POSITIVE ACCOUNTING THEORY, POLITICAL COSTS AND SOCIAL DISCLOSURE ...

POSITIVE ACCOUNTING THEORY, POLITICAL COSTS
AND SOCIAL DISCLOSURE ANALYSES: A CRITICAL LOOK





Markus J. Milne
Accountancy and Business Law
University of Otago
Dunedin
New Zealand

Ph: 64-3-479-8120
Fax: 64-3-479-8450
Email: mmilne@commerce.otago.ac.nz

---

POSITIVE ACCOUNTING THEORY, POLITICAL COSTS
AND SOCIAL DISCLOSURE ANALYSES: A CRITICAL LOOK*






ABSTRACT


This paper critically reviews the literature seeking to establish evidence for a positive accounting
theory of corporate social disclosures. It carefully traces through the original work of Watts and
Zimmerman (1978) showing their concern with the lobbying behaviour of large US oil companies
during the 1970s. Such companies were argued to be abusing monopolists and likely targets of self-
interested politicians pursuing wealth transfers in the form of taxes, regulations and other ‘political
costs’. Watts and Zimmerman’s reference to “social responsibility” is shown to be a passing
remark, and most likely refers to “advocacy advertising”, a widespread practice amongst large US
oil companies at that time. Subsequent literature that relies on Watts and Zimmerman to present a
case for social disclosures is shown to extend their original arguments. In the process, concern over
the “high profits” of companies is shown to diminish, and the notion of political costs is so
broadened that it blurs with other social theories of disclosure. Consequently, the positive
accounting based social disclosures literature fails to provide distinct arguments for self-interested
managers wealth maximising. This paper also shows that the empirical evidence gathered to date
in support of a positive accounting theory of social disclosures largely fails in its endeavour.






















* The author would like to thank Alan MacGregor, Carolyn Stringer, Gregory Liyanararchchi, Ros Whiting
and an anonymous reviewer for their helpful comments on earlier drafts of this paper.

---

INTRODUCTION

Along with numerous other rationales (e.g., decision usefulness, legitimacy theory, stakeholder
theory, critical or political economy theory—see Gray et al., 1995 for a review), positive
accounting theory or the political cost hypothesis has been suggested to explain why firms make
voluntary social disclosures. Based on the original work of Watts and Zimmerman (1978, 1986),
several empirical studies (e.g., Belkaoui and Karpik 1989; Ness and Mirza, 1991; Panchapakesan
and McKinnon, 1992; Lemon and Cahan, 1997) have directly sought to establish evidence for the
political cost hypothesis as an explanation for firms’ social disclosures. Several related empirical
studies have also sought to use the political cost hypothesis to explain other types of voluntary
disclosure, including value added statements (Deegan and Hallam, 1991), disclosures by statutory
authorities (Lim and McKinnon, 1993), and disclosures in pursuit of reporting excellence awards
(Deegan and Carroll, 1993).

Along with numerous others1, Gray et al., (1995, pp. 51-52) dismiss the positive accounting
arguments and literature on the grounds of the underlying assumptions of the theoretical
framework. As they suggest, positive theories are not about what (social) reporting should be, but
rather about what it is. As a basis for change and improvement, positive theories are seen to offer
little or no development of corporate (social) reporting. Gray et al. (1995, p.52 and fn. 11) go on to
admit, however, having been persuaded by many of the critiques, that they have not seriously
engaged this literature because they believe it to be “virtual rubbish” and prefer their position as
“heretics”.

On the face of it, however, and as a basis for explaining why firms are making social disclosures,
positive accounting explanations are less easily dismissed. Casual observation, for example, reveals
that positive accounting explanations rely on empirical evidence largely identical to that used in
support of other explanations (most notably, legitimacy theory) of social disclosures; explanations
which, incidentally, Gray et al. (1995) seem to find more acceptable. As Gray et al. (1995, p. 49;
2001) note, a number of empirical studies have shown strong associations between disclosure and
firm size, and between disclosure and type of industry. In fact, the size-disclosure relationship
appears empirically the most robust.2 Such results are claimed in support of legitimacy theory (e.g.,
Patten, 1991, Deegan and Gordon, 1996), as well as in favour of positive accounting theory. Lemon
and Cahan (1997, p. 79), in fact, could not have put the issue more clearly when they state:

Patten finds that the public pressure variables [firm size, and industry classification] are significant
while the profitability variables [ROA, ROE] were not. Patten interprets these results as supporting
legitimacy theory where a firm must satisfy an implied contract with the society it operates in. To
the contrary as firm size is commonly used to represent a firm’s political visibility, we interpret
Patten’s results as consistent with the political cost hypothesis.

If positive accounting theory is to be rejected as a basis for explaining why firms are making social
disclosures, then it requires a more rigorous examination of the arguments and empirical evidence
than has occurred to date. Of course, according to Watts and Zimmerman’s theory, the answer why

1 There have been numerous critiques of Watts and Zimmerman’s positive accounting theory. Gray et al. (1995)
themselves cite several of these including Christenson (1983), but also see Ball and Foster (1982), Tinker et al. (1982),
Lowe et al. (1983), Whittington (1987), Hines (1988), Whitley (1988), Sterling (1990), Chambers (1993) and Tinker
and Puxty (1995).
2 See, for example, Kelly (1981), Trotman and Bradley (1981), Belkaoui and Karpik (1989), Patten (1991),
Panchapakesan and McKinnon (1992), Adams et al. (1995; 1998), Deegan and Gordon (1996), Hackston and Milne
(1996), Lemon and Cahan (1997).


1

---

firms (managers) make social disclosures is because it is in their interests to do so3. Of interest to
this paper, however, is upon what basis, and particularly if at all, do researchers go about
demonstrating such a claim, and to what extent are such claims persuasive.

Watts and Zimmerman (1978, 1986) provide the stated theoretical basis for a number of social
disclosure studies and so their work is reviewed first in some detail. Direct reference to social
disclosure or social responsibility by Watts and Zimmerman themselves, however, appears
extremely scant. From the later section on social disclosure, it will become clear that the use of
Watts and Zimmerman’s work relies on interpretations and extensions to the original. In the
process of reviewing this work, it will be shown that the original arguments of Watts and
Zimmerman have been, if not misused, misunderstood or taken out of context, then only partially
argued and tested.

WATTS AND ZIMMERMAN (1978) AND SOCIAL RESPONSIBILITY

A review of Watts and Zimmerman’s original positive accounting theory papers (1978, 1979), their
subsequent book, Positive Accounting Theory (1986), and their own review paper on positive
accounting theory (1990), reveals that reference to “social responsibility” is only made in the
original 1978 paper. I shall turn to that reference next, but first it is worth noting that no title of any
paper they have either jointly or separately published since 1978 contains any reference to social
responsibility4. Furthermore, only 1 paper (Blacconiere & Patten, 1994) with any reference to
matters of social or environmental responsibility/disclosure in the title has been published in the
Journal of Accounting and Economics —a journal which has been strongly associated with positive
accounting theory, and which Watts and Zimmerman founded in 1979.

That social responsibility/disclosure is an issue that appears to be of only passing interest to Watts
and Zimmerman and their positive accounting theory project is further substantiated when one
carefully examines the original reference to it, and the context in which it was made. Watts and
Zimmerman (1978), builds on Watts (1977), and seeks to (p. 112):

…develop a positive theory of the determination of accounting standards. Such a theory will help
us to understand better the source of the pressures driving the accounting standard-setting process,
the effects of various accounting standards on different groups of individuals and the allocation of
resources, and why various groups are willing to expend resources trying to affect the standard-
setting process.

Watts and Zimmerman (1978, p. 113) assume that “individuals act to maximise their own utility”
and consequently “management lobbies on accounting standards based on its own self-interest.”
They go on to suggest their purpose is to “identify factors which are likely to be important
predictors of lobbying behaviour…”, and a key part of this task is to examine how accounting
standards affect management’s wealth.

Management wealth, it is argued, is a function of changes in share prices (via stocks and stock
options), and changes in cash bonuses (via compensation plans). Ordinarily, managers are predicted

3 As Sterling (1990, p. 117, emphasis in original) and others (e.g., Chambers, 1993) make clear, understanding why
firms choose certain accounting methods is not at issue. Utility maximisation is always the answer according to the
assumption of W&Z. The research objective is to demonstrate a set of relationships consistent with that assumption. Of
course, ‘self-interest’ could, quite reasonably on the available evidence to date, be claimed to lie at the heart of all
voluntary disclosure behaviour, even where such disclosure behaviour is made to appease certain constituents and
retain organisational legitimacy. W & Z’s assumption, however, is that accounting manipulations serve managers’
economic self-interest.
4 See their web-based publication lists at: http://www.simon.rochester.edu/fac/zimmerma/html/pubs.html,
and http://www.simon.rochester.edu/fac/watts/html/publicat.html.

2

---

to have greater incentives to lobby for accounting standards that lead to increases in reported
earnings and thereby management wealth. Since changes in cash flows and stock prices can also be
affected (reduced) by taxes, regulatory procedures (for regulated firms), information costs, and
political costs, however, managers also have to consider the effects reported earnings might have
on the likelihood that such costs could be imposed on the firm. In some cases, the extra costs of
accounting standards that lead to increases in reported earnings may outweigh the benefits (to
management) of the reported earnings, and so management is predicted to lobby against such
changes. More specifically, Watts and Zimmerman (1978, p. 118) argue that:

…managers have greater incentives to choose accounting standards which report lower earnings
(thereby increasing cash flows, firm value, and their welfare) due to tax, political, and regulatory
considerations than to choose accounting standards which report higher earnings and, thereby,
increase their incentive compensation. However, this prediction is conditional upon the firm being
regulated or subject to political pressure. In small, (i.e., low political costs) unregulated firms, we
would expect that managers do have incentives to select accounting standards which report higher
earnings, if the expected gain in incentive compensation is greater than the forgone expected tax
consequences.

Subsequent literature (e.g., Zmijewski and Hagerman, 1981; Healy, 1985, Press and Weintrop,
1990) has sought to refine these arguments and extend them beyond lobbying behaviour on
accounting standards to accounting method choice, where managers have the discretion to choose
among a set of accounting procedures. Reviewing this literature, Watts and Zimmerman (1986,
1990) highlight three key hypotheses as follows:

(1) The bonus plan hypothesis: Ceteris paribus, managers of firms with bonus plans are more likely to
choose accounting procedures that shift reported earnings from future periods to the current period.

(2) The debt/equity hypothesis: Ceteris paribus, the larger a firm’s debt/equity ratio, the more likely the
firm’s manager is to select accounting procedures that shift reported earnings from future periods to the
current period.

(3) The size hypothesis: Ceteris paribus, the larger the firm, the more likely the manager is to choose
accounting procedures that defer reported earnings from current to future periods.

So where within this argument does “social responsibility” fit? To understand this, we need to
further examine Watts and Zimmerman’s notion of political pressure, political costs and their ‘size
hypothesis’.

Political Costs and the Size Hypothesis
According to Watts and Zimmerman (1978, p. 115), politicians have the power to effect upon
corporations wealth re-distributions by way of corporate taxes, regulations, subsidies etc.
Moreover, certain groups of voters have incentives to lobby for the “nationalisation, expropriation,
break-up or regulation of an industry or corporation”, which in turn are seen to provide incentives
to politicians to propose such actions. This idea that politicians seek to intrude into the affairs of
corporations and redistribute wealth away from them comes from the earlier work of Stigler (1971),
Peltzman (1976) and Jensen & Meckling (1978).

To counter such pressure from politicians, Watts and Zimmerman (1978, p. 115) suggest:

… corporations employ a number of devices, such as social responsibility campaigns in the media,
government lobbying and selection of accounting procedures to minimize reported earnings. By
avoiding the attention that “high” profits draw because of the public’s association of high reported

3

---

profits and monopoly rents, management can reduce the likelihood of adverse political actions and,
thereby, reduce its expected costs (including the legal costs the firm would incur opposing the
political actions). Included in political costs are the costs labor unions impose through increased
demands generated by large reported profits.

The magnitude of the political costs is highly dependent on firm size.

The source of concern Watts and Zimmerman have in mind for the public and so politicians, then,
is “large profits” and their association with monopoly power. Environmental degradation by
companies, depletion of resources, exploitation of workers, the production of unsafe products, and
so on, as sources of public and political concern are not mentioned. It is the potential abuse of
monopoly power that lies at the heart of Watts and Zimmerman’s notion of political costs. Their
reference (1978, p. 115, fn. 12) to Menke’s US Congressional statements that “Huge accounting
profits, but not high profit rates, are an inevitable corollary of large absolute firm size. This makes
these companies obvious targets for public criticism.” further substantiates this point. Similarly,
their subsequent references (see 1986, p. 235, and fn. 3 to Alchian & Kessel, 1962, and Jensen &
Meckling, 1978) to the “size hypothesis” emphasise the concern of the press and politicians with
size of profits and potential monopoly abuses. As Watts (1977, p. 68) states in interpreting his work
with Zimmerman, “They [W&Z, 1978] argue that the corporate manager has an incentive to select
accounting procedures and to lobby with politicians and bureaucrats for accounting procedures
which reduce the net income reported in financial statements. Political entrepreneurs use “high”
profits to create “crises”.” Wong’s (1988, pp. 27-29) interpretation and examples of political costs,
too, are consistent with an emphasis on large profits. Watts and Zimmerman also reinforce this idea
in their 1990 review by stating “…political costs are a function of reported profits. Thus, incentives
are created to manage reported accounting numbers” (p. 133).

Precisely how “social responsibility campaigns in the media” were envisioned to fit into Watts and
Zimmerman’s notion of political costs is not made clear. Unless they involve significant
expenditures (both on actual social responsibility activities, and in the form of “advertising costs”),
by themselves they would unlikely provide much reduction in “large profits” and so reduce the
apparent source of concern to the public and politicians. Alternatively, or perhaps as well, these
“campaigns in the media” are believed to be devices to distract attention away from a firm’s large
profits, and so soften the image of abusive monopolies; in other words, to legitimise the firm’s
large profit making. Yet, again, these campaigns in the media may have little to do with profit
making and monopoly abuses, but with other aspects of business behaviour found unacceptable to
members of society.

Oil Companies and Advocacy Advertising
Some insights into Watts and Zimmerman’s reference to social responsibility campaigns in the
media are possible from their discussion of political costs. Nearly all of their examples come from
the oil industry (see, Watts, 1977, p. 68, W&Z 1978, p.115, p.121; W&Z 1986, p. 236-237;
Zimmerman, 1983; Holthausen and Leftwich, 1983, p. 88). Oil companies during the 1970s
dominated the largest of the US companies, and they were also subject to much public outrage and
political scrutiny during and immediately following the ‘oil crisis’ of 1973.5 Also during this same
period, a number of companies, including oil and energy companies, and especially Mobil Oil
(Inc), undertook what Sethi (1976, 1977a, 1977b) as referred to as ‘advocacy advertising’.


4

---

Advocacy advertising often undertakes an editorial style approach and represents a deliberate
attempt on the part of the corporation to present a point of view about a major public issue. The
strength of such advertising, according to Sethi, is that the content of the message is controlled and
defined in a manner favourable to the sponsor and the environment of the message is carefully
controlled, thus making otherwise one-sided viewpoints appear more objective. Furthermore, Sethi
(1977b) suggests that the sociocultural environment of the advertisements was one of traditional
American values, such that if one sought to raise objections to the advertisements’ content one must
invariably also be seen to be criticising such traditional values. Several examples of such
advertising are provided in the appendix.6

In the context of Watts and Zimmerman, Sethi’s work on advocacy advertising is particularly
illuminating for several reasons. First, advocacy advertising obviously comes much closer to
lobbying behaviour than annual report social disclosures. Sethi (1976), for example, reports that
many critics of advocacy advertising tried to have such advertising classified as “grass-roots”
lobbying —a practice specifically exempt for corporate tax expense deductions. Second, the role of
politicians and the media, and particularly the news media are clearly implicated as a rationale for
advocacy advertising. Raleigh Warner, Jr., Chairman of Mobil Oil, for example, suggested the
purpose of Mobil’s advocacy advertising campaign was to defend his company against slander and
unfair treatment in the media (Sethi, 1977a, p.52). Similarly, quotes found in Sethi (1977a) further
substantiate this rationale:

Corporations are constantly under attack from every quarter. Charges by consumer groups,
Congressional Committee and Subcommittee Chairmen, individual Congressmen and the various
organs of federal, state and local government get front-page headlines….Silence by a company in
the face of attacks upon its policies and practices is interpreted as an admission of guilt. Corporate
advertising provides one avenue of self-defense (Frederick West, Jr., president of Bethlehem Steel
Corporation in letter to Senator Hart 1974, quoted in Sethi, 1977a, p. 53).

It is understandable that the affected businesses may not like what they read about themselves in
the newspapers. They use the argument of media bias in objecting to these stories. This alleged
bias then becomes an excuse for publishing the kind advocacy advertisements that we’re beginning
to see. The problem has developed more markedly since the oil crisis, when oil and energy
companies have become more aggressive in buying advertising space to respond to public
criticism of their businesses (John B. Oakes, New York Times, quoted in Sethi, 1977a, p. 54).

Third, the evidence provided by Sethi (1976) suggests the direct impact of such advertising
campaigns on profits is likely to have been minimal. Citing evidence provided at the U.S. Senate
Hearings into Energy and Environmental Objectives in 1974, Sethi (1976, p. 10) suggests “actual
business spending on institutional goodwill advertising is small when measured in terms of total
advertising expenditures, sales revenues, or net profits.” He goes on to point out that the costs of
institutional advertising by the largest ten corporate advertisers in 1974 amounted to 0.046% of

5 During the 1970s several books were released critically examining the oil industry. See, for example, Blair’s (1976)
The Control of Oil, Sampson’s (1975) The Seven Sisters, and Engler’s (1977) The Brotherhood of Oil. The thrust of
many of these analyses was abuse of power, collusion and profiteering.
6 While the adverts shown in the appendix have been selected to illustrate the arguments here, Sethi (1977b, p. 16)
provides ample evidence that such advertising was dominated by oil companies, and especially so during the 1970s. For
example, he states “In 1971, of the top ten companies running institutional advertising, one was an oil firm, three were
public utilities, and one was an automobile concern. By 1972, the number of oil firms had risen to four, utilities
decreased to one, as did automobile manufacturers… in 1974… five of the top ten corporate advertisers were oil
companies…”. The content of such adverts were tracked by Benson & Benson, and during 1975 typically covered
energy (40%), economics/regulation (20%), ecology (15%), corporate social responsibility (10%) and capital
investment (9%). Oil companies and utilities dominated the energy advertisements, with Mobil and American Electric
Power dominating the economics/regulation advertisements, and 60% of all corporate social responsibility
advertisements were accounted for by four oil companies: Mobil, Gulf, Arco, and Exxon (see Sethi, 1977b, p. 55).

5

---

their sales revenues, and 0.67% of their net income. The potential of such advertising to preserve
existing and future profitability, however, should not be underestimated. According to Harold
Johnson, vice president of American Electric Power (quoted in Sethi, 1976, p.10), his company’s
advertisement campaign was a major factor in “persuading the White House, the EPA, and
important senators and congressmen to postpone the enforcement of the Clean Air Act standards
from 1975 to 1985.” Furthermore, according to Sethi (1976, p.10), Johnson believes how
government regulates business can and does affect corporate profits, and ultimately its survival, in a
major way. Such advertisement campaigns are seen as of major importance to a company’s
performance, and to the public it serves.

Fourth, and as already noted with respect to Johnson’s American Electric Power, many of these
advertisements were being used to counter sources of public and political concern other than “large
profit” making. In American Power’s case, the advertisements were clearly targeted to prevent or
stall the Clean Air Act, a piece of legislation targeting air polluters.

It seems likely, then, that Watts and Zimmerman’s notion of political costs was largely derived
from observations of the behaviour to and of US oil companies during the early and mid 1970s.7
Reference to social responsibility campaigns in the media was most likely based on the popular US
practice of advocacy advertisements at that time. Such advertising, however, seems to only partially
fit Watts and Zimmerman’s argument. It is clearly a means of lobbying behaviour, and lobbying
behaviour most often directly targeted at prospective legislation, an outcome of which could well
forestall reduction in future firm profitability. Whether such advertisements were intended to
reduce “high profits” or the visibility of high profits, however, is much less certain. Yet, the clear
objective for managers of such firms, according to Watts and Zimmerman’s theory, is to make
current profits lower and so remove the source of public and political concern.

While managers of firms may undertake social responsibility disclosure, media manipulation and/or
government lobbying, these activities seem unlikely by themselves to aid the management of
reported accounting numbers. Furthermore, as a means of lobbying against prospective legislation
that could reduce future firm profitability, targeted advocacy advertisements seem far more likely
to succeed than annual report social disclosures. Extending Watts and Zimmerman’s observations
about social responsibility campaigns in the media to social disclosures in annual reports seems to
be stretching too far. However, if managers are undertaking social responsibility disclosures in
annual reports as part of a strategy to reduce political costs and enhance their welfare, then at the
very least we should also expect to see evidence of those same firms undertaking other more direct
means to reduce reported profits (i.e., certain accounting method choices).

TESTING WATTS AND ZIMMERMAN’S POSITIVE ACCOUNTING THEORY

Before turning to studies of voluntary disclosure, several useful insights to testing Watts and
Zimmerman’s theory can be gained from briefly considering more conventional (i.e., accounting
choice) studies. First, their theory involves three, and I would say joint, predictors of accounting
choice behaviour. These three predictors, however, are not reinforcing, they imply opposing
incentives, and so trade-offs, for the management of earnings (Watts and Zimmerman, 1986, p.
246). Second, managers most likely have discretionary choices over a combination of accounting
procedures, and not just a single procedure like depreciation. Zmijewski and Hagerman (1981), for
example, show that four accounting procedures, each of which could be used in one of two ways
(i.e., income increasing or income decreasing), generates a possible set of 16 portfolio choices.
Consequently, Watts and Zimmerman’s model involves multiple (and joint) predictors of a

7 The more general point that Watts and Zimmerman’s theory is largely only applicable to the US has been made before
by others. See, for example, Lowe et al. (1983) and Peasnell and Williams (1986).


6

---

portfolio of procedures. As they suggest, however, it is difficult to specify a priori the relative
magnitude of the three opposing incentives (Watts and Zimmerman, 1986, p. 243), and so studies
typically use separate independent variables as surrogates for the offsetting incentives.
Notwithstanding this weakness, studies that seek to explain single procedures, rather than multiple
procedures (including perhaps social disclosure strategies) are less powerful tests of the hypotheses
(Watts and Zimmerman, 1986, p. 246). Likewise, studies that isolate and focus on single predictors
(e.g., political costs) of management behaviour are also incomplete and less powerful tests of their
theory.

In addition to concerns over correctly specifying an empirical model to test Watts and
Zimmerman’s theory, concerns have also been raised over the choice of proxies for the three
independent variables, and especially proxies for political costs.8 Watts and Zimmerman (1978)
originally used firm size (Fortune 500 rank of asset size) to proxy for political costs, but they and
others (see, for example, Ball and Foster, 1982; Whittred and Zimmer, 1990) have subsequently
criticised size as too noisy a proxy. Bujaki and Richardson (1997), for example, from a citation
review of the empirical use of firm size in accounting journals, claim to have found eighteen
distinct theoretical constructs for which size was used to proxy.9 They note that many of these
studies fail to attempt to establish either convergent or discriminant validity for size. While one
might disagree with their classification scheme, size is clearly found to proxy for more than
political costs.

Subsequent to Watts and Zimmerman (1978), empirical studies have tended to use or suggest a
wider range of measures to proxy for political costs. These include: profits, rates of return, risk,
capital intensity, industry concentration, industry membership, effective tax rates, number of
employees, number of shareholders, labour intensity, press coverage, and even social responsibility
disclosures themselves.10 In some cases, the metrics involve alternative measures of firm size (e.g.
number of employees). In other cases, however, they represent attempts to find alternative
measures of political visibility or political cost (e.g., capital intensity). In all cases, however, the
extent to which they are acceptable measures should depend upon the extent to which they can be
related back to the original arguments about “high profits” attracting public and political attention.
It will be seen that some studies, in suggesting or using measures of political cost, seem to ignore
the original arguments of Watts and Zimmerman, or, in the process of justifying their chosen
proxies, change the argument.

DISCLOSURE STUDIES AND WATTS AND ZIMMERMAN’S THEORY

Belkaoui & Karpik (1989)

Belkaoui & Karpik (1989) was one of the earliest and more comprehensive attempts to test Watts
and Zimmerman’s arguments with respect to annual report social disclosures. It is also perhaps the
only study that remains largely consistent with their original arguments. Specifically, Belkaoui and
Karpik (1989) invoke the debt/equity (measured by debt/assets and dividends/unrestricted earnings)
and political cost (measured by size, capital intensity and beta) hypotheses, but omit the bonus plan
hypothesis, when seeking to explain social disclosures. Politically visible firms are argued to
disclose more, while firms with high debt/equity ratios are argued to disclose less. Measures of

8 See Sterling’s (1990) remarks in his foonote #24, for example.
9 These included, for example, political costs, risk, analyst following, liquidity, management ability, expected returns,
trading frequency, bankruptcy likelihood, and social responsibility disclosure. It is notable in some cases, however, that
Bujaki and Richardson have confused empirical associations found between size and other measures (e.g., number of
social disclosure pages/words) as evidence of the use of size to proxy for such constructs/measures.
10 See, for example, Zmijewski and Hagerman (1981), Holthausen and Leftwich (1983), Watts and Zimmerman
(1986), Deegan and Hallam (1991), Panchapakesan and McKinnon (1992), Deegan and Carroll (1993), Lemon and
Cahan (1997).

7

---

social and economic performance are also included in their model, but social disclosure behaviour
is the only dependent variable tested. Evidence for both hypotheses is found in a regression model
that generates an adjusted R2 of 44%, with size, beta (risk) and leverage being the significant
variables. Social performance is also significant in the model.

Several issues arise from Belkaoui & Karpik (1989) that expose it as a weak assessment of Watts
and Zimmerman’s theory. Belkaoui & Karpik (1989, p. 38) acknowledge that “image-building and
public interest concerns may govern the decision to spend for social performance and to disclose
social information”, but they prefer to remain consistent with Watts and Zimmerman’s argument
and so emphasise that “specific and material expenditures are necessary to achieve social
performance goals. These same expenditures reduce net income.” The key to Watts and
Zimmerman’s argument, of course, is current income reducing accounting choices undertaken by
monopoly and large profit makers, and so the key to Belkaoui & Karpik’s (1989) argument is the
size of firms’ social spending (relative to current earnings). Belkaoui & Karpik (1989), however,
offer no evidence of firms’ social spending whatsoever, and also choose to use measures of firm
size other than profits.

To test their argument Belkaoui & Karpik (1989) use an index of social disclosure developed for
the Ernst and Ernst 1970s surveys of Fortune 500 annual reports, and a reputational ranking of
firms’ social performance based on a survey of business executives. Belkaoui & Karpik (1989)
show these two variables to be significantly correlated (r=0.55), but doubts must remain about the
extent to which actual (as opposed to perceived) social performance relates to social disclosures
(Wiseman, 1982; Deegan and Rankin, 1996). More importantly, however, there is no necessary
reason why social performance (either actual or perceived) and/or social disclosures should be
related to levels of social spending. The Ernst and Ernst index captures the variety of social
programmes a firm claims to be involved in. However, even if it were a true and accurate claim,
why should a firm involved in say ten different types of social responsibility programmes
necessarily be spending relatively more than a firm only involved in one type of activity?11
Likewise, why should having a reputation for doing “good” necessarily be a function of spending
levels? Not only might a reputation for doing good be based on non-cash donations (e.g., staff
time), there is also the prospect that the business executives’ reputational index is based on what
they read about each others’ companies’ social performance as disclosed in annual reports.
Belkaoui & Karpik (1989), then, while remaining consistent with Watts and Zimmerman and
arguing that levels of social spending are a way for self-interested managers to reduce current
period income, fail to test this proposition.

A second aspect of Belkaoui & Karpik’s (1989) study, and indeed most other (social) disclosure
studies, is their failure to take the opportunity to establish a wider basis for their findings. Clearly
the intent of social disclosure studies is to uncover management motives, and, in the case of
positive accounting theory, to uncover evidence of self-interested managers who are using social
expenditures and/or social disclosures to manipulate current reported earnings. As such, then, we
should expect these same managers to be exercising other accounting method choices consistent

11 Ernst and Enrst’s surveys of the 1970s involved analysing the content of the Fortune 500 annual reports against an
index of social responsibility activities (e.g., waste recycling, pollution abatement, training minorities, drug addiction
control). Early surveys identified 14 such items, while the later surveys listed 27 such items. In most cases, however,
companies either failed to disclose the level of expenditures involved or confined such disclosures to specific activities
(Beresford, 1974; Beresford and Feldman, 1976). Beresford (1974, p.43, also see Elias and Epstein, 1975, p. 37), for
example, notes “…disclosure of expenditures for social purposes is still quite small. The number was 73 [out of 500] in
1972 and 53 in 1971. Nearly 90 percent of the expenditure disclosures related to environmental controls, with the
remaining 10 percent relating mainly to charitable contributions. None of the other categories referred to above
included any significant number of companies which reported their socially responsible actions in dollar amounts.”
Beresford also noted at that time “most present disclosures can be characterised as imprecise, verbal rather than
quantitative, selective, non-normative and non-comparative.”

8

---