Collapsing under its own weight
Andrew Lo, director of MIT’s Laboratory for Financial Engineering describes how one rogue trader can impact global markets. Why? Because the software and systems on which it runs have become increasingly complex. And that means that a small change in the initial conditions can have unexpected consequences.
Interviewer: "Okay, so bad things will happen. I take it you are mainly concerned about the ripple effect when they do?"
Andrew Lo: "Exactly. The financial system as a whole is getting more complex. Financial institutions rely on ever more elaborate systems architecture and electronic communications across different counterparties and sectors. The number of parties involved, the nature of transactions, the volume of transactions as the market grows--taken together, the dynamics among these aspects of financial markets imply that the complexity is growing exponentially. No single human can comprehend that complexity. And as the system grows more complex, it is a well-known phenomenon that the probability of some kind of shock spreading through the system increases as well. Systemic shocks become more likely. Today, we are looking at some significant exposure to relatively rare events."
Andrew Lo was particularly disturbed by suggestions that a rogue trader, Jerome Kerviel at Societe Generale could hand-change records in their database system. "From news reports, it appears he was able to access internal financial databases and not only alter the stated holdings of the accounts he was trading, but was also able to circumvent the checks and reconciliation processes that were put into place to make sure these were accurate." The fraudulent trades Kerviel conducted may have actually triggered a worldwide "unwinding" of portfolios. "It's a larger-scale version of what happened in August of 2007--in particular, August 7, 8, and 9. A large number of quantitative equity hedge funds lost money on those dates simultaneously, yet there is no market event that you can point to that can explain why these funds lost money at the same time."
So what does Andrew Lo propose? A Capital Markets Safety Board, which like the NTSB, should look into every crash, not of the airframe variety, but of the financial kind, to make sure there isn't some systemic defect lurking under the code.
But will adding safety modules to the financial system help? Commenter Henkeeper in a previous thread points out that when complex systems are made even more complicated by additional safety systems sometimes their propensity to crash actually increases, citing Charles Perrow's book Normal Accidents.Perrow argued that "improvements may introduce new opportunities for disaster. Looking at an array of real and potential technological mishaps--including the Bhopal chemical-plant accident of 1984, the Challenger explosion of 1986, and the possible disruptions of Y2K and genetic engineering--Perrow concludes that as our technologies become more complex, the odds of tragic results increase. ... the Chernobyl nuclear accident, to name one recent disaster, was partially brought about by the failure of a safety system that was being brought on line."
posted by wretchardthecat at 1/31/2008 06:29:00 PM
21 Comments:
Wretchard quoted:
"The number of parties involved, the nature of transactions, the volume of transactions as the market grows--taken together, the dynamics among these aspects of financial markets imply that the complexity is growing exponentially. No single human can comprehend that complexity."
Early in my career, I had the problem of writing a big computer program that became so large and complex that I could no longer maintain it, even though I wrote every line of the program (the code had become crufty).
The lesson learned was it's not enough to write a computer program that simply works. The program must work, be well documented with appropriate comments (not too many) and have a logical style that is consistent with the process being modeled.
I now only program in the C (I had to unlearn Fortran because it was turning me into a bad programmer). I discovered years ago that there are excellent reasons why almost all of the world's largest programs are written in C, e.g. the Linux kernel, X-windows, etc. The guys at Bell Labs who wrote C were computer geniuses and their brilliance is reflected in C.
Nature has a self consistency that's quite beautiful and elegant. If one can capture nature's self consistency in the program then a marvelous transformation happens. One's relationship with the program ceases to be that of an innovator and becomes that of a participant. The program developes an intrinsic logic of its own that the programmer simply follows. Programming errors become less frequent because they stick out like sore thumbs. The program itself will tell the programmer how it wants to be innovated.
I know this sounds like hocus-pocus but I've talked to other people who have experienced this (geek nirvana).
The complexity is not the main problem.
The main problem is that there is nothing backing these positions.
No one knows what they are worth.
When everyone realizes this there are no buyers, the "value" crashes.
You have not seen anything yet.
Credit bubbles like this have inflated and crashed many times before.
The fraudulent trades Kerviel conducted may have actually triggered a worldwide "unwinding" of portfolios.
He's describing a retcon. Anybody who's familiar with the effects of balefire (a weapon in The Wheel of Time fantasy series) will immediately understand the eerie similarities of this particular "unwinding."
Seems to me the broader entanglement problem can only be solved by developing extra-market corrections. These would be ready-to-go responses, conscientiously composed and maintained to avoid and mitigate specific types of shock and error. An analogy that comes to mind is Flu transmission through the human-network. Our strategies to deal with this consist broadly of diagnosis, quarantine, treatment, and prevention.
Easier said than done, but since we cannot rely on the market to minimize system shock itself -- massive shock probabilities are built into the system due to entanglement -- we must have external devices ready to respond, treat, and adapt to any large errors which appear (and if we can become sophisticated, detect those subterranean trends which tend to cause avalanche).
But regardless, corrections will come, and the system will evolve. We just have to get better at crisis management.
And not to dispute the good doctor, but a small change in the initial conditions leading to drastic deviations in outcome -- aka the butterfly effect -- is not the whole story.
In complex systems, chaotic unpredictability on one level can look surprisingly ordered on the next. Oftentimes a complex system will "lock in" to certain patterns, certain phase "attractors" around which the "function" of the system will revolve. You could also have "nested" feedback mechanisms which keep the system returning to its equilibria even though the inputs may change in kind and degree. Of course, you can also have the opposite -- nested deviation amplifiers -- that autocatalyze the system to criticality. Both are well-documented attributes of natural systems.
The problem is we don't have a good way to predict these phenomena, because we don't have a way to accurately surmise all the parameter values in an nth-dimensional complex system like the global market.
So, in short, all I'm trying to say is that a tendency toward chaos is but a part of it. Large-scale structure and order can also emerge spontaneously for no apparent reason.
One “rouge trader” is not responsible for this credit crash.
Nor is it complexity that is the problem (it is interesting that the suggestions made here are to add more complexity or safeguards to solve that problem).
Failure is built into a fiat based pyramid scheme.
If there are no restrains (e.g., gold) placed on the creation of “money” (which today is debt) the money supply will collapse when no more debt can be serviced or created.
At the base of all of the complex derivatives (CDO’s MBS, swaps) is a simple flaw pointed out by the Mogambo Guru
“I buy a stock for $1, and then sell it for $2 (making a 100% profit).
And then that investor who bought it from me can make a 100% profit of $2 (doubling his money, too) by selling the stock to someone else for $4. And how does that guy who paid $4 for the stock make a similar profit? By selling the stock to someone else for $8!
I know what you are thinking. You figure, "Hey! This looks easy!"
So, now it is time to add up, so let's add up, which I am usually pretty good at, if the numbers are few, are all single-digit, I can write it all down and can get back to you sometime early next week with the answer. Fortunately, this is such a case.
So, how much profit was made from all of this selling? $7 (=$1 + $2 + $4)! How much money was spent? $15 bucks (=$1 + $2 + $4 + $8)! Hahaha!
Everybody so far has made a 100% profit, and yet twice as much was spent as was made in profits? A lousy 50% total ROI? Hahaha!
And the last guy, who may be otherwise known as the Last Fool In Line who bought at the exact top at the exact highest price, still has to find somebody to sell to at a profit, or the system goes into loss mode.
If he sells at $7, taking a $1 loss, then total profits for the system are reduced to $6 (=$1 + $2 + $4 - $1)! And yet the total amount spent in all this investing has climbed to $22 (=$1 + $2 + $4 + $8 + $7)!
Now ROI is 6/22= 27%!”
This system fails because it has the simple flaw that it is based on the idea that you can create wealth by printing money.
We used to worry about panic spreading among people. Now it spreads among "systems" at the speed of light -- before humans even know it's happening. So we need to calm the system automatically, have it self administer cyber Prozac at the speed of light. Then the entire world financial system can both panic and recover before humans know it happened.
Is it such a good idea to tell them?
What if the "Street" or the "Market" is cut off from a major source of funding such as oil petro-dollars as in the current case of the Middle East being cut off from the Internet?
Or does everthing stay in stasis for the weeks or months or quarters tht it takes to get the Arabs and their money back on board? The human being remains in Saudi Arabia, his money remains in a bank in Switzerland, and the major banking company he has just bought in America spends Monopoly money.
It's true that wealth is not created by printing money, but printing money (unfettered by artificial constraints) is often necessary and good.
The Persian Empire had enormous hoardes of gold which they thought equated to power, but they kept it in vaults and didn't spend it, so it did them no good and Alexander walked all over them.
Capitalism works by, essentially, time-travelling money from the future into the present. That's what "capitalizing" an asset means. If you've got growth, that future supply of wealth upon which to draw into the present is mathematically formally infinite. This is what the malthusians and marxists don't understand.
Printing money is only a problem if more is printed than the system requires to grease the wheels of useful capital investment, and to feed the pending transactions at the maximum velocity the money wishes to flow. Then it's inflationary, either in general prices, or certain asset bubbles.
So for example, simply handing cash to people who didn't produce anything is more likely to be inflationary than providing liquidity to those who are going to build assets that will have a future value; that future value was just time-travelled into the present in the form of cash to get it built!
And we all prosper as the pie grows bigger.
In other words, I favor tax cuts to "the rich" who paid taxes rather than handouts to those who already didn't pay any tax -- prices will just rise to absorb the latter and nobody gets ahead!
PS the Eurozone is much more likely to be in real recession later this year than the US (which will not enter a recession)...
Most people would say this is a gross over-charge on a huge credit card...
Or a crass fraud of over-borrowing for speculation.
From an accountant's point of view the EU accounting rules are weak compaired to the US (the Federal Reserve).
This highlights how a relative obscure trader by the name of Jerome Kerviel can destroy $7.2 billion without the watchful French Authorities knowing... Wink... Wink... Or, how massive Car-B-Qs go unpunished.
It also shows how a high-level banker, Mr. Noyer, of Bank of France, who knowingly hid losses can keep his job in EU (banks that did business with the French bank took the hit - and I hope it was not American banks).
In the US the same banker would have his head handed to him on a silver platter.
Wretchard knows that computer networks have a scale of Administrative privileges to low level user privileges. These are fire walls of sorts. But these administrative privileges can be, at the will of certain people, be delegated to lower individuals.
This can have a catastrophic affect on the whole network system should said privileges be misused. I speculate that this is what happened at Societe Generale (There must be computer logs of the transactions).
French trader Jerome Kerviel was given the keys to the treasury (or Administrative Privileges to the treasury) which caused grievous damage to said treasury.
People in the French banking system knew of the misuse but turned a blind eye to it for various reasons.
Here is my microscopic experience with rouge traders.
Very early in my accounting career a friend of my traded in the newly conceived stock options market were trades could be made in 100, 1000, 10,000 and so on, in derivative equivalent stock market.
My friend traded with a well known broker whose name will not be printed but shall be call ML.
My friend found what he thought was a “once in a life time” trade – but was a little short on capital. He desperately wanted to make the trade.
He knew he was a little short on capital but he called the ML “back office” in Cin... Ohio.
He then posed a hypothetical question to the back office girl.
If I make XXX trade and close it in the same day will I violate the capital requirement of ML?” The girl did not really know but said that it could be correct.
With that conversation my friend called his ML broker. The head broker was out of the office so the assistant broker took the call.
The assistant noted that my friend’s capital account was short of the amount necessary to make the trade. My friend then recounted the conversation with the “back office” girl and said it was “OK.”
The assistant broker “forced” the trade into the system. My friend had bet correctly and closed the trade the same day making a nice profit.
Next, things took a turn for the worst.
The head broker called and demanded to known why my friend was able to place the trade.
My friend said “the back office had indicated it was OK.” At that point, the ML broker said “Thanks, that’s enough to give my boss.”
A week later, my friend brought me the envelope containing the Federal Reserve Regulation “T” violation notice (it may have actually been a Reg. F violation but I don’t remember, it was worded as "trading while under capitalized").
I called around to a few lawyers. None really knew what to make of it but most said it was just a “traffic ticket.” I relayed that information to my friend. To my amazement my friend kept trading.
If this can happened in the US with its stringent Federal Reserve and watchful SEC, I can only guess the flimsy enforcement system of the new formed EU with its multi-lingual speaking regulators conversing through translators in a fast moving financial environment. It must be a mess.
I would guess it is like the 'Pirates of the Caribbean' robbing the various member of the EU. Johny D. gets the loot and looser gets the boot.
I would think that the EU has huge capital problems that it doesn’t care to disclose.
I also guess that their currency is overvalued. Their system of capital management is, at best, no better than the USA, and probably much worse!
If I were on the Board of the Federal Reserve, I would require a through review of all of the capital positions of the EU’s banks.
If necessary I would adjust the capital worthiness of the EU for said disaster, and require higher inter-bank interest rates to be paid to USA banks.
But, I am not. So, pay your money take your chances.
I think there is also a simpler mechanism underlying stock market troubles, at least in this country:
the tax system discourages straightforward "investing" for income (dividends/profits), and instead tilts the balance toward capital gains being the end-all, thus encouraging speculation, which the Founders were always concerned about.
Ideally you always wanted *both*, but today most of the emphasis is on the latter. The problem with investing/speculating for capital gains is that, like the Persians horded gold, its only realizable when you sell-- to buy again. Or, often worse, borrow against it.
Electronic systems, in reducing the friction on transactions (otherwise a good thing) inherently increase volatility, both in panics, bubbles, and the neverending though usually vain attempts at arbitrage.
One thing they teach in basic finance is how hard it is to make an "economic profit" (not profit in the usual accounting sense) that substantially beats the crowd.
One result of all this is the way public companies are always "grow, grow, grow", focusing on the short term stock price gyrations (which for the most part *don't* directly impact the company) instead of focusing on sustainable longer term profit.
The other thing I got out of the business courses I took for my Info Systems M.S., which I never heard explicitly stated was, was the disconnects between the way sthe various disciplines look at things (accounting, finance, economics). What do I mean?
The finance guys can talk all they want ROI, IRR etc and what a stock *should* be worth, and the accountants can calculate shareholder equity, but it is all trumped by economics -- supply and demand, IOW, at the end of the day the stock is only worth what people will pay for it -- which introduces a huge element of psychology the other domains ignore.
I had a prof make the same point about SDI, and this was my introduction to many of the concepts on discussion here. Essentially the systems would be so complex they could very easily spin out of control kinda like how program trading can quickly spur or deflate a market.
When I worked at a local international manufacturer and distributor maintaining one of their systems we were able to directly update their production IMS database (they had their prod DB2 more secure, but getting update access was not difficult).
In fact, it was a very common procedure. We had a Y2K bug lurking and messing things up every now and then. We'ld get a call from the clerk out on the floor reporting the tell-tale symptoms and we would pull up File-Aid and correct the offending record. Eventually we tracked down the Y2K bug and fixed it.
It is just another example of how adding layers upon layers upon layers of regulation and checks, and blocks, are circumventable by the dishonest and more often than not, make it hard for the honest to accomplish what they need to do.
All of this brings back memories of Nick Leeson bringing down Baring's Bank.
Funny how owning a piece of a company has become such a complicated matter. Long straddles, short straddles, collars, derivatives, puts, calls, long, short... With all of that complication I guess people can be forgiven forgetting what the stock is. It is a solid thing it is piece of ownership and control of a corporation.
It appears that “no one knew” that the European banks had invested heavily in the idiots in the U.S. who were buying homes under the assumption that the property would go up and up in price, enabling them to bail out and make a bundle before the real bill came due - or else refinance and escape reality that way.
And “no one knew” that the rogue trader in Europe was creating fake money by means of false accounts.
And when these two things that “no one knew” came together – blewie!
5 years ago today we lost the Shuttle Columbia. I will never forget waiting for those sonic booms that never came. And as one of those at NASA responsible put it “It was nobody’s fault.” But in the end, it was because NASA had rejected the very concept of an independent group that asked “What happens if…” And just as bad, they walked around muttering “Can’t happen…can’t happen…can’t happen” And indeed, what "could not happen" did not happen. Foam could not damage the Shuttle tiles - but a big chunk of silicone rubber could do massive damage, and it did.
In the words of Robert A. Heinlein, “It’s not the stuff you don’t know that gets you, it’s the stuff that you do know that is wrong.”
And often it is your ability to think about what is wrong that is the real problem. It's not so much the complexity of the systems that is the problem but the complexity of human systems designed to keep from thinking about the problems.
Newscaper,
Certain trading strategies such as futures are still tied to legitimate business operating and financing but after futures things become a lot less real.
I also agree with you much motivation for such speculation is based on tax law and regulation - that is devising new ways to get around such regulations.
Capital markets require both moral and communications clarity. I have found that a great many market participants cannot tell the difference between hedging and speculation. The higher you go in corporate hierarchy the less clarity you find. The financial journalists routinely add to the confusion. Most so-called hedge funds are really organized speculation funds. If market participants don't understand the difference between a hedge and a "spec" you increase the likelihood of one of Taleb's Black Swans swimming in.
The recent idea that you could get a higher return while reducing risk by packaging a bundle of mortgages (securitizing) can be justified mathematically, but is a violation of common sense risk management. It assumes that each of the individual mortgages had at least a minimal credit score. Now we find (surprise, surprise!) that such was not the case.
Every commenter in the thread has brought real world experience to the discussion. Long live the Belmont Club!
Let's back up. While there may be a story here about an individual event causing world wide outcomes, there's little evidence that's what happened here.
The original story in Bloomberg and WSJ quotes only the head of Soc Gen saying they lost 7 billion due to a "rogue trader" who compromised their systems (but didn't try to steal anything.) They quote no other sources in the bank. They quote no rival bank sources. They don't have anyone else to corroborate this story.
They say that the trader bypassed their elaborate uditing and risk management procedures. The articles don't ask any IT professionals from the finance world what that would take, or how difficult that is. They don't discuss the legal reporting requirements in the US vs. in France to discuss why that would, maybe, hypothetically, be easier there than here.
And then they state that Soc Gen's sell off may have triggered unwinding. But they quote no one on Wall Street who saw 7 billion dollars of exotic equities futures hit the market.
Why did no one see that? The obvious answer is because it didn't happen. There was no unwinding of Soc Gen's positions.
The whole story is unlikely. Mr. Lo says he's disturbed such a thing as the bypassing of the risk systems and auditing systems could happen. Perhaps a better word is SKEPTICAL.
The most likely actual story is that the traders at Soc Gen were all systematically bypassing the systems, or the systems werne't in place, or the systems had been ignored by management for years. Soc Gen now admits they were leveraged more than the worth of the bank--and NO ONE noticed. uh huh.
This is not a rogue trader story. This is a story about the danger institutions have when they don't understand the risk and the mechanisms out there for distributing that risk. This is a story about how a large system's suddenly move away from equilibrium can sink individual elements rather than the other way around.
Fundamentally, the problem is one of INFORMATION. The banks don't have it the way they'd like .They simply don't know their holdings, because it's nigh impossible to do so. In the US, legal restrictions on the cient-bank relationship, realtionships with other clients, and auditing/insider trading laws keep them from knowing their positions. But the complexity of the system does to--what is their exposure, TODAY, NOW? that's not an easy question to answer because how much an instrument is worth is not a fixed quantity. It's fundamentally a "it's worth what you THINK it's worth" answer--it's worth what you'll pay me for it.
and Soc Gen apparently didn't know how to price its exotic instruments afterall. And when people found out, Soc Gen lost.
Banks' real commodity is trust, not money. Their liquidity depends on that trust. And when the system moves away from equilibrium, that becomes more clear.
Most software is crap written by people who don't understand how it should work led by people who don't program with the whole project run by people who don't know the people that work for them.
So, good luck!!!
My guess is that the trading SW Kerviel used sucked and that he started doing a work-around like a lot of others and he got caught.
Eggplant,
I've had the same experience with complex systems, but don't forget that the sendmail daemon that Robert Morris subverted to crash the internet several years ago was written in C.
Now imagine very complex systems that are stacked: SQL databases layered under C programs controlled by TCL scripts controlled by shell scripts run by web interfaces.
Or more to the point SQL databases accessible by Windows applications.
It takes a *lot* of effort to audit these systems. It takes enormous effort to develop adequate access control systems. A lot of time the access control systems don't easily map as you ascend and descend the "stack" of controlling programs.
You may have database access controls that map to a group of users but then find that inappropriate people were added to the group.
In other words the distribution of responsibility that eases accessibility and managability also increase exposure to risk.
Our whole society is like this!
This is the challenge in large complex systems. This is why complex systems fail. Unfortunately, the controls put in place like Sarbanes-Oxley are designed to create a lot of paperwork, and only enforce the spirit of compliance, not actual rigorous testing.
Thankfully this model gets feedback. When people observe Societe Generale failing they will respond by examining their own system for vulnerabilities, and this is frankly more effective than anything ever proposed by a Congressman.
Andrew Lo was particularly disturbed by suggestions that a rogue trader, Jerome Kerviel at Societe Generale could hand-change records in their database system. "From news reports, it appears he was able to access internal financial databases and not only alter the stated holdings of the accounts he was trading, but was also able to circumvent the checks and reconciliation processes that were put into place to make sure these were accurate."
Seven billions dollars. Let's look at it another way, $7,000,000,000. That's one helluva stack of frogskins.
"Collapsing under its own weight"
More like, COLLAPSING UNDER ITS OWN GREED.
Others here have made passing mention of this. Since when is one single person given control over $7 Billion in assets? Yes, I know that Jerome Kerviel circumvented certain checks and balances but all of this was done with some degree of collusion by upper management.
The point remains that one single individual was entrusted with operations that would normally be assigned to an entire team of traders. This is merely a case of corporate greed. In seeking to reduce salary expenditures, the bank over-concentrated control into the hands of a single person. This is and always has been a recipe for disaster.
What happened to Soc Gen's internal review and auditing process? Why aren't the bank's CEO and comptroller being arrested for malfeasance? Most certainly, whoever wrote the software should be hung out to dry for incompetence, at least.
Worst of all, who will end up taking the actual hit? The bank's investors? Pardon me if I have severe doubts that any insurer is going to absorb this financial fiasco.
Ledger: This highlights how a relative obscure trader by the name of Jerome Kerviel can destroy $7.2 billion without the watchful French Authorities knowing... Wink... Wink... Or, how massive Car-B-Qs go unpunished.
I think that this may be a bit closer to the truth. Much of European financial policy hinges upon a socialistic fantasy world of money magically appearing when needed without concern for its actual origins or the ramifications arising from its expenditure. Few other explanations exist for the lax attitude of French authorities confronted with a BILLION dollars worth of cars being torched.
Allison says bank's real commodity is trust. I have to agree.
The longer I live the more I find how easy it is to break and bend rules. If one is noticed most often those who notice don't have the constitution to enforce the or carry through as specified.
That it most often comes down to people abiding by rules of their own volition.
A guess:
This is just the French Bank's way of finally writing off at least part of their huge losses that were incurred when Mr. Bush invaded Iraq and disrupted the lucrative oil deals the French had going with Sadam.
The whole story reeks of beaurocracy.
I feel priveliged to be able to sit in on this thread.
Especially Ledger.
Thank you, Wretchard!
I think that Allison is probably correct and that it took more a lone fraudster to bring down SocGen. Time will tell but I suspect that other people at the bank were long aware of holes in the risk-mgmt system.
Many of the commenters here ignore a key difference between markets and other complex systems. Markets change in response to the changing expectations of market participants.
Market participants have strong financial incentives to avoid future debacles, and will change their behavior in ways that reduce the risks of such things happening -- but also in ways that do not impede the orderly and profitable functioning of markets. By contrast, the main incentive faced by govt regulators, of markets as of new drugs and other technology, is to avoid debacles on their watch. They do not profit when markets are open and profitable, and therefore will tend to overregulate.
The markets have already corrected for the most recently discovered problem. Shares have been revalued to reflect newly revealed risks; trading-firm managers on whose watches the debacle occurred have been sacked. As with the market selloff during the summer of 2002, that was followed months later by Sarbanes-Oxley, the market adjusted immediately. The costly legislative "remedies" applied after the fact had no effect, but were credited because the problems they were intended to solve no longer occurred after the new laws were enacted.
While financial debacles are often very costly to market participants, they are a boon to politicians and officials in the regulatory bureaucracies. That is a bigger problem than the occasional market shakeup.
Post a Comment
<< Home