The Efficient Market Hypothesis

The Core Idea

The Efficient Market HypothesisDefinitionA theory stating that asset prices fully reflect all available information, making it impossible to consistently earn returns above the market average on a risk-adjusted basis. (EMH) is one of the most consequential ideas in modern finance. At its heart, it makes a deceptively simple claim: the price of a stock already reflects everything that is publicly known about it. If that's true, then no amount of clever analysis can reliably identify "bargains" — because the market has already priced in the information you're looking at.

The hypothesis is closely tied to the concept of a random walkDefinitionA model where future price changes are independent of past changes — each step is random. If stock prices follow a random walk, past movements tell you nothing useful about future direction.. If markets are efficient, tomorrow's price movement depends only on tomorrow's news. Since genuine news is, by definition, unpredictable, price changes themselves must be unpredictable. There is no pattern to exploit.

Burton Malkiel, the Princeton economist who popularised the idea in A Random Walk Down Wall Street (1973), put it memorably: a blindfolded chimpanzee throwing darts at a newspaper's stock listings could select a portfolio that performs as well as one carefully chosen by experts. His practical conclusion was not to literally throw darts — but to buy a low-cost index fundDefinitionA fund that passively tracks a market index (e.g. the S&P 500 or ASX 200) by holding all its constituent stocks. It aims to match market returns at minimal cost, rather than trying to beat them through stock selection. and hold it.

Key Principle

Efficiency doesn't mean markets are always right — it means they don't leave risk-free profit opportunities lying around. Mistakes happen, but they're not systematic or exploitable in advance.

Three Forms of Efficiency

The EMH is typically discussed in three progressively stronger forms, each defined by what kind of information is reflected in prices:

Weak Form

Prices already reflect all past trading data — historical prices, volume, and related statistics. If this holds, technical analysisDefinitionThe practice of studying charts of past price and volume data to identify patterns (e.g. "head and shoulders," "double bottoms") believed to predict future price movements. is useless. Looking at a chart of past prices cannot give you an edge, because whatever information was in those patterns has already been absorbed.

Semi-Strong Form

Prices reflect all publicly available information — not just trading history but also earnings reports, economic data, analyst research, and news. If this holds, fundamental analysisDefinitionAnalysing a company's financial statements, competitive position, industry conditions, and economic environment to estimate its intrinsic value and identify mispriced securities. cannot reliably produce excess returnsDefinitionReturns above what you would expect given the level of risk taken. Also called "abnormal returns" or "alpha." If you earn 12% when a comparable-risk benchmark earns 10%, your excess return is 2%., because the market has already incorporated all the same public data you're examining.

Strong Form

Prices reflect all information — public and private, including insider knowledge. Almost no serious economist believes the strong form holds literally. Insider trading laws exist precisely because private information can confer an advantage.

The Case Against: Predictable Patterns

Since the EMH gained academic dominance in the 1970s, researchers have spent decades trying to poke holes in it. They've found a number of anomaliesDefinitionStatistical patterns in asset returns that appear to contradict the EMH — such as the "January effect," momentum, or the outperformance of "value" stocks. Their existence is debated: they may reflect genuine inefficiency, or may be artefacts of flawed risk models. — patterns in stock returns that seem to offer predictable profits. Here are the most prominent.

Short-Term Momentum

Stocks that have risen recently tend to keep rising over the short term — and vice versa. Lo and MacKinlay (1999) found statistically significant positive serial correlationDefinitionA statistical relationship between a variable and a lagged version of itself. Positive serial correlation in returns means that a stock that rose yesterday is slightly more likely to rise again today. in short-run returns, meaning the market does seem to have some "memory." Behavioural economists attribute this to psychological bandwagon effectsDefinitionThe tendency for people to follow the crowd — buying because others are buying, which pushes prices further in the same direction. A form of herd behaviour. and the tendency for investors to underreact to new information, absorbing it only gradually.

However, the statistical dependence is tiny. Once you account for transaction costsDefinitionAll costs incurred when buying or selling a security: brokerage commissions, bid-ask spreads, market impact, and taxes. These costs eat into any theoretical trading profit. — commissions, spreads, slippage — the theoretical edge evaporates. Odean (1999) found that a sample of momentum traders actually performed worse than buy-and-hold investors, not better.

Long-Run Mean Reversion

Over longer horizons (3–5 years), the pattern flips. Stocks that performed strongly tend to underperform, and past losers tend to recover. Fama and French (1988) found that 25–40% of the variance in long-run returns could be explained by negative correlation with past returns. This is mean reversionDefinitionThe tendency for extreme values (unusually high or low returns) to move back toward the long-run average over time. A stock that has dramatically outperformed may subsequently underperform, and vice versa. — what goes up eventually comes back down.

This supports a contrarianDefinitionAn investment strategy that deliberately goes against the prevailing market sentiment — buying what is hated and selling what is popular, on the assumption that crowds tend to overreact. approach in theory. But in practice, Fluck, Malkiel and Quandt (1997) tested this over 13 years and found that while return reversal was statistically real, both past winners and past losers produced similar forward returns. The pattern existed, but couldn't be turned into actual profit.

Calendar Effects

The January effectDefinitionA historical pattern where stock returns — particularly for small-capitalisation stocks — tended to be abnormally high in the first two weeks of January. Largely attributed to tax-loss selling in December followed by reinvestment. — unusually high returns in early January, especially for small-cap stocks — was one of the most famous anomalies. Day-of-the-week effects (notably higher Monday returns) were also documented. The problem? These effects are small relative to trading costs, and most have weakened or disappeared entirely since being publicised. As Wall Street traders joke, the January effect now happens the previous Thanksgiving.

Value vs. Growth

Stocks with low price-to-earnings ratiosDefinitionShare price divided by earnings per share (P/E). A low P/E can indicate a stock is "cheap" relative to its earnings — or that the market expects earnings to decline. A high P/E may signal expected growth, or overvaluation. and low price-to-book ratiosDefinitionShare price divided by book value per share (P/B). Book value is total assets minus liabilities. A low P/B may indicate a "value" stock trading below the accounting value of its net assets. — so-called "value" stocks — have historically outperformed high-multiple "growth" stocks over long periods. Fama and French (1992) found that size and P/B together explained a substantial portion of return differences across stocks.

Does this prove inefficiency? Not necessarily. Fama and French themselves argued that the value premium might simply be compensation for risk that the standard Capital Asset Pricing ModelDefinitionThe CAPM is a model that relates expected return to systematic risk (measured by beta). Expected return = risk-free rate + beta × market risk premium. It implies the only risk that matters is the portion correlated with the overall market. fails to capture. Companies with low prices relative to book value are often in financial distress — they're cheap for a reason, and holding them carries higher risk.

There's also the real-world test. Actual "value" mutual funds have not consistently outperformed "growth" funds over the full historical record from the 1930s onward. And when Dimensional Fund Advisors launched a fund designed to capture the Fama-French value factor systematically, it generated negative alphaDefinitionThe return of an investment above or below a benchmark, after adjusting for risk. Positive alpha = outperformance; negative alpha = underperformance. The holy grail for active managers. over the 1993–1998 period.

"Many of the predictable patterns that have been discovered may simply be the result of data mining. Given enough time and massaging of data series, it is possible to tease almost any pattern out of most datasets." — Burton Malkiel

The Self-Destruct Mechanism

Here is the elegant irony built into the EMH: if an anomaly is real and profitable, it will attract capital. Traders will rush to exploit it, and in doing so, they will arbitrage it away. The January effect? Once known, investors buy in December and sell in early January — front-running the pattern until it vanishes. Any genuinely exploitable pattern sows the seeds of its own destruction.

III

Bubbles, Crashes, and "Irrational" Markets

Critics of the EMH point to dramatic episodes as proof that markets are sometimes deeply irrational.

The 1987 Crash

In mid-October 1987, the U.S. stock market lost roughly a third of its value in days. How can prices be "efficient" on October 1 and then collapse two weeks later with no obvious change in the economy?

EMH proponents counter that the environment did change. Long-term Treasury yieldsDefinitionThe return earned on a bond if held to maturity, expressed as an annual percentage. When yields rise, existing bond (and often stock) prices fall, because future cash flows are discounted at a higher rate. had risen from about 9% to 10.5% in the weeks prior. Congress threatened a "merger tax" that would curtail takeover activity. Treasury Secretary James Baker signalled willingness to let the dollar fall further. The Gordon Growth ModelDefinitionA formula for valuing a stock as the present value of its future dividends: Price = Dividend / (Required Return − Growth Rate). Small changes in the required return or growth rate can cause large swings in the implied fair price. shows why this matters:

Gordon Growth Model r = D/P + g → P = D / (r − g)

If the required return r rises from 11% to 13% while dividends and growth are unchanged, the model implies a price drop from $100 to roughly $67 — a one-third decline. No irrationality required; just a modest repricing of risk and discount rates.

The Internet Bubble

The late 1990s saw Internet stocks reach valuations that, with hindsight, were clearly absurd. Even Malkiel — a staunch EMH defender — conceded that the high-tech sector exhibited bubble-like behaviour. But he made an important distinction: there were no exploitable arbitrage opportunities available to rational investors at the time.

Why not? Because short sellingDefinitionSelling borrowed shares with the expectation of buying them back cheaper later. If the price falls, you profit; if it rises, your losses are theoretically unlimited. Requires borrowing the stock, which isn't always possible. an overvalued stock is extremely risky. A stock trading at twice its fundamental value might go to three times before it corrects. Professional arbitrageursDefinitionTraders who exploit price discrepancies between related securities. In theory, they push prices back to rational levels. In practice, capital constraints, borrowing costs, and timing risk limit their ability to correct mispricings. operate on short time horizons — if their clients see interim losses, they withdraw capital, forcing positions to be closed at the worst moment. Shleifer (2000) called this noise trader riskDefinitionThe risk that irrational traders push prices even further away from fundamental value before any correction occurs. This makes it dangerous for rational arbitrageurs to bet against mispricings, even when they're confident they exist. — the danger that irrational traders push prices further from fair value before any correction.

Important Distinction

A market can be "wrong" in hindsight and still be efficient in the sense that matters to investors — if the mispricing cannot be reliably identified and profited from in advance.

The Professional Track Record

For proponents of the EMH, the strongest evidence is the simplest: the people paid to beat the market usually can't.

The research is extensive and consistent. Over the decade ending 2001, roughly three-quarters of actively managed U.S. equity funds underperformed the S&P 500DefinitionAn index of 500 large-cap U.S. companies, widely regarded as the benchmark for U.S. equity performance. Comparable to the ASX 200 in Australia.. Over 20-year horizons, the median large-cap fund lagged the index by nearly two percentage points per year — almost entirely attributable to higher fees.

Worse still, there is virtually no persistenceDefinitionThe tendency (or lack thereof) for a fund's past performance to predict its future performance. If persistence is absent, a fund's strong recent results tell you nothing about whether it will continue to outperform. in outperformance. The top 20 funds of the 1970s, which returned nearly double the index, collectively underperformed in the 1980s. The 1980s' top performers lagged in the 1990s. The late-1990s stars — who tripled the index return in 1998–99 — did three times worse in 2000–01.

Of 355 equity funds in existence in 1970, only five outperformed the market by two or more percentage points per year over the next three decades. That's 1.4%. You'd expect a similar hit rate from pure chance. And you still face survivorship biasDefinitionA statistical distortion that occurs when poorly performing funds are merged or liquidated, removing their bad records from the dataset. This makes the surviving funds' average performance look better than the true average of all funds that existed.: 197 of those original 355 funds didn't even survive the period. Their poor records vanished from the data.

"I have personally tried to invest money, my client's money and my own, in every single anomaly and predictive device that academics have dreamed up… And I have yet to make a nickel on any of these supposed market inefficiencies." — Richard Roll, financial economist and portfolio manager

What This Means for Investors

The EMH doesn't claim markets are perfect. It claims they're efficient enough that you can't reliably profit from their imperfections after accounting for costs. The practical implications are significant:

Anomalies exist — but they're not free money. Short-term momentum, mean reversion, value premiums, calendar effects — the academic literature has documented them all. But once transaction costs, taxes, and timing risk are factored in, the exploitable edge shrinks to near zero. And once a pattern is widely known, the arbitrage mechanism tends to eliminate it.

Professional fund managers mostly don't add value. The data on this is remarkably consistent across decades and geographies. The median active fund underperforms its benchmark by roughly the amount of its fees. Past outperformance doesn't predict future outperformance.

Markets are fast, but not flawless. The EMH allows for temporary mispricings — sometimes large ones, as during the Internet bubble. What it denies is that any investor can systematically identify these in advance and profit from them. As Benjamin Graham wrote, the market may be a voting machine in the short run, but it is a weighing machine in the long run.

The boring advice remains the best advice. Buy a broad, low-cost index fund. Minimise fees and turnover. Stay diversified. Don't try to time the market. This was Malkiel's recommendation in 1973, and the subsequent fifty years of evidence have only strengthened it.

The $100 Bill Test

A finance professor and a student spot a $100 bill on the ground. The student reaches for it. The professor says, "Don't bother — if it were real, someone would have already picked it up." That, in essence, is the Efficient Market Hypothesis: if there were easy money to be made, it would already be gone.

◆ ◆ ◆