《计量经济学作用-虚拟变量回归.doc》由会员分享,可在线阅读,更多相关《计量经济学作用-虚拟变量回归.doc(23页珍藏版)》请在得力文库 - 分享文档赚钱的网站上搜索。
1、Four short words sum up what has lifted most successful individuals above the crowd: a little bit more.-author-date计量经济学作用-虚拟变量回归计量经济学作用-虚拟变量回归虚拟变量回归实验目的:分析19651970年美国制造业利润和销售额,季度的关系。实验要求:假定利润不仅与销售额有关,而且和季度因素有关(1) 如果认为季度影响使利润平均值发生变异,应如何引入虚拟变量?(2) 如果认为季度影响使利润对销售额的变化率发生变异,应如何引入虚拟变量?(3) 如果认为上诉两种情况都存在,又
2、当如何引入虚拟变量?(4) 对上述三种情况分别估计利润模型,进行对比分析。实验原理:最小二乘法原理实验步骤: 由于有四个季度,因此引入三个季度虚拟变量: 一、如果认为季度影响使利润平均值发生变异,应以加法类型引入三个虚拟变量,设其模型为:对模型进行回归,得到以下回归结果:Dependent Variable: YMethod: Least SquaresDate: 11/26/10 Time: 15:02Sample: 1965Q1 1970Q4Included observations: 24VariableCoefficientStd. Errort-StatisticProb.C6910
3、.4491922.3503.5947920.0019X0.0380080.0116703.2569140.0041D2-187.7317660.1218-0.2843900.7792D31169.320637.07661.8354460.0821D4-417.1182640.8333-0.6509000.5229R-squared0.517642Mean dependent var12838.54Adjusted R-squared0.416093S.D. dependent var1433.284S.E. of regression1095.227Akaike info criterion1
4、7.01836Sum squared resid22790932Schwarz criterion17.26379Log likelihood-199.2204F-statistic5.097454Durbin-Watson stat0.396350Prob(F-statistic)0.005810=6910.449-187.7317+1169.320-417.1182+0.038008 Se=(1922.350) (660.1218) (637.0766) (640.8333) (0.011670) t=(3.594792) (-0.284390) (1.835446) (-0.650900
5、) (3.256914)=0.517642=0.416093F=5.097454DW=0.396350二、如果认为季度影响使利润对销售额的变化率发生变化,应以乘法类型引入三个虚拟变量,设其模型为:=对上述模型进行回归,得到以下结果:Dependent Variable: YMethod: Least SquaresDate: 11/26/10 Time: 17:53Sample: 1965Q1 1970Q4Included observations: 24VariableCoefficientStd. Errort-StatisticProb.C7014.7571782.9323.934394
6、0.0009X0.0370680.0113223.2738960.0040X*D2-0.0009330.004302-0.2167760.8307X*D30.0079100.0040181.9685410.0638X*D4-0.0023850.004074-0.5852900.5652R-squared0.519733Mean dependent var12838.54Adjusted R-squared0.418624S.D. dependent var1433.284S.E. of regression1092.851Akaike info criterion17.01402Sum squ
7、ared resid22692129Schwarz criterion17.25945Log likelihood-199.1682F-statistic5.140331Durbin-Watson stat0.429628Prob(F-statistic)0.0055947014.757+0.037068-0.000933+0.007910-0.002385se=(1782.932)(0.011322)(0.004302) (0.004018) (0.004074)t=(3.934394)(3.273896)(-0.216776) (1.968541) (-0.585290)=0.519733
8、=0.418624F=5.140331DW=0.429628三、若上述两种情况都存在,应以加法和乘法相结合的方式引入三个虚拟变量,设模型为:对上述回归模型进行回归得到以下回归结果:Dependent Variable: YMethod: Least SquaresDate: 11/26/10 Time: 17:54Sample: 1965Q1 1970Q4Included observations: 24VariableCoefficientStd. Errort-StatisticProb.C10457.394075.1992.5661050.0207X0.0158680.0252650.6
9、280750.5388D2-4752.2575441.682-0.8733070.3954D3-3764.2085484.872-0.6862890.5024D4-4635.4645570.057-0.8322110.4175X*D20.0292070.0354260.8244670.4218X*D30.0311690.0346470.8996260.3817X*D40.0265770.0354750.7491760.4646R-squared0.546701Mean dependent var12838.54Adjusted R-squared0.348383S.D. dependent v
10、ar1433.284S.E. of regression1156.987Akaike info criterion17.20623Sum squared resid21417911Schwarz criterion17.59891Log likelihood-198.4747F-statistic2.756686Durbin-Watson stat0.464982Prob(F-statistic)0.044081=10457.39-4752.257-3764.208-4635.464+0.015868Se=(4075.199)(5441.682)(5484.872)(5570.057)(0.025265)t=(2.566105)(-0.873307)(-0.686289)(-0.832211)(0.628075)+0.029207+0.031169+0.026577se=(0.035426) (0.034647) (0.035475)t=(0.824467)(0.899626) (0.749176)=0.546701=0.348383F=2.756686DW=0.464982四、通过对三个模型进行对比分析可看出,第三个模型的参数估计值均不显著,模型一和二的销售额的参数估计显著,其余参数估计也不显著。方程都显著,但拟合程度都不是很好。-