentsConvergence achieved after 10 iterationsVariableCoefficientStd. Errort-StatisticProb.  C-83.1202838.65558-2.1502790.0482X20.0081610.0029092.8050080.0133X50.0885720.00547416.180410.0000AR(1)0.7426790.3034602.4473700.0272AR(2)-0.6933230.493573-1.4047030.1805R-squared0.997347    Mean dependent var405.7500Adjusted R-squared0.996640    S.D. dependent var396.5739S.E. of regression22.98927    Akaike info criterion9.320250Sum squared resid7927.600    Schwarz criterion9.569183Log likelihood-88.20250    Hannan-Quinn criter.9.368845F-statistic1409.735    Durbin-Watson stat2.100451Prob(F-statistic)0.000000Inverted AR Roots .37-.75i     .37+.75i经过修正，=0.96< =2.10<4-=2.20 ，所以拒绝原假设，认为该模型与上述=0.82070，相比有了明显好转。三、各因素对我国服务竞争力影响分析由以上回归数据以及相关检验，我们得出了各个变量与我国服务贸易出口的变动关系。结论是：中国服务业发展水平、货物贸易出口额、外商直接投资和对外开放度与服务贸易出口额都是正相关的；中国服务业发展水平对于服务贸易出口额影响作用最小；货物出口额、外商直接投资、对外开放度和第三产业就业人数对服务贸易出口额影响较大。这些因素均通过改善资源禀赋质量而对提高我国服务贸易国际竞争力具有重要的影响。1、服务贸易发展水平。从全国服务贸易发展来看，一般服务业发展水平较高的国家才是服务贸易大国，也就是说服务业发展水平是影响服务贸易竞争力基础性因素。而在中国这22年间，中国服务业发展水平对服务贸易竞争力影响作用小，也就反过来说明中国服务业发展水平较低，是制约是制约中国服务贸易竞争力的基础性因素。2、第三产业就业人数。基本上各国基于要素禀赋的比较优势来决定其服务贸易的流向和竞争能力，只是现在服务业尤其是生产者服务业的专业知识、技术和资本的密集型特点更加显著，因此对生产要素的理解和认识也要更加开阔，人力资本和附带专业技术管理的高级资本的影响也更大。3、对外开放度。纵观