Week 4: The Efficient Market Hypothesis

1. Why the Efficient Market Hypothesis matters

The subject of this lecture is the Efficient Market Hypothesis (EMH), the yardstick that behavioral finance as a whole positions itself against. The unit pursues three goals: explaining the core principles and assumptions of the EMH, following how it developed over time and what Eugene Fama contributed, and seeing why information sits at the heart of financial markets.

The EMH is one of the most influential frameworks in modern finance. It shaped how economists think about prices, returns and investment management, and it underlies valuation models such as the CAPM (Capital Asset Pricing Model) and arbitrage pricing models (APT). At the same time, it has been heavily criticised because it struggles to explain crises and bubbles. The central idea to remember is that the EMH sees the market as an information-processing mechanism: millions of traders each hold a small piece of information, and trading pushes all those pieces into one number, the price.

2. The historical roots: prices as information

George Gibson (1889): the stock exchange as a ballot box

George Gibson, writing about the stock exchanges of London, Paris and New York, claimed that once shares trade publicly on an open market, their price may be regarded as the best possible judgment available. He compared the exchange to a ballot box: someone who thinks a stock deserves a higher price buys it (a "vote" for a higher price), and someone who thinks it deserves a lower price sells. The resulting price is the result of the vote.

Charles Conant (1904): price as information

Charles A. Conant, in The Principles of Money and Banking (1904), wrote that without the information that prices supply we would be moving around blind. Prices work as signals that steer economic activity. Conant regarded markets as a decentralised communication network through which investors pass on what they know by trading. This idea anticipates the modern EMH.

The Soviet economy: what happens without prices

The Soviet Union is the opposite case. In a centrally planned economy, prices were set by bureaucrats, not by supply and demand. Without market prices, the system loses information: planners do not know which goods are scarce and which are in demand. The Soviet economy "operated in the dark". The lesson on the slide: the market, despite its imperfections, remains the most efficient mechanism we know for aggregating information.

The information industry: from pigeons to Reuters

Around the middle of the 19th century, information was the main source of competitive advantage. Early traders relied on carrier pigeons to move prices from one exchange to another. Paul Julius Reuter set up Reuters in 1851, at first with pigeons and afterwards with telegraph cables, making it the first firm to sell the speed of information as a business. The lesson is that the market has never been "perfect": it has always been a competition for information. A trader who learned a Paris price ahead of others in London could profit from it.

The era of instant information

Market efficiency depends on how accessible and how fast information is. With the Internet, investors got real-time access to prices, news and analysis, and information stopped being the privilege of a few institutions. But speed does not guarantee a rational reaction: fast information can also amplify crowd psychology. Technology made markets more efficient in processing news, but it also increased the potential for overreaction.

From the beeper to the smartphone

Brokers in the 1980s carried beepers so their offices could send them short messages fast. That marked the start of mobile information, and the speed of transmission turned into a competitive advantage. The reasoning matches the EMH: information has to reach every participant at once. The beeper has since given way to the smartphone, yet investor psychology has stayed the same. Faster communication alters how efficient the market appears, but not necessarily the way people act.

The lecturer closes this part by noting that the EMH is more than a theory: it also mirrors the historical progress in information dissemination. Every new technology (pigeon, telegraph, beeper, Internet, smartphone) cut the delay between an event and the moment prices reflected it.

3. Assumptions and formulation of the EMH

The information assumption

The core assumption is that new information that changes expected cash flows or discount rates shows up in prices at once. Remember that a share is worth the present value of the cash flows it is expected to generate, discounted at a rate that reflects its risk. News is relevant only when it alters one of those two inputs. The slide lists three supporting assumptions:

• Riskless profits get competed away, so no arbitrage opportunities persist.
• Investors behave rationally and make use of the data available.
• Information is public and accessible, and spreading it costs nothing.

A consistent implication: if information is already known, it cannot generate excess returns. If everybody already knows that a company will report higher profits, that knowledge is already in today's price.

Fama's formulation (1970)

Eugene Fama (1970) gave the classic definition: "In an efficient market, security prices fully reflect all available information." Three interpretations follow:

• Prices follow a random walk: future price changes cannot be predicted from available information.
• Once risk is accounted for, the expected excess return equals zero. Taking on more risk can raise your return, but it cannot raise your risk-adjusted return.
• Deviations come either from luck or from new information. When a fund outperforms the market, that is either chance or the arrival of truly new information.

The three forms of efficiency

Every form contains the one before it. A stronger form leaves less scope for excess returns. Most researchers agree that markets come close to weak-form efficiency in the short run, argue about the semi-strong form (where the bulk of the anomalies are found), and dismiss the strong form (insiders do make abnormal returns, which is the reason insider trading is against the law).

4. Testing the EMH

The EMH as a null hypothesis

In terms of method, the EMH serves as a null hypothesis: researchers take it as true and search for evidence that contradicts it. Testing happens via the predictability of returns. When returns cannot be predicted at all, the market counts as efficient. The majority of tests rest on the random walk idea. The reasoning is straightforward: when information is absorbed at once, price changes have to be random, since only new (and so unpredictable) information can move prices.

Market reaction speed

When markets are efficient, prices react immediately to new information. Any lag in the response would create a chance to profit, and so would signal inefficiency: if it took a price three days to absorb good news, you could buy once the news came out and gain from the gradual drift. The slide presents a Bloomberg intraday chart for 9 April 2025, on which the market shot up nearly vertically within a few minutes of a big news announcement and then held the higher level. A quick adjustment like this is taken as evidence of efficiency.

The random walk idea

When new information cannot be predicted, price changes cannot be predicted either, so prices follow a random walk. On the slide are 100 simulated Gaussian random walks: every path begins at zero and they fan out randomly as time passes, with some finishing well above the starting point and some well below. The history of a path tells you nothing about where it goes next.

Shiller's interpretation of the random walk

Robert Shiller emphasises that a random walk does not imply a "random market" where prices carry no meaning. What it implies is that all information has already been absorbed. Anyone able to forecast prices would trade on that forecast, and the market would take it in right away, so trying to predict is exactly what makes prices unpredictable. In this sense the EMH functions as a self-fulfilling equilibrium: the more investors trust in efficiency and search for mispricing, the more efficient the market gets.

The CRSP era

In the 1960s the University of Chicago set up CRSP (Center for Research in Security Prices), which supplied researchers with clean databases of US stock prices and made thousands of studies possible. These studies checked whether past price movements could forecast later returns. The first findings showed hardly any predictability, which reinforced faith in the EMH. Statistical return analysis turned into the main instrument for supporting the theory.

The joint hypothesis problem

Every test of the EMH needs a model of what the "normal" return should be (for example the CAPM). If we find abnormal returns, there are two possible explanations: the market is inefficient, or the pricing model is wrong. We cannot separate the two. This is the Joint Hypothesis Problem: we never test efficiency alone, but always efficiency together with an assumed model of expected returns.

This problem returns throughout the course: whenever a behavioral anomaly is found (momentum, value, post-earnings announcement drift), rational economists can argue that it is a missing risk factor. For your project, it means that any test of mispricing should use several benchmark models (CAPM, Fama-French three-factor, Carhart four-factor) and show that the result survives all of them.

Efficiency and portfolio management

When the EMH is true, fund managers are unable to beat the market on a systematic basis, and any extra return fits statistically with luck. According to Shiller, this conclusion gave rise to passive management: index funds and ETFs that just hold the entire market at very low cost. If beating the market is impossible, owning it cheaply is the smartest approach. The theory thus reshaped the modern investment industry in a big way.

5. Deviations from the EMH

The Medallion Fund vs the S&P 500 (1988 to 2018)

The Medallion Fund of Renaissance Technologies (a quantitative hedge fund) is the most famous challenge to the claim that nobody can systematically beat the market. The slides compare its returns with the S&P 500, and how $100 invested in 1988 would have grown.

Notice two things. First, Medallion had only one negative year (1989: −3.20%) in 31 years, and it earned its highest returns when the market crashed: +98.48% in 2000 (S&P −10.12%) and +83.38% in 2008 (S&P −38.49%). Second, this is not a one-year lucky streak: such consistency over three decades is extremely hard to explain as luck, which is what the EMH would require ("any excess return is consistent with luck"). Defenders of the EMH would point out that the fund is closed to outside investors, that it trades at very high frequency, and that it is a single survivor among thousands of funds. Still, it shows that some patterns in prices can be exploited.

The first signs of deviation

Despite the success of the early tests, patterns appeared that the EMH could not explain:

• Returns showed short-term positive dependence (momentum): stocks that went up tended to keep going up over the following months.
• Returns showed long-term reversal: stocks that went up a lot over several years tended to fall back later (the overreaction evidence of De Bondt & Thaler, 1985, covered in other weeks).
• Prices often overreacted to news.

These deviations were the first crack in the absolute form of the theory, because both momentum and reversal mean that past prices help predict future prices, which violates even the weak form.

The excess volatility problem: Shiller (1981)

Robert Shiller (1981) showed that stock prices are far more volatile than fundamentals justify. The argument: under the EMH, today's price equals the market's best forecast of the present value of future dividends. A forecast should vary less than the thing being forecast, because a good forecast smooths out surprises. Actual dividends of the S&P index move very little and very smoothly over time, yet prices fluctuate dramatically. So prices move more than any rational forecast of dividends could justify.

Excess volatility in practice: Beyond Meat (BYND)

The slides illustrate excess volatility with three Bloomberg charts of Beyond Meat (ticker BYND), the plant-based meat company. After its listing in 2019, the share price rocketed to a high of about $234.90 (July 2019) and then swung up and down by large amounts within months. Over 2019 to 2025 the price fell almost continuously to around $2 (a low of $0.52 in October 2025), and in October 2025 there was a sudden spike in price and trading volume. The company's actual sales and cash flows did not change anywhere near as dramatically as these price movements. The lecturer's interpretation: the difference is attributed to psychology and investor expectations: first a hype story about the future of food, later disappointment, and then a speculative short-lived rally.

Investor behavior

Investors' responses to information are not always rational. We see overreaction, overconfidence and herd behavior. Psychology produces trends that push prices away from fundamentals for a while. Shiller's view is that the EMH is a powerful theoretical framework, yet it falls short of capturing how real markets actually move.

Summary of the first decades of research

• What the EMH gets right is the rapid reaction of prices when new information arrives.
• What it does not account for is why anomalies and excess volatility keep existing.
• Markets are "approximately efficient", but not perfect.

This leads to the transition from statistical efficiency to the psychological effects that shape the market.

6. How textbooks changed their view

In the 1970s, the majority of university textbooks treated the EMH as a fundamental law of markets. The notion that nobody can "beat" the market became close to dogma, and courses concentrated on rational models like the CAPM and the APT. Shiller observes that the EMH came to stand for "scientific maturity": it suggested that economics had discovered the law of markets in the way physics had discovered the law of gravity.

In 1984 a widely used textbook described the EMH as a principle beyond dispute. Students learned that "the market is efficient, period". Psychological factors and anomalies were barely mentioned, and the text claimed that "security prices exactly reflect their value". The EMH was handled as a finished theory instead of one view among several.

Newer editions of that textbook took a different tone. The EMH is now described as a "useful approach" rather than an absolute law, and the authors add that further research is required to explain why security prices at times depart from the discounted cash flows we compute. The reading: the science adjusted to what the data showed. Markets are frequently efficient, though not perfectly so.

7. Technical analysis

Technical analysis before the EMH

Technical analysis is the attempt to forecast prices using patterns in historical prices and trading volume (charts). Until the EMH took over, it was the dominant way to read markets. Analysts thought prices moved in recurring patterns that exposed what investors expected. Shiller points out that before 1970 the claim that prices move at random looked almost ridiculous: everyone was convinced that "charts tell a story". Keep in mind that the weak form of the EMH states precisely that technical analysis is useless.

Psychological barriers

In the 1960s, as the Dow Jones index hit 1,000 points, technical analysts treated that level as a psychological barrier. Many investors reasoned: "1,000 points? It cannot be worth that much, I will sell." The index stayed near 1,000 for a long time without getting past it, and when it finally did, it quickly dropped back. A round number carries no economic significance, which makes this a textbook case of anchoring on an arbitrary reference point.

Resistance and support: the Nasdaq

The slide shows a Nasdaq chart with two horizontal lines. The upper line at about 25,370 is a resistance level: the index reached it several times and turned down each time, as if sellers were waiting there. The lower line at about 24,400 is a support level: the index fell to it several times and bounced back up, as if buyers were waiting there. Technical analysts trade on these levels: buy near support, sell near resistance, and expect a strong move when a level is "broken".

Patterns everywhere

The slide shows a catalogue of twenty chart patterns, each labelled bearish (signalling a fall) or bullish (signalling a rise): double top and double bottom, head and shoulders and inverted head and shoulders, rising and falling wedges, expanding triangles, triple tops and bottoms, flags, pennants, descending triangles and symmetrical expanding triangles. The point is that with enough named patterns, you can "see" a pattern in almost any price chart.

The head and shoulders pattern

One of the most famous patterns. It signals a trend reversal from bullish to bearish. It consists of a left shoulder (a peak), a higher peak called the head, and a right shoulder (a lower peak similar to the left one). The line connecting the low points between them is the "neckline"; when the price falls below the neckline, technical analysts expect a further decline. Technical analysis assumes such patterns have predictive value.

The role of technical analysis in modern finance

Even though its scientific footing is questionable, technical analysis is still used a lot in practice. A collective belief in such patterns can move markets, acting as a self-fulfilling expectation: when many traders see 24,400 as support and all put their buy orders at that level, those orders are what produce the support. So technical analysis matters socially: it shapes behavior even when it holds no real information. In sum, the EMH dismisses technical analysis as a way to forecast, while behavioral finance treats it as a topic for studying what people believe.

8. From efficiency to investor behavior

The majority of departures from the EMH come from the same source: human psychology. Investors do not process information like machines; emotions, social pressure and stories all affect them. The EMH leaves out the cognitive biases that cause prices to swing too far. As Shiller puts it, prices capture not just information but human stories as well.

Overreaction and underreaction

• Overreaction: the price moves too far in response to news and later partly reverses.
• Underreaction: the price responds too slowly, and keeps drifting in the direction of the news for weeks or months.
• Investors amplify price movements through imitative behavior (copying others).

Both create patterns that are inconsistent with a random walk: overreaction leads to reversals (negative correlation of returns over time), underreaction to drift and momentum (positive correlation).

Example: "Beautiful Tariffs" (April 2025)

On the slide is an intraday Bloomberg chart of the S&P 500 covering 9 to 11 April 2025, the period of the US tariff announcements. On 9 April the index leapt within minutes of a tariff-related statement. On 10 April much of that gain disappeared, and on 11 April the index crept higher once more. According to the lecturer, the reversal reveals that the first move was emotional rather than purely "rational". Set this beside the earlier slide on how fast prices react: one and the same type of chart displays a quick response (in line with the EMH) as well as an overshoot and correction (in line with overreaction).

Human psychology in markets: Shiller's four tendencies

As the slide sums it up: the EMH overlooks that investors are people, not machines.

Questioning "perfect rationality"

The EMH depends on the premise that market participants are completely rational. Shiller demonstrates that the evidence says otherwise: people respond emotionally, are swayed by social narratives, and prices frequently stay away from fundamentals for extended periods. Big bull markets stem not just from information but also from collective optimism.

The bubble paradox

Of all departures from the EMH, bubbles are the most dramatic. People buy since they notice others buying, not because fundamentals support the price. Higher prices in turn strengthen the conviction that "this time is different", pulling in yet more buyers in a feedback loop. According to the EMH, bubbles could not happen, since rational traders would keep selling overvalued assets until prices fell back to their true value.

Example: "Big in Japan"

The slide shows the Japanese TOPIX index from 1981 to 2025. During the late 1980s the index rose to a peak of roughly 2,900 at the end of 1989, driven by a stock and real estate bubble. It then collapsed and spent more than three decades below that peak, with lows near 700 in the early 2000s and in 2012. Only in recent years did it surpass the 1989 level (around 3,269 in October 2025). If the 1989 price had correctly reflected fundamentals, investors would not have had to wait 35 years to get their money back in nominal terms. The lesson: market psychology often dominates fundamentals.

The concept of "mass psychology"

According to Shiller, investing is a social activity rather than a private one. Information travels via stories, copying and trust, and not just through data. The result is a collective sentiment strong enough to shift whole markets, a social force the EMH leaves out.

From information to emotion

Having information does not guarantee rational behavior. Investors decide for themselves which events count as "important", and emotions lead to asymmetric reactions: a loss weighs more heavily than a gain of the same size (loss aversion, from prospect theory, Kahneman & Tversky 1979). This psychological weighting is why markets at times respond far too strongly to negative news.

From the EMH to "Narrative Economics"

Shiller proposes that narratives that spread like epidemics steer economic decisions: someone "catches" a story, hands it on, and it may go viral before dying out. Such stories mould expectations and the mood of the crowd. The EMH disregards this way of spreading and looks at data alone. Examples: tales of "quick profits" in crypto and in dot-com stocks caused excesses in both markets.

9. The random walk

Pearson's drunk man

Karl Pearson, the founder of modern statistics, coined the term "random walk", and he clearly was not thinking about finance. He illustrated unpredictability with a drunk man who takes each step in a random direction, so his path follows no pattern at all. Imagine you find him standing under a streetlight right now. Where will he be 10 minutes from now? Because the direction of his steps gives you nothing to go on, your best guess is in the same spot: a step left is as likely as a step right, so on average they offset each other. For stocks this means that today's price is the best prediction of tomorrow's.

Two models for prices: the random walk versus AR(1)

The random walk is the special case of the AR(1) with α = 0 and ρ = 1. When ρ is below 1, a shock does not last forever: part of it fades each period, so the series is pulled back to its mean. The closer ρ is to 1, the slower that pull.

Why this matters for investors: under a random walk, a stock that has fallen a lot is not "due" for a rebound. Under mean reversion it is, which would make contrarian strategies (buying past losers) profitable. Momentum and reversal anomalies are, in effect, evidence that returns are not a pure random walk.

Random walk vs the S&P 500

The slide plots the natural log of the S&P 500 from the late 1920s to the 2020s next to a simulated random walk with drift. Both lines rise from about 3 to about 8 or 9 over the century, with similar wiggles, crashes and recoveries, and they are hard to tell apart. The lesson: it is easy to "fool yourself" into seeing patterns in what may be pure chance (a reference to Nassim Taleb's book Fooled by Randomness). The many chart patterns of technical analysis can appear in a purely random series.

So... is the EMH wrong?

• It is not quite right to assume that markets move as a pure random walk.
• Markets exhibit trends.
• Chances to profit usually lie in the longer-term: ρ is near 1 but not exactly 1. Mispricing fades slowly, so you cannot see it from one day to the next, yet it adds up over years.
• Shiller's verdict: "It's a half-truth. Remember it, but don't put all your faith in it."