Bucking the Trend: The Informativeness of Analyst Contrarian Recommendations

• Author: Daniel Bradley,Christos Pantzalis,Xi Liu
• DOI: http://doi.org/10.1111/fima.12037
• Published date: 01 June 2014
• Date: 01 June 2014

Bucking the Trend: The Informativeness

of Analyst Contrarian

Recommendations

Daniel Bradley, Xi Liu, and Christos Pantzalis∗

This paper examines price reactions to analysts’ recommendationsissued in the opposite direction

of recent stock price movements. We find that upgrade and downgrade contrarian recommenda-

tions induce larger market reactions than noncontrarian recommendations, consistent with the

view that they are more informative. These results are strongest in the period before Regulation

Fair Disclosure, consistent with the view that private information was likely curbed after its im-

plementation. Contrarian downgrades are more likely to be issued by all-star analysts, but less

likely by experienced and busy analysts suggesting that contrarian recommendations are subject

to career concerns.

Analysts are conventionally viewed as important information agents who convey information

through earnings estimates and recommendation revisions. Many studies have found evidence

consistent with the view that individual analyst recommendation revisions are informative as

they are usually accompanied by significant market reactions at the time of the announcement

(Womack,1996; Green, 2006). Fur ther studies document cross-sectional variationin the market’s

assessment of recommendations’ information content based on various firm- and analyst-level

attributes such as firm size (Branson, Guffey, and Pagach, 1998), Institutional Investor’s all-star

status (Stickel, 1995), investment bank affiliation (Michaely and Womack, 1999), the timing of

the recommendation in relation to an earnings announcement (Ivkovic and Jegadeesh, 2004), and

others.

In this paper, we examine the information content of analysts’ recommendations conditional

upon past firm-level stock returns. Specifically, we focus on the market response to contrarian

revisions, those recommendation changes that contradict a sizeable recent stock price movement.

For instance, a contrarian downgrade would be issued after a sizeable upward movement in the

stock price. In contrast, trending revisions are those consistent with a sizeable recent change in

the firm’s stock price. We define stock price changes as sizeable if the past one-month return over

the period ending a week prior to the issuance of the revision exceeds one standard deviation in

return volatility where volatility is calculated over 60 months ending in month t−5.

A contrarian recommendation is considered bold when it goes against the recent stock price

movement. Jegadeesh and Kim (2010) find that market reactions to analysts’ recommendations

are stronger when they are farther from the consensus. Clement and Tse (2005) confirm that bold

earnings forecasts are more accurate than herding forecasts. In the presence of private information,

Bernhardt, Campello, and Kutsoati (2006) determine that analysts tend to issue contrarian earnings

forecasts that systematically overshootactual ear nings per share (EPS) in a direction opposite that

of the consensus forecast. That is, when an analyst possesses privateinformation and believes that

∗Daniel Bradley is the Bank of America Associate Professorof Finance at the University of South Florida in Tampa, FL.

Xi Liu is an Assistant Professor of Financeat Ohio University in Athens, OH. Christos Pantzalis is the Bank of America

Professor of Financeat the University of South Florida in Tampa, FL.

Financial Management •Summer 2014 •pages 391 - 414

392 Financial Management rSummer 2014

a firm’sear nings will fall short of EPS forecasts, she will systematically undershoot the consensus

forecast, while she will overshoot the consensus forecast if her private information suggests that

the firm will beat the EPS estimates. These studies suggest that contrarian revisions are more

likely to contain privateinfor mation. As such, theywould elicit a stronger market reaction relative

to other revisions. Werefer to this as the private information hypothesis that predicts heightened

market reactions to both upgrade and downgrade contrarian revisions.

Alternatively, investors may be reluctant to go against the recent stock price trend. Pervasive

empirical evidence exists suggesting that individual investorsare attracted to momentum (Kumar

and Lee, 2006) and there is evidence that institutional investors are momentum chasers, as well

(Chen, Hong, and Stein, 2002; Bennett, Sias, and Starks, 2003; Alti, Kaniel, and Yoeli, 2012).

Thus, investors may be reluctant to trade against the trend and the initial reaction to a contrarian

revision will be muted. We call this the reluctance hypothesis.

Finally, analyst bias may be related to contrarian revisions. For instance, in a sample of initial

public offering (IPO) firms, James and Karceski (2006) find that aff iliated analystsoften provided

“booster shots” to firms their bank recently took public. In our context, these would likely be

classified as contrarian upgrades if the return volatility exceeded our threshold. If analyst bias is

a factor, these biased optimistic recommendations would be discounted by market participants.

However, if analysts are positively biased, they would be hesitant to issue a negative downgrade

on their client’s firm. Thus, a downgrade from a conflicted analyst likely contains more bad news

than a typical downgrade resulting in a larger negativereaction. Therefore, the conflict hypothesis

predicts a smaller reaction to upgrades, but a larger reaction to downgrades.

In our sample of recommendation revisions from 1994 to 2012, we find that contrarian recom-

mendation revisions are more informative than trending recommendation revisions. In our event

time panel regressions, we confirm that the market reaction to contrarian upgrades is 0.27%

higher than trending upgrades during the (0,+2) day window surrounding the recommendation

announcement. This effect is similar for contrarian downgrades. The market reaction is 0.59%

lower over the corresponding window. These results are robust to controlling for firm size and

return reversals in our analyses indicating that our results are not mechanically driven by the

small-firm effect or bid-ask bounce. These results are strongest in the period before Regulation

Fair Disclosure (Reg FD) and are consistent with Cohen, Frazzini, and Malloy (2010) who find

that after Reg FD, the benefit of having personal connections to firm executives disappeared.

Taking our results as a whole, the evidence is most consistent with the notion that contrarian

revisions reflect private information.

Given that contrarian revisions tend to be more informative than other recommendation re-

visions, we explore characteristics that are related to their issuance. We find that analysts who

cover more firms are less likely to issue contrarian revisions, but less experienced and all-star

analysts are more likely to issue contrarian revisions. These results are consistent with the view

that analyst career concerns have an impact on analyst forecasting.

Our paper adds to the literature along several dimensions. While investment strategies based

on past returns are ubiquitous in the literature, most tend to focus on momentum or herding

strategies. Jegadeesh et al. (2004) suggest that analysts tend to recommend positive momentum,

high growth, and high-volume stocks. They also examine past returns with respect to analyst

recommendation changes and levels, but their focus is on subsequent six-month stock price

performance. Alternatively, our paper deals with the information content of recommendation

revisions rather than longer-term subsequent performance.

Our paper is also related to the work of Loh and Stulz (2011). They investigatethe infor mation

content of analyst revisions and find that approximately 12% of all recommendation revisions

result in visible stock price movement. Altinkilic and Hansen (2009) study analyst revisions on