Performance Measures in Earnings‐Based Financial Covenants in Debt Contracts

• Published date: 01 September 2016
• Author: NINGZHONG LI
• DOI: http://doi.org/10.1111/1475-679X.12125
• Date: 01 September 2016

DOI: 10.1111/1475-679X.12125

Journal of Accounting Research

Vol. 54 No. 4 September 2016

Printed in U.S.A.

Performance Measures in

Earnings-Based Financial Covenants

in Debt Contracts

NINGZHONG LI

∗

Received 16 July 2012; accepted 16 April 2016

ABSTRACT

This paper examines how performance measures are defined in major

earnings-based financial covenants in loan contracts to shed light on the eco-

nomic rationales underlying the contractual use of performance measures.

I find an earnings-based covenant is typically based on a performance mea-

sure close to earnings before interest, tax, amortization, and depreciation

expenses (EBITDA). However, my empirical analyses show that EBITDA is

less useful in explaining credit risk than earnings before interest and tax ex-

penses (EBIT) and even the bottom-line net income. Thus, measuring credit

risk cannot fully explain the choice of accounting performance measures

in earnings-based covenants. I conjecture that contracting parties choose an

EBITDA-related measure, instead of a measure calculated after depreciation

and amortization expenses (e.g., EBIT), to make the performance measure

∗University of Texas at Dallas.

Accepted by Philip Berger. I am extremely grateful to an anonymous referee for his or

her guidance. I thank Ray Ball, Ian Gow, Anya Kleymenova, Yun Lou, Chul Park (discus-

sant), Madhav Rajan, Scott Richardson, Lakshmanan Shivakumar, Bin Srinidhi, Irem Tuna,

Jeffery Wooldridge, Jingjing Zhang (discussant), and workshop participants at City University

of Hong Kong, London Business School, London School of Economics, University of Hous-

ton, University of Texas at Dallas, and Washington University, and participants of the 2011

AAA Annual Meeting and the 2011 CAPANA Annual Meeting for valuable comments. I am

grateful to Amir Sufi for sharing the loan agreement data online and Florin Vasvari for shar-

ing the cleaned bond transaction data. I thank Ying Huang, Xin Li, Bo Liu, Liping Lu, Connie

Neish, and Yu Xie for excellent research assistance, and the London Business School RAMD

Fund and University of Texas atDallas for financial support. All errors are my own.

1149

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

1150 N.LI

less sensitive to investment activities, which can be controlled through other

contractual terms, such as a restriction on capital expenditure, and provide

empirical evidence consistent with this conjecture.

JEL codes: G21; G32; M41

Keywords: performance measure; earnings-based covenant; debt contract;

incomplete contracting theory; credit risk

1. Introduction

The incomplete contracting theory, which emphasizes the efficient allo-

cation of control rights in debt-contracting relationships in the presence

of unforeseeable contingencies and agency conflicts between shareholders

and debtholders, views debt covenants as trip wires that give lenders an op-

tion to renegotiate loan terms by threatening default following a decline in

economic performance (e.g., Grossman and Hart [1986], Hart and Moore

[1988, 1990], Aghion and Bolton [1992], Dewatripont and Tirole [1994],

Garleanu and Zwiebel [2009]). While the theory generally argues that con-

tingent control rights allocation requires a contractible accounting signal

that exhibits considerable correlation with the state of nature, it is unclear

what specific economic rationales underlie the choice of accounting sig-

nals in debt covenants (Christensen and Nikolaev [2012]). This study aims

to shed light on this issue by examining how performance measures are de-

fined in major earnings-based financial covenants, including interest cov-

erage (IC), fixed charges coverage (FCC), debt-to-cash flows (DCF), min-

imum cash flows (CF), and debt service coverage (DSC) covenants, in a

large sample of loan agreements.1

Prior studies generally conjecture that performance measures in

earnings-based covenants are chosen to reflect borrowers’ credit risk (e.g.,

Ball, Bushman, and Vasvari [2008], Christensen and Nikolaev [2012]).

These studies argue that contracting parties improve the efficiency of a debt

contract by including in the contract a covenant that measures changes in

credit risk. Consistent with this argument, Christensen and Nikolaev [2012]

show that debt contracts’ reliance on earnings-based covenants is positively

associated with the ability of accounting information to explain credit risk.2

Relatedly, Ball, Bushman, and Vasvari [2008] find that when loan contracts

include performance pricing provisions, the likelihood that the perfor-

mance measure used is an accounting ratio, rather than a credit rating,

increases with the ability of accounting information to explain credit risk.

1Appendix C provides definitions of these covenants. Performance measures mean the

numerators of IC, FCC, and DSC covenants; denominators of DCF covenants; and measures

in CF covenants.

2Earnings-based covenants are called performance covenants in Christensen and Nikolaev

[2012].

PERFORMANCE MEASURES IN EARNINGS-BASED FINANCIAL COVENANTS 1151

My first set of analyses examines whether measuring credit risk can fully

explain the choice of accounting performance measures in earnings-based

covenants. I first manually code the detailed definitions of performance

measures in the five earnings-based covenants using a large sample of loan

contracts. I find that the contractual definitions typically begin with GAAP

(generally accepted accounting principles) net income but exclude transi-

tory earnings and interest, tax, depreciation, and amortization (D&A) ex-

penses. There could be many other adjustments, but their frequencies are

typically low. Thus, cash flows are rarely used as a performance measure in

the covenant measurement, and a typical performance measure is close to

earnings before interest, tax, and D&A expenses (EBITDA).3

I then examine whether the dominant use of an EBITDA-related mea-

sure in an earnings-based covenant is due to EBITDA being more useful in

measuring credit risk than a measure calculated after D&A expenses, such

as earnings before interest and tax expenses (EBIT) or even the bottom-

line net income. Measuring credit risk with bond yield, firm credit rating,

the probability of default estimated from the Black–Sholes–Merton option-

pricing model based on Merton [1974], as well as a composite measure

based on the first principal component of these measures, I find strong

evidence that EBITDA is less useful than EBIT and even the bottom-line

net income in measuring credit risk. These results hold in the sample

of firms that consistently use EBITDA-related measures in their earnings-

based covenants in my contract sample, as well as in a broader sample of

firms in the DealScan database. These findings, combined with the dom-

inant use of an EBITDA-related measure in an earnings-based covenant,

suggest that the contracting parties’ decision on which performance mea-

sure to use in earnings-based covenants is based on something more than

the ability of the measure to reflect credit risk.

I further explore why the majority of earnings-based covenants are based

on an EBITDA-related measure despite the fact that including D&A ex-

penses generally improves its usefulness in measuring credit risk. As ex-

isting theories provide little direct insight on this issue, my analyses are

primarily exploratory. Prior studies have shown that debt covenants and

financial reporting influence corporate investment decisions (e.g., Biddle

and Hilary [2006], Chava and Roberts [2008], Roberts and Sufi [2009],

Beatty, Liao, and Weber [2010a, b]). I conjecture that contracting parties

choose an EBITDA-related measure, instead of a measure calculated after

D&A expenses (e.g., EBIT and net income), to make the performance mea-

sure less sensitive to investment activities, because other mechanisms (e.g.,

restrictions on capital expenditure) can be used to control for these activi-

ties.4

3Throughout the paper (except in tables 2, A1, and A2), EBITDA is empirically defined as

earnings before extraordinary items plus interest, tax, and D&A expenses; EBIT is empirically

defined as earnings before extraordinary items plus interest and tax expenses.

4I thank the referee for proposing this argument.