推荐 计量经济学课程论文叶修华Word下载.docx

1、1.91%4.22.52109.619870.4316,475.103.66%4.92.28113.619880.5026,742.704.08%5.23.17118.319890.5756,981.404.83%7.22.612419900.7177,112.505.39%9.80.74130.719910.9277,100.504.25%11.9-1.55136.219921.17,336.603.03%12.52.03140.319931.37,532.702.96%141.53144.519941.57,835.502.61%15.12.82148.219951.88,031.702.

2、81%15.91.31152.4199628,328.902.93%232.54156.919972.38,703.502.34%24.4160.519989,066.901.55%26.93.15163199939,470.302.19%303.64166.620009,817.003.38%34172.220019,890.702.83%35.50.06177.120024.510,048.801.59%42.50.84179.9200310,301.002.27%40.31.55184200410,675.802.68%43.82.65188.9200510,989.503.39%43.

3、92.13195.32006511,294.803.24%49.51.71201.6200711,523.902.85%53.11.15207.34200811,652.003.85%55.6-0.48215.3数据来源：维基百科等网站。三、建立模型由数据分析，初步建立模型Y=b0+b1*X1+b2*X2+b3*X3+b4*X4+b5*X5+ui其中b0表示在没有任何因素影响下的NBA球员平均工资水平；b1表示美国GDP水平对NBA球员平均工资水平的影响；b2表示美国通货膨胀率对NBA球员平均工资水平的影响；b3表示NBA联盟工资帽对NBA球员平均工资水平的影响；b4表示美国的经济增长率对NB

4、A球员平均工资水平的影响；b5表示美国CPI水平对NBA球员平均工资水平的影响；ui为随机扰动项。（一）模型的参数估计及经济意义、统计意义上的检验 利用Eviews软件，做Y对X1、X2、X3、X4、X5的回归，回归结果如下表2：Dependent Variable: YMethod: Least SquaresDate: 12/03/12 Time: 09:37Sample: 1985 2008Included observations: 24VariableCoefficientStd. Errort-StatisticProb.C-2.8459031.356782-2.0975390.0

5、503GDP0.0012230.0003443.5520440.0023INF-13.269155.675898-2.3378060.0311SALARYGAP0.0386120.0260471.4823840.1555ECONOMICGROWTH-0.1049330.050898-2.0616420.0540CPI-0.0355520.011663-3.0482020.0069R-squared0.989471Mean dependent var2.561000Adjusted R-squared0.986547S.D. dependent var1.825273S.E. of regres

6、sion0.211710Akaike info criterion-0.054884Sum squared resid0.806778Schwarz criterion0.239630Log likelihood6.658606Hannan-Quinn criter.0.023251F-statistic338.3261Durbin-Watson stat0.890107Prob（F-statistic）0.000000由上表我们可以写出建立的回归方程：Y=-2.85+0.0012*X1-13.27*X2+0.039*X3-0.105*X4-0.036*X5+ui（-2.10） （3.55）

7、（-2.34） （1.48） （-2.06） （-3.05）R2=0.989 Adjusted R2=0.986 F-statistic=338.33 DW stat=0.891、 经济意义上的检验：该模型可初步通过经济意义上的检验，系数符号均符合经济意义，上表中的五大因素均可以在数量上增加NBA球员的平均工资。2、通过观察各因素的p值，我们发现出了salarygap意外的p值均小于0.05或与其极为接近，其精确度较为理想，同时R-squared=0.989，Adjusted R-squared=0.986，模型的拟合度很好。因此除salarygap的因素都对NBA球员平均工资水平有较大的影响

8、，同时我们也猜测模型中存在异方差，使得其他因素的影响的准确度受到了影响。因此需要进一步的异方差检验。（二）计量经济学检验样本数24，且模型为五元线性回归模型，利用怀特检验对异方差进行检验，可得结果如下表3：Heteroskedasticity Test: White1.360022Prob. F（20,3）0.4568Obs*R-squared21.61593Prob. Chi-Square（20）0.3617Scaled explained SS7.2437410.9958Test Equation: RESID2465.3184249.9478640.5346300.6300-0.0042

9、170.007276-0.5796650.6028GDP22.73E-061.38E-061.9776370.1424GDP*INF0.0363620.0495070.7344770.5159GDP*SALARYGAP-8.14E-050.000131-0.6217500.5782GDP*ECONOMICGROWTH-0.0004040.000204-1.9832440.1416GDP*CPI-0.0002610.000137-1.9097420.1522-36.24144108.8372-0.3329880.7611INF2455.3874383.80681.1865020.3208INF*

10、SALARYGAP-0.8927962.684853-0.3325310.7614INF*ECONOMICGROWTH-5.6588317.894198-0.7168340.5252INF*CPI-1.8078621.727274-1.0466560.37220.2119440.4314790.4912050.6570SALARYGAP20.0021160.0047540.4450670.6864SALARYGAP*ECONOMICGROWTH0.0304680.0155001.9656240.1441SALARYGAP*CPI0.0022200.0031530.7041060.53211.8

11、912971.1613141.6285840.2019ECONOMICGROWTH20.0296040.0197641.4978900.2311ECONOMICGROWTH*CPI0.0058590.0075750.7735050.49560.1089130.2319480.4695580.6707CPI20.0067980.0039641.7149210.18490.9006640.0336160.2384220.0374830.032711-4.3316700.003210-3.30087372.98004-4.0582002.3459520.456789 由上表可知R-squared=0

12、.900664，查表可得样本数为24，自由度为5的卡方分布的值11.0705，因为nR2=21.6211.0705,所以拒绝原假设，则如我们猜测的，模型存在异方差。我们初步分析原因，认为NBA的工资帽受到GDP，通货膨胀率，经济增长率和CPI的影响，所以可能会因此产生异方差，因此我们去掉这一因素，留下剩余的四个因素重现建立回归方程，结果如下表4:51-4.7599770.429613-11.079700.00000.0016430.0002018.157854-15.993725.536711-2.8886680.0094-0.1431720.045238-3.1648690.0051-0.0

13、394490.011716-3.3672430.00320.9881860.9856990.218279-0.0230320.9052700.2223965.2763840.042080397.31760.979294对于其异方差的检验，结果如下表5：53-1.3544050.934164-1.4498580.1810-0.0012740.001004-1.2691790.2362-6.48E-076.87E-07-0.9431690.3702-0.0121340.017096-0.7097490.49580.0001359.80E-051.3793080.20118.02E-057.68E-

14、051.0447530.323424.4110819.224051.2698200.2360-152.1414177.7162-0.8560920.4142-2.2827762.514051-0.9080070.38750.5863840.9637490.6084410.55790.4094320.1995532.0517480.0704-0.0103750.012189-0.8512030.4167-0.0092070.005693-1.6173250.14030.0796430.0570841.3951840.1964-0.0024670.002169-1.1378280.28460.75

15、95960.0377200.3856350.0544830.042705-3.1998350.016413-2.46355253.39802-3.0044992.0312152.7941290.143548可知R2=0.759596,同时对于样本为24，自由度为10的卡方分布临界值为18.31，则nR2=18.2318.31，所以接受原假设，表明残差是同方差的，不存在异方差性。根据表4得D-W检验，durbin-watson stat=0.979294，查表的样本为24，解释变量为4的dl为1.01 du=1.78，而DW值小于dl，存在正序列相关。利用迭代法对序列相关进行处理。一次迭代结果（

16、表6） 10:08 23 after adjustmentsConvergence avhieved after 1 2 iterations-5.8147730.660898-8.7982860.0012000.0003003.9931820.0009-8.4123154.567281-1.8418650.0830-0.0778230.035158-2.2135350.0408-0.0112610.017424-0.6462960.5267AR（1）0.6778980.1640304.1327700.00070.9936852.6580000.9918281.8019380.162892-0.5720020.4510740.27578612.57802Han