Saturday, October 18, 2008

Fools, or Knaves?

"But if regulators were oblivious to the danger, the question is why. In the case of Fannie Mae and Freddie Mac, the answer seems easy: Their massive lobbying machines thwarted every legislative attempt at reform. But what about the Fed, the Treasury and the [SEC], agencies that are not above politics but are known for their professionalism and expertise? Surely they had the capability and motivation to avoid a calamity of the type that is occurring. Why did they fail? The problem is that Wall Street and regulators relied on complex mathematical models that told financial institutions how much risk they were taking at any given time. The model predicts with 99 percent probability that institutions cannot lose more than a certain amount of money. ... As long as capital was greater than the value at risk, institutions were considered sound--and there was no need for hands-on regulation. Lurking behind the models, however, was a colossal conceptual error, the belief that risk is randomly distributed and that each event has no bearing on the next event in a sequence. ... What if risk is not shaped like a bell curve? What if new events are profoundly affected by what went before? The value-at-risk crowd focuses on each snowflake and resulting cause and effect. The complexity theorist studies the mountain. The arrangement of snow flakes is a good example of a highly complex set of interdependent relationships, so complex it is impossible to model. ... But compex systems are not confined to historical experience. Events of any size are possible, and limited only by the scale of the system itself. ... Financial systems overall have emergent properties that are not conspicuous in their individual components and that traditionall risk mangement does not account for", my emphasis, James Rickards (JR) at the Houston Chronicle, 3 October 2008.

It should be a felony for any financial institution holding federally insured deposits to lobby. Punishment: five years in the federal pen for the executives responsible and the lobbyist. Not up to five years, five years. The institution pays a fine of 1% of its last reported stockholders' equity as reported in Form 10-K, or if not publicly held, whatever it files with the OCC, for a first offense, 2% for a second offense, 4% for a third offense, etc., etc.. The SEC is known for "professionalism and expertise", by whom? The agencies JR mentions are hopelessly compromised. What makes JR think they believed VAR models were anything more than after the fact rationalizations for decreasing the amount of capital the institutions needed to operate?

JR confuses risk with uncertainty. I'll explain. A fair coin comes up heads about 50% of the time, actually say 49.99%, because at times, it stands on end, the coin having three dimensions. The coin is not a Flatland, 1884, Edwin Abbott character. There is no risk here, but the result of each coin toss in uncertain. Where a hurricane lands, a Houstonian matter of interest, is risky as you can't model everything with respect to hurricane paths.

We should not get carried away with the normal distribution. Even things we think are normally distributed may not be. Leta Hollingsworth (LH) wrote Children Above 180 IQ, 1942. Working with New York City (NYC) school children, this is from my memory, I read the book about 40 years ago, she identified 24 NYC school children who qualified. LH believed she found 80% of the children who would have qualified. Her result suprised her. Why? With a SD of 16.4 and mean of 100, 180 IQ is +4.88 SD. Such an observation should be seen in about one in a million cases. With 900,000 NYC school children assuming LH was correct, 180 IQ was at the 1/30,000 level, or 33X as frequent as a normal distribution would indicate. Fat tails are real.

Now for an IA war story. One of my last Chicago classes was business policy. The professor had been a large and prestigious consulting firm's senior vice-president before joining Chicago's faculty. He had two engineering degrees from a prestigious engineering school. The day's topic was oil industry rates of return (ROR). Many of his oil company clients as a consultant had complained about oil drilling's poor ROR, for a very "risky" activity. I thought for a minute and said, "the market doesn't reward risk you can diversify out of. Therefore your oil company clients didn't know what they were talking about". "Why is that IA?" "Because of a 'fallacy of composition', what is true for one thing may not be true for all things of its class". "Go on". "Suppose Big Oil drills 10,000 holes a year. While it is true any individual hole may be dry and 90% are, that means nothing for oil industry rates of return". "Why is that"? "Because oil holes are like coin flips. If Big Oil drills 10,000 holes, it knows beforehand that 900-1,100 will be productive. Since it drills lots of holes, where's the risk"? Then I went up to the blackboard and drew some diagrams of ROR for one well, two wells, and say 10,000 wells. The professor became livid and got red in the face. My whole class looked at me, horrified. The discussion ended. Why was this a hot topic? It was late 1973 and we had just seen the "oil shock".


Anonymous said...

The interesting thing about the oil shock of 73 is what proceeded it which was war followed by Nixon taking the ability of forgein holders of dollars to exchange those dollars for gold on 8/17/71.

The bubble in housing was the result of inflation due to currency devaluation from government spending which spilled into virtually all assert class, IMHOP the last bubble which is still inflating is in US treasuries.

Blissex said...

«Fat tails are real»

Obligatory reminder here of Mandelbrot's "The misbehavior of markets" and Taleb's "Fooled by randonmness".

But also that it is not just the distribution of events that matter, but also their correlation.

The rating agencies and the modelers for tier 3 credits were paid very very well to estimate not only that risk distributions were gaussian, but also that they were uncorrelated (or even anticorrelated), because generalized market losses and liquidity failed and mass bankruptcies had never occurred before :-).

So we had pools of debentures with a much lower risk rating than any of them.

«If Big Oil drills 10,000 holes, it knows beforehand that 900-1,100 will be productive.»

Generally yes, but this should also be qualified with "disregarding model risk and correlation in drilling sites".

Because in the medium-long term drilling sites are not chosen randomly:

* They are chosen according to specific models (intentionally).

* The easier areas are drilled before the less easy areas.

The 10% success rate is more a heuristic than a statistic...

But your overall argument mostly stands.

Independent Accountant said...

I agree with you. The "oil shock" was largely the Arabs compensating for prior US inflation. I recollect American oil companies and Saudi Arabia established a 1947 oil price of $1.30 a barrel. As I see it, the 1973 "oil shock" was largely the Arabs reestablishing the 1947 price in real terms. As for US long-term bonds, I'm with you and Jesse. We all see US bonds as a bubble.

Yes, and that's why there is real risk in drilling for oil as opposed to just uncertainty. In 1973 about 10% of "wildcat" wells drilled were productive. Yes, the 10% figure is not analogous to 49.99% of coin tosses coming up heads for a fair coin. Every square foot of the earth is different. Every well is different. My 1973 point was: oil company executives' estimates of reasonable ROR from drilling wells were much higher than the market's. As far as I could see, the market was right. I suspect, "wildcat" wells success rate is higher today with 3D and 4D seismic.

printfaster said...

Your note about SA setting the 1947 oil price at $1.30/bbl means a gold to oil ratio of about 30.

Let's see, at $100/bbl, that means a price of gold at $3000. Hmmm wonder how the Gulf people like that. Man we are getting cheap oil and cheating the daylights out of the Saudis.
Even at $50/bbl that is really
cheap oil.

Buy more t-bonds suckers.

Independent Accountant said...

I see no reason for a fixed ratio between the prices of oil and gold. As Uncle Sam prints more $$$, both will go up.
Domestic oil was about $3 a barrel from 1958 to 1973, until the "oil shock" when it went to about $11. Given the inflation from 1947 to 1973, you would be hard pressed to convince me oil was higher in 1973 than 1947. The US got used to very cheap oil. Does anyone remember how big cars were in 1973 as well as how much gas they consumed?
I think the oil exporting states are making a mistake keeping their foreign exchange reserves in dollars. I would rather have oil in the ground than 30-year US bonds yielding 4.25%

printfaster said...


One more thing on the "bell shaped curve". Technically it is not bell shaped, but cap shaped. The name for it is Poisson distribution.

The reason is simple. IQ cannot go below zero. I suppose if you included the dead, you would inflate the number at zero. This then "fattens" the tail so to speak.

The same is true of financial instruments. You cannot go below zero in the portion of oil wells or the number of successful investments.

Whenever pure positive numbers are involved in distribution you have Poisson.

Oh yes, Spengler does his own treatment of risk:

By his measure, the "gold" of the worlds currency is US real property. Similar to the Weimar escape which monetized Reich property to clear the loss of currency confidence for the new Mark.

I would like your comment on the monetization of US RE.

printfaster said...

IA on the price of oil/gold. The middle easterners are notorious for seeing wealth measured in gold. That is the only reason that I suggest the ratio may have some meaning. Not to the rest of the world, but by those releasing it from the ground.

At some point, they may see that Ts are not gold, but merely a tease.

By the way, I have felt for a long time that Fort Knox was being secretly emptied to pay for this dollar/gold discrepancy directly under the table to the Saudis after RMN closed the window. I could easily see them demanding real specie.

Independent Accountant said...

I read this article when it first appeared. I did not quote it as it is one of the few things Spengler wrote that made no sense to me. I see no reason the US ratio of wealth to income cannot rise. It may mean we have "capital deepening".
The French issued real estate backed paper currency in the 1790s, i.e., assignats and mandats. Both currencies became worthless.
While IQ cannot go below 0, I don't see this as a big problem. With current tests, SD = 15, 0 is -6.67 SD. An observation seldom seen. You can have negative returns from drilling oil wells if you get stuck with environmental liabilities.
I have heard rumors for decades that our gold at Fort Knox and the NY Fed isn't there. I don't know. No one ever hired me to inventory it.