Market Efficiency, Behavioural Finance and Anomalies

Introduction

Security analysis asks whether an investor can identify securities whose risk-adjusted returns differ from those justified by available information. Market-efficiency theory provides one answer: in an informationally efficient market, prices incorporate information rapidly, so persistent abnormal profits are difficult to earn after transaction costs and risk. Behavioural finance provides a complementary answer: investors are not always fully rational, and psychological biases, institutional incentives and limits to arbitrage can cause prices to depart temporarily from fundamental value.

The two perspectives should not be treated as simple opposites. Markets may be efficient with respect to some information, securities and periods, while displaying predictable patterns in other circumstances. A finance professional therefore needs to understand both the evidence supporting efficiency and the reasons anomalies may arise.

Learning Objectives

By the end of this chapter, a student should be able to:

1. Distinguish weak, semi-strong and strong forms of market efficiency.
2. Explain how information is incorporated into security prices.
3. Evaluate major empirical tests of market efficiency.
4. Explain behavioural heuristics, biases, loss aversion, overconfidence and herd behaviour.
5. Identify and critically assess important market anomalies.
6. Apply efficiency and behavioural concepts to investment and managerial decisions.
7. Distinguish abnormal performance from an intercept that is merely an artefact of model specification.

1. The Meaning of Market Efficiency

An efficient market is one in which prices reflect information sufficiently quickly and accurately that an investor cannot systematically earn risk-adjusted abnormal returns using that information. Efficiency is not the claim that every price is always correct. It is a claim about the profitability of trading strategies and the speed with which mispricing is corrected.

Let P_t denote the ex-dividend price of a security at time t, and let I_t denote the information set available at that time. A general fundamental-pricing representation is:

Here, CF_t+k is the cash flow at date t+k, and m_t,t+k is a stochastic discount factor reflecting both time value and risk. The notation E_t[·] means expectation conditional on I_t. This formulation is more general than simply writing price as an undiscounted expected future value. Under a constant discount rate r and a finite expected cash-flow stream, a simplified version is:

The simplified equation requires a constant, appropriate risk-adjusted discount rate and known payment dates. In practice, discount rates and cash flows may be uncertain and state-dependent.

An informationally efficient price can therefore be represented conceptually as the best risk-adjusted valuation conditional on available information, not as a guaranteed true value. New information can change expected cash flows, risk or discount rates, causing prices to change.

Information sets and the three forms

The forms are cumulative: strong-form efficiency implies semi-strong efficiency, and semi-strong efficiency implies weak-form efficiency. The reverse implications do not follow.

2. Empirical Tests of Efficiency

Weak-form tests

Weak-form tests examine whether returns are predictable from historical market data. Researchers use autocorrelation tests, runs tests, variance-ratio tests and trading simulations. If successive returns are independent, the autocorrelation at lag one, for example, should be close to zero:

where R_t is the return in period t, Cov(·) is covariance and σ(·) is standard deviation. A low autocorrelation does not prove efficiency: nonlinear dependence, time-varying volatility and transaction costs may still create or eliminate profitable strategies.

A trading rule must be evaluated using net returns:

A statistically significant pattern is not necessarily economically significant. The pattern must survive realistic costs, liquidity constraints, risk adjustment and out-of-sample testing.

Semi-strong tests and event studies

Semi-strong tests examine how prices respond to public announcements such as earnings, dividend changes, mergers, credit-rating changes or regulatory decisions. An event study estimates abnormal return around an event date.

A commonly used market model is:

where R_it and R_mt are raw returns on security i and the market, β_i measures sensitivity to market movements, and ε_it is an error term. In this raw-return specification, α_i is an intercept; it is not automatically Jensen’s alpha because it may reflect the risk-free-rate component and benchmark misspecification.

For performance evaluation, the excess-return market model is more appropriate:

where R_ft is the risk-free return and α_i^J is Jensen’s alpha relative to the market benchmark, subject to the model’s assumptions. A positive estimate does not prove skill: it may reflect omitted risk factors, data mining, estimation error or survivorship bias.

For an event study, abnormal return may be defined as:

where X_t represents the chosen benchmark information, such as market or factor returns. Cumulative abnormal return over an event window from τ₁ to τ₂ is:

A rapid, statistically reliable price adjustment concentrated near the announcement supports semi-strong efficiency. Delayed drift may indicate underreaction, but it may also result from a poorly specified benchmark, leakage, confounding news or measurement error.

Strong-form tests

Strong-form tests examine whether investors with privileged information earn abnormal returns. Evidence that corporate insiders, analysts or institutional investors possess information advantages challenges strong-form efficiency. Securities laws commonly prohibit trading on material non-public information. Consequently, evidence of insider profits is both an empirical issue and a governance issue; it does not imply that such trading is lawful or ethical.

Methodological cautions

Efficiency tests are joint tests of market efficiency and a model of expected return. Rejection may mean that markets are inefficient, or that the benchmark model is inadequate. Important problems include:

• Joint-hypothesis problem: abnormal returns cannot be measured without specifying expected returns.
• Multiple testing and data mining: searching many variables produces false discoveries.
• Survivorship bias: excluding failed funds or firms exaggerates performance.
• Look-ahead bias: using information unavailable at the decision date overstates strategy quality.
• Transaction costs and capacity: a strategy may work in a small sample but fail when capital is deployed.
• Risk variation: higher average returns may compensate for higher or unmeasured risk.
• Publication bias: striking findings are more likely to be published.

3. Behavioural Finance

Behavioural finance integrates psychology with economics. It relaxes the assumption that investors always process information rationally and recognises that arbitrage is risky, costly and limited. Behavioural explanations generally involve three elements: biased beliefs, preferences that differ from simple risk aversion, and limits faced by arbitrageurs.

Heuristics

Heuristics are mental shortcuts used to make decisions under uncertainty. They reduce cognitive effort but can produce systematic errors.

• Representativeness: investors classify a firm according to a familiar pattern and may extrapolate recent success too far.
• Availability: easily recalled events receive excessive weight; vivid market crashes may cause disproportionate pessimism.
• Anchoring: an initial price, forecast or purchase cost influences later judgments even when it is irrelevant.
• Affect heuristic: emotional impressions substitute for detailed probability assessment.

Major biases

Overconfidence causes investors to overestimate the precision of their information or their skill. It can produce excessive trading, concentrated portfolios and inadequate attention to contrary evidence. Managers may also overestimate acquisition synergies or forecast accuracy.

Confirmation bias leads decision-makers to seek evidence supporting an existing view and discount contradictory information. A disciplined investment committee should require a written bear case and independent challenge.

Disposition effect describes the tendency to sell winners too early and retain losers too long. It is related to mental accounting and the desire to avoid realising a loss.

Loss aversion means that the psychological discomfort from a loss is greater than the pleasure from an equal gain. In prospect theory, outcomes are evaluated relative to a reference point, and the value function is typically concave for gains, convex for losses and steeper on the loss side. Loss aversion can explain reluctance to sell losing securities, excessive demand for guarantees and preference for avoiding visible short-term losses even when long-term risk is higher.

Herd behaviour occurs when individuals follow others rather than their own information. It may arise from reputational concerns, compensation structures, social learning or genuine correlation in private information. Herding can amplify bubbles and crashes, but apparent herding is not always irrational: investors may rationally infer information from others’ actions.

Availability and familiarity bias can lead investors to overweight domestic companies, familiar industries or securities frequently discussed in the media, reducing diversification.

4. Market Anomalies

An anomaly is a recurring empirical pattern that appears inconsistent with a specified asset-pricing model or the efficient-market hypothesis. An anomaly is not automatically a risk-free profit opportunity.

Momentum and reversal

Momentum is the tendency for recent winners to outperform recent losers over an intermediate horizon. It is often interpreted as underreaction, gradual information diffusion or investor herding. Long-horizon reversal, by contrast, describes later underperformance of prior extreme winners and recovery of prior losers. Reversal may reflect overreaction, liquidity effects or changing risk.

Value and size effects

Value strategies buy firms with low prices relative to fundamentals such as book value, earnings or cash flow. The value premium may reflect investor overreaction, distress risk or other systematic risks. The size effect refers to the historical tendency of smaller firms to earn higher average returns than larger firms, although the effect varies across markets and periods and may be sensitive to measurement and liquidity.

Calendar and issuance effects

Examples include January seasonality, day-of-the-week patterns, post-earnings-announcement drift and underpricing of initial public offerings. Tax-loss selling, institutional trading schedules, information processing and changing sample composition may contribute. Once publicised, anomalies may weaken because investors trade on them.

Low-volatility and profitability patterns

Low-volatility stocks have sometimes produced higher risk-adjusted returns than standard models predict, despite the expectation that riskier securities should earn more. Profitability and investment patterns also appear in multifactor models. Their interpretation is contested: they may reflect omitted risk, leverage constraints, investor preferences or genuine mispricing.

5. Applying the Ideas to Portfolio Management

An efficient-market perspective discourages costly attempts to forecast short-term price movements using already known information. It supports diversification, low-cost implementation, attention to taxes and a risk-appropriate strategic asset allocation.

A behavioural perspective adds practical controls:

1. Write an investment policy statement with objectives, constraints and rebalancing rules.
2. Use position limits to control overconfidence and concentration.
3. Rebalance systematically rather than emotionally.
4. Separate research from trade execution where conflicts exist.
5. Record the information available at the decision date.
6. Use checklists, scenario analysis and independent challenge.
7. Evaluate performance against a pre-specified benchmark after fees and risk adjustment.
8. Distinguish a sound process from a lucky outcome.

Diversification reduces idiosyncratic risk but does not eliminate market, liquidity, inflation or model risk. Behavioural safeguards should therefore complement—not replace—financial analysis.

6. Managerial, Governance and Ethical Implications

Managers influence market efficiency through disclosure quality, investor relations and capital-allocation decisions. Timely, complete and non-misleading disclosure reduces information asymmetry. Selective disclosure, earnings manipulation and misleading forecasts undermine market integrity and can create legal liability.

Boards and compliance officers should establish insider-trading controls, restricted lists, blackout periods, wall-crossing procedures and records of material non-public information. Analysts must distinguish legitimate research from the misuse of confidential information. Fiduciaries should consider whether products exploit predictable behavioural errors through opaque fees, unsuitable risk or misleading presentation.

Artificial intelligence can scan filings, news and alternative data rapidly, but it introduces risks of overfitting, data leakage, unstable relationships, biased training data and crowded trades. A responsible process requires time-stamped data, out-of-sample validation, model-risk governance, explainability appropriate to the use case and human oversight.

7. Illustrative Case: The Delayed Earnings Response

Background. An equity analyst observes that firms reporting unexpectedly strong earnings often continue to outperform for several months.

Problem. The analyst must decide whether to recommend a momentum-based strategy or conclude that the evidence is unreliable.

Analysis. The analyst first defines an earnings surprise using information available before the announcement. Firms are sorted into portfolios and abnormal returns are measured using a pre-specified factor benchmark. The analysis includes bid–ask spreads, commissions, market impact, short-selling constraints and taxes. It uses a training period and a later out-of-sample period. The analyst also tests whether results are concentrated among illiquid firms or small firms.

Approach. A positive post-announcement cumulative abnormal return may indicate underreaction, but it could also reflect omitted risk, gradual information release, delayed analyst revisions or a benchmark error. The analyst therefore compares the strategy with a value-added risk model and reports gross and net results separately.

Outcome. Suppose the gross effect is positive but falls below estimated trading costs after accounting for realistic execution. The appropriate conclusion is not that the pattern never existed; rather, it is not an investable abnormal-profit opportunity for this institution under current conditions.

Learning points. A professional conclusion requires economic significance, not merely statistical significance; a valid information timestamp; controls for risk and costs; and out-of-sample evidence. An anomaly may be real, compensation for risk, or arbitraged away after discovery.

Challenges and Limitations

Market efficiency is difficult to test because expected returns are unobservable. Behavioural explanations can also become unfalsifiable if every anomaly is attributed to a different bias after the fact. Both approaches require disciplined hypothesis formation, robust data and replication. Investors should avoid treating a historical anomaly as a law of nature. Structural changes, regulation, technology, market access and competition can alter its profitability.

Chapter Summary

Market efficiency concerns the speed and completeness with which information enters prices. Weak-form efficiency concerns historical market data; semi-strong efficiency concerns all public information; and strong-form efficiency includes private information. Tests include return predictability studies, event studies and insider-performance analysis, but all depend on an expected-return model and are exposed to data, benchmark and trading-cost problems.

A correct pricing perspective uses discounted, risk-adjusted expected cash flows rather than an undiscounted future value. Similarly, a raw-return market-model intercept is not automatically abnormal performance; Jensen’s alpha is conventionally estimated from excess returns relative to a benchmark, subject to model assumptions.

Behavioural finance explains how heuristics, overconfidence, anchoring, confirmation bias, loss aversion, disposition effects and herding can influence investors and managers. Momentum, reversal, value, size, calendar, issuance and low-volatility patterns are important anomalies, but their economic meaning remains contested. Effective portfolio management combines diversification and cost control with behavioural safeguards, robust research design, ethical conduct and governance.

Key Terms

Market efficiency: The degree to which prices reflect relevant information.

Weak-form efficiency: Prices reflect historical market information.

Semi-strong efficiency: Prices reflect all publicly available information.

Strong-form efficiency: Prices reflect public and private information.

Information set: Information available to a decision-maker at a specified time.

Stochastic discount factor: A state-dependent factor converting future cash flows into current value.

Event study: A method for measuring price responses to a specified event.

Abnormal return: Actual return minus return predicted by a stated benchmark model.

Jensen’s alpha: The excess-return market-model intercept interpreted as risk-adjusted performance under the model.

Joint-hypothesis problem: The difficulty of separating market inefficiency from an incorrect expected-return model.

Heuristic: A mental shortcut used in judgment.

Overconfidence: Excessive belief in one’s knowledge or forecasting ability.

Anchoring: Excessive reliance on an initial reference value.

Confirmation bias: Favouring evidence consistent with an existing belief.

Loss aversion: Greater psychological sensitivity to losses than equal gains.

Disposition effect: Selling winners early and retaining losers too long.

Herd behaviour: Following the actions or beliefs of others.

Momentum: Continuation of intermediate-term relative performance.

Reversal: Later return reversal after extreme prior performance.

Limits to arbitrage: Risks and costs that prevent traders from immediately eliminating mispricing.

Review Questions

Short Answer Questions

1. What is the difference between weak-form and semi-strong efficiency?
2. Why is strong-form efficiency generally considered unrealistic?
3. What does an event study measure?
4. Explain loss aversion and the disposition effect.
5. Why is a raw-return market-model intercept not automatically Jensen’s alpha?

Long Answer Questions

1. Explain the three forms of market efficiency and evaluate the evidence used to test each.
2. Discuss the joint-hypothesis problem and its implications for empirical finance.
3. Explain how overconfidence, anchoring and confirmation bias affect portfolio decisions.
4. Compare momentum, reversal, value and size anomalies.
5. Discuss how disclosure, insider trading controls and investor protection support market integrity.

Analytical / Application Questions

1. A strategy earns a statistically significant gross return based on historical prices but loses money after spreads, commissions and market impact. Evaluate whether this contradicts weak-form efficiency.
2. An event study finds positive abnormal returns for ten days after an earnings announcement. Identify at least four alternative explanations and design tests to distinguish them.
3. A fund manager retains several losing positions because selling would “make the loss real,” while increasing trades after a successful year. Diagnose the behavioural biases and propose governance controls.

Multiple Choice Questions with Answers

Question 1

Which information set is central to weak-form market efficiency?

A. Historical prices and trading volume

B. All public financial statements

C. Private merger negotiations

D. Only macroeconomic forecasts

Correct Answer: A. Historical prices and trading volume

Question 2

In the excess-return market model, Jensen’s alpha is the intercept from regressing:

A. Raw security returns on raw market returns

B. Security excess returns on market excess returns

C. Dividend yield on inflation

D. Accounting earnings on book value

Correct Answer: B. Security excess returns on market excess returns

Question 3

If a security has a 12% return and the risk-free rate is 4%, its excess return is:

A. 4%

B. 8%

C. 12%

D. 16%

Correct Answer: B. 8%

Question 4

Which bias involves seeking information that supports an existing belief?

A. Availability bias

B. Confirmation bias

C. Representativeness

D. Loss aversion

Correct Answer: B. Confirmation bias

Question 5

Selling winning shares quickly while retaining losing shares is called the:

A. Disposition effect

B. Momentum effect

C. Size effect

D. January effect

Correct Answer: A. Disposition effect

Question 6

An event study’s cumulative abnormal return is calculated by:

A. Multiplying all daily returns

B. Adding abnormal returns across the event window

C. Subtracting the risk-free rate from dividends

D. Dividing market return by beta

Correct Answer: B. Adding abnormal returns across the event window

Question 7

Which statement best describes the joint-hypothesis problem?

A. Two investors must hold identical portfolios

B. Efficiency is tested together with an expected-return model

C. All anomalies occur in pairs

D. A market must have two exchanges

Correct Answer: B. Efficiency is tested together with an expected-return model

Question 8

Following others because of reputational pressure or social information is most closely associated with:

A. Herd behaviour

B. Anchoring

C. Tax-loss selling

D. Diversification

Correct Answer: A. Herd behaviour

Question 9

A general present-value model should incorporate:

A. Only the largest future cash flow

B. Undiscounted expected value only

C. Expected cash flows and a risk- and time-sensitive discounting mechanism

D. Historical trading volume only

Correct Answer: C. Expected cash flows and a risk- and time-sensitive discounting mechanism

Question 10

Which practice best reduces look-ahead bias?

A. Using information available only after the trade date

B. Time-stamping data and restricting tests to information available at each decision date

C. Removing unsuccessful observations

D. Selecting the benchmark after seeing results

Correct Answer: B. Time-stamping data and restricting tests to information available at each decision date