Wednesday, 18 April 2012

Imperfect markets

Markets are typically said to be the best way to distribute our resources. In general, I don't dispute this: I think that some necessities need to be distributed differently, to prevent poor people being priced out of the market, or to ensure coverage, but luxuries and items with plenty of substitutes suit markets well.
However, I do take issue with any suggestion that markets approach perfection, which is often assumed as axiomatic to economic theory. Markets cannot approach perfection. Perfect markets can approach perfection, but that's a tautology: we never deal with perfect markets.

This post is about the efficient market hypothesis, which I was reading about a couple of days ago during remedial GCSE maths (financial reporting work).
Weak efficiency is when prices can no longer be predicted from past movements. If they can be, then people are not acting on that possibility, which means that the price does not accurately reflect information. Prices should follow a 'random walk', in which each future movement is a new event.
Semi-strong efficiency is when prices incorporate all publicly available information.
Strong efficiency is when prices incorporate all information.
If markets are semi-strong efficient, inside information is required to beat the market.
If markets are semi-strong efficient or better, then someone with incomplete knowledge (i.e a human) should trust the market price instead of his own estimate of the price. The problem with this is that the market price is accurate precisely because it is an average of everyone's estimates, and therefore incorporates the information these people know. If one person free-rides on this accuracy, nothing changes, but when it becomes widely accepted that the market is right, information is removed from the estimate, which also becomes much more chaotic as everyone clusters around believing that the 'true' value is very close to the quoted market price, ensuring that small actions can have big effects. This is a double effect of the clustering of people's estimates and the removal of information. It will no doubt be exacerbated by the tendency of those with estimates furthest from the market price to doubt their information the most. Humans have an inbuilt fallacy called 'anchoring' which contributes to this.

Hence, if semi-strong or strong efficiency is accepted, it is rapidly lost by the free-rider problem of people taking advantage of the supposed information conveyed by the price. I have been told that this will be corrected by people who do know what is right taking the opportunity to make a profit, but this assumes that there is one person who knows accurately what the value is: the precise lack of which led to us creating the market in the first place. It also assumes that the market will self-correct. If there is a self-sustaining chaos, or movement away from the 'true' value, then any attempt to use knowledge will fail, being subject to chaos just like random bets.

Given this tendency of markets to degenerate to self-referential prices, there are two ways to beat the market: to use insider information in a semi-strong market, or to be lucky. In fact, if the market becomes self-referential, one can either invest in complex algorithms that predict human behaviour, in an attempt to understand how such self-referential human systems behave, or one can realise that the purpose of the market is no longer being served, and withdraw from it to start again.

Our big banks have, of course, gone for the complicated algorithm option.
So, why do we have hordes of speculators?

The solution would be to prevent the system being self-referential. We want a market that accumulates estimates and averages them, without these being biased by knowing the final result. We need to cut the feedback loop: the contributors to a market need to be blind to the price. That's a problem when the point of a market is to allow them to get a sensible and consistent price.
How else can we avoid the feedback loop that removes information? We can cut it through time, by delaying transactions until after bids, but delays come at a cost, and the next round in a continuous market will be affected by the last round no matter the timing.
In fact, we must tackle speculation itself: the ability to feed round the loop many times. We can't cut the loop perfectly, but if we make transactions cost, then following the loop round will incur a cost that will deter people who are not interested in what the market is for. This cost will hit legitimate users as well. We could also limit derivatives; these specifically refer to market price, and are therefore part of the self-referentiality of the market. Some are necessary, in order to deal with risk, but the freedom to overpower the market with them can disrupt the proper functioning of the market.
I would love to do a cost-benefit analysis of the liquidity added to the market for legitimate use weighed against the cost of greater market imperfection through self-reference. But sadly I don't know enough economics or maths.

No comments:

Post a Comment

Female entitlement

  There is a segment of society that claims to believe in equality and fairness; and yet refuses to examine the privileges of one half of ...