How we can predict the next financial crisis
-
0:01 - 0:02Once upon a time
-
0:02 - 0:09we lived in an economy of financial growth and prosperity.
-
0:09 - 0:13This was called the Great Moderation,
-
0:13 - 0:16the misguided belief by most economists,
-
0:16 - 0:19policymakers and central banks
-
0:19 - 0:23that we have transformed into a new world
-
0:23 - 0:27of never-ending growth and prosperity.
-
0:27 - 0:31This was seen by robust and steady GDP growth,
-
0:31 - 0:34by low and controlled inflation,
-
0:34 - 0:37by low unemployment,
-
0:37 - 0:40and controlled and low financial volatility.
-
0:40 - 0:47But the Great Recession in 2007 and 2008,
-
0:47 - 0:51the great crash, broke this illusion.
-
0:51 - 0:56A few hundred billion dollars of losses in the financial sector
-
0:56 - 1:00cascaded into five trillion dollars
-
1:00 - 1:02of losses in world GDP
-
1:02 - 1:05and almost $30 trillion losses
-
1:05 - 1:09in the global stock market.
-
1:09 - 1:15So the understanding of this Great Recession
-
1:15 - 1:20was that this was completely surprising,
-
1:20 - 1:21this came out of the blue,
-
1:21 - 1:24this was like the wrath of the gods.
-
1:24 - 1:26There was no responsibility.
-
1:26 - 1:28So, as a reflection of this,
-
1:28 - 1:31we started the Financial Crisis Observatory.
-
1:31 - 1:35We had the goal to diagnose in real time
-
1:35 - 1:38financial bubbles
-
1:38 - 1:44and identify in advance their critical time.
-
1:44 - 1:47What is the underpinning, scientifically, of this financial observatory?
-
1:47 - 1:52We developed a theory called "dragon-kings."
-
1:52 - 1:55Dragon-kings represent extreme events
-
1:55 - 1:58which are of a class of their own.
-
1:58 - 2:01They are special. They are outliers.
-
2:01 - 2:04They are generated by specific mechanisms
-
2:04 - 2:07that may make them predictable,
-
2:07 - 2:10perhaps controllable.
-
2:10 - 2:14Consider the financial price time series,
-
2:14 - 2:16a given stock, your perfect stock,
-
2:16 - 2:18or a global index.
-
2:18 - 2:21You have these up-and-downs.
-
2:21 - 2:24A very good measure of the risk of this financial market
-
2:24 - 2:27is the peaks-to-valleys that represent
-
2:27 - 2:29a worst case scenario
-
2:29 - 2:33when you bought at the top and sold at the bottom.
-
2:33 - 2:36You can look at the statistics, the frequency of the occurrence
-
2:36 - 2:39of peak-to-valleys of different sizes,
-
2:39 - 2:41which is represented in this graph.
-
2:41 - 2:44Now, interestingly, 99 percent
-
2:44 - 2:48of the peak-to-valleys of different amplitudes
-
2:48 - 2:51can be represented by a universal power law
-
2:51 - 2:55represented by this red line here.
-
2:55 - 2:59More interestingly, there are outliers, there are exceptions
-
2:59 - 3:01which are above this red line,
-
3:01 - 3:05occur 100 times more frequently, at least,
-
3:05 - 3:09than the extrapolation would predict them to occur
-
3:09 - 3:13based on the calibration of the 99 percent remaining
-
3:13 - 3:15peak-to-valleys.
-
3:15 - 3:20They are due to trenchant dependancies
-
3:20 - 3:23such that a loss is followed by a loss
-
3:23 - 3:26which is followed by a loss which is followed by a loss.
-
3:26 - 3:29These kinds of dependencies
-
3:29 - 3:35are largely missed by standard risk management tools,
-
3:35 - 3:38which ignore them and see lizards
-
3:38 - 3:42when they should see dragon-kings.
-
3:42 - 3:46The root mechanism of a dragon-king
-
3:46 - 3:49is a slow maturation towards instability,
-
3:49 - 3:50which is the bubble,
-
3:50 - 3:53and the climax of the bubble is often the crash.
-
3:53 - 3:57This is similar to the slow heating of water
-
3:57 - 3:59in this test tube reaching the boiling point,
-
3:59 - 4:02where the instability of the water occurs
-
4:02 - 4:05and you have the phase transition to vapor.
-
4:05 - 4:09And this process, which is absolutely non-linear --
-
4:09 - 4:11cannot be predicted by standard techniques --
-
4:11 - 4:15is the reflection of a collective emergent behavior
-
4:15 - 4:18which is fundamentally endogenous.
-
4:18 - 4:21So the cause of the crash, the cause of the crisis
-
4:21 - 4:24has to be found in an inner instability of the system,
-
4:24 - 4:30and any tiny perturbation will make this instability occur.
-
4:30 - 4:34Now, some of you may have come to the mind
-
4:34 - 4:37that is this not related to the black swan concept
-
4:37 - 4:39you have heard about frequently?
-
4:39 - 4:42Remember, black swan is this rare bird
-
4:42 - 4:45that you see once and suddenly shattered your belief
-
4:45 - 4:48that all swans should be white,
-
4:48 - 4:51so it has captured the idea of unpredictability,
-
4:51 - 4:53unknowability, that the extreme events
-
4:53 - 4:56are fundamentally unknowable.
-
4:56 - 4:57Nothing can be further
-
4:57 - 5:00from the dragon-king concept I propose,
-
5:00 - 5:04which is exactly the opposite, that most extreme events
-
5:04 - 5:07are actually knowable and predictable.
-
5:07 - 5:10So we can be empowered and take responsibility
-
5:10 - 5:13and make predictions about them.
-
5:13 - 5:16So let's have my dragon-king burn this black swan concept.
-
5:16 - 5:17(Laughter)
-
5:17 - 5:20There are many early warning signals
-
5:20 - 5:22that are predicted by this theory.
-
5:22 - 5:25Let me just focus on one of them:
-
5:25 - 5:28the super-exponential growth with positive feedback.
-
5:28 - 5:29What does it mean?
-
5:29 - 5:31Imagine you have an investment
-
5:31 - 5:34that returns the first year five percent,
-
5:34 - 5:37the second year 10 percent, the third year 20 percent,
-
5:37 - 5:40the next year 40 percent. Is that not marvelous?
-
5:40 - 5:43This is a super-exponential growth.
-
5:43 - 5:45A standard exponential growth corresponds
-
5:45 - 5:49to a constant growth rate, let's say, of 10 percent
-
5:49 - 5:53The point is that, many times during bubbles,
-
5:53 - 5:56there are positive feedbacks which can be of many times,
-
5:56 - 6:00such that previous growths enhance,
-
6:00 - 6:03push forward, increase the next growth
-
6:03 - 6:06through this kind of super-exponential growth,
-
6:06 - 6:09which is very trenchant, not sustainable.
-
6:09 - 6:12And the key idea is that the mathematical solution
-
6:12 - 6:16of this class of models exhibit finite-time singularities,
-
6:16 - 6:19which means that there is a critical time
-
6:19 - 6:22where the system will break, will change regime.
-
6:22 - 6:25It may be a crash. It may be just a plateau, something else.
-
6:25 - 6:28And the key idea is that the critical time,
-
6:28 - 6:30the information about the critical time is contained
-
6:30 - 6:36in the early development of this super-exponential growth.
-
6:36 - 6:40We have applied this theory early on, that was our first success,
-
6:40 - 6:44to the diagnostic of the rupture of key elements
-
6:44 - 6:46on the iron rocket.
-
6:46 - 6:49Using acoustic emission, you know, this little noise
-
6:49 - 6:51that you hear a structure emit, sing to you
-
6:51 - 6:55when they are stressed, and reveal the damage going on,
-
6:55 - 6:57there's a collective phenomenon of positive feedback,
-
6:57 - 6:59the more damage gives the more damage,
-
6:59 - 7:02so you can actually predict,
-
7:02 - 7:04within, of course, a probability band,
-
7:04 - 7:06when the rupture will occur.
-
7:06 - 7:08So this is now so successful that it is used
-
7:08 - 7:13in the initial phase of [unclear] the flight.
-
7:13 - 7:15Perhaps more surprisingly, the same type of theory
-
7:15 - 7:18applies to biology and medicine,
-
7:18 - 7:22parturition, the act of giving birth, epileptic seizures.
-
7:22 - 7:25From seven months of pregnancy, a mother
-
7:25 - 7:31starts to feel episodic precursory contractions of the uterus
-
7:31 - 7:35that are the sign of these maturations
-
7:35 - 7:39toward the instability, giving birth to the baby,
-
7:39 - 7:41the dragon-king.
-
7:41 - 7:44So if you measure the precursor signal,
-
7:44 - 7:50you can actually identify pre- and post-maturity problems
-
7:50 - 7:51in advance.
-
7:51 - 7:55Epileptic seizures also come in a large variety of size,
-
7:55 - 7:58and when the brain goes to a super-critical state,
-
7:58 - 8:02you have dragon-kings which have a degree of predictability
-
8:02 - 8:07and this can help the patient to deal with this illness.
-
8:07 - 8:10We have applied this theory to many systems,
-
8:10 - 8:12landslides, glacier collapse,
-
8:12 - 8:16even to the dynamics of prediction of success:
-
8:16 - 8:20blockbusters, YouTube videos, movies, and so on.
-
8:20 - 8:23But perhaps the most important application
-
8:23 - 8:26is for finance, and this theory
-
8:26 - 8:29illuminates, I believe, the deep reason
-
8:29 - 8:32for the financial crisis that we have gone through.
-
8:32 - 8:36This is rooted in 30 years of history of bubbles,
-
8:36 - 8:40starting in 1980, with the global bubble
-
8:40 - 8:42crashing in 1987,
-
8:42 - 8:44followed by many other bubbles.
-
8:44 - 8:47The biggest one was the "new economy" Internet bubble
-
8:47 - 8:49in 2000, crashing in 2000,
-
8:49 - 8:51the real estate bubbles in many countries,
-
8:51 - 8:54financial derivative bubbles everywhere,
-
8:54 - 8:56stock market bubbles also everywhere,
-
8:56 - 9:00commodity and all bubbles, debt and credit bubbles --
-
9:00 - 9:03bubbles, bubbles, bubbles.
-
9:03 - 9:05We had a global bubble.
-
9:05 - 9:09This is a measure of global overvaluation
-
9:09 - 9:13of all markets, expressing what I call
-
9:14 - 9:17an illusion of a perpetual money machine
-
9:17 - 9:21that suddenly broke in 2007.
-
9:21 - 9:25The problem is that we see the same process,
-
9:25 - 9:28in particular through quantitative easing,
-
9:28 - 9:31of a thinking of a perpetual money machine nowadays
-
9:31 - 9:35to tackle the crisis since 2008 in the U.S., in Europe,
-
9:35 - 9:37in Japan.
-
9:37 - 9:40This has very important implications
-
9:40 - 9:44to understand the failure of quantitative easing
-
9:44 - 9:46as well as austerity measures
-
9:46 - 9:49as long as we don't attack the core,
-
9:49 - 9:54the structural cause of this perpetual money machine thinking.
-
9:54 - 9:56Now, these are big claims.
-
9:56 - 9:59Why would you believe me?
-
9:59 - 10:02Well, perhaps because, in the last 15 years
-
10:02 - 10:05we have come out of our ivory tower,
-
10:05 - 10:08and started to publish ex ante --
-
10:08 - 10:11and I stress the term ex ante, it means "in advance" —
-
10:11 - 10:14before the crash confirmed
-
10:14 - 10:16the existence of the bubble or the financial excesses.
-
10:16 - 10:20These are a few of the major bubbles
-
10:20 - 10:24that we have lived through in recent history.
-
10:24 - 10:27Again, many interesting stories for each of them.
-
10:27 - 10:30Let me tell you just one or two stories
-
10:30 - 10:32that deal with massive bubbles.
-
10:32 - 10:34We all know the Chinese miracle.
-
10:34 - 10:37This is the expression of the stock market
-
10:37 - 10:40of a massive bubble, a factor of three,
-
10:40 - 10:43300 percent in just a few years.
-
10:43 - 10:46In September 2007,
-
10:46 - 10:49I was invited as a keynote speaker of a macro hedge fund
-
10:49 - 10:52management conference,
-
10:52 - 10:56and I showed to the conference a prediction
-
10:56 - 11:00that by the end of 2007, this bubble
-
11:00 - 11:02would change regime.
-
11:02 - 11:05There might be a crash. Certainly not sustainable.
-
11:05 - 11:10Now, how do you believe the very smart,
-
11:10 - 11:16very motivated, very informed macro hedge fund managers
-
11:16 - 11:18reacted to this prediction?
-
11:18 - 11:20You know, they had made billions
-
11:20 - 11:23just surfing this bubble until now.
-
11:23 - 11:24They told me, "Didier,
-
11:24 - 11:27yeah, the market might be overvalued,
-
11:27 - 11:29but you forget something.
-
11:29 - 11:31There is the Beijing Olympic Games coming
-
11:31 - 11:34in August 2008, and it's very clear that
-
11:34 - 11:37the Chinese government is controlling the economy
-
11:37 - 11:38and doing what it takes
-
11:38 - 11:43to also avoid any wave and control the stock market."
-
11:43 - 11:45Three weeks after my presentation,
-
11:45 - 11:47the markets lost 20 percent
-
11:47 - 11:50and went through a phase of volatility,
-
11:50 - 11:52upheaval, and a total market loss of
-
11:52 - 11:5570 percent until the end of the year.
-
11:55 - 11:57So how can we be so collectively wrong
-
11:57 - 12:02by misreading or ignoring the science
-
12:02 - 12:04of the fact that when an instability has developed,
-
12:04 - 12:07and the system is ripe, any perturbation
-
12:07 - 12:11makes it essentially impossible to control?
-
12:11 - 12:15The Chinese market collapsed, but it rebounded.
-
12:15 - 12:20In 2009, we also identified that this new bubble,
-
12:20 - 12:23a smaller one, was unsustainable,
-
12:23 - 12:26so we published again a prediction, in advance,
-
12:26 - 12:31stating that by August 2009, the market will correct,
-
12:31 - 12:33will not continue on this track.
-
12:33 - 12:36Our critics, reading the prediction,
-
12:36 - 12:40said, "No, it's not possible.
-
12:40 - 12:42The Chinese government is there.
-
12:42 - 12:44They have learned their lesson. They will control.
-
12:44 - 12:46They want to benefit from the growth."
-
12:46 - 12:50Perhaps these critics have not learned their lesson previously.
-
12:50 - 12:54So the crisis did occur. The market corrected.
-
12:54 - 12:56The same critics then said, "Ah, yes,
-
12:56 - 12:58but you published your prediction.
-
12:58 - 13:00You influenced the market.
-
13:00 - 13:03It was not a prediction."
-
13:03 - 13:06Maybe I am very powerful then.
-
13:06 - 13:08Now, this is interesting.
-
13:08 - 13:11It shows that it's essentially impossible until now
-
13:11 - 13:13to develop a science of economics
-
13:13 - 13:16because we are sentient beings who anticipate
-
13:16 - 13:20and there is a problem of self-fulfilling prophesies.
-
13:20 - 13:24So we invented a new way of doing science.
-
13:24 - 13:27We created the Financial Bubble Experiment.
-
13:27 - 13:29The idea is the following. We monitor the markets.
-
13:29 - 13:34We identify excesses, bubbles.
-
13:34 - 13:37We do our work. We write a report
-
13:37 - 13:41in which we put our prediction of the critical time.
-
13:41 - 13:44We don't release the report. It's kept secret.
-
13:44 - 13:47But with modern encrypting techniques,
-
13:47 - 13:51we have a hash, we publish a public key,
-
13:51 - 13:55and six months later, we release the report,
-
13:55 - 13:57and there is authentication.
-
13:57 - 14:02And all this is done on an international archive
-
14:02 - 14:06so that we cannot be accused of just releasing the successes.
-
14:06 - 14:09Let me tease you with a very recent analysis.
-
14:09 - 14:1317th of May, 2013, just two weeks ago,
-
14:13 - 14:14we identified that the U.S. stock market
-
14:14 - 14:17was on an unsustainable path
-
14:17 - 14:21and we released this on our website on the 21st of May
-
14:21 - 14:23that there will be a change of regime.
-
14:23 - 14:27The next day, the market started to change regime, course.
-
14:27 - 14:29This is not a crash.
-
14:29 - 14:31This is just the third or fourth act
-
14:31 - 14:34of a massive bubble in the making.
-
14:34 - 14:37Scaling up the discussion at the size of the planet,
-
14:37 - 14:38we see the same thing.
-
14:38 - 14:41Wherever we look, it's observable:
-
14:41 - 14:45in the biosphere, in the atmosphere, in the ocean,
-
14:45 - 14:48showing these super-exponential trajectories
-
14:48 - 14:51characterizing an unsustainable path
-
14:51 - 14:54and announcing a phase transition.
-
14:54 - 14:55This diagram on the right
-
14:55 - 14:58shows a very beautiful compilation of studies
-
14:58 - 15:02suggesting indeed that there is a nonlinear -- possibility
-
15:02 - 15:06for a nonlinear transition just in the next few decades.
-
15:06 - 15:10So there are bubbles everywhere.
-
15:10 - 15:12From one side, this is exciting
-
15:12 - 15:15for me, as a professor who chases bubbles
-
15:15 - 15:19and slays dragons, as the media has sometimes called me.
-
15:19 - 15:22But can we really slay the dragons?
-
15:22 - 15:25Very recently, with collaborators,
-
15:25 - 15:27we studied a dynamical system
-
15:27 - 15:30where you see the dragon-king as these big loops
-
15:30 - 15:34and we were able to apply tiny perturbations at the right times
-
15:34 - 15:39that removed, when control is on, these dragons.
-
15:39 - 15:42"Gouverner, c'est prévoir."
-
15:42 - 15:48Governing is the art of planning and predicting.
-
15:48 - 15:51But is it not the case that this is probably
-
15:51 - 15:54one of the biggest gaps of mankind,
-
15:54 - 15:58which has the responsibility to steer
-
15:58 - 16:00our societies and our planet toward sustainability
-
16:00 - 16:05in the face of growing challenges and crises?
-
16:05 - 16:10But the dragon-king theory gives hope.
-
16:10 - 16:13We learn that most systems have pockets of predictability.
-
16:13 - 16:18It is possible to develop advance diagnostics of crises
-
16:18 - 16:22so that we can be prepared, we can take measures,
-
16:22 - 16:25we can take responsibility,
-
16:25 - 16:27and so that never again will
-
16:27 - 16:31extremes and crises like the Great Recession
-
16:31 - 16:36or the European crisis take us by surprise.
-
16:36 - 16:37Thank you.
-
16:37 - 16:43(Applause)
- Title:
- How we can predict the next financial crisis
- Speaker:
- Didier Sornette
- Description:
-
The 2007-2008 financial crisis, you might think, was an unpredictable one-time crash. But Didier Sornette and his Financial Crisis Observatory have plotted a set of early warning signs for unstable, growing systems, tracking the moment when any bubble is about to pop. (And he's seeing it happen again, right now.)
- Video Language:
- English
- Team:
- closed TED
- Project:
- TEDTalks
- Duration:
- 17:01
Yuriko Hida commented on English subtitles for How we can predict the next financial crisis | ||
Thu-Huong Ha edited English subtitles for How we can predict the next financial crisis | ||
Thu-Huong Ha approved English subtitles for How we can predict the next financial crisis | ||
Thu-Huong Ha edited English subtitles for How we can predict the next financial crisis | ||
Morton Bast accepted English subtitles for How we can predict the next financial crisis | ||
Morton Bast edited English subtitles for How we can predict the next financial crisis | ||
Morton Bast edited English subtitles for How we can predict the next financial crisis | ||
Joseph Geni added a translation |
Yuriko Hida
7:08 - 7:13
qualifying?