Earnings Metrics, Information Processing, and Price Efficiency in Laboratory Markets

• Date: 01 June 2015
• Author: BRIAN J. WHITE,JESSEN L. HOBSON,W. BROOKE ELLIOTT
• Published date: 01 June 2015
• DOI: http://doi.org/10.1111/1475-679X.12080

DOI: 10.1111/1475-679X.12080

Journal of Accounting Research

Vol. 53 No. 3 June 2015

Printed in U.S.A.

Earnings Metrics, Information

Processing, and Price Efficiency

in Laboratory Markets

W. BROOKE ELLIOTT,

∗JESSEN L. HOBSON,

∗

AND BRIAN J. WHITE

†

Received 24 December 2012; accepted 16 February 2015

ABSTRACT

An enduring issue in financial reporting is whether and how salient summary

measures of firm performance (“earnings metrics”) affect market price effi-

ciency. In laboratory markets, we test the effects of salient earnings metrics,

which vary in how they combine persistent and transitory elements, on in-

vestor information search, beliefs about value, offers to trade, and market

price efficiency. We find that including transitory elements in salient earn-

ings metrics causes traders to search unnecessarily for further information

about these elements and to overestimate their effect on fundamental value

relative to a rational benchmark. In contrast, separately displaying persistent

elements in earnings increases the accuracy of traders’ value estimates. Prices

generally reflect traders’ beliefs about value, and prices are most efficient

when transitory elements are excluded from earnings metrics entirely. Our

study contributes to research on salience effects in financial reporting by

showing that including transitory elements in salient earnings metrics causes

∗University of Illinois at Urbana–Champaign; †The University of Texas at Austin.

Accepted by Haresh Sapra. We thank Robert Bloomfield, Robert Bowen, Qi Chen, Jane

Jollineau, Steve Kachelmeier, Lisa Koonce, Bob Libby, Eldar Maksymov, Mark Nelson, Mark

Peecher, Nick Seybert, Theo Sougiannis, Greg Waymire, Jeff Wilks, Flora Zhou, anonymous

reviewers for the 2013 FARS Midyear Meeting and 2013 Journal of Accounting Research Confer-

ence, and workshop participants at the 2013 Journal of Accounting Research Conference, Cornell

University, the Universities of Iowa, San Diego, and Texas and the BYU Accounting Research

Symposium for helpful comments. We also thank Stefan Palan and Marcelo Tyszlerfor z-Tree

assistance.

555

Copyright C, University of Chicago on behalf of the Accounting Research Center,2015

556 W.B.ELLIOTT,J.L.HOBSON,AND B.J.WHITE

inefficient information search and biased beliefs about value that can aggre-

gate to affect market prices. We also contribute to research in experimental

markets by showing that redundant disclosure is not always beneficial; redun-

dant disclosure of transitory earnings elements, in particular, appears to have

negative consequences for investor behavior and market efficiency.

JEL codes: C91; C92; G02; G14; M41

Keywords: earnings metrics; bounded rationality; information search; fun-

damental value; market price efficiency; laboratory markets

1. Introduction

An enduring issue in financial reporting is whether and how salient sum-

mary measures of firm performance (“earnings metrics”) affect investor

behavior and market price efficiency. A particularly important aspect of this

issue is how persistent and transitory elements are combined in earnings

metrics (Paton and Littleton [1940], Schipper and Vincent [2003], Dechow

and Schrand [2004]). For example, persistent and transitory elements may

be aggregated in a single earnings metric or disaggregated into separate

metrics or transitory elements may be excluded from prominent earnings

metrics altogether. The issue is also particularly timely, as the Financial

Accounting Standards Board (FASB) recently mandated that firms display

comprehensive income more prominently in order to “increase the promi-

nence of items reported in other comprehensive income” (FASB [2011],

p. 1).1In this study, we provide evidence relevant to this issue by testing

the effect of disclosing prominent earnings metrics in laboratory markets.

Although earnings have been a major focus of accounting research for

decades, our tests of how earnings metrics affect investor behavior and mar-

ket prices differ from previous studies in important ways. First, research on

salience effects in financial reporting has shown that more salient informa-

tion, including earnings information, is weighted more heavily in individ-

ual judgments and decisions (e.g., Hirst and Hopkins [1998], Maines and

McDaniel [2000], Elliott [2006]). While these studies show that salience

matters for individual judgments, it is also important to test, as we do here,

how these salience-affected individual judgments and decisions compare to

a rational benchmark in order to determine whether they are helpful or

harmful to investors’ judgments, and whether salience effects aggregate to

influence market prices.

Second, earnings metrics are among the most common forms of re-

dundant information disclosure; that is, the content of earnings can be

inferred from other information in the financial statements. Previous

research in experimental markets finds that, because investors are

1Many other comprehensive income items, such as unrealized investment gains and losses,

are by nature less persistent than operating items (e.g., Linsmeier et al. [1997], Chambers

et al. [2007], Bamber et al. [2010], Jones and Smith [2011]).

EARNINGS METRICS AND PRICE EFFICIENCY 557

boundedly rational (i.e., they display limited attention and processing

power, as in Hirshleifer and Teoh [2003]), providing redundant disclosures

that simplify information improves market price efficiency (Dietrich et al.

[2001], Hobson [2011]). However, these prior studies confound redun-

dancy with simplification. Importantly, earnings metrics, in the natural

setting and in this study, differ from the redundant information in previous

studies in that they do not necessarily simplify the processing task for

traders. Rather, they highlight earnings elements with differing persistence

characteristics. Users must then decide which other information to acquire

and how to impound earnings and other information into value before

deciding the price at which they are willing to trade. Because earnings

metrics are so central to financial reporting, it is important to test how

they affect information search, beliefs about value, offers to trade, and

ultimately market price efficiency.

We conduct our study in laboratory markets, in which three different

earnings metrics are prominently displayed: (1) earnings metrics that

aggregate persistent and transitory elements (“aggregated earnings”),

(2) earnings metrics that disaggregate persistent and transitory elements

(“disaggregated earnings”), and (3) earnings metrics that include only

persistent elements and exclude transitory elements (“core earnings”).

We compare each of these to a control condition in which no earnings

metric is saliently displayed, while allowing access under each alternative

to all earnings elements (both persistent and transitory). Using laboratory

markets provides several advantages, including the ability to establish

a rational benchmark for individual- and market-level measures and to

separately measure information search, beliefs about firm value, offers to

trade, and market aggregation.

The information available to traders in our markets is derived from the

model described by Ohlson ([1995, 1999] and subsequent work). Firm per-

formance each period includes both persistent and transitory earnings el-

ements. Because persistent elements are capitalized into value at a higher

rate than transitory elements, it is necessary to separately identify persistent

elements in order to derive fundamental value. As a result, increasing the

salience of persistent earnings elements by displaying them in a core or

disaggregated earnings metric should help boundedly rational traders to

estimate fundamental value more accurately. In addition, however, because

aggregated and disaggregated earnings metrics also increase the salience of

transitory elements (indeed, elevating their salience to be equal with that of

persistent elements), boundedly rational traders may overreact to transitory

elements in aggregated and disaggregated earnings markets. Thus, we ex-

pect the greatest information processing benefits in core earnings markets.

With respect to market prices, if prices reflect boundedly rational

traders’ on-average beliefs about fundamental value (Hirshleifer and Teoh

[2003]), then market prices should follow the same pattern as traders’

estimates of firm value relative to expected fundamental value. That

is, prices should be more efficient (i.e., they should deviate less from