Speech by SEC Staff:
Remarks at the University of Southern California Leventhal School of Accounting SEC and Financial Reporting Conference

by

Scott A. Taub

Deputy Chief Accountant; Office of the Chief Accountant
U.S. Securities and Exchange Commission

Los Angeles, CA
May 27, 2004

Good morning. It's an honor to be here for the second year in a row to address this audience. Last year when I was here, I started out by telling you how complicated my life had been recently. Amongst other things, I mentioned that I was the only one left in senior management in the Office of the Chief Accountant, and that I had a one-month old son at home. I then went on to talk for 20 minutes about the Sarbanes-Oxley Act and the recently passed SEC rules on internal control reporting. This year, everything is different. First of all, as most of you know, we now have a Chief Accountant. We also have another Deputy Chief Accountant. So I'm no longer alone. You'll be happy to know my son is doing well, but instead of being at home, he is currently in the back of the room wondering why daddy's head is so big on these screens. And you may be even happier to hear that I'm not planning to spend the next 20 minutes talking about the Sarbanes-Oxley Act at all.

One thing that hasn't changed from last year is that my words continue to be my own, and don't necessarily reflect the views of the Commission or my colleagues on the staff of the SEC.

Before I get to the topics I do want to discuss today, I think it's important to explain why I'm not talking about Sarbanes-Oxley. It isn't just because we're all tired of hearing about scandals, reforms, new rules, and big changes. It's actually because, after a long period of rule-writing, implementation, and hard work, we are, by and large, done. Just about all the new rules are in place, and most of them have been fully implemented by registrants and their auditors. With the notable exception of preparing for the first reports on internal controls, we have, at least at the SEC, gotten back to our day jobs. That is, we're now able to spend time on making sure that the rules and guidance that have been out there for some time are being applied in the manner that best represents their intent, and that provides the best information for investors.

Of course, we're still focusing on making improvements in financial reporting, and all of the topics I will discuss have that in common. Unfortunately, they have little else in common, so you'll have to forgive me if I make a few inelegant segues between topics.

Pensions

I'll start with one of the oldest accounting issues there is - pensions. The pension obligation is often a company's largest obligation, even if current accounting guidance doesn't always put the full obligation on the books. But I'd like to talk about the calculation of the pension obligation, rather than how that obligation is accounted for. As everybody knows, pension accounting involves the use of many different assumptions, and these assumptions can have huge effects on the calculations. One such assumption is the discount rate. The discount rate is meant to be the rate the company would have to pay to purchase high-quality investments that would provide cash sufficient to settle its current pension obligations. The SEC staff has previously made its views known that bonds rated Aa and higher would appear to meet the criterion of being quality. While the discount rates obviously vary from company to company, it would appear that, on average, rates being used are higher than the Aa bond index rate by between ¼ and ½ percent. In and of itself, this is not necessarily a problem, as a company might very well support its assumed discount rate in some other manner.

However, where disclosures do not indicate what evidence a company has looked to in determining its discount rate, we do generally ask for such support as part of a review of the company's filing. Unfortunately, we are often given responses that merely refer to rates used by other companies, rather than empirical support for the company's assumption. In some cases, registrant's have indicated that they relied on their actuaries to select a rate and their auditors to tell them if the rate was unreasonable. These are insufficient responses. No matter whom a company seeks out for advice in making assumptions like the discount rate, the company should have support for its assumptions that goes beyond a comparison with others. While referring to applicable index rates would provide acceptable support, other companies might support their rate in other ways. For example, by constructing a hypothetical portfolio of high quality instruments with maturities that mirror the pension obligation. In any event, we believe that size of pension obligations and the significance of the discount rate assumption to their calculation indicate that significant thought should be given to the rate's selection.

Contingencies

Sticking with age old issues, I want to talk for a few minutes about contingent liabilities. As background, remember that FASB Statement No. 5 covers the accounting for contingent liabilities, and requires accrual of payments for such liabilities if payment is both probable and estimable. Moreover, Statement 5 requires disclosures of the nature of any contingency, including the amounts that might be paid, if a loss is at least reasonably possible. In addition, the SEC's MD&A rules require discussion of items that might affect the company's liquidity or financial position in the future, including contingent liabilities. At the SEC, we constantly deal with questions relating to contingent liabilities. So I'd like to give you a few thoughts on accounting and disclosures in this area.

First, I would suggest that the recording of a material accrual for a contingent liability related to an event that occurred several years before should not be the first disclosure regarding that contingency. Rather, disclosures regarding the nature of the contingency and the amounts at stake should generally have already been provided under Statement 5 when the loss became reasonably possible. Vague or overly broad disclosures that speak merely to litigation, tax, or other risks in general, without providing any information about the specific loss contingencies being evaluated are not sufficient. Registrants and their auditors and attorneys should be critically assessing the claims against the company, and registrants should provide disclosures that discuss the nature of the claim, and the possible range of losses for any claim where the maximum reasonably possible loss is material. Furthermore, I should point out that Statement 5 and Interpretation 14 require accrual, for probable losses, or disclosure, for reasonably possible losses, of the most likely amount of the loss. While the low end of a range of possible losses is the right number if no amount within the range is more likely than any other, I find it somewhat surprising the number of instances where zero is considered the low end of a range with no number more likely than any other right up until a large settlement is announced. Interestingly, these situations are also often the ones for which no significant disclosure has been made in the financial statement before the settlement is announced. Those who complain about the detailed nature of some recent accounting pronouncements need only look to the accounting and disclosure of contingencies that fall into the scope Statement 5 to see why standard-setters sometimes move to more detailed guidance.

My second point is that income tax contingencies also fall within the scope of Statement 5. In other words, the accounting and disclosure requirements of Statement 5 apply to items of tax cushion. Statement 109 provides additional disclosure requirements for income tax items that arise as a result of temporary differences. Some of you no doubt are thinking, "If we follow that guidance, and comply with the disclosure requirements in Statements 109 and 5 - we'd be providing a virtual roadmap for the IRS." This issue has been with us for some time. A company's concern about maintaining confidentiality in this area is understandable, but it has to be balanced with the need for the company's investors, analysts, and regulators to gain a clear understanding of liquidity and the financial position and results of operations. Moreover, compliance with GAAP and SEC disclosure requirements is a responsibility a company voluntarily takes on when it decides to access the public markets. Confidentiality concerns do not excuse violations of GAAP or the SEC's disclosure rules.

My final point about contingencies relates to auditing. Having been an auditor, I am well aware that loss contingencies are one of the most difficult areas there is to audit. Representations from management and from attorneys sometimes seem to be all that there is to support an accrual or the lack of one. The difficulty in auditing these types of accruals, however, should cause the auditor to spend more time on them, not less. Auditors should seek to review the company's own analyses of the issues, including the support for the conclusions as to whether an accrual is necessary, and what the possible range of loss is. If the only procedures that can be performed are face-to-face discussions with company personnel and with outside counsel, those discussions should be held, and experienced auditors should be part of them. If a company's outside counsel is unwilling or unable to provide its expert views, the auditor should consider whether sufficient alternate procedures can actually be performed to allow the audit to be completed. Audit documentation should follow the same high standards that apply to other areas of the audit, as well. This, of course, includes the documentation of the audit of the tax contingency accounts. A note or short memo that indicates that qualified personnel from the audit firm held discussion of all relevant risks with company personnel is not sufficient. I would expect that the PCAOB inspection teams will be looking at the audit work done in these sensitive areas as they begin their first year of a full inspection schedule.

Segments

Let me turn now to the issue of disclosure of disaggregated information, more commonly known as segment disclosures. I know that segment disclosures have long been a source of pain for registrants, auditors, Commission staff, and standard-setters over the years. Unfortunately, they have also been a source of pain for users, the very parties who should be benefiting from them. While users are rarely united in terms of what information they would like to see in filings of public companies, they are virtually unanimous in the desire for more segment data. Statement 131 has increased the volume of segment data, but perhaps not as much as intended. One of the reasons for this is that there are many companies that take advantage of Statement 131's provisions that allow for aggregation of operating segments. While aggregation of immaterial segments is easy to understand, it is more difficult for users to accept aggregation that occurs because a company contends that material operating segments are so similar that they should be combined. Pursuant to Statement 131, this aggregation can only occur if the operating segments in question sell similar products or services created with similar production processes to similar customers using similar distribution systems in similar regulatory environments, and that those segments have similar economic characteristics. These criteria are all listed in paragraph 17 of the Statement 131, along with one more - that aggregation must be consistent with the objectives and principles of the standard.

It would appear that this should be a high hurdle to aggregation. The FASB makes clear in the basis for conclusions to Statement 131 that aggregation is acceptable in certain situations because "separate reporting of segment information will not add significantly to an investor's understanding of an enterprise if its operating segments have characteristics so similar that they can be expected to have essentially the same future prospects." That is, aggregation is OK if presenting the information separately won't provide much useful information to users of financial statements. In all of the discussions I've had with registrants on this issue, where I have asked why the company aggregated segment data, never once have I received an explanation that focused on whether the additional information would be useful to users, or whether aggregation in this instance is consistent with the objectives and principles of Statement 131. Instead, the discussions have all revolved around whether each of the six objective criteria I mentioned above have been met, and how the evaluation of similar economic characteristics should be performed. In addition, registrants often cite the competitive harm that would befall the company if it disclosed additional information. A better and more honest application of Statement 131 would allow aggregation only when providing the more detailed information wouldn't add anything to a user's understanding of the company's financial results and prospects.

Materiality

If my mentioning of pensions, loss contingencies, and segments didn't elicit enough negative reactions, I'll now throw caution completely to the wind and talk for a minute or two on materiality. Around 5 years ago, the SEC staff issued Staff Accounting Bulletin No. 99, which provides significant discussion of how materiality should be evaluated. While SAB 99 is long and provides detailed guidance in some areas, it hasn't resolved all of the issues regarding materiality evaluations, and has, unfortunately, had the effect of causing confusion in some cases about how quantitative and qualitative considerations on materiality should be analyzed. In the time that I have been serving as Deputy Chief Accountant, perhaps the most difficult discussions are the ones having to do with the evaluation of materiality of errors, so I'd like to provide you a few of my thoughts on this topic.

First, it is important to remember that accounting for something in a way that is not consistent with GAAP is still an error, whether the item is material or not. That is, just because something is small doesn't mean there are no accounting standards that apply to it. It just means that accounting for it wrong doesn't materially misstate the financial statements.

Second, registrants need to keep in mind that, just like every other area of financial reporting, the initial conclusion as to whether an error is material or not is the company's to make. Management should not merely ask its auditors whether an error must be corrected. Instead, an analysis should be done, considering all relevant qualitative and quantitative factors, to support a conclusion as to whether the error represents a material misstatement, alone or in combination with other errors. Only after the registrant has reached a conclusion as to the materiality of an item should an auditor do so. One more point on this - please don't outsource the decision to the SEC staff. Too often we get submissions from registrants seeking our concurrence with not restating for an error based on materiality, but with little analysis of why the error isn't material. In some cases, the submission has even indicated that either the registrant or the auditors have not yet reached a conclusion, but have decided to write to the SEC staff for help. We are not in a position to do the analysis - we don't know nearly as much about your situation as you and your auditors do, and we aren't able to make this decision for you. We are happy to provide our thoughts on materiality considerations after the company and its auditors have completed their analyses, but in the absence of conclusions from the company and its auditors, we would begin by presuming any error is material.

Finally, I'd like to mention another aspect of materiality analyses that is rarely talked about, the so-called income statement vs. balance sheet approach question. The income statement method of evaluating materiality, which is also called the current-period or rollover method, considers as an error the amounts recorded in the current period income statement that shouldn't have been. The balance sheet method, which is also called the cumulative or iron curtain method, considers as an error the total effect of all amounts that have been recorded in the company's books during the current or prior periods. So, if a company increases a reserve $10 more than necessary each year for three years, the rollover method treats the error as being $10 in each year, while the iron curtain method treats the error as $10 in the first year, $20 in the second, and $30 in the third.

Auditing standards briefly mention this debate in a footnote, noting only that "The measurement of the effect, if any, on the current period's financial statements of misstatements uncorrected in prior periods involves accounting considerations and is therefore not addressed in this section." As you might guess from my previous reminder that materiality conclusion must first be made by the company, rather than its auditors, I agree with the sentiment this debate is an accounting one, not an auditing one. Unfortunately, accounting literature doesn't deal with this at all.

During my time as an auditor with Arthur Andersen, I was taught to use the iron curtain method, and I think auditors generally believe it is the preferable method in most situations. Consider the example I just described, and assume that $10 would be immaterial to all periods, but $30 would be material. Under the rollover method, there is never a material error, unless the company wants to reverse the accrual in the fourth year, because doing so would put $30 out of period. So, under the rollover method, eliminating the overaccrual becomes a material error. Under the iron curtain method, the error would need to be corrected in year three, as it became material that year.

For various reasons, both methods are used in practice and have been accepted by auditors and regulators. This is often considered to be a policy election, and the fact that the accounting literature does not specify which method to use would seem to support, to some extent, acceptability of both methods. Interestingly, however, I have never heard of a company disclosing, in its summary of accounting policies or elsewhere, which method it uses to evaluate errors and why. Previous initiatives to provide guidance on which method should be used when have been made, with no substantive progress. Recent events have us at the SEC thinking that the time has come to provide more guidance in this area. While we aren't ready to provide that guidance right now, I can tell you that we are likely to be asking for more input and information in this area in the near future, with a view towards providing guidance in the area to resolve this question that has troubled accountants and auditors for so long.

Contingent Fees

Making yet another uneven transition, I'm going to switch now to an auditor independence issue that has put the Chief Accountant's Office in the newspapers in the past week. As everybody here is by now painfully aware, Sarbanes-Oxley included a number of new rules related to auditor independence. As promised, I won't talk about any of them now, but I do want to mention a portion of the SEC's auditor independence guidance that has existed for several years. In rules that went into effect in 2001, the SEC clarified that contingent fee arrangements between an auditor and client impair the auditor's independence. The rule in question provides an exception for fees that are contingent upon the determination of a court or other third-party regulator. The release accompanying the rule makes it clear that we intended the exception to apply to situations where the third-party rules on the fee itself, rather than a factor that is used to calculate the fee.

For example, if an accounting firm and audit client were to agree that the firm would receive 30% of any tax savings to the client resulting from tax advice provided by the firm, the fee would be a contingent fee and impair the auditor's independence, notwithstanding the expectation that tax return would be audited. In this case, the firm and client, not a court or government agency, would have agreed to the determination of the fee. The fact that a government agency might challenge the amount of the client's tax savings and thereby alter the final amount of the fee paid to the firm heightens, as opposed to reducing, the mutuality of interest between the firm and client. Accordingly, such fees impair an auditor's independence.

We understand, based on correspondence form the AICPA's Professional Ethics Executive Committee, that some have been interpreting this part of the SEC's rules differently than we intended, referencing an interpretation in the AICPA literature that would lead to a conclusion that fee arrangements like the one I just described are not contingent arrangements, because the IRS would be involved in reviewing the tax return. The SEC guidance does not include such an interpretation, and we do not believe such an interpretation would be consistent with clear language in the release that accompanied the 2001 independence rules. To ensure any remaining confusion is eliminated, Don Nicolaisen sent a letter to the Professional Ethics Executive Committee explaining our views. The letter is available on the SEC's internet site.

Don's letter also points out that other arrangements that function similar to contingent fee arrangements, such as so-called value-added fee arrangements that implicitly or explicitly include an assessment of the value-added based on a contingency, would also impair an auditor's independence. In substance, we believe that fees for any services provided by the auditor to the audit client should be subject to fee arrangements that provide for a final determination of the fee shortly after the work is performed, and based solely on the time, effort, and quality of the work, rather than based on contingencies related to benefits to be obtained from the work.

What Can We Do

Before I sit down and turn the microphone over to others, I'd like to provide a few short suggestions to those many participants in the financial reporting process that would like to help improve things, to provide better information in their public filings and documents. Often, it seems that it is too hard for one or a few people to make improvements, because the environment is so difficult today. But if you take the approach of looking at reporting not as a compliance exercise, but as a communication exercise, and make choices that provide more information, rather than those that minimally comply with the rules, there are many things that can be done.

For example, as Don Nicolaisen suggested last December, any company can improve its reporting by voluntarily presenting cash flows under the direct method. Users tell us they would prefer the direct method information, and Statement 95 provides a framework in which to provide it. As another example, related to a topic I touched on earlier, registrants could decide not to combine operating segments in their segment disclosures even in those instances where they do qualify. As the information required to be presented is that information which is already available and presented to the Chief Operating Decision-Maker, there is little cost involved in making this choice, and aggregating segments is never required.

Related to pension accounting, many constituents complain about the smoothing mechanisms of Statement 87. Well, nobody is required to use those smoothing mechanisms. Eliminate the pension corridor, and you'll be immediately providing more accurate financial statements. And there are plenty of others as well. Choosing accounting policies that recognize "other than temporary" impairments quicker, rather than slower, changing inventory accounting from LIFO to FIFO, and presenting income statement information by both function and nature would all seem to represent improvements to a company's reporting. For those who are truly interested in making improvements to financial reporting, these represent just a few of the opportunities that are currently available. I'd encourage everybody to think about the filings you work on, and identify just one or two similar items that you can use to improve your communication with users. In fact, I'd almost guarantee that you're in a better position that the standard-setters, the SEC, or anybody else, to find ways to improve your reporting.

Thank you for your attention. I look forward to taking questions later.

http://www.sec.gov/news/speech/spch052704sat.htm