bubbles crashes slides
TRANSCRIPT
-
8/6/2019 Bubbles Crashes Slides
1/49
1
Bubbles and Crashes
Dilip AbreuPrinceton University
Markus K. BrunnermeierPrinceton University
Hedge Funds and
the Technology Bubble
Markus K. Brunnermeier
Princeton University
Stefan Nagel
London Business School
-
8/6/2019 Bubbles Crashes Slides
2/49
2
Company X introduced a revolutionary wirelesscommunication technology.
It notonly provided support for such a technology but also
providedthe informational content itself.Its IPO price was $1.50 per share. Six years later it wastraded at $ 85.50 and in the seventh year it hit $ 114.00.
The P/E ratio got as high as 73.
The company never paiddividends.
Story of a typical technology stock
About RCA: READ Bernheim et al. (1935)The Security Market Findings and
Recommendations of a special staff of the 20th century fund - p. 475 and following
-
8/6/2019 Bubbles Crashes Slides
3/49
3
Story of RCA - 1920s
Company: Radio Corporation of America (RCA)
Technolgoy: Radio
Year: 1920s
Itpeake
d
at$ 397 in Feb. 1929,
do
wnto
$ 2.62 in May 1932,
1
1
2
2
3
3
time
$
Dec 25 Dec 50
(was < $ 14 till June 1945)
-
8/6/2019 Bubbles Crashes Slides
4/49
4
Internet bubble? - 1990sNASDAQ Combined Composite Index NEMAX All Share Index (German Neuer Markt)
38 day average
Chart (Jan. 98 - Dec. 00)
38 day average
Chart (Jan. 98 - Dec. 00) in Euro
Loss ofca. 60 %
from high of$ 5,132
Loss ofca. 85 %85 %
from high ofEuro 8,583
Why do bubbles persist?
Do professional traders ride the bubble or
attack the bubble (go short)?
What happened in March 2000?
Was it a bubble?
If it was a bubble, the question arises
Moving right along to the 1990s
-
8/6/2019 Bubbles Crashes Slides
5/49
5
Do (rational) professional ride the bubble?
South Sea Bubble (1710 - 1720)Isaac Newton
04/20/1720 sold shares at 7,000 profiting 3,500
re-enteredthe market later - ended up losing 20,000
I can calculate the motions ofthe heavenly bodies, butnotthe madness of people
Internet Bubble (1992 - 2000)
DruckenmillerofSoros Quantum Funddidntthink
thatthe party would end so quickly.We thought it was the eighth inning, and it was the ninth.
Julian Robertson of Tiger Fund refusedto invest ininternet stocks
-
8/6/2019 Bubbles Crashes Slides
6/49
6
The moral ofthis story is that irrational marketcan kill you
Julian said This is irrational and I wont play and
they carried
himout feet first.
Druckenmiller said This is irrational and I will playandthey carried him out feet first.
Quote of a financial analyst, New York TimesApril, 29 2000
Pros dilemma
-
8/6/2019 Bubbles Crashes Slides
7/49
7
Classical Question
Suppose behavioral trading leads to mispricing.Suppose behavioral trading leads to mispricing.
Can mispricings or bubbles persist in thepresence of rational arbitrageurs?
Whattype of information can leadtothebursting of bubbles?
-
8/6/2019 Bubbles Crashes Slides
8/49
8
Main Literature
Keynes (1936) bubble can emergebubble can emergeIt might have been supposedthatcompetition between expert professionals,possessing judgment and knowledge beyondthatofthe average privateinvestor, would correctthe vagaries ofthe ignorant individual leftto himself.
Friedman (1953), Fama (1965)Efficient Market Hypothesis no bubbles emergeno bubbles emerge
Ifthere are many sophisticatedtraders in the market, they may cause thesebubbles to burst before they really get under way.
Limits to ArbitrageNoise trader risk versus Synchronization risk
Shleifer & Vishny (1997), DSSW (1990 a & b)Bubble Literature
Symmetric information - Santos & Woodford (1997)
Asymmetric information
Tirole (1982), Allen et al. (1993), Allen & Gorton (1993)
-
8/6/2019 Bubbles Crashes Slides
9/49
9
Timing Game - Synchronization
(When) will behavioral traders beoverwhelmed by rational arbitrageurs?
Collective selling pressure of arbitrageurs
more than suffices to burstthe bubble.Rational arbitrageurs understandthat aneventualcollapse is inevitable.
But when?Delicate, difficult, dangerous TIMING GAME!
-
8/6/2019 Bubbles Crashes Slides
10/49
10
Elements of the Timing Game
Coordination at least > 0 arbs have to be outofthe market
Competition only first < 1 arbs receive pre-crash price.
Profitable ride ride bubble (stay in the market) as long as possible.
Sequential Awareness
A Synchronization Problem arises!Absentof sequential awareness
competitive elementdominates and bubble burst immediately.With sequential awarenessincentive to TIME THE MARKET leads to delayed arbitrage and
persistence of bubble.
-
8/6/2019 Bubbles Crashes Slides
11/49
11
model setup
introduction
preliminary analysis
persistence ofbubbles
public events
conclusion
price cascades and rebounds
-
8/6/2019 Bubbles Crashes Slides
12/49
12
Model setup
tt
0 t0+t0+
randomstarting
point
t0+
maximum life-span ofthe bubble
tradersare aware ofthe bubble
all tradersare aware ofthe bubble
bubble burstsfor exogenous
reasons
0
paradigm shift
- internet 90s
- railways- etc.
common action of arbitrageurssequential awareness(random t0 with F(t0) = 1 - exp{-t0}).
1
1/
pt
-
8/6/2019 Bubbles Crashes Slides
13/49
13
Payoff structure
Cash Payoffs (difference)Sell one share att- insteadof att.
pt- er - pt
wherept=
Execution price atthe time of bursting.
prior to the crash
after the crash
forfirst random orders up to
all otherorders
-
8/6/2019 Bubbles Crashes Slides
14/49
14
Payoff structure (ctd.), Trading
Small transactions c
osts ce
rt
Risk-neutrality but max/min stock position
max long position
max short position
due to capital constraints, margin requirements etc.
Definition 1: trading equilibrium
Perfect Bayesian Nash Equilibrium
Belief restriction: trader who attacks attime tbelieves that all traders who became aware ofthebubble priorto her also attack att.
Definition 1:
-
8/6/2019 Bubbles Crashes Slides
15/49
15
introduction
persistence ofbubbles
public events
conclusion
price cascades and rebounds
model setup
Preliminary analysis
preemption motive - trigger strategies
sell out condition
-
8/6/2019 Bubbles Crashes Slides
16/49
16
TriggerStrategies
Bursting date T*(t0
)=min{T(t0
+), t0
+ }
Role of Preemption Motive
Rules out coordinated sell outon Friday July 13th.
Bubble never bursts with strictly positive prob. at some t13.
Suppose it w
ould, then selling pressure w
ould
exceed
with prob>0.
Hence, price woulddrop already att13 incentive to sell out earlier
well defineddensity of bursting date (t|ti) for each arb.
Proposition 1: Trigger strategies.
Given c > 0, arb tinever sells outonly for an instant.He stays outofthe market at least until ti+ sells out.
Arb ti+ stays out until ti+ 2 exits and soon.
By trading equilibrium, arb ti stays out until ti+ exits.also illustrates failure ofstrategic complementarity
(pre-empt)
iftraders condition on calendar time
Proposition 1:
-
8/6/2019 Bubbles Crashes Slides
17/49
17
Sell out condition for periods
sell out attif
r i tio r te
benefit of ttacking cost ofattacking
RHSconverges to [(g-r)] as t
bursting date T*(t0)=min{T(t0+), t0+ }
h(t|ti)Et[bubble|] (1-h(t|ti) (g- r)pt
-
8/6/2019 Bubbles Crashes Slides
18/49
18
introduction
preliminary analysis
public events
conclusi
on
price cascades and rebounds
model setup
persistence of bubbles
exogenous crashes
endogenous crashes
lack of common knowledge
-
8/6/2019 Bubbles Crashes Slides
19/49
19
Persistence of Bubbles
Proposition 1: Suppose .
existence of a unique trading equilibrium
traders begin attacking after a
delay
ofperiods.
bubble does notburstdue to endogenous sellingpriorto .
Proposition 2:
-
8/6/2019 Bubbles Crashes Slides
20/49
20
Sequential awareness
t
traderti
ti-
since ti t0+
Distribution oft0
t0 t0+
since ti t0
ti
tk
Distribution oft0+
(bursting of bubble if nobody attacks)
t
tradertjtjtj-
t
tradertk
_
-
8/6/2019 Bubbles Crashes Slides
21/49
21
Conjecture 1: Immediate attack
Bubble bursts at t0+
when traders are aware ofthe bubble
Ift0< ti- , the bubble
would have burst already.
/(1-e-)
Distribution oft0
Distribution oft0+
tti- ti- ti+ti
-
8/6/2019 Bubbles Crashes Slides
22/49
22
Conj. 1 (ctd.): Immediate attack
t
Bubble bursts at t0+
Distribution oft0+
Bubble burstsfor sure!
hazard rate ofthe bubbleh = /(1-exp{-(ti+ -t)})
/(1-e
-
)
ti- ti- ti+ti
Distribution oft0
-
8/6/2019 Bubbles Crashes Slides
23/49
23
Conj. 1 (ctd.): Immediate attack
t
Bubble bursts at t0+
Bubble burstsfor sure!
hazard rate ofthe bubbleh = /(1-exp{-(ti+ - t)})
/(1-e
-
)
ti- ti- ti+ti
Distribution oft0
optimal time
to
attack ti+
i
delayed attack is optimaldelayed attack is optimalno immediate attack equilibrium!no immediate attack equilibrium!
bubble appreciation / bubble size
Recall the sell out condition:
_lower bound: (g-r)/ >/(1-e-)
-
8/6/2019 Bubbles Crashes Slides
24/49
24
t
hazard rate ofthe bubbleh = /(1-exp{-(ti+ + -t)})
ti- ti
Conj. 2: Delayed attack by arbitrary
Bubble bursts at t0 + + < t0 +
ti- + + ti+ +ti+
optimal todelayattack even moreeven more
conjecturedattack
attack is never successfulattack is never successful bubble bursts for exogenous reasons atbubble bursts for exogenous reasons at t0 +
lower bound: (g-r)/ >/(1-e-)
bubble appreciationbubble size
/(1-e-)
_
_
_
-
8/6/2019 Bubbles Crashes Slides
25/49
25
Endogenous crashes
Proposition 3: Suppose .unique trading equilibrium.
traders begin attacking after a delay of* periods.
bubble bursts due to endogenous selling pressureat a size ofpt times
Proposition 3:
arbitrageurs eventuallyburst bubblebut very late
(bridgebetween traditional analysis and Proposition 1)
-
8/6/2019 Bubbles Crashes Slides
26/49
26
Endogenous crashes
t
hazard rate ofthe bubbleh = /(1-exp{-(ti+ + - t)})
ti- ti- ti
lower bound: (g-r)/ >/(1-e-)
Bubble bursts at t0+ +*
ti- + +**ti++**ti+**
optimal
conjecturedattack
bubble appreciation
bubble size
_
-
8/6/2019 Bubbles Crashes Slides
27/49
27
Endogenous crashes - deriving *
In equilibrium traderti= t0+
bursts the bubble.
When she sells his shares her supportoft0 is [ti- , ti],hence his hazard rate is h = /(1-exp{-})
(1)
The bubble bursts atti= t0+ +*,hence it bursts at a size of egt*(*)bubble appreciation/ size = (g-r+z) / *(*) (2)
equilibriumh
(1)
bubble appreciationbubble size
(2)
*
-
8/6/2019 Bubbles Crashes Slides
28/49
28
Comparative statics
Role of information dispersion , Priordistribution oft0 F(t0) = 1 - exp{-t0}
the smaller, the larger*, the size of bubble
t0= 0, no infodispersion no bubble 0 distributions uniform [size is (g-r) ]
Dispersion ofopinion
as bubbles size
for exogenous crash
Role of momentum traders same as for More synchronization required More synchronization required
-
8/6/2019 Bubbles Crashes Slides
29/49
29
Lack of common knowledge
t0 t0+
standard backwards induction cant be appliedstandard backwards induction cant be applied
t0+everybody
knows ofthethe bubble
traders
know ofthe bubble
everybody knows thateverybody knows ofthe
bubble
t0+ 2 t0+ 3
everybody knows that
everybody knows thateverybody knows of
the bubble
(same reasoning applies for traders)
Ifone interprets as difference in opinion,lack of common knowledge gets a different meaning too.
-
8/6/2019 Bubbles Crashes Slides
30/49
30
introduction
preliminary analysis
persistence ofbubbles
conclusion
price cascades and rebounds
synchronizing events
model setup
-
8/6/2019 Bubbles Crashes Slides
31/49
31
Role of synchronizing events (information)
News may have an impactdisproportionateto any intrinsic informational (fundamental)content.
News can serve as a synchronization device.
Fads & fashion in information
Which news shouldtraders coordinate on?
When synchronized attack fails, the bubbleis temporarily strengthened.
-
8/6/2019 Bubbles Crashes Slides
32/49
32
Setting with synchronizing events
Focus on news with no informational content(sunspots)
Synchronizing events occur with Poisson arrivalrate .
Note thatthe pre-emption argumentdoes not apply sinceeventoccurs with zero probability.
Arbitrageurs who are aware ofthe bubble becomeincreasingly worried about itovertime.
Only traders who became aware ofthe bubble more thane periods agoobserve (look out for) this synchronizingevent.
-
8/6/2019 Bubbles Crashes Slides
33/49
33
Synchronizing events - Market rebounds
Proposition 5: In responsive equilibriumSell out a) always atthe time of a public eventte,
b) afterti+** (where **te - e -
without public event, they would have learntthisonly atte + e - **.
the existence of bubble attreveals thatt0>t- ** -
that is, no additional information is revealedtill te - e +**
density that bubble bursts for endogenous reasons is zero.
Proposition 5:
-
8/6/2019 Bubbles Crashes Slides
34/49
34
introduction
preliminary analysis
persistence ofbubbles
public events
conclusion
model setup
price cascades andrebounds
-
8/6/2019 Bubbles Crashes Slides
35/49
35
Price cascades and rebounds
Price drop as a synchronizing event.through psychological resistance line
by more than, say 5 %
Exogenous price dropafter a price drop
if bubble is ripe bubble bursts and price drops further.
if bubble is not ripe yet price bounces back andthe bubble is
strengthened for some time.
-
8/6/2019 Bubbles Crashes Slides
36/49
36
Price cascades and rebounds (ctd.)
Proposition 6:Sell out a) after a price drop ifip(Hp)
b) afterti+*** (where ***
-
8/6/2019 Bubbles Crashes Slides
37/49
37
Conclusion of Bubbles and Crashes
Bubbles
Dispersion ofopinion among arbitrageurs causes asynchronization problem which makes coordinatedprice corrections difficult.
Arbitrageurs time the market and ride the bubble.
Bubbles persist
Crashes
can be triggered by unanticipated news without anyfundamental content, since
it might serve as a synchronization device.
Rebound
can occur after a failed attack, which temporarily
strengthens the bubble.
(technological revolutions etc.)
-
8/6/2019 Bubbles Crashes Slides
38/49
38
Hedge Funds andthe Technology Bubble
Markus K. BrunnermeierPrinceton University
Stefan NagelLondon Business School
http://www.princeton.edu/~markus
-
8/6/2019 Bubbles Crashes Slides
39/49
39
reasons for persistence
data
empirical results
conclusion
-
8/6/2019 Bubbles Crashes Slides
40/49
40
1. Unawareness of Bubble Rational speculators perform as badly as others when market collapses.
2. Limits to ArbitrageFundamental risk
Noise trader riskSynchronization risk
Short-sale constraint
Rational speculators may be reluctant to go shortoverpriced stocks.
3. Predictable InvestorSentimentAB (2003), DSSW (JF 1990)
Rational speculators may wanttogo longoverpriced stock and
try to go short priorto collapse.
Why Did Rational Speculation Fail toPreventthe Bubble ?
About RCA: READ Bernheim et al. (1935)The Security Market Findings and
Recommendations of a special staff of the 20th century fund - p. 475 and following
-
8/6/2019 Bubbles Crashes Slides
41/49
41
data
reasons for persistence
empirical results
conclusion
-
8/6/2019 Bubbles Crashes Slides
42/49
42
Data
Hedge fund stock holdings
Quarterly 13 F filings toSEC
mandatory for all institutional investorswith holdings in U.S. stocks of more than $ 100 million
domestic and foreign
at manager level
Caveats: No short positions
53 managers with CDA/Spectrum data
excludes 18 managers b/c mutual business dominates
incl. Soros, Tiger, Tudor, D.E. Shaw etc.
Hedge fund performance data
HFR hedge fund style indexes
(technological revolutions etc.)
43
-
8/6/2019 Bubbles Crashes Slides
43/49
43
data
co
nclusio
n
reasons for persistence
empirical results
did hedge funds ride bubble?
did hedge fundstiming pay off?
-
8/6/2019 Bubbles Crashes Slides
44/49
45
-
8/6/2019 Bubbles Crashes Slides
45/49
45
Fig. 4a: eight of technology stoc s in hedge fund ortfolios versus eight in
ar et ortfolio
0.00
0.20
0.40
0.60
0.80
Mar-98 Jun-98 Sep-98 Dec-98 Mar-99 Jun-99 Sep-99 Dec-99 Mar-00 Jun-00 Sep-00 Dec-00
Proportion invested in NA DA high P/S stocks
Zw e ig- DiMenna
Soros
Hus ic
Market Portfolio
OmegaTiger
Did Sorosetc. ride thebubble?
46
-
8/6/2019 Bubbles Crashes Slides
46/49
46
Fig. 4b: Funds flo s, three- onth oving average
-0.20
-0.
-0. 0
-0.05
0.00
0.05
0. 0
Mar- 98 Jun-9 8 Sep-98 Dec-98 Mar- 99 Jun-9 9 S ep-99 Dec-99 Mar- 00 Jun-00 Sep-00 Dec-00
Fund flow sasproportionofassetsunder anagem ent
Quantu und (Soros)
Jaguar und (Tiger)
Fund in- andoutflows
47
-
8/6/2019 Bubbles Crashes Slides
47/49
47
0.00
0.10
0.20
0.30
0.40
0.50
0.60
-4 -3 -2 -1 0 1 2 3 4
Quarters around Price Peak
High P/S NASDAQ Other NASDAQ NYSE/AMEX
Share of equity held (in %)
Figure 5. Average share of outstanding equity held by hedge funds around rice ea s
of individual stoc s
Did hedge funds time stocks?
48
-
8/6/2019 Bubbles Crashes Slides
48/49
48
Figure 6: Performance of a copycat fund that replicates hedge fund holdings in the
ASDAQ high P/S segment
Mar-98 Jun-98 Sep-98 Dec-98 Mar-99 Jun-99 Sep-99 Dec-99 Mar-00 Jun-00 Sep-00 Dec-00
Total return index
High /S Copycat und All High /S ASDAQStocks
.0
2.0
3.0
4.0
Did hedge funds timing pay off?
49
-
8/6/2019 Bubbles Crashes Slides
49/49
49
Conclusion
Hed
ge fund
s were rid
ing the bubbleShort sales constraints and arbitrage riskare not sufficientto explain this behavior.
Timing bets of hedge funds were well
placed. Outperformance!Rules out unawareness of bubble.
Suggests predictable investor sentiment.
Riding the bubble for a while may havebeen a rational strategy.
Supports bubble-timing models
(technological revolutions etc.)