1、计量经济学论文eviews分析计量经济作业 我国旅游收入的计量分析一、经济理论陈述在研读了大量统计和计量资料的基础上,选择了三个大方面进行研究,既包括旅游人数,人均旅游花费和基本交通建设。其中,在旅游人数这个解释变量的划分上,我们考虑到随着全球经济一体化的发展,越来越多的外国游客来中国旅游消费。中国旅游的国际市场是个有发展潜力的新兴市场,尽管外国游客前来旅游的方式包罗万象而且消费能力也不尽相同,但从国际服务贸易的角度出发,我们在做变量选择时,运用国际营销的知识进行市场细分,划分了国际和国内两个市场。这样,在旅游人数这个解释变量的最终确定上,我们选择了国内旅游人数,入境旅游人数。这点选择除了理论
2、支持外,在现实旅游业发展中我们也看到很多景区包括成都的近郊也有不少外国游客的身影。所以,我们选取这两个解释变量等待下一步进行模型设计和检验。另外,对于人均旅游花费,我们在进行市场细分时,没有延续前两个变量的选择模式,有几个原因。首先,外国游客前来旅游的形式和消费方式各异且很难统计。我们在花大力气收集数据后,仍然没有比较权威的统计数据资料。其次,随着国家对农业的不断重视和扶持,我国农业有了长足发展。农村居民纯收入增加,用于旅游的花费也有所上升。而且鉴于农村人口较多,前面的市场细分也不够细化,在这个解释变量的确定上,我们选择农村人均旅游花费,既是从我国基本国情出发,也是对第一步研究分析的补充。所以
3、我们确定了城镇居民人均旅游花费和农村居民人均旅游花费。旅游发展除了对消费者市场的划分研究,还应考虑到该产业的基础硬件设施。在众多可选择对象中我们经分析研究结合大量文献资料决定从交通建设着手。在我国,交通一般分布为公路,铁路,航班,航船等。由于考虑到我国一般大众的旅游交通方式集中在公路和铁路上,为了避免解释变量的过多过繁以及可能带来的多重共线形等问题,我们只选取了前二者。即确定了公路长度和铁路长度这两个解释变量。其中,考虑到我国旅游业不断发展过程中,高速公路的修建也不断增多,在的确定过程中,我们已经将其拟合,尽量保证解释变量的完整和真实。二、相关数据三、计量经济模型的建立()()()()()()
4、我们建立了下述的一般模型:其中 1994-2003年各年全国旅游收入()待定参数 国内旅游人数 (万人)入境旅游人数 (万人)城镇居民人均旅游花费 (元)农村居民人均旅游花费 (元)公路长度(含高速)(万公里)铁路长度 (万公里)随即扰动项四、模型的求解和检验利用Eviews软件,采用以上数据对该模型进行OLS回归,结果如下: Dependent Variable: YMethod: Least SquaresDate: 12/23/10 Time: 01:56Sample: 1994 2003Included observations: 10VariableCoefficientStd. E
5、rrort-StatisticProb.C-340.50471357.835-0.2507700.0882X2-0.0016160.013520-0.1195290.1524X30.2323580.1280171.8150500.1671X46.3910521.7168883.7224630.0337X5-1.0467571.224011-0.8551870.0453X65.6734296.6672660.8509380.4573X7-474.3909355.7167-1.3336200.2745R-squared0.996391Mean dependent var2494.200Adjust
6、ed R-squared0.989174S.D. dependent var980.4435S.E. of regression102.0112Akaike info criterion12.28407Sum squared resid31218.86Schwarz criterion12.49588Log likelihood-54.42035F-statistic138.0609Durbin-Watson stat3.244251Prob(F-statistic)0.000944由此可见,该模型可决系数很高,F检验显著,但是、的系数t检验不显著,且的系数符号不符合经济意义,说明存在严重的多
7、重共线性。所以进行以下修正:一计量方法检验及修正多重共线性的检验:首先对Y进行各个解释变量的逐步回归, 由最小二乘法,结合经济意义和统计检验得出拟合效果最好的两个解释变量如下:Dependent Variable: YMethod: Least SquaresDate: 12/23/10 Time: 02:00Sample: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-3193.041606.2101-5.2672170.0012X49.7290031.4354426.7777
8、030.0003X5-1.1970362.059371-0.5812630.1293R-squared0.957285Mean dependent var2494.200Adjusted R-squared0.945081S.D. dependent var980.4435S.E. of regression229.7654Akaike info criterion13.95532Sum squared resid369544.9Schwarz criterion14.04609Log likelihood-66.77660F-statistic78.43859Durbin-Watson st
9、at0.791632Prob(F-statistic)0.000016继续采用逐步回归法将其余解释变量代入,得出拟合效果最好的三个解释变量,结果如下:Dependent Variable: YMethod: Least SquaresDate: 12/23/10 Time: 02:01Sample: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-3391.810514.1119-6.5974160.0006X20.0294140.0145252.0250420.0393X46.3
10、554592.0501753.0999590.0211X5-0.2845421.772604-0.1605220.1077R-squared0.974627Mean dependent var2494.200Adjusted R-squared0.961940S.D. dependent var980.4435S.E. of regression191.2739Akaike info criterion13.63446Sum squared resid219514.3Schwarz criterion13.75550Log likelihood-64.17232F-statistic76.82
11、334Durbin-Watson stat1.328513Prob(F-statistic)0.000035以上模型估计效果最好,继续逐步回归得到以下结果:Dependent Variable: YMethod: Least SquaresDate: 12/23/10 Time: 02:40Sample: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-1973.943441.5947-4.4700340.0066X2-0.0050950.011431-0.4457290.6744
12、X30.3282790.0806824.0688020.0096X44.6654851.1586654.0266020.0101X5-1.7140200.999029-1.7156860.1469R-squared0.994114Mean dependent var2494.200Adjusted R-squared0.989406S.D. dependent var980.4435S.E. of regression100.9150Akaike info criterion12.37329Sum squared resid50919.23Schwarz criterion12.52458Lo
13、g likelihood-56.86644F-statistic211.1311Durbin-Watson stat3.034041Prob(F-statistic)0.000009各项拟合效果都较好。虽然的t检验不是很显著,但考虑到其经济意义在模型中的重要地位,暂时保留。继续引入。Dependent Variable: YMethod: Least SquaresDate: 12/23/10 Time: 02:41Sample: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-2
14、034.155525.2137-3.8730040.0179X2-0.0070330.014095-0.4989770.6440X30.2995620.1286262.3289460.0803X44.7879861.3398883.5734230.0233X5-1.5118511.282385-1.1789370.1638X62.0623346.6592470.3096950.7723R-squared0.994252Mean dependent var2494.200Adjusted R-squared0.987067S.D. dependent var980.4435S.E. of reg
15、ression111.4976Akaike info criterion12.54959Sum squared resid49726.89Schwarz criterion12.73114Log likelihood-56.74797F-statistic138.3830Durbin-Watson stat3.130122Prob(F-statistic)0.000144根据以上回归结果可得,的引入使得模型中、的t检验均不显著,再考察二者的相关系数为0.949132,说明、高度相关,模型产生了多重共线性,因此将去掉。再将代入检验。Dependent Variable: YMethod: Lea
16、st SquaresDate: 12/23/10 Time: 02:42Sample: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-641.06701265.065-0.5067460.0190X20.0014320.0125790.1138380.9149X30.3157420.0794873.9722640.0165X45.6942291.4560423.9107590.0174X5-1.6317100.977195-1.6697900.1703X7-351.4600313
17、.6492-1.1205510.3252R-squared0.995521Mean dependent var2494.200Adjusted R-squared0.989921S.D. dependent var980.4435S.E. of regression98.43019Akaike info criterion12.30028Sum squared resid38754.01Schwarz criterion12.48183Log likelihood-55.50141F-statistic177.7916Durbin-Watson stat2.850083Prob(F-stati
18、stic)0.000087的系数为负,与经济意义相悖,因此也去掉。由此确定带入模型的解释变量为、。异方差性的检验:再对模型的异方差性进行检验:鉴于我们的样本资料是时间序列数据,选用ARCH检验。ARCH Test:F-statistic0.044061Probability0.839718Obs*R-squared0.056296Probability0.812449Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 12/23/10 Time: 02:43Sample (adjusted): 1995 2003
19、Included observations: 9 after adjustmentsVariableCoefficientStd. Errort-StatisticProb.C5197.7413188.9601.6299180.1471RESID2(-1)0.0792160.3773850.2099080.8397R-squared0.006255Mean dependent var5645.880Adjusted R-squared-0.135708S.D. dependent var6668.507S.E. of regression7106.603Akaike info criterio
20、n20.76857Sum squared resid3.54E+08Schwarz criterion20.81239Log likelihood-91.45855F-statistic0.044061Durbin-Watson stat1.810449Prob(F-statistic)0.839718这里Obs*R-squared为0.056296,P=0.8124490.05 所以接受,表明模型中随机误差项不存在异方差。再考虑P=3的情况:ARCH Test:F-statistic0.126837Probability0.938100Obs*R-squared0.787922Probabi
21、lity0.852354Test Equation:Dependent Variable: RESID2Method: Least SquaresDate: 12/23/10 Time: 02:46Sample (adjusted): 1997 2003Included observations: 7 after adjustmentsVariableCoefficientStd. Errort-StatisticProb.C206.96718303.9310.0249240.9817RESID2(-1)0.1623770.5363370.3027510.7819RESID2(-2)0.112
22、7990.5704270.1977460.8559RESID2(-3)0.3312760.5706580.5805160.6023R-squared0.112560Mean dependent var4377.448Adjusted R-squared-0.774879S.D. dependent var7000.432S.E. of regression9326.298Akaike info criterion21.41462Sum squared resid2.61E+08Schwarz criterion21.38371Log likelihood-70.95118F-statistic
23、0.126837Durbin-Watson stat1.521751Prob(F-statistic)0.938100这里Obs*R-squared为0.787922,P=0.8523540.05。所以仍然接受,表明模型中随机误差项不存在异方差。自相关性的检验:随机扰动项可能存在一阶负自相关。借助残差项和其一阶滞后项的二维坐标图进一步分析:由图示可看出,残差项和其一阶滞后项显然存在负自相关,然后利用对数线形回归修正自相关性,得到相应结果如下:Dependent Variable: LOG(Y)Method: Least SquaresDate: 12/23/10 Time: 02:52Samp
24、le: 1994 2003Included observations: 10VariableCoefficientStd. Errort-StatisticProb.C-8.7695512.012276-4.3580270.0073LOG(X2)0.3247890.3438680.9445160.0383LOG(X3)0.3840660.2277461.6863780.0225LOG(X4)1.4826830.3134874.7296430.0052LOG(X5)0.0057500.0689550.0833820.0468R-squared0.994678Mean dependent var7.740729Adjusted R-squared0.990421S.D. dependent var0.442977S.E. of regression0.043355Akaike info
copyright@ 2008-2022 冰豆网网站版权所有
经营许可证编号:鄂ICP备2022015515号-1