华东师范大学：《计量经济学》第13章 异方差：如果误差方差不是常数的结果（宫峰飞）

计量经济学课件 by Dr. FF Gong permanentemailaddressfgongathotmail.com httpblog.sina.comcn/ffgong 教材 Damodar N gujarati, Essentials of Econometrics, 3rd edition. McGraw-Hill Company, 2006 中译本:经济计量学精要 本课件已改正了其中的错误) 13-1

13-1 计量经济学课件 -----by Dr. F. F. Gong permanent email address: ffgong(at)hotmail.com http://blog.sina.com.cn/ffgong ƒ 教材： ƒ Damodar N.Gujarati, Essentials of Econometrics, 3rd edition. McGraw-Hill Company, 2006 ƒ 中译本：经济计量学精要 ƒ （本课件已改正了其中的错误）

第13章 异方差:如果误差方差不是常数的结果 要点:初步掌握异方差时的回归模型处理方法。不满足经 典的约束条件。 13-2

13-2 ƒ 第13章 ƒ 异方差：如果误差方差不是常数的结果 ƒ 要点：初步掌握异方差时的回归模型处理方法。不满足经 典的约束条件。 ƒ

■13.1异方差的性质 132异方差的后果 13.3异方差的诊断:如何知道存在异方差问题 134观察到异方差该怎么办:补救措施 13.5怀特异方差校正后的标准误和t统计量 136若干异方差实 137总结 13-3

13-3 ƒ 13.1 异方差的性质 ƒ 13.2 异方差的后果 ƒ 13.3 异方差的诊断：如何知道存在异方差问题 ƒ 13.4 观察到异方差该怎么办：补救措施 ƒ 13.5 怀特异方差校正后的标准误和 t 统计量 ƒ 13.6 若干异方差实例 ƒ 13.7 总结

131异方差的性质 见教材p285-6 ■例子 13-4

13-4 ƒ 13.1 异方差的性质 ƒ 见教材p285-6 ƒ 例子

异方差 X1 X2 PDI PDI Fig 13-1 (a)Homoscedasticity; (b) heteroscedasticity. 13-5

13-5 异方差 Fig 13-1 (a) Homoscedasticity; (b) heteroscedasticity

例13-1NYSE佣金率趋势 Table 13-1 Commission rate trends, New York Stock Exchange, April 1975-December 1978 X2 April 1975 45.7000027.6000015.00000 368000021.3000012.10000 34.50000204000011.50000 December 31.9000018.90000 1040000 March197650.30000338000019000010.80000 5000000334000019.5000 10.90000 46.7000031.10000184000 1020000 December47000003120000176000 10.00000 March19774.30000288000016.000009.800000 437000028.1000015.500009.700000 September404000026100001009.1000 14.000008900000 March1978402000025000013.900008.100000 43.1000027000014400008.500000 42.5000026900001440000 8.700000 40.70000 24.5000 13.70000 7.800000 Standard Name n Mean deviation Variance Minimum Maximum X11 5501630268 为 37234 27.600 X1610.0941783431806 Notes: X,= Commission rate, cents per share for 0 to 199 shares) X2=Commission rate, cents per share (for 200 to 299 shares) Commission rate, cents per share (for 1000 to 9999 shares) Xa=Commission rate, cents per share(for 10,000 Source: S Tinic and R West, The Securties Industry Unde Brokerage Commissions: Changes in the Structure and Perforn 13- Member Firms,The Bel oumalof Economcs, vol 11, no. 1

13-6 例13-1 NYSE佣金率趋势 Table 13-1 Commission rate trends, New York Stock Exchange, April 1975-December 1978

例13-1NYSE佣金率趋势图 o 0-199 shares 60.000 200-999 shares 56.842 53.684 △1000-9999 shares 50.526 10000+sha 47.368 s44.211 41.053 31579 28.421 25263 22.105 18947 15789 12.632 Q94737 6.3158 0.13323E-13 16 Time Fig 13-2 Commission per share, in cents, NYSE, April 1975 to 13-7 December 1978(based on data of Table 13-1)

13-7 例13-1 NYSE佣金率趋势图 Fig 13-2 Commission per share, in cents, NYSE, April 1975 to December 1978 (based on data of Table 13-1)

例13-2研发费用数据 Table 13-2 Innovation in America: Research and development (R&D) expenditure in the United States($, in millions ), 1988 Industry grouping Sales R&D expenses Profits 1. Containers and packaging 6375.3 625 185.1 2. Nonbank financial 116264 929 1,5695 3. Service industries 14,655.1 178.3 276.8 4. Metals and mining 21,869.2 258.4 2828.1 5. Housing and construction 264083 494.7 22259 6. General manufacturing 32,405.61,0830 3.751.9 7. Leisure time industries 35,107.7 1,620.6 2.884.1 8. Paper and forest products 402954 421.7 4645.7 9. Food 70,761.6 5092 5,0364 10. Health care 90552.8 6,620.1 13.8699 11.Aerospace 95294.0 918.6 4487.8 12. Consumer products 101.314.1 1,595310,278.9 13. Electrical and electronics 116,1413 6,1075 8.787.3 14. Chemicals 1223157 4.454.1 16,4388 15. Conglomerates 141.649.9 3,1638 9.7614 16.Office equipment and computers 175,025.8 13, 210.7, 774.5 17. Fuel 24143481,703.823,1685 18. Automotive 293.5430 9,528.2 184154 Note: The industries are listed in increasing order of sales volume 13-8 Source: Business Week, Special 1989 Bonus Issue, R&D Scorecard, pp. 180-224

13-8 例13-2 研发费用数据 Table 13-2 Innovation in America: Research and development (R&D) expenditure in the United States ($, in millions), 1988

例13-2研发费用-销售图 R&D2=266.917+0.0309 Sales se=(1002.961)(000834) =(02654)(3.696) 原回归结果和图错误 04610 E9=E35 =19299+0.0319X Eq(13.3 Sales (s, in millions Fig 13-3 Research and development (R&D)expenditure and 13-9 sales, U.S. industries, 1988

13-9 例13-2研发费用-销售图 Fig 13-3 Research and development (R&D) expenditure and sales, U.S. industries, 1988. i R & ˆ D Salesi 2 r = 266.1917 + 0.0309 se = (1002.961) (0.00834) t = (0.2654) (3.6996) = 0.4610 原回归结果和图错误

例13-2研发费用的残差-销售图 原图错误 ig 13-4 Residuals from the r&d regression( 13. 3 13-10

13-10 例13-2研发费用的残差-销售图 Fig 13-4 Residuals from the R&D regression (13.3). 原图错误