This page has been archived and commenting is disabled.
What The Oil Spill Can Teach About Financial Catastrophe Probability Forecasts
John Hussman once again provides a very insightful and glorious in its simplicity argument about why the "imminent oblivion" that was facing all of Western civilization is nothing but an obfuscation straw man meant to preserve the bondholder and equityholder interests in the insolvent financial firms, better known these days as the TBTF: "The only reason that bank "failures" in the Depression (and the
"failure" of Lehman) were problematic is that the institutions had to
be liquidated in a disorganized, piecemeal fashion, because there was
no receivership and resolution authority that could cut away the
operating entity and sell it as a "whole bank" entity ex-bondholder and
-stockholder liabilities. I put "failure" in quotations because there
is a tendency to think of such events as something to be avoided even
at the cost of public funds. Failure only means that bondholders don't
get 100 cents on the dollar. As I've repeatedly emphasized (and don't
believe can be emphasized enough), it is essential to invoke the word
"restructuring" wherever possible, because it immediately leads us to
seek constructive solutions between borrowers and lenders, without
public expenditure." In other words, as anyone who has ever looked at a Plan of Reorganization, liquidation does not equal restructuring. As the Lehman bankruptcy showed, there are perfectly salvageable pieces of any investment bank operation that can be promptly integrated into a different business (i.e. the Lehman North American Brokerage business that was acquired by Barclays). This is a topic we have also emphasized from day one - the Administration makes it seem like a bank failure immediately implies liquidation - this is not the case, and this is precisely what FinReg should have focused on. It did not. Yet it will have no choice but to do so, once a new and much larger crash occurs. As for the odds of that happening, Hussman has some other brilliant insights into the probabilities of "worst case scenarios" occurring, in an analysis driven by his observations on the GoM oil spill catastrophe.
"Is this something that could happen once in a million times? Is it something that could happen once in a thousand times, or once every 5,000 times? What exactly are the risks involved?"
President Barack Obama 5/17/10, addressing the Gulf oil spill
As we observe the recent oil spill in the Gulf of Mexico, the recent banking crisis, and the ongoing concerns about sovereign debt in Europe, one of the things that strikes me is that few analysts are much good at assessing probabilities for worst case scenarios.
We typically refer to the probability of some event Y as P(Y), and write the probability of Y, given some information X, as P(Y|X). So for example, the probability of a vehicle being a school bus might be only 1%, but given some extra information, like "the vehicle is yellow and full of children," the estimated "conditional" probability would go up enormously.
With regard to oil spills, however low one might have believed P( we'll have an oil spill ) to be, prior to the recent accident, the "prior" probability estimate should change given that we've now observed one of the worst oil spills in history. Even if the oil industry previously argued that the probability of an oil spill was one in a million, it's hard to hold onto that assessment after the oil spill occurs, unless your faith in the soundness of the technology is entirely unmoved in the face of new information.
See, if P( the technology is flawed | we had an oil spill) is 80%, and P( we'll have another oil spill | the technology is flawed ) is 80%, then regardless of how extremely unlikely you thought oil spills were before we observed one, or how unlikely you thought it was that the technology was flawed, you would now estimate P( we'll have another oil spill | we had an oil spill) at no less than 80% x 80% = 64%*.
While there are about 3800 oil platforms in the Gulf of Mexico, only about 130 deep water projects have been completed, compared with just 17 a decade ago. So in 10 years, applying a new technology, we've had one major oil spill thus far. Unless there is some a priori reason to assume that the technology is pristine, despite the fact that it has failed spectacularly, the first back-of-the-envelope estimate a statistician would make would be to model deep water oil spills as a "Poisson process." Poisson processes are often used to model things that arrive randomly, like customers in a checkout line, or insurance claims across unrelated policy holders. Given one major oil spill in 10 years, you probably wouldn't be way off the mark using an average "arrival frequency" of 0.10 annually.
From that perspective, a simple Poisson estimate would suggest a 90.5% probability that we will see no additional major oil spills from deep water rigs over the coming year, dropping to a 36.8% chance that we'll see no additional major oil spills from deep water rigs over the coming decade. Moreover, you'd put a 36.8% chance on having exactly one more major spill in the coming decade, an 18.4% chance on having two major spills, a 6.1% chance of having three major spills, and a 1.9% chance of having four or more major spills in the coming decade. This is quite a bit of inference from a small amount of data, but catastrophes contain a great deal of information when the "prior" is that catastrophes are simply not possible.
Given that the worst offshore oil spill in Australia's history happened only in November 2009 (which took months to shut down), this sort of estimate does not seem unreasonable. In any event, disasters contain information. It's no longer reasonable to apply previous risk estimates even after we've observed a major spill.
Similarly, before the housing crisis, it might have been tempting to shrug off mortgage defaults as relatively isolated events, since the price of housing had generally experienced a long upward trend over time. Indeed, historically, sustained declines in home prices could be shown to be very low probability events. But as the bubble continued, investors made little attempt to assess the probability of a debt crisis given that home prices had become detached from all reasonable metrics of income and ability to pay. Just as buy-and-hold investors assumed that the long-term return on stocks was constant at about 10%, despite the late 1990's valuation bubble, investors during the housing bubble kept looking at the "unconditional probability" P( credit crisis ) based on decades of normal housing valuations, when they should have recognized that the conditional probability P( credit crisis | extreme housing overvaluation and lax credit standards ) was probably higher. This turned out to be a profound oversight.
But that was evidently not a sufficient lesson. As soon as the surface appearance of the problem was covered up by an expensive and opaque band-aid of government bailouts and suspension of accounting transparency by the FASB, investors went right back to using those unconditional probability estimates. Indeed, until the spike in credit spreads that began a few weeks ago, the amount of additional yield investors demanded for taking credit risk had fallen back to the lows of 2007. We've had a major credit crisis, we have failed to restructure the debt underlying that crisis, and yet investors are approaching the market as if the debt has simply been made whole and we can continue along the former path.
Knowing that the cash flows from mortgage payments cannot possibly be adequate to service the original debt, that delinquencies continue to hit new records, and that there is an enormous overhang of nonperforming debt and unforeclosed homes - it seems utterly naive to assume that the problems we saw over a year ago have been adequately addressed.
What holds for oil also holds for red ink. Disasters contain information. It's no longer reasonable to apply previous risk estimates even after we've observed a major spill.
»
- 6028 reads
- Printer-friendly version
- Send to friend
- advertisements -
Forecasts that occur outside the range of "public credibility" or "political agenda" don't get taken seriously.
Much of the world we are living in would not have make a creditable forecast prior to Obama's election.
Standard statistics is based upon the naturally occuring behavior of inatimate, unbiased objects.
Projections of events that are predicated on human behavior are not unbiased and cannot be reliably projected using standard statistics.
and more important, did the actuaries predict Al and Tipper?
Would it be too corny if I said BP's oil spill was a "Black Swan?"
It was a "Black Pelican".
http://naturescrusaders.files.wordpress.com/2010/02/oilsoakedbird.jpeg
Unfortunately corny, but true. Nassim Taleb's 'black swan' neologism pertains to exactly this type of risk, that is, with very low probability but very high potential damage should the event in question occur. This type of risk is the hardest for humans to deal with rationally.
A classic example is the risk of dying in an airplane crash versus an auto crash. This risk is much higher in an auto, but few are afraid of making a car trip, while fear of flying is fairly common.
"See, if P( the technology is flawed | we had an oil spill) is 80%, and P( we'll have another oil spill | the technology is flawed ) is 80%, then regardless of how extremely unlikely you thought oil spills were before we observed one, or how unlikely you thought it was that the technology was flawed, you would now estimate P( we'll have another oil spill | we had an oil spill) at no less than 80% x 80% = 64%*."
That's the gambler's fallacy. Past events do not influence future probability. The chance would remain 80%.
If I just flipped a coin and it came up heads, flipping it again would not have a probablity of .25 that it would come up heads again.
Sheesh.
No, you're wrong. Hussman's reasoning is correct, but you wouldn't understand it unless you have a background in Probability Theory.
No I think he's right, unless you think future catastrophes would be correlated with this one. Probabilty doesn't "self-correct" for the outlying point.
It's like the old joke about how to avoid being blown up on a plane - carry a bomb yourself. After all the chances of a bomb on a plane are (say) one-in-a-million.... But the chances of there being *two* bombs....
Instead of simply asserting that I am wrong, how about saying how I am wrong, instead of grasping for some limp-dicked appeal to authority?
You should go back to trading dudes.
Actually, upon reading (rather than skimming) the article, it makes more sense since he is addressing the probability that "the technology" is itself flawed, not the probability of individual events.
So sue me.
Hussman states P(a, tech flawed >> causes oil spill) at 80%, then states P(b, another oil spill << tech flawed). These are the same statement. His derivation is bogus. And bogus is a technical term in the field of "that doesn't pass the stink test."
Why the heck are we playing with terms like "accident" and concepts of probability? Does *anyone* even want to smell "What the Queen has cooking!"
As I said earlier, what Hussman wrote makes perfect sense. It's a brilliant piece in my opinion.
No offense, but I don't think you would understand the math and therefore his reasoning unless you've studied Probability Theory.
http://en.wikipedia.org/wiki/Conditional_probability
Thanks, LDT, though I wouldn't let them off the hook so quickly. Any familiarity with structured thinking (in any discipline) should be sufficient to grasp what Hussman is stating. It's one of the reasons he is a must-read on a weekly basis: clear, accessible, insightful analysis with just a dash of wit and wonkiness. Though the investing style he employs for his two funds is based on a longer time horizon than mine, his assessment of the macro conditions and their remedy is spot-on.
Plus, his use of TA and options in a mutual fund setting is masterful. A person can learn a great deal by reading his weekly comments going back to 2003.
From that very Wikipedia link....
In other words, if A and B are independent, then the conditional probability of A, given B is simply the individual probability of A alone; likewise, the probability of B given A is simply the probability of B alone.
Check it.
To the *general* above: independent not dependent.
To the Hussman bogusness: redundant instances and independent.
Two errors in the same paragraph. I will revise my thought above to this:
Hussman has an incorrect understanding of a basic concept.
Coin flip one: 50/50
Coin flip two: 50/50
Coin flip N: 50/50
Independent!
Oil f-up one: 80%
Oil f-up two: 80%
Oil f-up N: 80%
Independent!
Now if A must happen to cause B to be activated then we have dependence, but that is not what is stated as going on here.
Back to hammer and nail, I see.
You are absolutely correct about the coin flip. Each flip is independent, and the independent probability is 50%. As you mention, the gambler's fallacy is thinking that after a heads (or a string of heads) that the coin is "due" for a tails, but each flip is independent of the other. We'll leave out the probability of streaks for simplicity.
However, Hussman clearly states in his preamble that he's positing conditional probabilities, not independent ones. Before the spill, BP might have said that the chance of a spill was 0.05%. OK, sure, whatever. We agree that your super cool ultra-deep drilling technology works because we have no evidence to the contrary prior to the spill (in fact, we have evidence that proves your "deep drilling technology" works because we're female MMS employees.)
But now that we know that there has been a spill, things start to get interesting. Given that there has been a spill, we know that it's likely that the drilling technology has flaws. In other words, P(BadTech|OilSpill) is pretty high (we'll use 80% per Hussman's swag.) We'll chalk the 20% chance that the tech isn't bad up to rumors about N. Korean torpedoes.
Independently, if we were 100% certain that the technology is bad, then there's a good chance there is going to be a new spill, or P(NewSpill|BadTech) is pretty high. (Confusingly, Hussman also swags 80% for this.) We'll chalk that 20% no-spill chance up to Sarah Palin's nighttime prayers. (Note that I have no knowledge of the probability that she first heard "Drill, Baby, Drill" while working as a female employee of MMS.)
It's easy to miss that P(BadTech|OilSpill) <> P(NewSpill|BadTech), especially since Hussman uses the same conditional probability value (80%) for both.
So an 80% chance of the tech being bad * 80% chance of a new spill if the tech IS bad = 64% chance of a new spill. Total swag and not meant to be an accurate probability of the next spill, but the point he's making is that given what we now know based on the existence of a spill, we can calculate probabilities that are MUCH higher than the wild-ass lowballs calculated before the accident and repeated by Obama. Similarly, fools today are pricing the stock market as if Sept 2008 had never occurred, instead of realizing that it provided valuable information about the conditional outcomes given our unresolved unhealthy economic conditions.
"...if A and B are independent..."
P(your having read Hussman's piece|using that quote) = 0%
Given coin flip one was a head. What is the probability that coin flip two is a head? Is coin flip two dependent on flip one?
Given that oil spill one occurred. Under exactly the same conditions what is the probability that oil spill two will occur? Is spill two dependent on one?
If all you have is a hammer, everything looks like a nail.
we are far more powerful than ever before, our screw ups cause more damage. if i understand correctly, the ultra deep drilling is a relatively new technology. are we sure the pressure changes down there cause side effects? haiti for example.
The probability of things not going as originally planned is almost always 100%. Most often the unplanned deviation is minor but "Murphy's Law" is prevalent worldwide. By nature, humans tend to be too optimistic about expect results.
If you assume generally that the two disasters Katrina and this one are not too distantly related, and the response is what it is, we are pretty bad off. Whether it is govt or private industry, there is a servere lack of Plan B stuff. As in worst case scenario planning. The info in these events and the past 30 months is not heartening. There is very little room for error as TBTF - wellll, yes. If it is too big, too deep, too anything, absent a Plan B is just unrealistic.
And for the Euro, better minds than mine predict 80% chance of collapse. I see Hussman is about there too. What are we ready for in the next 30 months??
The rapidity of novel catasrophic events is quickening.
"What are we ready for in the next 30 months??
The rapidity of novel catastrophic events is quick"
I'm so sick to my stomach, I can't even eat the popcorn. Must have a cog in my dissonance.
So what is the probability that some number this week will have the President's economic team use the term :Unexpectedly?
Just interested.
the probabilty of one in a thousand even reading the whole probabity thing is 1 in million,
so any event will be on the 10 oclock news
It seems to me that the essential problem is not whether different realizations of P() are independent, but the probability itself. This quantity is probably unknowable, given the small number of data points (actual oil spills). One might also say that the number of offshore wells where there has been no oil spill is relevant, which might be true if all these other wells were the same type, depth, crew experience, fail-safe measures, etc., but we know that this well was extraordinarily deep, so it's almost one of a kind.
The problem with such an endeavor is that the anti-drilling crowd can estimate the chance of a problem at 1 in 10, while the pro-drilling bunch can estimate it at 1 in whatever, and there's no way of knowing who's closer to the truth. In any case, however small the probability you give me, if the potential damage is high enough (can we estimate an upper bound on this spill yet?), then you have to decide that the risk is not worth taking.
Even if you have 1 million data points, you can't be sure to predict whether it will happen or not. Output and reality only temporarily meet, sometimes. In the end it either happens, or doesn't. You're either pregnant or not. When or how it occurs is never predicted by the output, it just either meets it this time, or doesn't as conincidence.
The probability model is merely a fictional representation in numbers of a dynamic process occuring inside a vastly more complex universe using human derived and limited variables in a foolish attempt to explain concretely, succinctly, accurately, and actionably something we want measured about the world around us for whatever reason. Again it does this through subsitution of what is with numbers that supposedly represent every possible variable affecting what is being measured. Again it substitutes REALITY with numbers.
Ding Ding Ding. We have a winner. The p scores (and whole statistical field) we abuse by treating them as stone cold facts will bite us in many more ways. Statistical outputs are guesses. Trying to use them as constants or even relying on them to be an accurate indicator of reality for another purpose (one example: variable in an equation) is extreme folly. So let's go Usain Bolt in attaching most things we depend on in our lives to this ?ideology?. Or laziness to have an excuse to pass the buck when it doesn't work. Who cares if I suspected it to be wrong, it said it wasn't.
Now apply this to how they are approaching
a) IMAC/IPAB in the deathcare commission via our gift to the insurance companies officially known as healthcare 'reform'. (note: I want real single payer)
1. Specifically - the use of QALY (Quality Adjusted Life Years), Comparative Effectiveness research, etc, etc. (basically the flaws in the process that determines what medical procedure OR strategy is worthy or not to be repaid..or time used...thus what is eventually practiced or abandoned, or in limited fashion absorbed by lower margins for a hospital/general practitioner).
It's also used to predict the effectiveness of things like mammograms, or self checking your nuts or breasts or other areas for lumps.
They are basically searching for something to show them what to cut (and not caring about whether or not it's a reflection of reality), and whatever these flawed studies, or outputs, p scores, z scores, whatever point to, will be targeted for cuts...by experts who just read a flawed output and acquiesced to it's presumed authority.
Anyone see a potential problem?
Nobody wants these aspects of healthcare 'reform'.
b) Global Warming
Look at how they use it to predict polar bear demise, shrinking ice caps, co2 levels, future warming by x year, who/what is causing it, etc, etc.
Because these things will happen, we must be causing it, and we must do whatever it takes to stop them from happening. Whoops.
There have always been ice caps, sea levels can rise forever, people can't be relocated over the course of centuries, co2 levels once they start rising will always and forever keep rising, no co2 ever settles or reacts with other elements, nor escapes into space, we can never figure out a way to bring co2 out of the atmosphere, the seafloor does not ever soak up co2, polar bears can't adapt and become regular northern bears living on land (nor the realization that dead ends occur in evolution...while the polar bear might be the mastodon, the saber tooth, the neanderthal of the bear species, we must attempt to save them and everything else by...destroying our economy, human (or inhumane) depopulation, and accept a lower standard of living.
The sun isn't a factor, it's cycles, it's potential supercycles, etc.
Nor is the absolute zero temperature at space basically meaning in computer speak, our earth is surrounded by the mother of all liquid nitrogen heatsinks called open and empty space.
But these models have everything perfect, they account for everything. Just like AIG's. Just like BP's. Except instead of economic or environmental disaster, this will be a complete societal and living standard disaster from our response which is completely our own doing by outguessing ourselves.
c)Derivatives
AIG did great with their models saying they'd never have to pay off the insurance they were peddling to a degree that would systemically harm AIG, let alone account for it's contagion.
D)cap and trade
It's a new derivatives scheme, built off the results poorly interpreted from part b (they can never be wrong), so it's a new thing that's a combo of b & c. We can only emit x amount of carbon, based on these metrics of water temperature, co2 levels, surface temperature, etc, etc.
I left out all the data fraud too fwiw. Even though it IS data fraud.
What are the odds of being wrong when data fraud is thrown inside of a faulty process, which is then used as the basis and bedrock of our supposed 'green future' and climate solutions?
Forget about the Sun's impact, or that Earth didn't have a stable temperature before the arrival of modern humans, or even that we have an incomplete understanding of the total dynamics of Earth's heatsink (the way it deals with heat and the absence of it, to balance itself out via mechanisms in nature).
But since this is all about quantification, we pretend to think we have it all down for certain, and that we have intelligently considered all pertinent factors that could throw off our results and have devised a way through cap and trade to set up an inflationary system, resulting in lower consumption through the cap and trade system to alter what we currently peg as the dynamics of Earth's surface/water temperature we consider to be our climate to correct it to an ideal we have figured out (temerature, co2 levels, air composition, water temp, acidity, sea level, glacier/polar cap coverage and have metrics for. (yeah right) We have forgotten nothing, and everything we accounted for, is accounted for perfectly.
Look around. We'll find these same sorts of analyses we've 'depended on' to be quite wrong in just about every way it was applied. Our house via this point is built on quicksand, which just happens to be located on Santorini/story of Atlantis/etc.
The people in power and those that oppose them, want to increase our reliance on this sham-wow. That's what we would consider dems and repubs and even independents to be fighting for in one way or the other.
Don't believe the folly. Don't implement these things in lieu of due diligence. Don't outsource thinking to a p score, a regression model, etc, etc. Having a wrong number to point out when the crap hits the fan is not an answer. You can't CYA with this bunk stuff. Yet some people believe in this crap like it's a religion, more kooky than a mythical mormislam. We can point to it like a straw man, saying 'well this is what it told us', it's not our fault that we should have known better.
Great article ZH as it brings up an important point about how we misuse probability.