Is Sell‐Side Research More Valuable in Bad Times?

• DOI: http://doi.org/10.1111/jofi.12611
• Date: 01 June 2018
• Author: ROGER K. LOH,RENÉ M. STULZ
• Published date: 01 June 2018

THE JOURNAL OF FINANCE •VOL. LXXIII, NO. 3 •JUNE 2018

Is Sell-Side Research More Valuable

in Bad Times?

ROGER K. LOH and REN ´

E M. STULZ∗

ABSTRACT

Because uncertainty is high in bad times, investors find it harder to assess firm

prospects and hence should value analyst output more. However, higher uncertainty

makes analysts’ tasks harder,so it is unclear whether analyst output is more valuable

in bad times. We find that in bad times, analyst revisions have a larger stock-price

impact, earnings forecast errors per unit of uncertainty fall, and analyst reports are

more frequent and longer. The increased impact of analysts is also more pronounced

for harder-to-value firms. These results are consistent with analysts working harder

and investors relying more on analysts in bad times.

WHILE THERE IS A LARGE LITERATURE ON sell-side analysts’ role as information

intermediaries, this literature mostly ignores the question of whether the state

of the economy affects the value of analyst output for investors.1There are good

reasons to believe that the usefulness and performance of sell-side analysts

depend on the state of the economy. It is well known that in bad times such as

∗Roger K. Loh is at Lee Kong Chian School of Business at Singapore Management University.

Ren´

e M. Stulz is at Fisher College of Business at Ohio State University, NBER, and ECGI. We

thank the Editor, Ken Singleton, an anonymous Associate Editor, and two anonymous referees

for valuable comments and suggestions. We also thank Marcin Kacperczyk, Oguzhan Karakas,

Jeff Kubik, Massimo Massa, Roni Michaely, Jay Ritter, Paola Sapienza, Siew Hong Teoh, Stijin

Van Nieuwerburgh, Mitch Warachka, Kent Womack, Frank Yu,and Jialin Yu; participants at the

Bank of America Merrill Lynch Asia Quant 2016 Conference, AFA 2014 Philadelphia meetings,

SMU-SUFE Summer 2013 Institute of Finance Conference, and at a seminar at the University

of Zurich for helpful comments. Brian Baugh, Andrei Gonc¸alves, David Hauw, Haoyuan Li, and

Shuyu Xue provided excellent research assistance. Disclosures: Roger received financial support

from the Sing Lun Fellowship and the Sim Kee Boon Institute for Financial Economics for data,

research assistance, and travel costs related to the project. Ren´

e received no funding related to this

project. Over the last three years, he received consulting income from major banks that employ

analysts, but none of that income was related in any way to issues concerning analysts and he

had no discussion with any of these institutions concerning analysts over that period. He is not

involved in any consulting involving analyst issues and does not anticipate such consulting in the

foreseeable future.

1For example, Womack (1996), Barber et al. (2001), and Kecsk´

es, Michaely,and Womack (2017)

show that stock prices react to the release of analyst recommendations and a drift follows after-

wards. Loh and Stulz (2011) show that some recommendation changes lead to a large noticeable

change in the firm’s stock price and that these recommendations can impact the firm’s information

environment. Bradley et al. (2014) report that, compared to earnings announcements or company

earnings guidance, recommendations are more likely to cause jumps in intraday stock prices. Oth-

ers find that analyst coverage reduces information asymmetry and improves visibility (Kelly and

Ljungqvist (2012)), disciplines credit rating agencies (Fong et al. (2014)), and affects corporate

policies (Derrien and Kecsk´

es (2013)).

DOI: 10.1111/jofi.12611

959

960 The Journal of Finance R

recessions and crises, there is greater variation in outcomes across firms and

over time (see, for instance, Bloom (2009)). Tothe extent that the role of analysts

is to make sense of firms amidst increased macro uncertainty, they should

be more important and hence should work harder in bad times. Increased

uncertainty, however, may make it more difficult for analysts to perform their

job. Further,the decline in trading volume and hence broker profits in bad times

may reduce performance rewards, leading to a decrease in analyst motivation.

It is therefore not clear whether analyst output is more valuable in bad times

than in good times. In this paper,we find that analysts are indeed more valuable

in bad times: the stock-price impact of their recommendation and earnings

forecast revisions is greater in bad times. We investigate possible explanations

for this finding and conclude that the evidence is consistent with analysts

working harder, and investors relying on analysts more in bad times.

We conduct our investigation using a sample of Institutional Brokers’ Esti-

mate System (I/B/E/S) Detail earnings forecasts from 1983 to 2014 and rec-

ommendations from 1993 to 2014. We define bad times in several ways. The

most obvious approach is to use prominent crises that have occurred during the

sample period, such as the October 1987 crash, the Long-Term Capital Man-

agement (LTCM) crisis of 1998, and the credit crisis of 2007 to 2009. We also

define bad times as recessions marked by the National Bureau of Economic

Research (NBER) and as high uncertainty periods in the Baker, Bloom, and

Davis (2016) policy uncertainty index (from www.policyuncertainty.com). Our

measure of the value of analyst output is the price impact, which captures the

extent to which analyst signals affect investors’ assessment of firm value and

hence is a measure of analysts’ contribution to firms’ information environment.

Using average two-day abnormal returns to stock recommendation changes,

we find that the stock-price impact of analysts is greater during bad times for

both downgrades and upgrades. Further, using the definition of influential rec-

ommendations as defined by Loh and Stulz (2011), who classify recommenda-

tion changes as influential if the stock-price reaction is statistically significant,

we find that both upgrades and downgrades are more likely to be influential

during bad times compared to good times. We also find that the market reacts

more strongly to earnings forecast revisions during bad times. Our evidence

of greater analyst impact during bad times is robust to controlling for firm

and analyst characteristics, including analyst fixed effects. We conclude that

analyst output is more useful for investors in bad times, in that it moves stock

prices more.

Notice that we focus on macro instead of firm-specific bad times. This is

because macro bad times are economically important and are more likely to

be exogenous to analysts. Prior studies such as Frankel, Kothari, and Weber

(2006) and Loh and Stulz (2011) show that analysts are more informative when

firm-level uncertainty is higher. While we already control for firm-level uncer-

tainty,to ensure that it is macro uncertainty that drives our results, we conduct

two sets of tests. First, we decompose a firm’s total stock return volatility into

market-, industry-, and firm-specific components. We find that the increased

impact of recommendation changes in times of high uncertainty is strongest

Is Sell-Side Research More Valuable in Bad Times? 961

when the market component is used to define high uncertainty. Second, we in-

vestigate whether the market simply reacts more to all types of firm news in bad

times (e.g., Schmalz and Zhuk (2017) find that earnings announcement reac-

tions are larger in recessions). Adapting the methodology in Frankel, Kothari,

and Weber (2006), we regress a stock’s daily absolute returns on a compre-

hensive set of dummy variables that capture important firm news events, in

particular, recommendation changes, reiterations, earnings announcements,

earnings guidance, dividend announcements, and insider trades. Interacting

these news dummies with an indicator for bad times, we find that not all firm

news events are associated with a greater impact in bad times. Importantly,

the market continues to react more to recommendation changes (and reitera-

tions) in bad times after taking into account all other news events and their

interactions with bad times. Our finding that analysts’ stock-price impact is

greater in bad times is therefore novel and robust.

We next find that analysts’ absolute forecast errors increase during bad

times, which raises the question of how their output can have more of an impact

on prices during these times. We show, however, that traditional measures of

analyst precision are not appropriate for comparing precision across good and

bad times. Rather,a relevant measure of precision is one that takes into account

the underlying uncertainty. In a simple Bayesian model, the extent to which a

new signal changes investors’ priors depends on both the weight that investors

put on the new signal and the weight that they put on their prior (e.g., Pastor

and Veronesi (2009)). As the precision of the signal increases relative to the

uncertainty associated with their prior, they put more weight on the signal.

Hence, in bad times, investors put more weight on a signal from an analyst if

the ratio of the precision of the signal to the uncertainty of the prior increases.

Such an outcome could occur even if the precision of the signal is lower in

bad times as long as the precision of the signal falls less than the increase in

the uncertainty about the prior. Thus, we can think of a relevant measure of

forecast error as a measure of forecast error per unit of underlying uncertainty.

Using prior volatility to normalize absolute forecast errors, we find that

this adjusted forecast precision actually increases during bad times (scaling

by prior volatility is similar to the approach that we use to define influential

recommendation changes). Importantly, however, the finding that analyst fore-

cast precision increases when measured against the underlying uncertainty

does not necessarily mean that analysts automatically become more useful to

investors. Kacperczyk and Seru (2007) show that the extent to which investors

rely on public information depends on the precision of their private informa-

tion. Thus, if analysts’ signals are public information, investors will rely less on

analysts in bad times if investors themselves have better private information.

We develop and test five possible, nonmutually exclusive, explanations for

why analysts might have more of an impact in bad times. First, we examine an

analyst reliance hypothesis that builds on Kacperczyk and Seru (2007). This

hypothesis predicts that investors rely on analyst information more during

bad times. During bad times, investors have to understand how the increase

in macro uncertainty affects firm prospects. Because of the increase in macro