Disentangling Managers’ and Analysts’ Non‐GAAP Reporting

• Date: 01 September 2018
• Published date: 01 September 2018
• Author: BENJAMIN C. WHIPPLE,JEREMIAH W. BENTLEY,KURT H. GEE,THEODORE E. CHRISTENSEN
• DOI: http://doi.org/10.1111/1475-679X.12206

DOI: 10.1111/1475-679X.12206

Journal of Accounting Research

Vol. 56 No. 4 September 2018

Printed in U.S.A.

Disentangling Managers’ and

Analysts’ Non-GAAP Reporting

JEREMIAH W. BENTLEY,

∗THEODORE E. CHRISTENSEN,

†

KURT H. GEE,

‡AND BENJAMIN C. WHIPPLE

†

Received 27 May 2015; accepted 19 January 2018

ABSTRACT

Researchers frequently proxy for managers’ non-GAAP disclosures using per-

formance metrics available through analyst forecast data providers (FDPs),

such as I/B/E/S. The extent to which FDP-provided earnings are a valid

proxy for managers’ non-GAAP reporting, however, has been debated exten-

sively. We explore this important question by creating the first large-sample

data set of managers’ non-GAAP earnings disclosures, which we directly com-

pare to I/B/E/S data. Although we find a substantial overlap between the

two data sets, we also find that they differ in systematic ways because I/B/E/S

(1) excludes managers’ lower quality non-GAAP numbers and (2) sometimes

provides higher quality non-GAAP measures that managers do not explic-

itly disclose. Our results indicate that using I/B/E/S to identify managers’

non-GAAP disclosures significantly underestimates the aggressiveness of their

reporting choices. We encourage researchers interested in managers’ non-

GAAP reporting to use our newly available data set of manager-disclosed

∗Isenberg School of Management, University of Massachusetts Amherst; †J. M. Tull School

of Accounting, TerryCollege of Business, University of Georgia; ‡Graduate School of Business,

Stanford University.

Accepted by Philip Berger. Wethank an anonymous referee, Terrence Blackburne, Nerissa

Brown, Dave Burgstahler, Michael Carniol (AAA Discussant), Matt DeAngelis (FARS Discus-

sant), Mike Drake, Kevin Johnson, Sarah McVay,and workshop participants at Brigham Young

University,the University of Washington, George Washington University, the University of Min-

nesota, the 2015 FARS Mid-year Meeting, and the 2015 AAA Annual Meeting for helpful com-

ments and suggestions. We also thank Jason Ashby,Alex Felsing, Enrique Gomez, Lisa Hinson,

Megan Jones, Elizabeth Schmidt, and Yushi Tian for their valuable research assistance.

1039

CUniversity of Chicago on behalf of the Accounting Research Center,2018

1040 J.W.BENTLEY,T.E.CHRISTENSEN,K.H.GEE,AND B.C.WHIPPLE

non-GAAP metrics because it more accurately captures managers’ reporting

choices.

JEL codes: M40; M41; M48

Keywords: non-GAAP earnings; financial analysts; voluntary disclosure;

earnings announcements; reporting quality

1. Introduction

Numerous studies in the non-GAAP disclosure literature examine the in-

centives that motivate managers’ non-GAAP reporting choices (e.g., Bhat-

tacharya et al. [2003], Doyle, Lundholm, and Soliman [2003]). A primary

challenge facing these studies, however, is that managers’ non-GAAP met-

rics have not been available in an archived data set, leading researchers to

either engage in costly hand collection or proxy for managers’ disclosures

using data from analyst forecast data providers (hereafter FDPs), such as

I/B/E/S. Some researchers, however, question the appropriateness of us-

ing FDP data to proxy for managers’ reporting because (1) FDPs calcu-

late earnings based on how analysts forecast firm performance (Thomson

Reuters [2013]) and (2) it is unclear whether analysts systematically devi-

ate from managers’ disclosed metrics. As a result, an ongoing debate in the

non-GAAP reporting literature questions the appropriateness of using FDP

data to proxy for managers’ non-GAAP disclosures (e.g., Easton [2003],

Lambert [2004], Berger [2005], Beyer et al. [2010]). Bradshaw and Soli-

man [2007, p. 736] highlight this point stating that:

There is no compelling evidence in the literature that identifies

[FDP] earnings as a predominantly analyst-driven, manager-driven, or

forecast data provider-driven phenomenon. This creates hindrances

to...researchers attempting to use data bases to interpret the individual

incentives and actions of the analysts, managers, and data providers.

We help settle this debate by creating the first large-sample data set

of manager-disclosed non-GAAP earnings metrics. We then directly com-

pare our manager data set to FDP-provided non-GAAP performance data,

specifically the I/B/E/S actual earnings number, and examine how well

FDP metrics capture managers’ non-GAAP reporting. Our analyses indicate

that, although the I/B/E/S and manager non-GAAP data sets largely agree,

they also differ in systematic ways, consistent with predicted differences in

managers’ and analysts’ non-GAAP reporting choices. Moreover, we pro-

vide evidence that the quality of non-GAAP metrics also differs across the

I/B/E/S and manager data sets. In particular, we find that I/B/E/S cap-

tures managers’ more informative non-GAAP disclosures, while it excludes

their more aggressive reporting. In addition, I/B/E/S sometimes con-

tains high-quality non-GAAP performance measures in periods when man-

agers do not explicitly disclose non-GAAP metrics. As a result, our analyses

suggest that using I/B/E/S to examine managers’ non-GAAP reporting

DISENTANGLING NON-GAAP REPORTING 1041

incentives underestimates the aggressiveness of their reporting choices.

Thus, we encourage researchers interested in examining managers’ non-

GAAP reporting incentives to use our publically available data set.1

To identify managers’ non-GAAP earnings disclosures, we (1) program-

matically search quarterly earnings announcements in SEC 8-K filings to

determine whether managers explicitly report a non-GAAP earnings per

share (EPS) metric and (2) extract managers’ non-GAAP EPS metrics from

the earnings announcements. This approach yields a data set of 115,370

quarterly observations for fiscal years spanning 2003 to 2012, and is sig-

nificantly larger than hand-collected data sets used in prior studies (e.g.,

Bhattacharya et al. [2003], Marques [2006]).2We compare our manager

data set to I/B/E/S and test for systematic differences. One reason the data

sets might differ is that Thomson Reuters defines I/B/E/S actual earnings

based on how analysts forecast firm performance, while the manager data

set captures managers’ observed non-GAAP reporting. If analysts do not

always mirror managers’ disclosures, then differences across the data sets

should occur in situations where we expect managers’ and analysts’ report-

ing choices to differ.

When we compare the manager and I/B/E/S data sets, we find that they

frequently provide the same non-GAAP earnings information. Neverthe-

less, 23% of the I/B/E/S non-GAAP numbers relate to instances where

managers do not explicitly disclose a non-GAAP EPS metric (which we la-

bel as the “I/B/E/S-Only” scenario), while 17% of the manager non-GAAP

numbers relate to instances where I/B/E/S does not contain a non-GAAP

EPS metric (which we label as the “Manager-Only” scenario).3We also find

that the differences in the data sets are not random, but occur in instances

where we predict that managers and analysts make different reporting

choices. For example, I/B/E/S-Only reporting is more likely to occur when

non-GAAP metrics are informative but costly for managers to report, such

as in the presence of transitory gains and when litigation risk is high. In

contrast, Manager-Only non-GAAP reporting is more likely to occur after

the implementation of SFAS 123R, when managers are reluctant to reduce

across-quarter comparability and lower their reported earnings metric by

expensing stock-based compensation. Overall, our evidence indicates that

using I/B/E/S to proxy for managers’ non-GAAP reporting systematically

misses some manager-disclosed non-GAAP numbers and incorrectly in-

cludes some analyst-adjusted metrics that managers do not explicitly report.

Next, we examine whether the quality of non-GAAP earnings varies with

differences across the data sets. Collectively, our evidence indicates that the

1The manager non-GAAP data set is publically available at https://sites.google.com/

view/kurthgee/data.

2We estimate that the accuracy with which we identify firms with and without non-GAAP

earnings is 95%, with an overall accuracy of non-GAAP EPS extraction of 86%.

3Moreover, even when both managers and I/B/E/S provide a non-GAAP performance

metric, the numbers sometimes differ (4% of their non-GAAP metrics).