Systematic noise
0 likes349 views
Solomon Asch's conformity experiments showed that people are inclined to conform their opinions to those of the group, even when the group's opinion conflicts with objective evidence. This tendency towards herding behavior has also been observed in individual investors. Studies of retail brokerage data found evidence that buying and selling decisions of individual investors are highly correlated and influenced by past stock returns, even when not driven by institutional investors, taxes, or risk preferences. The herding causes investors to concentrate their buying in fewer stocks than their selling and to be net buyers of stocks with unusually high trading volume. As a result, individual investors have the potential to systematically affect asset prices through their correlated behavior.
![Da+a & Method
Lakonishok, Shleifer and Vishny(1991)
HMit = |Pit E(Pit)| E|Pit E[Pit]|
= |(
Bit
Bit + Sit
) E(
Bit
Bit + Sit
)| AFit
where Pit : purchase proportion of all trade
E(Pit) =
Bit
Bit + Sit](https://image.slidesharecdn.com/systematicnoise-150615073305-lva1-app6892/85/Systematic-noise-36-320.jpg)
Systematic noise
- 1. Today, let’s s-art with Solomon Asch’s Experiment
- 2. ? A B C
- 7. …… $%@^&!$#!
- 9. ……
- 10. ⼩小綠綠隊友：A ！！！！！
- 11. ……
- 12. 獨眼⻯⿓龍：嘿神明隊友！你勒？
- 13. ……
- 14. 神明隊友：當然是A啊！
- 15. 從 眾 行 為 Herding Behaviour
- 18. 挑對隊友變成很重要的事情
- 20. 正 向 回 饋 Positive Feedback
- 21. 處 份 效 應 Disposition Effect
- 22. 資 訊 階 流 Information Cascades
- 25. 想要Starbuck, … 算了還是 City Coffee 好了！
- 30. Someday, . . . The influence of noise traders Will become apparent Fisher Black, 1986
- 31. Sys-ematic noise # Speakers: 張博能 | Department of Money and Banking Brad M. Barber, Terrance Odean, Ning Zhu Course of Financial Economics @ National Chengchi University Ccmail0111@gmail.com | @SCephan_Chang 2015/6/15
- 34. Discount Re+ail Period Number of householders Number of accounts Number of buys Mean buy value Number of sells Mean sell value Da+a & Method 1991/1 - 1996/11 1997/1 - 1999/6 66,465 665,533 104,211 793,499 1,082,107 3,974,998 $11,205 $15,209 887,594 3,219,299 $13,707 $21,170
- 35. How +o measure the Herding?
- 36. Da+a & Method Lakonishok, Shleifer and Vishny(1991) HMit = |Pit E(Pit)| E|Pit E[Pit]| = |( Bit Bit + Sit ) E( Bit Bit + Sit )| AFit where Pit : purchase proportion of all trade E(Pit) = Bit Bit + Sit
- 37. Da+a & Method Discount Broker Re+ail Broker All S+ocks Large Medium Small 0.0681 (< 0.001) *** 0.0758 (< 0.001) *** 0.0659 (< 0.001) *** 0.0637 (< 0.001) *** 0.1279 (< 0.001) *** 0.1138 (< 0.001) *** 0.1313 (< 0.001) *** 0.1250 (< 0.001) ***
- 38. Common Shift in risk-aversion ?
- 40. Da+a & Method Panel A: Large discount broker Group 1 with Group 1 Group 2 with Group 2 Horizon (L) Correlation of % buys in month t with % buys in month t+L Group 1 with Group 2 0 1 2 . . . 12 24 100% 100% 73.4% 48.2% 46.7% 47.7% 34.1% 33.1% 33.7% 9.7% 8.8% 9.6% 5.1% 3.1% 4.6% . . . . . . . . .
- 41. Da+a & Method Panel b: Large re+ail broker Group 1 with Group 1 Group 2 with Group 2 Horizon (i) Correlation of % buys in month i with % buys in month t+i Group 1 with Group 2 0 1 2 . . . 12 24 100% 100% 75.1% 56.7% 58.6% 55.8% 45.8% 46.4% 45.5% 22.7% 20.8% 23.1% 16.6% 22.6% 17.4% . . . . . . . . .
- 47. 3. Rational response +o correla+ed information
- 49. Unfil+ered da+a = Market orders + Limit Orders 1. Limit order driven?
- 50. Accounts = +axable accounts + +ax-deferred accounts 2. Avoid +ax losing
- 51. 3. Risk-aversion ? Use firms’ size as proxy for risk (Berk, 1995)
- 52. Let’s have discussion about Psychological biases
- 54. Overweight the impor+ance of past return when forecasting future’s
- 55. Sell winners, hold on+o losing Disposition Effect
- 56. Regression PBit = at + 12 j=1 bjtRt j + ctAVit + dtPBi,t 1 + it
- 57. 自 我 因 循 Self-herding
- 58. Separa+ely analyse buying and Selling, monthly Bit Pit = ab t + 12 j=1 bb jtRt j + cb t Bi,t 1 Pi,t 1 + b it Bit Pit = as t + 12 j=1 bs jtRt j + cs t Bi,t 1 Pi,t 1 + s it
- 59. Results: Influence of Past Returns on Buying & Selling (Discount)
- 60. Results: Influence of Past Returns on Buying & Selling (Re+ail)
- 61. The purchase and sale decisions of individual inves+ors are highly correla+ed. Summary
- 62. These are not primarily driven by： Summary > Passive response +o trades of institutional inves+ors > Tax loss selling > Risk preference
- 63. Inves+ors buy and sell s+ocks with strong past returns. Summary
- 65. Buying is concentra+ed in fewer s+ocks than selling Summary
- 66. Inves+ors are net buyers of s+ocks with unusually high trading volume. Summary
- 67. Individual inves+ors, sometimes referred +o as noise traders, have the po+ential +o affect asset prices because their noise is sys+ematic. Conclusion
- 68. Thanks for your at+ention :)