Anticipated Earnings Announcements and the Customer–Supplier Anomaly

• DOI: http://doi.org/10.1111/1475-679X.12158
• Author: JOSHUA MADSEN
• Published date: 01 June 2017
• Date: 01 June 2017

DOI: 10.1111/1475-679X.12158

Journal of Accounting Research

Vol. 55 No. 3 June 2017

Printed in U.S.A.

Anticipated Earnings

Announcements and the

Customer–Supplier Anomaly

JOSHUA MADSEN

∗

Received 29 January 2015; accepted 28 October 2016

ABSTRACT

I test whether the anticipation of earnings news stimulates acquisition of cus-

tomer information and mitigates returns to the customer–supplier anomaly

documented by Cohen and Frazzini (“Economic Links and Predictable Re-

turns.” The Journal of Finance 63 (2008): 1977–2011). I find that attention

to a firm’s publicly disclosed customers increases shortly before the firm an-

nounces earnings, and that customer stock returns predict supplier stock re-

turns shortly before, but not after, the supplier’s earnings announcement. I

further find some evidence that these predictable returns are increasing in

the level of customer information acquisition. These results are unique to an-

ticipated disclosure events and suggest that anticipation of supplier earnings

∗Carlson School of Management, University of Minnesota.

Accepted by David F. Larcker. This paper is based on my dissertation at the Uni-

versity of Chicago. I am grateful for the comments of my dissertation committee: Dou-

glas Skinner (chair), Ray Ball, Philip Berger, and Toby Moskowitz, as well as comments

from an anonymous referee, Ryan Ball, Andy Bodmeier, Hans Christensen, Ted Chris-

tensen, Steve Crawford, Michael Drake, Vivian Fang, Pingyang Gao, Frank Gigler, Jo˜

ao

Granja, Michael Iselin, Mo Khan, Christian Leuz, Jeff McMullin, Michael Minnis, Ma-

rina Niessner, Valeri Nikolaev, Shail Pandit, Spencer Pierce, Jake Thornock, Richard

Willis, Regina Wittenberg-Moerman, and seminar participants at the following univer-

sities: Chicago, UT Dallas, DePaul, Vanderbilt, Minnesota, Waterloo, Temple, Illinois,

and the 2013 BYU Accounting Research Symposium. Financial support was provided by

the University of Minnesota. An online appendix to this paper can be downloaded at

http://research.chicagobooth.edu/arc/journal-of-accounting-research/online-supplements.

709

Copyright C, University of Chicago on behalf of the Accounting Research Center,2016

710 J.MADSEN

announcements resolves investor limited attention to customer information

and accelerates price discovery of customer news.

JEL codes: G11; G14; M41

Keywords: customer–supplier anomaly; investor limited attention; antici-

pated earnings announcements

1. Introduction

Cohen and Frazzini [2008] (CF) present evidence that the stock returns in

month tof a firm’s publicly disclosed customers predict the firm’s stock re-

turn in month t+1 (customer–supplier anomaly). SEC filers are required

to disclose the names of material customers that represent 10% or more of

total sales,1and thus evidence that prices adjust with a lag to information

about customers is consistent with investor limited attention to material

customer information. The primary objective of this study is to test whether

the anticipation of a supplier’s scheduled disclosure stimulates acquisition

of customer information and attenuates market inefficiencies related to the

customer–supplier anomaly (anticipation hypothesis).

To test the anticipation hypothesis I examine customer information ac-

quisition and price discovery before a supplier’s earnings announcement.

Earnings announcements are a convenient setting to test the anticipation

hypothesis because they occur every quarter, are typically scheduled in ad-

vance, and generate significant price and volume reactions (see Kothari

[2001]). Investors thus have incentives to acquire information prior to

these scheduled disclosures (Kim and Verrecchia [1991], McNichols and

Trueman [1994]). However, this search for information need not be re-

stricted to the firm itself. If attention to the announcing firm’s customers

also increases before these announcements, then the resulting improved

price discovery could mitigate returns to the customer–supplier anomaly

documented by CF.

I use two primary tests to provide evidence that anticipation of supplier

earnings announcements (1) stimulates acquisition of customer informa-

tion and (2) improves price discovery of customer news and attenuates

market inefficiencies related to investor limited attention. I first exam-

ine variation in customer information acquisition around suppliers’ earn-

ings announcements. Although I do not know specifically who acquires the

customer information, increased acquisition of customer information be-

fore a supplier’s earnings announcement would suggest that the sched-

uled nature of these events elicits increased attention to customers by

the supplier’s investors. Second, I examine whether news about a firm’s

customers is priced immediately before the firm’s earnings announce-

ment. Specifically, I test whether customer news predicts the supplier’s

1See ASC 280-10-50-41 and SEC Regulation S-K section 101.

EARNINGS ANNOUNCEMENTS AND THE CUSTOMER–SUPPLIER ANOMALY 711

pre-announcement and/or post-announcement returns. Predictable pre-

announcement returns would suggest that information acquisition prior to

anticipated earnings announcements resolves investors’ limited attention

to customer information and improves price discovery. A pattern of pre-

dictable pre-announcement returns and insignificant post-announcement

returns would further suggest that the anticipation of supplier earnings an-

nouncements attenuates returns to the customer–supplier anomaly.

My results provide support for the anticipation hypothesis. For the first

test, I use daily Google search volume index (SVI) for customer tickers

and daily downloads of customer filings from the SEC’s EDGAR database

as measures of customer information acquisition. I document significant

spikes in customer information acquisition before suppliers’ earnings an-

nouncement dates, suggesting increased attention to customers by the sup-

plier’s investors. For the second test, I use a 60-day (i.e., approximately one

quarter) sales-weighted customer return as a measure of customer news.

Analyzing a large sample of supplier earnings announcement dates be-

tween 1990 and 2014, I find that customer news predicts suppliers’ three-

day pre-announcement returns, but fails to predict suppliers’ three-day

post-announcement returns. The pre-announcement returns are statisti-

cally greater than post-announcement returns, robust to various research

design specifications, and suggest that increased information acquisition

before anticipated disclosures resolves limited attention to customer infor-

mation and improves price discovery.

The anticipation hypothesis emphasizes the effects of anticipated disclo-

sures on customer information acquisition and price discovery. To test this

hypothesis, I exploit the scheduled and recurring nature of earnings an-

nouncements. To gauge whether the effects I document are specific to an-

ticipated supplier earnings announcements, I perform robustness tests us-

ing two less-anticipated supplier events: unbundled management forecasts

and pseudo events.2Results using these alternative, less-anticipated events

are either insignificant or of reduced magnitude relative to results using

supplier earnings announcements, providing additional evidence that the

increased customer information acquisition and price discovery are driven

by the anticipated nature of earnings announcements.

My two primary analyses suggest that customer information acquisi-

tion increases before supplier earnings announcements and that cus-

tomer returns predict supplier pre-announcement returns, but not

post-announcement returns. If the increased acquisition of customer

information is driving the predictable pre-announcement returns, then

these predictable returns should be increasing in the level of customer

information acquisition. I therefore attempt to directly link these two pri-

mary tests and examine whether the acquisition of customer information

2Pseudo events are defined as the non-earnings announcement date each quarter with the

maximum absolute return and above average trading volume.