1015 Didier Sornette PDF
1015 Didier Sornette PDF
1015 Didier Sornette PDF
billions US $
30’000 Stock markets losses
10’000
World GDP loss
The Global Bubble (2003-2008)
and
the illusion of the perpetual money machine (1980-2008)
Index of
over-
valuation
5
Code
Red,
by
John
Mauldin
and
Jonathan
Tepper
(Oct.
2013)
Easy
Money
will
lead
to
more
Bubbles
p l y
y
Positive feedbacks
su
p
p pl
s u
de
m nd
an a
d em
d
quantity quantity
Mechanisms for positive feedbacks in the stock market
!11
Positive feedbacks until 2008...
•Financial innovations
Real-estate loans and MBS as fraction of bank assets
Securitization of finance
Leverage
Under-estimating aggregate risks
•Outsourcing of risks
✴Deregulation and lack of oversight
✴Bad quantitative risk models in banks (Basel II)
✴Rating agency failures
✴Lowering of lending standards
✴Managers greed and poor corporate governance problem
•Facilitating factors
✦ Loans to poor’s
✦Freddy Mac and Fanny Mae...
✦Growth of over-capacity
✦Moral hazards
Positive feedback leads to SYNCHRONIZATION
Z. Néda, E. Ravasz, Y. Brechet, T. Vicsek and A.-L. Barabási, Self-organizing processes: The sound of many hands clapping Nature 403, 849-850 (2000)
Universal bubble scenario => ENDOGENEITY
and POSITIVE FEEDBACK
Charles Kindleberger, Manias, Panics and Crashes (1978)
Didier Sornette, Why stock markets crash (2003)
Financial bubbles, which we have been observing for over 400 years:
! exponential growth
p(t) ⇠ e t
! finite-time singularity
! 1
p(t) ⇠
! (tc t)1/
! power-law
|
p(t) ⇠ t1/|
Super-exponential growth
(positive feedbacks)
Korotayev, Malkov
Khaltourina (2006)
GDP
Multivariate endogeneous growth models and FTS
Case θ+β>1 : FTS
technology
capital
•Singularity theorems of Penrose and Hawking in General Relativity
!
•Formation of black holes (general relativity coupled to a mass field)
!
•Planet formation in solar system by run-away accretion of planetesimals
!
Finite-time •Tokamak physics of plasma turbulence (Zakharov equation of beam-driven
turbulence)
!
Singularity •Front genesis in meteorology
!
•Turbulence and Euler equation for inviscid fluids
!
•Rupture and failure of materials and engineering structures
!
•Earthquakes (Ruina-Dieterich state-and-velocity weakening solid friction)
!
•Micro-organisms chemotaxis aggregating to form fruiting bodies
!
•Surface instabilities to form spikes (Mullins-Sekerka instability)
!
•Jets from singular surfaces
!
•Fluid droplet snap-off
!
•Euler’s disk (rotating settling coin)
!
Famous historical bubbles
Semper Augustus
Source: Elliott Wave International; data source for South Seas, Global Financial Data
Positive feedbacks and origin of bubbles
Price
return(t+1)
return(t)
time
Prices in the learning-to-forecast market experiments
Next period returns r(t+1)versus current returns r(t) for
(Hommes et al., 2008).
group 2. Points on the diagonal correspond to constant
Five out of six markets exhibit long lasting bubbles with
growth rate (r(t+1)= r(t)), points above the diagonal (r(t
asset prices increasing to more than 15 times fundamental
+1)> r(t)) correspond to accelerating growth. Returns
value.
are defined as discrete returns:
r(t+1) = [p(t+1)/p(t)] − 1.
A. Hüsler, D. Sornette and C. H. Hommes Super-exponential bubbles in lab experiments: evidence for anchoring over-optimistic expectations on price,
Journal Economic Behavior and Organization 92, 304-316 (2013)
I Fulvio Corsi and Didier Sornette (2012)
II
Martin Pleisch Didier Sornette Fulvio Corsi (2013)
II
Synthetic tests with simulated FTS-GARCH time series
(false negative test)
ex-post diagnostic
ex-ante signal
t-statistic
ex-ante signal
NASDAQ Composite 1971-2001: estimated gamma (upper panel) and corresponding t-statistic (lower panel) in a FTS GARCH model
t-statistic
NIKKEI 1976-1992: estimated gamma (upper panel) and corresponding t-statistic (lower panel) in a FTS GARCH model
t-statistic
DAX 1991-2000: estimated gamma (upper panel) and corresponding t-statistic (lower panel) in a FTS GARCH model (yt = log(Pt))
Rational Expectation Bubbles and Crashes
(Blanchard-Watson)
!28
Stochastic finite-time singularity
The price
drives the
crash hazard
rate.
D. Sornette and J.V.
Andersen
A Nonlinear Super-
Exponential Rational
Model of Speculative
Financial Bubbles,
Int. J. Mod. Phys. C 13
(2), 171-188 (2002)
!29
!30
Contains two ingredients:
(1) growth faster than exponential
(2) growth of volatility
!31
Example of a “fearful” super-exponential bubble
!32
Hong-Kong
Patterns of price trajectory during 0.5-1 year before each peak: Log-periodic power law
!33
Real-estate in the USA
r n
tu
r e
ice
pr
Real U.S. House Prices between 1974 and 2014. Levels are shown in black and should be read on the left axis. Yearly growth rates
are shown in blue and should be read on the right axis. Three peaks in the growth rate coincide with a correction in the levels. When
the growth itself grows, the process becomes unstable and a correction follows (Source: Federal Reserve Bank of Dallas
international house price dataset, http://www.dallasfed.org/institute/houseprice/)
DISCRETE HIERARCHY
OF THE AGENT NETWORK
Presentation of three different mechanisms leading to discrete scale invariance, discrete
hierarchies and log-periodic signatures
37
Soramäki, K. et al. Physica A 379, 317–333 (2007).
network topology of the interbank payments transferred between
commercial banks over the Fedwire® Funds Service
Kimmo Soramäki, Morten L. Bech, Jeffrey Arnold, Robert J. Glass, and Walter E. Beyeler, Physica A 379, 317-333 (2006)
Rational expectation bubble models
with social interactions
Key idea: return-risk relationship also holds during bubbles via
the no-arbitrage (or close to no-arbitrage) condition.
Martingale hypothesis
(no “free lunch”)
Rationalthe
addresses expectation bubble
problem of the joint Rational expectation models
model of
estimation in the
the fundamental
presence of and of negative bubbles
an (unknown) fundamental value
bubble components and anti-bubbles
Bubble diagnostic if
(iii)
(iv) Dickey Fuller unit root test is rejected at 99.5% significance level
Li Lin, Didier Sornette, Diagnostics of Rational Expectation Financial Bubbles with Stochastic Mean-Reverting Termination Times,
The European Journal of Finance 19 (5-6) 344-365 (2013).
Extensions of the Log-Periodic Power Law model
Rationalthe
addresses expectation bubble
problem of the joint Rational expectation models
model of
estimation in the
the fundamental
presence of and of negative bubbles
an (unknown) fundamental value
bubble components and anti-bubbles
Rational expectation
addresses the critic of bubble
Granger Rational expectation bubble
andmodel in the
Newbold presence
(1974) of
and Phillips model with higher order solutions
mean-reverting
(1986) self-consistent
about spurious fits of non- of the RG
residuals
stationary price processes
A Consistent Model of ‘Explosive’ Financial Bubbles!
With Mean-Reversing Residuals!
L. Lin, R. E. Ren and D. Sornette (2009)
http://papers.ssrn.com/abstract=1407574
Rational Expectation formulation
t = tc Time
A. Johansen and D. Sornette, Financial “anti-bubbles”: log-periodicity in Gold and Nikkei collapses, Int. J. Mod. Phys. C 10(4),
563-575 (1999); Evaluation of the quantitative prediction of a trend reversal on the Japanese stock market in 1999, Int. J.
Mod. Phys. C Vol. 11 (2), 359-364 (2000)
Methodology for diagnosing bubbles
" Positive feedbacks of higher return anticipation
✴ Super exponential price
✴ Power law “Finite-time singularity”
!
" Positive feedback of negative spirals of crash
expectation
✴ Accelerating large-scale financial volatility
✴ Log-periodic discrete scale-invariant patterns
• Debt bubbles
6 months
7 years
Super-exponential growth
!59
Real-estate in the UK
Wei-Xing Zhou and Didier Sornette 2000-2003 Real Estate Bubble in the UK but not in the USA, Physica A 329, 249-263 (2003)
Real-estate in the USA
bubble peaking in Oct. 2007
62
data
positive bubble
negative bubble
The fingerprint of a positive (in green) and a negative bubble (in red) can be seen in the S&P500 before and during the credit crisis.
Notice the distinct pattern combining oscillations with a faster-than-exponential rise (or drop for a negative bubble) in the price. For
both bubbles, the LPPL model is fitted to the data in the window between t1 and t2.The critical time tc is given as a result of the model,
it is the most probable time for a change in regime. This may be a crash after a positive bubble or a rally following a negative bubble.
2006-2008 Oil bubble
Speculation vs supply-demand
D. Sornette, R. Woodard
and W.-X. Zhou, The
2006-2008 Oil Bubble
and Beyond, Physica A
388, 1571-1576 (2009)
(arXiv.org/abs/0806.1170)
Typical result of the calibration of the simple LPPL model to the oil price in US$ in shrinking windows with starting dates
tstart moving up towards the common last date tlast = May 27, 2008. !64
CORN GOLD
SOYBEAN
WHEAT
!65
Total liabilities of the U.S. financial and non- financial sectors divided by the GDP
!
The data are taken from the Flow of Funds accounts of the U.S. (http://www.federalreserve.gov/releases/z1/), the non-financial
sector includes the federal government, government sponsored entities, household and non-profit and non-financial business. The
smooth curves show the fits of the models.
This picture demonstrates that debt levels are on unsustainable tracks that, according to our
bubble models, are expected to reach a critical point towards the end of the present decade.
The efficiency of U.S. Non Financial debt. The dots tell you how many USD of GDP are
generated for each USD of U.S. Non Financial debt. The black double line gives an LPPL model fit and the
grey line gives the DS LPPL Trust index. When this index exceeds the 5% level, the process is not
sustainable and there is a substantial risk for a critical transition of the system.
MISLEADING METRICS: THE GREAT MODERATION
7500000
tru
th
5000000
believe
2500000 religio
n
0
1945 1952 1959 1966 1973 1980 1987 1994 2001 2008
rate of profit
Rate of profit and rate of accumulation: The
United States + European Union + Japan
* Rate of accumulation = rate of growth rate
of the net volume of capital * Rate of profit
= profit/capital (base: 100 in 2000)
!71
Financial Crisis Observatory
Zurich
Australia
Belgium
Canada
Advanced real-estate bubble detection
Switzerland
Germany
Denmark
Advanced real-estate bubble detection
Spain
Finland
France
Advanced real-estate bubble detection
UK
Ireland
Italy
Advanced real-estate bubble detection
Japan
South Korea
Luxembourg
Advanced real-estate bubble detection
Netherlands
Norway
New Zealand
Advanced real-estate bubble detection
Sweden
US
South Africa
Overview of the different asset classes and indices!
FCO Cockpit!
(end of January 2014 calculations, ETHZ FCO)
!
•
The DS LPPL ConfidenceTM indicator measures the sensitivity of the observed
bubble pattern to the chosen starting time. If the value is close to zero, it means that
the pattern was found only in a few of the data windows that were analyzed; if the
value is close to one, it means that the pattern is practically insensitive to the choice
of the data window. A very low value of this indicator means that the signal is not
robust and only pops up in one or two specific data windows, in that case, there is a
risk of over-fitting or data snooping and the result should be interpreted with care;
!
•
The DS LPPL TrustTM indicator measures how closely the pattern matches the
theoretical LPPL model, zero being a bad and one being a perfect match.
Zurich
Overview of the different asset classes and indices!
monitored in the FCO Cockpit!
(end of January 2014 calculations, ETHZ FCO)
436 systemic financial assets or indices in fixed income, equities, commodities and currencies
Clear bubble signals in Spanish, Irish and Italian government bonds and in
European financial subordinated bonds
!
recovery in the form of bubble exuberance
Energy products down -2%, agricultural products -12%, industrial metals -14% and
precious metals down a whopping -29%.
!
Weak negative signal bubbles
!
This is in great dissonance with the equities bull market that we are riding now.
GOLD
Gold Bullion(Zurich) kg(995) CHF
Significant weakening of some emerging markets currencies
!
Negative bubble for the Argentine Peso, the Turkish Lira and the
Indonesian Rupiah
!
Clear indication of an overreaction in this market.
Turkish lira vs ChF
Worrying exuberance in equity markets
!
1 out 3 markets are diagnosed in bubble regime
BRICS disappointed in the past year with the Brazilian Bovespa down -21%, the Russian RTS
Index -12%, the Indian BSE +7% and the Shanghai Composite -11%.
This is to be compared with a Dow Jones index +20% and Eurostoxx +19%.
strong bubble warnings are flashing in internet retail equities (where Amazon is part of),
life sciences, consumer finance, auto components, aerospace and defense, health care…
Bubble warnings in social network companies,
venture capital and dividend paying stock
Highlights