Accounting and Business Research

ISSN: 0001-4788 (Print) 2159-4260 (Online) Journal homepage: http://www.tandfonline.com/loi/rabr20

Signalling, Agency Theory and Accounting Policy

Choice

Richard D. Morris

To cite this article: Richard D. Morris (1987) Signalling, Agency Theory and
Accounting Policy Choice, Accounting and Business Research, 18:69, 47-56, DOI:
10.1080/00014788.1987.9729347

To link to this article: http://dx.doi.org/10.1080/00014788.1987.9729347

Published online: 28 Feb 2012.

Submit your article to this journal

Article views: 389

View related articles

Citing articles: 6 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=rabr20

Download by: [University of Toronto Libraries] Date: 02 December 2016, At: 08:56
Accounting and Business Research, vd. 18, No. 69, PP. 47-56, 1987 47

Signalling, Agency Theory and

Accounting Policy Choice*
Richard D. Morris
Abstract-Signalling and agency theories appear in the accounting literature to be competing theories. This article
demonstrates that they are actually consistent theories, in that one set of sufficient conditions of signalling theory
is at least consistent with one set of sufficient conditions of agency theory. Indeed, a considerable overlap exists
between the two theories: rational behaviour is common to both; information asymmetry in signalling theory is
implied by positive monitoring costs in agency theory; ‘quality’ in signalling theory can be defined in terms of agency
theory variables; and signalling costs are implicit in some bonding devices of agency theory. Examples are given
where both theories’ predictions about lobbying, accounting choices, and voluntary auditor selection are added
together.

Introduction tion is to consider the theories as competing sys-

In recent years, two seemingly competing theories tems.
have appeared in the accounting literature. Agency This article examines the logical relationship
theory,’ which describes the incentive problems in between signalling and agency theories. Far from
a firm caused by the separation of ownership and finding them competing, the conclusion of the
control of resources (the principal-agent problem) article will be that the two theories are consistent:
has been used to explain theoretically accounting that is, if one theory is ‘correct’ the other theory
choices, voluntary disclosure, voluntary ap- may also be ‘correct’. This opens up the possibility
pointment of auditors, and corporate lobbying of joining the two theories to provide fresh insights
about proposed accounting standards.* Signalling into the principal-agent problem, and into firms’
theory, originally developed to explain problems of accounting policy choices.
information asymmetry in labour market^,^ has Agency theory is described briefly in the first
been applied to questions of corporate dividend section, while an outline of signalling theory ap-
policy,4 capital structure decisions,’ voluntary dis- pears in the second section. The third section
closure,6managerial retention of ownership in new covers the steps involved in examining the re-
share issues,’ current value accounting,8and volun- lationship between two theories, and these steps
tary selection of auditors.’ The literature of agency are applied to agency and signalling theories in the
theory has tended to pay little attention to signal- fourth section. Some accounting applications of
ling, and signalling papers ignore agency theory. the finding that these two theories are consistent
Yet both theories are used to address similar appear in the fifth section. Conclusions are given
accounting issues. Consequently, an initial tempta- in the last section.

*The comments of Ray Ball and Peter Birrell are gratefully

Agency theory
acknowledged. Agency theory is concerned with the principal-
’Jensen & Meckling (1976). agent problem in the separation of ownership and
*Reviewed in Holthausen and Leftwich (1983), and in Kelly control of a firm (Jensen and Meckling, 1976),
(1983). Another branch of agency theory reviewed in Demski
and Kreps (1982) considers the form of optimal contracts
between different suppliers of capital (Smith and
between principal and agent. This literature is not considered Warner, 1979), and in the separation of risk bear-
here. ing, decision making and control functions in firms
’Spence (1973, 1974). Other non-accounting applications are (Fama and Jensen, 1983). If individuals act self-
in insurance (Rothschild and Stiglitz, 1976) and advertising
(Kihlstrom and Riordan, 1984).
interestedly, these separations produce conflicts.
4Bhattacharya (1979, 1980), Eades (1983), Kalay (1980), Agency theory demonstrates that the precip-
Penman (1982a), Miller and Rock (1985), John and Williams itators of these conflicts incur agency costs, which
(1985). they then have an incentive to reduce. The agency
SRoss (1977). costs of equity are, firstly, the decline in a firm’s
6Ross (1979).
’Leland and Pyle (1977), Downes and Heinkel (1982). value when shareholders perceive managers as not
*Forker (1984). pursuing the shareholders’ interest, especially
’Bar-Yosef and Livnat (1984). where managers act inefficiently or do not choose
48 A C C O U N T I N G A N D B U S I N E S S RES EA RCH

projects as profitable as shareholders would like, occur. Initially, sellers in a market are assumed to
and secondly, the costs of monitoring and bonding possess more information about their product than
managers so that they do pursue the shareholders' buyers. If buyers have no information about
interest. An optimum trade off exists between these specific products but do have some general percep-
two sets of agency costs. The first agency costs are tions (e.g. that p % of products offered will be
a manager's opportunity loss, if not reduced by faulty and that bad products should sell at $x and
monitoring and bonding, because his self- good products at $y > $x), then buyers will value
interested actions precipitate the costs and share- all products at the same price which is a weighted
holders build them into the firm's share price. average of their general perceptions.
On the other hand, the agency costs of debt are Sellers of above average quality products incur
borne by the equity holders. These costs include (a) an opportunity loss because their products could
the problems of excessive dividend payments, the sell at a higher price if buyers knew about the
issue of senior ranking debt, asset substitution and superior quality, while sellers of below average
underinvestment (Smith and Warner, 1979), to- products make an opportunity gain. Sellers of high
gether with bankruptcy and reorganisation costs, quality products have an incentive to leave the
and (b) the costs of monitoring and bonding. market-the phenomenon of adverse selection
Rational debtholders incorporate the likelihood of (Akerlof, 1970)-unless they can communicate
these agency costs in the price they pay for the their product's superior quality to buyers and thus
debt. Again, an optimum trade off exists between increase its price. This communication is done by
monitoring and bonding costs and other agency signalling: the publication of a device (e.g., a
costs of debt. The agency costs in (a) are oppor- product warranty) which acts as a prediction of
tunity costs to the manager and equity holders if superior quality. To be effective, the signal must
no action is taken to reduce them by monitoring not be easily copied by poor quality sellers. To
and bonding. ensure this, the assumption usually made is that
Monitoring and bonding devices include the signalling costs are inversely related to quality.
production of accounting reports, writing re- Also, the signal must be confirmable with actual
strictive covenants in debt contracts, and manage- product quality observed after purchase.
ment bonus plans geared to reported profits. As better quality sellers signal, buyers consider
There are two version of agency theory. In all the remaining sellers to be of poor quality. Their
Jensen and Meckling (1976), an optimum amount average price will be reassessed downwards accord-
of monitoring and bonding is reached beyond ingly. The best remaining sellers then try to screen
which the reduction of each dollar of perquisite themselves from the others-an iterative process
consumption costs more than the benefit gained. which continues as long as the increase in price
Consequently, residual agency costs remain. A obtained exceeds the signalling costs.
similar outimum occurs with the agency costs of
w .

debt. Extensions of the theory by Fama and Jensen

(1 983) and Jensen (1983) leave this fundamental The between two theories
result unchanged. The relationship between theories" can be assessed
In a second version of agency theory (Fama, in two alternative ways-by examining their un-
1980), agency costs are driven to zero by market derlying axioms and concepts, or by comparing
forces within and external to the firm. A form of their predictions.
full ex post settling up occurs which penalises
managers who deviate from maximising share- Axiomatic
holders' wealth. Thus, they are motivated to act in
the interests of shareholders. In essence, Fama's If two theories deal with the same subject at the
version of agency theory is a long run version of same level of reduction,ll under the axiomatic
Jensen and Meckling's. For this reason, Jensen and
version Of the be 'Once'- "A theory is a model or system consisting of (a) non-
contradictory, primitive assertions or axioms; (b) definitions of
trated on here. basic concepts; and (c) the conclusions or predictions which are
deduced from them. A theory's axioms and definitions are its
sufficient conditions (of which there may be several sets)-in
Signalling theory their presence the theory's conclusions follow. Necessary condi-
tions, if any, will be part of each set of sufficient conditions.
signalling theory addresses problems of informa-
Necessary conditions are axioms which, if negated, mean that
tion asymmetry in markets. The theory shows how the theory's conclusions can never hold,
this asymmetry can be reduced by the party with "The level of reduction refers to the level at which a problem
more information signalling it to others. Although is addressed by different disciplines. For example, alcohol
the theory was developed in the labour market, addiction is examined as a physiological disorder by physicians,
as a behavioural disorder by psychologists, and as a social
signalling is a general phenomenon applicable in problem by sociologists. There also can be different levels of
any market with information asymmetry. reduction within one discipline, such as the classical distinction
In most signalling models, the following steps between micro and macro economics.
 WINTER 1987 49

approach their relationship is fourfold. Firstly, the depicts, in the upper half, a series of now
theories may be equivalent: the same theory under contradictory axioms a, b, . . . , k ; and in the lower
different guises. Secondly, one theory may imply half a set of axioms Z,6,. . . ,E which contradict
the other: that is, one is a subset of the other. a, b, . . . ,k (ii contradicts a, 6 contradicts b, and so
Thirdly, the two theories may be consistent. This on). Theories I-V are shown as shapes embracing
means that if one is true the other is possibly true. subsets of these axioms. Axioms a, b , . . . , e are
Fourthly, they may be contradictory or competing sufficient conditions for theory I, axioms
explanations of the subject: if one is true, the other e , f ,g , . . . ,k, are sufficient conditions for theory I1
is false. and so on. Necessary conditions appear as shaded
Two steps are needed to investigate which of axioms. Theory I1 implies theories I11 and IV,I2
these four relationships applies to two theories A while theories I and I1 are consistent, as are
and B. Firstly, whether or not either theory’s theories I, I11 and IV. On the other hand, theory
axioms are necessary conditions must be identified. V contradicts each of the other theories.
To establish the equivalence of theories A and B,
both theories’ necessary conditions must be identi- Predict ion
cal. Similarly, for theory A to imply theory B, the The alternative approach to assessing the re-
necessary conditions of B must be the same as lationship between two theories (dealing with the
those of A, or at least a subset of them. If these same subject at the same level of reduction) is to
stipulations are not met, it is logically possible for compare their predictions or consequences. If these
theory A to hold in the absence of theory B, a conflict, the theories compete; if they are consis-
denial of equivalence and implication. On the other tent, so are the theories. (The four way
hand, consistency of theories A and B only requifes classification of the axiomatic approach is replaced
that their necessary conditions do not conflict. By by two categories: competing and consistent.)
contrast, if the two theories are competing, their As a means of assessing the relationship between
necessary conditions must conflict in at least one signalling and agency theories, prediction com-
particular. parison is rejected for two reasons. Firstly, com-
Secondly, each theory’s sets of sufficient condi- parison of predictions is not reliable if the theories’
tions must be examined. For equivalence, at least axioms conflict, because it is possible for the
one set of sufficient conditions for A must be theories’ predictions to conflict in one application
identical to at least one set of sufficient conditions and to be consistent in another. For example, the
of B. For theory A to imply theory B, at least one Ptolemaic and Newtonian systems of astronomy-
set of sufficient conditions of B must be entailed in competing theories axiomatically-yield the same
part-but not all-of one set of sufficient condi- predictions about planetary motion (and thus in
tions of A. Consistency of the two theories only this application are consistent theories) but give
requires that the sets of sufficient conditions of A conflicting predictions about other astronomical
do not conflict with those of B. However, if A and phenomena.
B are competing theories, at least one set of Secondly, in examining the predictions of two
sufficient conditions for each will be contradictory. theories, the assumption is that the theories’ con-
These two steps are illustrated in Figure 1 which structs are fully operationalised. At present, this is
not so with signalling and agency theories. For
example, in agency theory, leverage is used as a
proxy for closeness to accounting-based borrowing
limits, and bonus plans are usually represented in
empirical studies by a dichotomous variable. In
signalling theory, quality could be measured using
next year’s change in reported earnings or share
market abnormal performance. However, all of
these proxies are imperfe~t:’~ they measure the
horns I2In some investigations of the relationship between two
theories, the approach adopted is to demonstrate that the two
theories are special cases of a general or meta theory. The
general theory implies the special case theories. In Figure 1, this
is exemplified by theories 11, 111 and IV. Yetton and Crouch
(1983) provide an example of this approach in the leadership
literature.
”Ball and Foster (1982) outline the weaknesses of proxy
measures in agency theory. Abnormal share price performance
and net profit changes next year are imperfect proxies of quality
because, although they capture quality changes expected now
Fig. 1 by managers but known to investors later, they also capture the
~~
effects of unforeseen events arising in the ensuing year.
A.B.R. 18169-D
50 A C C O U N T I N G A N D BUSINESS RESEARCH

underlying theoretical construct with error. A make production/investment decisions in the

difficulty created is that predictions using imperfect first period which affect the quality of firms
proxies may compete (or be consistent) due to this in the second period.
measurement error, leaving unclear whether the (c) The quality of firms competing for equity and
underlying theories compete or not. debt funds in the capital market varies.
The remainder of the paper concentrates on (d) The actual quality of firms is objectively
assessing the relationship between agency and sig- observable, ex post, in period 2.
nalling theories using the axiomatic approach. (e) Ex ante, in period 1, information asymmetry
exists between the manager of each firm and
The relationship between agency theory and capital suppliers. The manager’s information
signalling theory about the firm is superior to that of capital
suppliers.
In the first section an outline of agency theory was (f) Signalling costs are inversely related to
given. The following assumptions constitute a set quality .
of sufficient conditions from which the features of
that theory can be derived.I4 The relationship between each theory’s assump-
tions will now be discussed.
All market participants are rational wealth
maximisers. Necessary Conditions
All firms operate in two periods.I5 Managers
make production/investment decisions in the Individual rational wealth maximisation (as-
first period which affect firms’ expected val- sumption 1) and separation of resource ownership
ues and variances in the second period. and control (assumption 4) are necessary condi-
Firms have external equity and debt tions of agency theory. Together, they imply
financing. conflicts of interest between managers of firms and
There is separation of equity and debt capital capital suppliers. Without them, the principal-
suppliers and managerial control in the firm. agent problem would not exist. Similarly, informa-
Each manager owns a fraction16of his firm’s tion asymmetry (assumption e) is a necessary
outstanding equity. condition of signalling theory, because without it
Each manager is remunerated by salary, per- the need for signalling does not exist.
quisite consumption, and returns on his Prima facie, these necessary conditions of each
equity in the firm. theory do not conflict. Positive monitoring costs
Monitoring and bonding are available at a (assumption 7) and separation of ownership and
cost proportional to the value of the firm; control (assumption 4) imply information asym-
and they reduce activities dysfunctional to metry between managers, investors and creditors.
capital suppliers at a reducing rate. Thus, information asymmetry is implicitly con-
tained in the sufficient conditions of agency theory
In the second section an outline of signalling given above. However, positive monitoring costs
theory was given. From it, the following set of are not a necessary condition of agency theory,
sufficient conditions for signalling in a capital because the principal-agent problem can still exist
market setting may be drawn. in the (unlikely) case of zero monitoring costs,
provided that bonding costs are positive. The latter
(a) All market participants are rational wealth ensures that residual conflicts remain even if all
maximisers. parties are fully informed.
(b) All firms operate in two periods. Managers Therefore, agency and signalling theories do not
share common necessary assumptions. Hence, the
I4For brevity, definitions of basic terms such as monitoring theories are not equivalent, nor does one imply the
and bonding are omitted. other.
”These may be thought of as the present for period one, and
the future for period two. Suficien t Conditions
I6In Jensen and Meckling (1976) this fraction commences at
100 percent and is reduced as the manager sells part of his The first four assumptions of both theories,
equity to outsiders. given above, are compatible except that agency
17Somemight object that this assumption is at variance with theory does not mention the concept of ‘quality’.
other applications of signalling theory in capital markets (e.g., However, because signalling is a general phenom-
Ross, 1977; Bhattacharya, 1980 Kalay, 1980; Talmor, 1981)
which assume that managers act in the interests of investors. enon applicable in any market with information
However, that assumption is not crucial, and is usually made asymmetry, the meaning of ‘quality’ can be taken
for convenience. The original exposition of signalling theory by from the market context being considered. There-
Spence (1973, 1974) had market participants acting in their own fore, quality may be defined in terms of a firm’s
interests, as did later applications of the theory by Rothschild
& Stiglitz (1976), Leland and Pyle (1977), Bhattacharya (1979), expected value, its risk, and the level of the man-
Hauaen & Senbet (1979), Downs & Heinkel (1982), Heinkel ager’s comr>ensation: three variables directly ad-
(198?), and Eades (1983). dressed b; assumptions (2) and (6) of agency
W I N T E R 1987 51

theory.’* The manipulation of these variables can the manager and the shareholder (Jensen and
produce agency costs. Thus, an above average Meckling, 1976). Many formal management com-
quality firm will have a higher expected value than pensation plans specify some threshold profit level
firms of comparable risk and levels of management below which no bonus is paid to managers (Healy,
compensation. 1985). This threshold will be negotiated by each
Information asymmetry (assumption e) of sig- manager and his shareholders. At the point of
nalling theory is implied by positive monitoring negotiation, the threshold acts as a prediction of
costs (assumption 7) and the separation of own- attainable future earnings since the manager acting
ership and control of capital (assumption 4) in in his own interest will want as much bonus as
agency theory, as stated previo~sly.’~ This infor- possible. Better quality managers may negotiate a
mation asymmetry leads to costs in the form of higher threshold to discriminate themselves from
opportunities foregone by the manager of an above poorer quality managers. If a manager believes his
average firm for raising equity or debt capital. Less firm to have a higher than average expected value,
will be paid for the firm’s equity or debt securities he will negotiate a compensation package, which,
than would be the case with no information asym- for example, contractually sets the maximum bo-
metry.” As losses from foregone opportunities, nus to be paid if the expected value is attained.
these costs are identical to agency costs of equity Also, to reduce the resulting horizon problem in
or debt. In short, the costs of information asym- which projects with high short term payoffs but not
metry are a subset of the construct agency costs. As necessarily high net present values are chosen,
with other agency costs, these costs of information above average firms will gear the bonus to some
asymmetry are borne by the manager of an above measure which takes account of all expected cash
average firm. Graphical proofs of this appear in flows, e.g. share price or average earnings. So,
the Appendix. The manager then has an incentive bonus schemes can act as a signal.
to signal his firm’s above average quality to reduce (2) Contractual debt covenants restricting the
this opportunity loss. amount of debt a firm can raise limit wealth
However, if the firm is below average, the agency transfers between equityholders and debtholders.
costs of information asymmetry are borne by Managers have an incentive to offer these protec-
investors since, in the absence of signalling, the tive debt covenants to increase the price at which
market regards the firm as average quality. This debt is sold, and indirectly to act as a signal about
conclusion alters a central result of agency theory: expected future earnings and expected levels of
that agency costs are always born by the agent. management compensation. Ross (1977) shows
With information asymmetry, some agency costs that where managerial Compensation contracts are
of below average firms are borne by investors.” in the form of a contingent contract and where
For signalling and agency theories to be consis- opportunities for wealth transfers to the equity-
tent, signalling costs (assumption f) must be borne holders are absent (or perfectly controlled), the
by the agent so that he has an incentive to signal contractual level of debt issued signals the expected
truthfully. Also, these signals should not be incon- value of the firm. Firms with higher contractual
sistent with monitoring and bonding devices (as- debt equity ratios have above average expected
sumption 7) in agency theory. That these condi- values. Maximum contractual debt equity ratios
tions hold can be illustrated by reference to three can also signal limits to a manager’s inefficiency
common bonding devices in agency theory. (perquisite consumption) because his inefficiency
(1) Management compensation plans are bond- will adversely affect shareholder’s equity on the
ing devices because they align the interests of balance sheet (e.g., by reducing reported earnings)
and raise the debt equity ratio towards its limit. So,
contractual debt equity ratios can act as signals as
well as bonding devices.
I8Quality may also be defined in terms of other variables such (3) Dividend constraints are bonding devices
as turnover, plant efficiency etc. However, it is sufficient for the
purpose of examining if agency and signalling theories are because they restrict the manager’s ability to trans-
consistent to show that ‘quality’ can be interpreted in terms of fer wealth to the equityholders. A manager’s incen-
agency theory variables. tive thus to pay dividends is positively related to
19Jensen & Meckling (1976) ignore information asymmetry, his firm’s leverage ratio (Kalay, 1978). Since the
claiming that it will not affect their theory in large equity
markets and where estimates are rational and errors are inde- manager (and the equityholders) bear the resultant
pendent across firms (p. 318). However, because of the incentive agency costs, he will attempt to reduce these by
for adverse selection to operate where information asymmetry offering protective debt covenants, such as the
is present, measurement errors will not be independent across dividend constraint. Other things being equal, the
firms (Ronen, 1979). greater the agency costs of debt, the tighter such a
20Barnea,Haugen and Senbet (1981) and Weston (1981) also constraint will be. However, a tight constraint
suggest that information asymmetry produces these costs.
21Thisobservation was first made by Ronen (1979). However, increases the chance of forced investment in nega-
he did not draw out the difference in who bears agency costs tive net present value projects, since the dividend
between above and below average firms. constraint may forbid distribution of surplus funds
52 A C C O U N T I N G A N D BUSINESS RESEA RCH

to shareholders. This is an overinvestment prob- theories, it is conceivably possible to combine them

lem. to yield predictions about accounting choices not
Firms with high growth potential, that is a large obtainable from either theory alone.22However,
number of positive net present value projects avail- the problems of operationalising both theories,
able, will find this overinvestment problem less mentioned earlier, make this a difficult task. At
costly than other firms and so will tend to offer present, the prediction of accounting choices can at
tighter dividend constaints to debtholders. In this least be improved by adding together the predic-
way, a tight dividend constraint can act as a signal tions from each theory. (This is possible logically
about the expected value of the firm. given the consistency of the theories.)23 Some
examples follow.
Signalling Costs
Each of these three bonding devices has a signal- Corporate Lobbying
ling cost function entailing opportunity costs. The Existing studies” of corporate lobbying about
manager incurs no penalty if he does not deviate proposed accounting standards have used agency
adversely from the contractual levels set by the theory to predict the decision to lobby and lob-
three devices. (There are also direct contract nego- bying position. Larger firms tend to lobby. There
tiation costs contained explicitly in agency theory.) is also some evidence that the decision to lobby is
If adverse deviation from contract does occur, the associated with leverage and the use of manage-
manager is penalised the more he deviates. These ment bonus schemes. Larger firms oppose account-
penalties include a decline in his ‘price’ in the ing standards which increase reported earnings,
market for managers and wealth opportunities while firms with accounting based debt or manage-
foregone for the firm and for the manager due to ment compensation contracts favour such stan-
his lack of diligence. dards. The opposite applies to proposed standards
Fama (1980) explicitly models these penalties in which reduce reported earnings.
his version of agency theory as the process of full Generally accepted accounting principles set a
ex post settling up. However, Jensen and Meckling lower bound on the fineness of disclosed account-
(1976) are silent about these penalties. Never- ing information. They also contain a degree of
theless, their version of agency theory is not incon- flexibility which is often criticised in the accounting
sistent with the existence of such penalties. Indeed, literature. Rational investors and creditors will
Jensen and Meckling discuss the role of factors expect management to use this flexibility in report-
other than direct monitoring and bonding outlays ing the firm’s performance in the most favourable
(p. 328) in controlling the behaviour of managers, way. Therefore, a tendency exists here for adverse
but do not include them in their model. selection to operate. To counteract this tendency,
high quality firms will wish to signal that they are
Conclusion not using this flexibility to mislead shareholders, or
Since the sufficient conditions outlined for sig- that they are not doing so as much as other firms.
nalling theory are at least consistent with those of One way of doing this is for such firms to be seen
agency theory, it can be concluded that the two as users of the ‘best’ accounting policies advocated
theories are consistent. Indeed, a considerable by the accounting profession, provided that these
amount of overlap exists between the theories policies do not allow low quality firms to masquer-
because several of the sufficient conditions for ade as high quality ones. Therefore, it is in the
signalling are implicit in agency theory. However, interest of high quality firms that accounting prin-
because information asymmetry, a necessary con- ciples allow them to reveal that they are in fact
dition of signalling theory, is not a necessary high quality.
condition of agency theory, signalling theory is not To ensure that accounting standards of this kind
implied by agency theory; nor are they equivalent are introduced, high quality firms will lobby the
theories. accounting profession. So too will poorer quality
firms, in whose interest it is that accounting stan-
dards do not provide fine information signals. So
Applications both types of firm have an incentive to lobby.
From the previous section, it is clear that some
constructs of agency theory-bonus plans, con-
tractual debt limits, and dividend constraints-can 22Sucha ‘meta-theory’ could also contain new concepts not
included in either signalling or agency theories.
also act as signals. An implication for accounting 2 3 F ~example,
r Puro (1984) argues that agency theory and the
research is that proxy variables used to measure economic theory of regulation are consistent, and takes to-
these constructs may capture both their agency gether the predictions from these theories to explain audit firms’
theory and signalling aspects. Results of empirical lobbying practices.
“These studies have investigated corporate lobbying about
studies using these variables must be carefully general price level accounting (Watts & Zimmerman, 1978),
interpreted as a result. foreign currency translation (Griffin, 1982, 1983; Kelly, 1982)
Given the consistency of signalling and agency and interest capitalisation (Dhaliwal, 1982).
WINTER 1987 53

Lobbying itself is not a signal, mainly because accounts, or one theory may be a subset of the
there seem to be no penalties for making a mis- other. Agency theory and signalling theory are not
leading lobbying submission. equivalent, nor does one theory imply the other,
For both high and low quality firms, the benefits because they do not share the same necessary
from lobbying should be higher the greater the conditions. However, the sufficient conditions of
information asymmetry between the firm and its signalling theory given in the fourth section are
investors. This will occur in firms with large num- at least consistent with those of agency theory.
bers of shareholders or geographically widespread Rational behaviour is common to both theories;
shareholdings. information asymmetry is implied by positive
monitoring costs in agency theory; quality can be
Accounting Policy Choice defined in terms of agency theory variables; and
In several studies,25 accounting policy choices signalling costs are implicit in some bonding de-
were related to size and debt covenants, and to a vices of agency theory. Therefore, agency theory
lesser extent to management compensation and signalling theory are consistent. Indeed, a
schemes. As with lobbying, larger firms favour considerable amount of overlap exists between
accounting policies which reduce reported earn- them.
ings, while firms with debt covenants and manage- The fifth section offered examples where both
ment compensation schemes based on reported theories’ predictions about lobbying, accounting
earnings favour income increasing accounting pol- choices, and voluntary auditor selection were
icies, ceteris paribus. The contribution of signalling added together (logically possible with consistent
theory is the prediction that higher quality firms theories). However, further research is needed to
will choose accounting policies which allow their combine signalling and agency theories into a more
superior quality to be revealed, while lower quality general ‘meta-theory’ which might yield insights
firms will choose accounting methods which at- into the principal-agent problem, and choice of
tempt to hide their poor quality. For example, a accounting methods, not obtainable from either
higher quality firm may voluntarily adopt segment theory alone.
reporting to disclose the superior risk/return
profile of its operations, but a low quality firm
would not. Similarly a high quality firm may
voluntarily disclose an earnings forecast, but a low Appendix
quality firm would not. As in lobbying, the incen- The objective of this appendix is to demonstrate
tive to signal by accounting policy selection should that information asymmetry can lead to agency
be highest where information asymmetry is great- costs which are borne by the agent in above
est: in firms with large and widespread numbers of average firms, and by the principal in below aver-
shareholders. Nevertheless, the number of ac- age firms. The graphical device employed has been
counting policies which act as valid signals may be used previously by Myers (1977), Kalay (1978) and
limited, given the present flexibility of generally Holthausen (1979) to demonstrate the wealth
accepted accounting principles. transfers that can occur between debtholders and
equityholders in a firm.
Voluntary Auditor Selection
Chow (1982) demonstrated how leverage, ac- Assumptions
counting based debt covenants and firm size were 1. Firms operate in two periods, t o = the
related positively to voluntary auditor ap- present and t , = the future. Figures 2 and 3
pointments by US companies when such ap- show t , outcomes in to present values.
pointments were unregulated. Signalling theory’s 2 . There is no debt financing.
prediction is that auditors will be appointed volun- 3. Where no information about specific firms is
tarily by higher quality firms in order to discrimi- available, they are treated alike by investors,
nate themselves from other firms in the market. whose general perceptions are determined
Again, this incentive should be greatest in firms by exogenous factors such as their general
with large and widespread numbers of share- knowledge about the corporate sector, the
holders. Further aspects of signalling and auditing state of the economy etc.
are discussed in Bar-Yosef and Livnat (1984). 4. Risk about the value of firms is depicted as
varying t , outcomes in different states of the
world.
Summary and conclusions 5. All states of the world are equally likely to
Any two theories at the same level of reduction occur.
may be either equivalent, consistent or competing 6. For convenience, states of the world have
been arranged in Figures 2 and 3 so that
25Holthausenand Leftwich (1983) and Kelly (1983) review value schedules are linear.
these studies. 7. Firms are wholly owned by managers who
54 ACCOUNTING A N D BUSINESS RESEARCH

intend to float shares in them to outside By signalling, the manager’s Compensation is

investors. OM,FD and he receives GJV,H from selling shares
8. Managers maximise their wealth by max- in the firm. If he does not signal, the manager
imising their expected compensation and the receives OAVD in compensation and share sale
price at which shares are issued. proceeds, and he incurs an opportunity loss of
9. Each manager’s compensation is a fixed AJV,V - M,GHF which is the manager’s net gain
proportion of the expected value of his firm from signalling. Hence, he has an incentive to
in t , . This compensation has priority of signal.
payment over returns to investors. In Figure 3, AV and MBC are again the value
10. Signalling costs are inversely related to the and management compensation schedules for the
value of the firm in any state of the world. average firm. Suppose that the manager of firm V,
Signalling costs rank after management knows that its value is less than 0 in most states
compensation in priority for payment. of the world, and that it is a riskier investment as
11. All firms have accepted projects with net shown by the steeper slope of OV,. If this informa-
present values of at least zero in all states of tion were known to investors, the manager of V,
the world. would only receive EV,F when issuing the shares
These assumptions differ slightly from those in and OEFD in compensation. Uninformed, the
the body of the paper in order to streamline the market would pay BVC for the shares, and the
exposition which follows. manager’s compensation would be OABCD.
Figure 2 shows AT, the value of all firms where Hence, the gain to him from investors not being
no signalling or monitoring is permitted. Investors informed is OAG - GV,V. No signalling will oc-
perceive all firms as having the same expected value cur, and uninformed shareholders bear the loss.
E(Y) and risk. Managers’ compensation is nego-
tiated as aE(O) shown as MBC. Since all states of
the world are assumed equally likely to occur, the
value of managers’ compensation and the value of
equity can be depicted as segments OABCD and
BOC of area OAOD, the value of the firm.
Suppose that the manager of one firm, JV,,
knows that its value exceeds O in all states of the
world, and is less risky than AO as shown by the
shallower slope of JV,. If the manager can commu-
nicate this to investors, he will receive a higher
price for the shares when they are floated, and his
compensation will increase to aE(V,) shown by
M,EF in Figure 2. Signalling will convey this
information to investors but at a cost given by GH.
The cost schedule is negatively sloped reflecting the States of the world
assumption that signalling costs are inversely re- Fig. 3
lated to value.

ul

E
8
+
3
0
c
c
.I-

C U J
3
d
G
0

c
c
$
$ Mi
“ M
A /’
0 ’

States of the world

Fig. 2
W I N T E R 1987 55

Similar Droofs could be offered for more com- Holthausen, Robert W. and Richard W. Leftwich (1983), ‘The
plex situations, with the Same results. The graph- Economic Consequences of Accounting Choice: Implications
of Costly Contracting and Monitoring’, Journal of Account-
ical proof also can be adapted easily to show ing and Economics, August 1983.
Similar effects of information asymmetry where Jaffe. J. F. (1974). Soecial Information and Insider Trading’. I ,

there is debt financing. Journal of Business, July 1974.

Jensen, Michael C. and William H. Meckling (1976), ‘Theory
of the Firm: Managerial Behavior, Agency Costs and Own-
ership Structure’, Journal of Financial Economics, October
References 1976.
Jensen, Michael C. (1983), ‘Organization Theory and Meth-
Akerlof, George A. (1970), ‘The Market for “Lemons”: Quality odology’, Accounting Review, April 1983.
Uncertainty and the Market Mechanism’, Quarterly Journal John, Kose and Joseph Williams (1985), ‘Dividends, Dilution
of Economics, August 1970. and Taxes: A Signalling Equilibrium’, Journal of Finance,
Ball, Ray and George Foster (1982), ‘Corporate Financial September 1985.
Reporting: A Methodological Review of Empirical Rv- Kalay, Avner (1978), ‘Towards a Theory of Corporate Divi-
search’, Journal of Accounting Research Supplement, 1972. dend Policy’, Unpublished PhD thesis, University of Roch-
Barnea, Amir, Robert A. Haugen and Lemma W. Senbet ester, 1978.
(1981), ‘Market Imperfections, Agency Problems, and Capi- Kalay, Avner (1980), ‘Signaling, Information Content, and the
tal Structure: A Review’, Financial Management, Summer Reluctance to Cut Dividends’, Journal of Financial and
1981. Quantitative Analysis, November 1980.
Bar-Yosef, Sasson and Joshua Livnat (1984), ‘Auditor Selec- Kelly, Lauren (1982), ‘Corporate Lobbying and Changes in
tion: An Incentive Signalling Approach’, Accounting and Financial or Operating Activities in Reaction to FAS No. 8’,
Business Research, Autumn 1984. Journal of Accounting and Public Policy, Winter 1982.
Bhattacharya, Sudipto (1979), ‘Imperfect Information, Divi- Kelly, Lauren (1983), ‘The Development of a Positive Theory
dend Policy and the “Bird in the Hand” Fallacy’, Bell Journal of Corporate Management’s Role in External Financial
of Economics, Spring 1979. Reporting’, Journal of Accounting Literature, Spring 1983.
Bhattacharya, Sudipto (1980), ‘Nondissipative Signaling Struc- Kihlstrom, Richard E. and Michael H. Riordan (1984), ‘Adver-
tures and Dividend Policy’, .~ Quarterly Journal of Economics, tising as a Signal’, Journal of Political Economy, June 1984.
August 1980. Leland, Hayne E. and David H. Pyle (1977), ‘Information
Chow, Chee W. (1982), ‘The Demand for External Auditing: Asymmetries, Financial Structure, and Financial Inter-
Size. Debt and Ownership Influences’, Accounting Review, mediation’, Journal of Finance, May 1977.
April 1982. Mayers, David and Clifford W. Smith (1981), ‘Contractual
Demski, Joel S. and David M. Kreps (1982), ‘Models in Provisions, Organizational Structure, and Conflict Control in
Managerial Accounting’. - . Journal of Accounting Research, Insurance Markets’, Journal of Business, July 1981.
Supplement 1982. Mayers, David and Clifford W. Smith (1982), ‘On the Cor-
Dhaliwal, Dan S. (1982), ‘Some Economic Determinants of porate Demand for Insurance’, Journal of Business, April
Management Lobbying for Alternative Methods of Account- 1982.
ing: Evidence from the Accounting for Interest Costs Issue’, Miller, Merton H. and Kevin Rock (1985), ‘Dividend Policy
Journal of Business Finance and Accounting, Summer 1982. under Asymmetric Information’, Journal of Finance, Sep-
Downes, David H. and Robert Heinkel (1982), ‘Signaling and tember 1985.
the Valuation of Unseasoned New Issues’, Journal ofFinance, Myers, Stewart C. (1977), ‘Determinants of Corporate Borrow-
March 1982. ing’, Journal of Financial Economics, November 1977.
Eades. Kenneth M. (1983). ‘Empirical Evidence on Dividends Penman, Stephen H. (1982a), ‘Tests of Dividend Signaling: A
as a Signal of Firm Viue’, Journal of Financial and Quan- Comparative Analysis’, unpublished paper, University of
titative Analysis, November 1983. California, Berkeley, 1982.
. Fama, Eugene F. (1980), ‘Agency Problems and the Theory of Penman, Stephen H. (1982b), ‘The Relative Informativeness of
the Firm’, Journal of Political Economy, April 1980. Alternative Insider Trading Measures’ unpublished paper,
Fama, Eugene F. and Michael C. Jensen (1983), ‘Agency University of California, Berkeley 1982.
Problems and Residual Claims’, Journal of Law and Eco- Puro, Marsha (1984), ‘Audit Firm Lobbying Before the Fi-
nomics, June 1983. nancial Accounting Standards Board: An Empirical Study’,
Forker, J. J. (1984), ‘Contract Value Accounting and the Journal of Accounting Research, Autumn 1984.
Monitoring of Managerial Performance: An Agency-Based Ronen, Joshua (1979), ‘The Dual Role of Accounting: A
Proposal’, Accounting and Business Research, Spring 1984. Financial Economic Perspective’, in James L. Bicksler (ed.)
Griffin, Paul A. (1982), ‘Foreign Exchange Gains and Losses: Handbook of Financial Economics (New York: North-
Impact on Reported Earnings’, Abacus, June 1982. Holland Publishing, 1979).
Griffin, Paul A. (1983), ‘Management’s Preferences for FASB Ross, Stephen A. (1977), ‘The Determination of Financial
Statement 52: Predictive Ability Results’, Abacus, December Structure: The Incentive-Signaling Approach’, Bell Journal of
1983. Economics, Spring 1977.
Haugen, Robert A. and Lemma W. Senbet (1979), ‘New Ross, Stephen A. (1979), ‘The Economics of Information and
Perspectives on Informational Asymmetry and Agency Re- the Disclosure Regulation Debate’, in Franklin R. Edwards
lationships’, Journal of Financial and Quantitative Analysis, (ed.), Issues in Financial Regulation (New York: McGraw-
November 1979. Hill, 1979).
Healy, Paul M. (1985), ‘The Effect of Bonus Schemes on Rothschild, Michael and Joseph Stiglitz (1976), ‘Equilibrium in
Accounting Decisions’, Journal of Accounting and Economics, Competitive Insurance Markets: An Essay on the Economics
April 1985. of Imperfect Information’, Quarterly Journal of Economics,
Heinkel, Robert (1982), ‘A Theory of Capital Structure Rele- November 1976.
vance Under Imperfect Information’, Journal of Finance, Smith, Clifford W. and Jerold B. Warner (1979), ‘On Financial
December 1982. Contracting: An Analysis of Bond Covenants’, Journal of
Holthausen, Robert W. (1979), ‘Bond Covenants and the Financial Economics, June 1919.
Choice of Accounting Techniques: The Case of Alternative Spence, A. Michael (1973), ‘Job Market Signaling’, Quarterly
Depreciation Methods’, Unpublished paper, University of Journal of Economics, August 1973.
Rochester, 1979. Spence, A. Michael (1974), Market Signaling: Informational
56 A C C O U N T I N G A N D BUSINESS R E S E A R C H

Transfer in Hiring and Related Screening Processes (Cam- Positive Theory of the Determination of Accounting Stan-
bridge, Mass.: Harvard University Press, 1974). dards’, Accounting Review, January 1978.
Talmor, Eli (1981), ‘Asymmetric Information, Signaling, and Weston, J. Fred (1981), ‘Developments in Finance Theory’,
Optimal Corporate Financial Decisions’, Journal of Financial Financial Management, 2, 1981.
and Quantitative Analysis, November 1981. Yetton, Philip and Andrew Crouch (1983), ‘Social Influence
Watts, Ross L. and Jerold L. Zimmerman (1978), ‘Towards a and Structure: Elements of a General Theory of Leadership’.
Australian Journal of Management, December 1983.