Speech by SEC Staff:
Remarks Before the 2006 AICPA National Conference on Current SEC and PCAOB Developments

by

Todd E. Hardiman

Associate Chief Accountant, Division of Corporation Finance
U.S. Securities and Exchange Commission

Washington, D.C.
December 12, 2006

As a matter of policy, the Securities and Exchange Commission disclaims responsibility for any private publication or statement of any SEC employee or Commissioner. This speech expresses the author's views and does not necessarily reflect those of the Commission, the Commissioners, or other members of the SEC staff.

Note: You can access the slides referenced in this speech at http://www.sec.gov/news/speech/slides121206teh.pdf

Can a large error be immaterial? Is it appropriate to consider the offsetting effects of multiple errors in concluding whether a single error is material? My first objective today is to share how we've addressed those questions when they've come up in registrant fact patterns. But I do have a second objective today — to begin talking about quarterly materiality. By starting this dialogue today, my hope is that you as members of the financial reporting community will engage in the discussion and help us work toward a common understanding. Before I begin, let me set some parameters to help you understand the context of my remarks.

You already know parameter number one: materiality judgments must consider all relevant facts and circumstances or in accounting shorthand all relevant quantitative and qualitative factors. What's most important today is parameter number two: If you hear anything in my remarks today that you believe suggests otherwise, refer back to parameter number one.

Can large errors be immaterial? The short answer is yes, in the right situation. But that's the end of the story. Let me start at the beginning.

Staff Accounting Bulletin No. 99 provides guidance on how to make materiality judgments. It provides guidance that helps answer the question: Can small errors be material? The example cited in the SAB is financial statement errors below 5%. And the guidance it provides includes an illustrative list of qualitative considerations that may cause a quantitatively small error to be material. The question we've addressed is the inverse of the example provided in the SAB. It goes like this: If small errors can be material, can large errors be immaterial? Our answer is yes; in certain circumstances it is possible for a large error to be immaterial. But the reality is that we just don't see those circumstances all that often. Here's an extremely simplified version of what we do see.

Registrant has an error that misstates pretax income by 20%. They look to the illustrative qualitative considerations cited in the SAB that can cause a small error to be material and observe that none of the considerations apply. They then conclude that the absence of those qualitative considerations supports their view that the 20% error is not material.

We do not believe that this type of analysis is sufficient. It does not answer the question: why are financial statements that misstate pre-tax income by 20% reliable? To illustrate this point using two of the qualitative considerations listed in the SAB, the fact that the 20% error does not affect debt covenants or executive compensation does not cause the 20% error to be immaterial. To support an assertion that a large error is immaterial, registrants need to look beyond the qualitative considerations listed in the SAB that identify when a small error may be material. They need to identify the considerations that cause the financial statements to be reliable notwithstanding the large error.

We took a stab at trying to identify the types of considerations that might cause a large error to be immaterial. You might find it instructive that we could only come up with two examples. The first is a break-even year. To use the illustration on your slide [see slide 67], if 2003 net income of $100,000 were misstated by $20,000 or 20%, it does not appear that the 20% misstatement by itself would cause the error to be material. But let me give two cautions about break-even years. First, a registrant that regularly has razor thin margins or net income is not what we mean when we refer to a registrant with a break-even year. Second, it may be difficult to objectively support that the most recently completed fiscal year is in fact a break even year, particularly when the evidence indicates that the registrant's turnaround will involve multiple years or when there is no clear trend evident from the registrant's historical earnings.

The second example fits best into the category of speculation on our part, rather than a live fact pattern we've actually seen. It seems like there might be circumstances where a large error relative to a discontinued operation that has been sold could be immaterial. To be clear, I'm not stating a conclusion, merely a possibility. Errors in discontinued operations can and often have been material. But it seems like circumstances could exist that an error related to a disposed discontinued operation that would not have affected the registrant's representations and obligations or the selling price it received could be immaterial.

Of course, in both examples I cite, break-even years and discontinued operations that have been sold, the registrant would need to consider all relevant quantitative and qualitative factors in reaching its materiality judgment. I should also caution you from attaching any significance to the 20% I used in this example. I used it as an illustration. I was not trying to define through my example what we think a large error is.

The second question I'll address is whether an individual error can be immaterial simply because of the existence of an offsetting error of equal magnitude. Let's look at an overly simplified example to illustrate the question. Assume a registrant has two errors. One error overstates net income by 30%; another error understates net income by 28%. For simplicity, also assume that both errors would be considered material. The question asks: Is it appropriate for the registrant to conclude that no adjustment need be made because on a combined basis net income is only overstated by 2% and the combined error would not be deemed material?

We think the answer is no. We believe registrants must evaluate each error individually, irrespective of its effect when combined with other errors. That evaluation should consider the effect of the individual error on each financial statement line item, including subtotals and totals, like gross profit and pretax income on the income statement and current assets and total liabilities on the balance sheet. If a registrant concludes that an individual error is material, irrespective of its effect when combined with other errors, they would need to restate their financial statements.

In a fact pattern we evaluated this year, a registrant had over 40 individual financial statement errors. Under their methodology, an individual error that was large relative to net income could be considered immaterial if two conditions existed (1) the combined effect of all 40 errors was quantitatively small and (2) none of the illustrative qualitative considerations identified in SAB 99 were met. In essence, they believed that the combined effect of all errors was a qualitative factor they could consider in evaluating whether the individual error was material. We did not believe that this methodology was appropriate.

Simply put, luck is not a qualitative factor that enters into a materiality judgment. Whether you have 2 errors or 40, the fact that those errors happen to offset is not a qualitative factor that we believe bears on the materiality evaluation of the individual error. By considering the offsetting effects of errors, this registrant failed to consider its surrounding circumstances. They did not have a single quantitatively small error as their methodology appeared to suggest. They had multiple errors, many of which were large relative to financial statement line items, subtotals and totals. In these circumstances, we believe each error had to be evaluated separately, not just on a combined basis. If the registrant concluded that an individual error was material, they would have to correct that error even if the aggregate impact on financial statement line items, subtotals, and totals was not material.

The registrant offered one more twist for us to consider. They indicated that if they had to restate their financial statements because one error was material, they would correct all 40 errors. If all 40 errors were corrected, the amounts on the face of the financial statement as originally reported would not differ significantly from the amounts in an as restated presentation. Would we still require restatement in this circumstance? Our answer was yes. How you report an error correction does not dictate whether an error is material. Using my initial example, if there is a substantial likelihood that a reasonable investor would consider the 30% error important, the fact that it is offset by the 28% error does not render it immaterial. The financial statements would need to be restated and that restatement would need to include disclosure that describes and quantifies both errors.

That brings me to my last topic — quarterly materiality. One of the first steps in assessing quantitative materiality is to consider the relative size of an error. By relative size, I mean the ratio of the error to a financial statement line item, like pretax income. If you look to your slide, it's the "x" or what some refer to as the denominator. For annual reporting periods, registrants often determine the relative size of an error by comparing it to annual amounts like annual pretax income and annual gross margin. For quarterly financial statements, the question is this: do you compare the error to quarterly amounts, like quarterly income, or annual amounts like estimated annual income?

A few weeks ago, I started to draft a speech that answers that question. But as I continued to discuss that speech with others, I came to believe that such a singular focus runs the risk of appearing to boil materiality down to the mechanics of a simple ratio. And yet, we know the materiality concept encompasses much more than that ratio.

So at the outset, let me eliminate that risk by reiterating what you already know. Materiality judgments are not limited to quantitative comparisons, they also include qualitative considerations. And if the judgment of a reasonable person would have been changed or influenced by the omission or misstatement of information, then that omission or misstatement is material regardless of what ratio you may consider in reaching your materiality conclusion. Nothing in my remarks today changes that fact.

But I do think it is important to begin a dialogue about that ratio. For one, it is an aspect of quantitative materiality. And two, there is something to discuss; we just do not believe the discussion is as broad as some have tried to cast it. Let me explain.

For accountants, one of the sources of this debate is Opinion 28, paragraph 29. As you can see from your slide, paragraph 29 says that in determining the materiality of an error, relate it to the estimated annual income and also to the trend of earnings. So what does it mean? And, what error is it referring to? Is it referring to errors carried forward from prior period financial statements, errors that originated in the current interim period financial statements, or all errors?

Some see the reference to annual income and believe that all errors in quarterly financial statements should always be assessed relative to annual amounts. They note that the Accounting Principles Board concluded in Opinion 28 that quarterly periods should be viewed primarily as an integral part of an annual period. And they highlight that there is often less rigor that goes into preparing quarterly numbers than there is for annual numbers. After all, there are shorter filing periods for quarterly financial statements than there are for annual financial statements. And quarterly financial statements are subject to review by independent auditors rather than audit. For these reasons, they believe that all errors in quarterly financial statements should always be assessed relative to annual amounts. This is the view that we believe casts the dialogue too wide. It has the practical effect of suggesting that you only need to get your annual numbers right. Any error that is immaterial to the year is fair game to book in a quarter, or even leave uncorrected, regardless of how it distorts or misstates quarterly results. And for that reason, we do not believe it complies with GAAP.

But there are two other views that we believe warrant further discussion. Both adhere to the requirement that an error that originates in a quarterly period must be evaluated relative to quarterly amounts of the quarter in which it originated. Where they differ is in their treatment of prior period errors.

One of those two views holds that all errors in quarterly financial statements, both originating errors and prior period errors, should always be assessed relative to quarterly amounts. Their mantra is "quarters matter" and they matter for a variety of reasons. First, they cite the move toward real time reporting that has gained significant momentum in the 33 years since Opinion 28 was issued, most recently in the changes to our current reporting framework under Form 8-K and the call in Section 409 of the Sarbanes Oxley Act for real time issuer disclosure. And second, they believe that investors, analysts and other users place equal or greater emphasis on quarterly results than annual results. Yet, one of the problems of this view is that it has the practical effect of requiring materiality judgments in annual financial statements to be made on the basis of what would be important to quarterly financial statements.

The other view holds that paragraph 29 of Opinion 28 relates to prior period errors. They note that whenever Opinion 20, and its successor Statement 154, reference "correction of an error" the context is prior period errors. And they read Opinion 28 in the same way. As originally issued, Opinion 28 had language in paragraph 25 that suggests its discussion about error corrections relates to prior period errors or in its words, errors in previously issued financial statements. On that basis, this view postulates that prior period errors could be viewed and assessed differently than current period errors. If a prior period error is not material to estimated annual amounts, they believe the error can be corrected in the quarter, with disclosure if the out-of-period amount would be material to the quarter. They also believe that since Opinion 28 does not reference current period errors, those errors would need to be evaluated relative to quarterly amounts of the quarter in which they originate.

Some might think this last view bridges the gap between the first two views. By using annual amounts to evaluate corrections of prior period errors, they think it captures the concept that quarters are an integral part of an annual period. And by using quarterly amounts to evaluate corrections of originating errors, they think it also captures the concept that quarters matter. At first glance, this view may also appear to resolve a number of the problems of the first two views, and it was the view that I originally thought I'd be supporting in this speech. But further thought has shown that this view is not necessarily a panacea. Perhaps its most notable weakness is that it suggests that in some circumstances it's okay for quarterly financial statements to be materially misstated so long as you tell investors they are. It also has the notable problem of being overly mechanical, such that the reason a prior period error is being corrected in a particular quarter doesn't seem to enter the analysis.

As our profession considers these views, we can't forget the views of non-accountants. Ask a non-accountant: Is an error immaterial to a quarter simply because it is expected to be immaterial to the year?" I think they'd say, 'No, look at the headlines.' Every quarter, company stock prices change depending on whether their quarterly results hit or miss analysts' estimates for that quarter. If you look, my guess is that you'll be hard-pressed to find an article that says something like: XYZ Company stock price remained unchanged today despite missing quarterly estimates because company believes it's on target to meet annual income estimates. That headline may be out there, I just haven't seen it.

Stepping back, it seems like the quarterly materiality issue is driven at least in part by the mechanics of financial reporting. Simple math says the sum of the four quarters must equal the year. But the fact remains that materiality judgments of annual financial statements do not consider quarterly effects. And so long as they don't, it seems like we'll struggle with the mechanics of how to account for errors that are immaterial to an annual period, but may be material to a quarterly period.

Unfortunately, I do not have the solution for this struggle, despite the fact that several of us have been trying to understand and discuss the issue for some time, especially since SAB 108 came out answering some of the questions about quantifying errors in annual financial statements. But I hope that this discussion will not be the last on the topic. That you as members of the financial reporting community will continue the discussion, so that we can all come to an understanding of how to think about materiality in quarterly financial statements in a way that best meets the needs of investors and the capital markets.