Annual report graphic use: a review of the literature.

Other empirical studies have measured financial graphics and compared the image to the actual data. We have learned that almost half of the reports in one study included at least one graph that was incorrectly prepared (Johnson, Rice, & Roemmich, 1980), that one in eight graphs in annual reports has errors in numbers and one in three has design errors that distort the numbers (Moen, 1990), that about one half of the graphs were misleading in the 35% of annual reports that had graphs (Courtis, 1997), that the manipulation is three times more likely to exaggerate, rather than underestimate, an upward trend (Beattie & Jones, 1992a, 1992c), that 30% of key financial graphs contain significant measurement distortion (Beattie & Jones, 1992a), and that at least 1 in 10 graphs in annual reports has an altered vertical scale that projects a more positive image than the real situation (Burgess, 2002). Somewhat similar to the Burgess findings, another study (Frownfelter-Lohrke & Fulkerson, 2001) determined that of 270 annual reports from 74 U.S. and non-U.S. companies, about one half did not include a scale and 17% did not begin with a zero baseline.

In addition to studies of the prevalence of graphs that contain distortion, we also see evidence that graph use may be contingent on favorable financial performance (Beattie & Jones, 1992c, 1999a, 2000b; M. J. Jones, 1994). Literature that expands the positive distortion of graphs beyond annual reports to other forms of financial reports includes IPO prospectuses (P. Mather, Ramsay, & Steen, 2000), CEO changes (Godfrey, Mather, & Ramsay, 2003), and proxy statements (Bannister & Newman, 2006).

Later in this article, following the Annual Reports and Communication section, is additional discussion of financial performance and how it relates to both writing and graphing.

Other experiments tested the effect of perception on various forms of graphs. Beattie and Jones (2002a) specifically focus on graph scale and how it affects the angle (the "slope parameter") of lines in line graphs. Their laboratory study concludes that the degree of the angle leads to bad judgments of corporate performance. They also experimentally test six levels of distortion in vertical bar graphs and find that no distortion in excess of 10% should be allowed (Beattie & Jones, 2002b). From a series of three experiments, we also learn that improperly designed graphs can mislead users and can cause them to make different decisions than they would have made if the graphs had been correctly prepared (Arunachalam, Pei, & Steinbart, 2002a: also see Arunachalam, Pei, & Steinbart, 2002b; Plumlee, 2002).

Some research examines just the dimensionality of graphs. When comparing two- and three-dimensional graphs, the two-dimensional graphs were more reliable--in both speed and accuracy--than were the three-dimensional graphs (Addo, 1994). Another study (Tractinsky & Meyer, 1999) draws a similar conclusion regarding the accuracy of two-dimensional graphs but also concludes that when the goal is to create a favorable impression, people preferred and created graphs with more three-dimensional depth.

Based on the considerable latitude graphic designers have in deciding how to present financial data in annual reports and its role in affecting readers' perceptions of the corporation, one might assume that public accounting auditors would measure and comment on such distortion. The next section looks at these issues.

ISSUES RELATED TO ACCOUNTANCY

Graphs in annual reports should accurately reflect data. Designers of annual reports may prepare graphs, and internal auditors may be involved as well. In the auditing function, accountants verify the accuracy of the data. Precisely how auditors treat or examine the graphs is an area that is still emerging, as we shall see.

Rather naively at one time, the National Association of Accountants (Andersen, 1983) proposed using graphs to represent accounting data because they can be reliable, can be understood, and can attract and hold attention (pp. 3-4), but the report barely mentions potential misunderstanding and misrepresentation of data. A 1978 Harvard Business Review article (Blake, Warner, & White, 1978) was not much more insightful. By 1995 (McCullar, 1995), 1999 (Fulkerson, Pitman, & Frownfelter-Lohrke, 1999), and 2000 (McNelis, 2000), the suggestions to accountants had improved substantially. An especially thorough review of guidelines for graphics in financial reports is directed toward accounting educators and their students (Hill & Milner, 2003).

Currently, auditors of U.S. annual reports are required to examine both the narrative and the financial sections, though the guidelines for review of the narrative section are brief and general. The Statement on Accounting Standards Number 8, "Other Information in Documents Containing Audited Financial Statements" ("AU Section 550," 1975), says the auditor must read "other information" to determine if it is "materially inconsistent" with the financial reports: this auditor's review includes the manner of presentation and the data. The "other information" includes graphs (Burgess, 2002). The Governmental Accounting Standards Board Statement 34 has parallel standards (Chase & Shoulders, 2003).

Articles with critiques of and suggestions for financial and accounting related graphs abound (e.g., Bannister & Newman, 2006; Lynch & Golen, 2002; Sugden, 1989; Werts, 2004; Wilson & Stanton, 1996).

The extent of the review of graphs in the narrative section is not well defined (Burgess, 2002, p. 52), but guidelines are emerging (Johnson et al., 1980: Taylor & Anderson, 1986). Auditors who are able to point out the existence of graphical alterations perform a real service to their clients (Burgess, 2002). Continuing research related to graph use and distortion in annual reports appears to be needed and important.

INTERNATIONAL IMPLICATIONS

The third major cluster of literature on graphs and annual reports includes international implications. Many articles review reports for a specific country or compare annual reports of different countries. (Numerous studies already cited fall into this category: Beattie & Jones, 1992a, 1992b, 1992c--United Kingdom; Beattie & Jones, 1999a, 1999b--Australia; Beattie & Jones, 2000a, 2000b--United Kingdom; Beattie & Jones, 2001--six countries; Beattie & Jones, 2002a, 2002b--United Kingdom: Canniffe, 2003--Ireland: Chevalier & Roy, 1993--Canada; Coles & Rowley, 1997--United Kingdom: Courtis, 1997--China; Frownfelter-Lohrke & Fulkerson, 2001--U.S. and non-U.S. companies: Godfrey et al., 2003--Australia; Green et al., 1993--Ireland; Platts & Tan, 2004--United Kingdom; Canadian Institute of Chartered Accountants, 1993--Canada.)

The countries most often reviewed (after the United States) are Canada, the United Kingdom, Australia. China. and other countries in Europe, and often the companies are multinational and/or employ English as a major language for their reports.

Graphic use and distortion appear to vary by country. In 1996, a study of 300 reports from top companies in Australia, Europe, and the United States found that French companies are inclined to exaggerate instead of underestimate trends, that in the United Kingdom graphs violate basic design principles, and that in the United States reports are more likely to include key performance variables in graphs if their financial situation is positive (Roy, 1998). Additional discussion of the effects of positive financial performance appears following the next section.

In 1997, the Institute of Chartered Accountants in England and Wales determined that graphs in annual reports in six countries are often poorly designed; the graphs with the most distortions were from France, the United Kingdom, and the United States (cited in Burgess, 2002, p. 45). Also using six countries (Australia, France, Germany, the Netherlands, the United Kingdom, and the United States), Beattie and Jones (2001) looked at 50 companies and concluded that there were substantial differences by country, with Germany being an outlier with noticeably fewer graphs of earnings per share (EPS) variables and a greater raw number of graphs.

Focusing on just Canadian annual reports, a thorough review of graphs concluded that a vast majority of Canadian companies use financial graphics and that a substantial number of the annual report graphics were potentially misleading because of their construction and design (Canadian Institute of Chartered Accountants, 1993, p. 124). The monograph compares eight studies of graphic distortion and then presents a list of 53 graphic discrepancies supported by the literature (p. 190). This Canadian extensive review of potential misrepresentation is often cited as it details the variety of distortion techniques.

Two surveys of reports from Hong Kong public companies concluded that 38% and 35% of the companies included graphics and that, in the latter survey, about one half of all the graphs violated sound principles (Courtis, 1997). This article extends earlier work by Beattie and Jones (1992c) but reports less graphic use than in the United States (as cited earlier).

Surveys of graphs used in Irish annual financial statements were used to selectively highlight specific areas where performance had improved (Green et al., 1993). As is often the case, the question is raised whether this positive distortion is intentional or a passive by-product of the publication process.