Friday, July 19, 2013

Brady on Speculation in Financial Markets

Food for thought from Michael Emmett Brady:
“There is a long 400–500 year history that demonstrates repeatedly, time and time again, that past and current speculation always leads to some kind of future economic problem.

Keynes recognized that financial markets, for the last 400–500 years since the introduction of modern, fractional reserve banking, exhibited the same speculative pattern over and over and over and over again. …. Obama, Bernanke, and Geithner … bailed out the Wall Street speculator crowd again, just as they were bailed out in the early to late 1980’s by Paul Volcker and late 1990’s–early 2000’s by Alan Greenspan. The result is that another bubble in the stock markets is being created. These financial bubbles are ergodic because the same pattern repeats again and again. New types of financial assets and financing are created by the banking industry. In the 1920’s, for example, these new financial assets were balloon payments for houses and margin account financing for stocks. The creation of these new types of assets is called securitization. The next step is debt leveraging. This allows speculators and speculating bankers to maximize their speculative debt financing. The growing bubble is fed by herding and copycat behavior that automatically leads to the creation of a larger and larger bubble. The next stage occurs as the bubble leads to a mania, which leads to a panic, which inevitably leads to a crash, which always leads to an economic downturn, recession, or depression of some sort. These kinds of events are stationary because they keep repeating over and over again. Their ultimate collapse can be predicted with a probability approaching 1. However, they are not normally distributed. One can’t use the normal distribution to describe the time series data in financial markets. The underlying processes are given by the Cauchy distribution.”
Michael Emmett Brady, September 18, 2009
I find the idea that the repeated rise and fall of bubbles per se in capitalism to be ergodic worthy of further investigation.

Of course, one needs a strict definition of ergodicity and stationarity.

But another issue is how one defines “bubble.” It is entirely conceivable that a small or moderate bubble might in fact stabilise, reach plateau and then further bull or bear markets may follow, instead of simply deflating in a significant way.

Of course, if one wants to limit the definition of “bubble” used here to large, debt-fuelled bubbles, which really destabilise asset prices wildly, then the idea that the collapse of such bubbles “can be predicted with a probability approaching 1” is not so unreasonable, even though I assume that such a probability value would be what Keynes called non-numerical (Keynes 1921: 160), and cannot be understood as in the same class as a priori probabilities.

Keynes, John Maynard. 1921. A Treatise on Probability. Macmillan, London.


  1. Related to this --not so much in terms of probability and the discussion of ergodic and non ergodic systems but-- in terms of the functioning of financial capitalism is William Janeway's thoughts, and in particular his book "Doing Capitalism in the Innovation Economy: Markets, Speculation and the State"
    I have a related post here in case it is of interest: which may be of interest.

  2. Tautological argument. Of course a bubble bursts. The metaphor is suggestive of that and the origin of the term suggests the same:

    Saying that a bubble bursts with a probability approaching 1 is like saying that a body in motion moves with a probability approaching 1.

    I also think this says something broader about the sort of thinking that Brady appears to be engaging in...

    1. I understand this possible critique, but it doesn't necessarily seem like a tautology to me: it looks like an inference derived from induction by simple enumeration:

      (1) instance 1 of a large asset bubble is accompanied by its collapse;

      (2) instance 2 of a large asset bubble is accompanied by its collapse;

      (3) instance 3 of a large asset bubble is accompanied by its collapse; etc. etc.

      Therefore instances of large asset bubbles will be accompanied by their collapse with a high degree of probability.

      Of course, one can complain that maybe the argument is unsound: perhaps it commits a fallacy of hasty generalisation or the data sample is too small, but one has to prove that too.

    2. But again, the argument is already inherent in the term "bubble" -- i.e. if we retroactively call an instance a "bubble" we are already saying that it would eventually collapse.

      Again, think of a body in motion. We can lay out three instances where a body in motion is moving. But the fact that it is moving is already implicit in the fact that we have designated it a "body in motion".

      If we had made the claim not that all bodies in motion move, but that all bodies move then we might be saying something. Similarly if we said that all instances of speculation resulted in collapse then that is a testable claim with content. But the other two are tautologies.

  3. I don't see the point really of Brady's neologism on what constitutes ergodicity. That a pattern is repeated again and again is certainly NOT in any way a tenable definition of ergodicity!