Mandelbrot Beats Economics in Fathoming Markets | Hacker News Reader

Mandelbrot Beats Economics in Fathoming Markets

bloomberg.com

69 pointsgoodwinb57 comments

It staggers me that so much prevailing thought still leans towards the idea of complex, highly connected systems as inherently stable, with their natural state as equilibrium. It dates back to the victorians, with their "All things Bright and Beautiful" view of god's creation, which was shattered by darwin. How long it takes us to learn the important lessons.

This view was prevalent in the world of ecology for decades, and led to all that guff about "nature's balance" and "spaceship earth" that we had to put up with for so long. It turned out that these ideas were based on falsified studies.

Adam Curtis addressed these ideas and more in his recent series of films, All Watched Over By Machines Of Loving Grace. Specifically, the second episode: "The Use and Abuse of Vegetational Concepts." The other episodes cover Ayn Rand and her effect on US economic policy, and the Selfish Gene theory and its murky origins in the wartorn Congo. And a whole lot of other stuff.

http://en.wikipedia.org/wiki/All_Watched_Over_by_Machines_of...

I highly recommend it, along with his other works, although you should be careful not to suspend your critical faculties in the headlights of his psychedelic filmmaking techniques.

I agree with your first 2 sentences but I think that Adam Curtis's 'films' should not be watched by anyone. Especially someone trying to think about something like stability and connected systems.

His train of thought narration and brilliant music is just opium for people who aren't concentrating.

I think people here are grown-up enough to cope, especially now that we've both warned them.

At the very least people should watch them simply for the treasure trove of historical accounts and strange characters contained within, even if they don't believe a word of the conclusions he draws from them.

I agree. I have watched almost all of his documentaries. In retrospect, everything looks like a con. He narrates history as if it was a big plan by the people in power but he forgets how utterly chaotic the world is. And yes, it is opium and I enjoyed watching his work.

You may enjoy this exquisitely observed parody, "The Loving Trap"

http://www.youtube.com/watch?v=x1bX3F7uTrg

Curtis basically falls for the animistic fallacy: that all events are caused by conscious agents. This is the same pathway that leads to both religion and conspiracy theorists.

I have always thought Curtis' overall theme (or one of them) was of powerful people drawing their grand plans, but always ultimately having them thwarted by external circumstances out of their control.

I personally felt that he

1. Identifies baddies and eeeevil plaaaans

2. Said plans go wrong because evil is dumb

3. Smugness.

Or something like that. Essentially his problem is mixing history with histrionics. He's not a documentarist, he's an entertainer. A music video director.

I think maybe a key mistake economists make regularly is equating an equilibrium with a lack of change.

In nature stability - on average - is often achieved by matching peaks and troughs, not by a lack of change (nature abhors a lack of change)

(see vacuum fluctuations)

>In nature stability

or at least in somewhat understandable models beyond simple linear ones:

http://mathworld.wolfram.com/StrangeAttractor.html

Negative feedback in markets often brings stability to a dynamical system and there is reason that there are negative feedback loops in markets.

A popular model of a negative feedback relationship in markets is called the 'Demand curve'.

Can this still be said when we're considering the economy of a whole country, or the whole world?

Both. Trade exists across invisible lines, even when the owners of the invisible lines forcibly try to prevent them.

There is also positive feedback in markets due to irrational exuberance and herd behaviour.

> It staggers me that so much prevailing thought still leans towards the idea of complex, highly connected systems as inherently stable, with their natural state as equilibrium.

I think it's more that

1. Systems that spin irreversibly out of equilibrium usually fall apart and so are not observed; and

2. For a long time, there was a dearth of mathematical tools for dealing with complex or chaotic systems.

Most attacks on economists about believing in this or that $obviously_ridiculous_belief are strawmen, giving credit neither for the subtleties of the actual concepts nor the self-awareness of economists themselves.

This is a good overview, though as one minor quibble, the critique of "rational agents" is separate from the critique of equilibrium models: even if all economic actors are rational agents, that doesn't imply that everything converges instantly to nice equilibria with an absence of feedback loops, attractors, and the other typical nonlinear-dynamical-system pathologies. In fact most agent simulations in AI that use rational agent models still find all sorts of that weirdness going on.

"Rational agents" in the context of economics does not mean people make the best choices for maximizing capital preservation. It means that human behavior is inherently rational because only an individual knows if his choices were in his best interest.

So it's basically meaningless?

The way it's typically used today is meaningless.

It came from economists asserting the difference between modeling human behavior vs. the behavior of physical objects. In that sense it is not meaningless. It's an important distinction that in order to model human behavior, we have to accept that people's goals are myriad and constantly shifting, and that we can only come to know their goals or preferences by them being revealed in action.

>"economists asserting the difference between modeling human behavior vs. the behavior of physical objects"

Is it really that different? Try to attach an object to several springs and see if it behaves "rationally".

It is very different.

Physical objects do not have free will. They don't act towards goals. They conform to unchanging mathematical patterns.

Humans do not. What I do today cannot be accurately used to determine what I will do tomorrow.

Ah, but that's assuming something very much in debate. Humans are actually quite predictable in many ways, though not as well as springs (currently). Aggregate behavior can often be predicted to reasonable accuracy, at least as well as with other complex, non-human systems like the weather. And even individual behavior can be predicted with enough information (and neuroscientists can even predict some specific decisions seconds in advance, given the right instrumentation). Just knowing some demographic and contextual information about a person hugely reduces your error on predicting what their "choices" will be.

It's a little different because the weatherman isn't forcing his predictions on you.

The debate takes on a different dimension when you're talking about political policy, when the weatherman passes a law forcing everyone to wear raincoats on days of high chances of precipitation.

I disagree completely. What about all inherent biases which make humans do things not in their best interest (they just think it is in their best interest, but many times they don't think, they panic) ?

See: - Loss aversion http://en.wikipedia.org/wiki/Loss_aversion - Sunk cost effects: http://en.wikipedia.org/wiki/Sunk_cost - Status quo bias: http://en.wikipedia.org/wiki/Status_quo_bias

As a trained economist I can attest to the fact that most micro economist view these issues you raised as basically irrelevant. I understand that it is basically damning experimental evidence but it is completely disregarded.

I'm not very up on microeconomics, but isn't behavioral economics a hot field currently? Or are they still seen as outside the mainstream?

The problem is those bias' can only be revealed through action.

No matter what you assert somebody's preference is, the preference is only revealed through action.

You can say me going to football games is not in my best interest, but the fact that I act by going to a football game demonstrates that it is.

This requires asserting as an axiom that people never make mistakes or get tricked, though, are always aware of all relevant information, and are in possession of an accurate mental model of how their actions are likely to affect themselves and the world. I mean, you can't really infer that someone's preference was to fall off a cliff by the mere fact that they did fall off it; they might not have realized that the cliff was there, among other possibilities.

> This requires asserting as an axiom that people never make mistakes or get tricked

Not having all the information doesn't negate that I'm the one making the choice. I think you're still referring to "rationality" in the logical sense of 2+2=4 being rational, not in the context of economics.

> I mean, you can't really infer that someone's preference was to fall off a cliff by the mere fact that they did fall off it; they might not have realized that the cliff was there, among other possibilities.

I didn't say all preference is revealed through a single action. I was trying to illustrate that action is the only way to determine preference.

For example, if I have a choice of going to McDonald's or Wendy's, you can't know which one I actually prefer unless I act and choose one over the other (thus preference is ordinal, not cardinal).

If I go to Wendy's, it would be absurd to say that I preferred McDonald's, since I chose to go to Wendy's. Thus, only through action is preference revealed. In the same way, it's absurd to say that someone else made a mistake when they bought an Apple product, because you're speaking of your preferences, since you can't actually know what's going on the head of the other person.

What "rational agent" means in the economic context is, that if Person A goes to Wendy's that says nothing about Person B's preference for Wendy's, since both have free will unlike inanimate objects.

Capitalism forces people to try to be rational.

To claim that humans are inherently rational is to deny the widespread existence of psychological pathologies.

Like the hyperrationality of sociopaths?

or depression

I'm surprised nobody has pointed out that the Euro is a tragedy of the commons.

"Pooling sovereign debt absolves the most irresponsible nations from confronting their unsustainable spending by forcing more responsible nations to pick up the tab. All of the incentives are weighted in favor of irresponsibility and none to responsibility. No pie-in-the-sky plan by the EU to dictate budgets to its members will ever work. The members will either ignore such interference or, as has already happened, cook the books to make it appear that they are doing so." – From Philipp Bagus' The Tragedy of the Euro

Unlike in the US, in europe the central bank can't lend to euro zone countries. Countries must borrow money from the banks and banks (like Goldman, Morgan Stanley, Deutsche Bank, etc) lend money to countries at highly speculative rates. If you had taken the time to inform yourself before opening your mouth you'd have known that countries like Portugal and Spain had relativley low deficits (much lower then the US) before the financial crisis. Please stop with this 'northern european countries superior monkey' narrative.

The ECB purchases member countries' debt. They move bad or weak assets off the balance sheets of the weak member banks into the balance sheets of the ECB. The stated purpose of the European Financial Stability Facility is to pool sovereign debt. The owners and shareholders of the european central bank are the central banks of member countries.

I didn't say anything about northern European countries being superior. It's a completely factual statement to say some countries are more solvent than others. It's not a value judgement.

The ECB buys souvereign debt incurred by countries that have no option but to borrow money from banks that demand very high interest rates - much higher than the interest rates the ECB charges comercial banks.

The purpose of the European Financial Stability Facility is to prevent german and french banks to go bankrupt in case any of the attacked countries default.

"It's a completely factual statement to say some countries are more solvent".

If by some countries you mean the northern countries, this is false: Spain had a superavit before the financial crisis - Germany did not.

Do not confuse opinions with facts.

For those confused readers, superavit is the Spanish and Portuguese word for surplus.

I don't know what your trip is with "northern countries." I never brought up specific countries or mentioned Spain or Portugal.

I was simply pointing out a fact that the ECB pools sovereign debt, and because of that will suffer from the tragedy of the commons - perverted incentives.

This article is a straw man.

I don't think most economists would seriously suggest that the market lives, should live, or ideally would live in a static equilibrium. I know that most economists wouldn't say they can predict the fluctuations of the market. The ones that think they can, of course, get disproportionate amounts of airtime on CNBC or wherever, so it's an understandable misapprehension.

I agree with you. I stopped reading at

"almost without exception, economists since Adam Smith have viewed economic systems as being in balance or equilibrium, and as having a natural tendency to return there after any disturbance. In this view, crises can be understood only as anomalies, the consequences of unusual outside shocks."

That quote shows a huge lack of contextual knowledge. A more accurate quote could be that since BEFORE Adam Smith Economists have disagreed on this point. (Menger/Walras, Rothbard/Schumpeter etc etc etc)

This article is pretty idiotic and yet another another attempt by new economists to try to excuse the terrible financial policies of the last 10 years under the silly and completely wrong slogan "nobody could have possibly predicted it would all turn out this way." This of course was the same excuse used for the housing crisis (even though most of the country actually predicted it) and for the Iraq debacle.

Take this quote "Nothing in mainstream “neoclassical” finance theory explains these persistent crises." That may be true but if so it merely reflects a weakness of neoclassical finance theory and not of our knowledge of economics as a whole.

In fact Keneyesian theory explains the boom and bust cycle as well as the periodic crises very well and has a pretty good solution for dealing with them. In fact we did deal with them pretty well during the postwar period of expansion and prosperity. Then of course we started progressively departing from Keneysian theory under the tutelage of the Chicago school and the crises, as if by clockwork, started intensifying and the crashes started getting worse.

So the answer is obvious, the theory is well known. The problem is that the new economists do not want to admit it because (i) they do not want to admit they are wrong and (ii) they sure as hell do not want the correct medicine that Keynesian theory prescribes.

So they go on with this ridiculous farce where they pretend that the economics crises are some unexplained phenomenon.

I'm always perplexed by the notion of trying to "model" an economy. I don't understand how aggregate statistics about a market reveal any insight into how to create sustainable value. I feel like it's similar to knowing the past winning numbers on a Roulette table, because it provides no actionable insight to future winning numbers.

On the contrary. Taking your analogy further:

In economy, you find out that whenever "Gentleman Jim" bets, he tends to win 70% of the time, rather than the 49% everyone else gets. Now, depending on policy, you either:

a) forbid jim to play b) readjust jim's token-to-money conversion ratio so he is on par with other players c) invest your money with jim

Note, though, that this being a zero sum game, anything other than (a) will bankrupt the house....

In what sense is this a zero sum game?

Sure if economics could be reduced to gambling it is zero-sum but betting in economics is betting on those who will create value(ideally).

Nassim Taleb addresses this sort of thinking. He calls it a "ludic fallacy", which is the over application of games to our down detriment. http://en.wikipedia.org/wiki/Ludic_fallacy

There are many models considered common knowledge in economics that possess strong explanatory and predictive power.

Though in recent times people tend to pick and choose which models they base their thinking on, depending upon their chosen political agenda.

Brief example: There has been a lot of political hand waving about possible inflation or even hyperinflation. If you look at a version of the Phillips Curve, unemployment and inflation have an inverse relationship. High unemployment = low inflation or worse. And inflation has remained low, while deflation has actually been more of a threat, and is actually a problem in many countries.

> Phillips Curve

Wikipedia seems to think that the 1970s in the US show that it doesn't work.

Yes, the Philips Curve is pretty much the poster child example of the Lucas Critique. That is, observed invariants often cease to be invariant when policy changes. When we well and truly went off the gold standard the Philips Curve ceased to have predictive power for the US economy.

More details: http://www.themoneyillusion.com/?p=9677

What you are describing is a Random Walk, 50/50 chance of going up or down, the previous data point holds no significance for the future. Mandelbrot discusses how this viewpoint is wrong in "Misbehavior of Markets" and describes market patterns using two movements: short trends and even shorter bursts. His multifactoral model uses these two ideas to model a market, however prediction is always difficult.

For a little background on this debate: http://tuvalu.santafe.edu/~jdf/papers/farmer0606.pdf

Wait, did economists think at one point that the distribution of movements in a stock market followed a Gaussian pattern? I mean, amateur statisticians approximate things to Gaussian distributions all the time to make the math easier, but that's hardly a problem unique to economics.

As to the Omori distribution, have they actually succeeded in making forward looking predictions with it, or were they just fitting a model to past data? Even if it is a real phenomenon, I can think of a way to extract money from the market by making it go away off the top of my head, so I don't imagine it will last long now that it's been reported in public.

The parts about equilibrium models being taken too seriously are well taken, though.

When they say "fat end of the tail," a Gaussian distribution is not necessarily implied. They are simply saying that when you go to the extreme ends of any distribution, almost any statistical model will begin to fall apart. If you look normalized data and look at a Q-Q plot, you will no doubt see problems at the extremes of the distribution, making those predictions more difficult.

As for the Omori distribution, I do not know anything about it, but I have studies similar distributions for predicting future Olympic running records,tallest human alive, etc, and these types of distributions rarely produce practically feasible results. If this model works well, I will be thoroughly impressed.

The only field that I know which deals with extreme events is Ruin Theory, but it is currently a very limited field. It may be possible for someone to adapt the field to study Macroeconomics, but even that may not be very informative.

At humanities current understanding of economics, I would argue that Black Swan Theory is the only practical way to understand huge economic shifts. Perhaps we will understand economic markets well enough to develop more complicated models, but that seems far in the future.

Q-Q plot: http://en.wikipedia.org/wiki/Q-Q_plot Ruin Theory: http://www.worldscibooks.com/etextbook/5943/5943_chap01.pdf

Black swan theory is an interesting read.

Yes, surprisingly, a large portion of statistical research in finance and economics assumes normal distributions for model formulation, prediction, and error checking.

It is a fast oversimplification to say that economists lose and Mandelbrot wins. Eugene Fama (http://en.wikipedia.org/wiki/Eugene_Fama) whose Phd thesis coined the concept of stock prices following a random walk simultaneously wrote about Mandelbrot distributions (http://www.jstor.org/pss/2350971).

"The (Mis)behavior of Markets" is a great read on the subject, by Mr. Mandelbrot himself.

"God doesn't play dice". And several decades later Mandelbrot discovered that God plays fractals.

Reply on news.ycombinator.com