Taleb vs economists

Everyone seems to be hailing Nassim Nicholas Taleb as the man who saw the crisis coming. I have a problem with this. It’s not that what Taleb says about risk is wrong. Quite the opposite. It just strikes me as trivially true.
We’ve known for ages that returns are non-Gaussian, that extreme events are more common than a normal distribution predicts, and that risk can’t be quantified simply, if at all. The Black Swan, then, was just an entertaining if a little egocentric way of telling us what we already knew.
So, when I read in it (p43) that bankers “are not conservative at all; just phenomenally skilled at self-deception by burying the possibility of a large, devastating, loss under the rug” I thought: “But surely they’ve learnt from statistics and experience by now. Their risk management can’t be as terrible as Taleb claims. I know bosses are stupid, but they can‘t be this gibberingly, imbecilically, carpet-chewingly, moronically cretinous, can they?” I suspect most economists thought my way.
It looks like we were wrong and Taleb right.
But this isn’t because Taleb had any great insights into the nature of risk. It‘s because he thought banks‘ risk managers were idiots, whilst economists didn’t think so - not even me.
In doing this, however, we were just following economists’ standard procedure - of assuming that agents were if not rational then at least not wholly stupid.
For me, all this is very troubling.  It suggests that what we economists have to learn from Taleb has nothing to do with the nature of risk - we‘ve all known that - but about others’ rationality. We should ditch the assumption - which in a sense is mere courtesy - not only that others are rational but even the weaker assumption that they are nearly so. Perhaps we should indeed regard them merely as “empty suits.”
But this is a vastly greater departure from standard practice than anything Taleb has suggested about the nature of risk.
All of which leaves me genuinely puzzled. Were banks risk managers really that bad? Are bosses an order of magnitude stupider than even I had thought? Or is something else happening? Help me.

Comments

[William]

Not stupid, just responding to incentives. If pay depends on taking risks, but not sensitive to the downside of those risks, then risks are taken. Under these circumstances, risk management becomes risk whitewashing.

[tom s.]

You can be silly and rational.

If everybody else is "burying the possibility of a large, devastating, loss under the rug" and making a lot of money by doing so, how long can you hold out from making the same silly assumptions? Each month you have a lower return than your competitors in another bank is another month you have to convince people that the lump under the rug is real.

Being rational means responding to the incentives in front of you, and those incentives don't always reflect long term reality. As you have been very effectively telling us for some time.

[tom s.]

I see william beat me to it.

[Giles]

Chris

Excellent as ever. I can't read Taleb because of his annoyingly egocentric style. Trying to read him and Niederhoffer together is almost more than the human frame can take.

I don't think it is about human irrationality, but the way people behave rationally within institutions. The short-term/long-term incentive problems you are perfectly well versed in. What is less frequently commented upon is the way power works within companies. Risk managers have no power - see the Economist's interview with one a few weeks back - and principal-agent issues abound throughout this area. What is rational for equity may be irrational for society; what is rational for the trader may be irrational for the shareholder, and the risk-manager may similarly have no incentive to play Cock Robbin when 90% of the time the sky is not falling.

Giles

[dirigible]

Incentives are a very good way of making people *act* stupidly.

[chris]

@Tom, William. Good points, but you've just relocated the stupidity.
If one bank consistently makes more than another, why don't shareholders think: it's just taking more tail risk? Surely, the less profitable bank is then attractive.
It's only if shareholders misjudge risks that the less profitable but safer bank has an incentive to act daft.
Aren't we just reviving the point that shareholders are useless at disciplining management?

[Matt M]

"It is better to be conventionally wrong than unconventionally right" is a truism for a reason.

Individual risk managers may well be smart enough to know that they shouldn't be treating risks as Gaussian random variables, but as a class they can easily allow themselves to doublethink their way around this objection.

When the bad consequences of an assumption only manifest themselves (by definition!) rarely then institutional laziness can take over. There is little to no incentive for an individual employee to tilt at windmills, nor their direct managers. Unfortunately, the incentives only appear at the ownership (or possibly upper management) level, while these individuals are ignorant of the specific shortcomings of the models their risk management teams are using. All this level of management sees is output. Profit per hour of manpower.

Ownership is divorced from knowledge. Bad consequences result. I have difficulty seeing how this can be easily rectified. Expertise in this area is relatively rare, dispersed among a few tens of thousands of individuals with strong math skills and specific knowledge of markets. Efficiency dictates that these individuals should control much more than their share of capital, but principal-agent issues make it difficult or impossible to align their incentives with those of the owners of capital.

[Nick Drew]

"he thought banks‘ risk managers were idiots ... ": for the most part, he's wrong about this

"... whilst economists didn’t think so"

but they were wrong too, and Giles' view is the right one: risk managers have no power.

[Preachy Preach]

I was reading the (alarmingly prescient in retrospect) 2006 book by Satyajit Das "Traders, Guns and Money" the other day, in which he makes it quite plain (with the benefits of 25 years experience in derivatives) that risk management was more of a box-ticking exercise than anything else. Also, he's very good (and clear) on the nuts and bolts of the derivatives business (including the infamous CDSes), without, unlike Taleb, coming across as an arrogant prick.

He has an anecdote about being head-hunted to be head of risk for a bank. He requested as his salary the the cost of a put option on the value of the shares held by the bank's board at the end of each financial year, on grounds that if he was going to be the fall guy should the bank go down...

[Matt M]

Chris, it is not only the shareholders which have difficulty controlling the actions of a firm's employees. Managers often/usually know less about what their employees are doing than their employees do. The lower you are on the food chain the more you know that what you are doing is not quite right, but the less incentive you have to speak up.

The larger a firm/partnership is the more long-term incentives to do things properly is divorced from the expertise that knows when things are not being done properly. Conversely the smaller a firm is the less leveraged expertise is. With current incentive structures there is therefore some sort of optimum size to financial firms. The better your incentive structures get the larger this size is, which adds to general welfare.

[anotherplanet]

I know it's a boring answer, but 'herd mentality'. Someone's successful with a new 'instrument', so everyone else piles in, lowering the returns and negating the advantage. At least when it all goes wrong there's safety in numbers - "he told me to, miss", or "they were doing it too mum". the alternative : to say, um... this doesn't add up, I think I'll forego these short-term gains and handsome bonuses and look at the big picture, takes a bit more nerve.

Plus: as the innovations get more and more complex, so more and more managers and investors have to rely on the judgement of the decreasing pool of rocket scientists that understand them - increasing the risk of a miscalculation.

As finance gets more and more specialised the disjunction between the professors who understand the equations and the traders who deal the products means that too much is lost in translation. We see the same process in the application of technology to major programmes like NHS computerisation. Those designing the software don't know enough about running hospitals; those managing healthcare don't understand the systems they're commissioning.

Perhaps it all comes down to the asymmetries in information that economic theory isn't too good at accommodating?

[bobvis]

Chris, I think your confusion comes from thinking of firms as rational versus people as rational.

In "The Black Swan", Taleb describes bankers as picking up pennies in front of a steamroller. He says this is stupid.

However, it is not stupid for them even as it might be stupid for their firms. As Taleb notes, bankers brought their firms to ruin in the 1982 Latin American crisis, the Savings & Loan crisis a few years later, and the 1987 market crash. These bankers lost millions. However, how many of them personally lost more millions than they had made previously?

In each of these prior cases, bankers did not necessarily lose *all* of their money. For most of us in non-banking. It will be difficult to earn 1 million or 2 in our lifetimes. They can earn that in a single business cycle. The government ordinarily insulates them from the steamroller, so in the long run they don't don't personally do so badly.

Notice how you see banking *firms* in the bailout line. You do not see *bankers* asking for a way to pay their rent. Bankers themselves are plenty rational.

Banking firms are not rational. They do not have brains to be rational or irrational. They are just organizational structures. That is why you see them repeatedly doing stupid things.

[reason]

Chris,
I think you are missing something here. What has happened (you know the bail out - privatise the profit and socialise the loss)? You are forgetting Keynes quip about (paraphrased) "A prudent banker is imprudent in the same way as everybody else."

[reason]

I also remember a quote from the US that no mortgage bank could afford to completely ignore the lower lending standards of their competitors. They wouldn't get any business. And this quote was a very traditionally conservative banker. The problem is that in a competitive insurance market (and that is essentially also what banks are), the least prudent set the price (I've pointed this problem out before).

[Mike Woodhouse]

The poor old gamekeepers always get it in the neck when the poachers have had a good day, don't they?

From having worked with a good number of them, I'd assert that risk managers do a fine job identifying and controlling exposure to the risks of which they are aware.

But how do you assess your exposure to an unidentified risk? That's where the Black Swan comes in. With hindsight, of course, the whole situation is easily explained and risk controllers vilified for the oversight. Rather unfair - they don't get publicly praised for all the risks they identified and controlled for which didn't cause havoc in the financial sector.

I do suspect that the presence of the risk function creates such a sense of security that when something does blow, the impact is much greater because unidentified risks are accumulated enthusiastically due their not being controlled.

So the next crisis could be even worse than this one.

[reason]

I'm also wondering if the models don't all contain autocorrelated ceterus paribus assumptions. (Each forgetting their own influence of the market).

[tolkein]

Part of the problem is that too many people bought the story about how risk is much better understood without realising that, even if it was true - and from my own experience in product design, I don't think it was true - firms will adopt , say lending practices, which give them the level of risk they had before. This meant that when tail events occurred, they would do so more frequently and when they occurred the consequences would be nastier. Telling bank management this (and showing that tail events were more common than believed) was a fruitless exercise. They just said they had better ways of measuring risk.
From their viewpoint this was sensible. If I was wrong and they listened to me they would miss out on lending opportunities profits and bonuses. If I was right everybody else would be in the same boat. As an individual, what is the rational course of action?

[rus bowden]

Let's say what we know, what we see, without attributing genius to anyone: the complaint.

A manager comes in and collects her bonuses based on the short term. Then, if she is smart, she gets out before everything piles up too high. Whether she gets out "in time" or not, the next manager comes in and straightens out "the mess" (always a mess) left by the previous. It's always the last one who is to blame. This new manager, through the standardized use of smoke, mirrors and short term figures, holds on to the position, firing those he can blame, setting up the newer (or older) and better ways, until his mess is too much for him to clean up, and he moves on. This can all be done, and may best be done willy nilly. The manager does not ever have to know that there might be someone better at the job than she is. She may actually believe that she has the best way of running things. She may even think that her way is less devastating, and in a pure sense, optimal for business. After all, no one else is doing anything better. And sometimes the proof is in the longevity of holding the short term position, a la, "I'm the only one who lasted in that position more than two years!"

Yours,
Rus

[ryanothebeach]

They are NOT that stupid, they are simply complicit. Complicit in ignoring the obvious for a share of the spoils. Criminal in my view.

[rus bowden]

It works best if they are stupid-- I should say ignorant. Not only does the emperor have no clothes, but no mirror either. This would be in the sense of ego dynamics. Bear in mind, the people who get the axe and are kept under their thumbs, are usually perceived by them as not as good as they are. Beliefs are often convenient opinions. People are prone to be prejudice against others, thus propping themselves up in their own minds.

Yours,
Rus

[Patrick]

Not everyone! "NNT", as he styles himself in "Black Swan", was stating the obvious; his book is a highly annoying read - because he actually has nothing new to say.

A deeply irritating individual.

[dearieme]

Unless something can be done about it, it doesn't much matter why bankers behave as if they are stupid.

[Luis Enrique]

How is the word stupid being used here? One definition might be something like "below average intelligence" but that's not much use, because we want to be able to say that very intelligent people may do stupid things. But what's a stupid thing to do? Something with bad consequences, judged in hindsight? That's no use, either.

I'm not trying to deny there is such a thing as a stupid thing to do, and that bankers can do stupid things, I just think the word is being flung around rather readily.

What are we looking for?

1. the agent could reasonably have been expected to foresee that what they were doing was going to end badly and
2. the agent didn't have some reason for doing it anyway such as greed, or facing a choice like "do it, or lose your job", or, were unable to do something about for some other reason.

In a hierarchical organization, point 2 can perhaps absolve workers of stupidity, if not greed, and perhaps even managers, but the buck stops with shareholders. Which is of course your point Chris.

But I still don't see why different ownership structures, like co-operatives, are going to be less institutionally stupid.

Also, it is possible to observe individually rational but collectively harmful behaviour in many contexts, like say in the behaviour of farmers in Sub-Saharan Africa. People are a bit less ready to fling around the word stupid in that context.

[Cabalamat]

As numerous people have pointed out, it's not stupidity it's responding to incentives.

I suggest a law forcing these people who've been paid huge bonuses to pay them back.

I was browsing through The Black Swan in a bookshop the other day, and have come to the same conclusions of other people -- it's a glossy rendition that doesn't have anything new to say.

[Matt M]

"I suggest a law forcing these people who've been paid huge bonuses to pay them back. "

Leaving aside the fact that you are suggesting an ex post facto law, why should we be rewarding stupidity on the part of the owners/upper managers? They entered into a valid contract with their employees. It is their problem if they were insufficiently clever to design the proper ownership and incentive structures to get the results they wanted. Clawing that money back will certainly do nothing to incentivise them to be smarter in the future.

[rus bowden]

Beyond the qualification that managers have been stupid in that they have followed their pay plans into short term gains, they also may have been, in the larger part, ignorant of ramifications, that there was an outside world being devastated through the mechanisms of their actions. This is important, because policy and pay plans lead to ignorant people being placed in charge of things, even qualifying best for these positions. If you're savvy and argumentative, you don't get the job, or stay clear of it altogether in the first place.

Surely, their were underhanded connivers involved making their smash and grabs. After all, this was the game laid out for them. But more important to the theme here, is that many manager didn't know any better.

Have you guys seen this, for instance, where arguments were being made in Congress that look crazy now in hindsight:

Shocking Video Unearthed Democrats in their own words Covering up the Fannie Mae, Freddie Mac Scam that caused our Economic Crisis:
http://www.youtube.com/watch?v=_MGT_cSi7Rs

By the way, I qualify the posting of this prejudiced video at the Bryan Appleyard blog post, in the first link above, and now here: Nassim on Newsnight:
http://www.bryanappleyard.com/blog/2008/10/nassim-on-newsnight.php#c5480892812911581582

Yours,
Rus

[passer by]

Maybe irrationality is amplified in dynamic environments?

[Dipper]

at the risk of becoming very hoarse, the FT had an excellent article in April on the sub-prime losses at UBS. The Risk department demonstrated that the bank was taking an enormous risk, but they were ignored.

[DRDR]

If the managers aren't stupid, merely responding to incentives, then doesn't that mean the principals who designed the organization structure are stupid? Also, it's not that economists blindly assume people are rational, it's that in the case of industry, we assume that evolutionary forces will drive the bad managers out of their positions or the principals out of business. So the real question, why is this not happening?

[Patrick]

The key problem is that the risks and rewards lie in different places. If traders and bankers can make millions riskily and then their firm goes under they are still millionaires. The solution would seem to lie in putting the risk of personal wipe-out into the mix.

Shareholders have failed utterly to align the full lifecycle risk / reward profile of their staff with their own needs as shareholders. Many shareholders probably aren't that upset anyway as it is, again, not their own money at risk (because most shareholders are paid staff in fund management institutions). Their bonus may be at risk but few fund managers will get sacked when the whole market tanks.

[John Emerson]

Maybe bankers believed in Econ 101 and equilibrium and so on. That's what econ teaches, regardless of what the top economists think.

[gaddeswarup]

John Kay in 2003 http://www.johnkay.com/in_action/268
"Evolution favours Taleb distributions. The gene pool of dreadful French drivers is depleted by road accidents - but only at the rate of 8,000 a year. Most young Frenchmen make it to their dates. Evolution within organisations has a similar bias. Someone who makes steady, small gains ranks as a safe pair of hands and is promoted until he meets his apotheosis."
Bloomber today http://www.bloomberg.com/apps/news?pid=20601087&sid=aDVgqxiT9RSg&refer=home
"Taleb's `Black Swan' Investors Post Gains as Markets Take Dive " (from a comment in Economist's view)

[Snickers]

Taleb talks about financial markets because that's what he knows from practical experience working in the trenches.

His critique, though expressed in terms of his personal experience, is really about the thin underpinnings of post-industrial societies.

"For me, all this is very troubling. It suggests that what we economists have to learn from Taleb has nothing to do with the nature of risk - we‘ve all known that - but about others’ rationality."

Will you ignore that feeling of being troubled, or explore it? Taleb is in the same boat as John Robb ("Brave New War"): the emperor has no clothes, now what?

As an exercise, take what NNT has been saying about risk management and apply that not to financial markets but how our food arrives at table. (See Pollan in this week's NYT Magazine for an example -- no direct NNT connection but clearly on the same track.)

It is not irrational to question others' rationality -- that's how money is made in the markets. Taleb is urging us to ask the basic market questions at meta levels. He seems wary of making recommendations but it's clear he thinks we need to emphasize redundancy, resiliancy, manageable scale and most importantly, humility.

[jonpo]

simple game theory - risk managers comply with the expectation of risk management practices, because the payoffs are greater that way.. until the payoffs aren't... then the risk managers may defect and profess never to have really beleived their own voodoo but just that they were copying everyone else tit for tat because it was more profitable for them. risk management practices may be irrational but they can stay irrational for longer than it takes to get a ferrari and a very comfortable living.

[prufrock]

No incentive is provided to make individuals think at "long term survival": just tell them they will receive the bulk of the bonus at the end of their working life. If the company is still alive either... nuts!

[Kenny]

"They are NOT that stupid, they are simply complicit. Complicit in ignoring the obvious for a share of the spoils. Criminal in my view."

[William]

chris: "Aren't we just reviving the point that shareholders are useless at disciplining management?"

why yes we are. when managers are allowed to make up their own incentives, those incentives are less likely to align with shareholder value. After all, it's easier to sell things (in this case mortgages & the like) than it is to make a profit from them. So incentives become aligned with sales rather than profits.

just like the internet boom: you lose money on each transaction, but you make up for it in volume. (that's a joke, only funny for those who didn't invest in internet stocks)

[Tony Ericson]

So much dazzling "economic" erudition in the preceding comments. Does it all help us understand what has really happened? Felt to me more like angels on a pin head arguments?
As an engineer, I learned the Laws of Thermodynamics. Amongst them were two relevant concepts -
- All types of Perpetual Motion Machines are impossible.
- All transactions involving Energy are "lossy".
Wealth is the Energy that powers our social existence. The same rules apply.
We have had 12 years of the leaders of our economy disguising "lossiness" - imagining they had a Perpetual Motion Machine of Wealth.
You asked for help. 37 years ago ago. in a number of classmates, we attempted to transcribe the Economic Theory we were being then taught via Thermodynamic Laws. We achieved an understanding that worked.
Maybe it is time for some of today's younger, fresher, brains to have a go? Perhaps new insights may be found? God knows they are needed.

[Bob B]

"I was browsing through The Black Swan in a bookshop the other day, and have come to the same conclusions of other people -- it's a glossy rendition that doesn't have anything new to say."

Any comments about this on Mandelbrot in 2004?

"In the sixties Mandelbrot already showed that extreme price swings are more likely than some of us think or incorporate in our models. . . "
http://lstat.kuleuven.be/research/reports/2004/report2004_08.pdf

But then as Hegel observed long ago:

"What experience and history teach is this – that people and governments never have learned anything from history, or acted on principles deduced from it." [Philosophy of History]

[passer by]

What is certain, if you post about Taleb at the moment you get a lot of hits :0)

The math is easy, the more certain or settled things become, (18 years of growth, low inflation ect) the more likely a Black Swan will burst the bubble, because the system over this time becomes more sensitive to errors, or in economic terms it gets more highly geared.

This is probably how the universe itself was born, and probably how Mankind came to dominate this planet (for now), both in genetics and Human Psychology, Mutations, flux, irrationality all pretty well add up to the same thing.