harnessing the wisdom of crowds - harnessing the wisdom of crowds zhi da, university of notre dame...

Download HARNESSING THE WISDOM OF CROWDS -   HARNESSING THE WISDOM OF CROWDS Zhi Da, University of Notre Dame Xing Huang, Michigan State University Second Annual News

Post on 25-Apr-2018

219 views

Category:

Documents

5 download

Embed Size (px)

TRANSCRIPT

  • 1

    HARNESSING THE WISDOM OF

    CROWDS

    Z h i D a , U n i v e r s i t y o f N o t r e D a m e

    X i n g H u a n g , M i c h i g a n S t a t e U n i v e r s i t y

    S e c o n d A n n u a l N e w s & F i n a n c e C o n f e r e n c e

    M a r c h 8 , 2 0 1 7

  • 2

    Many important decisions in l i fe are

    made in a group

    F O M C m e e t i n gB o a r d m e e t i n g

    J u r y

    Who wants to be a millionaire

    A s k t h e a u d i e n c e

  • 3

    Wisdom of crowdsJ e l l y b e a n s i n t h e j a r e x p e r i m e n t

    Ask a group of people to guess:

    How many jelly beans in the jar?

    A large groups average answer to a question involving

    quantity estimation is generally as good as, and often

    better than, the answer provided by any individual in

    that group

    Law of large number requires independence

    94

    13599

    77

    67

    97

    112

    102

    59161

    68

    78

    47

    126

    80

    106

    95 jelly beans

  • 4

    CONS: MEN THINK IN HERDS

    Herding can reduce the accuracy of a groups average answer

    The more influence we exert on each other, the more likely it is that we will believe the same things and make the same mistakes. That means its possible that we could become individually smarter but collectively dumber.

    ---James Surowiecki, The Wisdom of Crowds.

    Sequential sett ing: Pros vs. Cons

    95

  • 5None of us is as dumb as al l of us.

  • 6

    CONS: MEN THINK IN HERDS

    Herding can reduce the accuracy of a groups average answer

    The more influence we exert on each other, the more likely it is that we will believe the same things and make the same mistakes. That means its possible that we could become individually smarter but collectively dumber.

    ---James Surowiecki, The Wisdom of Crowds.

    Additional information production may improve the accuracy of private signals.

    We become both individually smarter and collectively smarter.

    PROS:GENERATE ADDITIONAL INFORMATION PRODUCTION

    Sequential sett ing: Pros vs. Cons

    95

  • 7

    WE STUDY THE QUESTION:

    WE QUANTIFY THE IMPACT OF HERDING ON ECONOMIC OUTCOMES

    Do individuals herd in a sequential setting and reduce the usefulness of information aggregated across individuals (i.e., the wisdom of crowds)?

    The empirical challenge is that individuals information set is usually unobservable

    We overcome the empirical challenge by directly measuring and randomizing on individuals information set

    In this paper

  • 8

    WHY ITS A GOOD SETTING?

    Forecasts and realizations are clearly measured and easily observable The forecasters do not have direct influence on realizations Corporate earnings are of crucial importance

    Our sett ingA c r o w d - b a s e d e a r n i n g s f o r e c a s t p l a t f o r m

    ( E s t i m i z e . c o m )

  • 9

    M A I N A N A LY S I S

    Estimize.com related Sample statistics

    1D AT A A N D S T AT I S T I C S

    2

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    I N F L U E N T I A L U S E R A N D H E R D I N G

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    3R A N D O M I Z E DE X P E R I M E N T

    4

    Are users herding more with influential users?

    Does herding lead to return predictability

    Roadmap

  • 10

  • 11

    Open web-based platform founded in 2011, where users can make earnings forecasts

    Diverse user group: buy-side | sell-side | independent analysts | other working professionals | students

    Estimize consensus is more accurate than WS | IBES consensus, also complementary

    Jame et. al. (2016) | Adebambo and Bliss (2015) Now available on Bloomberg

    Various incentives for making forecasts

    Competition, monetary and professional prizes Reputation, useful track record Altruism, social preference

    Estimize.com

    Scoring system [-25, 25]:- Positive if more accurate than WS

    consensus- Awarded on an exponential scale,

    which encourages independent opinion

  • 12

    Sample period for :

    Main analysis: 2012/03 - 2015/03 Randomized experiment: Q2 and Q3 in 2015

    Ticker-quarter-forecast panel:

    2147 quarterly earnings 2516 users covering 730 stocks: mostly large-growth

    An average release (a sequence of forecasts for a ticker-quarter):

    20 forecasts from 16 users

    User activity on Estimize.com is tracked by Mixpanel

    Events: release page, estimate page, submit estimates, etc. Timestamp, location, device, etc.

    Sample statistics

  • 13Release page view

    Viewing activity = 1: If a user spent more than 5 seconds on the release page before making her own forecast

  • 14

    M A I N A N A LY S I S

    Estimize.com related Sample statistics

    1D AT A A N D S T AT I S T I C S

    2

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    I N F L U E N T I A L U S E R A N D H E R D I N G

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    3R A N D O M I Z E DE X P E R I M E N T

    4

    Are users herding more with influential users?

    Does herding lead to return predictability

    Roadmap

  • 15

    Note: Fixed effects subsume the need to control for stock or user characteristicsClustered standard errors account for autocorrelations in forecast errors

    Herding behavior

    Information weighting regression, Chen and Jiang (2006, RFS)

    FE 0Dev 0 0 : overweight on public information0 0 : overweight on private information

    I n d i v i d u a l s h e r d m o r e w h e n t h e y v i e w o t h e r s f o r e c a s t s

    More weight on the consensus forecast after viewing the release page

  • 16

    Influence of herding on forecast accuracy Herding makes individual forecast more accurate but reduces the accuracy of the consensus forecast

  • 17

    Note: Nonzero Views = 1: if a user spent more than 5 seconds on the release page before making her own forecast

    CTA dummy = 1: if the forecast is submitted during the last three days before announcements

    Individuals absolute forecast error decreases after viewing release page

    Influence of herding on forecast accuracyH e r d i n g m a k e s i n d i v i d u a l f o r e c a s t m o r e a c c u r a t e

  • 18

    Note: LnNumView = ln (1+ the percentage of forecasts with release views)Std Dev(FE)/Abs(Median(FE)) controls for uncertainty

    Consensuss absolute forecast error increases as there are more viewing activities within a release

    Magnitude: 0.0551 X ln(1+1) = 3.82 centsThis is more than the distance between the perfect forecast and the forecast with median Abs(FE) (3 cents)

    The consensus of group without viewing activity wins more than 50% with statistical significance

    Influence of herding on forecast accuracyH e r d i n g m a k e s c o n s e n s u s f o r e c a s t l e s s a c c u r a t e

  • 19

    Note: Consistent bias indicator = 1: if the bias in both early period and CTA period are in the same direction.

    LnNumView = ln (1+ the percentage of forecasts with release views)

    More viewing activity in the close-to-announcement period, the biases in these two periods are more likely to be consistent.

    Influence of herding on bias persistence

    Early period Close-to-announcement period

  • 20

    VIEWING ACTIVITY MAY BE ENDOGENOUS

    Less informed users are more likely to view others forecasts. Consistent with less accurate consensus. But inconsistent with more accurate individual forecast.

    ADDRESS THIS CONCERN:

    randomizing users information sets

    Endogeneity concern

  • 21

    M A I N A N A LY S I S

    Estimize.com related Sample statistics

    1D AT A A N D S T AT I S T I C S

    2

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    I N F L U E N T I A L U S E R A N D H E R D I N G

    Herding behavior The influence of herding on

    forecast accuracyo Individual forecasto Consensus forecast

    3R A N D O M I Z E DE X P E R I M E N T

    4

    Are users herding more with influential users?

    Does herding lead to return predictability

    Roadmap

  • 22

    RANDOMIZED EXPERIMENTS

    Pilot round (Q2 in 2015): 13 stocks are randomly selected Second round (Q3 in 2015): 90 stocks are randomly selected

    WHAT WE DO

    Randomly select users and disable the release page Ask them to make earnings forecasts (blind forecasts) Afterwards, the release page is restored They can immediately revise their forecasts (revised forecasts) Others not selected still view the original release page and make forecasts (default forecasts)

    Bl ind experiments

  • 23Blind view

  • 24

    Note: Estimate view, Avg #releases, #tickers, and abs(FE) are based on users forecasts before the experiment

    Users in blind and default group are similar in observable user characteristics.

    Blind vs. DefaultU s e r c h a r a c t e r i s t i c s

  • 25

    HERDING BEHAVIOR

    Default forecasts put more weight on Estimizeconsensus relative to blind forecasts

    Blind vs. DefaultH e r d i n g a n d c o n s e n s u s a c c u r a c y

    CONSENSUS

Recommended

View more >