ANNUAL REPORT 2007

 
CRITICAL ACCOUNTING POLICIES AND ESTIMATES
 
Our significant accounting policies are discussed in Note 1 within Notes to Consolidated Financial Statements.
 
 
At the beginning of our 2007 fiscal year, we adopted the Financial Accounting Standards Board (FASB) Interpretation No. 48 “Accounting for Uncertainty in Income Taxes” (FIN No. 48), which affects our process for estimating tax benefits and liabilities, as further discussed in the “Income taxes” section beginning on page 27. The initial application of FIN No. 48 resulted in a net decrease to accrued income tax and related interest liabilities of approximately $2 million, with an offsetting increase to retained earnings. Refer to Note 1 within Notes to Consolidated Financial Statements for further information on FIN No. 48.
 
 
In September 2006, the FASB issued SFAS No. 157 “Fair Value Measurements” in order to establish a single definition of fair value and a framework for measuring fair value in generally accepted accounting principles (GAAP) that is intended to result in increased consistency and comparability in fair value measurements. SFAS No. 157 also expands disclosures about fair value measurements, with the intention of improving the quality of information provided to users of financial statements. The standard applies whenever other authoritative literature requires (or permits) certain assets or liabilities to be measured at fair value, but does not expand the use of fair value. SFAS No. 157 was originally effective for financial statements issued for fiscal years beginning after November 15, 2007, and interim periods within those years with early adoption permitted. In early 2008, the FASB issued Staff Position (FSP) FAS-157-2, which delays by one year, the effective date of SFAS No. 157 for all non-financial assets and non-financial liabilities, except those that are recognized or disclosed at fair value in the financial statements on a recurring basis (at least annually). The delay pertains to items including, but not limited to, non-financial assets and non-financial liabilities initially measured at fair value in a business combination, reporting units measured at fair value in the first step of evaluating goodwill for impairment under SFAS No. 142 “Goodwill and Other Intangible Assets”, indefinite-lived intangible assets measured at fair value for impairment assessment under SFAS No. 142, and long-lived assets measured at fair value for impairment assessment under SFAS No. 144 “Accounting for the Impairment or Disposal of Long-Lived Assets”. We plan to adopt the portion of SFAS No. 157 that has not been delayed by FSP FAS-157-2 as of the beginning of our 2008 fiscal year, and plan to adopt the balance of its provisions as of the beginning of our 2009 fiscal year. For the Company, balance sheet items carried at fair value on a recurring basis (to which SFAS No. 157 applies in 2008) consist primarily of derivative financial instruments. Balance sheet items carried at fair value on a non-recurring basis (to which SFAS No. 157 will apply in 2009) consist of assets held for sale and exit liabilities. Relevant to the “Intangibles” section beginning on page 23, we also use fair value concepts to test various long-lived assets for impairment and to initially measure assets and liabilities acquired in a business combination. We do not currently expect the adoption of SFAS No. 157 in 2008 to have a material effect on the measurement of the Company’s financial


22

assets and liabilities. We are continuing to evaluate the impact the standard will have on the determination of fair value related to non-financial assets and non-financial liabilities in post-2008 years.
 
 
In December 2007, the FASB issued SFAS No. 141 (Revised 2007) “Business Combinations” and SFAS No. 160 “Noncontrolling Interests in Consolidated Financial Statements,” which are effective for fiscal years beginning after December 15, 2008. These new standards represent the completion of the FASB’s first major joint project with the International Accounting Standards Board (IASB) and are intended to improve, simplify, and converge internationally the accounting for business combinations and the reporting of noncontrolling interests (formerly minority interests) in consolidated financial statements. We will adopt these standards at the beginning of our 2009 fiscal year. The effect of adoption will generally be prospectively applied to transactions completed after the end of our 2008 fiscal year, although the new presentation and disclosure requirements for pre-existing noncontrolling interests will be retrospectively applied to all prior-period financial information presented.
 
 
SFAS No. 141(R) retains the underlying fair value concepts of its predecessor (SFAS No. 141), but changes the method for applying the acquisition method in a number of significant respects including the requirement to expense transaction fees and expected restructuring costs as incurred, rather than including these amounts in the allocated purchase price; the requirement to recognize the fair value of contingent consideration at the acquisition date, rather than the expected amount when the contingency is resolved; the requirement to recognize the fair value of acquired in-process research and development assets at the acquisition date, rather than immediately expensing; and the requirement to recognize a gain in relation to a bargain purchase price, rather than reducing the allocated basis of long-lived assets. Because this standard is generally applied prospectively, the effect of adoption on our financial statements will depend primarily on specific transactions, if any, completed after 2008. We are currently evaluating the effects that SFAS No. 141(R) is likely to have on potential post-2008 transactions.
 
 
Under SFAS No. 160, consolidated financial statements will be presented as if the parent company investors (controlling interests) and other minority investors (noncontrolling interests) in partially-owned subsidiaries have similar economic interests in a single entity. As a result, the investment in the noncontrolling interest, previously recorded on the balance sheet between liabilities and equity (the “mezzanine”), will be reported as equity in the parent company’s consolidated financial statements, subsequent to the adoption of SFAS No. 160. Furthermore, consolidated financial statements will include 100% of a controlled subsidiary’s earnings, rather than only the parent company’s share. Lastly, transactions between the parent company and noncontrolling interests will be reported in equity as transactions between shareholders, provided these transactions do not create a change in control. Previously, acquisitions of additional interests in a controlled subsidiary generally resulted in remeasurement of assets and liabilities acquired; dispositions of interests generally resulted in a gain or loss.
 
 
Management is currently evaluating the impact of adopting SFAS No. 160 on the Company’s financial statements. Presently, there are no significant noncontrolling interests in any of the Company’s consolidated subsidiaries. Therefore, we currently believe that the impact of SFAS No. 160, if any, will primarily depend on the materiality of noncontrolling interests arising in future transactions, including those entered into during 2008, to which the financial statement presentation and disclosure provisions of SFAS No. 160 will apply.
 
 
Our critical accounting estimates, which require significant judgments and assumptions likely to have a material impact on our financial statements, are discussed in the following sections on pages 23-27.
 
Promotional expenditures
Our promotional activities are conducted either through the retail trade or directly with consumers and involve in-store displays and events; feature price discounts on our products; consumer coupons, contests, and loyalty programs; and similar activities. The costs of these activities are generally recognized at the time the related revenue is recorded, which normally precedes the actual cash expenditure. The recognition of these costs therefore requires management judgment regarding the volume of promotional offers that will be redeemed by either the retail trade or consumer. These estimates are made using various techniques including historical data on performance of similar promotional programs. Differences between estimated expense and actual redemptions are normally insignificant and recognized as a change in management estimate in a subsequent period. On a full-year basis, these subsequent period adjustments have rarely represented more than .4% (.004) of our Company’s net sales. However, as our Company’s total promotional expenditures (including amounts classified as a revenue reduction) represented nearly 30% of 2007 net sales, the likelihood exists of materially different reported results if different assumptions or conditions were to prevail.
 
Intangibles
We follow SFAS No. 142 “Goodwill and Other Intangible Assets” in evaluating impairment of intangibles. We perform this evaluation at least annually during the fourth quarter of each year in conjunction with our annual budgeting process. Under SFAS No. 142, goodwill impairment testing first


23

requires a comparison between the carrying value and fair value of a reporting unit with associated goodwill. Carrying value is based on the assets and liabilities associated with the operations of that reporting unit, which often requires allocation of shared or corporate items among reporting units. The fair value of a reporting unit is based primarily on our assessment of profitability multiples likely to be achieved in a theoretical sale transaction. Similarly, impairment testing of other intangible assets requires a comparison of carrying value to fair value of that particular asset. Fair values of non-goodwill intangible assets are based primarily on projections of future cash flows to be generated from that asset. For instance, cash flows related to a particular trademark would be based on a projected royalty stream attributable to branded product sales. These estimates are made using various inputs including historical data, current and anticipated market conditions, management plans, and market comparables.
 
 
We also follow SFAS No. 142 in evaluating the useful life over which a non-goodwill intangible asset is expected to contribute directly or indirectly to the cash flows of the Company. An intangible asset with a finite useful life is amortized; an intangible asset with an indefinite useful life is not amortized, but is evaluated annually for impairment. Reaching a determination on useful life requires significant judgments and assumptions regarding the future effects of obsolescence, demand, competition, other economic factors (such as the stability of the industry, known technological advances, legislative action that results in an uncertain or changing regulatory environment, and expected changes in distribution channels), the level of required maintenance expenditures, and the expected lives of other related groups of assets.
 
 
At December 29, 2007, intangible assets, net, were $5.0 billion, consisting primarily of goodwill and trademarks associated with the 2001 acquisition of Keebler Foods Company. Within this total, approximately $1.4 billion of non-goodwill intangible assets were classified as indefinite-lived, comprised principally of Keebler trademarks. While we currently believe that the fair value of all of our intangibles exceeds carrying value and that those intangibles so classified will contribute indefinitely to the cash flows of the Company, materially different assumptions regarding future performance of our North American snacks business or the weighted-average cost of capital used in the valuations could result in significant impairment losses and/or amortization expense.
 
Stock compensation
In December 2004, the FASB issued SFAS No. 123(R) “Share-Based Payment,” which generally requires public companies to measure the cost of employee services received in exchange for an award of equity instruments based on the grant-date fair value and to recognize this cost over the requisite service period. We adopted SFAS No. 123(R) as of the beginning of our 2006 fiscal year, using the modified prospective method. Accordingly, prior years were not restated. With the adoption of this pronouncement, stock-based compensation represents a critical accounting policy of the Company, which is further described in Note 1 within Notes to the Consolidated Financial Statements.
 
 
In 2006, our adoption of SFAS No. 123(R) resulted in an increase in the Company’s corporate SGA expense and a corresponding reduction to earnings and net earnings per share, due primarily to the recognition of compensation expense associated with employee and director stock option grants. No such expense was recognized under our previous accounting method in pre-2006 periods; however we were required to disclose pro forma results under the alternative fair value method prescribed by SFAS No. 123 “Accounting for Stock-Based Compensation”. Using reported results for 2006 and pro forma results for 2005, the comparable impact of stock compensation expense is presented in the following table:
 
                         
   
    Stock-based
       
    compensation expense     Diluted EPS
 
(millions, except per share data)   Pre-tax     Net of tax     impact  
   
 
2006:
                       
As reported comparable
  $ 31     $ 20     $ .04  
SFAS No. 123(R) adoption impact
    65       42       .11  
 
 
As reported total
  $ 96     $ 62     $ .15  
 
 
2005:
                       
As reported comparable
  $ 18     $ 12     $ .03  
Pro forma incremental
  $ 58     $ 37       .09  
 
 
Pro forma total
  $ 76     $ 49     $ .12  
 
 
 
 
Accounting for stock compensation under SFAS No. 123(R) represents a critical accounting estimate, which requires significant judgments and assumptions likely to have a material impact on our financial statements. Due to the need to determine the grant-date fair value of equity instruments that have not yet been awarded, the actual impact on future results will depend, in part, on actual awards during any reporting period and various market factors that affect the fair value of those awards. Additionally, while the timing and volume of grants associated with a particular year’s long-term incentive compensation are within our control, the timing and volume of “reload” option grants are not. Reload options are awarded to eligible employees and directors to replace previously-owned Company stock used by those individuals to pay the exercise price, including related employment taxes, of vested pre-2004 option awards containing this accelerated ownership feature. Under SFAS No. 123(R), these reload options result in additional compensation expense in the year of grant and for 2007 and 2006, represented approximately one-third of the Company’s total stock option expense.


24

The Company has not granted options containing an accelerated ownership feature since 2003; however, the potential requirement to award reload options over the contractual 10-year term of the original grants could continue to significantly impact the amount of our stock-based compensation expense for a number of years.
 
 
We estimate the fair value of each stock option award on the date of grant using a lattice-based option valuation model for annual grants and a Black-Scholes model for reload grants. These models require us to make predictive assumptions regarding future stock price volatility, employee exercise behavior, and dividend yield. Our methods for selecting these valuation assumptions are explained in Note 8 within Notes to Consolidated Financial Statements. In particular, our estimate of stock price volatility is based principally on historical volatility of the options granted, and to a lesser extent, on implied volatilities from traded options on the Company’s stock. For the lattice-based model, historical volatility corresponds to the 10-year contractual term of the options granted; whereas, for the Black-Scholes model, historical volatility corresponds to the expected term, which is currently 2.25 years. We decided to rely more heavily on historical volatility due to the greater availability of data and reliability of trends over longer periods of time, as compared to the terms of more thinly-traded options, which rarely extend beyond two years. At year-end 2007, historical volatilities using weekly price observations ranged from approximately 22% for 10 years to 11% for 2.25 years, while implied volatilities averaged approximately 20% for traded options with terms in excess of six months. Based on this data, our weighted-average composite volatility assumption for purposes of valuing our option grants during 2007 was 17.5%, as compared to 17.9% for 2006. All other assumptions held constant, a one percentage point increase or decrease in our 2007 volatility assumption would increase or decrease the grant-date fair value of our 2007 option awards by approximately 4%.
 
 
To the extent that actual outcomes differ from our assumptions, we are not required to true up grant-date fair value-based expense to final intrinsic values. However, these differences can impact the classification of cash tax benefits realized upon exercise of stock options, as explained in the following two paragraphs. Furthermore, as historical data has a significant bearing on our forward-looking assumptions, significant variances between actual and predicted experience could lead to prospective revisions in our assumptions, which could then significantly impact the year-over-year comparability of stock-based compensation expense.
 
 
SFAS No. 123(R) also provides that any corporate income tax benefit realized upon exercise or ve sting of an award in excess of that previously recognized in earnings (referred to as a “windfall tax benefit”) will be presented in the Consolidated Statement of Cash Flows as a financing (rather than an operating) cash flow. If this standard had been adopted in 2005, operating cash flow would have been lower (and financing cash flow would have been higher) by approximately $20 million as a result of this provision. For 2007 and 2006, the corresponding reduction in operating cash flow attributable to windfall tax benefits classified as financing cash flow was $15 million and $22 million respectively. The actual impact on future years’ operating cash flow will depend, in part, on the volume of employee stock option exercises during a particular year and the relationship between the exercise-date market value of the underlying stock and the original grant-date fair value previously determined for financial reporting purposes.
 
 
For balance sheet classification purposes, realized windfall tax benefits are credited to capital in excess of par value within the Consolidated Balance Sheet. Realized shortfall tax benefits (amounts which are less than that previously recognized in earnings) are first offset against the cumulative balance of windfall tax benefits, if any, and then charged directly to income tax expense, potentially resulting in volatility in our consolidated effective income tax rate. Under the transition rules for adopting SFAS No. 123(R) using the modified prospective method, we were permitted to calculate a cumulative memo balance of windfall tax benefits from post-1995 years for the purpose of accounting for future shortfall tax benefits. We completed such study prior to the first period of adoption and currently have sufficient cumulative memo windfall tax benefits to absorb projected arising shortfalls, such that 2008 earnings are not currently expected to be affected by this provision. However, as employee stock option exercise behavior is not within our control, the likelihood exists of materially different reported results if different assumptions or conditions were to prevail.
 
Retirement benefits
Our Company sponsors a number of U.S. and foreign defined benefit employee pension plans and also provides retiree health care and other welfare benefits in the United States and Canada. Plan funding strategies are influenced by tax regulations. A substantial majority of plan assets are invested in a globally diversified portfolio of equity securities with smaller holdings of debt securities and other investments. We follow SFAS No. 87 “Employers’ Accounting for Pensions” and SFAS No. 106 “Employers’ Accounting for Postretirement Benefits Other Than Pensions” (as amended by SFAS No. 158, “Employers’ Accounting for Defined Benefit Pension and Other Postretirement Plans”) for the measurement and recognition of obligations and expense related to our retiree benefit plans. Embodied in both of these standards is the concept that the cost of benefits


25

provided during retirement should be recognized over the employees’ active working life. Inherent in this concept is the requirement to use various actuarial assumptions to predict and measure costs and obligations many years prior to the settlement date. Major actuarial assumptions that require significant management judgment and have a material impact on the measurement of our consolidated benefits expense and accumulated obligation include the long-term rates of return on plan assets, the health care cost trend rates, and the interest rates used to discount the obligations for our major plans, which cover employees in the United States, United Kingdom, and Canada.
 
 
To conduct our annual review of the long-term rate of return on plan assets, we model expected returns over a 20-year investment horizon with respect to the specific investment mix of each of our major plans. The return assumptions used reflect a combination of rigorous historical performance analysis and forward-looking views of the financial markets including consideration of current yields on long-term bonds, price-earnings ratios of the major stock market indices, and long-term inflation. Our U.S. plan model, corresponding to approximately 70% of our trust assets globally, currently incorporates a long-term inflation assumption of 2.7% and an active management premium of 1% (net of fees) validated by historical analysis. Although we review our expected long-term rates of return annually, our benefit trust investment performance for one particular year does not, by itself, significantly influence our evaluation. Our expected rates of return are generally not revised, provided these rates continue to fall within a “more likely than not” corridor of between the 25th and 75th percentile of expected long-term returns, as determined by our modeling process. Our assumed rate of return for U.S. plans in 2007 of 8.9% equated to approximately the 50th percentile expectation of our 2007 model. Similar methods are used for various foreign plans with invested assets, reflecting local economic conditions. Foreign trust investments represent approximately 30% of our global benefit plan assets.
 
 
Based on consolidated benefit plan assets at December 29, 2007, a 100 basis point reduction in the assumed rate of return would increase 2008 benefits expense by approximately $44 million. Correspondingly, a 100 basis point shortfall between the assumed and actual rate of return on plan assets for 2008 would result in a similar amount of arising experience loss. Any arising asset-related experience gain or loss is recognized in the calculated value of plan assets over a five-year period. Once recognized, experience gains and losses are amortized using a declining-balance method over the average remaining service period of active plan participants, which for U.S. plans is presently about 13 years. Under this recognition method, a 100 basis point shortfall in actual versus assumed performance of all of our plan assets in 2008 would reduce pre-tax earnings by approximately $1 million in 2009, increasing to approximately $7 million in 2013. For each of the three fiscal years, our actual return on plan assets exceeded/(was less than) the recognized assumed return by the following amounts (in millions): 2007–($99); 2006–$257; 2005–$39.
 
 
To conduct our annual review of health care cost trend rates, we model our actual claims cost data over a five-year historical period, including an analysis of pre-65 versus post-65 age groups and other important demographic components of our covered retiree population. This data is adjusted to eliminate the impact of plan changes and other factors that would tend to distort the underlying cost inflation trends. Our initial health care cost trend rate is reviewed annually and adjusted as necessary to remain consistent with recent historical experience and our expectations regarding short-term future trends. In comparison to our actual five-year compound annual claims cost growth rate of approximately 6%, our initial trend rate for 2008 of 8.5% reflects the expected future impact of faster-growing claims experience for certain demographic groups within our total employee population. Our initial rate is trended downward by 1% per year, until the ultimate trend rate of 4.75% is reached. The ultimate trend rate is adjusted annually, as necessary, to approximate the current economic view on the rate of long-term inflation plus an appropriate health care cost premium. Based on consolidated obligations at December 29, 2007, a 100 basis point increase in the assumed health care cost trend rates would increase 2008 benefits expense by approximately $16 million. A 100 basis point excess of 2008 actual health care claims cost over that calculated from the assumed trend rate would result in an arising experience loss of approximately $9 million. Any arising health care claims cost-related experience gain or loss is recognized in the calculated amount of claims experience over a four-year period. Once recognized, experience gains and losses are amortized using a straight-line method over 15 years, resulting in at least the minimum amortization prescribed by SFAS No. 106. The net experience gain arising from recognition of 2007 claims experience was approximately $6 million.
 
 
To conduct our annual review of discount rates, we use several published market indices with appropriate duration weighting to assess prevailing rates on high quality debt securities, with a primary focus on the Citigroup Pension Liability Index® for our U.S. plans. To test the appropriateness of these indices, we periodically conduct a matching exercise between the expected settlement cash flows of our plans and bond maturities, consisting principally of AA-rated (or the equivalent in foreign jurisdictions) non-callable issues with at least $25 million principal outstanding. The model does not assume any reinvestment rates and assumes that bond investments mature just in time to


26

pay benefits as they become due. For those years where no suitable bonds are available, the portfolio utilizes a linear interpolation approach to impute a hypothetical bond whose maturity matches the cash flows required in those years. As of three different interim dates during 2007 and four different dates during 2006, this matching exercise for our U.S. plans produced a discount rate within +/- 20 basis points of the equivalent-dated Citigroup Pension Liability Index®. The measurement dates for our defined benefit plans are consistent with our Company’s fiscal year end. Thus, we select discount rates to measure our benefit obligations that are consistent with market indices during December of each year. Based on consolidated obligations at December 29, 2007, a 25 basis point decline in the weighted-average discount rate used for benefit plan measurement purposes would increase 2008 benefits expense by approximately $15 million. All obligation-related experience gains and losses are amortized using a straight-line method over the average remaining service period of active plan participants.
 
 
Despite the previously-described rigorous policies for selecting major actuarial assumptions, we periodically experience material differences between assumed and actual experience. As of December 29, 2007, we had consolidated unamortized prior service cost and net experience losses of approximately $.6 billion, as compared to approximately $.9 billion at December 30, 2006. The year-over-year decline in net unamortized amounts was attributable primarily to the favorable impact of rising discount rates on our benefit obligations. Of the total unamortized amounts at December 29, 2007, approximately 50% was related to discount rate reductions prior to 2007, with the remainder primarily related to net unfavorable health care claims cost experience (including upward revisions in the assumed trend rate.) For 2008, we currently expect total amortization of prior service cost and net experience losses to be approximately $41 million lower than the actual 2007 amount of approximately $99 million. As discussed on page 14, total employee benefits expense for 2008 is expected to be slightly lower than the 2007 amount, due to increases in the discount rate environment and the on-going phase in of favorable asset returns.
 
 
Assuming actual future experience is consistent with our current assumptions, annual amortization of accumulated prior service cost and net experience losses during each of the next several years would decrease versus the 2008 amount.
 
Income taxes
Our consolidated effective income tax rate is influenced by tax planning opportunities available to us in the various jurisdictions in which we operate. Judgment is required in evaluating our tax positions to determine how much benefit should be recognized in our income tax expense. We establish tax reserves in accordance with FIN No. 48 (which we adopted at the beginning of 2007). FIN No. 48 is based on a benefit recognition model, which we believe could result in a greater amount of benefit (and a lower amount of reserve) being initially recognized in certain circumstances. Prior to the adoption of FIN No. 48, our policy was to establish reserves that reflected the probable outcome of known tax contingencies. Favorable resolution was recognized as a reduction to our effective tax rate in the period of resolution. The initial application of FIN No. 48 resulted in a net decrease to the Company’s consolidated accrued income tax and related interest liabilities of approximately $2 million, with an offsetting increase to retained earnings.
 
 
The Company evaluates a tax position in two-steps in accordance with FIN No. 48. The first step is to determine whether it is more-likely-than not that a tax position will be sustained upon examination based upon the technical merit of the position. In weighing the technical merits of the position, we consider the facts and circumstances of the position; we assume the reviewing tax authority has full knowledge of the position; and we consider the weight of authoritative guidance. The second step is measurement; a tax position that meets the more-likely-than not recognition threshold is measured to determine the amount of benefit to recognize in the financial statements. While reviewing the ranges of probable outcomes, the Company records the largest amount of benefit that is greater than 50 percent likely of being realized upon ultimate settlement. The tax position will be derecognized when it is no longer more-likely-than not of being sustained.
 
 
For the periods presented, our income tax and related interest reserves have averaged approximately $170 million. Reserve adjustments for individual issues have rarely exceeded 1% of earnings before income taxes annually. Significant tax reserve adjustments impacting our effective tax rate would be separately presented in the rate reconciliation table of Note 11 within Notes to Consolidated Financial Statements.
 
 
The current portion of our tax reserves is presented in the balance sheet within accrued income taxes and the amount expected to be settled after one year is recorded in other noncurrent liabilities. Likewise, the current portion of related interest reserves are presented in the balance sheet within accrued other liabilities, with the amount expected to be settled after one year recorded in other noncurrent liabilities.
 


27

print this page section: previous section next section page: previous page next page