Slashdot Mirror
The Perils of Simplifying Risk To a Single Number
A few weeks back we discussed the perspective that the economic meltdown could be viewed as a global computer crash. In the NYTimes magazine, Joe Nocera writes in much more depth about one aspect of the over-reliance on computer models in the ongoing unpleasantness: the use of a single number to assess risk. Reader theodp writes: "Relying on Value at Risk (VaR) and other mathematical models to manage risk was a no-brainer for the Wall Street crowd, at least until it became obvious that the risks taken by the largest banks and investment firms were so excessive and foolhardy that they threatened to bring down the financial system itself. Nocera explores the age-old debate between those who assert that the best decisions are based on quantification and numbers, and those who base their decisions on more subjective degrees of belief about the uncertain future. Reliance on models created a 'false sense of security among senior managers and watchdogs,' argues Nassim Nicholas Taleb, who likens VaR to 'an air bag that works all the time, except when you have a car accident.'"
For an EXCELLENT article about this, read Malcolm Gladwell's "Blowing up", which you can find online for free:
http://www.gladwell.com/2002/2002_04_29_a_blowingup.htm
Watch the Teaser Trailer for "The Lightning Thief" Her
1 (?!)
42 ;)
you just had a car accident and the airbag didn't blow up
"DRM is like the Ford Pinto: it's a smooth ride, right up the point at which it explodes and ruins your day."-C.Doctorow
The problem isn't so much reducing risk to a single number -- it's the risk of reducing it to the wrong single number. Put simply, VaR (as measured and used by most banks) tells you how bad it will get 2 or 3 times a year. Great -- it's a fantastic measure for management to have. But to use it (as regulators did, and managers were seduced to), as a proxy for how bad it would get once every thirty years was nonsense -- that simply wasn't what it was measuring.
http://www.nytimes.com/2009/01/04/magazine/04risk-t.html?_r=1&pagewanted=print
http://www.nytimes.com/2009/01/04/magazine/04risk-t.html?pagewanted=all
Hmmmm. Math or "subjective degrees of belief about the uncertain future".
I've always operated on the principle that they were all lying, thieving, immoral, unethical, and greedy fucking bastards that were ready to bend you over for a nickel. Seems my supposition is being proven correct more and more each day.
Until recently, it was the smaller guys in the stock market that were getting screwed and the whole system kept the thievery down to a manageable level. Now from the largest, to smallest, they all seem to be getting destroyed, American in ruins, and the previously rich and powerful with outstretched hands at the Feds.
Of course maybe that is too cynical, but I always saw the stock market as rigged from the beginning. What do I know though? :)
the odds are stacked way against US. we note that without a legitimate replacement, folks greed/fear/ego based ?thinking? sends then right back to the scene of the crime for some additional punishment. better days ahead.
So what's the VaR of using VaR? :)
When you don't have a leg to stand on, don't even get up.
Simplifying risk to a single number is as dangerous as simplifying a decision to buy or sell to a single boolean value.
It's something that has to be done at some point regardless of how the risk is estimated and how all the information is used to make the final decision.
A bit more to the topic: One just has to find the correct numbers and use them well for decision making. It will be more reliable than hand waving and can actually be analyzed.
Some people enjoy "69", but "42" has always been a lifesaver throughout the last 30 years...as long as I have a towel. Hmm... I thinking the "69" people also may want to have a towel as well.
First, don't forget that Taleb is selling something. Very smart guy, but he wants to make a good living too.
VaR isn't something I'd want to be without, but you clearly can't depend on models alone when your assumptions are uncertain. That's what the whole mess with CDOs and such comes down to- bad assumptions.
With the mortgage-backed market (e.g. sub-prime), the assumption was that N number of borrowers would default in X period of time. If they had the models would've been fine, but in reality they didn't, and the basis for the assumptions was horribly incorrect. Why those assumptions were incorrect is another story.
I don't think that the problem is a single number it is connectivity. You might think that if you have three investments with a 10% risk of losing £1,000,000 the chances of all three of them losing £1,000,000 is 0.1*0.1*0.1 = 0.001 or 0.1%. The thing is if one loses that much then the markets may lose confidence meaning the others go down too - they are not independent probabilities.
Garbage in - Garbage out, much of the input data to the VaR was questionable.
The problem in the financial world is one of thinking there's a single factor called "risk". In fact there are many, interlinked factors: The risk the business will go bust is one - however from that sprout a whole range of subsidiary risks: from losing all your investment to getting back 95% of it.
Similarly with mortgage risk and any other type of investment. What the financial markets need is a better understanding of the causal links between risks and to price the returns on investments accordingly.
That will be a *big* job, and one that will take years or decades to iron the bugs out of.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
Indeed we did. And I think we came to the consensus that it was a load of bollocks.
Confucius say, "Find worm in apple - bad. Find half a worm - worse."
Nocera explores the age-old debate between those who assert that the best decisions are based on quantification and numbers, and those who base their decisions on more subjective degrees of belief about the uncertain future.
Why is anyone still making this distinction, as we now know that the only self-consistent numerical representation of risk follows directly from our subjective degrees of belief about the uncertain future? Furthermore, we have known this for over a generation... isn't it about time that the knowledge start filtering into the popular discourse?
While Bayesian methods are not always all that useful for practical problems (I use them on occasion in my work) the conceptual foundations and deeper understanding of the nature of plausible reasoning and its relation to probability theory needs to be more widely understood.
One of the big take-home messages from the Bayesian revolution is that probability theory is nothing but quantification of what we do subjectively, insofar as our subjective impressions are self-consistent, so the only people who are still debating quantitative vs subjective approaches as such are people who do not understand the question.
Blasphemy is a human right. Blasphemophobia kills.
There is a guy called Steve Keen who more or less predicted the collapse using an ODE system. When he presented his findings (some time ago) he claims that the economists in the crowd thought that a 6 variable ODE was so complicated that he had to have made a mistake.
An air bag that works all the time, except when you have a car accident.
A question for the powerful minds of /.
How is the risk of driving with that airbag in your car compared to a normal one?
- Greater.
- Smaller.
- Exactly identical.
- Unknown until you open the box.
Is here any roleplayer that does NOT know how using an artificial value to describe "real" problems automatically leads to some people "playing the system" instead of playing the game?
Nobody here ever had a munchkin in his troupe? A powergamer? A minmaxer? Someone who learned the rules and immediately started to look for loopholes, how to play by the rules without actually taking them serious?
Now why did anyone think this would be different when real money is involved, and thus the incentive to abuse the rules way higher?
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
Value at Risk (VaR) is simply business/finance jargon for quantiles of a probability distribution, heaavily promoted by JP Morgan since the 1980s and widely adapted by financial and non-financial institutions alike. It's usually the 95 or 99 percentile of the value of some portfolio per day/week/month/year.
Now there is a lot of criticism about using wrong distributions (usually assumed Normal/Gaussian). Economists have known this for a very long time: pretty much anybody who as studied the subject knows most financial returns are not Gaussian. Even in an undergraduate course I had to calculate VaRs with thick tailed and asymmetric distributions. I'm not sure if this is done in practice in big firms though. The amount data to estimate the statistics are typically far bigger than in class assignments, and assuming Gaussianity makes the calculations very easy.
I think BTW that it's more sensible to recognize that variances and correlations are changing over time, and use proper dynamic models to account for that, rather than simply replacing the Gaussian with a thick tailed skewed distribution.
So Mr. Mathematically-savvy Man, why don't you go ahead and transform economics for the better? I'm sure there are many more "obvious" things out there to come up with.
VaR is a pretty decent risk measure on a micro scale. The real problem with it is that VaR constraints tend to make banks less diversified, introducing systemic risk. When things go sour, banks are forced to sell off similar assets, and because all of the banks tend to hold assets with similar risk, markets fluctuate all the more.
It is telling that a broad index of hedge funds is better resistant against risk than an index of banks.
Risk models are largely irrelevant because the only risk anyone in the financial sector is really interested in minimizing is the risk that they will get fired. The way to do that is to do almost exactly the same thing as everybody else, no matter how mind blowing stupid it is. Plenty of people realized that banks etc were not nearly as sound as commonly believed years ago. Those that tried to act on this were fired long ago since they weren't making as high a ROI as those willing to invest in dodgy hedgefunds etc. Rational market my ass.
http://rareformnewmedia.com/
If the Government becomes a lawbreaker, it breeds contempt for law;
And then there are the computer models that predict climate. As a single number. Temperature of the atmosphere. For the next 100 years. Scary?
Yes, no kidding. As flawed as some of these models are they definitely do not treat all investments as independent.
One of the best features of a VaR model is to offset positions against each other. As an incredibly simple example lets say that after a day a of trading one prop desk has a long ibm position and another prop desk has an equally sized short ibm position. Its important to understand that those two positions effectively cancel each other, and your VaR from them is $0.
This concept gets much, much more complicated when dealing with derivatives and relations across asset classes.
Ummmm...shouldn't that be 0 then?
My blog
Objective or subjective models don't mean anything to people who only care about short-term performance. Whether the investment is good or bad in the long term doesn't matter to an investment manager who stands to get a seven figure bonus based on the current year's numbers. So what if the company fails next year? Not his problem.
I think there should be a 10 year moratorium on the invocation of the central limit theorem in financial mathematics.
These jackasses assume that everything vaguely bell shaped is a normal distribution and then fall to pieces when it turns out that the real distribution has fatter tails.
Liberal economics -- not liberal politics, quite the opposite most of the time -- explicitly derives its conclusions from three assumptions: that individuals make rational decisions, that they have access to information, and that they are free to buy/sell.
Those are pretty reasonable assumptions, and, when they hold, the conclusions tend to hold.
The difference with physics is that when physicists start saying "assuming that this body is of negligible mass and at non-relativistic speeds" they don't end their exposé with "thus we have a solution to the three body problem for three super massive black holes at 0.999 c"
Social psychology has shown repeated instances where rationality is seriously impaired. For example, social proof can make us all really stupid. And cognitive dissonance is a bitch. What do those words mean? When a million idiots do something stupid, you're very likely to think it's a very good idea, too. And the longer you've been doing something stupid without negative consequences, the less likely you are to stop.
Add to that the fact that those "investment vehicles" were designed to conceal information, specifically financial risk, and right here you have two out of three pillars of classic economic theory missing. Is it any wonder the whole thing went down?
Finally, I wonder if any free marketer / libertarian types actually read any Adam Smith. I remember reading a quizz, which unfortunately I can't find anymore, Marx vs. Smith, in which you were asked to identify whom had written what. Very hard to tell them apart in some cases.
If it works for Dick Clark, it works for me.
"I give Bank of America a 77, because of their assets and because I can really dance to it. "
Vincent J. Murphy
Spandex Justice
Yes giving out so much in bonuses made people focus on short term gains, rather than stable returns, some models were wrong, and hundreds of other factors. One factor that I think is over looked in this is the number of times the computer programs/ algorithms etc were ignored. I have heard off friends working in the industry that computer models that favoured a less risky lower return approach were binned or ignore. Unwelcome risk reports were made to disappear. It was not computers or mathematicians that are to blame here. It is the people who thought they were smarter than the rest of us. Those who thought it would never end, well all things come to an end, and the sick thing is these are the ones who made the millions and left everyone else in the S***.
Anyone sane enough to borrow money wouldnt ask.
Anyone asking isnt sane enough to use borrowed money.
Its been the lending policy for years, and I am sure it will continue this way in the future. They continually throw offers to borrow money at my wife who has excellent credit, and enough money in the bank to not need what they are offering.
So, could you summarize, in a number, how risky it is to use a single number to represent risk?
The problem with using a single number is simple: It is easily gamed and there's lots of incentive to do so.
So people will sell you worthless junk that technically has a high number rating because if you're relying on the number you'll pay them for their worthless junk.
Fanatically anti-fanatical
As an example, suppose that the distribution suggests the chance of losing 50 million dollars is +3 sigma for some measure. The problem is that there is a subtle effect - say panic, herd effect or some interaction of derivative models - which only becomes significant around the 3 sigma mark. The result could be that the exposure at a 4 sigma event is billions of dollars. A proper risk model would need to take this into account
My conclusion based on what I have read so far is that the physicists (in particular) involved in developing quantitative models would have benefited from a lot more exposure to real world experiment. They would then have had more of a clue about the unreliability of data away from the mean, scatter, and the importance of the fact that in physics subtle errors turn out to be signs that the model is wrong - e.g. relativistic effects only become important at a significant fraction of c.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
Economics is not a science.
Science is the application of the scientific method. When's the last time you saw an economist perform an experiment where only one variable was at play?
If he explores all forms and substances Straight homeward to their symbol-essences; He shall not die.
OK. And can you give us a single number of the risk of simplifying to a single number?
A friend of mine is a risk assessment quant who was working at Lehman right up to the point where they declared bankruptcy. I asked him about this article the other day. He said that their models started telling them something was very wrong back in 2007. The problem was that Fuld (the CEO) refused to believe what the models were saying.
The most accurate model in the world won't help if you don't pay atention to the results it produces.
There's also apparently an issue with the classical VaR models depending on transparent pricing, which these real estate instruments lack. So some of the most troublesome assets apparently weren't in the model.
"The purpose of computing is not numbers, but insight" - Hamming
"Simulations are like political prisoners, either can be tortured to tell you what you want to hear." - Unknown
But beyond the clever idea that we shouldn't all become mindless drones to our simulations if we don't understand the underlying principles that went into them is the problem that, at least in the financial world, risk and reward became disconnected, mortgage brokers being the perfect example. Mortgage brokers don't get paid unless they sell a mortgage. Their income depended not on the soundness of the mortgages they sold, but merely on the number regardless of how small a chance there was of the mortgagee managing to support the payments. Every time this bad debt got sold someone made a commission, so it was repackaged and resold dozens and dozens of times.
Also, when people talk about all the money the banks lost on this bad debt, it's not like it went up in smoke. That money went somewhere. Some people made phenomenal killings in the bad debt market.
Jealously hoarding mod points since 2007.
You might think that if you have three investments with a 10% risk of losing £1,000,000 the chances of all three of them losing £1,000,000 is 0.1*0.1*0.1 = 0.001 or 0.1%.
No, no-one who actually calculates and uses VaR thinks that. Anyone who has done any statistics, like all finance quants, will correctly take into account covariances. The actual problem is the interpretation of the "correct" VaR, and relying on it too heavily.
I'll give you the actual definition of VaR. If you calculate the VaR(10 day, 5%) to be $100,000, this means that there is a 5% chance that the loss on your portfolio over a 10 day period will be larger than $100,000, or that your profit will be larger than $100,000 assuming a symmetric distribution. It's when people think "Oh that's great, we can ONLY lose $100,000" when you have a problem. The actual loss could be ANY value larger than $100,000.
It's hardly a perfect statistic, since there are still many assumptions involved. However, it's still a decent estimator and it's better than making a wild guess based on gut feelings. Despite what most people currently believe, a lot of brainpower has gone into developing financial theories and some stuff is pretty damn good. The financial industry deserves some bashing, but it frustrates me when people spread incorrect information; at least complain about the right things.
VaR is a pretty decent risk measure on a micro scale. The real problem with it is that VaR constraints tend to make banks less diversified, introducing systemic risk. When things go sour, banks are forced to sell off similar assets, and because all of the banks tend to hold assets with similar risk, markets fluctuate all the more.
Risk is really a bunch of intertwingled probability functions, which are probably infeasible to calculate. This seems remarkably similar to quantum mechanics and entanglement, so perhaps economic modelling would be a good use for quantum computers? Ignoring the intertwingling is probably about as silly as trying to use classical methods to calculate quantum interference, too...
I saw Nassim Nicholas Taleb speak a few months ago, and my immediate reaction was that he was seriously misguided.
His argument, if I understood it correctly, was that all risk modelling is dangerously faulty because it is impossible to know the impact of an event which is extremely rare (a black swan). Thus, while we can say that it may only have a 0.001% chance (or whatever) of occurring, we can't say with any certainty what the effect will be.
My problem with that, specifically in relation to financial institutions, is that the maximum loss is capped. A hedge fund can't lose more than it owns/manages. The maximum risk to a â5bn fund is precisely â5bn. Therefore, once you have a conservative estimate of the probability of the event occuring and know the maximum exposure, models are useful again.
I'd say that if you use a single number, the peril would be 52.994.
UTF-8: There and Back Again
Reality's an untamed beast
That's difficult to master,
But models are quite docile
And give you answer faster.
From a pome I saw in a computer book from the 70's, can be found online here http://www.langston.com/Fun_People/1993/1993AFE.html
vi +
Even if it's a complex number? Or would that make it too imaginary?
Instead of complicated analysis based on people taking on too much debt and asset bubbles and government regulation or lack thereof, just blame everything on George Bush. It is easier than thinking and a bunch of people will think you are insightful.
Except for ending slavery, the Nazis, communism, & securing American independence, war has never solved anything.
The real risk score can be a single number however the number changes as time progresses. The investments the banks made were relatively safe in the beginning however as time moved on those investments became riskier relative to market conditions. An investor (bank) needs to be in a position to unwind an investment as soon as the risk score exceeds a certain level. Of course as the investments begin to unwind the risk scores increase for others with similar investments and you have a crash. The key is to diversify the risk and to assume the risk is often greater than the calculated score.
VaR is a pretty decent risk measure on a micro scale.
No, it's not. More precisely it is but only for 99.9% of days. That's Taleb's point. You look at the VaR for 3 years in a row and it's always right, you grow to trust it, and then the big whoop comes.
The real problem with it is that VaR constraints tend to make banks less diversified
Actually it's other way around. It's the property to diversify as much as you please that VaR provides. That's why improper calculation of risks involved caused such a wide spread of problems.
What we need is for the upside/downside/inside risks that all the banks are exposed to, to be made public. That way customers can decide for themselves which bunch of cowboys to entrust with their cash. Sadly, as we've found out, they're all as bad as each other.
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
http://www.nakedcapitalism.com/2009/01/woefully-misleading-piece-on-value-at.html
In summary, the NY Times piece has basic assumptions of how VaR works all wrong specifically that the models assume normal distributions.
What you appear to be trying to describe is the neo-liberal paradigm. That's not really what I'm talking about, although it is my opinion that it is complete bollocks, but that's just my opinion.
My point is that you can't take liberal economic theory, keep the conclusions and expect them to hold when you've clearly removed the starting assumptions.
On top of that, what you write isn't even logical:
Or maybe, the assumptions are:
- that individuals make decisions which are more rational than if someone else makes it for them
What does "more rational" mean? Classical economic theory assumes that someone is rational in that it will buy something at a lesser price if it can, and will attempt to sell the least of something it's got (good, service, labor ...) for as much money as it can. That's it. How can you be less rational?
In any case, if there is government intervention, which is I suppose what you are against, it's got nothing to do with the rational part of the argument, it has to do with the freedom part of it. And I haven't talked about this.
- that they have access to better information
Again, what does that mean? If gov't regulation forces companies to be more transparent (a la Sarbannes-Oxley), it means less freedom for the company but more information for the market as a whole. It's once more an impact on the third assumption but clearly not on the second.
I disagree, accurate information is required for the market to choose the most efficient price/product.
"In the long run we're all dead" Keynes famously said; and in that case it means that, sure, if we wait long enough, people will stop trusting liars and crooks and they will be weeded out, but by that time damage will have been done, and the market will not have functioned optimally in the mean time.
The problem isn't the models; models predict behavior quite well. The problem is trying to reduce the predictions of models to a single number; that does not work.
1 being worst 10 being best.
..to swallow this whole "accidental" BS economic meltdown theory. One, that all these fancy super high paid corporations and Cxx whatevers and governmental regulators are all incredibly stupid, and that their alleged computer programs didn't wargame out the obvious conclusions about trying to sell convoluted contracts with hardly any basis in economic reality. I call BS on both assumptions, which leaves the obvious, it has been a series of really high level crimes, an economic and political power coup, designed and implemented in order to transfer real wealth upstream into much fewer hands and to give those megathieves even more global political power. If you look at it as a crime instead of an accident, it makes a lot more sense given the results we are seeing now.
How many people would it take to pull off? Not many, a few dozen tops, some heads of central banks and a few national treasuries and big investment houses, and that's it.
It was engineered, it wasn't an accident. They want as much control over the planet as they can get, and there are only two ways to do that, fight a lot of messy wars, or just completely control the currencies and credit and stocks and bonds situations. Manufactured boom and bust cycles, manufactured "liquidity" problems, various laws and etc that let a small handful of people control all the major currencies. That's some serious power there, held by mostly unelected and unaccountable globalist thieves who have no loyalty to anything but their own wealth accumulation and power. And the thieves get the bully pulpit to "explain" what happened and it is sucked down in the press as the real deal.
Something about foxes and henhouses comes to mind.
Furthermore, engineered systems have two separate control systems: normal operating controls and independant safety controls. Never the twain shall meet, for often the normal controls exacerbate the situation and must be pre-empted by the safety controls. The more advanced the normal controls (optimization), the more advanced the safeties have to be.
None of this is present in finance. VaR may be all well and good as a normal operating measure, but does nothing in the tail which will blowup. I do not see anything as a tail safety measure institutionalized. What measures are taken are done on "gut feel".
Risk, in financial terms, is a measure of the variability of returns, i.e. the standard deviation of the returns. A well-diversified portfolio generally reduces the variability due to the individual risks of investments being uncorrelated. Harry Markowitz, the father of portfolio theory, pointed out that these quants all assumed that a basket of mortgages is highly uncorrelated and thus well diversified. However, in a broad real estate downturn, they all become very highly correlated. Therefore, if your standard deviation WAS 10%, it suddenly becomes 50% or more, which rapidly changes your VaR from a handful of millions to several billion overnight. VaR, being an oversimplification, didn't take that into account and all the big investment firms suddenly had billions of dollars at risk and billions of dollars of losses without realizing it. It's simply a matter of garbage-in, garbage-out, something my Portfolio Analysis prof drilled into our head and hopefully gets drilled into the heads of Wall Street CEOs.
I'm just stating the so-called liberal economic theory. It's not my opinion; it's a fact that it is a starting point of that theory.
The Fedederal Reserve bank announces that there's going to be an upturn in the market, if everyone believes them then they go on a buying spree and the index goes up. If no one believes the Fed then they go on a selling spree and the index tanks - and they fire Alan Greenspan and hire on a new head of the the Fed ..
davecb5620@gmail.com
You are right, but the grand parent does have a point though.
Most banks understand the idea that the 5% losses will be very large numbers and use some sort of non-linear models to estimate those losses - there clearly were errors, but they were not of the order of magnitude to sink the banks.
What really killed them is really the sudden outbreak of correlations. First it was the real estate slowdown, then it became the credit crunch, then later equity markets crashed and then they bought down Europe and emerging market countries. Somewhere along the line it also caused bankruptcies in companies and started killing the real economy. These things were supposed to be uncorrelated (or supposed to have low correlations) - when they became correlated it killed the whole country. If the correlations had held, then a slowdown in the real estate markets should not have caused the bond markets to tank (they are really products that compete with each other for investor money and should move in opposite directions) - the tanking of bond markets then caused Private equity companies to sit on the beach, which in turn killed the revenue to the banks--again another unexpected correlation.
On the other hand, I would think that the extreme "Black Swan" has not really materialized. S&P used to move up and down by a percentage and a half a day on the average and now it moves up by 4-5% a day. These are same order of magnitude, even if it is unusually high. Bond default rates used to be 0.5% for AAA, now it is maybe 3%. Housing prices used to move up by a couple of percentages a month, now it is falling by 5-10%. None of this is really a Black Swan - a grey swan maybe. When S&P falls by 20% a day or when default rates hit 15% for AAA or when housing prices fall by 50% a month -- That would be a Black Swan. It would indicate that the US market is a different animal from what we thought (and acts similar to Russia/China/India markets).
http://slashdot.org/submission/1062723/Cheap-mobile-data-plan?art_pos=2
Is Australia a number? Because it's that simple.
... is that any probability of risk figures have to be based on the available data. Part of that data is the correlation between a portfolio of highly rated financial instruments vs that of those with a lower rating.
When the bond rating agencies take a portfolio of junk bonds or mortgages, wave their magic wand over them and grant them a rating of AAA, the whole risk evaluation system is thrown off. How they (the bond rating companies) can sell (at very high prices) essentially worthless information to their customers and escape any liability (so far) is beyond me.
Have gnu, will travel.
Math works great. A single number is fine. The problem is when the information used to get that single number is based on lies. The math correctly calculated the answer they wanted to see after they put in the wrong information. This could be taken two ways, that fraud is bad, or that math is bad because people who do fraud can use bad assumptions to prove anything. And, in the era of fear of science we have going here, we know which it will be. Ever notice how Enron wasn't about evil people making up numbers, but how "accounting" is to blame? "They got it on the books with dummy corporations and such" but no mention that much of what they were doing was illegal at the time, and all the laws since wouldn't change the outcome (they lied on financials, their auditors knew and didn't report it because to do so would get them fired from a huge account). It wasn't an accounting error. It wasn't loopholes. It was a massive criminal conspiracy to generate wealth from smoke and mirrors. That they used math doesn't make the math wrong, yet it gets blamed.
So I guess the answer is, "A single accurate number is all you need. If you expect fraud, then multiple independent numbers would help identify areas where the numbers deviate from their expected values."
Learn to love Alaska
The biggest problem with those creating financial instruments from home loans is that no one tested their models with systemic housing price decreases.
Economist Arnold Kling said that many years ago, Freddie Mac actually did "stress testing" of their portfolios under a 20% systemic real estate market downturn, but during the early 2000's they abandoned this technique.
CDOs did a good job of reducing the risk of early repayment and "random" defaults on mortgages. However it ended up concentrating the risk for a systemic market downturn.
Unfortunately, I am sure that some time in the future there will be another huge systemic risk that both government and the private sector will miss and we'll get hit again. The only thing we can do is keep economic freedom high in the period in-between to allow the economy to restructure (less jobs in building, more in health care, for example) in order to return to growth.
Had the GP bet on that, he'd be destroyed by short term market manipulation that several entities, including your federal government, make several times a week, like almost everybody that betted against WallStreet.
Rethinking email
Probability of something going wrong: 1.0
Rethinking email
I've never understood why it cost so much to manage 401k funds if the whole stock market had really been reduced to a simple risk figure. Where's the quality of jugement I was paying my 401k fund manager for ?
When stocks were rising, then I suppose retirement investors looked the other way as some of their profits were taken as management fees. But now they are going down and we're still paying managers to manage them downward
I hope that there will be more focus on simpler 401k options that don't need so much management overhead, maybe even some federally managed funds with no management fees.
Nullius in verba
It's pretty clear what they (classical liberal economists) mean by rational, information and freedom. The definition is part of the theory.
And since this is a theoretical model, it is also understood that nothing in reality fits perfectly.
When people are rational most of the time, are reasonably informed, and have some freedom to buy/sell, market will work for the greater good. That's the theory.
I'm just saying that here people weren't informed, and weren't being rational due to social proof + commitment; and that there's no need to invoke the dreaded loss of freedom to realize that the whole system couldn't work according to freemarket fundies' theories.
Access to information is never perfect -- being subject to scarcity like all other goods
Really? That's a very peculiar statement to make in this day and age, and on this particular site.
In essence, you seem to be treating freedom as an independent (even insignificant) aspect of economics, when in practice you cannot assume either rationality or optimal access to information without it.
I get it, you're a libertarian. You defend your opinions, if only just by parroting your usual lines.
Me, I'm just looking at the underlying theory. Rationality, information, freedom. Three conditions. Two of them are missing; whether the third is present or not is moot.
What's so hard about it?
Disclaimer: IAAMFPHDS (I am a mathematical finance PhD student).
While quants could accurately gauge the historical covariance of different assets in a portfolio, what they failed to take into account is that there is correlation in the tails of the distribution.
An example of this is that, back in the good old days, there was a degree of correlation between the Dow and the FTSE 100. If the FTSE 100 went up, it was a decent indicator that the Dow would also be up, but by no means a sure thing. However, during the crisis, the two indices practically moved in lockstep.
The moral of the story is that in the rare event that things get bad, correlation tends to spike. The models failed to take account of this, which is part of the reason we're in this mess.
Slashdot: news for Apple. Stuff that Apple.
Sweet, I love to hear from people who know more than me. Yeah, I already knew intuitively what you're saying, but I definitely couldn't have articulated it like you did. Thanks.
...knowing that human relationships, "networking", "gut feeling", and trust are far more reliable than cold numbers, entrusted their money to Bernie Madoff.
Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
It sure looks like the market has gotten incredibly removed from reality. Huge sums of money are invested in side-bets such as options and dervitives, with less actually invested in owning shares of actual companies.
Getting "blown up" happens to the derivitives traders, not those who hold actual stocks. That is what has taken down the firms that spectacularly cratered earlier this year - they were holding options and swaps that they or the other party could not actually cover when the markets shifted.
The problem with VAR is not the measure itself, which is assuredly useful if one understands the limitations.
The problem is that once any risk measure (that is say, 95%+ 'reliable') becomes institutionalized as the gold standard, catastrophic failure of the financial system is inevitable (at least according to the general black swan theory).
Why? Because any firm that doesn't optimize profit against the risk criteria is going to have a lower P/R, and will lose capital to firms who are more 'efficient' at investing as long as things are 'normal'. This will result in the firm either folding, or being acquired.
If the firm does optimize to the risk criteria however, they stick a ton of risk into the tails of the risk distribution, which isn't measured, and so they'll get taken out when the black swan hits (ie. a rare event occurs, and all that hidden risk smacks them upside the head).
(and yes, I know this is a very simplified explanation).
No, it's not. More precisely it is but only for 99.9% of days. That's Taleb's point. You look at the VaR for 3 years in a row and it's always right, you grow to trust it, and then the big whoop comes.
That's all in the game. This is the case with any risk measure, alone or in combination with others. The worst-case scenario is the Dies Irae; you cannot take it into account if you want to do any business at all.
Actually it's other way around. It's the property to diversify as much as you please that VaR provides. That's why improper calculation of risks involved caused such a wide spread of problems.
The point was (and I am sorry if it didn't come across clearly), that though banks are better diversified on the micro level, this leads to them holding similar portfolios, introducing systemic risk. That is, banks, as an aggregate, are difersified poorly as a result (among others) of Basel-induced VaR contsraints.
Standardized, regulated exchanges usually come about when a market already existed before, but it would be desirable to have a more transparent, reliable market clearinghouse. With stocks, people invented shares long before anyone opened an actual stock-exchange: you write out a contract on paper agreeing that, in return for $x, so-and-so now owns 1 share of your company, and you have a contract somewhere specifying how many total shares there are, when/if new shares can be issued, etc. It's basically what happens when you try to expand a small partnership to more than a few people and bring in investors.
Once you get enough of these fractional-ownership certificates floating around, each with slightly different rules, and disputes start arising about who owns what and what that means, the logical next step is to agree on some relatively standardized method of fractional ownership, and a central clearinghouse to trade the certificates. Which is what stock markets are.
On the other hand, moving that sort of business to stock markets also increases the number of market participants and frequency of transactions by reducing entry and transaction costs---it's impossible, for example, to "day-trade" paper stock certificates in person directly with their owners thousands of times per day. That has positive and negative effects---positive in that it increases available capital and gives retail investors more parity of access compared to large investors, and negative in that it makes the whole system more volatile and sensitive to chaotic-systems effects.
10 PRINT CHR$(205.5+RND(1)); : GOTO 10
Wow. It looks like most people don't get it.
If Mr A gave Mr B billions of dollars of The Public's Money to play at a casino and both Mr A and Mr B got filthy rich when times were good, and when it blows up all that happens is Mr B loses his job and Mr A keeps his job by blaming Mr B or saying BS like "perfect storm/everyone was doing it".
Why then should Mr A and Mr B be doing things differently?
After all, in the following year, Mr A passes billions to Mr C who does pretty much the same thing as Mr B. And Mr B? He's hired by Mr D who wants Mr B to make him richer (just like he did for Mr A).
AFAIK, not long after LTCM blew up, its founder John Meriwether still managed to get hundreds of millions of dollars to start a hedge fund.
What I see are individuals making pretty rational decisions, those decisions sometimes just happen to be bad for a lot of other people. But why should those individuals care?
Their conscience should bother them? The last I checked the Economists leave the conscience stuff to "The Invisible Hand". People laugh at the religious, guess who really has even less of a clue on how things work? At least the religious have some idea about the "Invisible Hand" sort of stuff.
It's hilarious that you have all those people saying/writing stuff like "When Genius Failed".
That's like the sheep saying the wolves have failed just because the wolves dropped 95% of a billion sheep over a cliff, whilst "only" managing to stuff themselves to the brim with 1% of the billion sheep. I'm sure the wolves were a bit upset about the whole thing, but hey there are billions more sheep...
Yeah I see failure alright. Go figure where.
You want to reduce the risk of stuff blowing up, and how big they blow up? It has nothing to do with creating better financial models or better economic theories.
It has to do with making and enforcing rules like: if too many sheep die, we shoot and skin the wolves responsible. Simple as that.
All that transparency and regulation is worthless if at the end of the day the wolves get away.
Taleb overlooks the problems with the other side though. If you have a mandated risk management system, the hopefully rare, occurrences of model error, or just the "long part" of the distribution mean that the whole banking sector can fail together rather than just losing a few banks. To much regulation essentially means the regulators picking the risk model for you.
I think a mixed approach is the best, which I think Basel 2 is trying to force (but could do much better). You need to have some controls to ban hair brained approaches. But banks should have the most interest in their own survival. Allowing flexibility to model their own risk should lead to more innovation; banks have an incentive (and the money) to model better, less capital required = greater ROA, and also investing opportunities open up. As well, it should diversify the sector more which should allow it to continue even if one area gets clobbered.
Here's my experience doing software and system project mgmt for engineering-related stuff (DoD, DOE, NASA). It's very anecdotal, very notional and works out to be usually very valid, provided sufficient peer review and/or experience.
First, for any component or decision characterize a risk value of (complexity/maturity). So something highly complex not done before is high risk. Something simple with many precedents for success is low risk.
Next, map risk in terms of susceptibility to sustainability (USAF OT&E terms), or IOW - probability of (a bad thing's) occurrence to its impact on the system or the user to be able to do the job via a work-around or some other "operate-through" procedure. Stated another way, if something is susceptible to some risk, will it be able to sustain operations in the event of a failure?
So, just because something had a high risk value doesn't mean to NOT go that way - because it might have a low probability of occurrence (a feature very seldom used for example (based on use cases)) and a work-around with high confidence (or whose impact is too low to care about).
On the other hand, something with a low risk value might need re-assessment or re-work under the conditions that it has a high chance of use (therefore a high chance of occurring) and no acceptable work-around (and it impacts the user so that he can't do his job).
So. I guess I can safely say that I'm glad I'm not in economics because I can't for the life of me take the above life experience - that is that risk assessments and decisions are multi-dimensional - and then say that it CAN make sense to reduce risk to a scalar value - for any system, be it technical, economic or social or what-have-you.
Pathological kinda promises Path + Logical - but instead, you get stuck with pathetic.
That's all in the game. This is the case with any risk measure, alone or in combination with others. The worst-case scenario is the Dies Irae; you cannot take it into account if you want to do any business at all.
Here's the thing: you cannot put it "into the game" when you use incorrect distribution (as they did with VaR). Otherwise pretty much everything (including the world end) can be taken into account given the right price (and possibility of payments in hard currency). The main problem will become the "gambler ruin" (in real "act of God" situations) which will drive premiums even higher than the distribution calls for. Plus, by all accounts, 2008 was not the year the world ended (only seemed that way).
So VaR as used by them was not even a proper "risk measure" - that was one of mistakes they did thinking that it was.
Sure, if I were working for a Wall Street bank and used a more realistic model I would get my ass fired in no time for what it would look like being so risk averse. But that's a systematic problem with businesses today.
The point was (and I am sorry if it didn't come across clearly), that though banks are better diversified on the micro level, this leads to them holding similar portfolios, introducing systemic risk. That is, banks, as an aggregate, are difersified poorly as a result (among others) of Basel-induced VaR contsraints.
Ok. So they diversified they asses to such extent that overall financial system "connectedness" grew high enough to make it unstable.
> After seeing the rampant fraud committed by the global financial elite, I'm very inclined to agree with you. What we need isn't just a number that quantifies risk, but also a number that quantifies trust.
Won't they just game that number, too? Once you fix the rules for any system, people will start to attack them. And, based on what I've seen in online games, they'll find a way to break them. Especially when there's real money at stake, not just virtual gold and items (though even those can be converted to cash these days...).
This is related to the paradox of high standards: once you set your standards too high, the only people who "meet" them are the worst cheaters.
It's hardly a perfect statistic, since there are still many assumptions involved.
I think you're understating the imperfection(s), notably the assumption of the constant covariance matrix over the forecast horizon. Under dramatic increases in variance, the correlations tend to drift toward 1.
From what I can tell the Austrian school has just taken the conclusions of the old school liberal economists, dispensed with the basic requirements and pushed it to its absurd extreme. The only thing they keep saying is that when things go bad, that's because the gov't has intervened too much. It's like religion; something bad happened to you? Didn't pray to $deity enough. You prayed 23h a day? Well what were you wasting an hour sleeping for anyway.
As for the information thing, again, what I understand it to mean is that there is a reasonable amount information, the more the better, obviously, but more specifically not disinformation. Only "austrians" have that weird non-linear view whereby something either is 100% kosher or it's Soviet Russia. In the real world there are plenty of shades of grey, plus there's two color components.
For example the "austrian" critic of FDR and the Great Depression is laughable. Supposedly it happened because of too much interventionism. Well the US 1929 is about as non-interventionist as you can get and see what it did. By contrast Western Europe in the past 50 years is very much Soviet Russia in comparison, and if those fascistoid fucktards were right, we'd be living in slums eating dirt.
likens VaR to 'an air bag that works all the time, except when you have a car accident.'"
Or an airbag that works in an accident but takes your head as it does.
Nasty!
Invest in Diseased Livestock
Never shake hands with a man you meet in a fertility clinic.
The crash in a nutshell: Moral hazard. The US Federal Government created a moral hazard with guarantees against private risk. There's no downside if the government won't let you fail, so why be safe and when you can be daring? And so financial markets made overly risky investments. Worthless securities became valuable commodities.
This isn't a failure of capitalism, but a failure of government tinkering. The common claim that Bush/Greenspan/etc were laissez faire is a laughable assertion. Fannie, Freddie and the Fed are quasi-government institutions, and NOT private organizations operating in a free market. The market is not perfect by any stretch of the imagination, but that does not imply that government fiddling about can do better.
When faced with collapsing financial markets, the question to ask is not what government can do, but what the government can undo.
Don't blame me, I didn't vote for either of them!
This is a more general problem with scoring in general. Instructors deal with this issue often -- how do you reduce students' capabilities to a single quantized grade, and fairly? It's not possible...
Any score is the result of a set of variables (think of it as a vector in a vector space). Scoring effectively takes the magnitude of the vector. Now you can play various games by "warping" the space, adding weighing coefficients and whatnot, but in the end there is a huge amount of information loss, and as Taleb mentions, there can always be a black swan, where you discover yet another dimension exists in your space, only the sampled data happened to be correlated for a little bit.
When launching an attack against someone in your sig, be sure you get their UID correct. Especially when IT IS ONLY TWO FUCKING NUMBERS.
The dangers of knowledge trigger emotional distress in human beings.
Stuff VaR, I find it hard enough to get any clients to create and maintain a Vision Document (RUP - or any other similar document from another framework) for their project. This ha always proved to be the biggest risk.
Then there are those stunning Risk Registers with the Likelihood and Impact factors assigned during the one and only "risk workshop", that is akin to decision by committee, that are never looked at again. Does anyone know who I apply to for those hours of my life spent in Risk Workshops back.
My $0.02 is spent.
The weathers here - Wish you were beautiful
News flash: You get what you measure. Be sure you understand what you're measuring.
There's no problem with reducing risk to a single number. You just have to be aware of what that number represents -- it's not the absolute chance of risk in all scenarios, it's the predicted risk assuming everything works according to your model.
The problem here has nothing in particular to do with risk analysis, it's the same problem people have with statistics and measurements in general -- they aren't willing to take the time to understand what the numbers mean.
When the "facts" are designed, doctored, fudged, or simply ignored to allow unbridled greed and self interest to run it's course, one can look at the misdiagnosis of the facts, but in the end, all you have to do is look at the primary relationship between the ruling motivation and the behavior in question. None of what's happened is surprising, and it's happened before. We allowed people to make crazy bets and called it investment during the 1920s, and got the great depression. We allowed people to do precisely the same thing in 1980s through 2008, and suddenly we act like we've never seen this beast before.
Surely there is nothing new under the sun!... at least in area of primate behavior...
I've known that this crash was coming for over 5 years, predicted it for this year 5 years ago and for this Sept last winter.
For 3 years, everyone I've spoken with about the economy has said we were in a recession.
Despite all this, our financial and economic "experts" are claiming that they were "taken by surprise."
So we gave them more phony money and sent them to fix it!?
The whole idea that lending money to people who lend money to people who lend money is somehow "increasing the money supply," is faulty at it's roots. Each layer in this creates another layer of risk, so the systemic risk is a function of how much money is lent how deep.
Add to this the fact that finance has become JIT (just in time) as manufacturing did 30 years ago.
JIT works fine so long as the system is stable, but a bottleneck can choke the system. In the 80's it became impossible to import chromium and within a few weeks there was no stainless steel to be had in the country.
Since modern companies operate on borrowed money, dependent upon their ability to borrow continuously, a tightening of credit sends shock ripples through the system--which reinforce each other.
Additionally, our economy world-wide is now essentially a single system, all of the elections and economic decisions are made on the same time-table, which also reinforces any abnormalities.
Add to this the always handy criminals exploiting the system--and their gullible clients (too good to be true usually is.) And the fact that all major currencies are based only upon the faith of those accepting them, and the system is unstable at best.
Note that our response, world-wide, has been to inject a substantial percentage of the value of currency in circulation into the economy without any corresponding increase in value.
This is a classic way to inflate money.
There were good reasons that the US Constitution required all US currency to be made of either gold or silver.
Since all of the basic materials for our life are in commodities markets--which can be manipulated with only a small percentage of the value of the materials, once these markets become controlled by speculators and financiers rather than producers and consumers of the materials, prices become disconnected from the real value of the materials. No one who looked at oil prices for the past couple years, and who recognized the recession, could possibly believe that the oil was worth what the prices in the market.
This is not a new problem, (check out the tulip crash,) but it is subject to analysis and repair--if we want to repair it.
There are many people who do not want this fixed as it is much easier to make money exploiting problems within the system.
In this way it is exactly like the political systems. There are some major flaws in all of our political systems which are retained because they permit the people in charge to stay there.
It is interesting that: 1) the US gov't has no official definition for "depression." 2) All official and most unofficial pronouncements have been exceedingly reluctant to admit to "recession"--I've only heard one economist us the D word and that was a couple weeks ago--I expect that this will surpass the 1930's in depth and effect by a huge margin.
No entity, individual, organization or country can spend more value than they have or can create indefinitely. The usual plan of politicians when they do such "borrowing" is to ensure that the effects come to fruit in the NEXT administration.
Our mortgage problems now are a direct result of not repairing the entire system when the S&L crash occurred (in large part due to government auditors saying that the S&L's need to make more aggressive loans....)
The US went from having the lowest savings rate to having a negative savings rate. This effectively means that all those toys actually belonged to the banks (who do NOT want them!)
Since the banks don't have any use but to sell them, and the people who would buy h
Those who trusted their bankers to handle what they didn't understand sure feel better now, don't they?
I am amazed there are people out there still trying to rationalize the pretty obvious by fancy means.
All this talk about risk is utter nonsense, as long as it is considered sane to lend to people that can't possibly pay back all your fancy wording is worth squat.
I have seen billions thrown at banks and car companies. Where is the legislation ensuring that people lacking any creditworthiness will not get a loan?
IANAL but write like a drunk one.
VaR is meaningless. As the article pointed out, all it says is that you won't lose more than X 99% (or some other percentage) of the time. VaR tells you nothing about the performance of your portfolio in really bad conditions (that 1%). Because of this, it actually encourages traders to take risky bets where most of the time the trade does fine and 1% of the time it is bad enough to wipe out the whole company. VaR only gives you the threshold. It says nothing of what happens once you cross the threshold. The "advantage" (curse) of VaR is that it is intuitive and easy to understand. Of course you cannot summarize the entire balance sheet of a company like Citigroup in a single number that is so easy to understand. There are certain risk measures that are more complicated to define, but give some description of the tail distribution (i.e., give some idea of how bad things will get in that 1%).
These are so-called Coherent Risk Measures and Convex Risk Measures. They encourage diversification (rather than discouraging it as VaR does) and penalize taking on huge amounts of risk for minimal profits. They are not perfect either--ideally one would want to have a sense of the distribution (which is a whole function), rather than just a single number--but they are far better than VaR. VaR is encouraged by regulators and is accepted by banks b/c it allows them to take big risks and make lots of money most of the time. Using better risk measures would go a long way to cleaning up the global financial system.