Sell‐Side Analyst Research and Stock Comovement

• Date: 01 September 2014
• Author: YEXIAO XU,VOLKAN MUSLU,MICHAEL REBELLO
• Published date: 01 September 2014
• DOI: http://doi.org/10.1111/1475-679X.12057

DOI: 10.1111/1475-679X.12057

Journal of Accounting Research

Vol. 52 No. 4 September 2014

Printed in U.S.A.

Sell-Side Analyst Research

and Stock Comovement

VOLKAN MUSLU,

∗MICHAEL REBELLO,

†AND YEXIAO XU

†

Received 29 August 2011; accepted 11 May 2014

ABSTRACT

We document that a stock’s price around a recommendation or forecast co-

varies with prices of other stocks the issuing analyst covers. The effect of

shared analyst coverage on stock price comovement extends beyond analyst

activity days. A stock’s daily returns covary with the returns of other stocks with

which it shares analyst coverage. These links between stock price comove-

ment and shared analyst coverage are consistent with the coverage-specific

information we find in earnings forecasts; analysts who cover both stocks in

a pair expect future earnings of the stocks to be more highly correlated than

do analysts who cover only one stock from the pair. Collectively, our evidence

indicates that analyst research produces coverage-specific spillovers that raise

price comovement among stocks that share analyst coverage. The strength of

these spillovers is comparable to spillovers from broad industry and market

information in analyst research.

1. Introduction

Researchers are debating whether sell-side analysts produce stock-specific

or broad (market- or industry-wide) information. Piotroski and Roulstone

[2004] and Chan and Hameed [2006] document that broad stock indices

∗University of Houston; †University of Texas at Dallas.

Accepted by Philip Berger. Weare grateful to an anonymous referee, Gus De Franco, Stan

Markov, David Reppenhagen, Daniel Taylor, Kelsey Wei,and seminar participants at Southern

Methodist University, University of Houston, University of Texas at Dallas, 2011 FMA Asian

conference, 2011 AAA annual conference, and 2012 AAA FARS mid-year meeting for helpful

comments.

911

Copyright C, University of Chicago on behalf of the Accounting Research Center,2014

912 V.MUSLU,M.REBELLO,AND Y.XU

better explain returns of stocks with higher levels of analyst coverage,

suggesting that analysts convey primarily broad information. In contrast,

Liu [2011] finds that analyst recommendation revisions convey primar-

ily stock-specific information. Based on cross-sectional evidence, Crawford,

Roulstone, and So [2012] show that analysts convey primarily stock-specific

(broad) information for stocks that are already covered (not covered) by

other analysts.

We join the debate by arguing that analysts also convey coverage-specific

information, that is, information that emphasizes commonalities among

stocks in their coverage. This information lies within a wide spectrum whose

opposite extremes are marked by stock-specific and broad information.1

Consider analysts who concurrently cover Oracle and SAP, stocks from the

enterprise software segment of the computer software industry. The ana-

lysts can produce information that is useful in evaluating both stocks by,

for example, focusing on potential clients or software releases of competi-

tors. This information is not specific to either Oracle or SAP, yet it is more

focused than information on the computer software industry or the mar-

ket. Similarly, consider analysts who cover only large oil companies such

as Exxon and Chevron. The analysts can focus on shared issues relating to

vertical integration, which are largely irrelevant for small oil companies.

There are also ample opportunities for analysts to produce valuable

coverage-specific information when they cover stocks from multiple indus-

tries by, for example, emphasizing the linkages between the stocks and their

shared economic exposure. Analysts covering both original design man-

ufacturers (e.g., Dell) and original equipment manufacturers that supply

components (e.g., Asustek) may focus on the supply chain between these

companies. Similarly, analysts covering stocks from different industries but

with common economic exposure, for example, manufacturing bases in

China, can highlight the effect of this exposure.

When deciding on the information mix in her research, an analyst faces

a tradeoff similar to that of a firm that produces related but differentiated

products: The firm can typically lower its overall cost by using common

1We have searched for explicit coverage-specific information in analyst reports for a small

set of stocks in 2010. The reports frequently contain references to other stocks analysts cover.

Consistent with Franco, Hope, and Larocque [2014], most of these references are about

competitors, suppliers, and customers. While consistent with the use of coverage-specific in-

formation, this evidence is not conclusive since, in many cases, analysts who do not cover

competitors, suppliers, or customers also mention these stocks. However, some references

strongly suggest the use of coverage-specific information. In these instances, analysts refer

to other firms they cover that are not competitors, customers, or suppliers. Moreover, only

these analysts appear to find these firms relevant since other analysts do not mention these

firms. Examples of such references include one report that uses metrics for SAP and Yahoo

to justify the analyst’s valuation of Apple; an analysis of business ties between Discovery Com-

munications and Mattel to justify the latter’s valuation; the use of metrics on H&R Block (tax

services), Republic Services (waste management), and DigitalGlobe Inc. (satellite imaging) to

value Cintas Corp. (industrial uniforms); and metrics on Arch Coal (coal mining) and Ashland

Inc. (chemicals) to evaluate Vulcan Materials (construction materials).

SELL-SIDE ANALYST RESEARCH AND STOCK COMOVEMENT 913

inputs across its product lines. However, common inputs lower product

differentiation, limiting the firm’s ability to maximize revenue by fully

exploiting demand for each product. Similarly, an analyst can lower her

overall research cost by emphasizing coverage-specific information, which

is useful for multiple stocks, and downplaying stock-specific information,

which is useful for only one stock. However, coverage-specific information

will make the analyst’s research on individual stocks less distinct and less

informative for investors, limiting the analyst’s ability to fully meet investor

demand for research on individual stocks. Since the analyst’s reward, that is,

her compensation and continued employment, depends on the aggregate

investor demand for her research, substituting coverage-specific informa-

tion for stock-specific information may decrease the analyst’s reward.2Em-

phasizing broad industry and market information will similarly yield cost

savings while lowering the analyst’s reward. Relying on broad information

has an additional drawback since the analyst’s reward will not fully reflect

information her research conveys about stocks outside her coverage.

Like the firm whose optimal input mix will equate the marginal net prof-

its from each product-specific and common input type, an analyst’s optimal

information mix will equate her marginal net reward (i.e., reward net of

cost) from each information type. The marginal net reward from each in-

formation type is likely to decline with its increased use, effectively prevent-

ing the analyst from focusing on only one information type in equilibrium.3

We hypothesize that coverage-specific information will be part of the

mix because, when the analyst first introduces coverage-specific informa-

tion in her research, she will enjoy significant cost savings yet sacrifice lit-

tle in terms of her ability to produce stock-specific information that meets

investor demand. As the analyst relies more on coverage-specific informa-

tion, she will find it progressively more difficult to acquire and process new

coverage-specific information and thus lower her research cost. Moreover,

the increased reliance on coverage-specific information will make it increas-

ingly difficult to differentiate her research for individual stocks. Overall,

coverage-specific information will raise investor expectations about shared

economic exposure among stocks that the analyst covers. Given that in-

vestor expectations of shared economic exposure determine comovement

between stock returns (Merton [1973]), investors’ raised expectations will

result in higher return comovement than what covariance of stock fun-

damentals would predict. We refer to these predictions collectively as the

Coverage-Specific Spillover Hypothesis.

2By producing high-quality research on stocks they cover, analysts can attract new under-

writing business (Krigman, Shaw, and Womack[2001]), generate brokerage income (Jackson

[2005]), and earn “All-Star” recognition (Ljungqvist et al. [2007]). Analysts are rewarded for

each of these outcomes (Groysberg, Healy, and Maber [2011]).

3Veldkamp [2006] models an equilibrium in which analyst research is informative for sev-

eral assets.