If you have been
reading my blog recently, you know that I have not written kindly about the
integration of rational expectations and the efficient markets hypothesis into
Austrian economics. This might seem strange to any readers who believe that
markets work. I sympathize with that perspective, but I believe it is
inappropriate to assume this. As I have noted about some modern Austrian
macroeconomic models, reliance on RE and EMH assumes disequilibrium out of
existence unless it is caused by an exogenous shock. RE and EMH may approximate
market outcomes 99% of the time, or even 99.9% of the time, but the remainder,
(1 – p), needs to be understood, especially if its source is endogenous. Imagine if I
have a 1/100 or 1/1000 chance of being struck by lightning every time I step
outside. The variance of outcome in the case I am struck is high. According to
a quick Google search, there is a 10% chance that I die, and some spectrum of
other outcomes. I am not interested in the median result – say, I spend one week
in the hospital – if 1 in 10 times I end up dead. Furthermore, in a world of
fat tailed probability distributions, which seem to better describe the world
of finance than Gaussian distributions, I may not want to optimize expected
value if I am not sure how fat the tails are. In other words, I may
know that my map of possible outcomes does not cover the entirety of possible
outcomes, but I am unsure to what extent it is lacking. This is what George
Shackle and Israel
Kirzner refer to as surprise: outcomes that one is
absolutely ignorant of. If it so happens that the possible outcomes I am
missing from my probability distribution have much higher expected losses than
any outcome within my probability distribution, I may want to consider not optimizing in a way suggested by
RE.
None of this is to
say that RE and EMH are not useful. They provide a teleological compass for
fruitful economic analysis. In many cases, however, median outcomes do not
matter. In these cases, RE and EMH are better described as an Achilles heel of
economic analysis.
What is needed is an
economic framework that allows for emergent processes, a non-equilibrium
framework. We do not know what the economy will look like in ten years – even
absent exogenous shocks – neither should agents in our models. (If you
disagree, extend the time frame to 100 years.)
What is needed is not
that Austrian economics “catch-up” to the modern equilibrium paradigm, “but an
effort to create the contemporary analytics that might have been created had
Austrian macro theorizing continued to evolve since 1940 with the same
robustness that it exhibited before 1940 (Wagner, 98).”
What is needed is a
macroeconomics that foregoes determinism associated with RE and EMH and
concentrates on rules that guide decision-making and human action. Individuals
do not approach “truth” directly. They integrate information according to
heuristics, and then act to achieve their ends according to their interpretation of information.
Those who endeavor to
develop this research program must model in terms of heterogeneous agents whose
actions are determined according to rule sets. The outcomes of interactions
between agents acting according to rule sets and the transformation of rule
sets over time may shed light on market processes in a way that equilibrium
theorizing cannot.
No comments:
Post a Comment