The Efficient (Adaptative?) Market (Policy?) Hypothesis

I have been reading a bit about cognitive biases, which of course has led to a nice discussion about whether financial markets truly are as "random" as is commonly believed.  After wandering through some rather interesting articles, I stumbled upon an alternative paradigm described by Andrew Lo as the Adaptive Market Hypothesis.  Lo introduces the EMH with the classic joke:

...an economist [is] strolling down the street with a companion. They come upon a $100 bill lying on the ground, and as the companion reaches down to pick it up, the economist says, “Don’t bother—if it were a genuine $100 bill, someone would have already picked it up”.

Lo then goes on to describe the various "real world" critiques of the EMH, spanning the range of cognitive biases including loss aversion, miscalibration of probabilities, and regret.  Lo does admit that, in spite of these biases, the EMH can still be robust; arbitrage should take advantage of these irrational biases and price them out of the market.  However, Lo finds that this static conception of markets, as a collection of entities constantly reaching for a predetermined equilibrium, unsatisfactory for the highly dynamic world of finance.  Lo summarizes his new approach, termed the Adaptive Market Hypothesis, as an evolutionary approach to market efficiency.  In his words:

Prices reflect as much information as dictated by the combination of environmental conditions and the number and nature of "species" in the economy

This is a very interesting approach, as it takes into account the environment  in the formulation of market efficiency.  Lo's definition also leads room for evolving financial strategies, asymmetric information, and changing market size.  It posits that while the EMH can be true, it is only true in certain situations, and that, ala Grossman and Stiglitz, markets necessarily have some degree of inefficiency to justify their existence.  In a sense, the AMH is the general case of the EMH; the EMH is the asymptotic case of the AMH.  In markets of infinite agents with infinite cognitive capacity, the EMH holds.  In anything less, the AMH is more satisfactory.  Some Federal Reserve research on foreign exchange markets has also provided evidence that the AMH can function as a better explanatory framework.

However, as Scott Sumner reminds us in his post on the EMH, any criticism needs to have practical implications.  And Lo does have a long list of qualitative insights, , spanning many paradigm differences between the AMH and the EMH.  Fundamentally, these implications boil down to the need to think of markets as organic, evolving systems with changing preferences and strategies.

When I was thinking about these two paradigms, I found that a clarifying question in the context of these two theories is "Why is there no analogous 'Efficient Policy Hypothesis'?"  A certain webcomic humorously illustrates one reason why the hypothesis can't hold:


In the framework of asymptotic cases, the Efficient Policy Hypothesis lies at the other extreme of the AMH, opposite to the EMH.  Whereas efficient markets have a large number of agents with relatively little individual power over the market, policy is comprised of one agent, the government with very large power over the market.  Additionally, while markets can adaptive quickly to new situations, policy, especially if it requires democratic deliberation, has no such capacity.  This kind of thinking then creates a new mechanism through which the failure of policies (and nations) occurs.  It occurs not as simply a malevolence on the part of regulators, but as the natural result of unnatural selection.  And perhaps this does have "practical implications" on the design of policies; for societies to be robust, regulators must adapt.