Empirical findings on market efficiency

Capital-market efficiency implies that price-relevant information is quickly and fully reflected in stock prices. The market-efficiency theory is based on the mechanisms and forces of arbitrage. If a piece of price-relevant information is not yet incorporated in the current stock price, there will be powerful economic incentives to uncover it and to trade on it. Consequently, the stock price will adjust until it fully reflects all available price-relevant information. This implies that capital markets might be efficient to some information systems, but not to others (Fama 1970, Ball 1972, Beaver 1998). Market efficiency should, therefore, be assessed for a given set of available information. For instance, the capital market may well be efficient when it comes to immediate reflection of price-relevant accounting information. But the market may be inefficient when it comes to private information (i.e. insider information). Fama (1970) classifies market efficiency in three different forms: weak, semi-strong and strong. If the capital market is strongly efficient, which is an unrealistic assumption, all information, even private information held by the managers, is reflected in stock prices. All information is already in the public domain. Thus, there is no information

asymmetry, and thereby, no need for accounting (Ronen 1974, Bromvich 1977, Barth and Landsman 1995, Field et al. 2001). If the capital market is semi-strongly efficient, however, financial statements become an important low-cost provider of information. Under semi-strong efficiency, all publicly available information, including financial-accounting information, is reflected in stock prices. The more private information that is made publicly available, e.g. through financial statements, the more information is reflected in stock prices (Beaver 2002).

Market-efficiency tests found in the accounting literature fall into two categories:

event studies and cross-sectional tests of return predictability (Kothari 2001).

These studies provide tests of semi-strong market efficiency. Event studies comprise short-term and long-term event studies. These studies investigate abnormal returns over narrow windows surrounding the event (short-term event studies) or over longer periods following the event (long-term event studies). The investigated events could be earnings announcements, dividend announcements, announcements of restructuring or merger plans or announced changes in accounting methods. Cross-sectional tests of return predictability investigate whether accounting-based trading rules can be used to form portfolios of stocks that perform abnormal returns. Such tests generally use accrual measures or market-to-book ratios to form these portfolios (Kothari 2001, Beaver 2002). Both lines of literature are discussed briefly below. Other literature investigating capital-market efficiency may also be relevant. Still, the primary concern is whether the capital market is efficient when it comes to reflecting accounting information; not information from other sources than the financial statement.

Short-term event studies provide joint tests of information content and market efficiency. Given price-relevant information, event studies bring evidence on the

impact, speed and unbiasedness of the market reaction to that information. Strong evidence is found for quick and unbiased market responses to earnings announcements, merger and restructuring announcements and dividend announcements (Kothari 2001). Still, these studies do have some methodological challenges. An important issue is to ensure that the event, for instance the earnings announcement, is not published simultaneously with other announcements (Lev 1989). This may confound the association between the short-window abnormal return and the event. The second issue is to identify the day on which the information is actually revealed to the capital market. To avoid missing the actual day, the return window is usually set equal to a few days centred on the announcement day.

The evidence from short-term event studies supports semi-strong market efficiency (Kothari 2001). The studies by Lee (1992) and Landsman and Maydew (2002) may serve as illustrative examples. Lee (1992) uses intra-day returns and trading-volume data to investigate the market reactions to earnings announcements. He reports a statistically significant price reaction of the same sign as the earnings surprise within 30 minutes of the earnings announcement. No statistically discernible price effect is discovered afterwards. The shift in trading volume is also short-lived: less than two hours for large trades and a few hours for small trades. Landsman and Maydew (2002) investigate the market reactions to earnings announcements over three decades. They find that stock-return volatility and trading volume are significantly larger on earnings-announcement days and that the activity reverts to normal immediately afterwards.

Long-term event studies generally investigate post-announcement drifts in stock returns following an event. There are several potential reasons for the

post-announcement drift. Likely candidates are economic irrationality among the investors and market frictions. A post-announcement drift can be defined as the predictability of abnormal returns following certain events (Kothari 2001).

Numbers of studies have demonstrated large abnormal returns following well-published events such as earnings announcements, initial public offerings (IOPs), seasoned public offerings (SPOs) and analysts’ long-term forecasts (Kothari 2001). These findings seriously challenge the market-efficiency hypothesis. Post-announcement drift is found to have the same sign as unexpected earnings in earnings announcements. This has led to the general conclusion that capital markets underreact to earnings announcements.

Important seminal articles discussing post-announcement drift are Rendleman, Jones and Latane (1987), Freeman and Tse (1989) and Bernard and Thomas (1989, 1990). Bernard and Thomas (1989) investigate the post-announcement drift in changes in quarterly earnings. Earnings surprises are calculated as the difference between earnings for a given quarter one year and earnings of the same quarter the previous year. They demonstrate that buying stocks in firms reporting surprisingly high quarterly earnings, selling short stocks in firms reporting surprisingly low quarterly earnings and holding this position for 60 days following the announcement, will give a significantly high abnormal return. As Bernard and Thomas (1989) point out, it is a wellknown fact that quarterly-seasonal earnings changes are positively correlated. Thus, if a firm reports surprisingly high earnings this quarter compared to the same quarter last year, it is likely that its future quarter earnings will be surprisingly high as well. Rational investors should anticipate this and be willing to bid up the price of the firms’ stocks in response to surprisingly high quarterly earnings, but Bernard and Thomas (1989) find that this is not the case. These findings suggest that the capital market underestimates the

positive correlation between quarterly earnings changes (Bernard and Thomas 1989, 1990, Ball and Bartov 1996). Surprisingly, the post-announcement drift has not disappeared even several decades after its first discovery (e.g.

Narayanamoorhy 2006). Later studies, however, have found that the significance of the announcement drift varies according to certain characteristics of the firms and the capital market. Bhushan (1994) for instance, demonstrates that the post-announcement drift is mainly driven by relatively smaller firms, firms with stocks having relatively larger bid-ask spreads, stocks that are less frequently traded and less closely followed by analysts. Other studies report evidence suggesting that the post-announcement drift is less strong in firms having more institutional ownership (Bartov, Radhakrisknan and Krinsky 2000) and more timely analysts’

forecast revisions (Zhang 2008). This last finding is consistent with Bhushan (1994). The results for short-term and long-term event studies are puzzling. The short-term event studies demonstrate evidence consistent with market efficiency.

In constrast, the long-term event studies suggest the opposite that price-relevant information is reflected in stock prices with substantial time lag following the events. This last evidence is inconsistent with market efficiency (Beaver 2002).

One potential reason for these contradicting results is different methodology.

Long-term event studies are believed to suffer from more serious methodological problems than short-term event studies. Likely problems in long-term event studies are omitted variables and survivorship bias (Lev 1989, Kothari 2001).

Post-announcement drifts could be due to an omitted priced risk factor. The omission of this risk factor will affect the estimate of expected return and thereby the estimate of abnormal return. Thus, the post-announcement drift could be the result of an under or misspecified return model rather than evidence of market inefficiency (Kothari 2001). In contrast, short-term event studies are believed to

suffer from fewer problems of misestimated expected returns (e.g. Brown and Warner 1985). Common expected return in short-term event studies is about 0.05% per day. The misestimation of expected return due to risk mismeasurement is likely to be less than 0.01-0.02% per day. This is small compared to a common abnormal return of 0.5% in these studies (Kothari 2001). Due to fewer methodological problems, more confidence can be placed on the evidence from short-term event studies than the evidence from long-term event studies. Despite this, there is no doubt that the mounting evidence on the post-announcement drift still represents a serious challenge to the market-efficiency hypothesis.

A different line of literature investigates market responses to accounting-method changes. These studies are similar to the event studies in that they investigate the market response to a certain event, in this case, change in accounting methods.

The accounting-method changes have no (apparent) cash-flow effects. An efficient capital market is, therefore, predicted not to be misled by its effects on net earnings and net-asset values. Thus, no market response to accounting-method changes is consistent with an efficient capital market (Watts and Zimmerman 1986:72, 1990, Beaver 1998: 135, Kothari et al. 2010). These tests, however, are problematic. Changes in accounting methods are not exogenous, but endogenous.

A voluntary change in accounting methods could reflect opportunistic reporting incentives or signalling incentives. Likewise, a mandatory change in accounting methods could be the result of lobbying effort of different interest groups (Watts and Zimmerman 1986). For instance, the decision to capitalise research and development costs could be driven by the desire to affect the outcomes of earnings-based compensation contracts. Alternatively, capitalisation might signal that the research and development activity is expected to provide economic benefits. This suggests that accounting-method changes might have cash-flow

effects. Early studies, however, report findings consistent with market efficiency.

Ball (1972), for instance, investigates accounting changes in net earnings and reports no significant market response to these changes, which is consistent with market efficiency. Likewise, Beaver and Duke (1973) find no significant market response to changes in depreciation methods. Some later evidence is inconsistent with the market-efficiency hypothesis. For instance, capital markets are not found to be able to undo the effects on net earnings when firms choose between pooling and purchase accounting. Vincent (1997) compares price-earnings ratios of firms using the pooling-of-interests method with those using the purchase method for business combinations. The earnings numbers of the pooling-method firms are restated as if these firms used the purchase method. She finds that the price-earnings ratios of the pooling-method firms are higher than those for purchase-method firms, suggesting that firms using the purchase purchase-method are placed at a disadvantage. Taken together, the results on accounting-method changes and market efficiency are somewhat mixed and inconclusive (Kothari 2001).

The evidence on market responses to accounting accruals is mainly found to be inconsistent with market efficiency. These studies are not considered as event studies. Rather, they are investigating cross-sectional predictability of abnormal returns without addressing particular events (Kothari 2001). Sloan (1996) is an important seminal study in this line of literature. Net earnings consist of cash flows and net accruals. The cash-flow component is found to be more persistent and less likely to be incurred by measurement errors than the accrual component in net earnings. Since accruals are less persistent and more subject to measurement errors than cash flows, the capital market is predicted to respond more strongly to changes in earnings caused by the cash-flow component in earnings than the accrual component. Sloan (1996) reports evidence inconsistent with these

predictions. Rather, the evidence suggests that the capital market overestimates the persistence of accruals and underestimates the persistence of cash flows. This questions whether the capital market effectively distinguishes high from low quality earnings. Lev and Nissim (2006) report evidence consistent with Sloan (1996), but they conclude that the accrual anomaly is less severe for firms having institutional investors. In contrast, Kraft, Leone and Wasley (2007) report evidence inconsistent with Sloan (1996) and Lev and Nissim (2006). They add variables such as capital expenditures to the analysis and find that the mispricing of accruals disappears. A recent survey by Kothari et al. (2010) questions whether prior research findings can reject the market-efficiency hypothesis. They conclude that an overwhelming body of evidence suggests that stock prices largely anticipate the economic substance of the information in financial statements. They argue that “(...) the evidence of market inefficiency is much like waves over deep sea waters – the tranquillity of deep waters underneath swamps any indication of turbulence from waves on the top” (Kothari et al 2010:278). Still, it is reasonable to question whether the capital markets are efficient.