Analyst responsiveness and the post-earnings-announcement drift

https://doi.org/10.1016/j.jacceco.2008.04.004Get rights and content

Abstract

This study examines the responsiveness of analyst forecasts to current earnings announcements. The results show considerable cross-sectional variation in analyst responsiveness and suggest that this variation is related to the costs and benefits associated with prompt forecast revisions. More importantly, this study finds that with responsive forecast revisions, more of the market reaction takes place in the event window and less in the drift window, suggesting that analyst responsiveness mitigates the post-earnings-announcement drift and facilitates market efficiency.

Introduction

The efficient market hypothesis implies that in a (semi-strong) efficient market, upon receiving new information, investors instantaneously adjust their expectations with respect to future earnings, which in turn are reflected instantaneously in stock prices. However, researchers have documented evidence inconsistent with this implication. One of the most persistent anomalies is the post-earnings-announcement drift, whereby stock prices continue to drift for a long period after earnings announcements. Since this phenomenon was first documented by Ball and Brown (1968), it has survived numerous robustness checks, including extensions to more recent data (e.g., Bernard and Thomas, 1989; Chan et al., 1996). As Fama (1998) puts it, the post-earnings-announcement drift is an anomaly that is “above suspicion.”

A number of studies have attempted to explain the post-earnings-announcement drift. Notably, recognizing the importance of analysts as information intermediaries, some studies focus on the role of analysts in either mitigating or contributing to the drift (e.g., Abarbanell and Bernard, 1992). Overall, these studies find that analysts fail to fully incorporate information in earnings announcements and that such underreaction contributes to the post-earnings-announcement drift.

In this paper, I argue that there are at least two aspects of analyst underreaction to earnings announcements—underreaction in magnitude and underreaction in time. The extant literature on analyst underreaction largely focuses on underreaction in magnitude, allowing no specific role for when analyst forecasts are made.1 However, analyst reaction timeliness, and the implications of this timeliness for the post-earnings-announcement drift (and possibly other forms of market underreaction), are also important: if analysts’ forecast revisions fully reflect prior earnings news but are only made long after the information becomes available, such revisions are only efficient in magnitude but not in time. In other words, these revisions are still inefficient.2 Thus, to the extent that analyst forecasts affect the market's expectations of future earnings, one expects to continue observing returns drift after the earnings announcements.

This paper focuses on this relatively unexplored aspect of analyst underreaction by examining the responsiveness of sell-side security analysts’ forecast revisions after quarterly earnings announcements (hereafter, “analyst responsiveness”) and its effect on the post-earnings-announcement drift. While analysts do not necessarily revise their forecasts each time they receive new information, their forecast revisions tend to cluster to a greater extent after earnings announcements than after other corporate information events (Bagnoli et al., 2005). This is not surprising because of the important valuation implications of earnings and the rich information set frequently accompanying earnings announcements.3

Bernard and Thomas (1989) are among the first to attribute the post-earnings-announcement drift directly to the speed of investors’ response to new information. Some practitioners further attribute the slow market reactions specifically to analysts’ slowness in revising their earnings forecasts. For example, John Bogle Jr., president of Bogle Investment Management, argues that “[s]hare-price momentum results from earnings-estimate momentum. Analysts are afraid to go out on a limb. That causes estimates to change much more slowly than they should.” (Clements, 1999). The literature has provided some theories as to why market participants, including analysts, may be slow in reacting to new information. For instance, Barberis et al. (1998) suggest that market underreaction is consistent with conservatism in the psychology literature, defined as the slow updating of beliefs in the face of new information. Daniel et al. (1998) present a model in which investors overweigh the value of their private signals and underweigh the information content of important public information such as earnings announcements. Finally, Hong and Stein (1999) suggest that market participants may require additional private information to convert the news in earnings announcements into a judgment about future earnings.

I focus on the responsiveness of analysts’ first forecast revisions for the next quarter after the current quarterly earnings announcements. This focus follows Bernard and Thomas (1990), who suggest that the post-earnings-announcement drift is caused by investors’ failure to promptly recognize the autocorrelation structure of quarterly earnings. I define an analyst as being responsive if she revises her forecast within two trading days after the earnings announcement (i.e., trading days 0 and 1 with respect to the announcement date). Overall, I find that there is significant cross-sectional variation in analyst responsiveness at both the analyst level and the firm level during my sample period of 1996–2002.4 Depending on the year, about 26–53% of analysts revise their forecasts within the responsive window and about 58–76% of firms are followed by at least one responsive analyst. Both percentages show an increasing trend during the sample period.

Further tests find that analyst responsiveness is related to the trade-off between the costs and benefits associated with responsiveness. In particular, when the firms are larger, when the earnings announcements are accompanied by conference calls or managerial guidance, or when the earnings announcements are for fourth fiscal quarters, analysts are more likely to be responsive because it is less costly for them to obtain additional information and revise their forecasts promptly. The tests also show that analyst responsiveness is increasing in the size of the employing brokerage house, which is a proxy for the resources and support available to analysts. Finally, analysts are more likely to be responsive when there is higher competition among analysts, suggesting that analyst responsiveness increases with the benefits potentially associated with being responsive.

The main research question of the paper is whether analyst responsiveness affects how the market reacts to earnings announcements. I find that the earnings response coefficient in the event window is significantly higher for firm-quarters with responsive analysts and that the corresponding post-earnings-announcement drift is significantly lower. In other words, the result suggests that with responsive analysts, more of the market reaction takes place in the event window and less in the drift window.5 The result holds after I control for the determinants of analyst responsiveness. I also investigate the relation between analyst underreaction in magnitude and in time. I find no discernable differences in analyst underreaction in magnitude between firm-quarters with and without responsive analysts. This result suggests that underreaction in time and underreaction in magnitude are not necessarily correlated, ruling out underreaction in magnitude as an alternative explanation for the main results of the paper.

This study contributes to the literature by providing evidence consistent with lack of analyst responsiveness as an explanation for the post-earnings-announcement drift or other market underreaction phenomena (Bernard and Thomas, 1989; Clements, 1999). This evidence suggests that the speed at which market participants—specifically, analysts—incorporate new information into their forecasts for future earnings is indeed associated with the extent of market underreaction to earnings announcements.

This study also adds to the literature on the efficiency of analyst forecasts by highlighting the importance of examining analyst forecast timing. As discussed earlier, this literature has largely focused on analyst underreaction in magnitude by examining the serial correlation in analyst forecast errors (e.g., Abarbanell and Bernard, 1992; Easterwood and Nutt, 1999). However, both the timing and the magnitude of analysts’ reactions to public information are important because market efficiency hinges on both the instantaneity and the completeness with which stock prices reflect available information.

A few other studies have also examined the timing of analyst forecast revisions around information releases such as earnings announcements. Stickel (1989) shows that analysts avoid revising prior to earnings announcements and more frequently revise immediately after the announcements (see also Ivković and Jegadeesh, 2004; Bagnoli et al., 2005). The current study extends this work, as it not only examines the timing of analyst forecast revisions, but does so with a specific interest in its determinants and its implications for market efficiency. This extension is important because it helps us understand the cross-sectional variation in analyst responsiveness and how analyst responsiveness affects the post-earnings-announcement drift, one of the most robust market inefficiency phenomena.

Finally, this study contributes to the literature that examines the role of analysts as information intermediaries. While empirical evidence regarding this role is ample, unlike the extant literature that tends to focus on different analyst characteristics (e.g., analyst experience or reputation), the current study suggests a specific mechanism through which analysts play their role as information intermediaries. Specifically, the results suggest prompt analyst forecast revisions help market participants process and react promptly to the information contained in earnings announcements and that such revisions significantly mitigate the magnitude of the post-earnings-announcement drift.

The paper proceeds as follows. Section 2 describes the sample and provides descriptive statistics on analyst responsiveness. Section 3 examines the determinants of firm-level analyst responsiveness and Section 4 examines the effects of analyst responsiveness on market reactions to earnings announcements. Section 5 discusses additional analyses. Section 6 concludes.

Section snippets

Sample and descriptive statistics

The sample starts with all spilt-unadjusted I/B/E/S individual analyst forecasts for quarterly earnings per share (EPS) with fiscal period ending between 1996 and 2002. I delete observations with zero analyst-specific identification code6 or missing CUSIP. I also exclude observations with (i) a forecast date on or after the corresponding earnings announcement date or (ii)

Identifying determinants of analyst responsiveness

In this section, I examine the determinants of analyst responsiveness. The analysis is performed at the firm level, because the determinants will be control variables in the market reaction tests, which are necessarily at the firm level. I propose the determinants along the lines of the costs and benefits of revising forecasts promptly.

Analyst responsiveness and the post-earnings-announcement drift

As explained in Section 1, to the extent that analyst responsiveness facilitates market reactions to earnings announcements, I expect to see lower post-earnings-announcement drift for firms with responsive analysts. While this expectation might seem obvious, there are least two reasons why that might not be the case. First, it is possible that investors overweigh their own interpretation of the earnings announcements relative to analysts’ interpretations, even though investors tend to

Underreaction in time and in magnitude

The aspect of analyst underreaction examined in this paper is analyst underreaction in time, which has not been examined in detail in prior research. However, the literature has examined another important aspect, namely, underreaction in magnitude. Specifically, this literature posits that if analysts underreact to the news in an earnings announcement, their forecast errors are expected to be autocorrelated and the extent of such autocorrelation measures the degree of underreaction in

Concluding remarks

A common view of the existing explanations for underreaction anomalies such as the post-earnings-announcement drift proposes that investors are slow in updating their expectations of future earnings upon receiving new information. This study investigates whether analysts, as important information intermediaries, are on average responsive to earnings announcements and whether responsive analysts’ prompt forecast revisions can help facilitate market reactions to earnings announcements.

I show that

References (31)

  • J. Abarbanell et al.

    Tests of analysts’ overreaction/underreaction to earnings information as an explanation for anomalous stock price behavior

    Journal of Finance

    (1992)
  • M. Bagnoli et al.

    Analyst estimate revision clusters and corporate events, Part I

    Annals of Finance

    (2005)
  • R. Ball et al.

    An empirical evaluation of accounting income numbers

    Journal of Accounting Research

    (1968)
  • E. Bartov et al.

    Investor sophistication and patterns in stock returns after earnings announcements

    The Accounting Review

    (2000)
  • V.L. Bernard et al.

    Post-earnings-announcement drift: delayed price response or risk premium?

    Journal of Accounting Research

    (1989)
  • Cited by (0)

    I thank Mei Cheng, Bjorn Jorgensen, Dawn Matsumoto (the referee), Jim Ohlson, Stephen Penman, K.R. Subramanyam, Jacob Thomas, Ross Watts (the editor), Richard Willis, Kent Womack, Paul Zarowin, and participants at the Columbia Burton Workshop and 2005 FARS mid-year conference for helpful comments and suggestions. I also thank I/B/E/S for providing analyst forecast information.

    View full text