Direct Evidence on the Informational Properties of Earnings in Loan Contracts

• Published date: 01 May 2017
• Author: SCOTT D. DYRENG,RAHUL VASHISHTHA,JOSEPH WEBER
• DOI: http://doi.org/10.1111/1475-679X.12168
• Date: 01 May 2017

DOI: 10.1111/1475-679X.12168

Journal of Accounting Research

Vol. 55 No. 2 May 2017

Printed in U.S.A.

Direct Evidence on the

Informational Properties

of Earnings in Loan Contracts

SCOTT D. DYRENG,

∗RAHUL VASHISHTHA,

∗

AND JOSEPH WEBER

†

Received 18 November 2015; accepted 2 February 2017

ABSTRACT

Using a sample of firms that disclose the realizations of earnings used for

determining covenant compliance in loan contracts, we provide direct evi-

dence on the informational properties of earnings used in the performance

covenants included in debt contracts. We find that the earnings measure used

in performance covenants does not exhibit asymmetric loss timeliness and

has significantly greater cash flow predictive ability than GAAP measures of

earnings. We suggest that these results reflect the idea that contracting parties

∗Fuqua School of Business, Duke University; †Sloan School of Management, Massachusetts

Institute of Technology.

Accepted by Douglas Skinner. We thank Robert Hills for valuable research assistance,

as well as Ryan Ball, Mary Barth, Qi Chen, Richard Frankel, Aloke Ghosh, Christian Leuz,

Greg Miller, Katherine Schipper, Richard Sloan, Mark Soliman, Mohan Venkatachalam, two

anonymous referees, and seminar participants at Baruch College, Carnegie Mellon Uni-

versity, Columbia University, Duke University, FARS Conference (2015), Florida State Uni-

versity, Journal of Accounting Research Conference (2016), University of Illinois – Chicago,

University of Miami, University of Michigan, University of Missouri, Northwestern Univer-

sity, Nazarbayev University, Pennsylvania State University, University of Southern California,

University of Utah, Yale University, and Southeast Summer Accounting Research Confer-

ence (2014) for helpful comments and discussions. We thank Ted Christensen for shar-

ing data on pro forma earnings. An online appendix to this paper can be downloaded at

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

371

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

372 S.D.DYRENG,R.VASHISHTHA,AND J.WEBER

design accounting rules for performance covenants to enhance their efficacy

as “tripwires.”

JEL codes: G32; M40; M41

Keywords: earnings properties; debt contracts; cash flow prediction; con-

servatism

1. Introduction

One of the key roles of accounting is to generate information about a

firm’s performance that is useful for writing debt contracts. Surprisingly,

we know very little about the informational properties of the actual earn-

ings measures that are used in covenants, probably because realizations of

these numbers are not easily obtainable. In this study, we identify a sample

of firms that disclose performance covenant earnings realizations, as de-

fined in their debt contracts, and examine the properties of these earnings

measures.

There is considerable debate surrounding the properties of accounting

earnings that are desirable for use in debt covenants. Several studies argue

that because lenders are particularly concerned with downside risk, they

prefer conservative earnings measures (e.g., Watts [2003a, b]). They argue

that conservative accounting facilitates more timely creditor intervention

by reducing the verifiability threshold of losses that can trigger covenant

violations, allowing lenders to mitigate their downside lending risk. Oth-

ers (e.g., Leuz [2001], Gigler et al. [2009]) highlight that while lowering

the verifiability threshold of losses facilitates timely creditor intervention, it

also increases the incidence of costly “false positives” caused by losses that

are not indicative of future performance. Still other research suggests that

accounting rules in debt contracts are designed to remove transitory items

that are not indicative of future performance (e.g., Li [2010]). The argu-

ment is that transitory items can dampen the ability of earnings to predict

future cash flows, leading to unnecessary covenant violations that induce

costly renegotiations and inefficient wealth transfers.

We directly examine the extent of conservatism and cash flow predic-

tive ability of earnings used in performance covenants of private debt con-

tracts. We focus our attention on performance covenant earnings because

these covenants represent the main contractual mechanism through which

borrowers’ periodic earnings performance directly affects contractual out-

comes (e.g., Christensen and Nikolaev [2012]). To execute our tests, we

use key word searches of public filings using Lexis-Nexis and 10-K Wiz-

ard to identify a set of firms that disclose earnings realizations as defined

under the accounting rules specified in the debt contract for use in per-

formance covenants (hereafter, PERF COV EARNINGS).1After imposing

1We provide an example of a disclosure in the appendix.

INFORMATIONAL PROPERTIES OF EARNINGS IN LOAN CONTRACTS 373

necessary data requirements, this procedure yields a sample of 128 firms,

corresponding to 1,721 observations.

For this sample of firms, we compare the informational properties of

PERF COV EARNINGS to those of two FASB GAAP-based earnings mea-

sures: net income as defined by FASB GAAP (hereafter, NET INCOME)and

earnings before interest, taxes, depreciation, and amortization (hereafter,

EBITDA).2Our main focus is to examine whether PERF COV EARNINGS

differ from NET INCOME; however, benchmarking PERF COV EARNINGS

against EBITDA is also useful for exploring to what extent contractual ad-

justments beyond interest, taxes, and depreciation shape the properties of

PERF COV EARNINGS.

Our initial descriptive analyses reveal marked differences in the magni-

tudes of PERF COV EARNINGS and NET INCOME and EBITDA.Wefindthat

PERF COV EARNINGS is greater than NET INCOME (EBITDA)formore

than 99% (84%) of our observations. The magnitudes of the differences

are also quite large: for the median firm, NET INCOME (EBITDA) is lower

than PERF COV EARNINGS by about 87% (10%). Overall, the descriptive

evidence suggests that, on average, the contractual adjustments made in

debt contracts are economically large and income increasing.

In the next analysis, we use the Basu [1997] measure of asymmetric timely

loss recognition to examine the extent of conditional conservatism in PERF

COV EARNINGS and FASB GAAP earnings. Consistent with prior research,

we find that NET INCOME exhibits asymmetric loss timeliness for our sam-

ple firms. However, we also find that PERF COV EARNINGS is not condition-

ally conservative.

We next compare the cash flow predictive ability of these three earnings

measures by using the adjusted R2from regressions of one-year-ahead op-

erating cash flows on each of the three different earnings measures. We

find that the adjusted R2from the regression of PERF COV EARNINGS is

both statistically and economically (over 50%) larger than the adjusted R2

from corresponding regressions for GAAP-based earnings measures. These

results suggest that PERF COV EARNINGS has significantly greater cash flow

predictive ability than FASB GAAP-based earnings measures.

Because our analysis is based on a small sample of firms that disclose

contractual earnings, a concern is that these firms may have unusual debt

contracts and our inferences may not generalize to the broader sample

of borrowers with performance covenants. An examination of the relevant

disclosure rules (see section 2) suggests that our sample is likely to contain

firms for which performance covenants are a material term in their debt

contracts and the violation of these covenants could materially affect the

2GAAP does not require firms to report EBITDA. When we use the abbreviation EBITDA,

we are referring to the sum of the firm’s NET INCOME, depreciation and amortization ex-

pense, interest expense, and tax expense, as reported on the firm’s income statement (per

Compustat).