Graphical sleight of hand: how can auditors spot altered exhibits that appear in annual reports?

EXECUTIVE SUMMARY

* GRAPHS IN CORPORATE ANNUAL REPORTS may have an altered vertical scale and often report a more favorable picture than warranted by the underlying financial information. Auditors can perform a real service for their clients by pointing out the existence of graphical alterations in financial statements and helping companies avoid potential "earnings management" problems.

* AUDITORS ARE NOT REQUIRED to corroborate or test additional financial information reported in corporate annual reports, such as data in ratios and graphs. However, auditors should consider whether such information is materially inconsistent with the financial statements.

* A PROPERLY CONSTRUCTED GRAPH that begins with a vertical scale of zero has no disparity between the picture and the numbers. There are high-tech and low-tech solutions auditors can use to detect vertical scale alterations less obvious to the eye.

* ASSESSING MATERIAL INCONSISTENCIES involves auditor judgment. Auditors should consider the materiality guidance outlined in the SEC's SAB no. 99 in preparing financial statements and performing audits of them.

* PRACTITIONERS ENGAGED TO EXAMINE MD&A should explore the materiality implications of graphical alterations and to what extent they lack reliability or representational faithfulness. Intentionally altered graphs mask trends and the real-life events they purport to convey.

At least one in 10 graphs in corporate annual reports has an altered vertical scale that depicts a more positive picture than the underlying financial information warrants. The vertical scale modifications can change readers' judgments, alter trends and may be deemed intentional by regulators if they seem to magnify favorable information. Auditors can perform a real service for their clients by pointing out the existence of graphical alterations in financial statements and helping companies avoid potential "earnings management" problems with the SEC and investors.

Evidence of altered graphs in corporate annual reports spans more than a decade. Vertical scale alterations are reported around the world. According to a 1997 study of 300 annual reports from six countries by the Institute of Chartered Accountants in England and Wales, the design and construction of graphs in annual reports are often poor. Measurement distortions, including vertical scale alterations, were the worst in annual reports from France, the United Kingdom and the United States.

WHAT ARE AUDITORS LOOKING AT?

Auditors review financial graphs for material inconsistencies with the financial statements as part of the process described in Statement of Auditing Standards no. 8, Other Information in Documents Containing Auditing Financial Statements, or in a separate engagement to examine the MD&A where most graphs appear as set forth in Statement on Standards for Attestation Engagements no. 8, Management's Discussion and Analysis. The graphs portray trends and forward-looking information investors want in annual reports and that the SEC requires for displays of performance in proxy statements (Executive Compensation, 17 CFR 229.402(1)) and registration statements of mutual funds (Registration Form Used by Open-Ended Management Investment Companies, 17 CFR 239.15A and 17 CFR274.1 lA).

Auditors are not required to corroborate or test additional financial information reported in corporate annual reports, such as data in ratios and graphs. However, SAS no. 8 says an auditor should read and consider whether such information, or its presentation, is materially inconsistent with information appearing in the financial statements (see "Regulatory Considerations," page 51).

HOW TO SPOT ALTERATIONS

The most obvious vertical scale alterations display broken scales or gradations that begin at a number greater than zero. The observer can detect less obvious alterations by looking at the graph itself and its numbers. If the numbers appear about the same each year but the bars do not, then the graph was altered to present a different picture.

Exhibit 1, page 47, illustrates how altering the vertical scale can change what a graph is supposed to represent. The graph on the left properly displays a vertical scale beginning at zero. The one on the right begins the vertical scale at $200. The vertical scale alteration magnifies the small changes between years and mistakenly portrays variability in otherwise stable data. Even if the vertical scale was omitted, an auditor glancing at the numbers and the graph could detect a disparity between the amount of change portrayed in the picture (measured in centimeters) and the amount of change reported in the numbers (measured in dollars). A properly constructed graph that begins with a vertical scale of zero has no disparity between the picture and the numbers.

Both charts in exhibit 1 were prepared using Microsoft Excel's Chart Wizard (available by clicking on the chart icon at the top menu of Microsoft Excel). After following a few steps that include selecting the type of chart desired (for example, bar graph, pie chart) and highlighting the source data to be graphed (axis information and financial information), the Chart Wizard inserts a graph into the worksheet. Chart Wizard automatically selects a vertical scale origin greater than zero if the data are relatively stable and unchanging. This common charting practice of altering the vertical scale is done in an effort to magnify changes thought to be too small to see.

There are high-tech and low-tech solutions auditors can use to detect vertical scale alterations less obvious to the eye. The high-tech solution is to recreate the client's graph using software such as Chart Wizard. The client's choices (color, size, dimension, type) can be duplicated through chart design or by editing with a right-click of the mouse over any aspect of the chart. To ensure the vertical scale begins at zero, the auditor should right-click on the vertical scale. Scale options should be set at a minimum value of zero (or a negative value if appropriate for the data), a maximum value near the highest value being graphed--such as in exhibit 1 on the left-hand side--and an X axis that crosses at zero. The auditor's newly created graph should portray the same visual as the client's graph if the latter is properly constructed.

Differences in the two charts' vertical scales can also be detected using a low-tech method. Bring out a ruler and compare the change in centimeters with the change in numbers between two bars in the client's graph. Any difference in the change in centimeters and numbers indicates a vertical scale alteration. Measuring vertical scale alterations is illustrated in the formula below, the graphical discrepancy index, an old method developed by statisticians and still used in graphical construction:

Graph discrepancy index = 100 [[percentage change in centimeters/percentage change in dollars] - 1]

Examine the graph on the right in Exhibit 1. The graph portrays a 101.2% increase between years X2 and X4 measured in centimeters: 100 [(8.25 cm-4.1 cm)/4.1 cm]. The increase in dollars between the years is only 1.96%: 100 [($208-$204)/$204]. The alteration is measured as, 100 [[101.22/1.96] - 1] equating to 5,064. High amounts of alteration occur when the vertical scale omits a large portion of the dollars from the graph. In the graph on the right in exhibit 1, because the scale is set at $200 the first $200 of each bar is omitted from the display; the bars show a range of only $4 to $8 from year X2 to year X4. This representation creates a large distortion in the picture. For example, the difference between years X2 and X3 looks dramatic on the graph to the right, but on the properly constructed graph on the left, the difference is barely noticeable. (Note that production of the exhibits for this article does not permit an exact representation of these measurements.)

ALTERATIONS DISTORT PERCEPTIONS

The author reviewed 70 annual reports from U.S. companies (50 Fortune 500 companies and 20 smaller public companies representing a cross-section of industries) and found 64 of them displayed graphs. Thirteen had obvious vertical scale alterations ranging from 55% to 913%, with a median alteration level of 153%. Exhibit 2, page 48, displays the range of alteration effects found in the annual reports. A hypothetical data set is illustrated with distortion levels of 0%, 56%, 162% and 976%. All distortion levels found in the annual reports change the information portrayed in the graph and hinder perceptions.

Two questions will assist auditors in applying these survey results to their reviews of graphs in annual reports: Is there a materiality threshold level of alteration to avoid, and do the types of data (stable vs. volatile) make a difference? Auditors need to know whether there is some acceptable level of alteration and whether the common charting practice of magnifying small changes in constant, unchanging data is a good idea.

The author studied 80 financial statement readers' perceptions (the survey participants were 40 stockbrokers, bankers and CPAs and 40 third- and fourth-year business students) of bar graphs with varying alteration (0%, 5%, 25% and 55%). Very minor alterations were studied to determine whether slight changes in the graph made a difference in readers' perceptions. The stable trend bar graphs viewed by the participants had relatively constant numbers, which appeared as a straight line when graphed; the variable trend bar graphs showed erratic numbers that appeared in a zigzag pattern. Each reader was asked to describe "the growth from year X2 to X4" in net income on a seven-point scale ranging from very little change (one) to very large change (seven) in all graphs that were part of the survey.

A separate group of CPAs measured the true growth in net income by looking at raw numbers of stable and variable trends taken from the financial statements. The real growth in the numbers was compared with the perceived growth in the numbers when properly graphed to ensure the numbers and the illustration communicated the same thing--no material inconsistencies. Then the author altered the vertical scales in the stable and variable trend graphs so the reader could assess the impact of the distortion.

SLIGHTEST CHANGE CAN BE TROUBLE

The responses by survey participants indicated even the slightest manipulation in a graph led people to think there was more growth in the numbers than was actually present. This trend continued as the graphical manipulation increased--larger alterations in the picture created greater perceptions of growth. There was a steady increase in alteration effects with each alteration level (readers looked at distortions of 5%, 25% and 55%). The average participant perceived 0.3 points to 1 point higher growth (on a seven-point scale in the altered graphs than nonaltered graphs. The effects were all statistically significant when comparing each individual's response with altered and nonaltered graphs at all discrepancy levels and data trends. The results indicated vertical scale alterations, even when barely noticeable (5%), affected perceptions. Virtually no amount of graphical manipulation was acceptable; any alteration in the vertical scale caused a sizable difference in readers' perceptions.

A second finding of the survey indicated manipulating a company's stable

trend numbers was worse than manipulating volatile trend information because any alteration in stable data gave the mistaken appearance of volatility. Readers perceived 0.56 points to 1 point more growth in the picture when the stable data were altered, and only 0.3 points to 0.64 points more growth in the picture when the variable data were changed. Companies should not alter the vertical scale of stable trends to magnify small changes because readers are sensitive to any modifications made.

The author compared the reactions of both seasoned and novice financial statement readers. As might be expected, the professionals were less affected by graphical alterations than students. When asked whether they used the numbers, the pictures or a combination of both, most professionals said "mostly numbers" and most students said "a combination of numbers and pictures."

The magnitude of the numbers may affect perceptions. The numbers displayed in the stable and variable trend graphs were larger with each level of alteration. Readers may perceive larger changes with bigger numbers, even though the percentage change between years remains small.

WHAT TO DO ABOUT INCONSISTENCIES

Auditors should consider the possibility that when companies alter graphs to present the best picture possible, this may create impressions of earnings manipulation, which can affect the quality of a company's earnings reporting and ultimately lower market values. Practitioners engaged to examine corporate financial statements should explore the materiality implications of altered graphs and whether vertical scale alterations create material inconsistencies. Depending on the circumstances and the significance of any inconsistency found, the auditor may conclude the financial statements, the annual report or graphs require revision. If the client does not make the appropriate changes, the auditor can decide to revise his or her report to include an explanation describing the material inconsistency, withhold the use of his or her report in the document or formally withdraw from the engagement.

Exhibit 1: Large Alteration Effect

Hypothetical company
Net income--0% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Hypothetical company
Net income--5,064% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Note: Table made from bar graph.
Exhibit 2: Alteration Effects Found in Annual Reports

Hypothetical company
Net income-0% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Hypothetical company
Net income-56% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Hypothetical company
Net income-162% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Hypothetical company
Net income-976% distortion

X1  $205
X2  $204
X3  $209
X4  $208
X5  $205

Author's Study

The author reviewed 70 annual reports from U.S. companies (50 Fortune 500 companies and 20 smaller public companies representing a cross-section of industries) and found 64 of them displayed graphs. Thirteen had obvious vertical scale alterations ranging from 55% to 913%, with a median alteration level of 153%.

Regulatory Considerations

Assessing material inconsistencies involves auditor judgment. FASB Concepts Statement no. 2, Qualitative Characteristics of Accounting Information, defines materiality as "the magnitude of an omission or misstatement of accounting information that, in light of surrounding circumstances, makes it probable that the judgment of a reasonable person relying on the information would have been changed or influenced by the omission or misstatement." The SEC takes a more conservative position on materiality than FASB. Staff Accounting Bulletin (SAB) no. 99, Materiality, specifies materiality guidance in preparing financial statements and performing audits of those financial statements. The SEC warns against overreliance on materiality thresholds to the exclusion of qualitative considerations, and reporting on any intentional errors, regardless of materiality. SAB no. 99 specifies that any intentional error in the financial statements, regardless of materiality, is unacceptable. (For a full discussion of SAB no. 99, see "Earnings Management and the Abuse of Materiality," JofA, Sept.00, page 41.)

Statement on Standards for Attestations Engagement (SSAE) no. 8 outlines considerations of qualitative factors and advises practitioners engaged to examine and report on MD&A that "assessing the significance of misstatement for some items in MD&A may be more dependent upon qualitative than quantitative considerations. Qualitative aspects of materiality relate to the relevance and reliability of the information presented." Practitioners engaged to examine MD&A should note that graphical alterations have materiality implications to the extent they lack reliability or representational faithfulness.

Intentionally altered graphs mask trends and the real-life events they purport to convey. Graphical alterations can cause reader misperceptions. Whether such graphs create material inconsistencies with the financial statements is a matter of auditor judgment, requiring consideration of existing authoritative guidance and the author's practical instruction.

DEANNA OXENDER BURGESS, CPA, PhD, is an assistant professor of accounting at Florida Gulf Coast University in Fort Myers. Her e-mail address is [e-mail].