Module 2

Theory of Under –reaction and Over-

reaction
 Trend of Asset Prices
• In short run…exhibit momentum
• In long run …tendency towards reversal or fundamental
reversion
• It is empirically observed ( out of the sample and out of the
period tests)
• Inertia of three types
1. Short term (less than one month)
2. Medium term (9-12 months)
3. Long term (3-5 years)
Thus these departures from classical assumptions of strict
rationality must be addressed.
 BSV (1998) and DHS(1998) argument

• Price is f ( representative heuristic)

• Thus short term momentum is by default
• Continue…continue….overreact….long term
reversal
• After certain time, continuance is unaffordable
so reversal is inevitable.
 Heterogeneous Beliefs of Two Pillars
• Two important agents (1) news watchers and (2) momentum
traders.
• Their beliefs can be heterogeneous but bounded rationality and
interaction
• News watchers forecast on private signals about fundamentals and
act slowly diffusion ….under-reaction
• Momentum traders chase trend and earn by momentum strategies
in the beginning
• Later price reversal and everything comes back to equilibrium
• Biggest paradox is earnings follow random walk model instead of
trending or mean reverting ( investors love time series rather than
sporadic events)
Who can time the gradual diffusion of news?

• Early momentum buyers impose a negative externality on

the “late” momentum buyers
• News watchers slow movers and momentum traders exploits
arbitrage benefits
• Speed of momentum increases culminating into overreaction
• Thus the very existence of under reaction sows the seeds of
overreaction ( by luring momentum traders)
• Thus the source of under reaction and over reaction is ….”
gradually diffusing news about the fundamentals” .It is also
called as Grand Daddy of under reaction
 Empirical Observations
• Assets Returns………..positive autocorrelation
at short horizons…..momentum (trending)
• Asset Returns…………negative correlation at
longer horizons…..long term reversal (mean
reverting)
• Impulse response function may be useful to
judge how long reaction will continue
 Cont.
• High (E/P, or C/P or BE/ME) ………..tend to have poor post
earnings growth (losers) ………overreact………high future returns
• Low (E/P, or C/P or BE/ME) ………..tend to higher post earnings
growth (winners) ………overreact………low future returns
• High price….IPO……poor long term post event
return…….overreaction………(it is part of market correction to
fundamental value and mean reverting)…..but if IPO value
weighted……abnormal return shrink for all benchmark, and
they are not reliably different from zero……………thus this
anomaly is largely restricted to tiny firms…..source of bad
model problems
 Cont.
• Divesting firms …..become merger
target…..under reaction…….post event
abnormal returns for parent and spin offs are
positive
• Repurchase ……..under react………..positive long
term post event abnormal return
• Dividend initiations…..positive……..under react
• Dividend omissions…..negative……..under react
 Cont.
• Investors in mean reverting regime….change
in earnings…..investors under reacts….wait for
further confirmation…..thus response is
delayed…..momentum persists
• Investors in trending regime…..change in
earnings….investors extrapolate the
earnings…..overreaction…..reversal of long
term return
 Cont.
• Stock split……positive abnormal return after
the split………..brings drift in the price
 Emerging Cross Sectional Anomalies
• Chance deviations to be expected under market
efficiency
• Persistent across time and space
• Size, value and momentum
• Pattern lead to events studies…
1. People want premia rationally as per Sharpe ratio
2. Can be exploited by utility function with strong habit
persistence for predictive variations in market return
3. Observed irrationality
 Contrarian Investment, Extrapolation and
Risk Trends
• Variety of investment strategies that involve buying out of
the favor (value) stocks have outperformed glamour
strategies
• Actual future growth rates of earnings, cash flow, etc. of
value stocks are higher than glamour stocks
• Value strategies are less risky and more rewarding provided
you stay for longer invested (However metaphysical
explanation of risk should be commensurate with returns).
• The plausible explanation is data snooping and higher ex
post return
 Cont.
• We can conjecture the preference of both
individuals and institutional investors in agency
context
• Individual investors focus on glamour
strategies by extrapolating growth past growth
rates and make judgmental errors. Putting
excessive weight on recent past history is most
common judgmental error in psychological
experiments.
 Cont.
• Institutional investors also gravitate towards
glamour stocks in order to show that they are
prudent investors and they justify easily to
their sponsors. It is also related to the career
of money managers of these institutions who
are given the responsibility of performing
better than benchmark in short term. Thus
they are more tilted towards the glamour
stocks
 Cont.
Money managers can’t afford to
underperform with respect to their peers for
any non trivial period of time , for if they do
so, their sponsors will withdraw the fund.
Value strategies will generate positive payoffs
in the long run. It is too late and money
managers may lose their jobs. Thus they are
deprived of the arbitrage benefits emanating
from value stocks.
 Behavioral Theory of Asset Pricing
• It should rest on assumptions about investors behavior
that are either a priori plausible or consistent with
causal observation
• Evidence required with parsimony and unified evidence
( slowly diffusing news about future fundamentals)
without any other exogenous source of investor
sentiment and liquidity disturbances)
• Prediction must be testable and prone to validation
• It has bounded rationality referring to the ability to
process small subset of information in an unbiased way)
 Testable Predictions in the Theory
• Under reaction and overreaction are more
pronounced in small and low analyst coverage
firms where information diffuse very slowly
• There should be relationship between
momentum traders’ horizon and pattern of
return auto correlation
• The moment the news reaches to public
domain from private one, under- reaction is
surpassed by overreaction
 Psychological Biases have important role to
play
• Investors’ overconfidence about the precision of private
information…..( It implies negative long lag autocorrelations,
excess volatility, and when managerial actions are correlated
with stock mispricing, public event based return
predictability).
• Biased self attribution causing asymmetric shift in investors
confidence as a function of their investment outcomes
( positive short lag autocorrelations or momentums or short
run earnings drift.
• Confidence is also time dependent, priori dependent, and
framing. Also =f( age, education, experience, status of wealth)
 Few Mixed Explanations
• Investors +analysts……..generate information
1. By interviewing management
2. Verifying rumors
3. Analyzing financial statements
• Investors may boast of overconfidence and their
data generating skills and can think of lesser pricing
errors in forecasting….underestimation of errors
• Stock market over-react to private signal and under
react to public signal
 Cont.
• Repurchase ……….proxy of under-valuation……..predict positive
abnormal returns
• Equity offering ….proxy of overvaluation…….predict negative
abnormal return
• Cash flow or earnings surprise….initially confidence
increases…..cause same direction average stock price trend
…..later reversal…..overreaction
• Investors are quasi rational (Bayesian optimizers)
• Smart traders should dominate rational ones but never happens
• Analysts or investors may form “Herd” in few stocks when they
receive some private information prior to others
 Explanation of Self Attribution Theory
• Too strongly attribute events that confirm the validity of
their actions to high ability (Take credit for past success
and blame external factors for failures)
• Rather than Events that disconfirm the actions to
external noise or sabotage
• It adds confidence if the public information is in
congruent with his beliefs
• But confidence does not fall commensurately when
public information contradicts his private information
• HEADS I WIN, TAILS ITS CHANCE
 Cont.
• Notion of cognitive dissonance in which
individuals internally suppress information that
conflicts with past choices
• Good news after buy and bad news after sale
are boasting ones
• Investors does something….further confirmed
by public information ……become
overconfident…..momentum…..carry on…..long
run….long term reversal
 Explanation of Overconfidence
• More prevalent in diffuse task ( in judgment based
rather than lab one)
• High for task where feedback is delayed
• Lead to underestimating of forecasting errors
• People overestimate their abilities and perceive more
favorably than they are viewed by others
• “perhaps the most robust findings in the
psychology of judgment is that people are
overconfident” in all professionals.
 BSV arguments
• Representative biases…..too much weigh to
recent pattern of data……too little to the
properties of the population they belong to
• Conservatism biases…….slow updates of
models or information in the face of new
evidence
 DHS arguments
• Investors are of two types
1. Informed …………….prone to judgmental bias
emanating from overconfidence and biased
self attribution
2. Uninformed …………..no judgmental
biases………..passive investors
 Road ahead
• Under-reaction and overreaction are equally frequent and
probable. It is chance results.
• It is randomly split
• It can not be timed
• Thus indirectly leading to market efficiency
• Over-reaction is a long term anomaly….we have to be
cautious and sensitive to methodology, time period…..it is
highly likely that it may disappear in the return generating
process. Long term anomalies are statistically and
economically marginal and can be reduced by measurement
biases.
 Road Ahead
• Anomaly may disappear if methodology of
estimating abnormal return is reasonably changed.
It is on shaky footing. It is highly likely that anomaly
could be the outcome of replication and robustness
checks that followed publication of original studies
• On the other hand Under-reaction is
only Pyrrhic victory for market
efficiency.
 Return Metric
• BHAR ( Buy and Hold Abnormal Return)
• AAR (Average Abnormal Return)
• Time: daily, monthly, quarterly, yearly
 Other Interpretations
• Anomalies may be ubiquitous but attribute to
bad model problems
• It could be methodological illusions
• Are you dredging for anomalies……rewarding
occupation …………..called as “data snooping”
evidences.
• Can be spurious correlations
• Publication hungry professors also distort the
information (publish or perish)
 Cont.
• Bad models contaminated with mugged or
recurring factors only.
• We require beta and alpha from out of the
period and out of the sample for solving bad
model problems
• Sorting can’t be restricted to size and
BE/ME……….styled managers
 Cont.
• Anomaly is no guarantee against efficiency