These study notes are based on Chapter 11 the Exam 9 syllabus reading Investments by Bodie, Kane, and Marcus. This chapter describes three variations of the efficient market hypothesis, and corresponds to learning objective A9 on the syllabus.
easypackages::packages("dplyr", "ggplot2", "DT", "data.table")
options(scipen = 999)The Efficient Market Hypothesis is the idea that the price of stocks already reflect all available information. There are three variations on this hypothesis, described below. The various forms of the hypothesis can be tested empirically by comparing actual performance to a benchmark when new information about the stock becomes available. The method varies based on the benchmarking model used, but in general, assume that the return on a stock in time \(t\) is given by \(r_t\), and the market return in the same time period is \(r_{M, t}\). A typical benchmark formula is of the form \[ r_t = a + br_{M,t} + \epsilon_t \] The abnormal return is the residual \(\epsilon_t\). A key consideration is that if this model is being used to assess the impact of a specific event (e.g. annoucement of a merger), then the regression parameters must be determined using data from a time frame that is separated from the event. This may need to be well in advance of the actual event in order to avoid capturing leakage of information about the event in the regression coefficients. Because of the leakage effect, market efficiency tests are often performed using cumulative abnormal returns to ensure that the test captures all abnormal stock movement that might have been caused by the new information.
Note that an implication of the fact that abnormal returns are defined relative to a risk-adjusted benchmark indicates that they are joint tests of both market efficiency and the risk-adjustment procedure.
There are three barriers that prevent us from determining experimentally whether the market is efficient or not:
Magnitude: if there were market inefficiencies, they would likely be small relative to the overall volatility of the market and therefore impossible to detect statistically.
Selection bias: if there were successful trading strategies based on market ineffiencies, they would likely be kept secret. Therefore, the absence of such strategies cannot be taken as evidence of market efficiency.
Lucky events: due to random chance, there will always be some trading strategy that turns out to be successful, so the existince of a successful strategy can’t be taken as evidence against market efficiency.
The weak form of the efficient market hypothesis is that stock prices incorporate all information that can be derived by examining past market trading data. In particular, this implies that analysis of market trends will not result in a superior investment strategy.
There are broadly two forms of tests of the weak form of the EMH:
Returns over Short Horizons: test for serial correlations in stock returns. A positive correlation would be indicative of a momentum effect, while negative correlation would be indicative of a correction. Broad market indices exhibit weak positive serial correlation, though not strong enough to inform a trading strategy. The momentum effect is stronger for portfolios of the best-performing stocks than it is for the aggregate market index.
The two trends noted above tend to be complementary: short-term over-reactions lead to long-term corrections in stock prices.
The semi-strong form of the efficient market hypothesis is that all publicly available information is reflected in the stock price, including information such as the firm’s product line, quality of management, balance sheet, earnings forecasts, etc.
Examples of empirical anomalies that do not reconcile with the semi-strong efficient market hypothesis include:
The P/E effect is the observation that portfolios with low price-to-earnings ratios have higher returns than high price-to-earnings stocks. Given the simplicity of calculating price-to-earnings ratios, a more plausible explanation is that the risk-adjustment procedure is incorrect. (All else equal, a riskier stock will sell for a lower price, so it will have a lower P/E ratio and higher expected returns.)
The Small-Firm-in-January Effect is the observation that portfolios of small firms tend to have higher returns; although small firms are riskier in general (and should have higher returns), this is true even after adjusting for risk via the CAPM. Historically, this effect appears to be strongest in the first two weeks of January. However, this could just be due to the risk-adjustment procedure not fully capturing the risk associated with small firms. Examples include:
The Neglected Firm Effect attempts to explain the small firm effect as a result of smaller firms being less well-researched by institutional investors, which makes them riskier investments than large, highly-monitored firms. The comparison is done between firms that are highly researched, moderately researched, and neglected, based on the number of institutions holding the stock. From this perspective, the neglected firm effect is not a market inefficiency, but a risk premium associated with limited information on the firm.
Liquidity Effects: small firms may also require a premium due to their lower liquidity relative to large firms.
Book-to-Market Ratios: firms with high book-to-market ratios tend to have higher returns, suggesting either that they are underpriced, or that the book-to-market ratio is acting as a proxy for a risk factor not accounted for by the CAPM.
Post-Earnings-Announcment Drift: the earnings surprise is the difference between the actual earnings and the expected earnings prior to the announcement. Grouping firms into deciles according to earnings surprise, the change in excess return is not immediate, but shows some “sluggishness” before it is fully incorporated into the price. The typical pattern following a good-news announcement is a big jump on the day of the announcement, followed by a gradual but steady increase in cumulative abnormal earnings.
An important consideration is that after these results were published, the anomaly being described has generally disappeared (with the exception of the book-to-market effect). This suggests that as this information is disseminated among the investment community, the market becomes more efficient.
The strong form of the efficient market hypothesis states that stock prices reflect all information related to the stock, including information only known to company insiders.
Markets are generally not considered to be strong-form efficient due to legal restrictions on insider trading.
Insider trades are reported through the SEC Official Summary; however, the abnormal returns associated with insider buying are generally not of sufficient magnitude to overcome transaction costs.
Increases in cumulative abnormal earnings prior to an announcement have been taking as evidence that limitations on insider trading have not been fully effective, though the pre-announcement abnormal earnings are less than would be expected if the market were strong-form efficient.
A consequence of the efficient market hypothesis is that stock prices follow a random walk: changes in price are random and unpredictable.
This must be the case, because if price changes were predictable, it would indicate that not all information is reflected in the price of the stock.
Movements in the stock price are not driven by irrational behaviour, but rather, by changes in the available information about the stock. In other words, it is the underlying information that is unpredictable, and this is what is reflected in the stock price.
Technical analysis is the search for predictable patterns in stock prices.
Based on the idea that we don’t need to know the future prospects of the firm, because regardless of the reason for a change in price, if the stock price adjusts slowly enough then we can identify a trend that can be exploited during the adjustment period.
An example of technical analysis is tracking the relative strength of a stock over time; this is the ratio of the stock’s performance to a broad market index. If the ratio is increasing, then it may continue for a long enough period to offer profit opportunities.
Key components of technical analysis are resistance levels and support levels, which are the maximum and minimum price of a stock, respectively. These levels are considered to be set by market psychology.
Implications of the efficient market hypothesis for technical analysis include:
The EMH is fundamentally opposed to key assumptions of technical analysis. In particular, the weak form of the EMH essentially says that technical analysis is without merit.
If the market truly believed in resistance levels, then no one would by stocks that are priced slighly below the resistance level because they have no growth potential but plenty of room to fall. Therefore, this lower price level becomes the new resistance level.
A technical rule that appears to work will ultimately be self-destructing, because as a profitable trading rule begins to be used, it will eventually be reflected in stock prices. As a result, technical analysis becomes a continual search for new trading rules, which are destroyed by overuse, and need to be replaced by new rules.
Fundamental analysis assesses stock prices based on the firm’s earnings and dividend prospects, future interest rates, and risk. It is based on calculating the present value of all future payments a stockholder will receive, recommending purchase if this quantity exceeds the current stock price. Implications of the efficient market hypothesis for this type of analysis include:
The semi-strong form of the hypothesis implies that most fundamental analysis will fail, because the publicly-available earnings data used in the analysis should already be incorporated into the stock price.
Fundamental analysis will only provide value of the analyst can discover insights that are not known to other analysts. It only works if the analyst can find firms that are better than everyone else’s analysis indicates, because the market price will already reflect the competitors’ analysis.
A passive investment strategy aims to establish a well-diversified portfolio of securities without any attempt to find over-priced or under-priced stocks.
It is typically characterized by a buy-and-hold strategy.
It is consistent with the efficient market hypothesis view that active portfolio management is largely a waste of time.
Rationale for adopting such a strategy is that the costs of obtaining differential insights do not justify the benefits for most investors. For large portfolios, the small incremental increase in return may generate sufficient gains to justify the cost of active portfolio management.
Avoiding frequent buying and selling reduces transaction costs.
A typical approach is to buy an index fund which replicates the performance of a stock index such as the S&P 500.
There are legitimate reasons for engaging in portfolio management that are unrelated to an attempt to “beat the market”:
Risk appetite: Selection of a well-diversified portfolio that is tailored to the investor’s appetite for systematic risk.
Taxation: High-income investors may prefer tax-exempt municipal bonds while low-income or tax-exempt investors may not. There may also be a need to consider the portfolio’s balance of dividends versus capital gains due to differential tax treatment of the two. Generally high-income investors prefer capital gains because they are taxed less heavily and realization of gains can be deferred.
Employment: The investor is already implicitly invested in their employer. For example, if the investor gets a bonus based on their company’s performance, they should generally not invest in the same industry, since the investor is already over-invested in the industry, and this would exacerbate the lack of diversification.
Age: Older investors living off savings will want to avoid long-term bonds and favour conservation of principal. Younger investors may prefer long-term inflation-indexed bonds.
Considerations when assessing the performance of a mutual fund include:
Comparisons to a market index may not be appropriate, since mutual funds tend to have considerable weight in small firms. Therefore, a fairer benchmark would incorporate the performance of small firms.
Standard benchmark is a four-factor model that augments the Fama-French model with a momentum factor (the factor portfolio is constructed based on prior-year stock return). This benchmark encompasses a wide range of “style choices” for mutual funds.
The \(\alpha\) values need to be adjusted to account for mutual fund fees. Typically, funds will have a positive \(\alpha\) before fees, but a negative \(\alpha\) after. This phenomenon has been observed in both equity and bond funds.
It is important to compare across time periods, to ensure that performance persists over time. Generally, even when performance persists, it is signifincantly reduced in subsequent periods. A possible explanation is that skilled mutual fund managers attract new funds until the complexity of managing the extra funds results in a deterioration of performance.
An important split is between broker-sold funds and direct-sold. The latter tend to have consistently positive \(\alpha\), which can be explained by direct investors likely having more financial literacy than those who purchase through brokers.