Financial Economics — Information, Market Efficiency and Price Discovery

Financial Economics — For Prices to Carry Information, There Must Still Be a Reason to Research

“The market is efficient” is a sentence used far too often. Unless we separate which information set prices are said to reflect as a conditional expectation, and who produces that information and why, counterexamples can be found anywhere.

1. The weak, semi-strong and strong forms differ in their information sets

Fama’s classification divides types of predictability.

Information sets of the efficient market hypothesis
Form	Information prices are said to reflect	Evidence that counts as rejection
Weak	Past prices and volumes	Risk-adjusted excess returns from technical analysis alone
Semi-strong	Public accounts and news	Systematic drift that persists after disclosure
Strong	Even inside information	Insiders earning excess returns on average

The strong form is close to rejected from the outset, which is why insider-trading rules exist. The weak and semi-strong forms do not claim that “nobody can make money”; they are closer to claiming that no rule based on public information leaves anything after subtracting risk and trading costs.

A random walk is neither a necessary nor a sufficient condition for efficiency. If risk premiums vary, returns can be predictable without any arbitrage. Conversely, prices can contain errors that last a long time if arbitrage capital is constrained.

2. Grossman–Stiglitz: if prices reflect everything, nobody researches

If information could be obtained at zero cost, an equilibrium in which prices fully reflect it would be possible. In reality research is costly. If prices carried 100% of the information, the profit from research would be zero, nobody would research, and prices could not carry the information.

Equilibrium in information production

E[\pi \mid \text{research}] - c = E[\pi \mid \text{no research}]

With a positive research cost c, prices reflect information only partly, so that the researcher's expected profit covers the cost.

So one cannot say at the same time that “news goes straight into prices” and that “research is worthless”. Price discovery is a costly activity, and its reward remains in the form of temporary errors or bid-ask spreads.

In numbers: let research cost 2 won per item, with a research profit of 5 won when the information is not yet in the price and 0 won when it already is. If $p$ is the probability that the information remains unreflected, the researcher’s expected profit is $5p$ won. At $5p=2$ , that is $p=0.4$ , researching and not researching are equally attractive. In equilibrium prices reflect only 60% of the information, and the remaining 40% gap pays for research.

3. Anomalies: risk, mispricing or costs?

Differences in average returns not explained by market beta — small stocks, value stocks, momentum — are reported again and again. There are three lines of interpretation.

Risk: compensation for losing more in financially distressed times
Mispricing: overreaction and underreaction, arbitrage under constraints
Costs and bias: data mining, trading costs, delisting bias

Some anomalies disappear or weaken after publication, either because the discovery itself invites arbitrage or because they fail to replicate out of sample.

Check your understanding

If semi-strong efficiency holds, is a strategy that earns risk-adjusted excess returns from chart patterns right after corporate disclosures possible? Disclosures are public information and chart patterns are past prices, so if the semi-strong form holds, the strategy’s expected excess return is zero. Whether drift remains in actual data is the evidence for rejection.

References

Eugene Fama, “Efficient Capital Markets,” Journal of Finance (1970)
Sanford Grossman and Joseph Stiglitz, “On the Impossibility of Informationally Efficient Markets,” American Economic Review (1980)
Korea Exchange KIND, Listing disclosure system (Korean)