Saturday, October 13, 2007

ust another Regression toward the Mean

The United States consumes nearly 25% of the world’s liquid petroleum resources (“oil”) yet has only 4.5 % (approximately) of the world’s population. Over the next decade or so, the U.S. will consume somewhere in the neighborhood of 10% of the world’s oil, and within 20 years, 5% of the world’s oil. Worse, the gross production of oil will also decline during the period. How can I be so sure? Because a Regression toward the Mean is a certainty, the only area up for debate is the rate of regression.

First, a definition from

“If random variables X and Y have standard deviations SX and SY and correlation coefficient ρ the slope of the regression line is given by

p x SY/SX

A consequence of this is that a change of 1 standard deviation in X is associated with a change of ρ SDs in Y. Unless X and Y are perfectly correlated ρ will be less than unity. Thus, for a given value of X the value of Y that would be predicted by the regression line is always fewer SDs from its mean than X is from its mean. Regression to the mean will occur if | ρ | < 1, so in practice it always occurs.

Note that regression toward the mean is more pronounced the less the two variables are correlated, i.e. the smaller | ρ | is.”

What are the 2 variables? U.S. oil consumption, and non-U.S. oil consumption. You disagree? You have a better chance of making an argument that pi is not 3.1415… and e is not e = 2.7182…

The funny thing, and I do this for a living, is trying to convince someone that is “certain” this is not so but who could not define pi or e – yet still has strong convictions, based on not one shred of research or hard data, on energy supplies. The late Malcolm Forbes once said that “being rich does not make a man right more often, just harder to correct”.

I received an email recently from a fellow Wall Streeter who said he did not subscribe to the “doom and gloom” of my analysis. I wrote him back:

“It matters little what you or I subscribe to. What
does matter is being correct and then placing your bets

The very words "doom and gloom" are put out there to
belittle other viewpoints of resource capacity. For
every seller there is a buyer (or we would have no
market) and only one of them is correct in their assessment of future

Further, "Doom and Gloom" cannot be measured. Oil
imports, BTU's consumed, bushels of wheat and corn -
these can all be counted and measured. If the answer on the right
side of the equals sign ( = ) translates into what
ever you call “doom and gloom” - well, the data does not
really care about your semantics...”

Correct me if I am wrong, but an overweight, hypertensive patient who dismisseshis physician’s conclusions as “doom and gloom” is a fool, no?

Let’s get back to the issue:

In other words, over the next 20 years, the U.S. will go from consuming over 20.3 million barrels per day (“bpd”), every day, to consuming less than 5 million bpd, every day. No more casual drives through the country side, no more silly car commercials, no more heating more than 1000 square feet of living space and then, only if you are very wealthy and only if the political realities are accepting of such.

A great shock is going to be felt in the markets. All of the pension liabilities, current valuations and multiples, insurance tables, medicare, social security, etc… EVERYTHING will need to be “reassessed”. A nice way of saying most financial assets will lose most of their value.

I speak only in constant U.S. dollars. I have no opinion on what reaction the U.S. Government, and its Department of Treasury and Federal Reserve Bank, might have. After all, Oil has not risen much, if at all, in price for Canadian consumers as their currency is not being devalued as is the U.S. dollar is.

Oh, and by the way. The EIA’s data for July world oil production is out. Production of crude and condensate continues to decline, India and China’s imports continue to rise, and U.S. imports continue to fall… but more on this in a future post.

Yours for a better world,

Mentatt (at) yahoo (dot) com

No comments: