10 accounting theory questions.

THE ACCOUNTING REVIEW American Accounting Association Vol. 90, No. 1 DOI: 10.2308/accr-50861 2015 pp. 301–320

Unintended Consequences of Lowering Disclosure Thresholds

Kirsten Fanning

University of Illinois at Urbana–Champaign

Christopher P. Agoglia M. David Piercey

University of Massachusetts Amherst

ABSTRACT: In recent years, regulators have considered several initiatives to lower the threshold for disclosing risks to investors. We examine two ways in which disclosing more

risks can actually lower investors’ perceptions of risk. Utilizing an experiment, we find evidence of two unintended consequences on different types of investors. First, we

demonstrate that the addition of low-probability risks to a disclosure can dilute (rather than

add to) more probable losses, leading certain investors to lower their perceptions of overall

risk. Second, since lowering the threshold changes the overall composition of the disclosure

by adding low-probability losses, firms could adopt a tactic of minimization that characterizes

the entire disclosure as unimportant, presenting the lowest risks most saliently, using

compliance with the low threshold as a plausible reason for giving a lengthy disclosure of

generally unimportant risks. Our findings suggest that such a tactic can be persuasive.

Keywords: disclosure thresholds; dilution effect; persuasion tactics; investor judgment.

Data Availability: Contact the authors.

I. INTRODUCTION

I nvestors have long asserted that because disclosure thresholds are too high, firms are often not

disclosing enough information for investors to appropriately allocate their resources (Financial

Accounting Standards Board [FASB] 2008; Desir, Fanning, and Pfeiffer 2010). Echoing

We thank John Harry Evans III (senior editor), Lisa Koonce (editor), two anonymous reviewers, Brooke Elliott, Steve Glover, Jeff Hales, Ling Harris, Kevin Jackson, Scott Jackson, Bill Messier, Mark Peecher, Mike Peters, Ray Pfeiffer, Chad Simon, Jason Smith, Jane Thayer, Brad Tuttle, Scott Vandervelde, David Wood, and Flora Zhou for their helpful comments on this paper. We also thank workshop participants at Drexel University, Georgia Institute of Technology, University of Illinois, University of Massachusetts Amherst, The University of Mississippi, University of Nevada, Las Vegas, University of South Carolina, and Villanova University, as well as participants at the 2010 AAA Accounting, Behavior and Organizations Section Research Conference, the 2010 Brigham Young University Accounting Research Symposium, and the 2011 AAA Annual Meeting for their insights and helpful suggestions. We are grateful to Jeff Hales and Jane Thayer for generously sharing their experimental materials with us.

Editor’s note: Accepted by Lisa Koonce.

Submitted: August 2011 Accepted: June 2014

Published Online: July 2014

301

investor sentiment, regulators have recently considered initiatives designed to lower the threshold

for disclosure of potential losses. For example, regulators have recently proposed or implemented

lower disclosure thresholds for risks related to litigation, derivatives, financing receivables and

credit quality, pensions, liquidity and interest rates, and climate change (e.g., FASB 2010, 2011b,

2012b; Securities and Exchange Commission [SEC] 2010). As (then) FASB Chairman Robert Herz

(2008) noted, lower disclosure thresholds would provide information earlier to existing and

potential investors ‘‘to give them a greater understanding of the risks companies are facing’’ and to ‘‘improve their ability to make informed investment decisions.’’ In contrast, managers and attorneys from many companies are strongly opposed to lowering disclosure thresholds. They argue that

investors will be inundated with a sea of information, which will require investors to wade through

lengthy disclosures of unlikely, and largely unimportant, potential risks to uncover the truly

important ones (Credit Suisse Group 2008; Keating Muething & Klekamp PLL 2008). One reason

that this issue is contentious is because firms argue that increased disclosures will lead to worse

judgments if investors do not incorporate the information appropriately, while investors and

regulators believe that increased disclosures will lead to better judgments, since they will have more

information.

One might normally expect that disclosing more risks of potential losses would increase

investors’ perceptions of disclosed risk. However, this movement toward greater disclosure may

have unintended consequences. We suggest two ways in which disclosing more risks can decrease investors’ perceptions of risk. First, lowering a disclosure threshold naturally means including

additional risks in the disclosure that were previously thought too inconsequential to disclose. If

some investors allow the low-probability risks to dilute the more probable risks in their overall

perceptions, then the requirement to disclose more bad news could actually lead to more positive

assessments of the firm. Second, lowering the threshold changes the overall composition of the

disclosure through the addition of low-probability risks. This provides firms with an opportunity to

design their disclosure strategically to influence investors’ risk perceptions downward, a strategy we

call a ‘‘minimization tactic.’’ By presenting the lowest risks most saliently, such a strategy would effectively characterize the entire disclosure as unimportant, using compliance with the low

threshold as a plausible reason for a lengthy disclosure of generally unimportant risks. Standard

setters have stressed a desire to understand the consequences of their standards on ‘‘all types of investors’’ (FASB 2012a; emphasis in the original), specifying a wide range of investors that includes long, short, and potential investors.

1 Thus, we examine the effect of lowering a disclosure

threshold on a range of investor types. Our theory predicts that the effects will be different

depending on the type of investor.

The debate over the FASB’s recent proposal to lower the threshold for disclosing loss

contingencies provides us with an appropriate setting in which to examine these two unintended

consequences on different types of investors. 2

To test the effects of lowering a disclosure

threshold, we conduct an experiment in which we provide investors with financial statements with

legal disclosures for a hypothetical pharmaceutical company and ask the investors to assess the

likelihood of material loss due to disclosed pending litigation. Participants were randomly

1 Long investors are those who have taken a current investment position in a company. Short investors, or short sellers, are those who have taken an investment position against a company. Potential investors are those who do not have an investment position in a company.

2 Currently, generally accepted accounting principles (GAAP) require that loss contingencies, such as pending litigation, should be disclosed when the probability is at least reasonably possible that a material loss may be incurred. In a 2008 Exposure Draft, the FASB proposed lowering the probability threshold for disclosure of loss contingencies from ‘‘at least reasonably possible’’ to ‘‘more than remote’’ (FASB 2008, paragraph 5) in order to ‘‘expand the population of loss contingencies that are required to be disclosed’’ (FASB 2008, vi ). More recently, the FASB had also considered lowering the disclosure threshold even further to include some remote losses (FASB 2010, 2011a).

302 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

assigned to a long, potential, or short investment position, and to one of three disclosure

conditions. Participants in all conditions received financial information for two firms. Following

Thayer (2011), the underlying economics of the two firms were held constant, and participants in

the long and short investment conditions selected one of two firms in which to take their assigned

investment position. Thayer’s (2011) approach allows participants to actively make an investment

decision, thus bolstering the strength of the manipulation, without firm choice impacting the

results. Adapted from Han and Tan (2010), participants in the potential investor condition

received the same information about the two firms and evaluated one of the two firms, selected at

random.

We manipulate disclosure by presenting potential losses in one of three ways. Specifically,

participants received disclosures with: (1) a higher threshold, (2) a lower threshold, or (3) a lower

threshold with a minimization tactic. All participants learned about the same high-probability loss

contingency involving a lawsuit exceeding the higher threshold. However, because the purpose of

lowering a disclosure threshold is to increase the amount of information disclosed in financial

statements, we provided participants in our lower threshold conditions disclosure information

concerning three additional lower-probability lawsuits that were not disclosed in the higher

threshold condition because they did not meet the higher threshold. In the minimization tactic condition, the firm designs the disclosure to minimize investor perceptions of negative information

in the disclosure. Such a minimization tactic would portray the entire disclosure as a lengthy list of

generally low-risk losses, which are relatively unimportant, but must be disclosed under the low

threshold.

The results of our experiment are consistent with our predictions. First, when the disclosure

threshold is higher, investors assess the firm’s disclosed risks consistent with their directional goals.

That is, long investors assess the lowest and short investors assess the highest likelihood of firm

loss, with potential investors falling in between. Second, we find evidence of a dilution effect for

some investors under the lower threshold. Specifically, because the additional disclosures resulting

from a lower threshold involve lawsuits with low probability of losses, adding these risks actually

decreases some investors’ perceptions of disclosed risk by diluting the influence of the higher- probability lawsuit.

3 Consistent with prior psychology research, we find that this dilution effect is

larger for potential investors than for long or short investors. Finally, we find evidence that a

minimization tactic allowed by the lower threshold can persuade some investors to perceive less

risk. Specifically, when firms adopt the tactic, short investors perceive risk to be as low as long

investors’ already relatively optimistic risk assessments. As Hales, Kuang, and Venkataraman

(2011) note, a disclosure strategy is persuasive when it makes it harder for even short investors to

hold pessimistic beliefs about a company. Our findings suggest that firms can successfully use a low

threshold, which is an attribute of an accounting standard, to persuade investors to see bad news

less negatively.

The findings of our study have a number of important implications. For example, we contribute

to the extant literature by providing new evidence that directional goals can moderate the dilution

effect when individuals encounter post-decisional information. We also find that, in our setting, use

3 Consistent with prior research (Liberman and Ross 2006, 171–173), we focus on participants’ assessments of the disclosed litigation, rather than disclosed and undisclosed litigation, to provide a clean test of the dilution effect. Specifically, normatively, participants’ perceptions of the risk of disclosed litigation should not decrease when we add more risks to the disclosure; instead, such a decrease reflects the dilution effect (Liberman and Ross 2006, 172). At the same time, a similar decrease in undisclosed risk could be for normative reasons. Specifically, when a disclosure threshold is lowered, one would expect undisclosed risk to decrease simply because some previously undisclosed risks are now disclosed. While this decrease in undisclosed risk would go in the same direction as the dilution effect, it would not necessarily reflect the dilution effect. Thus, focusing on disclosed litigation risk assessments allows us to isolate non-normative dilution effects present in participants’ judgments.

Unintended Consequences of Lowering Disclosure Thresholds 303

The Accounting Review January 2015

of a particular minimization persuasion tactic was most effective in influencing short investors, who

actually had the greatest incentive not to be persuaded. Further, our experimental design allows us

to test for effects of accounting standard changes, holding other factors constant (Kachelmeier and

King 2002). Our findings should, therefore, be of interest to investors, as well as to regulators

contemplating/implementing changes to disclosure requirements (e.g., FASB 2010, 2011b, 2012b;

SEC 2010), because we present evidence of potential unintended consequences of lowering

disclosure thresholds. That is, while regulators seek to warn investors of more risks that the firm faces by lowering a disclosure threshold for bad news (FASB 2008; Herz 2008), doing so may

actually lead many investors to perceive less overall risk of potential losses. Investors and regulators should be aware that none of our three investor types (long, potential, or short) perceived

greater overall risk from the lower disclosure threshold. Further, under certain conditions, potential

and short investors perceived less risk.

The next section discusses the background and related research and develops the hypotheses.

Sections III and IV present the method and the results of our experiment. Section V offers

conclusions and implications.

II. THEORY

As we develop our theory and hypotheses, we first consider investors’ judgments under a

reasonably high disclosure threshold, such as the existing Statement of Financial Accounting

Standards (SFAS) No. 5’s ‘‘reasonably possible’’ threshold for the likelihood of contingent losses. Next, we consider the incremental effects of lowering the disclosure threshold to include lower-

probability losses. Finally, we consider the incremental effects of firms adopting a specific

‘‘minimization tactic’’ that lowering the disclosure threshold potentially allows. In each case, we consider this for investors with long, short, and no investment positions.

Investors’ Judgments under a Higher Disclosure Threshold

Prior research has examined how investors with a long, short, or no investment position

evaluate earnings guidance information to form earnings forecasts (Han and Tan 2010; Hales et al.

2011; Thayer 2011). Because long investors have made a decision to invest in a company, they

receive higher payoffs if the company performs relatively well in the future. Short investors, who

have decided to invest against a company, have an economic stake in the company performing

relatively poorly in the future (Hales 2007). These attributes give each type of investor a distinct

directional goal for the way they would like to encounter and interpret information about the future

performance of the company (Thayer 2011). While they have opposing goals, both types of

investors would prefer to see their decisions as justifiable. Thus, individuals’ directional goals tend

to bias how they interpret and assess information, so that they form beliefs that are influenced by

their preferences (Kunda 1990). They generally do not overtly bias their own beliefs in ways that

would be deliberate or obvious to themselves, but rather engage in just enough biased processing to

justify reaching a preferred conclusion (Boiney, Kennedy, and Nye 1997). In contrast, potential

investors have not made a decision to invest in a company and do not have an existing economic

stake in either better or worse performance for that company. These investors do not hold

directional goals and are, therefore, relatively objective, compared to long and short investors (Han

and Tan 2010).

Prior literature suggests that investors holding different positions in a company are influenced

by their respective goals. Prior research examining investors’ forecasts of quarterly earnings finds

that long, short, and potential investors generally form different predictions about future earnings

when evaluating the same earnings guidance information. Specifically, long (short) investors tend

to forecast earnings higher (lower), consistent with what they would like to believe about the

304 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

company’s future performance. Lacking strong directional goals, potential investors’ forecasts

generally tend to fall in between those of long and short investors (Han and Tan 2010; Hales et al.

2011; Thayer 2011). Although forecasting quarterly earnings per share requires a very different

judgment process, we expect a similar effect to occur in response to a company’s disclosure of

risks. For example, in our loss-contingency setting with the pending litigation disclosed by a

company under the higher threshold of SFAS No. 5, we expect that long and short investors’

perceptions of litigation risk will be influenced by their directional goals. This suggests the

following hypothesis:

H1: Under a higher (SFAS No. 5) threshold for disclosure, investors’ perceptions of litigation risk will be lowest for long investors and highest for short investors, with potential

investors’ perceptions in between.

Expanding Disclosure of Risks to Investors: The Debate over Lowering the Threshold for Disclosing Loss Contingencies

The recent proposal to lower the disclosure threshold for loss contingencies (FASB 2008,

2010, 2011a) provides a timely setting to explore the larger debate over the potential effects of

initiatives to expand disclosure of risks to investors. Under the current SFAS No. 5 accounting

standard for disclosure of loss contingencies, management must disclose pending lawsuits if they

believe the likelihood of a material loss is ‘‘at least reasonably possible’’ (FASB 2008). However, investors contend that too often, firms are not disclosing loss contingencies until a loss has

already been realized (Desir et al. 2010). Many investors believe that even when there is a fairly

high likelihood of material loss from an unresolved lawsuit, firms attempt to justify non-

disclosure by asserting that the relatively high ‘‘reasonably possible’’ threshold is not met (cf. Piercey 2009). Investors argue that this leaves them unable to appropriately incorporate

contingencies into their judgments, which could lead to a suboptimal allocation of their resources

(Desir et al. 2010). In response, the FASB issued an Exposure Draft (FASB 2008) that proposes

to lower the probability threshold for disclosure from ‘‘at least reasonably possible’’ to ‘‘more than remote.’’ This would result in the disclosure of previously undisclosed lawsuits that fall between these two thresholds.

4 As the FASB (2008) notes, lowering the probability threshold for

disclosure in this way would require many firms to disclose a greater number of lawsuits. The

FASB (2010, 2011a) has even been considering lowering the threshold further to include some

remote losses.

Many managers and attorneys are strongly opposed to expanding loss contingency disclosures.

A Wall Street Journal editorial noted that: ‘‘Senior litigators from 13 companies, including Pfizer, General Electric, DuPont, Boeing, and McDonald’s have signed a letter to [then] FASB Chairman

Robert Herz, objecting to the plan. ‘Too often, lawsuits are filed for publicity or to pressure

companies, only to be dropped later,’ they wrote’’ (Wall Street Journal 2008, A12). Herz (2008) later responded, ‘‘The new disclosures are aimed at providing information earlier to existing and potential investors in order to give them a greater understanding of the risks companies are facing.

We believe that information would improve their ability to make informed investment decisions.’’ Thus, firms argue that increased disclosures will lead to worse judgments if investors do not

incorporate the information appropriately, while investors believe that the additional disclosure will

lead to better judgments.

4 Simon (2002) reviews prior accounting research that suggests that the change from ‘‘reasonably possible’’ to ‘‘more than remote’’ would change the disclosure threshold from lawsuits with at least an approximately 57 percent probability of loss to those with more than a 12 percent probability of loss.

Unintended Consequences of Lowering Disclosure Thresholds 305

The Accounting Review January 2015

The Effects of a Lower Disclosure Threshold on Investors’ Judgments

Under a lower disclosure threshold such as the one proposed by the FASB (2010), firms would

be required to disclose additional low-probability losses. Many firms opposing the proposed lower

threshold have expressed to the FASB their belief that investors will overreact to additional low-

probability losses made salient under a lower threshold, resulting in an unwarranted increase in

investors’ perceptions of risk (Elsbree and Milne 2009). However, contrary to these expectations,

the dilution effect from psychology would suggest that disclosure of additional low-probability

losses could dilute the influence of a reasonably possible loss, leading some investors to assess risk

lower even though more potential losses are disclosed. Specifically, investors’ perceptions of the overall disclosed litigation risk could be lower due to

the additional low-risk lawsuits if these additional lawsuits dilute the higher-probability losses as

they form their perceptions. Dilution is a basic and widespread cognitive bias that has been

demonstrated across a variety of tasks and contexts (Tetlock and Boettger 1989; Glover 1997;

Liberman and Ross 2006 ). Dilution occurs when the presence of less diagnostic information

weakens the influence of more diagnostic information on individuals’ judgments. For example,

research in consumer judgment settings suggests consumers judge products less favorably when a

moderately positive product attribute is provided along with a highly positive attribute, compared to

when only a highly positive attribute is provided (Jo, Nakamoto, and Nelson 2003; Liberman and

Ross 2006 ). Thus, consumers may naturally tend to average rather than sum the attributes, with the

result that moderate information can dilute the influence of more extreme information.

Similarly, some investors may assess a firm’s risk by ‘‘averaging’’ rather than summing potential losses in a disclosure. For example, under a lower threshold for contingencies, disclosures

would include low-probability lawsuits for investors to average in with, and thereby dilute, a

reasonably possible lawsuit. Thus, as low-probability lawsuits are added to firms’ disclosures, they

may weaken the influence of higher-probability lawsuits on investors’ overall perceptions of

disclosed litigation risk. 5

This would lead investors to assess the risk of loss due to disclosed

litigation lower, even though more lawsuits would be disclosed.

However, there are circumstances under which dilution is less likely. Prior research shows that

the dilution effect is less likely to occur among individuals who are inclined to only incorporate

some of the information into their judgments. For example, this can occur when individuals are told

that they need not use the information carefully (Tetlock and Boettger 1989), when they seek to

save time (Glover 1997), or due to confidence in their expertise (Gilovich and Griffin 2002).

Although investors ordinarily exhibit the dilution effect, prior research suggests that those who

have already taken an investment position become less likely to incorporate further information into

their judgments in order to feel good about that prior decision. Before a decision is made,

individuals are less sensitive to the risk information can pose to a preferred outcome since they have

not yet committed to a decision (Ditto and Lopez 1992). However, after individuals have already

made a decision that they cannot change, they would like to believe that they made a good choice

(Piercey 2009). Once they have satisfactorily justified that decision to themselves, incorporating the

implications of additional information into their judgments runs a risk of threatening their ability to

maintain a self-serving illusion of defensibility for the past decision (Fazio and Williams 1986;

Köneke 2009; Sweeney and Gruber 1984). Under these circumstances, individuals are less likely to

5 Most studies on the dilution effect examine the impact of adding non-diagnostic information to more diagnostic information on individuals’ judgments. Normatively speaking, adding non-diagnostic information should have no impact on individuals’ judgments. In contrast, we examine the impact of adding less diagnostic information (Jo et al. 2003; Liberman and Ross 2006 ). While not irrelevant, lower-probability losses should increase perceptions of disclosed litigation risk less than a reasonably possible loss. Normatively, however, adding low-probability losses to a disclosure should not reduce disclosed litigation risk (Liberman and Ross 2006 ).

306 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

assimilate additional information into their judgments (Taber, Lodge, and Glathar 2001; Lerman,

Croyle, Tercyak, and Hamann 2002). In our setting, long and short investors have already made and

justified the decision to take a long or short investment position in a company. As a result, these

investors would be less likely than potential investors to incorporate additional information into

their judgments and, therefore, would exhibit a smaller dilution effect (Tetlock and Boettger 1989).

Prior research suggests two interesting effects of lowering a threshold on investor judgment.

First, lowering a disclosure threshold to warn investors of additional low-probability losses may

lead some investors to decrease their overall perceptions of disclosed litigation risk, even though more lawsuits are disclosed. Second, this dilution effect would tend to be larger for potential

investors than for long and short investors. This suggests the following hypothesis:

H2: Relative to litigation risk assessments made under a higher disclosure threshold, a lower disclosure threshold will lead potential investors to decrease their litigation risk

assessments more than long and short investors decrease their litigation risk assessments.

A Minimization Tactic for Reporting under a Lower Threshold

Changing to a low disclosure threshold alters the overall composition of firms’ disclosures, and

this presents disclosing firms with an opportunity to use a specific disclosure strategy that we label a

‘‘minimization tactic.’’ Such a strategy would minimize investor perceptions of, and concerns about,

negative information contained in the disclosure. This type of minimization tactic would

characterize the entire portfolio of disclosed losses as generally unimportant, presenting the lower

likelihood losses most saliently, using compliance with the low threshold as the reason for giving

investors a long and tedious disclosure of less important risks. The specific components are integral

to this strategy. 6

For example, firms cannot plausibly characterize the overall disclosure as

unimportant while presenting the highest likelihood losses prominently. The ability to use

compliance with the low threshold standard is also an integral part of this strategy because it gives

firms a plausible reason for providing lengthy disclosures of information that they are characterizing

as unimportant. 7

In contrast, this sort of a minimization tactic is not applicable under a higher threshold, such as

the current SFAS No. 5 rule. First, the current standard’s higher ‘‘reasonably possible’’ threshold

means that firms only disclose lawsuits with a significant likelihood of resulting in a material loss.

Obviously, firms could not plausibly try to represent disclosures containing only ‘‘reasonably

possible’’ losses as unimportant. Second, while firms are not expressly prohibited from voluntarily

disclosing additional lower-probability lawsuits under the current rule, archival evidence indicates

that firms generally choose to disclose as little as they reasonably can (Desir et al. 2010). Further, as

Penman (1980) suggests, firms cannot plausibly call their own voluntary disclosures unimportant

(i.e., if their own disclosures are so unimportant, then why would they voluntarily disclose them?).

Prior research suggests that such a minimization tactic is likely to be effective. Specifically,

communicators can influence others’ opinions by emphasizing particular considerations over other

6 The minimization tactic is a multi-faceted construct consistent with research on specific persuasion strategies. Examining a specific persuasion strategy necessarily requires examining a specific, multi-faceted construct, where the more specific the persuasion strategy tested, the more multi-faceted the construct will be (Perreault and Kida 2011).

7 Research suggests that managers prefer to strategically present themselves and their firms as favorably as possible, and that managers can use features of accounting standards to achieve a strategic and biased portrayal of their company (Hackenbrack and Nelson 1996). For example, Honeywell’s (2010, 3) annual report portrays their position more favorably by using a persuasion strategy that was allowed by a change in the reporting standard with language that clearly appears strategic: ‘‘All of this progress is more obvious within the financial results now that we have rid ourselves of that devil monkey, pension accounting.’’

Unintended Consequences of Lowering Disclosure Thresholds 307

The Accounting Review January 2015

relevant considerations when giving their interpretation of an issue (e.g., Maule and Villejoubert

2007). For example, Shankar and Tan (2006) find that, in an effort to persuade their reviewers,

audit preparers employ language that emphasizes evidence that supports their preferred conclusions

and downplays evidence that contradicts their conclusions.

Hales et al. (2011) illustrate the empirical pattern in which an effective disclosure strategy can

persuade long and short investors differently. They test the effects of using vivid (versus pallid)

language in optimistic earnings guidance on the judgments of long and short investors. Absent the

vivid language, investors’ earnings forecasts reflect their directional goals, with long investors

providing more optimistic forecasts and short investors providing more pessimistic forecasts. Using

the vivid language had no impact on long investors’ forecasts, which remained at their original

optimistic level. However, short investors, who had the greater potential to be persuaded without

the vivid language, converged to long investors’ optimistic judgments when the vivid language was

used, indicating that the persuasion tactic was effective.

In our setting, we expect that the minimization tactic firms can use with a lower disclosure

threshold would have the largest effect on investors with the most potential to be persuaded.

Similar to Hales et al. (2011), it is difficult to say a priori whether long investors would be incrementally influenced by the minimization tactic, since their judgments are likely to be

already relatively optimistic absent this disclosure strategy. It is also difficult to say whether

potential investors would be incrementally influenced by this minimization tactic, since H2

predicts that their judgments would otherwise decrease the most under the lower threshold as

a result of dilution. In contrast, short investors are the most likely to be influenced by this

minimization tactic, since their judgments would otherwise be the least optimistic and have

the greatest potential to be affected. Short investors still have directional goals to be as

pessimistic as they can reasonably justify, but it is possible that persuasion strategies can be so

effective that individuals with opposing directional goals have difficulty finding information to

support their preferred conclusions (Hales et al. 2011). This discussion suggests the following

hypotheses:

H3a: A minimization tactic allowed by a low disclosure threshold will lead short investors to decrease their litigation risk assessments more than long investors decrease their

litigation risk assessments.

H3b: A minimization tactic allowed by a low disclosure threshold will lead short investors to decrease their litigation risk assessments more than potential investors decrease their

litigation risk assessments.

Note that we do not predict a priori whether potential investors would be persuaded more than long investors. The relatively larger dilution effect predicted for potential investors (H2) could

lower their perceptions of litigation risk enough that there would be little opportunity for the

minimization tactic to incrementally lower their perceptions of litigation risk further.

III. METHOD

Participants

One hundred fifty individuals enrolled in M.B.A.-level courses served as our participants. 8

On

average, participants had approximately seven years of professional business experience. Sixty-

eight percent had previously invested in the stock market and 93 percent planned to invest in the

8 The Institutional Review Board Human Subjects Committee granted approval of this study.

308 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

future. Actual investment experience does not influence our participants’ responses (all p � 0.33), consistent with prior research suggesting that graduate business students are good proxies for

individual investors (e.g., Han and Tan 2010; Hales et al. 2011). Intentions to invest also do not

influence investors’ responses (all p � 0.22).

Overview of the Experimental Task and Procedures

Participants completed an experimental task that occurred in two phases. In the first phase of

the experiment, the instrument presented two hypothetical pharmaceutical companies and explained

how participants would be paid. Depending on their experimental condition, participants’ payoffs

either rewarded them for selecting the firm that would be the better performer, rewarded them for

selecting the firm that would be the worse performer, or paid them a flat fee for evaluating one of

the two firms chosen for them at random. After selecting a firm, or having one randomly selected

for them, participants received in the second phase the selected firm’s income statements and

balance sheets for each of the last three years, along with financial statement disclosures regarding

pending litigation. After examining the firm’s financial statements and legal disclosures,

participants assessed litigation risk, the likelihood that the firm would incur a future material loss due to the disclosed litigation, on an 11-point scale, where 0¼ ‘‘material loss not at all likely’’ and 10 ¼ ‘‘material loss very likely.’’

We focus on participants’ assessments of disclosed litigation, as opposed to all potential litigation, in order to provide a clean test of the dilution effect. Specifically, a decrease in

disclosed litigation risk assessments under the lower disclosure threshold demonstrates the dilution effect, because participants’ perceptions of the risk of disclosed litigation should not (normatively) decrease by adding more risks to the disclosure (Liberman and Ross 2006, 172).

However, a similar decrease in undisclosed litigation risk under the lower threshold, while seemingly consistent with the dilution effect, could be for normative reasons. That is, given that

the purpose of lowering this disclosure threshold is to increase the number of lawsuits disclosed,

undisclosed risk could decrease under the lower threshold simply because some of it is now

disclosed. Thus, assessments of all litigation cannot precisely demonstrate non-normative dilution effects present in participants’ judgments. However, assessments of the disclosed litigation can demonstrate this, providing a cleaner test of H2. Finally, after assessing disclosed

litigation risk, participants answered a series of case-related and demographic questions,

including manipulation checks. The post-experimental questionnaire also informed participants

how much money they earned.

Independent Variables

Two independent variables (Investor Position and litigation Disclosure Type) were manipulated between participants, resulting in a 3 3 3 experimental design. We manipulated

Investor Position at three levels: long, short, and potential. For this manipulation, we use two important features of Thayer’s (2011) experimental design. First, Thayer (2011) points out that

real long and short investors make a deliberate choice when picking a company to invest in or

against. Consequently, following Thayer (2011), our long and short investors choose between two companies in which to take a long or short position (see Figure 1), where their payoffs will

depend on this choice. This allows investors to experience making a decision, and this gives them

reason to want to feel good about their past decision. This is significant because our theory in H2

predicts differences in the dilution effect based on investors’ desire to maintain feelings of

justifiability for past decisions. When selecting between the two firms, long (short) investors

knew that they would be paid $15 if their chosen firm outperformed (underperformed) the other

firm, or $5 otherwise (adapted from Thayer [2011]) (see, also, Hales 2007; Han and Tan 2010;

Unintended Consequences of Lowering Disclosure Thresholds 309

The Accounting Review January 2015

Hales et al. 2011). In contrast, potential investors have not yet chosen a firm to either invest in or

against, and do not have directional goals for the firm’s performance. Accordingly, our potential

investors received the same information about the two companies (Figure 1), and were told that

they would receive a flat $10 for evaluating one of the two firms, selected at random (adapted

from Han and Tan [2010]).

In the first phase of the experiment, we provide summary information for each firm, which

included a unique set of three financial statement ratios and four qualitative statements (Figure 1).

Following Thayer (2011), we hold constant the underlying economics of the two firms so that firm

choice would not impact our results. That is, all six financial statement ratios (three for each firm)

were derived from the same underlying set of financial statements. 9

Further, all of the summary

information was pretested on 59 Master’s of Accounting students to ensure that the two firms were

considered similarly attractive ( p¼0.865) and chosen with similar frequency ( p¼0.369). Because the summary information for both firms is based on the same underlying set of financial statements,

the subsequent financial statements and disclosures participants received and evaluated in the

second phase of the experiment were identical regardless of which of the two firms was selected.

During the second phase of the experiment, participants read the financial statements of the firm that

either they selected or was selected for them ( potential investors), and assessed the disclosed

FIGURE 1 Excerpt from Background Experimental Materials

All participants were given this summary information prior to evaluating the selected firm. Long and short investors were asked to pick one of these two firms in which to take their assigned investment position. Potential investors were also given this information, but were not asked to pick a firm. Instead, potential investors were told they would be asked to objectively evaluate one of these two firms (subsequently chosen at random by the computer).

9 Summary information included net income growth rate, net profit margin, and return on equity for one firm and, similarly, revenue growth rate, operating profit margin, and return on assets for the other firm. Thus, the specific ratios for each firm differed, but each firm’s summary information contained one growth, one profit, and one return ratio, respectively.

310 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

litigation risk. Participants’ subsequent perceived litigation risks are independent of the firm

selected (all p � 0.27). The second independent variable, Disclosure Type, was manipulated at three levels and refers

to whether the legal disclosures were in compliance with: (1) a higher threshold (based on the

current SFAS No. 5 threshold); (2) a lower threshold (based on the FASB [2010] proposal) and

presented objectively by the company (lower threshold-objective condition); or (3) a lower

threshold with a minimization tactic (lower threshold-minimization condition). The disclosure in the

higher threshold condition contained one reasonably possible lawsuit. Management introduced the

litigation by stating that the existing accounting standard requires disclosure of lawsuits that

management believes have at least a reasonably possible likelihood of resulting in a material

liability. The two lower threshold conditions each contained the same reasonably possible lawsuit

as the higher threshold condition, as well as three additional lawsuits that are less than reasonably

possible, but more than remote. In all three disclosure-type conditions, each lawsuit was

individually identified as having either a ‘‘reasonably possible’’ or a ‘‘slightly more than remote’’

likelihood of leading to a material loss. Thus, in both lower threshold conditions, objective and

minimization, the financial statements provided the same set of lawsuits and disclosed the same

likelihoods of loss for each.

In the objective condition, management relayed the facts of the pending litigation in a

straightforward and direct manner. Management introduced the litigation by stating that the

accounting standard requires disclosure of lawsuits that management believes have more than a

remote likelihood of resulting in a material liability. Order of presentation was based on likelihood,

with the reasonably possible lawsuit disclosed first, followed by the three lower-probability

lawsuits. Headings were provided for each individual lawsuit to help distinguish one plaintiff from

the next.

In the minimization condition, management adopted the minimization tactic that uses the

lower threshold to minimize investor perceptions and concerns regarding negative information

contained in the disclosure. This minimization tactic characterizes the entire disclosure of

lawsuits as relatively inconsequential by diverting attention from higher-probability lawsuits and

blaming the largely unimportant disclosure on compliance with the standard. 10

Specifically,

management buried the reasonably possible lawsuit in the middle of the disclosure without

individual headings for each lawsuit, while strategically emphasizing compliance with the

standard’s low disclosure threshold as the reason for burdening investors with a lengthy list of

predominantly inconsequential lawsuits. 11

The design of our minimization tactic condition draws

on actual firm comment letters to the FASB regarding the proposed amendment to SFAS No. 5,

actual financial statement disclosures, prior persuasion research, and prior accounting literature

(Credit Suisse Group 2008; Keating Muething & Klekamp PLL 2008; Maule and Villejoubert

2007; Shankar and Tan 2006 ).

10 As noted in the development of H3, because this tactic relies on the lower threshold to provide a plausible reason for a lengthy disclosure of information characterized as unimportant, it cannot be utilized under the higher threshold. Thus, it is not fully crossed within the higher threshold condition. Consequently, we use a three-level factor to test two unintended consequences of lowering a disclosure threshold (see Piercey [2009] for a similar three-level design).

11 A pretest with 29 senior accounting undergraduates suggests that when compared side-by-side, the disclosure in the minimization condition is seen as more strategic than the disclosure in the objective condition (means ¼ 8.17 versus 4.48, t ¼ 7.97, on a scale where 0 ¼ ‘‘not at all strategic’’ and 10 ¼ ‘‘very strategic’’; p , 0.001). In addition, the disclosure in the minimization condition is viewed as less objective than the disclosure in the objective condition (means¼3.82 versus 8.04, t¼9.03, where 0¼ ‘‘not at all objective’’ and 10¼ ‘‘very objective’’; p , 0.001). Further, all pretest participants also selected the minimization condition disclosure as the presentation that is the less objective and more opportunistic of the two.

Unintended Consequences of Lowering Disclosure Thresholds 311

The Accounting Review January 2015

IV. RESULTS

Manipulation Checks

Responses to manipulation check questions suggest that the manipulations were successful.

Participants correctly identified their respective compensation scheme based on their investment

position 99 percent of the time. Participants correctly recalled the number of lawsuits disclosed 89

percent of the time, and the probability threshold 89 percent of the time. Our conclusions are

unchanged if we exclude participants who failed the manipulation checks from the analyses.

Tests of Hypotheses

Table 1 presents the descriptive statistics, overall ANCOVA results, and tests of hypotheses for

participants’ litigation risk assessments. The ANCOVA includes participants’ prior beliefs about industry litigiousness as a covariate.

12 The Investor Position 3 Disclosure Type interaction (F ¼

2.37, p ¼ 0.055, Table 1, Panel B) is displayed graphically in Figure 2. H1 predicts that long investors’ perceptions of litigation risk under the higher threshold will be

lowest, short investors’ perceptions will be highest, and potential investors’ perceptions will be in

between. As Figure 2 and Table 1 show, investors’ average litigation risk assessments are 4.92,

6.73, and 7.18 for long, potential, and short investors, respectively. A Jonckheere-Terpstra test

confirms the predicted ordering of investor judgments under the higher threshold (z ¼ 2.96, p ¼ 0.002, Table 1, Panel C).

13 This result supports H1.

We supplement our test of H1 by conducting pairwise contrasts within the predicted ordering

of investor judgments. As predicted, long investors’ risk assessments are significantly lower than

short (t¼�3.33, p , 0.001) and potential investors’ assessments (t¼�2.80, p¼0.003). However, short investors’ assessments are only directionally higher than potential investors’ assessments (t¼ 0.64, p¼0.260). While the predicted ordering of these three means is statistically significant, prior research suggests reasons why potential investors’ judgments would be more likely to differ more

from long investors’ judgments than from short investors’ judgments. Specifically, while neutral

individuals do not have directional goals, features of the task can move their judgments closer to

one of the two groups with opposing directional goals (Piercey 2009). Neutral individuals attempt

to adjust their judgments when receiving information from a non-neutral source (Wegener and Petty

1995). In our setting, potential investors are aware of the reporting company’s directional incentives

and are, therefore, likely to move closer to short investors’ more pessimistic judgments. 14

When we

use ordered contrast weights that reflect potential investors’ possible tendency to err on the side of

pessimism (�1 for long investors, þ0.333 for potential investors, and þ0.667 for short investors), we obtain results similar to the Jonckheere-Terpstra test (t¼3.62, p , 0.001). Overall, our primary test of H1 in Table 1, Panel C supports H1. Our supplemental tests also provide additional support

for H1 and are consistent with prior research.

H2 predicts that the dilution effect will be larger for potential investors than for long or short

investors. In our setting, this refers to a tendency for the low-probability losses added under a low

12 Individuals who believe the industry to be more litigious tend to assess litigation risk higher across experimental conditions. As such, we include litigiousness as a covariate (F ¼ 10.22, p ¼ 0.002, Table 1, Panel B). We obtain statistically similar results at conventional significance levels and reach the same conclusions if the covariate is removed from the model.

13 This Jonckheere-Terpstra test is based on the ANCOVA model, including the effects of the covariate (Chow and Liu 1998), and is used to test hypotheses of the type A , B , C.

14 There still may be a difference between potential and short investors’ judgments, however, because short investors can make a similar adjustment for the company’s potential optimism and have additional directional goals to evaluate the company pessimistically.

312 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

TABLE 1

Litigation Risk Assessments

Panel A: Descriptive Statisticsa

Disclosure Type

Investor Position Higher

Threshold Lower-

Objective Lower-

Minimization

Short investors 7.18 7.43 4.76 (1.54) (1.05) (2.27)

[16] [12] [17]

Potential investors 6.73 5.54 4.93 (2.15) (1.68) (2.49)

[19] [18] [16]

Long investors 4.92 4.95 4.75 (2.37) (1.89) (2.07)

[21] [16] [15]

Panel B: Analysis of Covariance

Source Sum of Squares df

Mean Square F p-value

Industry Litigiousness (covariate) 42.36 1 42.36 10.22 0.002 Investor Position 59.32 2 29.66 7.16 0.001 Disclosure Type 59.55 2 29.78 7.18 0.001 Investor Position 3 Disclosure Type 39.37 4 9.84 2.37 0.055 Error 580.34 140 4.15

Panel C: Tests of Hypotheses

Test Statistic p-valueb

H1: Within the higher threshold disclosure: Short investors’ assessments . potential investors’ . long investors’ 2.96 0.002

H2: Higher threshold versus lower threshold-objective disclosure: Potential investors’ assessments decrease more than long and short investors’ �1.57 0.059

H3: Lower threshold-objective versus minimization disclosure: Short investors’ assessments decrease more than long investors’ (H3a) �2.32 0.011 Short investors’ assessments decrease more than potential investors’ (H3b) �1.98 0.025

a Means, (standard deviations), and [n] are presented. Means are the adjusted means from the ANCOVA, which includes

the covariate Industry Litigiousness. Participants provided their perceptions about the industry’s litigiousness on an 11- point scale, where 0 ¼ ‘‘not at all litigious’’ and 10 ¼ ‘‘very litigious.’’ b

All hypotheses were tested within the ANCOVA. Given the directional expectations suggested by our theory and hypotheses, these tests are one-tailed. A Jonckheere-Terpstra test that is based on the ANCOVA and includes the effects of the covariate Industry Litigiousness (Chow and Liu 1998) is used to test H1. H2 is tested with a contrast using weights of�1 andþ1 for potential investors in the higher threshold and lower threshold-objective conditions, respectively,þ0.5 and�0.5 for long investors, andþ0.5 and�0.5 for short investors. H3a (H3b) is tested with a contrast using weights of �1 and þ1 for short investors in the lower threshold-objective conditions and lower threshold-minimization conditions, respectively,þ1 (�1) for long ( potential) investors, and 0 (0) for potential (long) investors. H2 and H3 are tested with t- tests.

(continued on next page)

Unintended Consequences of Lowering Disclosure Thresholds 313

The Accounting Review January 2015

disclosure threshold to reduce, rather than increase, investors’ overall perceptions of disclosed

litigation risk. As shown in Figure 2, potential investors’ judgments appear to decrease more from

the higher threshold to the lower threshold-objective condition than do long and short investors’

judgments. Specifically, we compared the change in potential investors’ litigation risk assessments

to the average change among long and short investors’ assessments, using contrast weights of �1 and þ1 for potential investors in the higher threshold and lower threshold-objective conditions, respectively,þ0.5 and�0.5 for long investors, andþ0.5 and�0.5 for short investors. The decrease in potential investors’ litigation risk assessments is significantly larger (t¼�1.57, p¼0.059, Table

TABLE 1 (continued)

Dependent Variable: Litigation Risk Assessments. Participants assessed the likelihood that the firm will incur a future material loss due to the disclosed litigation. Participants recorded their responses on an 11-point scale, where 0 ¼ ‘‘material loss not at all likely’’ and 10 ¼ ‘‘material loss very likely.’’ Independent Variables: Investor Position is manipulated as participants’ position in the firm (long, short, or potential investor). Disclosure Type manipulated whether participants received pending litigation disclosures in compliance with: (1) a higher threshold (i.e., ‘‘at least reasonably possible’’); (2) a lower threshold (i.e., ‘‘more than remote’’) presented objectively; or (3) a lower threshold (i.e., ‘‘more than remote’’) with a minimization tactic.

FIGURE 2 Litigation Risk Assessments

Dependent variable (vertical axis): Litigation Risk Assessments. Participants assessed the likelihood that the firm will incur a future material loss due to the disclosed litigation. Participants recorded their responses on an 11-point scale, where 0 ¼ ‘‘material loss not at all likely’’ and 10 ¼ ‘‘material loss very likely.’’ All means reported above are adjusted means from the ANCOVA presented in Table 1, which includes the covariate Industry Litigiousness. Participants provided their perceptions about the industry’s litigiousness on an 11-point scale, where 0¼ ‘‘not at all litigious’’ and 10 ¼ ‘‘very litigious.’’ Independent variables: Investor Position is manipulated as participants’ position in the firm (long, short, or potential investor). Disclosure Type manipulated whether participants received pending litigation disclosures in compliance with: (1) a higher threshold (i.e., ‘‘at least reasonably possible’’); (2) a lower threshold (i.e., ‘‘more than remote’’) presented objectively; or (3) a lower threshold (i.e., ‘‘more than remote’’) with a minimization tactic.

314 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

1, Panel C) than the concurrent change in long and short investors’ assessments. 15

This result

supports H2.

Simple effects tests find that potential investors exhibited a significant dilution effect, while

long and short investors did not. Specifically, potential investors’ litigation risk assessments are

significantly lower in the lower threshold-objective condition than in the higher threshold condition (5.54 versus 6.73, t ¼�1.77, p ¼ 0.040), even though the lower threshold-objective condition included the same reasonably possible lawsuit as the higher threshold condition, plus three additional lawsuits. Thus, the three low-probability lawsuits diluted, rather than added to, the

impact of the reasonably possible lawsuit in potential investors’ perceptions of disclosed litigation

risk. In contrast, there are no significant differences between the lower threshold-objective condition and the higher threshold condition for short investors (7.43 versus 7.18, t¼0.32, p¼0.748) or for long investors (4.95 versus 4.92, t ¼ 0.03, p ¼ 0.969). These findings are also consistent with H2.

H3 predicts that the minimization tactic allowed by a low disclosure threshold will lead short

investors to decrease their litigation risk assessments more than either long (H3a) or potential

investors (H3b). Results are consistent with our theory and expectations. Specifically, short investors’

litigation risk assessments are significantly lower in the lower threshold-minimization condition than in the lower threshold-objective condition, suggesting that the minimization tactic persuaded short investors (4.76 versus 7.43, t ¼�3.47, p , 0.001). In contrast, there are no significant differences between the minimization and objective conditions for potential investors (4.93 versus 5.54, t¼�0.87, p ¼ 0.386) or for long investors (4.75 versus 4.95, t ¼�0.28, p ¼ 0.779). Our formal tests of H3 indicate that the effect of the minimization tactic strategy is significantly larger for short investors than

for either long investors (t¼�2.32, p¼0.011, Table 1, Panel C) or for potential investors (t¼�1.98, p ¼ 0.025, Table 1, Panel C).16 These results support H3a and H3b.

For completeness across all nine cells, we performed additional Jonckheere-Terpstra tests of

Investor Position within the two lower threshold conditions, in addition to our earlier Jonckheere- Terpstra test of H1 within the higher threshold condition. Figure 2 shows that investors’ judgments are also influenced by long, potential, and short Investor Position within the objective condition (4.95, 5.54, and 7.43, respectively; z ¼ 3.39, p , 0.001). However, Figure 2 also shows that the judgments of long, potential, and short investors converge toward the relatively optimistic

judgments of long investors within the minimization condition (4.75, 4.93, and 4.76, respectively; z ¼0.11, p¼0.91). This pattern of results is consistent with our overall theory and our primary tests of H2 and H3.

Supplementary Analysis: Future Earnings Perceptions

Because our theory relates to how our manipulations of the litigation disclosures would

influence investors’ perceptions of disclosed litigation risk, we use investors’ perceptions of

15 We combine long and short investors as both have directional goals, and the theory underlying H2 generates similar predictions for both with respect to the dilution effect. Accordingly, H2 predicts that the dilution effect will be larger for potential investors than for both long and short investors. However, individual comparisons of potential versus short and potential versus long investors produce results that are directionally consistent with H2, although less significant ( p ¼ 0.081 and p ¼ 0.102, respectively).

16 Unlike our test for H2, in our tests of H3a and H3b, we do not use a combined test comparing the effect of the minimization tactic on short investors to its average effect on potential and long investors. The theory underlying H3 uses different reasons for predicting that long investors and potential investors will each react less to the minimization tactic than will short investors. The distinction involves long investors’ relative optimism before the effects of the minimization tactic versus potential investors’ larger dilution effect before the effects of the minimization tactic. Accordingly, H3 predicts that short investors’ judgments will decrease more than either long or potential investors’ judgments, and we separate each test. As noted in footnote 15, H2 combines long and short investors because the prediction for H2 is based on both groups having a directional goal, which is not the case for the H3a and H3b comparisons.

Unintended Consequences of Lowering Disclosure Thresholds 315

The Accounting Review January 2015

‘‘material loss due to the disclosed litigation’’ as our dependent variable for testing H1 through H3.

However, these perceptions of litigation risk could also influence subsequent investment-relevant

judgments. Thus, participants also assessed the extent to which the firm’s earnings in future periods

would be affected by the disclosed litigation. Participants recorded their responses on an 11-point

scale, where 0 ¼ ‘‘future earnings not affected at all’’ and 10 ¼ ‘‘future earnings affected greatly.’’ Table 2 presents cell means and ANCOVA results for the future earnings perceptions variable.

Although future earnings perceptions are likely to impound other noisy sources of variance because

they are a less direct operational measure of our theoretical construct than litigation risk

assessments, they should be positively correlated with investors’ litigation risk assessments. As

expected, we find future earnings perceptions to be significantly correlated ( p , 0.001) with

litigation risk assessments.

Further, the H1 predicted ordering of participants’ litigation risk assessments under the higher threshold (short . potential . long investors) also holds for their perceptions of the effect of the disclosed litigation on future earnings (Jonckheere-Terpstra z¼3.37, p , 0.001, Table 2, Panel C). Similar to our H2 litigation risk assessment results, we find that potential investors exhibit a significant dilution effect for future earnings perceptions (t ¼�3.11, p ¼ 0.001), while long and short investors do not (t¼�1.55, p¼0.12, and t¼1.31, p¼0.19, respectively). More specifically, we find that the dilution effect in potential investors’ future earnings perceptions is significantly greater than the concurrent (lack of ) change in long and short investors’ judgments (t¼�2.66, p¼ 0.004, Table 2, Panel C). Again, individual comparisons of potential versus short and potential versus long investors produce results that are directionally consistent ( p ¼ 0.001 and p ¼ 0.104, respectively). Finally, as with our H3a and H3b litigation risk assessment results, we find that the

persuasive effect of the minimization tactic on future earnings perceptions is greater for short investors than for either long (t ¼�3.75, p , 0.001, Table 2, Panel C) or potential investors (t ¼ �2.84, p ¼ 0.003, Table 2, Panel C). Thus, results with the future earnings variable are consistent with those obtained with the litigation risk variable, and are consistent with and supportive of our

theory and hypotheses.

V. CONCLUSION

With the recent discussion surrounding changes to disclosure requirements, our study is

designed to provide insight to parties interested in or affected by such changes. For example, our

experimental design allows us to test for effects of accounting standard changes, holding other

factors constant, as called for by Kachelmeier and King (2002). Our findings should be of interest to

investors, as well as to regulators contemplating or implementing changes to disclosure

requirements (e.g., FASB 2010, 2011b, 2012b; SEC 2010), because we present evidence of

potential unintended consequences of lowering disclosure thresholds. Contrary to the FASB’s

(2008) stated objective of expanding disclosure of bad news to warn investors of more risks that the firm faces, investors and regulators should be aware that for no investor type (long, potential, or

short) did lowering a disclosure threshold increase perceptions of overall risk, and potential and short investors’ perceptions of risk actually decreased. For researchers, we provide new evidence that directional goals can moderate the dilution effect when individuals encounter post-decisional

information. Further, we find that in our setting, use of a particular minimization persuasion tactic

was most effective on short investors who had the greatest incentive not to be persuaded.

Limitations of our study include the fact that we do not test all possible disclosure strategies

that a firm could pursue, including strategies that incorporate some, but not all, elements of the

minimization tactic. Because of our motivation to test for a specific unintended consequence of

lowering disclosure thresholds, our intention was to explore the implications of a specific, multi-

faceted persuasion strategy that is made available by the proposed lower disclosure threshold and

316 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

TABLE 2

Supplementary Analysis: Future Earnings Perceptions

Panel A: Descriptive Statisticsa

Disclosure Type

Investor Position Higher

Threshold Lower-

Objective Lower-

Minimization

Short investors 6.74 7.87 5.33 (1.78) (0.94) (2.33)

[16] [12] [17]

Potential investors 5.97 4.00 4.23 (1.47) (1.75) (2.50)

[17] [18] [16]

Long investors 4.64 3.81 5.01 (1.81) (1.83) (2.17)

[21] [16] [15]

Panel B: Analysis of Covariance

Source Sum of Squares df

Mean Square F p-value

Industry Litigiousness (covariate) 27.65 1 27.65 7.58 0.007 Investor Position 129.14 2 64.57 17.71 ,0.001 Disclosure Type 22.59 2 11.29 3.10 0.048 Investor Position 3 Disclosure Type 74.76 4 18.69 5.13 0.001 Error 510.56 140 3.65

Panel C: Supplemental Tests

Test Statistic p-valueb

Within the higher threshold disclosure: Short investors’ assessments . potential investors’ . long investors’ (Test 1) 3.37 ,0.001

Higher threshold versus lower threshold-objective disclosure: Potential investors’ assessments decrease more than long and short investors’ (Test 2) �2.66 0.004

Lower threshold-objective versus minimization disclosure: Short investors’ assessments decrease more than long investors’ (Test 3a) �3.75 ,0.001 Short investors’ assessments decrease more than potential investors’ (Test 3b) �2.84 0.003

a Means, (standard deviations), and [n] are presented. Means are the adjusted means from the ANCOVA, which includes

the covariate Industry Litigiousness. Participants provided their perceptions about the industry’s litigiousness on an 11- point scale, where 0 ¼ ‘‘not at all litigious’’ and 10 ¼ ‘‘very litigious.’’ b

All tests were conducted within the ANCOVA. Given the directional expectations suggested by our theory, these tests are one-tailed. A Jonckheere-Terpstra test that is based on the ANCOVA and includes the effects of the covariate Industry Litigiousness (Chow and Liu 1998) is used for Test 1. Test 2 uses a contrast with weights of �1 and þ1 for potential investors in the higher threshold and lower threshold-objective conditions, respectively,þ0.5 and�0.5 for long investors, andþ0.5 and�0.5 for short investors Test 3a (3b) uses contrast weights of�1 andþ1 for short investors in the lower threshold-objective conditions and lower threshold-minimization conditions, respectively, þ1 (�1) for long ( potential) investors, and 0 (0) for potential (long) investors. Tests 2 and 3 are tested with t-tests.

(continued on next page)

Unintended Consequences of Lowering Disclosure Thresholds 317

The Accounting Review January 2015

that our theory suggests would lead to this unintended consequence (Perreault and Kida 2011).

Thus, we tested a complete implementation of the strategy and compared it to another disclosure

condition that is devoid of the strategy, presenting the risks as objectively as possible. Our approach

follows Kerlinger and Lee’s (2000, 459) principle to ‘‘design, plan, and conduct research so that the

experimental conditions are as different as possible’’ along the theoretical construct of interest. Still,

future research could separately examine the facets of our minimization tactic in order to isolate the

extent to which the persuasion effect that we document is driven by a particular facet or by their

joint use. Future research could also examine other specific reporting strategies in our setting,

including those that do and do not overlap with the minimization tactic.

REFERENCES

Boiney, L. G., J. Kennedy, and P. Nye. 1997. Instrumental bias in motivated reasoning: More when more is

needed. Organizational Behavior and Human Decision Processes 72 (1): 1–24.

Chow, S. C., and J. P. Liu. 1998. Design and Analysis of Animal Studies in Pharmaceutical Development.

New York, NY: Marcel Dekker, Inc.

Credit Suisse Group. 2008. Comment Letter 100 on Exposure Draft: Disclosure of Certain Loss

Contingencies: An Amendment of FASB Statements No. 5 and 141(R). Norwalk, CT: Financial

Accounting Standards Board.

Desir, R., K. Fanning, and R. Pfeiffer. 2010. Are revisions to SFAS No. 5 needed? Accounting Horizons 24

(4): 525–545.

Ditto, P. H., and D. Lopez. 1992. Motivated skepticism: Use of differential decision criteria for preferred

and nonpreferred conclusions. Journal of Personality and Social Psychology 63 (4): 568–584.

Elsbree, D., and J. Milne. 2009. Disclosures of Certain Loss Contingencies Final Comment Letter Summary

Memorandum. (August 6). Norwalk, CT: Financial Accounting Standards Board.

Fazio, R. H., and C. J. Williams. 1986. Attitude accessibility as a moderator of the attitude-perception and

attitude-behavior relations: An investigation of the 1984 presidential election. Journal of Personality

and Social Psychology 51 (3): 505–514.

Financial Accounting Standards Board (FASB). 2008. Exposure Draft: Disclosure of Certain Loss

Contingencies: An Amendment of FASB Statements No. 5 and 141(R). Norwalk, CT: FASB.

Financial Accounting Standards Board (FASB). 2010. Proposed Accounting Standards Update:

Contingencies (Topic 450): Disclosure of Certain Loss Contingencies. Norwalk, CT: FASB.

Financial Accounting Standards Board (FASB). 2011a. Current Technical Plan and Project Updates.

Norwalk, CT: FASB.

Financial Accounting Standards Board (FASB). 2011b. FASB Improves Employer Disclosures for

Multiemployer Pension Plans. Norwalk, CT: FASB.

Financial Accounting Standards Board (FASB). 2012a. Investors—Your Views Are Critical in Helping Us

Improve Financial Reporting. Norwalk, CT: FASB.

Financial Accounting Standards Board (FASB). 2012b. Proposed Accounting Standards Update: Financial

Instruments (Topic 825): Disclosures about Liquidity and Interest Rate Risk. Norwalk, CT: FASB.

TABLE 2 (continued)

Dependent Variable: Future Earnings Perceptions. Participants assessed the extent to which the firm’s earnings in future periods would be affected by the disclosed litigation. Participants recorded their responses on an 11-point scale, where 0 ¼ ‘‘future earnings not affected at all’’ and 10 ¼ ‘‘future earnings affected greatly.’’ Independent variables: Investor Position is manipulated as participants’ position in the firm (long, short, or potential investor). Disclosure Type manipulated whether participants received pending litigation disclosures in compliance with: (1) a higher threshold (i.e., ‘‘at least reasonably possible’’); (2) a lower threshold (i.e., ‘‘more than remote’’) presented objectively; or (3) a lower threshold (i.e., ‘‘more than remote’’) with a minimization tactic.

318 Fanning, Agoglia, and Piercey

The Accounting Review January 2015

Gilovich, T., and D. Griffin. 2002. Introduction—Heuristics and biases: Then and now. In The Psychology of Intuitive Judgment, edited by Gilovich, T., D. Griffin, and D. Kahneman, 1–18. Cambridge, U.K.: Cambridge University Press.

Glover, S. M. 1997. The influence of time pressure and accountability on auditors’ processing of

nondiagnostic information. Journal of Accounting Research 35 (2): 213–226.

Hackenbrack, K., and M. W. Nelson. 1996. Auditors’ incentives and their application of financial

accounting standards. The Accounting Review 71 (1): 43–59.

Hales, J. 2007. Directional preferences, information processing, and investors’ forecasts of earnings.

Journal of Accounting Research 45 (3): 607–628.

Hales, J., X. J. Kuang, and S. Venkataraman. 2011. Who believes the hype? An experimental examination

of how language affects investor judgments. Journal of Accounting Research 49 (1): 223–255.

Han, J., and H.-T. Tan. 2010. Investors’ reactions to management earnings guidance: The joint effect of

investment position, news valence, and guidance form. Journal of Accounting Research 48 (1): 81– 104.

Herz, R. H. 2008. FASB seeks to inform investors, not whack companies (letter to the editor). Wall Street Journal (August 18): A4.

Honeywell. 2010. Honeywell 2010 Annual Report. Morristown, NJ: Honeywell International, Inc.

Jo, M. S., K. Nakamoto, and J. E. Nelson. 2003. The shielding effects of brand image against lower quality

countries-of-origin in global manufacturing. Journal of Business Research 56 (8): 637–646.

Kachelmeier, S. J., and R. R. King. 2002. Using laboratory experiments to evaluate accounting policy

issues. Accounting Horizons 16 (3): 219–232.

Keating Muething & Klekamp PLL. 2008. Comment Letter 20 on Exposure Draft: Disclosure of Certain Loss Contingencies: An Amendment of FASB Statements No. 5 and 141(R). Norwalk, CT: FASB.

Kerlinger, F. N., and H. B. Lee. 2000. Foundations of Behavioral Research. 4th edition. New York, NY: Thomson Learning.

Köneke, V. 2009. Psychology of Self-Insight: Motivated Reasoning and Self-Deception. Norderstedt, Germany: Druck and Bindung.

Kunda, Z. 1990. The case for motivated reasoning. Psychological Bulletin 108 (3): 480–498.

Lerman, C., R. T. Croyle, K. P. Tercyak, and H. Hamann. 2002. Genetic testing: Psychological aspects and

implications. Journal of Consulting and Clinical Psychology 70: 784–797.

Liberman, V., and L. Ross. 2006. Idiosyncratic matching and choice: When less is more. Organizational Behavior and Human Decision Processes 101 (2): 168–183.

Maule, J., and G. Villejoubert. 2007. What lies beneath: Reframing framing effects. Thinking and Reasoning 13 (1): 25–44.

Penman, S. H. 1980. An empirical investigation of the voluntary disclosure of corporate earnings forecasts.

Journal of Accounting Research 18 (1): 132–160.

Perreault, S., and T. Kida. 2011. The relative effectiveness of persuasion tactics in auditor-client

negotiations. Accounting, Organizations and Society 36 (8): 534–547.

Piercey, M. D. 2009. Motivated reasoning and verbal vs. numerical probability assessment: Evidence from

an accounting context. Organizational Behavior and Human Decision Processes 108 (2): 330–341.

Shankar, P. G., and H-T. Tan. 2006. Determinants of audit preparers’ workpaper justifications. The Accounting Review 81 (2): 473–495.

Simon, J. 2002. Interpretation of probability expressions by financial directors and auditors of UK

companies. The European Accounting Review 11 (3): 601–629.

Sweeney, P. D., and K. L. Gruber. 1984. Selective exposure: Voter information preferences and the

Watergate affair. Journal of Personality and Social Psychology 46 (6 ): 1208–1221.

Taber, C. S., M. Lodge, and J. Glathar. 2001. The motivated construction of political judgments. In Citizens and Politics: Perspectives from Political Psychology, edited by Kuklinski, J. H., 198–226. New York, NY: Cambridge University Press.

Tetlock, P. E., and R. Boettger. 1989. Accountability: A social magnifier of the dilution effect. Journal of Personality and Social Psychology 57 (3): 388–398.

Unintended Consequences of Lowering Disclosure Thresholds 319

The Accounting Review January 2015

Thayer, J. 2011. Determinants of investors’ information search: Credibility and confirmation. The Accounting Review 86 (1): 1–22.

U.S. Securities and Exchange Commission (SEC). 2010. Commission Guidance Regarding Disclosure Related to Climate Change. Washington, DC: SEC.

Wall Street Journal. 2008. FASB’s lawyer bonanza. (August 7): A12. Wegener, D. T., and R. E. Petty. 1995. Flexible correction processes in social judgment: The role of naı̈ve

theories in corrections for perceived bias. Journal of Personality and Social Psychology 68: 36–51.

320 Fanning, Agoglia, and Piercey

The Accounting Review January 2015