Accounting and Valuation:
How helpful are recent accounting innovations?
Bradford Cornell
California Institute of Technology
Pasadena, CA 91125
626-564-2001
Wayne R. Landsman
Kenan-Flagler Business School
University of North Carolina
Chapel Hill, NC 27599-3490
919-962-3221
July 2006
We thank Bruce Miller, Ken Peasnell, Cathy Shakespeare, and Shu Yeh for helpful comments. We also acknowledge funding from the Center for Finance and Accounting Research and the KPMG Fellowship Fund at UNC-Chapel Hill.
21
Abstract
Policy debate in accounting has focused on the methods by which individual line items are aggregated to produce competing measures of “earnings”. In a previous paper, Cornell and Landsman (2003), we argued that this focus is misguided. Investors are more than capable of choosing between competing measures of earnings, or developing their own measures, provided that they have complete and clear information regarding the component items. In this paper, we examine the extent to which innovations in accounting reporting rules regarding employee stock options, derivatives, and asset securitizations provide the necessary component level data for these items. We conclude that while progress is being made, the current level of component disclosure remains inadequate. There is still a long way to go before accounting reports provide investors with the requisite component information to perform accurate, independent valuations of companies.
21
Introduction
In a previous paper, Cornell and Landsman (2003), we examine the issue of the “quality of earnings”. The debate over the quality of earnings arose because, on the grounds that GAAP earnings supposedly is not the best measure of the true profitability of a business, companies and security analysts introduced a host of pro forma alternatives. The widespread use of non-GAAP measures then became an issue of concern for regulators. The SEC’s former chief accountant, Lynn Turner, among others, argued that pro forma releases, “Often they appear to be trying to lead investors away from the real number, from real net income.”[1] The concern that investors were being misled by pro forma numbers resulted in the requirement in the Sarbanes-Oxley Act (2002) that firms provide a reconciliation between pro forma earnings and GAAP earnings if they release pro forma earnings.
The debate over the proper measure of earnings also spawned an academic literature that examines the subject. Studies including, Brown and Sivakumar (2001), Bradshaw and Sloan(2002), Francis, Schipper and Vincent (2003), and Lougee and Marquardt (2001) use regression techniques to estimate the “quality” of competing earnings measures based on the valuation information they contained. The results of the studies turned out to be a mixed bag with the findings depending on the earnings measures being compared, the time period and the sample of companies chosen, and the metric employed.
The thesis put forth in Cornell and Landsman (2003) is that from a valuation perspective the entire debate regarding earnings quality was both misguided and theoretically unresolvable. Put simply, there is no consistently meaningful way to condense all the historical financial data that is relevant for forecasting future performance into one measure (or a time series of one measure). Furthermore, attempts by regulatory bodies, such as the SEC, or accounting standard setting bodies, such as the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board, to determine an appropriate definition of pro forma income distracts attention from far more critical problems involving omissions and ambiguities in the constituent items that make up any measure of earnings.
To return to first principles, the primary reason for public disclosure of accounting information is to promote an efficient allocation of capital. A prerequisite for efficient allocation is that market values, on which the allocation of capital is based, reflect, to the greatest extent possible, true economic values. That requires, using the words of Warren Buffett, that investors be able to “understand” the businesses in which they are considering investing.[2] In this endeavor, the precise fashion in which accounting numbers are aggregated is of minor importance. The work of Ohlson (1995) and Feltham and Ohlson(1995) makes it clear that as long as a clean surplus accounting method is employed, the differences between the different measures of income cancel out in the present value relation. More importantly, it is unlikely that any set of accounting aggregates will be sufficient to allow an investor to properly value a company. What is critical, Cornell and Landsman argue, is that all the component data that feed into any aggregates are fully and transparently reported. Using that component data, and other value relevant information, investors can develop their own techniques for valuing the company. These techniques may involve the use of GAAP or pro forma aggregates developed by the company, but they may not.
Furthermore, the research and scholarship necessary to determine how financial data should be aggregated and used to value companies is publicly available. What is not public, unless the company discloses it, are details related to individual accounting line items that allow investors to evaluate the economic performance of a company. This information is critical for answering questions such as: How have revenues grown over time and why? What has been the cost of producing those revenues in terms both of marginal outlays and capital expenditures? To what extent have past expenditures provided the company with growth options for the future? Consequently, Cornell and Landsman claim that the primary role of regulation and accounting standard setting should be to insure that the individual line items are presented in a clear and complete fashion and at an adequate level of disaggregation. Given adequate data on the individual line items, investors can construct their own measures of “earnings”.
Viewed in this light, the basic accounting items: revenues, cost of good sold, sg&a, operating assets, working capital, depreciation, one-time charges and the like can be thought of as the primary components that comprise the information needed to value a firm. The various earnings measures are linear combinations of these primary components constructed by investors based on their analysis of the firm. Different investors will put the basic components together in different ways and thereby arrive at different valuations for the firm. The market then aggregates the diverse estimates of value into a price that affects the ultimate allocation of capital. If investors have accurate information only about broad aggregates, no matter how well constructed the aggregates may be, the market price can only reflect that method of aggregation. As a result, information is lost and pricing is likely to be less efficient.[3]
In this paper, we examine innovations in three widely discussed and controversial accounting standards to determine the extent to which those innovations make it more likely that investors will have the component data they need to properly value businesses. The first is SFAS 123 which relates to accounting for employee stock options. The second is SFAS 133 which covers accounting for derivatives and hedging activity. The third is SFAS 140, which deals with accounting for asset securitizations. We find that in each case some progress has been made but the rules still do not require firms to disclose details on the component data in what, from a valuation perspective, is an adequate manner.
Accounting Standards
SFAS 123: Employee Stock Options
From an economic standpoint, granting options is costly to a company in that doing so dilutes the claim of current shareholders to future dividends. In addition, increases or decreases in the value of outstanding options represent real economic costs or benefits. Accurate valuation requires taking account of these costs and benefits associated with employee stock options. The challenge to accounting is to provide the information required for investors to undertake this task.
Statement of Financial Accounting Standards No. 123 (Revised), Share Based Payment (FASB, 2004) requires the cost of employee stock options grants be recognized in income using grant date fair value by amortizing the cost during the employee vesting or service period. This requirement removed election of fair value or intrinsic value cost measurement permitted under the original recognition standard, Statement of Financial Accounting Standards No. 123, Accounting for Stock-based Compensation (FASB, 1995). Until recently, most firms elected to measure the cost of employee stock options using intrinsic value. However, for such firms, SFAS No. 123 required they disclose a pro forma income number computed using a fair value cost for employee stock option grants, as well as key model inputs they use to estimate fair values. To illustrate our discussion of the SFAS 123 disclosure, Table 1 extracts information from the Stock Option plan footnote from the Coca Cola Company 2005 annual report. The top panel of Table 1 lists the fair value per option as well as assumptions Coca Cola used when estimating the fair value of options granted during the most recent three fiscal years, 2003, 2004, and 2005. Coca Cola states in the disclosure that they use the Black-Scholes-Merton option pricing model to estimate grant date fair values. The bottom panel of Table 1 shows a summary of share activity during the same period. Because Coca Cola elected to measure the cost of employee stock options using fair value beginning in 2002, they do not report pro forma income. Instead, the compensation charge associated with employee stock options is reported elsewhere in the financial statements as a component of selling, general and administrative expenses.
Under SFAS 123, there are five possible accounting events associated with the granting of options: grant date measurement, compensation recognition, exercise, expiration, and forfeiture. Although SFAS 123 provides for grant date fair value measurement of the cost of the options granted to employees, there are no financial statement effects at grant date. Table 1 indicates that that in 2005, Coca Cola granted to employees 34 million options with an average fair value of $8.23 per option, for a total grant fair value of $279.82 million. Typically, straight line amortization of the grant date fair value is applied, whereby the compensation cost of the options is recognized as compensation expense during the service or vesting period. A table in Coca Cola’s financial statement footnotes listing the components of selling, general and administrative expenses states that in 2005 Coca Cola recorded an employee stock option compensation expense of $324 million. The financial statement effects of such recognition is a debit to compensation expense and a credit to paid-in capital—employee stock options, so there is no net effect on shareholders’ equity before tax.[4] Upon exercise by employees, the debits to cash and paid-in capital—employee stock options are offset by credits to common stock and additional paid-in capital. The Consolidated Statements of Shareowners’ Equity (not tabulated here) reveal that Stock Issued to Employees Exercising Stock Options increased Common Stock by $2 million and Capital Surplus by $229 million. Options that expire unexercised result in paid-in capital—employee stock options being closed out into additional paid-in capital. Under SFAS 123, FASB prefers basing compensation expense using an estimate of expected forfeitures. However, the FASB permits (and most firms use) the more simple reversal of prior compensation for employees whose options will not vest because of forfeiture. The obvious problem with this approach is it can artificially inflate income (by reducing compensation expense), and it is generally not possible for investors to determine how much of the recognized employee stock option compensation expense is attributable to current year’s amortization of current and past option grants and the reversal associated with forfeitures. Not only can investors not determine the effect of forfeitures on current year’s compensation expense, but they also cannot even determine the number of forfeitures from the SFAS 123 disclosure, since the FASB permits firms to aggregate the number forfeitures and expired options, which Table 1 indicates is 7 million for Coca Cola in 2005.
Aside from the income distortion that results from forfeitures, critics of SFAS 123 express concern that grant date accounting fails to reflect the true cost of employee option grants to shareholders. Most notably, options that expire because stock prices fall or fail to rise during the exercise period result in no dilution of ownership claims. That is, SFAS 123 fails to provide for financial statement effects of changes in the options’ fair value beyond grant date. A recent study by Landsman, Peasnell, Pope, and Yeh (2006) shows analytically that dilution effects of employee option grants can only be correctly captured by including in income changes in the options’ fair value beyond grant date. The study then attempts to prove this point empirically by showing that accounting earnings and equity book value adjusted to reflect changes in option fair values beyond grant date better capture the dilution effects of option grants—i.e., the true cost of the options to current equity holders—as reflected in current equity market values.[5] However, when conducting their empirical analysis, the authors have to rely on a variety of assumptions to estimate the fair value change component of income because the SFAS 123 disclosure fails to provide this information.[6] This makes it difficult for the study to determine empirically whether investors price this income component. Of course, investors are similarly affected by this lack of information. Regardless of whether this component of income should be recognized in income, a simple solution to the problem would be for the FASB to improve the SFAS 123 disclosure by simply adding an extra column to the option activity table showing fair value estimates of options outstanding, granted, exercised, expired, and forfeited (the latter two should, of course, be disaggregated in the table!).
The recent option backdating controversy provides another example of a disclosure problem, at least as far as valuation is concerned. As of this writing, numerous firms are under SEC and Justice Department investigation for backdating option grants in the period from 1992 to 2002. The critical valuation question is whether the backdating was adequately communicated to the company's shareholders, and whether any income effects were properly reflected in earnings. For example, because firms that backdate options choose a grant date with a lower stock price than that on the actual grant date, the options are effectively in-the-money at grant date. During the backdating period virtually all firms that backdated option grants elected to measure the cost of option grants using intrinsic value, reported earnings should have been reduced by the amortization applicable to affected fiscal years, including the year of the grant. Because backdating is typically not reflected properly in earnings, some companies that have recently admitted to backdating of options have restated earnings for past years. If backdating causes investors to misinterpret a company’s economic earnings in a given period, because they do not have adequate information, inaccurate valuations can result.