Efficient Capital Markets
By Steven L. Jones and Jeffry M. Netter
The efficient markets theory (EMT) of financial economics states that the price of an asset reflects all relevant information that is available about the intrinsic value of the asset. Although the EMT applies to all types of financial securities, discussions of the theory usually focus on one kind of security, namely, shares of common stock in a company. A financial security represents a claim on future cash flows, and thus the intrinsic value is the present value of the cash flows the owner of the security expects to receive.1
Theoretically, the profit opportunities represented by the existence of “undervalued” and “overvalued” stocks motivate investors to trade, and their trading moves the prices of stocks toward the present value of future cash flows. Thus, investment analysts’ search for mispriced stocks and their subsequent trading make the market efficient and cause prices to reflect intrinsic values. Because new information is randomly favorable or unfavorable relative to expectations, changes in stock prices in an efficient market should be random, resulting in the well-known “random walk” in stock prices. Thus, investors cannot earn abnormally high risk-adjusted returns in an efficient market where prices reflect intrinsic value.
As Eugene Fama (1991) notes, market efficiency is a continuum. The lower the transaction costs in a market, including the costs of obtaining information and trading, the more efficient the market. In the United States, reliable information about firms is relatively cheap to obtain (partly due to mandated disclosure and partly due to technology of information provision) and trading securities is cheap. For those reasons, U.S. security markets are thought to be relatively efficient.
The informational efficiency of stock prices matters in two main ways. First, investors care about whether various trading strategies can earn excess returns (i.e., “beat the market”). Second, if stock prices accurately reflect all information, new investment capital goes to its highest-valued use.
French mathematician Louis Bachelier performed the first rigorous analysis of stock market returns in his 1900 dissertation. This remarkable work documents statistical independence in stock returns—meaning that today’s return signals nothing about the sign or magnitude of tomorrow’s return—and this led him to model stock returns as a random walk, in anticipation of the EMT. Unfortunately, Bachelier’s work was largely ignored outside mathematics until the 1950s. One of the first to recognize the potential information content of stock prices was John Burr Williams (1938) in his work on intrinsic value, which argues that stock prices are based on economic fundamentals. The alternative view, which dominated prior to Williams, is probably best exemplified by John Maynard Keynes’s beauty contest analogy, in which each stock analyst recommends not the stock he thinks best, but rather the stock he thinks most other analysts think is best. In Keynes’s view, therefore, stock prices are based more on speculation than on economic fundamentals. In the long run, prices driven by speculation may converge to those that would exist based on economic fundamentals, but, as Keynes noted in another context, “in the long run we are all dead.”
Stock returns and their economic meaning received scant attention before the 1950s because there was little appreciation of the role of stock markets in allocating capital. This oversight had several contributing factors: (1) Keynes’s emphasis on the speculative nature of stock prices led many to believe that stock markets were little more than “casinos,” with no essential economic role; (2) many economists during the Great Depression and the immediate post–World War II era emphasized government-directed capital investment; and (3) the modern corporation, and the resulting need to raise large sums of capital, was a relatively recent development. But the invention of computing power in the 1950s, which made rigorous empirical analysis with large data sets more feasible, brought renewed attention from academic researchers.
In 1953, British statistician Maurice Kendall documented statistical independence in weekly returns from various British stock indices. Harry Roberts (1959) found similar results for the Dow Jones Industrial Index, and later, Eugene Fama (1965) provided comprehensive evidence not only of statistical independence in stock returns, but also that various techniques of “chartists” (i.e., technical analysts) had no predictive power. While this evidence was generally viewed as supporting the random walk model of stock returns, there was no formal understanding of its economic meaning, and some mistakenly took this randomness as an indication that stock returns were unrelated to fundamentals, and thus had no economic meaning or content. Fortunately, the timely work of Paul Samuelson (1965) and Benoit Mandelbrot (1966) explained that such randomness in returns should be expected from a well-functioning stock market. Their key insight was that competition implies that investing in stocks is a “fair game,” meaning that a trader cannot expect to beat the market without some informational advantage. The essence of the “fair game” is that today’s stock price reflects the expectations of investors given all the available information. Therefore, tomorrow’s price should change only if investors’ expectations of future events change, and such changes should be randomly positive or negative as long as investors’ expectations are unbiased. This revelation had its roots in the developing rational expectations theory of macroeconomics, and thus, some economists refer to the EMT as the “rational markets theory.” It was later recognized that the “fair game” model allows for the expectation of a positive price change, which is necessary to compensate risk-averse investors.
In 1970, Eugene Fama published his now-famous paper, “Efficient Capital Markets: A Review of Theory and Empirical Work.” Fama synthesized the existing work and contributed to the focus and direction of future research by defining three different forms of market efficiency: weak form, semistrong form, and strong form. In a weak-form efficient market, future returns cannot be predicted from past returns or any other market-based indicator, such as trading volume or the ratio of puts (options to sell stocks) to calls (options to buy stocks). In a semistrong efficient market, prices reflect all publicly available information about economic fundamentals, including the public market data (in weak form), as well as the content of financial reports, economic forecasts, company announcements, and so on. The distinction between the weak and semistrong forms is that it is virtually costless to observe public market data, whereas a high level of fundamental analysis is required if prices are to fully reflect all publicly available information, such as public accounting data, public information regarding competition, and industry-specific knowledge. In strong form, the highest level of market efficiency, prices reflect all public and private information. This extreme form serves mainly as a limiting case because it would require even the private information of corporate officers about their own firm to be already captured in stock prices.
A simple way to distinguish among the three forms of market efficiency is to recognize that weak form precludes only technical analysis from being profitable, while semi-strong form precludes the profitability of both technical and fundamental analysis, and strong form implies that even those with privileged information cannot expect to earn excess returns. Sanford Grossman and Joseph Stiglitz (1980) recognized that an extremely high level of market efficiency is internally inconsistent: it would preclude the profitable opportunities necessary to motivate the very security analysis required to produce information. Their main point is that market frictions, including the costs of security analysis and trading, limit market efficiency. Thus, we should expect to see the level of efficiency differ across markets, depending on the costs of analysis and trading. Although weak-form efficiency allows for profitable fundamental analysis, it is not difficult to imagine a market that is less than weak form but still relatively efficient in some sense. Thus, it can be useful to define the efficiency of a market in a more general, continuous sense, with faster price reaction equating to greater informational efficiency.
While most of the empirical research of the 1970s supported semistrong market efficiency, a number of apparent inconsistencies arose by the late 1970s and early 1980s. These so-called anomalies include, among others, the “small-firm effect” and the “January effect,” which together document the tendency of small-capitalization stocks to earn excessive returns, especially in January. But financial economists today attribute most of the anomalies to either misspecification of the asset-pricing model or market frictions. For example, the small-firm and January effects are now commonly perceived as premiums necessary to compensate investors in small stocks, which tend to be illiquid, especially at the turn of the year. Fama (1998) also notes that the anomalies sometimes involved underreaction and sometimes overreaction and, thus, could be viewed as random occurrences that often went away when different time periods or methodologies were used.
More serious challenges to the EMT emerged from research on long-term returns. Robert Shiller (1981) argued that stock index returns are overly volatile relative to aggregate dividends, and many took this as support for Keynes’s view that stock prices are driven more by speculators than by fundamentals. Related work by Werner DeBondt and Richard Thaler (1985) presented evidence of apparent overreaction in individual stocks over long horizons of three to five years. Specifically, the prices of stocks that had performed relatively well over three- to five-year horizons tended to revert to their means over the subsequent three to five years, resulting in negative excess returns; the prices of stocks that had performed relatively poorly tended to revert to their means, resulting in positive excess returns. This is called “reversion to the mean” or “mean reversion.” Lawrence Summers (1986) showed that, in theory, prices could take long, slow swings away from fundamentals that would be undetectable with short horizon returns. Additional empirical support for mispricing came from Narasimhan Jegadeesh and Sheridan Titman (1993), who found that stocks earning relatively high or low returns over three- to twelve-month intervals continued the trend over the subsequent three to twelve months.
These apparent inefficiencies contributed to the emergence of a new school of thought called behavioral finance (see behavioral economics), which countered the assumption of rational expectations with evidence from the field of psychology that people tend to make systematic cognitive errors when forming expectations. One such error that might explain overreaction in stock prices is the representative heuristic, which holds that individuals attempt to identify trends even where there are none and that this can lead to the mistaken belief that future patterns will resemble those of the recent past. On the other hand, momentum in stock returns may be explained by anchoring, the tendency to overweight initial beliefs and underweight the relevance of new information. It follows that momentum observed over intermediate horizons could be extrapolated over longer time horizons until overreaction develops. This does not, however, imply any easily exploitable trading strategy, because the point where momentum stops and overreaction starts will never be obvious until after the fact.
Resistance to the view that stock prices systematically overreact, as well as to the behavioral interpretation of this evidence, came along two fronts. First, Fama and Kenneth French (1988) found that stocks earn larger returns during more difficult economic conditions when capital is relatively scarce and the default-risk premiums in interest rates are high. Higher interest rates initially drive prices down, but eventually prices recover with improved business conditions, and hence the mean-reverting pattern in aggregate returns. Second, adherents of the EMT argued that the cognitive failures of certain individuals would have little influence on stock markets because mispriced stocks should attract rational investors who buy underpriced and sell overpriced stocks.
Critics of the EMT responded to both of these charges. In response to the Fama and French evidence, James Poterba and Lawrence Summers argued that the mean-reverting pattern in aggregate index returns is too volatile to be explained by cyclical economic conditions alone. They claimed that excessive mean reversion resulted from prices straying from fundamentals, similar to Shiller’s excess volatility story. As to whether the marginal trader is fully rational or subject to systematic cognitive errors, Andrei Shleifer and Robert Vishny (1997) and others noted that, while market efficiency requires traders to act quickly on their information out of fear of losing their advantage, mispricing can persist because it offers few opportunities for low-risk arbitrage trading. For example, how should one have responded during the bubble in Internet-based stocks of the late 1990s? Most of these stocks were difficult to short sell, and even if it was possible, a well-informed, fully rational short seller faced the risk that less than fully rational traders (also known as “noise traders”) would continue to move prices away from fundamentals. Thus, the market will not necessarily correct as soon as rational traders recognize mispricing. Instead, the correction may come only after the mispricing becomes so large that noise traders lose confidence in the trend or rational traders act in response to the additional risk introduced by the noise traders.
The most striking examples of apparent inconsistencies with the EMT are the 1987 stock market crash and the movement of Internet stock prices beginning in the late 1990s. Some economists, admittedly a minority, believe that the 1987 crash and the Internet run-up and fall are consistent with market efficiency. For example, Mark Mitchell and Jeffry Netter (1989) argued that the large market decline in the days before the market crash in 1987 was triggered by an initially rational response to an unanticipated tax proposal, which in turn triggered a temporary liquidity crunch (or panic) due to much higher sales volumes than the market was prepared to handle. The exchanges, traders, and regulators learned from this experience making markets more efficient. Burton Malkiel (2003a, 2003b), analyzing the Internet bubble, notes that Internet company values were difficult to determine, and while traders in most cases were wrong after the fact, there were no obvious unexploited arbitrage opportunities.
Regardless of whether it is the exception or the rule, the favorable market conditions of the late 1990s for technology and Internet-based stocks illustrate the stock market’s critical role in resource allocation. A firm whose stock has appreciated rapidly finds it easier to raise additional funds through a secondary offering because higher prices mean a smaller percentage ownership of the firm needs to be offered to raise a given amount of capital. Favorable conditions also make it easier for privately held firms to raise funds through an initial public offering (IPO) of stock. Furthermore, a so-called hot IPO market entices venture capital firms to invest funds in hot industries and sectors in hopes of taking their firms public in such a favorable market. Many view these favorable market conditions as consistent with the market’s valuation of growth options and the motivating incentive necessary to make the fundraising portion of venture growth and creation possible. But while favorable market conditions can attract the investment capital necessary to grow a fledgling new industry, the market for technology and Internet-based stocks in the late 1990s appears to have overheated and, in hindsight, directed too much investment capital toward this sector. Thus, by the late 1990s, the return an investor in this sector could have rationally expected had fallen below what economic conditions could justify, as well as below what most investors actually anticipated.
While prices may take long, slow swings away from fundamentals, the EMT is still useful in at least two important ways. First, over shorter horizons, such as days, weeks, or months, there is considerable evidence that the EMT can explain the direction of stock price changes. That is, the response of stock prices to new information reasonably approximates the change in the intrinsic value of equity. Second, the EMT serves as a benchmark for how prices should behave if capital investments and other resources are to be allocated efficiently. Just how close markets come to this benchmark depends on the transparency of information, the effectiveness of regulation, and the likelihood that rational arbitragers will drive out noise traders. In fact, the informational efficiency of stock prices varies across markets and from country to country. Whatever the shortcomings of capital markets, there appears to be no better alternative means of allocating investment capital. In fact, the privatization movement of the 1990s and early 2000s suggests that most governments, including China’s, now recognize this fact. Thus, academic inquiry in this area is likely to focus more on the conditions that explain and improve the informational efficiency of capital markets than on whether capital markets are efficient.
For an excellent review of the debate on market efficiency, see Shiller 2003 for the behavioral finance view, and Malkiel 2003a for the proefficiency view.