Accounting, Organizations and Society

The profit versus loss heuristic and firm financing decisions

Matt Pinnuck

a,

⇑

, Chander Shekhar

b

a

Department of Accounting, University of Melbourne, Parkville, VIC 3010, Australia

b

Department of Finance, University of Melbourne, Parkville, VIC 3010, Australia

abstract

This paper examines the extent to which the profit versus loss heuristic directly affects

debt issuance decisions. We hypothesize that reporting a loss and its use as a heuristic

rather than firms’ economic fundamentals has an impact both on the decision to raise

external debt finance and on the choice between debt and equity financing. The results

are consistent with the hypothesis. We find that there is a sharp and economically-signif-

icant discontinuity around the zero-earnings threshold in the level of debt issues. Firms

reporting small losses issue significantly less debt than firms reporting small profits. We

also find that the loss heuristic has an impact on the choice between debt and equity in that

loss firms issue less debt relative to equity. Taken together the results are consistent with

the notion that profit versus loss heuristic impacts the debt issuance decision and provide

explanations that add to those offered by the traditional theories.

1. Introduction

There is a considerable body of psychology literature on

the widespread use of heuristics in making decisions (Fred-

erick, 2002; Kahneman, Tversky, & Slovic, 1982; Simon,

1956; Tversky & Kahneman, 1974).

1

This literature has

shown that a widely used heuristic is categorization or

classification, namely the grouping of objects into categories

based on some similarity among them. Categorization sim-

plifies problems of choice and of processing of vast amounts

of information (Hilton & von Hippel, 1996; Macrae & Boden-

hausen, 2000; Reed, 1972; Rosch & Lloyd, 1978).

2

In accounting, one of the oldest and most visible firm

classifications is that of firms as either reporting an account-

ing profit or an accounting loss, giving rise to two broad cat-

egories of firms that could be considered good or bad. In this

paper we argue that the profit versus loss binary classifica-

tion of firms represents a significant, simple and powerful

heuristic which could be used as a reference point to classify

firms (as good or bad) and in turn determine the cost of fi-

nance. This leads to two questions regarding its impact on

debt issuance decisions

3

: What is the impact of the profit ver-

sus loss heuristic on the decision by firms reporting losses to

raise external debt? Given a decision to raise external finance

by firms reporting a loss, how does the profit versus loss heu-

ristic affect the choice between debt and equity financing?

In standard corporate finance theory, projects are exe-

cuted if they have positive net present value. If a company

needs external financing the capital market will provide

the funds. In this setting there is no reason to expect a re-

gime change (or a discontinuity) in the decision to raise

external finance at the zero-earnings threshold. In an

efficient and rational capital market there should be little

or no difference in the external capital raising patterns

http://dx.doi.org/10.1016/j.aos.2013.09.003

⇑

Corresponding author. Tel.: +61 3 83443544.

E-mail address: mpinnuck@unimelb.edu.au (M. Pinnuck).

1

As discussed by Simon (1956) since time and cognitive resources are

limited, we cannot optimally analyse the data the environment provides us.

Instead, natural selection has designed minds that implement rules-of-

thumb ‘‘algorithms,’’ ‘‘heuristics’’ or ‘‘mental modules’’ selectively to a

subset of cues.

2

While the use of categorization in human decision making has long

been recognized in the psychology literature it is only recently that the

economics literature has begun to study the role of categorization in

economic decision making see Barberis and Shleifer (2003) as an example.

3

Existing literature has examined the impact of this earnings heuristic

on the actions of the firm by focussing on how it creates incentives for

earnings management activities by small profit firms (Dechow, Richardson,

& Tuna, 2003; Roychowdhury, 2006).

Accounting, Organizations and Society 38 (2013) 420–439

Contents lists available at ScienceDirect

Accounting, Organizations and Society

journal homepage: www.elsevier.com/locate/aos

between small-loss and small-profit firms because there is

little or no difference in economic fundamentals. The fun-

damental proposition in this paper is that the profit-ver-

sus-loss heuristic is a material consideration in managers’

external debt financing decisions, due to the costs attribut-

able to the heuristic effect (rather than to economic funda-

mentals), associated with debt financing while reporting a

loss. We posit that the costs associated with the profit ver-

sus loss heuristic may result in debt issuance decisions that

are different from those implied by traditional theories.

We assume corporate managers are rational but that

external agents providing finance may use heuristics (either

due to bounded rationality and/or to reduce the transaction

costs of processing information) to aid decision making. Our

main argument is that the profit versus loss heuristic (here-

after the PL heuristic) could affect the decision to raise

external debt finance because potential debt stakeholders

may employ this heuristic to determine the terms of trans-

actions with the firm (Conlisk, 1996; Degeorge, Patel, & Zec-

khauser, 1999). Lenders, for example, could use the binary

classification of firms into profits versus losses as a heuristic

and offer better terms to profit versus loss-reporting firms

(Degeorge et al., 1999, p. 7). The resulting higher cost of

external financing implies that corporate managers may

be reluctant to raise external debt finance whenever their

firms report a negative income, leading to our first hypoth-

esis that the PL heuristic results in firms reporting an

accounting loss issuing lower amounts of debt.

Our second hypothesis is that given that a loss-report-

ing firm’s decision to raise external finance, the PL heuristic

impacts its choice between debt and equity financing.

Holding constant other determinants of capital issuance,

this choice will depend on the relative magnitude of the

impact of PL heuristic on the cost of debt and equity. We

argue, in Section 2, that the PL heuristic will impose greater

discrete costs on debt relative to equity financing for a

numbers of reasons. Given this, loss reporting corporate

managers may prefer to raise finance using equity as op-

posed to debt.

We find that there is a sharp and economically signifi-

cant discontinuity around the zero-earnings threshold in

the issuance of debt. Small-loss firms issue significantly

lower amounts of both debt than small profit firms, and

this regime change cannot be explained by economic fun-

damentals or by standard theories of debt issuance. We

also find evidence supporting the view that the PL heuristic

results in loss firms issuing less debt relative to equity. We

also provide some indirect evidence supporting our

hypothesis – we find that when a firm initiates a credit-rat-

ing the PL heuristic results in lower ratings (after control-

ling for economic fundamentals) for loss-reporting firms

relative to profit-reporting firms. This is consistent with

Kisgen (2006) who reports that firms nearing a credit rat-

ing downgrade issue less debt relative to equity because of

concerns about receiving a further ratings downgrade.

To the extent that firms can decide whether they report

profits or losses by using earnings management tech-

niques, endogeneity and self-selection could be an alterna-

tive explanation for the results. This concern is partially

mitigated in two ways. First, our results are robust to stan-

dard approaches to address endogeneity and self-selection.

We employ the instrumental variable approach and 2SLS

regression which does not affect the results. Further, and

recognizing the well-documented limitations of finding a

valid instrument, we employ additional OLS model specifi-

cations which also leave the primary results unchanged.

Second, we note that self-selection actually provides some

support for our basic hypothesis – if reporting a loss versus

reporting a profit does affect the cost of finance then it

would lead to (at least some) self-selection as firms that

wish to raise finance would manage their earnings and re-

port profits to escape these penalties. Notwithstanding this

observation and our robustness tests, it is still possible that

firms self-select (for reasons unrelated to our hypothesis)

and our tests have been unable to adequately control for

this phenomenon. This should be recognized as a limita-

tion of the study.

This paper makes a number of contributions. First, we

contribute to the financial accounting literature that seeks

to understand how the properties of financial accounting

information impacts decision making and resource alloca-

tion. We show how one of the most significant properties

of accounting – binary classification – can give rise to heu-

ristic decisions which directly affects real economic deci-

sions. As recognized by Dye (2002) financial reporting is,

at its roots essentially a process of binary classification: lia-

bilities are current or non-current; leases are capital leases

or operating leases; expenditure is an asset or an expense,

etc. Notwithstanding the importance of this binary classifi-

cation property to financial accounting, there is very little

research that directly examines its impact on decision mak-

ing and in turn on real resource allocation. One of the oldest

binary classifications in accounting, is that of firms as either

reporting an accounting profit or an accounting loss. We

show that this binary classification per se appears to have

an effect on economic decision making.

More generally, we also contribute to the behavioural

financial accounting literature which has focused on how

behavioural biases affect security pricing of accounting

information.

4

We show that behavioural biases in the pro-

cessing of external financial accounting information can also

affect firms’ internal real economic and financing decisions.

The study is therefore related to the behavioral management

accounting research that has explored the relative influence

of positive and negative information and has shown that in

evaluating balanced scorecards evaluators will exhibit a

negativity bias such that negative performance information

is weighted more heavily than positive performance infor-

mation (Kaplan, Petersen, & Samuels, 2012). Our work is also

related to some management research that has explored the

relative influence of positive and negative information and

finds, for example that in employee performance evalua-

tions negativity bias has a significant effect on performance

evaluations (Brannick & Brannick, 1989; Ganzach, 1995).

4

There is a very large volume of financial accounting literature which

has focused on how less than perfect rationality can give rise to security

mispricing in the reaction to financial information. Prominent examples are

the research that has examined existence and explanations for post-

earnings announcement drift (Bernard & Thomas, 1989) and the accrual

anomaly (Sloan, 1996). There is very little research however that has

examined how less than perfect rationality in the processing of external

financial reports can directly affect firms’ internal decisions.

M. Pinnuck, C. Shekhar / Accounting, Organizations and Society 38 (2013) 420–439 421

Our work also contributes to both the empirical corpo-

rate finance literature that examines the security-issuance

decision, and to the emerging literature on behavioral cor-

porate finance. Baker, Ruback, and Wurgler’s (2008) survey

of this literature suggests that reference points could play a

role in the corporate finance decisions but they find the

area is relatively unexplored. To date the impact of the

PL heuristic on the security issuance decisions has not been

investigated in the empirical literature. Notwithstanding

the large volume of research there is little consensus as

to the theories that explain the security issue decision.

5

The main theories of security issuance are the trade-off

model and pecking order (Myers, 1984; Myers & Majluf,

1984). Our study and findings do not rule out these theories

as explanations for debt issuance. Rather we suggest that the

profit versus loss heuristic has incremental explanatory

power which is independent of existing theories.

The remainder of this paper is set out as follows. Sec-

tion 2 develops the hypotheses. Section 3 describes the

data. Section 4 presents our empirical tests and results.

Section 5 concludes.

2. Existing literature and hypotheses

In this section we first set out existing theories of debt

issuance and the associated predictions regarding the rela-

tion between the level of earnings and external debt finance.

We then develop the hypotheses being tested in this paper.

2.1. Existing theories of capital issuance and their relation to

earnings

Existing theories and explanations for debt issuance

have focused primarily on the presence of agency costs,

taxes and bankruptcy costs (the trade-off model), informa-

tion asymmetry (pecking order theory) and mispricing

(market timing behavior). In the trade-off model, agency

costs, taxes and bankruptcy costs push firms to increase

debt as earnings increase (Fama & French, 2002, pp. 8–

9).

6

A pure trade-off theory predicts a positive relationship

between profitability and external debt financing. Trade-

off theory’s prediction follows from both the ability and

the willingness of firms to issue debt. More profitable firms

are able to service higher levels of debt and have a greater

incentive to exploit tax shields associated with debt.

Pecking order theory states that when there is informa-

tion asymmetry between managers and outsiders, firms

prefer the following order when using external sources to

finance new projects: internal funds, debt and then equity.

According to Myers (1984) due to adverse selection, firms

prefer internal to external finance. When outside funds

are necessary firms prefer debt to equity because of lower

information costs associated with debt issues. The underly-

ing logic is that the valuation of debt is less information

sensitive than equity and therefore subject to lower ad-

verse-selection costs. Pure pecking-order theory predicts

a negative relationship between the level of earnings and

external debt financing because as the level of earnings

and thus internal funds increases, the need for external

debt financing decreases. Several papers point to a nega-

tive relation between leverage and profitability as evidence

supporting the pecking order’s prediction of managers pre-

ferring to use internal funds before turning to debt (e.g.

Fama & French, 2002; Titman & Wessels, 1988).

We remain agnostic as to which theory (or combination

thereof) explains debt issuance.

7

The common assumption

across both theories is that economic agents involved in cor-

porate finance, both managers and investors, are rational. The

common implication across both theories is that whatever

the functional form is of the relationship between earnings

and external debt financing, it is the same across all levels

of earnings.

8

Our study replaces the traditional rationality

assumptions with a specific set of behavioral assumptions –

the use of heuristics by lenders – which leads us to hypothe-

size the relationship, between earnings and external debt

financing has a discontinuity at the zero-earnings threshold.

2.2. Hypotheses

In developing our hypothesis we assume corporate man-

agers are rational and they are responding to the use of the

PL heuristic by the external agents providing the finance.

9

The underlying argument in this paper is that the use of the

PL heuristic categorizes firms into either a profit or a loss

group and there is a long historic convention of profits being

considered good and losses being considered bad. This, in

5

Empirical research on the pecking order shows mixed results with

some studies confirming the theory (see, e.g., Bayless & Chaplinsky, 1991;

Hovakimian, Opler, & Titman, 2001; Shyam-Sunder & Myers, 1999) and

other papers rejecting it (see, e.g., Fama & French, 2005; Frank & Goyal,

2003; Hewge & Liang, 1996; Leary & Roberts, 2005). A growing literature

shows that firms time the market with equity issues (e.g., Baker & Wurgler,

2002; Gomes & Phillips, 2005; Henderson, Jegadeesh, & Weisbach, 2006).

More recently, Dittmar and Thakor (2007) postulate that firms issue equity

when agreement between management and investors is high regardless of

firm valuation.

6

Note that the term trade-off is used in different ways by different

authors. For some authors it means that bankruptcy and taxes are being

balanced (Kraus & Litzenberger, 1973). For other authors it includes

agency-based arguments (Fama & French, 2002).

7

In a recent study Leary and Roberts (2010) note that models incorpo-

rating a broad range of determinants from previous capital structure

studies perform significantly better at explaining issuance decisions.

8

As a typical example Fama and French (2002) predict a positive

relationship between earnings and leverage due to the trade-off model and

negative relationship between earnings and leverage due to the pecking-

order model (see Table A.2, p. 32). These predictions are tested by Fama and

French (2002) using an OLS regression across all sampled firms regardless

of their levels of earnings. This effectively implies that the functional form

of negative relation between external financing and earnings is constant

across all levels of earnings.

9

As recognized by Baker et al. (2008) research in behavioral corporate

finance employs two distinct approaches. The first emphasizes that

investors are less than fully rational. It views managerial financing and

investment decisions as rational responses to securities market mispricing.

The second approach emphasizes that managers are less than fully rational.

In this paper we are implicitly adopting the first approach as an assumption

underpinning our hypotheses development. We however acknowledge

that, CEOs themselves may be irrational and could be directly affected by

the PL Heuristic in the decision to raise finance. The finance literate has

documented that CEOs are subject to some behaviorial biases, such as over-

confidence (Malmendier & Tate, 2005). It is therefore possible that CEOs

may also be affected by the psychological forces underpinning the PL

heuristic.

422 M. Pinnuck, C. Shekhar / Accounting, Organizations and Society 38 (2013) 420–439

turn, leads to the fundamental hypothesis that the PL heuris-

tic imposes additional costs on loss-reporting firms that wish

to raise external capital, and this represents a material consid-

eration that affects managers’ decision to raise debt finance.

Further support for a differential effect of profits (good) versus

losses (bad) is provided by the psychology research that has

examined the relative impact of negative and positive out-

comes (when they are of similar type and magnitude) on indi-

viduals’ judgments and decisions (Baumeister, Bratslavsky,

Finkenauer, & Vohs, 2001; Rozin & Royzman, 2001; Skowron-

ski & Carlston, 1989). In summarizing research findings on

this issue, Baumeister et al. (2001, p. 323) state, ‘‘When equal

measures of good and bad are present, however, the psycho-

logical effects of bad ones outweigh those of the good ones.’’

Existing theories of firm financing suggest that the past

profitability of a firm and hence the amount of retained

earnings available should be an important determinant of

external debt financing. Capital budgeting holds that firms

issue external finance (either equity and/or debt) in order

to invest the proceeds in positive net present value pro-

jects (Brealey & Myers, 2002). A priori, there is no reason

to expect that firms that differ only with respect to which

side of zero earnings threshold they fall on, should differ in

their opportunity set of available net present value pro-

jects. However, we conjecture that the profit versus loss

heuristic is a material consideration in managers’ external

debt financing decisions due to the additional direct and

indirect costs associated with external debt finance for

loss-reporting firms.

There are three arguments for why the PL heuristic

could impose additional discrete costs on debt financing

for loss-reporting firms. First, the transaction-costs argu-

ment of Burgstahler and Dichev (1997a, p. 122) suggests

that stakeholders could use the PL heuristic to determine

the terms of transactions with the firm because it is too

costly to retrieve and process detailed information about

earnings for all firms. Lenders, for instance, could use the

binary classification of firms into profits versus losses as

a heuristic and offer better terms to profit compared to

loss-reporting firms. DeGeorge et al. (1999) suggest that

banks may grant loans only to firms that report positive

earnings – that is banks use a threshold of zero earnings

as an initial screen – as judiciously adjusting interest rates

in response to differential performance may be ‘‘too hard’’

(p. 7). Support for this argument is provided by the results

from Jiang (2008) who finds evidence that firms that miss

the benchmark of reporting a profit incur a greater cost of

debt as proxied by bond yield spread.

The second argument is related to the effect of debt

contract covenants. Both Begley and Freedman (2004)

and Beatty, Weber, and Yu (2008) find evidence (from a re-

view of public debt contracts) that profits and losses are

treated asymmetrically in debt contracts. Begley and

Freedman (2004) find that in public debt covenants the

dividend covenant penalizes losses more heavily than it re-

wards profits. Specifically, dividend pools typically include

only 50% of net profit but deduct 100% of net losses. Beatty

et al. (2008) note that for firms with net worth covenants,

reporting a profit only partially increases the covenant

slack, whereas reporting a loss reduces the slack by 100%.

As contracts are rigid, the existence of these covenants im-

poses asymmetric costs on debt issuance when firms re-

port a loss. This could lead to an aversion by loss-

reporting firms to issuing debt. Finally, issuing debt while

reporting a loss could result in a ratings downgrade, which

imposes further second-order type costs on the firm.

Kisgen (2006) examines the extent to which credit ratings

directly affect capital-structure decisions and finds that

firms near a credit rating downgrade issue less debt relative

to equity. This behavior is attributed to the economically

significant discrete costs of rating changes.

10

Therefore

loss-reporting firms may be reluctant to issue debt as

issuance may attract the attention of rating agencies and

the reported loss act as a heuristic for a ratings downgrade.

Therefore if external stakeholders use the PL heuristic

to assign a higher cost to debt for a loss reporting firm then

managers may be reluctant to issue debt.

11

Collectively,

these arguments lead to the primary hypothesis that the

PL heuristic is a material consideration in managers’ deci-

sions to raise external debt finance. The primary testable

implication of the PL heuristic hypothesis is that there is a

discontinuity or kink in relationship between the level of re-

ported earnings and external financing at the zero earnings

threshold. Stated as a hypothesis:

H1. At the zero-earnings threshold there is a discontinuity

in the issuance of debt.

The equity market may also assign a higher cost of equi-

ty finance to firms reporting an accounting loss due to the

effects of the PL heuristic. For example stocks of loss-

reporting firms could be mispriced due to firms being

10

Kisgen (2006, p. 1036) argues the discrete costs of rating changes

include ‘‘for instance, several regulations on bond investment are based

directly on credit ratings: credit ratings levels affect whether particular

investor groups such as banks or pensions funds are allowed to invest in a

firms’ bonds and to what extent investor groups such as insurance

companies or brokers–dealers incur specific capital requirements for

investing in a firm’s bonds. Ratings can also provide information to

investors and thereby act as a signal of firm quality. If the market regards

ratings as informative, firms will be pooled together by rating and thus a

rating change would result in a discrete change in a firm cost of capital.

Ratings changes can also trigger event that result in a discrete costs for the

firm, such as a change in bond coupon rate, a loss of contract, a required

repurchase of bonds, or a loss of access to the commercial paper market.’’

11

This avenue – through which the PL heuristic could affect capital

structure – may be considered as a specific setting whereby the market

timing theory exerts influence on capital structure decisions. Given the well

documented empirical shortcomings of the trade-off and pecking order

models (see e.g. Fama & French, 2005; Frank & Goyal, 2003; Welch, 2004),

market timing is currently the most strongly supported of the three major

theories of financial policy (Deangelo, Deangelo, and Stulz (2007). The

market timing vein of capital structure literature maintains that firms issue

equity in order to take advantage of mis-pricing by new investors (Baker

et al., 2008; Stein, 1996). Managers admit to market timing in anonymous

surveys. Graham and Harvey (2001) find that two-thirds of CFOs agree that

‘‘the amount by which our stock is undervalued or overvalued was an

important or very important consideration’’ in issuing equity. Therefore, if

the PL heuristic results in mispricing of firms reporting losses and if these

firms believe that they are undervalued then they are less likely to issue

securities of any type (i.e. both debt and equity). Firm reporting losses with

temporarily underpriced stocks therefore have an incentive to postpone an

offering until the stock price has recovered. Survey evidence lends some

plausibility to timing in the debt market as well. In particular, Graham and

Harvey (2001) find that interest rates are the most cited factor in debt

policy decisions: CFOs issue debt when they feel ‘‘rates are particularly

low.’’

M. Pinnuck, C. Shekhar / Accounting, Organizations and Society 38 (2013) 420–439 423

penalized for not reporting a profit.

12

We some present

some preliminary evidence consistent with the PL heuristic

having an affect on the decision to raise external equity.

Therefore, given the decision to raise external finance,

how does the PL heuristic affect the choice between debt

and equity financing? Controlling for empirically estab-

lished determinants of firm financing decisions the impact

of the PL heuristic will depend on the relative magnitudes

of the discrete costs imposed by the heuristic on issuing

either debt or equity. We argue the PL heuristic will im-

pose greater discrete costs on debt relative to equity

financing for three reasons.

13

First, and as argued earlier, issuing debt in the presence

of a loss is likely to result in a ratings downgrade with all

the consequential indirect costs. Issuing equity does not

give rise to the costs associated with ratings changes. Kis-

gen (2006) finds that firms near a credit rating downgrade

issue less debt relative to equity. This suggests the PL heuris-

tic will result in less debt being issued relative to equity.

Second, empirical evidence suggests that the relative

importance of the three earnings benchmarks – reporting

a profit, reporting an earnings increase, and beating the

consensus analysts forecast – differs between the debt

and equity markets. Jiang (2008) finds that reporting a

profit is the most important of the three benchmarks for

the debt market. In contrast, Brown and Caylor (2005) find

that reporting a profit is the least important benchmark in

the equity market and Payne and Thomas (2009) find that

missing the zero earnings threshold has no negative conse-

quences. Third, debt issuance is subject to public debt cov-

enants which treat profits and losses asymmetrically.

Equity issuance is not subject to such contracts. While ulti-

mately an empirical question, these arguments suggest

that the PL heuristic will have a greater effect on debt rel-

3. Research design and results

3.1. Sample selection and descriptive statistics

To construct our sample we begin with the set of firms

in the CRSP-Compustat merged database over the period

1976 through to 2006.

14

We begin in 1976 as it is the ear-

liest year for which we have IBES data which we can use

to determine whether or not a firm has analysts’ coverage.

15

The final sample includes all available firm-years except for

those for which the data for all the variables used in the

study was unavailable. Our primary hypothesis is that the

PL heuristic will have an impact on debt issuance. It is also

possible that the PL heuristic can have an impact on equity

issuance. As we compare the relative impact of the PL heu-

ristic on debt versus equity issuance we need to measure

both debt and equity issuance.

Following prior research we use Compustat Cash flow

statement data to measure debt and equity issues (see

Frank & Goyal, 2003; Kisgen, 2006). To determine whether

a firm issues debt, consistent with Frank and Goyal (2003)

and Kisgen (2006) we use Compustat variable Debt (Long-

Term) Issuance (Data Item #111) which represents the

amount of funds generated from the issuance of long-term

debt.

16

To determine whether a firm issues equity we use

Compustat variable Sale of Common and Preferred Stock

(Data Item #108) which represents funds received from

issuance of common and preferred stock. We treat our secu-

rity issuance decision as binary. This is both because our

hypothesis is based on a binary argument and, as recognized

by Kisgen (2006) and others, equity offerings have high

transaction costs and significant economies of scale and

may be considered binary decisions. A similar argument

424 M. Pinnuck, C. Shekhar / Accounting, Organizations and Society 38 (2013) 420–439