Asymmetric Trading Costs Prior to Earnings Announcements: Implications for Price Discovery and Returns

• DOI: http://doi.org/10.1111/1475-679X.12189
• Date: 01 March 2018
• Published date: 01 March 2018

DOI: 10.1111/1475-679X.12189

Journal of Accounting Research

Vol. 56 No. 1 March 2018

Printed in U.S.A.

Asymmetric Trading Costs Prior

to Earnings Announcements:

Implications for Price Discovery

and Returns

TRAVIS L. JOHNSON

∗AND ERIC C. SO

†

Received 29 May 2015; accepted 15 June 2017

ABSTRACT

We show that the cost of trading on negative news, relative to positive news,

increases before earnings announcements. Our evidence suggests that this

asymmetry is due to financial intermediaries reducing their exposure to an-

nouncement risks by providing liquidity asymmetrically. This asymmetry cre-

ates a predictable upward bias in prices that increases preannouncement, and

subsequently reverses, confounding short-window announcement returns as

measures of earnings news and risk premia. These findings provide an alter-

native explanation for asymmetric return reactions to firms’ earnings news,

and help explain puzzling prior evidence that announcement risk premia

precede the actual announcements. Our study informs methods for research

centering on earnings announcements and offers a possible explanation for

∗McCombs School of Business, The University of Texas at Austin; †Sloan School of Man-

agement, Massachusetts Institute of Technology.

Accepted by Christian Leuz. We thank two anonymous referees, Audra Boone, Zhi

Da, Emmanuel De George, Amit Goyal (WFA discussant), Nick Guest, Terry Hender-

shott, S.P. Kothari, Charles Lee, Russ Lundholm (Stanford discussant), Paul Tetlock, Ro-

drigo Verdi, Frank Zhang (Illinois discussant), and seminar participants at the 2014

Citi Quant Research Conference, Cornell University, London Business School, MIT,

NASDAQ Economic Research, Stanford University Summer Camp, The University of

Texas at Austin, University of Illinois, and the 2015 Western Finance Association Meet-

ing for helpful suggestions. An online appendix to this paper can be downloaded at

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

217

Copyright C, University of Chicago on behalf of the Accounting Research Center,2017

218 T.L.JOHNSON AND E.C.SO

patterns in returns around anticipated periods of heightened inventory risks,

including alternative firm-level, industry-level, and macroeconomic informa-

tion events.

JEL codes: E44; G12; G14; G21; M40; M41

Keywords: earnings announcements; liquidity provision; transaction costs;

announcement returns; risk premia; bad news; asymmetric reaction

1. Introduction

A vast literature spanning accounting, finance, and economics studies stock

returns around firms’ earnings announcements. Researchers use stock re-

turns around earnings announcements to understand revisions in investor

expectations about firm value, risk premia driven by the release of earnings

news, and determinants of the equity market reaction to earnings news,

such as the credibility of financial reporting. Similarly, liquidity and trading

volumes around earnings announcements are commonly used to gauge the

extent of attention or disagreement among investors, information asymme-

try, and the news conveyed to investors.

This study provides both theoretical and empirical evidence that re-

turns, liquidity, and trading volume around earnings announcements are

asymmetrically influenced by frictions in the financial intermediary sec-

tor, which is composed of investment banks, broker-dealers, and market-

making firms, among others. Intermediaries provide liquidity by serving as

the trade counterparty in response to imbalanced demand between buy-

ers and sellers. In doing so, they are forced to take temporary positions

in the security, referred to as inventory, and thus expose themselves to

price fluctuations (inventory risk). Due to this exposure, the compensation

that intermediaries demand varies with their inventory positions as well as

the security’s risk profile. We show that these factors create asymmetries in

trading costs and, in turn, price discovery surrounding earnings announce-

ments by altering investors’ incentives to trade.

Our central hypothesis stems from prior evidence that intermediaries are

positively exposed to the market and hold positive average inventory posi-

tions, indicating that they are likely exposed to increased risks associated

with earnings announcements.1As a result of this exposure, we predict

that intermediaries demand greater compensation for providing liquidity

to sell orders, which would exacerbate their exposure, relative to buy or-

ders, which would shield them from announcement risks by helping them

1For example, Brunnermeier and Pedersen [2009] report that broker-dealer firms have

median market betas above one. Adrian and Shin [2010] show that financial intermediaries’

leverage is highly procyclical and argues that this is because they expand their positions in

response to market booms. Similarly, studies using proprietary data on market makers’ inven-

tory positions show that they tend to hold positive inventories (Madhavan and Smidt [1993],

Comerton-Forde et al. [2010]).

ASYMMETRIC TRADING COSTS PRIOR TO EARNINGS ANNOUNCEMENTS 219

reduce net exposure before the announcement (referred to as “getting

flat” in the industry). This asymmetry acts like an endogenous short-sale

cost by discouraging selling, which in turn causes an upward bias in prean-

nouncement price discovery and returns that reverses postannouncement.

We refer to this hypothesis about the causes and effects of asymmetric trad-

ing costs (ATCs) as the ATC hypothesis. We formalize the ATC hypothesis

by developing a model of trading before an information event involving a

financial intermediary with positive exposure to the announcing firm’s eq-

uity. Our model predicts the following chain of events: the intermediary sets

asymmetric prices for liquidity to reduce its inventory before the announce-

ment, causing a positive bias in preannouncement price discovery, which

leads to an upward bias in preannouncement returns that corrects once

the news is released. The upward bias may initially seem counterintuitive

because, for most agents, the desire to sell would lead to the opposite—a

decrease in average prices before announcements. However, when liquid-

ity providers in our model seek to reduce their inventories, they use their

pricing power to induce buy demands and ensure that average prices are

above fundamental value.

Our first set of empirical results supports the cornerstone of the ATC hy-

pothesis, that intermediaries provide liquidity asymmetrically before earn-

ings announcements. Using short-term return reversals as a proxy for the

compensation intermediaries demand for this, as in Campbell, Grossman,

and Wang [1993], we show that reversals become increasingly asymmetric

before announcements. Reversals associated with net selling pressure (i.e.,

recent losers) increase several days before announcements and peak at

more than three times normal levels immediately before announcements,

whereas no discernible trend exists for buying pressure. These results are

consistent with our model’s prediction that intermediaries demand greater

compensation for providing liquidity to sellers, relative to buyers, and thus

that traders incur higher relative costs of impounding negative news into

preannouncement prices.

A natural question is whether the magnitude of an intermediary-based

story could explain the magnitude of our findings. We estimate that the

asymmetry in liquidity provision in our sample peaks at 54 basis points

(bp) on day t−1, which is based on the spread in return reversal mag-

nitudes across quintiles of prior day net buying/selling pressure. A basis of

comparison for this estimate is the returns of unconditional return-reversal

strategies, which peak in our sample at around 100 bp before earnings an-

nouncements. Another reasonable benchmark of comparison is the typical

magnitude of price pressures due to specialist inventory positions, which

Hendershott and Menkveld [2014] estimate is 49 bp, with a half-life of

0.92 days. These comparisons suggest that our results are not only econom-

ically meaningful but also comport with plausibility bounds established by

research.

Our second set of empirical results provide support for our model’s pre-

diction that investors respond to ATCs by trading more aggressively on