The Use of Graphs in Annual Reports. Evidence from European Listed Banks

The use of graphs in annual reports. Evidence from European listed banks.

Abstract

Graphs are important tools in a firm’s overall disclosure strategy. They can be used in order to communicate information in a concise and eye-catching way to serve users’ interest. However, graphs have also been used in a selective and distorted way, to give a more favourable impression about the firm performance to the reader, and serve the preparer’s interest. Using a sample of 47 major European listed commercial banks, this paper studies the nature and differences in graph usage in major European listed banks, and whether institutional settings and banks’ financial performance had any influence on the way banks used graphs. We found a significant and positive relationship between the use of financial performance graphs and the increase in banks’ overall financial performance as well as significant positive association between the increase of some specific performance variables and the number of their graph representations. We also found significant differences among the countries studied in term of graphs’ use and design. Anecdotal evidence of measurement distortion of financial performance graphs has been found. Such findings support impression management and institutional theories.

1. Introduction.

During the past few decades, the annual report of large listed companies has been transformed from a primarily formal, legal document into a public relations document in which the financial statements are almost relegated to a ‘technical appendix’ (Lee, 1994; Hopwood, 1996; Beattie et al, 2008). Listed companies use visual representation in their annual reports to communicate financial and non-financial information, augmenting the financial statements and related notes with a variety of additional material, like graphs.

The use of graphs has historically functioned for data representation to facilitate managerial decision-making (Masini, 1947). Beattie and Jones (2008) illustrate the main motivations for graph use and design choices by annual reports’ preparers to serve the interests of the annual reports’ users. Graphs allow management to present information in a flexible way as they usually fall outside the framework of accounting regulation[1]. In addition, they are an ‘eye-catching’ presentational format. Not only do they attract the reader’s attention, especially when in colour and/or highlighted, but also facilitate comparisons and the identification of trends. Graphs also synthesize key performance indicators and enable management to present key financial and non-financial information in a readily accessible form, even for the less expert users. In addition, being visual, graphs allow ‘spatial’ rather than ‘linguistic’ decoding. The reader can, therefore, use ‘sight’ (the dominant visual sense) to ‘see’ the data more directly and clearly. Graphs are ‘memorable’ as human beings tend to retain pictorial and graphical representations better than numbers (DeSanctis, Jarvenpaa, 1989). Last but not least, graphs are ‘international’ as they tend to be independent of language.

Besides the above-mentioned ‘altruistic’ motivations (Merkl-Davies and Brennan, 2007) for graph use and design choices, a major concern is the use of graphs to serve managerial interests rather than users’ interests with the potential outcome that the message conveyed is no longer neutral and unbiased (e.g., Beattie and Jones, 1992; 2008; Dilla and Janvrin, 2010). Whether graphs are used to pursue the informational needs of the users of the annual report, rather than the preparer’s own interest is still an unsolved question.

The relevance and role of graphs have been documented in previous studies. Several single-country studies have been conducted (e.g., Johnson et al, 1980; Beattie and Jones, 1992, 1999; Courtis, 1997; Mather et al, 2000; Godfrey et al, 2003; Ianniello, 2009), mostly on Anglo-American firms (see Beattie and Jones, 2008 for a review), however the disclosure of financial information using graphs has been the subject of few comparative studies (Beattie and Jones, 1997; 2001; Frownfelter and Fulkerson, 1998). In addition, no study has yet been focused on the use of graphs in banks, as the latter have often been excluded as they were expected to have different graphical reporting practices from non-financial companies (e.g., Beattie and Jones, 2001).

Therefore, the main purpose of this study is to investigate the nature and differences in graph usage in major European banks, comparing the way variables are graphed with the findings of previous studies that focused on non-financial companies.

The remainder of this paper is structured as follows. In the next section, the prior research on the use of graphs in non-financial companies is discussed. Hypotheses will be developed accordingly. In section three, we present our research methodology, including sample selection and classification of graphs. Our findings follow in Section four. In section five we discuss our results and conclude.

2. Literature review and hypotheses’ development.

The primary focus of financial report should be to provide a true and fair view about a company’s performance (IASB, 1989) in order to serve users’ interest, however academic literature has documented the incentives for, and ways in which, management seeks to create a more favorable view of the company’s performance than is warranted. Examples of such practices include earnings management (for a review, see Healy and Wahlen, 1999), accounting narratives (for a review, see Merkl-Davies and Brennan 2007), the misuse of photographs (e.g., Graves et al, 1996), as well as graphs.

Previous literature has found that the self-serving motivation is likely to arise and determine a selective use of graphs (e.g., Beattie and Jones, 1992; 2000a). Companies’ managements may have incentives to represent their companies’ performance in the best possible light, potentially resulting in selective financial misrepresentation (Tweedie and Whittington, 1990; Revsine, 1991), with graphs being adopted and designed specifically to manipulate the financial signals sent to annual reports’ users, enhance their perception of corporate performance, and lead users of financial information to sub-optimal decisions (e.g., Beattie and Jones, 1992; 1997; Dilla and Janvrin, 2010). In financial graphs, selectivity is likely to occur when a company graphs variables when there is a favorable trend (e.g., rising operating profit) and elects not to graph variables with unfavorable trends (e.g., declining operating profit). The absence of graphs tends to conceal poor performance, while companies use graphs to make good performance more salient to the users (Beattie and Jones, 1992; Dilla and Janvrin, 2010). The outcome of such impression management behavior is that the information conveyed with graphs is no longer neutral (Beattie and Jones, 2008). Previous studies have found that selectivity has been used to highlight financial performance variables through the use of graph (Beattie and Jones, 1992; 2000). Hence, we expect that:

Hypothesis 1a: Financial performance indicators are more likely to be graphed in annual reports of banks with good, rather than bad, financial overall performance.

Hypothesis 1b: Financial performance indicators are more likely to be graphed in the annual reports of banks with good, rather than bad, performance in terms of the variable graphed.

Positive accounting theory (Watts and Zimmerman, 1986) predicts that managers of highly visible firms, exposed to public scrutiny and media and regulator’s attention (such as oil and gas industry, Watts and Zimmerman, 1978), may deem it undesirable to make large increases in firm performance salient to annual reports’ users, as among those users there are the regulators. As regulators are not fully informed as it is costly for them to become informed about whether firm performance represents monopoly profits or not, drawing attention to high profits is more likely to increase political costs, and an highly visible firm may want to avoid changes in regulations that would either constrain their activities and/or impose more taxes on them. Along this line, Dilla and Jarvin (2010) found that large non-financial companies with greater performance increases are less likely to voluntarily graph key financial indicators.

Banks operate in a highly regulated industry that is under the attention of media and regulators. Therefore, their potentially selectiveness on financial performance graphs could be driven by the potential political costs that they could incur by drawing attention to their high performance. Hence, we expect that:

Hypothesis 2a: Financial performance indicators are less likely to be graphed in the annual reports of companies with good, rather than bad, financial performance.

Hypothesis 2b: Financial performance indicators are less likely to be graphed in the annual reports of banks with good, rather than bad, performance in terms of the variable graphed.

As any other organizational practice, financial reporting practices do not develop in a vacuum, due to the firms’ embeddeness in a nexus of formal and informal rules, rather they are the result of macro social processes (e.g., DiMaggio & Powell, 1983) and are likely to reflect the underlying environmental influences that affect firms in different countries (e.g, Ball et al., 2000; Haniffa & Cooke, 2002; Archambault & Archambault, 2003). National accounting practices vary because of environmental and cultural factors. Thus, the adoption of graphs, as well as the type of variable graphed is likely to vary across countries (Beatty and Jones, 2000b, Ianniello, 2009).

Hypothesis 3a: Graphs usage and the variables graphed will differ among banks belonging to different countries.

Nobes (1983, 1998) classified international accounting systems into micro, Anglo-Saxon practices, and macro, continental European practices. Micro-based accounting practices are typified by comparatively weak governmental influence, strong accounting professions and comparatively active equity markets. The focus is on the provision of a ‘fair’ presentation of the accounts and the portrayal of economic reality for the benefit of investors. By contrast, macro accounting practices are typically characterized by strong governmental influence on accounting, relatively weak accounting professions, and less active equity markets. In micro-based countries, financial reporting is geared up to satisfying investors comparatively more than in macro-based countries where the needs of alternative annual reports’ users, such as other stakeholders, tend to dominate. These pressures may make the financial performance representation in the annual report relatively more important in micro-based, rather than macro-based, countries. Based on this classification, the U.K. may be classified as micro-based country, whereas France, Germany, Italy and Spain have a macro orientation (Nobes, 1983). Hence, we expect that:

Hypothesis 3b: Financial performance indicators are more likely to be graphed in the annual reports of UK banks than in those of continental European banks.

3. Research method

3.1 Sample and data gathering

Using the database Bankscope, we selected the European financial banks based in the largest five European economies and listed during the whole 2006 year. Listed subsidiaries of a holding bank that was already in the sample, financial firms that were not commercial banks and firms whose annual report was no publicly available were dropped. The final sample comprises 47 commercial banks, listed in the top five European countries: Germany, France, UK, Italy and Spain. Among these 47 commercial banks, there are also 11 banks now considered, after the global financial crisis, systemically important financial institutions[2].

We gathered consolidated annual reports from the firms’ websites and collected data about the graph title, the graph location in the annual report (page and presence in the highlights section), the graph category-topic, the graph type (e.g., column), and whether the graph refers to the whole group or to segmental areas, such as specific business divisions, countries and/or subsidiaries The data checklist was first pilot-tested on 9 banks to ensure clarity and completeness in the data collection among the authors.

After the data collection process, in order to group graphs into main categories, we first identified general titles of the graphs, then broader keywords and, at the end, the main topic/category each graph belongs to.

Banks’s market and accounting performances were collected from Bankscope database.

3.2 Method

Our main variable is represented by the number of specific variables graphed.

Following previous literature (Beattie, Jones, 1992; 2001), selectivity was investigated by testing the association between the number of the financial performance variables graphed and the increase in bank’s overall financial performance (Hypotheses 1a and 2a), as well as by testing, for each financial performance variable, the association between the number of times a specific variable graphed and the increase in the specific financial performance variable (Hypotheses 1b and 2b). Hypotheses 3a and 3b were tested by analyzing the association between the country a bank belong to with the total number of variables graphed, as well as with the topic of variables graphed.

Moreover, we have investigated impression management practices by providing some anecdotal evidence of key financial variables’ graphs that present material measurement distortion. Measurement distortion occurs when the variations of the measures depicted in the graph are not proportional to the variations of the real data and, thus, there is a violation of the fundamental principle of the graph construction (Tufte, 1983). Based on previous literature (Beattie and Jones, 2008), we have measured graph distortion with the graph discrepancy index (GDI) developed by Taylor and Anderson (1986), which is a variation of Tufte’s (1983) lie factor.

GDI=AB-1×100

Where:

GDI = Graph Discrepancy Index

A = height of last column-height of first columnheight of first column×100

B = percentage variation of the performance measure

Positive values of the GDI express an exaggeration of a trend while negative values an understatement[3]. However, to understand whether the distortion has been favorable or unfavorable for the firm, we have to analyze the nature of the variable and its trend. In general, the distortion is favorable for the firm when the GDI is higher than 0 and there has been an upward trend of the variable graphed or when the GDI is lower than 0 and there has been a declining trend of the variable graphed. On the contrary, the distortion is unfavorable for the firm when the GDI is lower than 0 and there has been an increase in the value of the variable graphed or when the GDI is higher than 0 and there has been a decrease in this value.

As a GDI within the values of +/-10% does not seem to affect the audience perception of a variable trend depicted in a graph (Beattie, Jones, 2002), we have considered this level as the cut-off level to split between material and not material distortion. We have also taken into account a +/-5% GDI level as our threshold for material distortion, following other previous papers (Beattie, Jones, 1992; Mather et al., 2000).

4. Key findings

Descriptive Statistics

Across the five European countries analyzed, we identified 42 banks out of 47 (89.86%) that inserted graphs in their 2006 annual reports. Graphs adoption in banks is in line with what found by previous studies in non-financial firms (Beattie et al, 2008.). We identified a total number of 1,050 graphs and 1,243 variables graphed. This misalignment is due to the fact that 12% of the total graphs reported more than one variable (details are reported in Table 1). By comparing and contrasting the five European countries analyzed, we found that the number of graphs, as well as the number of variables graphed is higher in Spain and the UK. In such countries, banks graphed on average, respectively, 78 and 28 variables; whilst Italian, German and French banks represented graphically less than 20 variables (see table 1). Hereafter the focus of the paper will be on the variables graphed.