Sign reversal in the relationship between income smoothing and cost of debt

• Author: Edward Owens,Dan Amiram
• DOI: http://doi.org/10.1111/jbfa.12295
• Date: 01 January 2018
• Published date: 01 January 2018

DOI: 10.1111/jbfa.12295

Sign reversal in the relationship between income

smoothing and cost of debt

Dan Amiram1,2 Edward Owens3

1ColumbiaUniversity Graduate School of

Business,New York, NY 10027, United States

2CollerSchool of Management, Tel Aviv Univer-

sity,TelAviv 6997801, Israel

3GoizuetaBusiness School, Emory University,

Atlanta,GA 30322, United States

Correspondence

EdwardOwens, Goizueta Business School, Emory

University,Atlanta, GA 30322, United States.

Email:ed.owens@emory.edu

Abstract

Despite the fact that income smoothing by managers is a perva-

sive phenomenon that has been widely researched, extant litera-

ture provides incomplete evidence on how smoothing is associated

with cost of debt in general, and in the private loan market in par-

ticular. The institutional factors associated with private loan con-

tracts, combined with the theoretical motivations for smoothing,

make it unclear whether smoothing will be positively, negatively,

or not associated with loan spread. Using both cross-country and

within-country analyses on an international sample of private loans,

we predict and provide evidence that income smoothing is associ-

ated with lower cost of debt when the threat of private benefits con-

sumption by managers is low, but is associated with higher cost of

debt when the threat of private benefits consumption bymanagers is

high. We provide the first evidencein the literature that the garbling

effect of smoothing can predictably dominate the signaling view of

smoothing in debt contract design, and we identify private benefits

consumption threat as the feature of the contracting environment

that empirically reveals a sign reversal in the relationship between

smoothing and cost of debt.

KEYWORDS

debt contracts, income smoothing, private benefits

1INTRODUCTION

Despite the fact that income smoothing by managers is a pervasive phenomenon that has been widely researched,

extant literature provides incomplete evidence concerning how borrower income smoothing is associated with cost

of debt in general, and in the private loan market in particular.The institutional factors associated with private loan

contracts,combined with the theoretical motivations for smoothing, makeit unclear ex a nte whether smoothing will be

positively,negatively, or not associated with loan spread. Thus far, extant literaturehas examined settings where only

the signaling effect of smoothing dominates, and therefore concludes that smoothing is associated with lower cost of

debt. In this study, we extendthe literature. Specifically, we take advantage of an international setting to predict and

identifya theoreticallymotivated feature of the contracting environment that reveals a sign reversal in the relationship

40 c

2017 John Wiley & Sons Ltd wileyonlinelibrary.com/journal/jbfa JBus Fin Acc. 2018;45:40–71.

AMIRAM ANDOWENS 41

between smoothing and cost of debt, thereby providing the first evidence in the literature that the garbling effect of

smoothing can predictably dominate the signaling effect of smoothing in debt contract design.

Income smoothing is the exercise of managerial discretion to alter the time profile of earnings to reduce the vari-

ability of a firm's reported income stream (e.g.,Beidleman, 1973; Fudenberg & Tirole, 1995; Trueman & Titman, 1988).1

The literature focuses on two keyalternative motivations for income smoothing. On one hand, managers may smooth

earnings in an effort to truthfully conveytheir private information about underlying economic earnings and associated

risk to capital providers and other marketparticipants (i.e., the information signaling view of smoothing) (e.g., Trueman

& Titman, 1988). Alternatively,managers may smooth earnings (regardless of their private information about longer-

term economic earnings) in an effort to avoid intervention by outsiders to facilitate consumption of private benefits

from the firm (i.e., the information garbling view of smoothing) (e.g., Fudenberg & Tirole, 1995).2,3

The information signaling and information garbling views of smoothing yield directionally opposing predictions con-

cerning the association between observed smoothing and cost of debt. The signaling view suggests that smoothing is

an attempt by managers to signal that economic earnings are less volatile than would be inferred from unsmoothed

(reported) earnings, and therefore smoothing suggests lower probability ofdefault (e.g., Merton, 1974). In contrast, the

garbling view suggests that smoothing is associated with higher risk of private benefits consumption. Tothe extent that

private benefits consumption extracts firm wealth at the expense of debtholders (Lin, Ma, Malatesta, & Xuan, 2011),

the information garbling view of smoothing would suggest higher loss given default and higher probability of default

(Fudenberg & Tirole, 1995). In the context of these alternative views of smoothing, ceteris paribus, we predict cost of

debt will be lower if lenders take the information signaling view of smoothing, and will be higher if lenders take the

information garbling view of smoothing, ceteris paribus. Moreover, as lenders only lose money if a borrower actually

defaults, we expect the effects of smoothing on cost of debt to be stronger when the probability of default is higher.

Extant literature provides mixed evidence on the association between smoothing and cost of capital in the equity

market. This evidence does not translate into debt marketsbecause creditors have different payoff functions than do

equityholders, and can protect themselves through other contracting mechanisms (e.g., financial covenants). Existing

evidence from research on the effects of smoothing in debt markets provides evidence that either smoothing is asso-

ciated with lower cost of debt, or that there is no association. That is, extant literature does not document anysetting

or conditions where smoothing is associated with higher cost of debt, which is noteworthy given the strong theoret-

ical literature that suggests that smoothing can, in some situations, represent garbling by management, which could

negatively affect lenders.

Our conjecture is that extant literature has not examinedthe association between smoothing and cost of debt in a

setting or manner in which the garbling effect can reveal itself. Evidence from the US bond marketthat smoothing is

associatedwith lower cost of debt is not generalizable to broader credit markets, because the bond market is populated

by relatively high quality, transparent borrowers. Thus, the garbling view of smoothing seems relatively unlikelyto

manifest (Bharath, Sunder,& Sunder, 2008).

Studies that examinesmoothing in the private loan market find either a negative or no association between smooth-

ing and cost of debt, either because they do not focus on the subset of borrowers where garbling effects are most likely

to dominate, or because lenders can obtain private information from borrowers, which may mitigate any association

1The overallvolatility of a firm's earnings relative to cash flows (i.e., ‘income smoothness’) is determined by two components. The first component, which we

refer to as ‘fundamental smoothness’,is determined by the inherent volatility of the underlying business processes that generate earnings (which is not a

function of managerial reporting discretion). The second component, which we refer to as ‘income smoothing’orsimply‘smoothing’,is the portion of overall

incomesmoothness that is determined by managerial reporting choices (i.e., managerial discretion), which themselves are affected by managerial incentives.

2We use the term ‘private benefits’ as a general term that encompasses severalconcepts, including both value-extractive private benefits and non-value-

extractiveprivate benefits, both of which arise from managers’ ability to control the firm. Aghion and Bolton (1992) refer to these categories as ‘pecuniary’

and‘non-pecuniary’, or ‘monetary’ and ‘non-monetary’ private benefits, respectively.To be more precise, we choose the terms value-extractiveand non-value-

extractive,because, for example, there can be non-monetary private benefits that nonetheless are value extractive (e.g., shirking). Value-extractive private

benefitsarise from actions such as empire building, rent extraction, perk consumption and expropriation (Tirole, 2001), which reduces the value of the firm to

minority shareholders and debt holders. In contrast,non-value-extractive private benefits include social and professional connections, status and opportuni-

tiesthat arise from continually controlling and running the entity (Aghion & Bolton, 1992).

3Wenote that in this context, the terms ‘signaling’ and ‘garbling’ refer to the intentions of the manager who is making the smoothing choice. Extant literature

sometimesrefers to these as ‘informative’ and ‘opportunistic’ motivations for smoothing, respectively (e.g., Dechow, Ge, & Schrand, 2010).

42 AMIRAM ANDOWENS

between smoothing and contract terms. We view the latter possibility as unlikely,because if managers indeed smooth

to facilitate consumption of private benefits, it is unlikelythey will reveal their motivation to lenders in private commu-

nications. On the other hand, if managers smooth to signal, it is indeed plausible that managers may also supply private

information to lenders that reinforce this signal. Importantly, we do not assert that private lenders are the only (or,

even, the primary) audience for signal-based smoothing.4Accordingly,managers would still need to smooth income to

signal their private information to other market participants with whom no private channels of communication exist

(e.g., equityholders, bondholders).5

Our conjecture is that lenders’ view of smoothing is shaped by their assessment of the likelihood that managers

will consume private benefits, which is a function of the extent of enforcement and penalties imposed in a given envi-

ronment if managers are caught consuming private benefits. Hereafter, we refer to environments with weak (strong)

enforcement and weak (strong) punitive consequences for private benefits consumption as posing a ‘high (low) threat’

of private benefits consumption. In a probabilistic sense, we conjecture that in high (low) threat environments, lenders

are more likely to takethe information garbling (signaling) view of smoothing. 6Intuitively,if there are harsh penalties

for consuming private benefits, lenders will assess a lower likelihoodthat managers are attempting to consume private

benefits, ceteris paribus, and their view of smoothing will be shaped accordingly.7This intuition provides the basis for

our empirical design, where our key design choice concerns how to empirically measure the threat of private benefits

consumption in the contracting environment.

Firm-levelgovernance measures provide one such option for measuring the threat of private benefits consumption.

However, because managers choose both the extentof smoothing and firm-level governance mechanisms, there are

standard endogeneity concerns with using firm-level measures. Accordingly,for our primary analyses we use an inter-

national sample, which enables us to exploit plausibly exogenous (from the perspective of a manager) country-level

variation in the threat of private benefits consumption in the contracting environment.

Consistent with our predictions, we provide evidence that smoothing is associated with lower cost of debt in coun-

tries characterized by low threat of private benefits consumption, and associated with higher cost of debt in countries

characterized by high threat of private benefits consumption. Further,we find that the effects of smoothing on cost

of debt are particularly pronounced for firms with relatively high credit risk, consistent with the intuition that lenders

would be particularly concerned about the implications of smoothing for expected credit loss in cases where it is more

likely that borrowers will default.

We note that it is unlikelythat lenders view smoothing as information signaling (garbling) for all firms in a low (high)

threat environment. That is, within either a low or high threat environment, firm-specific threat of private benefits

consumption probably influences the relationship between smoothing and cost of debt. Although pursuing an analysis

based on firm-level governance characteristicsnaturally raises endogeneity concerns, we repeat our tests using mea-

sures of firm-level variation in the threat of private benefits consumption within a given country.Specifically, we use

percentage of closely held shares (e.g., Iliev,Lins, Miller, & Roth, 2015) for our primary non-US sample, and repeat our

analysis using a firm-specific managerial entrenchment index (Bebchuk, Cohen, & Ferrell,2009) with a larger US-only

sample. Consistent with our primary findings, in both cases smoothing is negatively (positively) associated with cost of

debt for firms with relatively low (high) threat of private benefits consumption.

Although our analyses focus on the association between smoothing and cost of debt, private lenders have loan con-

tracting terms other than interest rate at their disposal that may likewise be affected by income smoothing, such as

4Forexample, Dou, Hope, and Thomas (2013) provide evidence that firm-supplier relationships can provide smoothing incentives.

5Indeed, the basic logic of our study holds eveni f the lender can directly obtain allinformation from the borrower other than whether the borrower is con-

sumingprivate benefits or not.

6Consistentwith our conjecture, in informal interviews with bank loan officers in countries where it is easier to consume private benefits, loan officers indeed

indicated that they become ‘nervous’ about a borrower if its financial statements looked too ‘stable.’ In contrast,the loan officers are reassured when they

observesmooth earnings in countries where it is harder to consume private benefits. Having said this, our predictions also follow if lenders infer their view of

smoothingfrom other information sources (i.e., not relying on environmental characteristics), as long as borrower smoothing motivations are correlated with

theprivate benefits consumption threat.

7Weformalize this intuition in Appendix B.