Why markets are mostly efficient and people mostly are not

There is an apparent contradiction at the heart of behavioural finance. The evidence that individual investors make systematic, repeatable errors is overwhelming. The evidence that market prices are hard to beat is also overwhelming. Both cannot be surprising at the same time — if everyone is predictably wrong, why is the price so stubbornly right?

The resolution is the single most useful idea in this course, and it is worth getting exactly right before we spend eleven lessons cataloguing human error. Market efficiency is a claim about prices, not about people. A price is not an average of opinions. It is the point at which the marginal buyer and the marginal seller, each risking real capital, stop wanting to trade. Errors that are random across a population cancel in that aggregation. Errors that are correlated across a population do not — and the whole discipline of behavioural finance is the study of when they correlate.

What "efficient" actually claims

A market is informationally efficient with respect to a given information set if prices already reflect that information, so no strategy based on it earns an abnormal return after costs and after adjusting for risk.

Eugene Fama's 1970 survey organised this into three nested versions. Weak form efficiency says past prices and volumes are already reflected, so pure chart-reading cannot earn an abnormal return. Semi-strong form says all publicly available information is reflected — annual reports, SNB policy statements, ad hoc disclosures on SIX. Strong form says even private information is reflected, which is plainly false, and is enforced against by insider trading law rather than by economics.

The mechanism is competition, not virtue. Suppose a Swiss industrial announces that a large order has been cancelled. Nobody needs to poll investors. A handful of analysts and traders who follow the company update their estimates, act on them, and their buying or selling pressure moves the quote within seconds. It takes very few informed participants to set a price, because price is set at the margin. The remaining thousands of holders can be inattentive, biased or asleep, and the price will still be approximately correct.

Paul Samuelson made the deeper point in 1965: if prices already incorporate everything anticipated, then what moves them next must be, by definition, the unanticipated part. Price changes therefore look close to random — not because markets are chaotic, but precisely because they are well-anticipated. The random walk is a consequence of efficiency, not an assumption behind it.

Why "mostly" is the load-bearing word

Perfect efficiency is logically impossible, and Sanford Grossman and Joseph Stiglitz proved it in 1980. Gathering information costs money. If prices already reflected everything, nobody would be paid for gathering information, so nobody would gather it, so prices could not reflect it. Equilibrium requires that informed traders earn just enough to cover their research costs. Markets must be slightly inefficient in order to be mostly efficient. The mispricing that persists is roughly the size of the cost of finding it.

That has a direct implication for you. The residual inefficiency is not free money lying on the pavement. It is the wage paid to people with better data, better models and lower trading costs than you have, for doing work you are not doing.

There is a second complication, known as the joint hypothesis problem. You can never test efficiency on its own. Any test asks whether returns were abnormal, and "abnormal" requires a model of what normal is. If you find that some group of stocks outperformed, you cannot separate "the market was wrong" from "my risk model was incomplete". This is why decades of anomaly research remain contested: value and momentum premia might be mispricings, or they might be compensation for risks the model omits.

Third, arbitrage has limits. In textbooks, mispricing is eliminated by traders who buy the cheap thing and short the expensive one at no risk. In practice, as Andrei Shleifer and Robert Vishny set out in 1997, the arbitrageur manages other people's money, faces margin calls, and can be forced to close a correct position at the worst moment. Two documented cases make the point vividly. Royal Dutch and Shell were legally entitled to a fixed split of the same cash flows, yet traded at persistently different relative valuations for years. When 3Com floated a small stake in Palm in 2000, the market briefly valued 3Com's remaining Palm shares at more than the whole of 3Com. These were not subtle errors, and they were not arbitraged away quickly, because doing so required capital that could survive being wrong for a long time.

So: mostly efficient. Prices are usually a very good summary of available information, occasionally a bad one, and you cannot easily tell which regime you are in from the inside.

Where the inefficiency actually lives: in you

The empirical record on individual investors is not ambiguous. Brad Barber and Terrance Odean's study of tens of thousands of US discount brokerage accounts in the 1990s found the average household underperformed a simple market benchmark by a modest margin, while the most active fifth of traders trailed by several percentage points a year. The driver was not stock selection alone. It was turnover. Trading itself was the cost.

The pattern repeats among professionals. The SPIVA scorecards published by S&P Dow Jones Indices, which track active funds against their benchmarks in Europe including Switzerland, consistently show that a large majority of active equity funds trail their index over ten-year horizons once fees are included. This is arithmetic before it is psychology: in aggregate, all investors hold the market, so the average actively managed franc must earn the market return before costs and less than it after costs.

Your disadvantage relative to a market price is structural, not a personality flaw:

• The price aggregates thousands of independent estimates. You have one.
• The price updates continuously. You update when you happen to look.
• Errors cancel across a crowd. Yours accumulate in the same direction, because your biases are stable — which is the subject of the rest of this course.

The Swiss frictions that make trading worse than it looks

Switzerland changes the cost arithmetic in ways worth stating explicitly.

Turnover tax (Umsatzabgabe). Swiss securities dealers levy stamp duty of 0.15% on Swiss securities and 0.30% on foreign ones, split between the two contracting parties, so a private client typically sees 0.075% or 0.15% charged on each trade. It is charged on the transaction, win or lose.

Work it through. Take a CHF 200,000 portfolio of foreign equities and turn it over once in a year. That is CHF 200,000 sold plus CHF 200,000 bought, so CHF 400,000 of transaction value. At the client's 0.15% share, stamp duty alone is CHF 600. Add brokerage of, say, 0.2% a side — CHF 800 — and the bid-ask spread you cross twice. You are into the region of CHF 1,400 to CHF 2,000, roughly 0.7% to 1.0% of the portfolio, paid for the privilege of having changed your mind. Your new holdings must beat your old ones by that margin just to break even.

Depth, and where the enforcement mechanism gets slow. Efficiency is not a uniform property of "the market"; it is a function of how many informed participants are competing over a given security and how much capital stands ready to trade it. The Swiss franc bond market is small by global standards next to the dollar and euro markets, and much of it trades thinly and by appointment rather than on a continuous order book. The same is true at the bottom of the SIX equity list. In those segments the spread you cross is wider, the price you see may be hours old, and the arbitrage machinery that makes the efficiency argument work has fewer hands on it. This cuts both ways: thin markets are where genuine mispricing is most likely to survive, and also where it is most expensive to act on. Assuming a small, illiquid holding can be exited at the last quoted price is one of the more reliable ways to be surprised.

Home bias. Swiss investors hold far more Swiss equity than Switzerland's low single-digit share of global market capitalisation would justify. And Swiss equity is concentrated: a handful of large caps in pharmaceuticals and food dominate the index. A "diversified Swiss portfolio" can be a bet on three companies and one currency.

Currency. The franc has a long history of real appreciation. A foreign asset that performs well in dollars or euros can still lose value in francs, which means your reported result mixes two decisions — the asset and the currency — that deserve to be evaluated separately.

What follows from all this

The practical conclusion is not "markets are perfect, do nothing". It is a shift in where the burden of proof sits.

Treat the market price as a strong prior — the pooled estimate of everyone paying attention, backed by their money. Disagreeing with it is allowed, but it requires an answer to a specific question: what do I know, or what am I willing to bear, that the marginal price-setter does not? Legitimate answers exist. You may have a genuinely longer horizon than a fund manager measured quarterly. You may be willing to hold illiquidity. You may know an industry from the inside. "I read an article and it seemed compelling" is not on the list.

And notice which side of the ledger you actually control. You cannot make the market less efficient. You can make yourself less inefficient — by lowering turnover, cutting costs, diversifying, and building a process that survives your own emotions. The remaining lessons in this course are that project. The next one starts with the asymmetry that does the most damage: loss aversion.

What to take away

• Efficiency is a property of prices set at the margin by informed capital, not a claim that investors are rational; a market can be hard to beat while most of its participants are behaving badly.
• Grossman and Stiglitz showed perfect efficiency is self-defeating — the surviving mispricing is roughly the size of the cost of finding it, which is why it is a wage for research rather than free money.
• The joint hypothesis problem means every claim of "the market is wrong" is also a claim about your risk model, and you usually cannot separate the two.
• Limits to arbitrage explain why obvious mispricings such as Royal Dutch/Shell and 3Com/Palm persisted: correcting them required capital that could survive being right too early.
• In Switzerland, stamp duty of 0.075% to 0.15% per trade, plus commissions and spreads, can cost most of a percent a year at full annual turnover — and in thin segments such as the franc bond market or the small end of SIX, the wider spread and slower price discovery mean both that mispricing survives longer and that acting on it costs more.

Sources

• Eugene F. Fama, "Efficient Capital Markets: A Review of Theory and Empirical Work", Journal of Finance, 1970.
• Paul A. Samuelson, "Proof That Properly Anticipated Prices Fluctuate Randomly", Industrial Management Review, 1965.
• Sanford J. Grossman and Joseph E. Stiglitz, "On the Impossibility of Informationally Efficient Markets", American Economic Review, 1980.
• Andrei Shleifer and Robert W. Vishny, "The Limits of Arbitrage", Journal of Finance, 1997.
• Kenneth A. Froot and Emil M. Dabora, "How Are Stock Prices Affected by the Location of Trade?", Journal of Financial Economics, 1999 — the Royal Dutch/Shell case.
• Owen A. Lamont and Richard H. Thaler, "Can the Market Add and Subtract? Mispricing in Tech Stock Carve-Outs", Journal of Political Economy, 2003 — the 3Com/Palm case.
• Brad M. Barber and Terrance Odean, "Trading Is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors", Journal of Finance, 2000.
• S&P Dow Jones Indices, SPIVA Europe Scorecard (published semi-annually), for active-versus-index results including Swiss equity funds; and the Swiss Federal Act on Stamp Duties for the turnover-tax rates on domestic and foreign securities.