Tuesday, September 13, 2011

Put not your faith in labels

Any time there is a serious economic downturn, mainstream economics drops in popular and intellectual esteem, even though mainstream economics generally does not pretend to have abolished the business cycle while it explains why reliable, systematic prediction of future economic conditions is impossible.

However, serious economic downturns are almost invariably the result of bad policy, and economists and economics will be berated for failing to give successful advice. (Indeed, one view is that serious downturns can be manifestations of market predictions of bad policy.)

This loss of esteem is generally unfair about microeconomics. In the words of former head of the Department of Prime Minister and Cabinet, Dr Michael Keating, microeconomics is successful at “producing robust predictions of general tendency”. It is rather less unfair about macroeconomics, which has failed to achieve even a common analytical language.

Of course, some folk can be correct in predicting some particular event or trend, but it will generally not be continually the same folk. Outlier predictors are, of course, a statistical possibility—consider having a large number of people predicting heads or tails: some folk will continue to get it right though fewer and fewer as more throws are made, but that is a result of that a particular sequence will occur and, with enough varied predictors, some people will pick that sequence; it is not that they have some insight into “heads or tails”.

That said, economics also creates rods for its own back, and it does so by some poor naming of major ideas. Typically, such names misleadingly over-claim and either provoke hostile responses from the unsympathetic or encourage poor thinking or both.

Information and expectations

Consider 'rational expectations': it should really be labelled consistent expectations since the fundamental notion is that expectations of agents in the model of the economy should be consistent with the model. That is, you should not base any economic model on the mere presumption that the modeller has significantly and persistently better insight into the operation of the economy than the (other) agents in the economy. It is fundamentally a principle of analytical humility but, unless one can provide reasons why the modeller would have—in a systematic and continuing way—better information than people for whom a great deal is at stake, it is a sensible analytical principle.

The principle is a formal expression of the reality that people react to information about the economy: which includes (implicit or explicit) models of the economy. But labelling the principle ‘rational expectations’ is misleading and invites misreading as putting some very high value on the “quality” of expectations rather than their common cross-agent limitations. (Yes, the notion is that agents are rational in their use of information but rationality is a general economic principle: the principle involved here is specifically about analytical consistency in application of information.)

Then there is ‘the efficient market hypothesis (EMH)’: which should be labelled the informed market hypothesis, since the notion is that prices will reflect information available to agents. The problem with terming it the efficient market hypothesis is that it then looks like a principle of market perfection, which it is not. (There is a strong version[, a semi-strong] and a weak version: I am talking here of the weak version.) The weak version does imply that open markets will the best way of pricing assets, but that does not entail that markets are perfect: merely there is no systematically better mechanism for pricing assets. It does not even imply that information is transferred instantaneously.

Nor does EMH imply that asset price bubbles are impossible: on the contrary, it is precisely because we cannot predict new information that we cannot predict turning points, so asset price bubbles become possible (since if we could reliably and systematically predict turning points, prices would not rise to a level for people to be caught by them when the bubble bursts) while expectations of capital gain both motivate agents and are part of the information feeding into prices. Asset bubbles are clear enough in hindsight, when we have the information about how they ended: specific information not available to the participants in the bubble (hence there are always folk denying that any bubble exists: and if income on said assets rises to “catch up with” the expected capital gains, they will be correct).

If EMH was labelled the ‘informed market hypothesis’ it would be less misleadingly named and less of an affront to folk not enamoured of markets. (And yes, EMH is about the efficiency implications of use of information but, again efficiency—or its lack—is a general feature of economic mechanisms; EMH is specifically about markets and information.)

Putting consistent expectations and informed markets together—the alert reader will have noticed that they both enjoin the analyst to take information flows seriously and not presume one is a privileged observer—suggests that considerable scepticism about regulation is appropriate. (Particularly discretionary regulation, where the approval of officials is required.) As there is no reason to think regulators will be systematically better informed than economic agents in general. (Noting that regulation is based on some implicit or explicit model of behaviour.) Indeed, there is good reason to think that discretionary regulation will make markets more chaotic, rather than less, by narrowing the use of information and generating perverse incentives: an expectation that has considerable empirical support (pdf), particularly in the experience of command economies.

There is a large debate about central banking in particular around the discretionary/rule/open markets possibilities (central bankers should have discretion; they should operate according to some policy rule; they should be abolished). Hence Swedish economist Lars Svensson’s suggestion to target the forecast; so policy action and market information work together.

Note: I am using “layperson friendly” characterisations of both “rational expectations” and EMH: for much more sophisticated discussion of both and critiques thereof, see Stephen Williamson’s review essay(pdf) on John Quiggin’s Zombie Economics.


Another case of poor labelling is the Austrian economics concept of 'malinvestment'. What is or is not a good investment significantly depends on larger economic conditions. What is a great idea in New York may be a really dumb one in Port-au-Prince. The notion of malinvestment is that unwarranted monetary expansion misleads folk about the future path of economic activity. As conditions change, investments based on such unwarranted expectations are "exposed" and need to be liquidated to free resources to go to more valuable uses.

But the label implies (in compete contradiction of Austrian value subjectivism) that being a malinvestment is an intrinsic quality of an investment. If so, the level of economic activity becomes irrelevant to the level of malinvestment. So, you can happily advocate any amount of restrictive "adjustment" because the level of "bad investments" wasting resources is set.

[Read the rest at Skepticlawyer or at Critical Thinking Applied.]

No comments:

Post a Comment