[tt] NS: Why economic theory is out of whack

[Premise Checker]

This is why laissez-faire is more important than ever: we simply do not 
understand what is going on, so it is best to let the market sweep out the 
bad judgments.

What is the case, not really emphasized here, is that the economy has 
emergent properties that follow well known probability *distributions*
(Yule, Pareto, normal, lognormal, Poisson) *without* any known probability 
*process* which result in these *distributions*.

Why economic theory is out of whack
http://www.newscientist.com/article.ns?id=mg19926651.700&print=true
19 July 2008
Mark Buchanan

See also: Editorial: Economic theory just isn't up to scratch [attached]

WHEN you next sit down to watch the TV news, listen out for a
telling phrase. At some point the newscaster will say something
like: "The financial markets reacted to the report with a sharp
fall..." Don't believe a word of it. The markets rarely react to
news in this way.

Earlier this year, physicist Jean-Philippe Bouchaud and colleagues
at Capital Fund Management in Paris studied the news feeds produced
by Dow Jones and Reuters that provide real-time reports of items of
potential interest to investors. Looking at more than 90,000 news
items relevant to hundreds of stocks over a two-year period, they
studied how "jumps" in stock prices - sudden, large movements - were
linked to news items.

They weren't. Most such jumps weren't directly associated with any
news at all, and most news items didn't cause any jumps. "Jumps seem
to occur for no identifiable reason," Bouchaud says
(www.arxiv.org/abs/0803.1769).

This finding flies in the face of traditional economic theory, which
insists that markets are mostly in equilibrium, reflecting an
overall balance of economic forces. Markets change, the theory says,
when those forces change: for example, when good news about a
company increases demand for its stock, making its price go up. In
this view, dramatic changes can only follow from correspondingly
dramatic causes. Bouchard's evidence says that, in fact, markets
have unruly internal dynamics all their own, with rallies and
crashes emerging seemingly from nowhere.

Evidence against the simple "equilibrium" view of economics is
piling up from other sources too. Take the recent worldwide credit
crisis. Its main cause, the most sophisticated computer models now
suggest, may be a fundamental tendency for markets to evolve, like
an uncooled nuclear reactor, towards a dangerously unstable state.
Everything from observations of irrationality in traders to the
statistics of market fluctuations is telling us something is wrong
with received wisdom, and a growing band of researchers has formed
the view that we desperately need to develop a new theory of
economics. "If we don't address the problems, there's absolutely no
doubt that other extreme crises will occur in future," says Didier
Sornette, an econophysicist at the Swiss Federal Institute of
Technology in Zurich.

So what might this new economics look like? The standard theory of
financial markets, shaped in large part by American economists
Milton Friedman and Eugene Fama in the 1950s, is founded on the idea
that the prices of stocks and other securities should tend towards
their proper values. There are two reasons for this. First,
investors have a strong incentive - the potential loss of their own
money - to work out how much an investment is really worth. As
rational people, they shouldn't be willing to pay too much for a
stock, or sell it for too little.

Second, the information gathered by millions of investors should in
effect be pooled by the buying and selling in the market, making the
market price an even better match to the true value of the stock
than any individual can arrive at alone. Any temporary mispricing,
the theory claims, should quickly get wiped out as some clever
investor jumps on it with an eye to an easy profit. In this way,
market forces should tend to iron out any problems long before they
get unduly large. An unexpected rise or plunge in values just cannot
happen unless there has been some correspondingly good or bad news.

This tells us straight away that something about the model is
flawed. We are currently experiencing what may be the worst
financial crisis since the 1930s. Wall Street firms have already
lost billions, and the US government has had to save at least one
from outright collapse. Some analysts forecast that losses could
ultimately exceed a trillion dollars.

The crisis was triggered by the bursting of a bubble in the US
mortgage market that had grown to grotesque proportions, thanks to
lax banking regulations and complex financial instruments that hid
risks in what appeared to be safe packages.

On top of that, there was the issue of "moral hazard". As economists
have been pointing out for some years, many common financial
incentives induce people to act for their own short-term benefit,
while saddling someone else - often their clients or the firm they
work for - with longer-term risks. In the case of the sub-prime
mortgage market, for instance, brokers were collecting commissions
on mortgages that required no deposit and no proof of income. Since
the brokers were not lending their own money, it was for them a
risk-free business. Meanwhile, investment banks took on these risky
loans and lumped them together into "collateralised debt
obligations" (CDOs). Once the risks were safely blurred, the banks
were able to sell the CDOs on at a healthy profit.

Alarming as this sounds, it should be fine if you really believe in
individuals' good sense and equilibrium economics. Investors will
simply factor the risks of the sub-prime mortgages into the value of
mortgage-backed securities, and adjust their expectations, putting
realistic prices on everything.

Unfortunately, equilibrium thinking has hit the wall. In 2007 a
global panic saw stock markets plunge. "A striking feature of the
crisis is that the situation appeared to be driven by emotion," says
physicist and former hedge-fund manager Doyne Farmer, now at the
Santa Fe Institute in New Mexico. "The word 'fear', which is not an
equilibrium concept, appeared in almost every newspaper article
covering these events."

The crisis also illustrates another shortcoming of equilibrium
thinking: a tendency to underestimate the likelihood of sudden large
events. Compared with the normal distribution of random events
represented by the bell curve, the statistics of financial
fluctuations have fat tails. In other words, large price
fluctuations are more likely than one might at first sight expect.

Failure to appreciate this has led to a number of big losses by
"quant" hedge funds, which use complex mathematical algorithms to
analyse the markets. While analysts insisted this was just bad luck,
they had in fact based their calculations on an incorrect
understanding of the statistics of the market, according to
economist Brad DeLong of the University of California, Berkeley.
"They said things like, 'Our strategy was fine, we were just hit by
a 16-standard-deviation event,'" he says. This reflects erroneous
equilibrium thinking that assumes the tail of the curve is slender.
"Tails are fat," says DeLong.

Perhaps we should have seen this coming. Some economists have long
argued that the movement of opinions and information between people
tends to amplify market movements, leading inevitably to fat tails.
Bouchaud and colleague Olivier Guedj found strong evidence for the
idea four years ago. Using data on analysts' forecasts of US,
European, UK and Japanese stock earnings over the period 1987 to
2004, they looked at how well their predictions had turned out. The
data showed, for starters, that they were generally over-optimistic
- so much so that a more successful strategy would have been simply
to assume that the following year's earnings would be the same as
the current year's. Tellingly, Bouchaud and Guedj found that the
analysts tended to make forecasts that were similar to those other
analysts had already announced, even when this went against
available information (www.arxiv.org/abs/cond-mat/0410079). They
flock like sheep in Prada shoes.

Virtual markets

A new generation of financial market simulations is starting to take
this flocking behaviour into account. The idea is to include more
detail on what makes people buy and sell, and how the opinions or
actions of one investor can influence others. "Traditional economic
models really don't even try to capture these dynamics," says Stefan
Thurner, head of the Complex Systems Research Group at the Medical
University of Vienna in Austria.

While this may seem like a considerable omission, it's not an easy
one to put right. Market dynamics can be bewilderingly complicated,
with thousands or even millions of participants - ranging from banks
and investment funds down to individual punters - all interacting
with one another. One of the most practical ways to get a handle on
how these elements interact is to build computer models populated by
artificially intelligent "agents" that buy and sell among
themselves, mimicking the activity of real markets. According to
economist Blake LeBaron of Brandeis University in Waltham,
Massachusetts, such models have already had some impressive success
at reproducing stock histories. "These models seem to fit real
markets - not only the fat tails, but trading volume and other
measures too," LeBaron says. "Traditional models just don't go very
far in reproducing any of this."

Meanwhile, Sornette has been investigating the effect of herding
behaviour on the rallies and crashes that seem to be inherent in
financial markets. His models have shown that news from the real
world does have an effect on markets - though not in the informed
and rational way one might expect from the classical equilibrium
view. "Paradoxically, it is investors trying to learn the relevance
of new information - often by watching others - that amplifies price
swings," Sornette says.

He developed the model with his student Georges Harras. Each day
their agents decide what to do using three types of information:
public news; what they hear from friends or others in their social
network; and any private information they may have themselves. Over
time, the agents gauge how effective each kind of information is in
helping them make good decisions, and they adjust their behaviour
accordingly.

Sornette and Harras found that, as in any real market, prices in
their artificial market never completely stabilise but continue to
move up and down more or less chaotically. The researchers could,
however, do something that is impossible in a real market: look at
how individual players' decisions were linked to those ups and
downs. This showed that the public and private information tends to
keep prices around realistic values, as the classical equilibrium
model says it should. The joker in the pack is information that
flows through social networks, and gets spread by word of mouth.
This, it turns out, creates groups of people coordinated in their
actions, which in turn leads to bubbles - stocks that become priced
too high or too low. Curiously, these bubbles can triggered by
nothing more than a random streak of news, which then becomes
amplified by social feedback.

The model is also providing insights into the origin of market
crashes, suggesting that here too received wisdom is wrong. Most
financial analysts look for the origins of a crash in specific
events immediately beforehand. Sornette and Harras's model, by
contrast, indicates that is has more to do with a progressive
linking together of investors' decisions and expectations over
months or years. This reinforces any problems - which in turn leads
to general market instability (see "Financial flocking").
Eventually, Sornette reckons, the markets reach a state like an
avalanche waiting to happen. "Anything can trigger the avalanche
once the system is ripe," he says.

This kind of instability may have a lot to do with the events that
triggered the current credit crisis. In recent unpublished work,
Thurner, Farmer and Yale University economist John Geanakoplos have
developed an agent model of the securities market that includes
hedge funds, banks and ordinary investors. The model's hedge funds
try to identify momentarily mispriced securities, and make a profit
by buying or selling in the expectation that the price will return
to a realistic value in the future. As in the real world, they
"leverage" their investments by borrowing from the banks.

The simulations have revealed some alarming consequences of this
kind of activity. With no leverage, a hedge fund can only lose its
own investors' money, but as leverage increases it can also lose
money it has borrowed from a bank, possibly putting that bank into
difficulties. "Lots of leverage begins to pose the threat of
failures cascading through the market," says Thurner.

Intriguingly, the risk of cascades like this occurring doesn't
increase gradually. Things go smoothly until the amount of leverage
reaches a certain threshold, at which point the model shows the
market undergoing a sudden change, loosely akin to a physical phase
transition, like water freezing into ice. Increasing levels of
credit create stronger links between market players, heightening the
chance that the failure of one can put an unsustainable burden on
others, triggering further failures. In the simulations, once the
level of leverage passes a certain threshold, it becomes
overwhelmingly likely that a single chance failure will send waves
of trouble through the entire market. Avoiding future crises will
mean identifying where the real-world market's "freezing point" is -
and keeping levels of leverage low enough to steer clear of it.

Geanakoplos cautions that this work remains speculative, but the
idea of increasing leverage bringing disaster corresponds well with
history, says Sornette, who has studied the dynamics of a number of
market crashes. "All bubbles I have studied have been associated
with increasing access to easy money, whether it's lower margin
requirements, lower interest rates, more foreign investments, and so
on," he says. Whether this idea can be put to work depends not only
on identifying the threshold for trouble, but also on the regulatory
authorities' willingness to try new approaches. They are certainly
needed, Sornette reckons. "The natural reaction to a crisis is to
update and upscale regulation and supervision," he says, "but this
has repeatedly failed to ensure even medium-term stability in the
past." Now could be the time for a move away from equilibrium
thinking.

See also: Editorial: Economic theory just isn't up to scratch

Related Articles

Financial markets driven wild by hormones
http://www.newscientist.com/article.ns?id=mg19826525.000
19 April 2008
Comment: Are economic bubbles so bad?
http://www.newscientist.com/article.ns?id=mg19826615.600
18 June 2008
Revealing order in the chaos
http://www.newscientist.com/article.ns?id=mg18524881.300
26 February 2005

Weblinks

Doyne Farmer's web page:
http://www.santafe.edu/~jdf/SFI%20Template/About%20Me.html
Econophysics Forum at the University of Fribourg
http://www.unifr.ch/econophysics/
Complex Systems Research Group, University of Vienna
http://www.complex-systems.meduniwien.ac.at/
+++++

Editorial: Economic theory just isn't up to scratch
http://www.newscientist.com/article.ns?id=mg19926653.400&print=true

IT IS hard to find a chink of light in the financial gloom looming
over the US. Last week, US federal regulators seized control of
IndyMac, one of the nation's largest savings banks, after a
crippling run on its deposits. Then the government announced plans
to inject billions into the nation's two largest mortgage-finance
companies amid panic over their massive debts and dwindling
reserves. Just a few years ago we were told that the unprecedented
sophistication of modern financial engineering had systematically
reduced market risks, making events of this kind all but impossible.
So what happened?

In part, it is a rerun of an old story - a tale of greed and lax
oversight fuelling a wave of speculative investment that was both
unrealistic and unsustainable. Think 1990s dot.com bubble or
17th-century "tulipmania". House prices, as anyone with any
historical knowledge could tell you, do not always go up, yet many
seem to have bought into the idea that they would ("Why markets go
bad"). Some got rich by perpetuating that notion and playing the
market with other people's money.

The crisis also seems to have been worsened by a woefully archaic
theoretical understanding of markets. In assessing market risks,
regulators still rely on what is called "equilibrium theory", whose
conceptual roots lie in 19th-century physics. This holds that market
behaviour reflects a balance between forces. Market values change,
the theory says, only in response to external influences - new
information about a company, for example, or a real change in the
housing supply. Beyond these forces, markets have no real internal
dynamics of their own.

Given how rumours drive markets up and down, and the way investors
flock like sheep and follow the words of various gurus, this is
clearly unrealistic. While some economists acknowledge as much, and
focus their study on "puzzling" deviations from equilibrium, only a
few researchers have actually tried to build new foundations and
move beyond the restrictions of equilibrium thinking. They have
pioneered computer models that mimic markets by simulating the
behaviour of individuals, banks, hedge funds and other players,
including regulators. These "agent-based" models do not assume
equilibrium from the outset, but instead let market behaviour emerge
naturally from the actions of interacting participants. As a result,
the modellers can experiment with raising interest rates, say, or
adjusting regulations to ease credit, and then see what happens.

One potentially pertinent insight from such models is that too much
easy credit can be a dangerous thing. As participants borrow ever
larger sums of money to amplify the potential profit from their
investments, their actions tie the fates of different participants
ever more closely together. This pushes the market toward greater
instability, making collective financial meltdown an almost
inevitable consequence.

Agent-based modellers admit their simulated markets need
improvement. The most vexing problem remains validation - ensuring
that the models give legitimate insight into real markets. Even so,
agents already models outperform traditional ones in predicting core
market features such as the statistics of daily or weekly price
fluctuations.

It is odd then that most economists seem uninterested, and only a
small minority have embraced the new approach. This hesitation
appears to stem in large part from cultural inertia. While most of
science is embracing the power of computer simulations to gain
insights into complex systems with many interacting components, most
economists remain dismissive of any work that is not based on strict
mathematical proof.

Although the present crisis was not caused by poor economic models,
those models have extended its reach by nurturing the complacent
view that markets are inherently stable. And while no one should
expect better models alone to prevent future crises, they may give
regulators better ways to assess market dynamics, detect early signs
of trouble and police markets. The cultural inertia of academic
economics should not block anything that helps shed light on the
markets that exert such a powerful influence over our lives. The
regulation of markets ought to be based on the very best science,
even if that does mean abandoning some of economists' most cherished
ideas.

Related Articles

Financial markets driven wild by hormones
http://www.newscientist.com/article.ns?id=mg19826525.000
19 April 2008
Comment: Are economic bubbles so bad?
http://www.newscientist.com/article.ns?id=mg19826615.600
18 June 2008
Revealing order in the chaos
http://www.newscientist.com/article.ns?id=mg18524881.300
26 February 2005
It's the economy, stupid
http://www.newscientist.com/article.ns?id=mg18224425.200
10 April 2004
Firm forecast
http://www.newscientist.com/article.ns?id=mg16221835.100
24 April 1999
Market forces
http://www.newscientist.com/article.ns?id=mg15821285.500
4 April 1998

Weblinks

Doyne Farmer's web page:
http://www.santafe.edu/~jdf/SFI%20Template/About%20Me.html
Econophysics Forum at the University of Fribourg
http://www.unifr.ch/econophysics/
Complex Systems Research Group, University of Vienna
http://www.complex-systems.meduniwien.ac.at/

E-mail me if you have problems getting the referenced articles.