Monday, September 15, 2008

Lehman Brothers (1850-2008)

Lehman Brothers were established, under this name, in 1850 when Mayer Lehman, the youngest brother (the guy on the right side), joined his older brothers, Henry Lehman (the real founder) and Emanuel Lehman (the guy on the left side), and emigrated from Bavaria to Alabama. Their first businesses were based on using cotton as a currency.

The company has grown and survived many twists and turns, world wars, Great Depression, internal battles, and many bankruptcies of competitors in the financial ecosystem. However, it failed to survive the subprime-related crisis and today, after 158 years, it filed for bankruptcy, listing USD 613 billion of debts, making it the largest bankruptcy ever.

Well, such things happen. I think that there are too many momentum speculators, derivatives, and investments detached from the fundamentals. The risk of many types of investments - arising at various time scales - has not always been properly quantified. From this viewpoint, it's great that one company focusing on these things will evaporate, together with the employees who are also focusing on things different from the actual content of the economy and the products.

What is less great is an additional period of irrationality and volatility that we can expect. Many speculators will extrapolate various downward trends dramatically, causing additional problems to the world economy. I wonder why so many people are doing these irrational things and adding so much noise to the system. Cannot they try to invent their own realistic picture how the world should look like and what the prices should be, and simply sell overpriced things and buy the undervalued ones?

Right now it also seems rather likely - because of the excessive volatility we have seen - that futures traders and other speculators (and not supply and demand) have been the main players behind the dramatic fluctuations of the oil price during the last year. That's too bad because speculators should normally help to quantify the true value of things - like commodities - and stabilize the prices because they should know when the price is undervalued or overpriced. This idealized picture apparently didn't occur during the last year or so.

There are too many gadgets that try to extrapolate recent trends, whatever time scale they choose for the extrapolation. Such extrapolations inevitably lead to increasing leverage, growing bubbles, bursting bubbles, instabilities, and skyrocketing irrationality. There exist surprising additional sources of such instabilities. Other investment formats that superficially look like stabilizing effects actually act as destabilizing ones.

For example, there exist twin-win funds that allow investors to earn money both in the case when a price increases as well as in the case when it drops. For example, if the oil price increases by X%, you get 0.95 times X% from your initial investment after 5 years (aside from the money you have paid). If it drops by X%, you receive 0.5 times X%, another positive number! This sounds great but it is probably not hard to achieve. The expert who manages the fund simply buys oil for your money, but as soon as the price drops below the initial level, he shorts oil.

You might think that the existence of twin-win funds would stabilize the oil price because the fund manager is motivated to keep the oil price constant because when it is constant, he won't have to pay you much. ;-) However, when I thought about the situation twice, such a reasoning turned out to be flawed.

Whether someone is motivated or not is not important. What matters is the actual impact of the decisions he is led to make. The fund manager doesn't directly determine the oil price. You must look what he is actually doing to eliminate the risk and to get the money that the investors will demand (plus some profit). For the sake of transparency, I will be talking about the strategy of a fund manager who doesn't rely on others (option sellers) who would be parts of the system. In other words, my fund manager below plays the role of all the traders who are needed to make the fund work. The conclusions of my discussion will be universal; if you considered a more complicated strategy, involving option seller etc. (a topic intensely debated in the fast comments: do options influence oil price? LM: They do!) and you would include all of their decisions into your research of the oil price, the oil price would be affected pretty much in the same way. OK, so what does my fund manager do?

If the oil price exceeds the initial level, he must actually own the oil, so that he will be able to pay his investors if the oil price increases significantly. On the other hand, when the oil price drops below the initial level, he has to short oil.

This means that such managers are going to short oil just when it decreases below the initial level (and buy it if/when it returns above the initial level). If you think what it means, it brings instability to the system because buyers stabilize the price when they "rationally" buy the product when the price goes below their perceived "fair interval". And they sell it when it gets above their perceived "fair interval". This is the standard sign determining the relationship between the supply and demand that helps to stabilize prices. But the twin-win fund manager is doing just the opposite. He sells (or shorts) oil when it gets below the initial value and buys it when it gets above the initial value. ;-)

You can see that no one else in our example is buying or selling oil because of the existence of the twin-win fund. It follows that the twin-win fund magnifies market fluctuations. But honestly speaking, we shouldn't forget that the fund doesn't influence anything at time scales longer than 5 years because at the end, it sells all the oil that it bought and buys back all the oil that it shorted. ;-) But 5 years could be too long a time and the economies can be shattered earlier than that.

This kind of instability from similar financial tools is going to be rather generic. Does it mean that your humble correspondent is going to defend some kind of regulation? Well, I have mixed feelings about it. But yes, if the government wants to punish a certain kind of behavior of the market players, the behavior of those who destabilize the system - who are doing things that can be demonstrated to have a destabilizing effect - should be among the punished ones. Such things should be taken into account when various policies (e.g. tax policies) are being designed. I could tell you formulae for Lumo's friction taxes that would moderate hysterias. ;-)

In the ideal capitalist world as I imagine it, the fluctuations should be much lower.

Meanwhile, the bloody fate of Lehman Brothers could remind the greedy investors that their strategies based on the analysis of the momentum (the time derivatives of the prices) - and not the fundamentals (the absolute value of the prices) - could turn out to be just another form of lottery.

What do you think?

Bonus: Famous companies save the world from climate change

(Hat tip: Demesure)

There have been two great companies that passionately led the global efforts to save the world from climate change. We should follow their examples. The names of these two companies were:
  • Enron
  • Lehman Brothers
While Enron did everything it could to make the U.S. sign the Kyoto Protocol, the Lehman report "The business of climate change II" has been enthusiastically praised by the environmentalists, to use a polite word for the loons.

The two climate change reports represent 50% of the recent Intellectual Capital :) of Lehman brothers (PDF files below the last link). You can buy this capital for 0.3% of the price one year ago. James Hansen is a part of the package.

Will you join these two wise companies that can compare costs and benefits and quantify all the risks so well? ;-) Isn't it cool that these crooks are gone?

Al Gore and Lehman brothers

Yes, it's true. Check the National Post (Canada) where Richard Lindzen says the following about Al Gore:
... And he's on the on the board of Lehman Brothers who want to be the primary brokerage for emission permits. ...
Rae Ann has pointed out that the information is not quite accurate but it has some true core, see the fast comments. More details at IceCap.US and ClimateAudit.ORG.


  1. interesting point. If prices (or rather log prices to get rid of the currency and quantisation) were truly a Brownian process, then you are right, the price difference at time T+deltaT is independent of the time derivative.

    Unfortunately, prices are not Brownian walks. My reference for this is "Theory of Financial Risk and Derivative Pricing - from Statistical Physics to Risk Management" by Bouchaud and Potters - two physicists. They argue convincingly that for example the Dow Jones index is not a Brownian motion since extreme events happen far too often compared to a exp(-x^2) diffusion model.

    If you have some control over these deviations from diffusion, maybe you can make some money (better than lottery) from looking at time derivatives...

    Your description of the win-win situation misses another problem: The broker has to buy the oil when he already knows that it has gone up. Thus he always has to buy at too high a price and to short at too low a price.

  2. Dear Robert,

    concerning the twin win fund, I don't quite understand how your comment differs from mine, except that yours sounds more vague.

    The strategy I recommended the manager is to buy the oil "right" above the initial price, and sells it "right" below it. There's no objective point of telling whether the initial price is too low or too high - it's defining level of the situation, exactly because the manager has to assume it is a Markovian process if he wants to eliminate risks. Of course, in reality, he can't buy and sell millions of times, so he will only act when the price deviates a few pips from the initial ones.

    Concerning the Brownian motion, you seem to mix Brownian motion and more general Markovian processes. Brownian motion might make big jumps Gaussian-unlikely but it doesn't generally follow from the Markovian character of the evolution.

    If someone assumes that the analysis of trends and momenta can't lead to useful profit (and I am not really claiming that, see below), it is equivalent to assuming that the process is Markovian. But that doesn't imply that there are no big jumps or that they decrease in the Gaussian way: the latter only follows for Brownian-like motion.

    I haven't claimed that the evolution is Brownian motion and, in fact, I haven't even claimed that it is Markovian. I just claimed that the speculators assuming and abusing the non-Markovianity of the evolution are counterproductive for the markets' stability and help to deviate the markets from the optimal equilibrium. And if something is punished by state policies and taxes, this is what should be punished.


  3. "Concerning the Brownian motion, you seem to mix Brownian motion and more general Markovian processes."

    I guess you wanted to say martingales. A Markov process can still be well predictable to allow for technical analysis.

  4. What I meant by "above/below the price" is that the broker will not buy when the price is only infinitessimally above the reference price since in that case he would we constantly trading. He would wait until the price is significantly up and then be paying significantly more. The point being that you can only tell the price goes up after it happened.

    You are right about me confusing gausseanity and markov (memoryless) evolution of prices. My confusion came about since the law of large numbers relates the two when sum many distributions.

  5. Dear robert, thanks, now I think we are in agreement. Yes, he has to sacrifice an interval around the initial price, to avoid excessive frequency of trading.

    Incidentally, the question how many times a Brownian motion curve crosses y=0 (i.e. is the trader constantly trading) is interesting. This number can be clearly zero because Brownian motion is continuous and it is normal for such a function, y(x), to avoid y=0 for long intervals on the x axis.

    But when you choose a region where y switches from mostly negative to mostly positive values, how many times will you cross y=0? Is it finite or infinite? Is the typical intersection a "fractal"? Is it countable?