Impression management in corporate press releases: an international perspective

(1)

Impression management in

corporate press releases: an

international perspective

Master thesis 2013

Author: Joram Prinsen 1736175 Coordinator: Prof. mr. drs. J.Th. Degenkamp

Managers are assumed to act as rational, self-interested entities, adopting an impression management strategy to pursue utility maximization. Using a specific writing style is supposed to contribute to this strategy for the purpose of influencing the readers‟ perception of corporate narrative disclosures. The aim of this thesis was to explore the relationship between the firms‟ performance and the adopted writing style in press releases regarding the final year results from European listed companies. In addition this research investigated the role of the legal system in which the firm operates, comparing writing style under the common law and the civil law legal system. The LIWC analysis software was used to derive the dependent variables of writing style: positive emotion words, negative emotion words, complexity and certainty. The results did provide support for the relationship between a low firm performance and a writing style signifying for more negativism, less positivism and a more complex writing style. Thus, under poor financial year results, management attempts to obfuscate unfavourable outcomes by striving to a higher text complexity. In addition, firms operating under a common law legal system adopt a writing style using more positive and negative emotion words in their press release.

(2)

1. Introduction

Daniel Kahnemann, who won the Nobel Prize for Economics in 2002, and his colleague Amos Tversky, studied in the eighties the perception of decision problem outcomes and the shift of preference when the same problem is framed in a different way. The respondents had to choose between two strategies to combat a disease, which is expected to kill 600 people. The group of respondents was asked to choose either one of the following options: A: 200 people will be saved; or B: there is 1/3 probability that 600 people will be saved and 2/3 probability that nobody will be saved. Although it is expected that 200 people will survive in both scenarios, the majority of the respondents chose option A. Then, a second group of respondents was submitted to the same problem with an alternative formulation of the options: A: 400 people will die; or B: there is 1/3 probability that nobody will die, and 2/3 probability that 600 people will die. Now, the majority chose option B. However the problems are identical, the change from the described outcomes in terms of „people saved‟ in group 1, to „people die‟ in group 2, is accompanied by an evident shift in decision making. Apparently, writing style is of significant relevance in “framing” texts and therefore in influencing the readers‟ perception of the text (Kahnemann & Tversky, 1981).

1.1 Definitions

Derived from psychology and social sciences, “framing” is regarded as the judgment of choices and opinions, which is of “significant rhetorical interest in building support, convincing investors, and sustaining confidence” (Ginzel, Kramer, & Sutton, 1993). These cornerstones of framing are in accordance with, and form part of the general aspects of the corporate governance structure of an organization. Broadly defined, “corporate governance consists of the institutional structures, legal rules, and best practices that determine which body within the corporation is empowered to make particular decisions” (Bainbridge, 2011). Subsequently, the corporate governance structures and rules within the organization determine how the members of that particular body are chosen and the norms that should guide decision-making. This broadly defined definition points out corporate governance principles derive from divergent sources and take many forms. Some principles derive from rules of best practice or from social norms, while others derive from the hard law of corporation statutes and judicial opinions enforced by legal sanctions. In addition, corporate governance derives from the principle of the framing of textual outcomes. Taken together,

(4)

these divergent sources collectively specify the rules and procedures in order to give direction for the decision making process (Bainbridge, 2011).

In the context of these principals of corporate governance, management may contribute to achieve the firms‟ objectives by “framing” organizational outcomes. Therefore, management may adopt an impression management strategy, which consists largely of emphasizing on favorable organizational outcomes and/or obfuscating unfavorable organizational outcomes (Merkl-Davies, Brennan, & McLeay, 2011).

Knowing this, a link could be established between on the one hand the respondents of the experiment conducted by Kahnemann and Tversky (1981), and organizational stakeholders in general on the other hand. The respondents of the respective experiment shifted their preference of outcomes and perception of the problem due to a different writing style of the presented problem. The link can be established now with organizational investors, who base their investor decisions and their perception of the company partially on the corporate disclosures, like press releases. In another word: investors base their perception of the company on corporate disclosures, which may be subject to managers‟ impression management strategy. This study will examine the use of a specific writing style as a tool for impression management in corporate press releases from European listed companies, thereby making a comparison between “common law” and “civil law” countries.

Impression management is generally regarded as the process in which one attempts to control and shape the impressions that others form of the entity (Schlenker, 1980). This study includes a theoretical and empirical study of corporate impression management thereby focusing on different elements of the adopted writing style of corporate press releases. Following scholars like Abrahamson & Park (1994); Cho, Roberts, & Patten (2010); Courtis (1995); Merkl-Davies, Brennan, & McLeay (2011), and Rogers, Von Buskirk, & Zechman, (2011); this study will focus on writing styles of press releases as a tool for managing stakeholder impressions.

This study will have an international perspective, with a distinction being drawn between common law countries and civil law countries. The main research question is:

(5)

To what extent does firm performance explains differences in writing styles between firms and to what extent do we observe different writing styles between divergent legal systems?

1.2 Relevance

Prior research on impression management already emphasizes on the use of writing style, language and verbal tone as a tool for managing stakeholders impressions. However, in the existing literature there are still opportunities to contribute to the literature regarding impression management in the context of corporate reporting. This study can contribute to the existing literature for a number of reasons:

First, although prior research established on the use of language and verbal tone in environmental disclosures (Cho, Roberts, & Patten, 2010), the readability of the „Management Discussion and Analysis‟ (Li, 2008), the concealment of negative organizational outcomes from shareholders in the president‟s letter (Abrahamson & Park, 1994), the word choice in the context of impression management in chairman‟s statements (Merkl-Davies, Brennan, McLeay, 2011), country differences in attributional properties of performance explanations in management commentary reports (Aerts & Tarca, 2010) and the use of optimistic either pessimistic tone in corporate disclosures (Rogers, Von Buskirk, & Zechman, 2011), to my best knowledge no prior research focused particularly on the use of different writing styles in corporate press releases.

Second, although prior research focused on the use of impression management in the United States (Abrahamson & Park, 1994; Cho, Roberts, & Patten, 2010; Li 2008), in the United Kingdom (Merkl-Davies, Brennan, & McLeay, 2011), by observing differences in attributional properties between the common law countries the UK, the USA, Canada and Australia (Aerts & Tarca, 2010) and one made a comparison between the readability of press releases in the USA, the UK, Canada, New Zealand and Hong Kong (Courtis, 1995), no other research has been conducted on writing styles in press releases of firms from several countries and belonging to divergent accounting rules systems. Furthermore, no other research made a comparison in writing style as a tool for impression management in corporate disclosures between firms belonging to the common law legal system and firms belonging to the civil law legal system.

(6)

Third, although prior research found empirical support for the relationship between the firm performance and positivism either negativism of the president‟s letter (Abrahamson & Park, 1994), between the firm performance and the optimistic tone of the firm disclosures (Cho, Roberts, & Patten, 2010), between the firm performance and the use of “certainty” in the environmental disclosures (Cho, Robert, & Patten, 2010), and between the firm performance and the use of positive and negative words and “complexity” in the chairman‟s statement (Merkl-Davies, Brennan, & McLeay, 2011), to my best knowledge no prior research has been conducted on the relationship between firm performance and writing style in corporate press releases.

Last, I found significant inconsistencies in prior research concerning the relationship between the firm performance and the writing style used in organizational disclosures. Abrahamson & Park (1994) found a significant correlation between firm‟s negative financial performance and the negativity of the president‟s letter. In line with this result, Merkl- Davies, Brennan, & McLeay (2011) found that firms reporting negative organizational outcomes use significantly less positive words and significantly more negative words in their disclosures. In line with these results, other studies found that reading difficulty increases as firms exhibit poor performance (Adelberg, 1979; Courtis, 1998; Li, 2006).

However, in a study conducted by Cho, Roberts, & Patten (2010), investigating language and verbal tone used in corporations‟ environmental disclosures, they found that firms with lower environmental performance use a more optimistic tone in their disclosures. These results are quite contrary and make this paper of much interest to contribute to the existing literature regarding writing style in corporate narrative disclosures.

2. Theory

2.1 Impression management

Considering a general perspective from psychology and sociology, impression management is aimed at shaping the perceptions of a wide range of outside partners (Merkl-Davies & Brennan, 2011) and is concerned with “studying how individuals present themselves to others to be perceived favorably by others” (Hooghiemstra, 2000). Based on the framework of

(7)

Merkl-Davies & Brennan (2007), impression management is divided in two broad categories: attribution and concealment.

2.1.1 Attribution

Merkl-Davies & Brennan (2007) identified two broad categories of impression management, which they divide into seven impression management strategies. One of these strategies is attribution, whereas the other six are used for concealment. Attribution refers to the propensity to ascribe positive outcomes to internal causes and negative outcomes to external causes (Merkl-Davies, Brennan, & McLeay, 2011). Basically, the attribution recognizes an unfavorable event, but denies responsibility to it by pointing to external factors. A denial occurs for unfavorable events “by commenting on negative outcomes as something that happened in spite of internal events or actions that normally should have led to the contrary” (Aerts, 2005). The attribution seeks to contribute to innocence and therefore reduces responsibility for unfavorable outcomes. Another type of attribution is a form of justification whereby the unfavorable outcomes are recognized and accepts responsibility to it, but simultaneously reduces its negative implications by pointing to its character as a tool for achieving higher goals (Aerts, 2005).

2.1.2 Concealment

The other six impression management strategies identified by Merkl-Davies & Brennan (2007) as tools for shaping the perceptions of outside partners are all used for concealment. Concealment refers to the use of writing styles in which unfavorable outcomes are obfuscated or by emphasizing favorable outcomes (Abrahamson & Park, 1994; Cho, Roberts, & Patten, 2010; Merkl-Davies & Brennan, 2007). Of these six concealment strategies, two attempt to hide bad news, and four attempt to emphasize on good news. These two obfuscating strategies attempt to hide bad news either by making the text more difficult to read (reading ease

manipulation) or by using persuasive language (rhetorical manipulation) (Rahman, 2012).

The other four concealment strategies attempt to emphasize good news by focusing on positive words and themes (thematic manipulation), by visually biasing the disclosed information (visual and structural manipulation), by choosing measures which show the financial performance favorably (performance comparison) or by disclosing only one of many financial numbers in order to emphasize on a certain financial performance (choice of

(8)

Reading ease manipulation attempts to obfuscate unfavorable outcomes by issuing longer and less readable disclosures (Li, 2008), in order to confuse or distract the reader (Courtis, 2004). Merkl-Davies & Brennan (2007, p. 133) noted that “syntactical complexity makes texts more difficult to read and this is regarded as a proxy for obfuscation”. However, there is a certain limit to the manipulation of the readability of text, because when a text is too complex, the purpose of the impression management will be surpassed (Rahman, 2012).

Rhetorical manipulation is a manipulative technique that attempts to obfuscate unfavorable outcomes by using more persuasive language (e.g. passive voice and use of pronouns) (Merkl-Davies, Brennan, & McLeay, 2011), “by emphasizing on not what firms say but rather how they say it” (Merkl-Davies & Brennan, 2007).

Concealment might be achieved by thematic manipulation of the disclosures whereby the management exhibits a self-serving bias in their choice of firm performance items to be disclosed (Cho, Roberts, & Patten, 2010), for instance by “including more favorable pro forma earnings numbers in corporate narratives or by displaying positive organizational outcomes more prominently than negative organizational outcomes by means of positioning or highlighting” (Merkl-Davies, Brennan, & McLeay, 2011).

Visual and structural manipulation attempts to manipulate the readers‟ perception of a firms‟ performance by affecting the way management displays the information. In order to accomplish this manipulation of perception, management repeats certain information, and positive information tends to be more emphasized and more prominently placed than negative information (Cho, Roberts, & Patten, 2010, Rahman, 2012). Visual manipulation could be accomplished by the use of visual effects such as font style, font size, bold text, colored text etc. in order to emphasize on certain parts of a text (Courtis, 2004).

Firms choose their comparisons and measures for their current financial numbers in a way that is as favorable as possible. Therefore, management will choose the lowest possible prior-period benchmark in order to report the highest possible year-on-year earnings, or may compare their financial numbers with reference points such as industry average or competitors‟ financial numbers in a way that is positive for the organization. (Rahman, 2012)

(9)

The last impression management strategy identified by Merkl-Davies & Brennan (2007) used for concealment is the disclosing of only certain financial numbers, and the omitting of other less favorable numbers in order to enhance the perception of the reader (Rahman, 2012). Conclusively, in essence corporate attempts at concealment and attribution focus primarily on manipulating either the presentation (language and verbal tone) or the disclosure (quantity, thematic content and attribution) of the disclosed information (Merkl-Davies & Brennan, 2007).

2.2 Underpinning theories

Regarding impression management in corporate narrative disclosures, several theories are conceivable underlying writing style as a tool for impression management. Recent thinking about management has been influenced by both economic approaches to governance such as agency theory, and sociological and psychological approaches such as the legitimacy theory. In impression management, agency theory is a generally accepted underlying theory, which depicts subordinates as individualistic, opportunistic and self-serving.

Alternatively to agency theory, stewardship theory depicts subordinates as collectivists, pro-organizational and trustworthy. Stewardship theory defines situations in which managers are not motivated by individual goals, but rather are stewards whose motives are aligned with the objectives of their principals. From stewardship theory, in impression management managers are not assumed to act as rational, self-interested entities, but rather as stewards whose behavior is ordered such that pro-organizational, collectivistic behaviours have higher utility than individualistic, self-serving behaviours (Davis et al., 1997).

In stakeholder theory, stakeholders are defined as “any group or individual who can affect or is affected by the achievement of the firm‟s objectives” (Freeman, 1983). Stakeholders of the firm include stockholders, creditors, employees, customers, suppliers, public interest groups, and governmental bodies. Given the stakeholder theory, a major objective of the firm is to attain the ability to balance the conflicting demands of various stakeholders in the firm (Roberts, 1992). In stakeholder theory managers may be prompted to engage in impression management in order to respond to the concerns of various stakeholder groups (Merkl-Davies & Brennan, 2011).

(10)

In contrast, legitimacy theory has its roots in the idea of a social contract between the corporation and society (Magness, 2006). In legitimacy theory managers may be prompted to engage in impression management in order to conform to social rules and norms (Ng & Tseng, 2008). To conform to social rules and norms is particularly prevalent with respect to social and environmental reporting, where managers may use writing style as a tool to convince the reader of their good intentions (Alexander, 2009). Managers may also engage in impression management as a means of legitimizing actions and in order to justify decisions (Hooghiemstra, 2000).

This thesis will put forward the agency theory as the predominant perspective on impression management in a corporate reporting context. Agency theory focuses on the relationship between managers and investors, which is primarily based on utility maximization. Since managers operate „in an environment in which their remuneration and wealth is linked to the financial performance of the companies that employ them, management has economic incentives to disclose messages that convey good performance more clearly than those conveying bad performance‟ (Rutherford 2003, p. 189). Agency theory provides therefore a narrow view of impression management, as it focuses on the relationship between managers (agents) and investors (principals), focuses on reporting bias regarding the financial firm performance, and conceptualizes impression management as the strategic use of discretionary narrative disclosures (Merkl-Davies & Brennan, 2011).

2.2.1 Agency Theory

In the context of corporate governance, the agency theory focuses on the relationship between the agent and the principal, thereby emphasizing on impression management as a tool for discretionary disclosure of information. Different stakeholders, like managers and investors, may have different incentives to disclose information, which can be categorized in incentives based on an economic perspective, incremental information, hubris or retrospective sense-making (Merkl-Davies & Brennan, 2011).

The different incentives of stakeholders are the result of a distinction of ownership by agents (management) and the principals (investors) (Eisenhardt, 1989). The relationship between the principal and the agent is named the agency relationship. Problems arise, when the agent and the principal do not pursue the same interests, thereby creating an agency problem

(11)

(Eisenhardt, 1989; Merkl-Davies & Brennan, 2011), or information asymmetries arise whereby the agent is assumed to have private information to which the principal cannot costlessly gain access (Baiman, 1990). Conflicts of interest and information asymmetries can arise when negative organizational outcomes occur (Abrahamson & Park, 1994). In this situation, both managers and investors are considered as rational, self-interested decision makers (Merkl-Davies & Brennan, 2011), assuming agents as opportunists favoring their own interests over the principals‟ interests (Abrahamson & Park, 1994).

The underlying model of man in a basic agency theory framework is the Resourceful,

Evaluative, Maximizing Model, described by Jensen & Meckling (1994). In their model, they

state this kind of man as a rational human being who pursues utility maximization and self-interest. Utility can be non-pecuniary benefits like the amount of charitable contributions and personal relationships, but also pecuniary benefits like a higher remuneration (Van Zijl, 2012). In impression management, the agency theory is generally recognized as the main supporting theory. Managers are assumed to act rationally to maximize their utility by exploiting information asymmetries to mislead investors about financial performance and prospects (Merkl-Davies, Brennan, & McLeay, 2011). This manifests itself in the adoption of one of more of the seven impression management strategies identified by Merkl-Davies & Brennan (2007), generally leading to reporting bias by the emphasis of positive organizational outcomes and obfuscation of negative organizational outcomes in corporate narrative disclosures. This research emphasizes on the assumption that managers act rationally to maximize their utility by influencing the writing style they adopt in corporate press releases relating to the final year results.

2.2.2 Theoretical underpinning of impression management

Following the extensive research on the agency theory by Eisenhardt (1989), the major focus is on the goal conflict between principal and agent whereby the management may have different incentives for discretionary disclosure of information. Necessary for the insight, this research investigates different incentives and means of discretionary disclosure of information. Prior research states incentives for discretionary disclosure of information (Merkl-Davies & Brennan, 2011); means of manipulation of disclosures (Macintosh, 2009); reasons for voluntary corporate disclosures (Lang & Lundholm, 2010) and managers‟ motives to conceal unfavorable outcomes (Abrahamson & Park, 1994). Becoming aware of these incentives will contribute to our understanding of the reason behind impression management

(12)

and will be helpful in the understanding of the results and in providing a well-founded conclusion.

In accordance to Merkl-Davies & Brennan (2011), incentives for discretionary disclosure of information can be categorized in four explanations. The first incentive is based on an economic perspective. Management is prone to manipulation of the presentation and disclosure in order to mislead investors about organizational outcomes (make awareness of the similarity with “concealment”). Merkl-Davies & Brennan (2011) name three alternative incentives for discretionary disclosure of information. The first alternative is the incremental information explanation, which is based on investor rationality and efficient markets. This efficient market hypothesis assumes that biased reporting would lead to higher cost of capital and reduced share price performance. Since managers‟ compensation is linked to share price performance, managers have no incentives to involve in impression management. Another alternative explanation for discretionary disclosures is managerial hubris, which refers to excessive individual pride that expresses itself in “a subconscious cognitive bias” (Merkl-Davies & Brennan, 2011). Individuals who are under a certain degree of hubris may take irrational actions and decisions, which may lead to discretionary disclosure of information. The final incentive for discretionary disclosures given by Merkl-Davies & Brennan (2011) is the result of managerial retrospective sense-making, which refers to managers who retrospectively assign causes to organizational actions and events (Merkl-Davies & Brennan, 2011).

Now prior literature on agency theory and different incentives for discretionary disclose of information has been examined, it is useful for the notion of discretionary disclosures to reflect on some philosophical considerations. According to Macintosh (2009), standard setters as GAAP hold to the idea that disclosures are ex post statements of some ex ante increase in the enterprise‟s wealth. However, contra this idea, in practice financial accounting executives participate in manipulation of disclosures in order to meet the capital markets‟ inexhaustible demand for increasing stock price which will contribute to executives‟ earnings levels. In his paper, Macintosh (2009) explains five means of manipulation of disclosures.

First, many accounts allow for a great proportion of subjectivity on the part of statement preparers. While balance sheet accounts like cash are easily to be confirmed, other accounts including R&D could be treated as an expense, as well as a future benefit. Such accounts as

(13)

R&D, along with accounts as goodwill and depreciation expenses are not as clear-cut as they look and are subject to a certain degree of subjectivity.

Second - and which is supported by Cho, Roberts, & Patten (2010, p. 432) - standard setters as GAAP are constantly changing; contain contradictions and ambiguities, thereby leaving opportunities for manipulation of disclosures. Furthermore, GAAP is notorious by statement prepares for its complexity and difficulty in composing financial statements.

A third means of manipulating disclosures is the widely spread practice of structuring transactions in order to achieve beneficial tax accounting results.

The manipulation of real activities is a fourth means of manipulating disclosures and involves strategic timing of sales and expenses in order to affect reported earnings.

The last means covers financial officers entering into negotiations with the corporations‟ external auditors regarding important accounting issues which have significant influence on their financial statements (Macintosh, 2009: p. 152-154).

Summarized, according to prior literature, many reasons might be suggested for discretionary disclosure, - and manipulation of information. Now prior literature on incentives for discretionary disclosure of information and means of manipulating has been examined, this section will be finalized section with a brief statement derived from prior literature why managers may or may not be willing to voluntary disclose information. Lang & Lundholm (2010) studied the relationship between firms‟ discretionary disclosure and stock price, concluding that managers increase firms‟ discretionary disclosures in order to higher the firms‟ share price, confirming the statement of Merkl-Davies & Brennan (2011) that managers are “rational, self-interested decision makers.”

Opposite, Abrahamson & Park (1994) state why it is not in managers‟ interest to reveal negative organizational outcomes. First, managers fear dismissal after negative organizational outcomes, and fear for reputational damage, making employment after dismissal more difficult to obtain. Second, managers experience that if they reveal negative outcomes, shareholders will sell their shares, thereby lowering firms value and managers wealth if they own company shares.

(14)

Conclusively, this theory section is summarized as follows: impression management is aimed at shaping the perceptions of a wide range of outside partners (Merkl-Davies & Brennan, 2011) that traditionally is separated in the techniques attribution and concealment. The focus of these techniques will primarily be on either language and verbal tone or the quantity, thematic content and attribution of the disclosed information (Merkl-Davies & Brennan, 2007). The underpinning key element of impression management is the agency theory, which serves as an explanatory and supporting theory regarding the role of attribution and concealment in conflicts of interest and information asymmetry between agents and principals (Eisenhardt, 1989). Underlying the agency theory and referring to prior research on impression management by Merkl-Davies & Brennan (2011), Macintosh (2009), Lang & Lundholm (2010) and Abrahamson & Park (1994), managers and investors may have different incentives, motives and means to (voluntary) disclose information.

2.3 Common law & Civil law legal system

This study will examine to what extent firm performance explains differences in writing styles between firms, thereby adopting an international perspective. To accomplish an international perspective, this research will control for the legal system of the country in which the firm operates, thereby making a distinction between countries adopting the common law system and countries adopting the civil law system.

Laws in general are not written from scratch, but rather adopted on a voluntary basis centuries ago or due to forced imposition by past colonial powers (Mintz, 2005). Distinction is made between common law, which originates in England (The UK en The USA) and the civil law family which derives from Roman law. The civil law family can be further subcategorized into French (Belgium, The Netherlands, part of Italy, Western parts of Germany), German (Germany and parts of Italy) and Scandinavian Law (La Porta, Lopez de Silanes et al. 1998). Laws vary substantially between these countries: civil law gives fewer rights to investors than common law does, shareholders in common law countries have the strongest protection and French civil law countries the weakest protection (Galle, 2012).

Ball, Kothari, & Robin (2000) compared the quality of financial reporting in several common law-countries (Australia, Canada, the USA and the UK), to reporting quality in civil law

(15)

countries (France, Germany, Japan). With regard to standards and regulations, a number of contrasts exist between common law and civil law countries. In common law countries, like the United Kingdom, the accounting standards are set in the private sector and are primarily applicable on investors. In contrast with common law countries, in civil law countries the corporate governance structure is primarily set by governments and therefore the company structure is subject to more political influence than under a common law legal system. As a result of a larger amount of political influence under civil law, additional constituencies, such as banks, business associations and labor unions, are presented within the corporate governance structure (Scott, 2011, §13.7.2). In effect, under the civil law system there is less information asymmetry, since important constituencies are insiders rather than outsiders (Ball, Kothari, & Robin, 2000).

Although this study is the first study examining writing styles in press releases under divergent legal systems, some prior research already considered the effect of a company‟s national institutional environment on the explanatory statements in company‟s management commentary. Aerts & Tarca (2010) examined properties (like complexity) of attributional framing of financial performance outcomes and considered how they differ between countries. In order to make the comparison, they selected four common law countries, knowing the USA, Canada, the UK and Australia. These countries have in common to adopt a common law legal system where public information dissemination is essential for an efficient capital market, but they face differences in expected regulatory and litigation costs. From these countries, the UK is the shared country for which both Aerts & Tarca (2005) and this study investigate as a common law country. Different from mandatory regulations in i.e. the USA, narrative reports in the UK are provided in response to voluntary best-practice guidelines. The minimal nature of mandatory rules in the UK minimizes regulatory and litigation costs in these countries. From large common law countries, the UK is perceived to have the lowest regulatory and the lowest litigation costs, leading up to that UK companies are expected to have less detailed and less formal explanations of performance (Aerts & Tarca, 2010).

Jones & Pittman (1982) examined an alternative prior cross-country research on institutional settings. They argue that the more institutionalized the environment, the less management adopts an impression management strategy. A higher degree of scrutiny, for example the

(16)

mandatory reports by the SEC in the USA, is associated with higher regulatory and litigation costs and therefore with a lower likeliness of impression management to be involved.

The required data is derived from listed companies located in several European countries, namely Belgium, France, Italy, The Netherlands, Spain (civil law) and the United Kingdom (common law). Choosing these countries enables us to add a dummy variable to the regression model and thereby testing for different writing styles between common law and civil law countries.

3. Hypotheses

Now the theoretical framework regarding writing style as a tool for managing stakeholders‟ impressions is established, the problem will be concretized. This paragraph will discuss the research questions in order to structure the problem and the results. Prior research shows firm performance to be causally linked to the degree of positivity either negativity of corporate disclosures. However, these studies show inconsistent results, which make this research question highly interesting and will contribute to answering the main question. Abrahamson & Park (1994) conducted an agency theory-based research on concealment of negative organizational outcomes. They found that both declining financial performance and changes in financial performance show correlation with “negativity” in President‟s letter. This suggests that when financial performance is low, agents account for them in president‟s letters. This result affirms the findings by Merkl-Davies, Brennan, & McLeay (2011), who examined a psychological research on impression management in corporate narratives. They found that firms that report negative organizational outcomes use significantly less positive emotion words and significantly more negative emotion words than firms that report positive organizational outcomes. Furthermore, other studies by Adelberg (1979), Courtis (1998) and Li (2006) confirm these results, finding that reading difficulty of firms‟ disclosures increases as firms exhibit poor performance. Figure 1 presents the results from prior research.

(17)

In figure:

However, a major comparing study has been conducted which shows a very different result on this matter. Cho, Roberts, & Patten (2010) conducted a research on the language of corporate environmental disclosures finding that poorer environmental performers compose their environmental disclosures in a more optimistic language tone. This indicates that the used writing style in these environmental disclosures focuses on attribution of favorable outcomes to internal causes and unfavorable outcomes to external causes. The findings of the studies regarding firm performance and positive either negative characteristics of firms‟ disclosures is summarized in Figure 1. These results are quite contrary. Therefore, I state firm performance to be of significant influence on writing styles in organizational disclosures, hence providing a non-directional hypothesis.

Hypothesis 1a is stated as follows:

H1a. The positive either the negative nature of firms’ press releases will be related to the firms’ performance.

 Relationship between “firm performance” and “positive/negative words”

Negativity presidents‟ letter

↑

(Abrahamson & Park)

Negative firm performance Negative emotion words in chairmen‟s Statements

↑

(Merkl-Davies, Brennan, & McLeay)

Positive emotion words in chairmen‟s Statements

↓

(Merkl-Davies, Brennan, & McLeay)

Optimistic tone in environmental disclosures

↑

(Cho, Roberts, & Patten)

Figure 1

Summary of prior research on the relationship between firm performance and positive or negative writing style in corporate disclosures.

(18)

As depicted in Figure 1, prior research shows agents could account for unfavorable organizational outcomes by the use of more negative and less positive emotion words (Abrahamson & Park, 1994; Merkl-Davies, Brennan, & McLeay, 2011) or they attempt to obfuscate those unfavorable organizational outcomes by the use of a more optimistic tone (Cho, Robert, & Patten, 2010). Besides the possibility of the use of more positive and/or negative emotion words as an agents‟ tool for implementing an impression management strategy, other tools are applicable in practice. Disclosure of unfavorable organizational outcomes could be accomplished by using “convoluted, less certain language” (Jones, 1996), in order to mask internal attributions (Cho, Roberts, & Patten, 2010). The term certainty refers to a writing style that indicates “resoluteness, inflexibility, completeness, and a tendency to speak ex cathedra” (Hart, 2001). In other words, obfuscation of unfavorable outcomes could be achieved by using an irresolute, flexible, incomplete and pedantic writing style. Cho, Roberts, & Patten (2010) investigated the use of “certainty” in US corporate environmental disclosures finding a significant relationship between low firm performance and the use of a less certain writing style. This suggests that firms with unfavorable outcomes seek to contribute to innocence for these unfavorable outcomes by using a less certain writing style. I expect therefore agents to use a less certain writing style in case of disclosure of unfavorable outcomes.

Hypothesis 1b is stated as follows:

H1b. The use of a less “certain” writing style in firms’ press releases will be related to a lower firm performance.

 Positive relationship between “firm performance” and “certainty”.

Agents striving for implementing impression management in their disclosures are expected to attribute favorable organizational outcomes to internal causes and negative outcomes to external causes; or either implementing a writing styles in which unfavorable outcomes are obfuscated or by emphasizing favorable outcomes (Abrahamson & Park, 1994; Cho, Roberts, & Patten, 2010; Merkl-Davies & Brennan, 2007). Besides implementing a positive either negative writing style and/or using a certain writing style in corporate disclosures as a tool for impression management, a third category of implementing an impression management strategy is to strive to a higher text complexity. Merkl-Davies, Brennan, & McLeay (2011) found that chairmen‟s statements of unfavorable outcomes are characterized by a more

(19)

complex writing style than those of statements reporting favorable outcomes. This suggests unfavorable outcomes to require more complex and detailed disclosures than in case of favorable outcomes. According to Bloomfield (2008), document length may be interpreted as an indicator of text complexity and is therefore a proxy for complexity of disclosures. Following the finding by Merkl-Davies, Brennan, & McLeay (2011) that disclosure of unfavorable outcomes results in higher text complexity, and the finding of Li (2008) that low firm performance is related to longer sentences,

Hypothesis 1c is stated as follows:

H1c. A complex writing style in firms’ press releases will be related to a lower firm performance.

 Negative relationship between “firm performance” and “word count”.

This paper studies the relationship between a firms‟ performance and different aspects of writing style as tool for impression management, namely the amount of words used and the use of positive emotion words, negative emotions words and words signifying for certainty in a firms‟ press release regarding the final year results. Furthermore, this study adopts an international perspective by comparing writing styles between common law countries and civil law countries. No other studies have ever been examined on writing styles of corporate disclosures under divergent legal systems. Therefore, hypothesis 2 concerning divergent legal systems is based on assumptions by prior research with a corresponding subject. Prior research on divergent legal systems is based on the presumed height of the firm‟s regulatory and litigation costs and thereby with managers‟ tendency to adopt an impression management strategy.

Aerts & Tarca (2010) conducted a research on country differences in the attributes of managers‟ explanations of performance in management commentary. They found that firms with minimal mandatory rules, like in the UK (compared to the USA), are exposed to lower regulatory costs and litigation costs and are therefore expected to disclose less detailed and less formal explanations of performance.

Other research on country differences and divergent legal systems examined by Jones & Pittman (1982) confirm this result. They found that firms operating in highly institutionalized

(20)

environments, and thus which are subject to strict regulations, are exposed to higher regulatory and litigation costs and therefore with a lower likeliness of impression management to be involved.

Ball, Kothari, & Robin (2000) conducted a research on international institutional factors, comparing financial reporting quality between common law and civil law countries. They argue that in civil law countries, national accounting standards are established and enforced by the government and hence firms are subject to more political influence that under common law.

Based on above mentioned prior research on international institutional factors, this research assumes that firms operating under stringent governmental accounting rules, have a lower tendency to adopt an impression management strategy, due to higher regulatory and litigation costs. That includes: Firms in common law countries that disclose a press release concerning the final year results, use a writing style which is in a less extent “aimed at shaping the perceptions of a wide range of outside partners” (Merkl-Davies & Brennan, 2011).

Hypothesis 2 is stated as follows:

H2. Firms operating under a common law legal system are less likely to adopt an

impression management strategy in the writing style of press releases than under a civil law system.

4. Research methods

This paper will emphasize on writing styles of press releases from several countries representing six language regions, namely Dutch, English, French, German, Italian and Spanish. Based on several characteristics of „writing style‟, this paper will examine to what extent firm performance explains differences in writing styles of corporate press releases. Furthermore, this study will examine the influence of the legal system adopted in a country on the applied writing style in press releases relating to the final year results.

(21)

In order to test hypotheses H1a, H1b, H1c and H2, I obtained a representative sample of press releases regarding the final year results of European listed companies, and proxies were developed for “positive words,” “negative words,” “certainty” and “complexity” using LIWC text analysis software. In addition, control variables were defined that serve as a control mechanism for other important factors that, based on prior research, may be related to managers‟ use of a specific writing style in their press releases. The details of the data method are described below.

4.1 Sample selection

The press releases were derived by obtaining the English releases concerning the final financial results over the years 2009, 2010 and 2011. The press releases were freely obtainable through the investor relationship page or the press/news page on the corporate website. In the event the press release regarding the final year results were not available, the press release regarding the preliminary results were used. To be included in the sample, the sample firms had to meet the following criteria:

1. The firm had to be listed on at least one European stock exchange. 2. They had to have a fiscal year ended in 2009, 2010 and in 2011.

3. They had to have a press release available over both the years 2009, 2010 and 2011. 4. They had to have a press release regarding the final year results for the years 2009,

2010 and 2011. If not available, the firm had to have a press release available regarding the preliminary results instead of a press release regarding the final year results.

From the sample of 4496 company-press release combinations, there were such 2251 organizations, which met all the four criteria and constitute the final sample. Subsequently, I used the LIWC program for the measurement of our sample.

(22)

4.2 Variables

4.2.1 Dependent variables

Regarding the dependent variables, I used Linguistic Inquiry and Word Count (LIWC), a computerized text analysis program developed by the psychologists Pennebaker, Booth, & Francis (2007), for the purpose of analyzing linguistic style. Following Merkl-Davies, Brennan, & McLeay (2011), this research uses the LIWC software in order to measure the dependent variables of writing style of press releases. The LIWC program consists of two components, the program itself and a dictionary file, thereby comparing each word in the press release with a dictionary file. The dictionaries are the core of the LIWC program and refer to a series of words divided in prebuilt or customizable categories (Tausczik & Pennebaker, 2010). The LIWC software analyses the press release on a word-by-word basis and calculates the number of words that match with the pre-defined dictionary file, providing a percentage of words matching with the dictionary. For example, the dictionaries file for “Tentative words” consisting of words like “maybe,” “perhaps” and “guess.” (Merkl-Davies, Brennan, & McLeay, 2011).

Following prior research on linguistics, several elements of writing style can be distinguished. Cho, Roberts, & Patten (2010) studied language and verbal tone by measuring “optimism” and “certainty.” They state optimism and certainty in environmental disclosures measuring the presence of the optimistic words and words suggesting certainty. A high optimism is been associated by words as: “highest standards” and “improvement”; a low optimism by “discharge” and “costly”, high certainty by words as “million” and “both in year 200x and 200x” and low certainty by “we believe” and words as “could”, “can” and “changes”. Making use of a quite similar research method, Abrahamson & Park (1994) studied the concealment of negative organizational outcomes making use of a computer-assisted content analysis. They measured “negativity” by means of a database, similar to Cho, Roberts, & Patten (2010), using text analysis on words like “loss” and “disappointing,” denoting for negativity in the presidents letter. Furthermore, Merkl-Davies, Brennan, & McLeay (2011) use a content analysis on the psychological dimension of word use making use of four dimensions of the LIWC program, i.e. “word count”, “positive emotion words”, “negative emotions words” and “cognitive complexity.”

(23)

Following prior research on writing style in corporate disclosures, this research will make use of the LIWC software to measure the dependent variables. The LIWC program is necessary to provide the LIWC dictionaries match percentages and the total word counts necessary in the regression model. In order to explain to what extent firm performance explains differences in writing styles; the relationship will be examined between the firms‟ Return on Assets (Earnings Before Interest and Taxes / Total Assets) and, according to the corporate press releases the following linguistic indicators:

Table 1. Linguistic indicators

Linguistic indicator LIWC

1) The amount of positive emotion words % matching with positive emotion words

2) The amount of negative emotion words % matching with negative emotion words 3) The amount of “certainty” words % word matching with “certainty” words 4) The total word count, which counts for

“complexity” (Bloomfield, 2008).

Log # total word count

LIWC: Linguistic Inquiry and word count

4.2.2 Independent and control variables

Three independent variables are tested in this study: the independent variable firm

performance, and the variables firm size and industry classification that serve as control

mechanism for other factors, rather than firm performance that may influence the applied writing style in press releases.

4.2.2.1 Firm performance

The variable firm performance is measured as the Return On Assets. Following prior research on the relationship between firm performance and positivism on the president‟s letter by Abrahamson & Park (1994), research on the relationship between firm performance and the optimistic tone of firm disclosures executed by Cho, Roberts, & Patten (2010), and based on Rogers, Von Buskirk, & Zechman, (2011), who control for firm performance in their research

(24)

on disclosure tone, in this study firm performance will be computed as the firms‟ Earnings Before Interest and Taxes (EBIT) divided by the firms‟ Total assets.

4.2.2.2 Firm size

Prior research on corporate disclosures indicates that the extent of disclosure is significantly related to firm size. Li (2006) found bigger firms to disclose more complex annual reports, and Cho & Patten (2007) find that the extent of disclosures is significantly positively related to firm size. Cho, Roberts, & Patten (2010) indicate this relationship probably exists because larger companies carry out a higher visibility and therefore tend to disclose more information. Since this study tests for the relationship between firm performance and the total amount of words used in the press releases as a proxy for complexity and as a tool for impression management, the test will control for firm size. Following Abrahamson & Park (1994) and Merkl-Davies, Brennan & McLeay (2011), firm size will be computed as the Total Assets of each company.

4.2.2.3 Industry classification

The second independent control variable is the industry in which the company operates. Cho & Patten (2007) indicate that companies operating in stressful industries systematically provide more extensive disclosures than companies operating in less stressful industries do. According to Aerts (2005), impression management may be affected by the industry in which the company operates. Given that this study tests for the amount of words used in the press releases as a proxy for complexity, the test will control for industry classification. Following prior research on impression management by Abrahamson & Park (1994); Li (2006); Cho, Roberts, & Patten (2010), Merkl-Davies, Brennan, & McLeay (2011) and Aerts, Cheng, & Tarca (2013), industry type is included in the sample as a control dummy variable clustered in SIC-codes.

(25)

The following four-digit SIC-codes are classified as dummy control variables:

Table 2. SIC industry classification

SIC Civil Industry classification

SIC_1 Consumer Durables, NonDurables,

Wholesale, Retail, and Some Services (Laundries, Repair Shops)

SIC_2 Manufacturing, Energy, Utilities

SIC_3 Business Equipment, Telephone and

Television Transmission

SIC_4 Healthcare, Medical Equipment, and Drugs

SIC_5 Mines, Constructions, Transport, Hotels, Bus

Services, Entertainment, Finance

4.2.3 Common law / Civil law

Our data consists of press releases of firms from two legal systems: the common law system and the civil law system. This study will have an international perspective, focusing on press releases from the common law countries in the United Kingdom, and the civil law countries Belgium, France, Germany, Italy, The Netherlands and Spain. By choosing these countries, this study adopts a broad international perspective on the applied writing styles in press releases, thereby making a comparison between common law and civil law countries. In order to make the comparison in the adopted writing style in corporate press releases between common law countries and civil law countries, a dummy variable has been added to the regression model.

The following model has been used in the regression analysis:

Writing style = a + x*FirmPerformance + x*FirmSize + x*SIC + x*Law

Conclusively, prior research on writing style in corporate disclosures has been executed through a computerized content analysis, making use of databases for the measurement of

(26)

“positivity,” “negativity”, “certainty”, “complexity” and “word count.” Following prior research on analysis of writing style, this study makes use of the LIWC software, thereby conducting content analysis by means of: word count – which serves as a proxy for complexity of disclosures according to Bloomfield (2008), certainty words, negative emotion words and positive emotion words. This study examines the relationship between firm performance, which has been calculated as (EBIT/Total Assets), and the LIWC scores for %positive words, %negative words, %certainty words and #total words, in order to answer the research question: “To what extent does firm performance explains differences in writing styles between firms”. In addition, a dummy variable has been added for the legal system the firm belongs to, testing for the influence of the legal system on the writing style. Moreover, this study controls for firm size and industry classification, which may, rather than firm performance, influence writing style in corporate press releases.

5. Results

5.1 Descriptive statistics

Our final sample for the analysis consists of 2251 press releases from European listed companies originating from Belgium, France, Germany, Italy, The Netherlands, Spain and the United Kingdom. The press releases are concerned with the financial year results and were derived from the investor relationship page on the corporate website of the selected firms.

Table 3 presents the descriptive statistics for the dependent, independent and control variables. One can derive from Table 3 that the average press release contains 3.5% positive emotion words, 1% negative emotion words, 0.9% words signifying for certainty and consists on average of about 4700 words. We added value 1 to common law countries and value 0 to civil law countries, thus from the table one can deduce that the ratio common law: civil law countries is 40/60. This is in line with our desire for a balanced final sample.

All variables except for the dummy variables (SIC and LAW) are winsorized to the 1st and the 99th percentile to reduce the effect of outliers.

(27)

Table 3. Descriptive statistics

Descriptive statistics Code Mean Standard

deviation

Skewness Kurtosis N

Dependent variables

Positive emotion words PosEmo 3,499 1,009 0,143 -0,124 2251 Negative emotion words NegEmo 1,005 0,520 0,453 0,361 2251 Certainty words Certainty 0,911 0,352 0,326 1,024 2251 Complexity (word count) Complexity 4694 4838 1,938 4,521 2251

Independent variable

Firm performance (ROA) FirmPerformance 0,055 0,115 -2,338 10,316 2251

Control Variables

Firm size FirmSize 81408919 218654869 4,583 23,197 2251 Legal system Law 0,410 0,492 0,372 -1,864 2251

In order to test for normal distribution in our sample, the skewness and kurtosis statistics of the different variables are tested. Skewness measures the symmetry of the distribution and kurtosis measures the relative concentration of values in the center of the distribution. In both statistics, a value of zero means a normal distribution of the sample data. The generally accepted range of skewness lies between -1 and 1, and between -3 and 3 for the kurtosis statistics. Within these ranges, the distribution approaches normality.

One can derive from Table 3 the independent variable FirmPerformance, the dependent variable Complexity and the control variable FirmSize are outside these rules of thumb. The data outside the accepted ranges need to be transformed since these data are not normally distributed. FirmSize and Complexity are positively skewed, so the arithmetic mean is greater than de median and greater than the mode. This means the final sample consist of more smaller than larger firms, and a greater proportion of shorter press releases than longer press releases. Following the suggestions on right skewed data by Decoster (2001), for both FirmSize and Complexity the natural logarithm of the variables is calculated. The independent variable FirmPerformance is negatively skewed, which means a greater proportion of better performing firms than lower performing firms. Following Decoster (2001), we calculate the exponent of the independent variable FirmPerformance. For the

(28)

FirmPerformance data, transformation by the exponent is the best possible option to approach a normal distribution. Table 4 provides the descriptive statistics after the transformation of the data.

Table 4. Descriptive statistics after transformation

Descriptive statistics Code Mean Standard

deviation

Skewness Kurtosis N

Dependent variables

Positive emotion words PosEmo 3,499 1,009 0,143 -0,124 2251 Negative emotion words NegEmo 1,005 0,520 0,453 0,361 2251 Certainty words Certainty 0,911 0,352 0,326 1,024 2251 Complexity (word count) Complexity 7,921 1,118 -0,315 -0,526 2251

Independent variable

Firm performance (ROA) FirmPerformance 1,063 0,110 -1,190 5,948 2251

Control Variables

Firm size FirmSize 16,110 2,319 -0,313 0,081 2251 Legal system Law 0,410 0,492 0,372 -1,864 2251

All skewness and kurtosis statistics are within the accepted range. Except for FirmSize, however from all possible data transformations, the exponent transformation leads to the best possible approach for a normal data distribution.

5.2 Analysis

To analyze the relationship between the dependent, independent and control variables the variables are tested in four models. In each model, the independent variable FirmPerformance and the control variables FirmSize, Law and SIC remain equal. Model 1 tests for the dependent variable Complexity, Model 2 for PosEmo, Model 3 for NegEmo and in Model 4 tests for Certainty. The statistics of the models are provided in Table 5. The VIF statistics is used to detect multicollinearity. A VIF-value of more than 10 is a significant indicator of multicollinearity and denotes for contaminated coefficients. All models are tested whether or not they satisfy the assumptions underlying a regression model, based on each‟ models linearity, normality, homoscedasticity and independence. No evidence was found that these assumptions were violated.

(29)

5.2.1 Complexity

Model 1 includes the dependent variable Complexity. The results of the model indicate that the R2 and the adjusted R2 (R2 = 0.356, adjusted R2 = 0.355) are significantly different from zero at the p < 0.001 level. The adjusted R2 of 0.355 indicates that 35.5% of the variations in the Complexity (word count) of the press releases are caused by the independent variables. The highest VIF statistic found in this model is 1.311, which indicates no serious levels of multicollinearity.

Looking at the different coefficients in Model 1, we see that FirmPerformance is significantly negative related to Complexity at the p < 0.01 level. This is according to our expectation that managers attempt to obfuscate unfavorable organizational outcomes by striving to a higher text complexity. Looking at the control variables, we see that the legal system is positively related to Complexity at the p < 0.001 level. Since Law is a dummy variable with common law (= 1) and civil law (= 0), common law firms indicate for a more complex writing style in their press releases. This is in accordance with our expectations that firms operating under stringent governmental accounting rules, which are applicable on common law countries, are more likely to disclose more extended explanations of performance. This is in line with Aerts & Tarca (2010).

Next, the results indicate also a significant positive relationship between FirmSize and Complexity at the p < 0.001 level. This is consistent with our expectation. Larger firms carry out a higher visibility and therefore tend to disclose press releases that are more complex and are more extensive. This result is in line with research on writing style executed by Li (2006) and by Cho, Roberts, & Patten (2010).

Last, the results indicate a significant relationship between SIC and Complexity at the

p < 0.10 level. This is consistent with our expectations and in line with prior research on writing style by Aerts (2005) and Cho & Patten (2007). They argue that impression management may be affected by the industry in which the company operates and that companies operating in stressful industries systematically provide disclosures that are more extensive.

(30)

Table 5. Results Regression analysis

Dependent variables Model 1

Complexity Model 2 PosEmo Model 3 NegEmo Model 4 Certainty

Standardized β Standardized β Standardized β Standardized β

Intercept **** **** **** **** Independent variable FirmPerformance -0,048 *** 0,108 **** -0,167 **** 0,021 Control variables SIC 0,030 * -0,092 **** -0,074 **** -0,032 Law 0,653 **** 0,447 **** 0,265 **** 0,077 ** FirmSize 0,304 **** -0,011 * -0,105 **** -0,105 R2 0,356 0,214 0,144 0,007 Adjusted R2 0,355 0,212 0,142 0,006 F-value 310,818 **** 152,694 **** 94,403 **** 4,239 ** Highest VIF (collinearity) 1,311 1,311 1,311 1,311

(31)

5.2.2 Positive emotion words

Consistent with Model 1, Model 2 includes the independent variable FirmPerformance and the control variables SIC, Law and FirmSize. The dependent variable included in Model 2 to test for writing style is PosEmo. The overall statistics of Model 2 show a decrease in the F-statistics from 310.818 in Model 1 to 152.694 in Model 2. The model however, is still significant at p < 0.001 level. The results indicate that the R2 and the adjusted R2 (R2 = 0.214, adjusted R2 = 0.212) are significantly different from zero at the p < 0.001 level, which means that the aggregate of the variables FirmPerformance, SIC, Law and FirmSize have significant influence on the dependent variable PosEmo. The independent and the control variables are identical to Model 1, thus the highest VIF-value remains at 1.311, which indicates no serious levels of multicollinearity.

Looking at the different coefficients in Model 2, we see that FirmPerformance is significantly positive related to PosEmo at the p < 0.001 level. This is consistent with a part of the expectations. Based on prior research on writing style as a tool for impression management we hypothesized without a positive or negative direction. Findings in prior research (include Merkl-Davies, Brennan, & McLeay, 2011) showed a significant positive relationship between firm performance and “positivity” or the use of more positive emotion words in corporate disclosures. Thus, our result is consistent with this part of the expectations. However, we hypothesized two-tailed because other research (include Cho, Roberts, & Patten (2010)) show divergent results, finding that firms with a lower performance use a more optimistic language tone in their disclosures. This part of the expectations is not consistent with our results. Conclusively, the results indicate a significant positive relationship between the firm performance and the amount of positive emotion words used in corporate press releases.

Looking at the control variables, we see that the legal system is significantly positive related to PosEmo at the p < 0.001 level. Since the dummy variable for common law has a value of (=1), and the dummy for civil law has a value of (=0), common law firms indicate for the use of more positive emotion words in their press releases. This is not according to the expectation. A possible reason for the positive relationship is that common law firms operate under more stringent governmental accounting rules than civil law firms do. Using more positive emotion words in their disclosures enables managers to compensate for these strict accounting rules allowing them to use a positive writing style as an impression management tool.

Impression management in corporate press releases: an international perspective