The Economic Sociology of Asset Market Efficiency

Brayden King asks:

Pub Sociology: A sociology of market efficiency: According to orthodox views of market efficiency, smart money should be able to correct for the irrationality of bad investors, bringing prices back to fundamentals even when the majority of investors over- or undervalue certain stocks. This is the role of arbitrage. Yet, we often see that markets still produce inefficient outcomes. Why?... Our sociological intuition tells us that market structure - the social relations between actors, practices, and meaning - ought to mediate the extent to which markets operate more or less efficiently. This is one of the most important insights that economic sociology has to offer, I think (and one of the primary conclusions in my dissertation). We should be able to show that the structure of relations in a market has a real impact on the efficiency of market pricing and choice.

Ezra Zuckerman, who I think is one of the brightest and most interesting scholars in the field, has a lot to say about this. In his 2004 piece* on structural coherence and market valuation, he argues the very this very point:

I challenge the assumption made by the [efficient market hypothesis] that the social structural environment typical of financial markets always has the necessary features to support the highly sophisticated social learning necessary for incorrect models to valuation to be driven from the market...

*Zuckerman, Ezra. 2004. “Structural Incoherence and Stock Market Activity.” American Sociological Review 69: 405-32.

Well, it is explicit in DeLong, Shleifer, Summers, and Waldmann (1990), "Noise Trader Risk in Financial Markets," Journal of Political Economy, that as a matter of economic theory rational, sophisticated investors are not guaranteed to earn higher expected returns on their portfolios than are noise traders if rational investors have short horizons, and if noise traders on average concentrate their long positions in the assets about which their opinions irrationally fluctuate.

It is implicit in DSSW (1990) that as a matter of economic theory incorrect models of valuation are not only not driven from the market but exist happily in it and can come to dominate it if:

1. Rational investors have short horizons.
2. Rational investors are risk averse.
3. Noise traders on average concentrate their long positions in the assets about which their opinions irrationally fluctuate.
4. Noise traders' misperceptions are correlated across noise traders.
5. New entrants into the market look back at recent realized returns to decide what valuation strategies to adopt.
6. There is enough fundamental risk to curb rational investors' willingness to take large positions against noise traders.

Perhaps it is time to make this explicit as well: here's a memo.

And I am still waiting for my copy of Braydon King's dissertation...

Comments

[Steve Carr]

There's a fourth premise that has to be true for DSSW to prevail:

Noise traders must all (or almost all) be making the same mistake when it comes to valuing the asset in question. In this model, all the uninformed and misinformed are misinformed in the same way: they all agree (incorrectly) that the asset is under- or overvalued.

This seems to me a deeply questionable assumption. While it's plain that there are times when all uninformed investors are in agreement (the late 1990s comes to mind), there is no theoretical reason to assume that misinformed/uninformed investors are going to be misinformed in the same way. On the contrary, our starting assumption should be that errors are going to be distributed normally (or close to it), so that noise traders will be misvaluing assets in uncorrelated ways, with the result that the market will in fact price the asset correctly (or as correctly as is possible). To make the case that DSSW describes the way markets work in the real world, you have to demonstrate that noise traders in fact consistently value assets in the same way.

[c.]

I'm pretty sure that's not true.

[Jack]

Steve, I'm not at all sure that assuming the views of the noise traders is random is a good assumption. It would of course be nice if it were true but I find it hard to imagine the investor being modelled under such an assumption.

Also the ones you suggest can be ignored for the reasons you suggest. In effect this isn't really a model of the whole market (more precisely of every bit of the market. It works quite well for the market as a whole). The effects described won't obtain for all securities but neither will all the samples from the distribution turn out the mean.

It is also not the case that elaborating the model would make it go away. For example an even more astute arbitrageur might spot the dominance of noise traders and come along for the ride making life even harder for the fundamentalist arbitrageurs. That is certainly what momentum traders think they are doing. AIM had scripts highlighting positive and negative language in press releases and news coverage so that they could respond instantly.

As for the final test I don't think it is quite right. You would just have to show that sometimes uninformed investors I think the proposition that all uninformed opinion is a wash would be quite hard even to simulate even without introducing realistic biases such a flawed bayesian learning or common attention (What percentage of companies are in the news in a given week? How many people look at performance tables before finding an investment? How many people pay attention to stocks that are not at one end or other of some league table?) and news sources which can do quite a good job of explaining concentration on particular stocks.

[Barry]

There's an old saying, 'the market can remain irrational longer than you can remain solvent'.

[Silent E]

Brad's Memo:
"This memo picks up a thread left implicit in
DSSW (1990): the interaction between fundamental risk and the
possibility that noise traders might flourish in the market. In the
domain of economic theory at least, even relatively small increases
in fundamental risk curb the enthusiasm of rational investors to bet
against noise trading. Such fundamental risk substantially impairs
the market’s ability to function as a social value calculating
mechanism, in the sense that it cannot then be expected to drive
incorrect models of valuation from the market. Indeed, as long as
conditions 1-3 plus the following are met:
4. New entrants into the market look back at the mean of
recent realized returns to decide what valuation strategies to
adopt (and ignore the variance).
5. There is sufficient fundamental risk to curb the appetite of
sophisticated investors from making long and risky bets
against noise trading.
Then, at least as long as we are climbing this particular branch of
economic theory, the market does not support the highly
sophisticated social learning necessary for incorrect models to
valuation to be driven from the market. There are forms of noise
trading—incorrect models of valuation—that survive in a stable
fashion in the market, no matter how long the run in which it has to
learn. There are forms of noise trading that come to dominate the
market."

I imagine this is at its worst in bubbles, where the noise traders become totally divorced from 'rational' investing considerations. But there is no good way to time the end of a bubble and short it. Large bubbles attract sufficiently large numbers of new entrants that sophisticated investors find themselves outnumbered (and, more importantly, out-capitalized) by the combined forces of Noise and Inexperience. The bubble is entirely pyschological, and trying to time it's end is as much a speculative exercise as trying to ride it.

[anne]

http://www.calvorn.com/gallery/photo.php?photo=5482

Male Baltimore Oriole Feeding Chicks
New York City--Central Park.

Hmmm....

[tina]

I can't wait for the DeLong and King piece that shatters the barrier between economics and sociology!

[anne]

Tima, I do agree for we are still neglecting sociology:

http://www.nytimes.com/2005/06/09/business/09scene.html?ex=1275969600&en=d364cb0cbad9e57f&ei=5090&partner=rssuserland&emc=rss

June 9, 2005

The Mysterious Disappearance of James Duesenberry
By ROBERT H. FRANK

UNLESS you are a professional economist nearing retirement, the name James S. Duesenberry is probably unfamiliar. By itself, that is unsurprising, because he wrote primarily for academic audiences while on the Harvard economics faculty from 1946 to 1989. The real surprise is that most academic economists under 50 have also never heard of Mr. Duesenberry.

This is puzzling because his theory of consumer behavior clearly outperforms the alternative theories that displaced it in the 1950's - a striking reversal of the usual pattern in which theories are displaced by alternatives that better explain the evidence. His disappearance from modern economics textbooks is an intriguing cautionary tale in the sociology of knowledge.

[anne]

Look to the marvelous work of dear Orlando Patterson in sociology and ask how we might use this in economics.

[palinal]

Tina- and birdwatching.

[anne]

http://www.calvorn.com/gallery/photo.php?photo=5459&u=4|11|...

House Wren Singing
New York City--Central Park, Maintenance Field.

Well, it is that kind of morning :)

[wkwillis]

There's an old saying, 'the market can remain irrational longer than you can remain solvent'.

Posted by: Barry | June 14, 2005 04:19

This should have the words 'leveraged and' inserted firmly between 'you' and 'solvent'.

[anne]

Palinal, I was asked only yesterday whether bird watching was an art. Surely an interesting question, especially so if watching is at all combined with telling of what we see. But I am improving at even noticing, or so I imagine :)

[Josh Narins]

Prof DeLong,

Although no details have been hammered out, I was wondering what you'd suggest reading, or what you'd write yourself, concerning the self-congratulatory efforts of Bush/Blair and the hypothetical Africa debt deal.

As I understand it, Gordon Brown is putting forward the deal, Bush and other G8 people will hammer out details in early July.

AEI is all over it, which means it _must_ be supporting corporate plutocracy (their only apparent goal, if sometimes tempered, as far as I can tell).

[anne]

http://www.nytimes.com/2005/06/14/opinion/14tue2.html

A First Step on African Aid

The Bush administration took a modest but important step in the right direction recently when it reached agreement with six other rich countries to ease the burden of debt in Africa. The arrangement would eliminate $40 billion of debt owed by 18 of the world's poorest countries, including 14 in Africa. President Bush deserves credit for agreeing to provide additional money to make sure the World Bank does not go out of business when the payments stop coming. The bank and the International Monetary Fund are the largest creditors to poor countries.

But enough patting on the back; the real mountain remains to be climbed. Mr. Bush has yet to sign on to the meat of the proposal by Prime Minister Tony Blair of Britain to increase aid to Africa by $25 billion a year. That is the amount that Britain and many development economists say is essential to achieve the United Nations' goal of halving global poverty by 2015. Indeed, the debt relief package actually comes to the relatively small sum of about $1.5 billion a year, not all for Africa....

[meno]

Speaking as one who helps build risk-management systems: One of the key things here is that the 'rational investors' who you want to be doing your arbitrage are often very, very risk averse.

Banks do not like to take risks. Oh no, they do not.

[Ian Whitchurch]

Quick background - I do fundraising for junior oil explorers in Australia. I've been in the middle of some "exciting" times, as the value of shares in oil companies fluctuate by hundreds of percent during the drilling of wells. While I dont quite play the market for a living, my ability to raise funds is determined by how well I analyse stocks.

Theoretically, Econometrica in 1984 (round about p1000) has some useful articles about rationalisable expectations.

Here's the short version of how the world works, as I understand it.

We have "fundamentals" players, who spend lots of time and effort gathering information on fundamental values.

We have "momentumt" players, who carry out the very cheap method of buying stuff thats going up, and selling whats going down.

The price of a stock, P is X at time t. At t+1, it's X+v.

P(t)'s price is different from P(t+1)s price because it's bid up or down.

If only fundies are buying, v is positive because information has been discovered that makes the company more valuable. If only fundies are selling, v is negative because information has been discovered that makes the company less valuable.

A momentum trader can make money by copying what the fundies are doing, only without spending all that time and effort analysing companies - if they are going up, good information exists and you should buy. If they are going down, bad information exists, and you should sell.

The problem is that fundies buy intially (pushing P(t+1) to X+v1), then momentum players see that and buy, pushing P(t+2) to X+v1+v2.

Momentum traders then assume that even more good information exists, because the price has gone up even further. They then buy more, pushing the price up even further.

This model doesnty explain why bubbles pop, but it does explain why they happen, because momentum traders mistake other momentum traders buying for buying by people with better information.

Ian Whitchurch

[Robert Bell]

Seems like Fama and French should talk you ...

http://gsb.uchicago.edu/faculty/french_asset.pdf

The notion of correlation between the beliefs of uninformed investors also figures into their "correction" of CAPM pricing.