This case presents the question whether a retailer may account for unverified inventory shrinkage, and if so, whether the method used by Wal-Mart Stores, Inc. & Subsidiaries is permissible. The Commissioner of Internal Revenue (Commissioner) appeals from the tax court's2 decision reversing the Commissioner's determination of federal tax deficiencies. We affirm the tax court.

The parties have stipulated the relevant facts. Wal-Mart Stores, Inc., was the parent company of a group of affiliated corporations, including Kuhn's-Big K Stores Corp. (Kuhn's), Big K Edwards, Inc. (Edwards), and Sam's Wholesale Clubs (Sam's).3 The Taxpayer filed consolidated tax returns for the fiscal years ending in late January of 1984 (referred to herein as the 1983 taxable year), 1985, 1986, and 1987. The Taxpayer's principal place of business was in Bentonville, Arkansas.

During the taxable years at issue, Wal-Mart operated its stores as mass discount retailers, carrying between 60,000 and 90,000 different merchandise items in each store. Wal-Mart purchased more than $22 billion in merchandise, turning its inventory over as often as 4.5 times per year. Sam's ran its stores as discount warehouses, carrying between 3,500 and 5,000 different merchandise items, acquiring more than $2.6 billion in merchandise. The Taxpayer's operations grew at a resounding pace from 1983 to 1986. For example, the number of Wal-Mart stores increased from 642 to 980 and the number of Sam's stores increased from 3 to 49. The Taxpayer utilized an extensive distribution and tracking system to maintain optimal inventories at each store. The Taxpayer's inventory system is commonly revered as the finest in the retail industry.

For both financial reporting and tax purposes, the Taxpayer used the accrual method of accounting and maintained a perpetual inventory system. Under the perpetual inventory system, the cost or quantity of goods sold or purchased is contemporaneously recorded at the time of sale or purchase. The system continuously shows the cost or quantity of goods that should be on hand at any given time. The Taxpayer performed physical inventories to confirm the accuracy of the inventory as stated in the books, and made adjustments to the books to reconcile the book inventory with the physical inventory.4

The Taxpayer's physical inventories were taken at its stores in rotation throughout the year. The Taxpayer did not take physical inventories during the holiday season (November, December, and the first week of January). The Taxpayer refers to this technique, which is common in the retail industry, as cycle counting. Cycle counting is necessitated by the difficulty in conducting physical inventories at every store on the last day of the year. This technique also provides management with feedback on the effectiveness of its inventory management and facilitates the use of experienced personnel to conduct the physical inventories.

Forty-five days prior to conducting a physical inventory in one of its stores, Wal-Mart's internal audit department would send the store a preparation package, which included instructions on how to prepare for the physical count. Each physical count was then conducted by a team of independent counters (18 to 40 persons) and representatives from Wal-Mart's loss prevention department (1 to 2 persons), internal audit department (1 to 3 persons), and operations division (1 to 2 persons). Wal-Mart's independent auditors, Ernst & Young, also sent representatives to randomly selected physical counts to test their accuracy. The independent counters generally counted every inventory item. The results of the physical count were then reconciled with the book inventory. The reconciliations were reviewed by Wal-Mart's internal audit department. Generally, Wal-Mart did not record the results of a physical inventory in its books until the following month.5

Sam's conducted its physical inventories in the same manner except that physical counts were usually taken twice a year and recorded the very next day. Sam's also periodically conducted item audits, counting the goods on hand for a particular merchandise unit and recording those results the next day. The physical inventories of both Wal-Mart and Sam's usually revealed shrinkage.

Shrinkage (or overage) is the difference between the inventory determined from the perpetual inventory records and the amount of inventory actually on hand. Because shrinkage reduces profits, the Taxpayer has devoted extensive resources to monitoring and mitigating shrinkage. There are many causes of shrinkage, including employee theft, customer theft, vendor theft, damage, breakage, spoilage, accounting and recording errors, errors in marking retail prices, cash register errors, markdowns taken and not recorded, errors in accounting for customer returns, and errors in accounting for vendor receipts and returns.

Because the Taxpayer did not conduct a physical inventory at year-end, its perpetual inventory records did not account for any shrinkage that may have occurred during the period between the date of the last physical inventory and the taxable year-end. The parties refer to this period as the stub period. Left unadjusted, the Taxpayer's book records would overstate income because the stub period shrinkage results in a decrease to ending inventory, thus increasing the cost of goods sold and reducing gross income.6

In adjusting its books to reflect stub period shrinkage, Wal-Mart estimated stub period shrinkage for each store monthly by multiplying a retail shrinkage rate by the store's sales during that month. At new stores, the retail shrinkage rate was fixed by management at 3% of sales in the 1983 and 1984 taxable years and 2% of sales in the 1985 and 1986 taxable years. Wal-Mart used that fixed rate from the date the store opened until its first physical inventory, which rarely took place less than six months after opening. After taking the first physical inventory at a new store, Wal-Mart computed a shrinkage rate for that store by dividing the store's shrinkage at retail, as verified by the first inventory, by the store's sales for the period starting with the date the store opened and ending on the inventory date. After Wal-Mart took a second inventory, it computed the store's shrinkage rate by dividing the store's shrinkage, as verified by the first two inventories, by the store's sales starting with the opening date and ending on the second inventory date. After the third inventory, it computed the shrinkage rate by dividing the store's shrinkage, as verified by the first three inventories, by the store's sales starting with the opening date and ending on the third inventory date. After taking a fourth and each subsequent inventory, the retail shrinkage rate was based on a rolling average of the historical shrinkage over the preceding three inventories.

The retail shrinkage rate was subject to certain floor and ceiling percentage limitations that were set by Wal-Mart's management, based primarily on a weighted five-year average of shrinkage. If the computed shrinkage rate was below the floor limitation, or if it was an overage, the computed rate was replaced by the floor limitation. Likewise, if the computed rate exceeded the ceiling limitation, it was replaced by the ceiling limitation.

Sam's consistently estimated the stub period shrinkage at .2% of sales. That rate was determined by management based on their analysis of historical results from warehouse operations.

Each physical inventory would reveal a difference (estimation error) between the estimated shrinkage, as reflected in the perpetual inventory records, and the verified shrinkage. At that time, the Taxpayer would adjust its inventory records to reflect any estimation error. Consequently, for each taxable year, the Taxpayer's total adjustments to its inventory records for shrinkage would include (1) any shrinkage estimates for the period from the start of the taxable year until the physical inventory date, (2) any estimation error adjustment, and (3) any stub period shrinkage estimates. The Taxpayer also recomputed the retail shrinkage rate for the store and used that new rate to estimate shrinkage until the next physical inventory. Thus, if the Taxpayer overestimated or underestimated a store's shrinkage in year one, it would adjust that store's shrinkage rate in year two. In accordance with the cycle counting technique, this was a continuous process throughout the year for each of the Taxpayer's stores.

Wal-Mart estimated shrinkage for each store, but not for each department within each store. It used a series of computations to allocate the estimated stub period shrinkage to each department. Once these allocations were made, Wal-Mart used the adjusted ending inventory to make its LIFO computations, see Treas. Reg. § 1.472-1, which were made on a division-wide basis and not at the individual store level.

The Taxpayer reported shrinkage on an aggregate basis for both financial and tax purposes. The practice of estimating shrinkage as a percentage of sales is prevalent in the retail industry. Ernst & Young issued unqualified opinions to the effect that the Taxpayer's financial statements for the subject years conformed with Generally Accepted Accounting Principles (GAAP). Ernst & Young periodically reviewed the Taxpayer's accounting method for estimating shrinkage during the stub period and never recommended a change.

The Commissioner disallowed the Taxpayer's use of shrinkage estimates. The Commissioner issued the Taxpayer two notices of deficiency, one in the amount of $9,937,544.68 for the 1983 taxable year and $4,084,255.28 for the 1984 taxable year and the other in the amount of $9,381,626.40 for the 1985 taxable year and $8,206,962 for the 1986 taxable year.7 The Taxpayer filed a timely petition in the tax court seeking redetermination of the resulting deficiencies.

The tax court, following a prior decision, see Dayton Hudson Corp. and Subsidiaries v. Commissioner, 101 T.C. 462, 465, 1993 WL 473332 (1993) (Dayton Hudson I ) (12-5 reviewed decision), held that a method of computing shrinkage, including estimated shrinkage, is permissible if "sound." Citing Treasury Regulation section 1.471-2(a), the tax court defined a sound inventory method as one that (1) conforms to the best accounting practice in the trade or business and that (2) clearly reflects income. The tax court then applied the two-part test to conclude that the Commissioner had abused her discretion in finding that the Taxpayer's inventory methods were not sound.

The Commissioner appeals, asserting that the regulations prohibit the reduction of ending inventory by estimated shrinkage. The Commissioner further contends that, even if shrinkage estimates are permissible, the Taxpayer's method of accounting did not clearly reflect income.

We first address whether all shrinkage estimates are prohibited from being used to reduce ending inventories and then turn to whether the Taxpayer's particular method of accounting for estimated shrinkage is permissible.

A. Use of Shrinkage Estimates to Reduce Ending Inventories

The general rule for inventory accounting is that:

Whenever in the opinion of the [Commissioner] the use of inventories is necessary in order clearly to determine the income of any taxpayer, inventories shall be taken by such taxpayer on such basis as the [Commissioner] may prescribe as conforming as nearly as may be to the best accounting practice in the trade or business and as most clearly reflecting the income.

I.R.C. § 471(a).8 Determining the propriety of a taxpayer's reduction of ending inventories by estimated shrinkage requires construction of the relevant sections of the Internal Revenue Code (Code) and the regulations.9 Thus, this issue presents a question of law subject to plenary review. See Musco Sports Lighting, Inc. v. Commissioner, 943 F.2d 906, 907 (8th Cir.1991).

It is undisputed that the use of inventories is necessary to determine income because the Taxpayer is engaged in the sale of merchandise. See Treas. Reg. § 1.471-1. An inventory method must (1) conform as nearly as may be to the best accounting practice in the trade or business, and (2) clearly reflect income. See Treas. Reg. § 1.471-2(a). The tax court analyzed the Taxpayer's inventory method under this two-prong test. The Commissioner contends that independent of this two-prong test, Treasury Regulation section 1.471-(2)(d) precludes the use of all shrinkage estimates. The regulation provides in relevant part:

Where the taxpayer maintains book inventories in accordance with a sound accounting system in which the respective inventory accounts are charged with the actual cost of the goods purchased or produced and credited with the value of goods used, transferred, or sold, calculated upon the basis of the actual cost of the goods acquired during the taxable year (including the inventory at the beginning of the year), the net value as shown by such inventory accounts will be deemed to be the cost of the goods on hand. The balances shown by such book inventories should be verified by physical inventories at reasonable intervals and adjusted to conform therewith.

Treas. Reg. § 1.471-2(d).

The Commissioner interprets the regulation to provide that (1) the Taxpayer's book inventories, without adjustment for estimated shrinkage, will be deemed to reflect the costs of goods on hand at year-end, and (2) book inventories may be adjusted to reflect shrinkage only when a physical count is conducted. The Taxpayer contends that adjusted book inventories are deemed to reflect the cost of the goods on hand at year-end, if the adjustment is made in accordance with a sound accounting system.

The regulatory scheme recognizes that shrinkage occurs, as it provides that book inventories "should be verified by physical inventories at reasonable intervals and adjusted to conform therewith." The Commissioner does not dispute that the proper way to account for verified shrinkage is to reduce book inventories, resulting in an increase to cost of goods sold. Nonetheless, the Commissioner asserts that book inventories may be reduced only by verified shrinkage and not with estimates. Nothing in the text of the regulation, however, distinguishes between estimated shrinkage and shrinkage that has been verified by physical count. We will not read in such a distinction.

Our reading comports with the background of the Commissioner's regulatory scheme. Prior regulations construing the predecessor to section 471 provided that physical inventories must be taken at year-end. See Regs. 45, art. 1588(3)(B) (1921). The Commissioner dispensed with that requirement in 1922. See Regs. 45, art. 1582, as amended by T.D. 3296, I-1 C.B. 40 (1922) (providing that physical inventories need not be taken at year-end, but only at reasonable intervals). In removing the burdensome year-end physical inventory requirement, the Commissioner opened the door to the industry practice of estimating shrinkage during the stub period. Nothing in the Commissioner's regulatory scheme, which provides that book inventories may substitute for year-end physical inventories, prohibits the reduction of ending inventories to account for estimated inventory shrinkage. Moreover, the accrual method of accounting contemplates the use of estimates. See, e.g., Treas. Reg. § 1.461-1(a)(2)(ii). Accordingly, we construe Treasury Regulation section 1.471-2(d) to provide that book inventories, as adjusted for estimated stub period shrinkage, are "deemed to be the cost of the goods on hand" provided that (1) such inventories are maintained in accordance with a sound accounting system; (2) goods are valued at actual cost; and (3) physical inventories are taken at reasonable intervals.

Here, the second and third criteria are not at issue. The only controversy is whether the Taxpayer "maintain[ed] book inventories in accordance with a sound accounting system." The regulations do not define "sound accounting system," and we find no plausible construction of the phrase that would exclude the use of shrinkage estimates. See Dayton Hudson I, 101 T.C. 468. In mandating a "sound accounting system," we find that the regulation prescribes no higher standard than that of section 471(a), which requires that an inventory method must conform "to the best accounting practice in the industry" and result in a clear reflection of income.10 I.R.C. § 471(a).

The phrase "best accounting practice in the industry" is synonymous with GAAP. See Thor Power Tool Co. v. Commissioner, 439 U.S. 522, 532, 99 S. Ct. 773, 58 L. Ed. 2d 785 (1979). The tax court found, and the Commissioner concedes on appeal, that the Taxpayer's use of shrinkage estimates conformed to GAAP. We thus conclude that Treasury Regulation section 1.471-2(d) permits the Taxpayer's method of estimating shrinkage provided that such method resulted in a clear reflection of income. We will analyze the Taxpayer's particular method in Part IIB.

We reject the Commissioner's argument that permitting the reduction of ending inventories by estimated shrinkage runs afoul of the Supreme Court's decision in Thor Power Tool. In Thor Power Tool, the taxpayer, in accordance with GAAP, wrote-down certain "excess inventory" to its net realizable value, which in most cases was its scrap value. 439 U.S. at 530, 99 S. Ct. 773. However, despite taking the inventory write-down, the taxpayer continued to offer the inventory for sale at full price. See id. at 529, 99 S. Ct. 773. The Court applied the regulations to the facts and held that the Commissioner properly disallowed the taxpayer's write-down as not clearly reflecting income because the taxpayer's method "was plainly inconsistent with the Regulations." Id. at 538, 99 S. Ct. 773. In Thor Power Tool, the regulations clearly articulated when a taxpayer could value inventory below its market value, which is distinct from the present situation because, here, the Taxpayer's method is not clearly covered by the relevant regulations.

The Commissioner additionally contends that the Taxpayer's reduction of ending inventories for estimated shrinkage was actually a deduction for a reserve, which may not be taken unless Congress has specifically so provided. See id. at 541-44, 99 S. Ct. 773. A reserve is used to set aside funds for an "estimated future loss" or expense. Id. at 542, 99 S. Ct. 773. Here, the Taxpayer estimated the amount of shrinkage that had actually occurred during the stub period, not a loss that may occur in the future. Therefore, estimated shrinkage is not an impermissible reserve.

In sum, the Commissioner's arguments are unavailing, and we find nothing within the relevant statutes, regulations, or case law that prohibits the use of shrinkage estimates.

B. Taxpayer's Method of Accounting for Shrinkage

The general rule for methods of accounting provides, "Taxable income shall be computed under the method of accounting on the basis of which the taxpayer regularly computes his income in keeping his books." I.R.C. § 446(a). For all relevant years, the Taxpayer has used the same method of accounting for book and tax purposes. Nonetheless, "if the method used does not clearly reflect income," the Commissioner has broad authority to prescribe a method that "does clearly reflect income." I.R.C. § 446(b). In the two notices of deficiency issued to the Taxpayer, the Commissioner prescribed a method that does not account for shrinkage until verified by a physical count.11 The tax court determined that the Commissioner had abused her discretion in acting under section 446(b), because the Taxpayer's method reflected income clearly.

The "clearly reflect income" standard is a technical standard which is not easily resolved by resorting to "experience with the mainsprings of human conduct." Commissioner v. Duberstein, 363 U.S. 278, 289, 80 S. Ct. 1190, 4 L. Ed. 2d 1218 (1960). Consequently, we find that the tax court's conclusion that a particular method of accounting resulted in a clear reflection of income is a conclusion of law, or at least a mixed question of law and fact, subject to de novo review. See, e.g., Black Hills Corp. v. Commissioner, 73 F.3d 799, 804 (8th Cir.1996); cf. Ford Motor Co. v. Commissioner, 71 F.3d 209, 212 (6th Cir.1995) (holding that the clear reflection of income standard presents a question of ultimate fact, which is reviewed de novo). Nonetheless, we respect the tax court's purely factual findings unless they are clearly erroneous, and we accord the Commissioner's determination the same discretion. See Black Hills Corp., 73 F.3d at 804.

The Commissioner's discretion to prescribe a method that clearly reflects income cannot be disturbed unless it is clearly unlawful or plainly arbitrary. See Thor Power Tool, 439 U.S. at 532-33, 99 S. Ct. 773. However, the Commissioner may not require a taxpayer to change from an accounting method that reflected income clearly, merely because the Commissioner believes that her method more clearly reflects income. See, e.g., Louisville and Nashville R.R. v. Commissioner, 641 F.2d 435, 438 (6th Cir.1981). We find that the Taxpayer's method resulted in a clear reflection of income because it complied with GAAP, it was consistent, and it produced accurate results.

The Code does not define the phrase "clearly reflect income." The regulations, however, provide some guidance. Treasury Regulation section 1.446-1(a)(2) states that an accounting method "which reflects the consistent application of generally accepted accounting principles in a particular trade or business in accordance with accepted conditions or practices in that trade or business will ordinarily be regarded as clearly reflecting income." Here, the tax court found that the Taxpayer's method of accounting for stub period shrinkage consistently applied GAAP and was in accordance with the best accounting practice in the retail industry.

The Commissioner does not dispute that finding, but instead asserts that the tax court placed undue emphasis on financial accounting principles. We disagree. Compliance with GAAP will ordinarily "pass muster for tax purposes," though it does not create a presumption of validity. See Thor Power Tool, 439 U.S. at 540, 99 S. Ct. 773. The tax court did not grant the Taxpayer an improper presumption. After stating that the Taxpayer's compliance with GAAP is not dispositive, the tax court carefully examined the Taxpayer's method. We will do the same.

Wal-Mart's accounting method resulted in a consistent estimation of stub period shrinkage. Treasury Regulation section 1.471-2(b) provides that "[i]n order clearly to reflect income, the inventory practice of a taxpayer should be consistent from year to year, and greater weight is to be given to consistency than to any particular method of inventorying or basis of valuation." Wal-Mart's estimated shrinkage rates were generally based on objective factors--historical shrinkage averages--that did not leave room for manipulation from year-to-year. Other than setting new store rates, Wal-Mart had no discretion in calculating estimated shrinkage.

Moreover, Wal-Mart, an industry leader in inventory management and shrinkage control, used the same accounting method for taxes, financial reporting, and evaluations of daily operations. Although Wal-Mart made minor alterations to its method of estimating shrinkage during the years in issue, we find no clear error in the tax court's factual finding that Wal-Mart consistently calculated shrinkage estimates.

Accuracy is also indicative of an accounting system that clearly reflects income. See Caldwell v. Commissioner, 202 F.2d 112, 115 (2d Cir.1953) (stating that "income should be reflected with as much accuracy as standard methods of accounting practice permit"). Obviously, in permitting cycle counting, the regulations do not require absolute accuracy. See Treas. Reg. § 1.471-2(d) (providing for verification and adjustment of book inventories). We agree with the tax court's finding that Wal-Mart's method of estimation produced reasonably accurate results. In reaching that conclusion, the tax court analyzed the overall accuracy of Wal-Mart's method by considering the testimony of the various expert witnesses presented by the parties.

We note that it is difficult, if at all possible, to devise an infallible method of comparing estimated shrinkage to what actually transpired. The only verified calculation of shrinkage is for the physical inventory period. However, because each physical inventory period covers two different tax years, there is no direct proof available to show exactly what proportion of the shrinkage occurred during the stub period and what proportion occurred during the year-end-to-physical-inventory period. Consequently, we must rely on indirect proof.

The tax court first compared the total amount of booked shrinkage as a percentage of sales for the taxable year to the total amount of verified shrinkage as a percentage of sales during the physical inventory period:

Taxable Year         Booked         Verified
    1983              1.47            1.48
    1984              1.11            1.06
    1985              1.36            1.37
    1986              1.00            1.00

The figures for booked shrinkage consist of both the amount of estimated shrinkage recorded by Wal-Mart for the taxable year and the estimation error adjustment for the just-concluded physical inventory period. The figures for verified shrinkage are for the physical inventory period.

As the tax court recognized, this comparison has its limitations in that it compares shrinkage from two different time frames. Nonetheless, the comparison is useful because these periods overlap. Moreover, the inclusion of the estimation error adjustment in the figures for booked shrinkage shows the effect that Wal-Mart's adjustments for physical inventories had on book inventories. This reconciliation was a fundamental part of Wal-Mart's inventory method. The Commissioner argues that the inclusion of adjustments for overestimates from the previous year masks the true extent of an overestimate of shrinkage in the current year, potentially resulting in a perpetual tax deferral if the taxpayer consistently overestimates shrinkage. But this argument assumes that there is a stub period shrinkage estimation error in each succeeding year that is at least equal to the original error. This is unlikely considering that Wal-Mart's estimated shrinkage rate is self-adjusting in accordance with prior verified shrinkage.

In an attempt to demonstrate the inaccuracy of Wal-Mart's method, the Commissioner compares estimated shrinkage, without adjustment for prior estimation errors, to verified shrinkage. The comparison is limited to the physical inventory period only. The Commissioner's analysis again ignores the estimation error adjustment, which is a fundamental part of Wal-Mart's accounting method. Nevertheless, even if we accepted the Commissioner's flawed analysis, it would not persuade us that Wal-Mart's method was inaccurate. If prior estimation errors were excluded from the calculation of estimated shrinkage, the difference between verified and estimated shrinkage during the physical inventory period follows:

Comparison of Verified Shrinkage to Unadjusted Estimated Shrinkage

                    Stated as a Percentage of Sales
                 Estimated Shrinkage  Verified Shrinkage    Difference
     1983               1.72                 1.48              .24
     1984               1.58                 1.06              .52
     1985               1.52                 1.37              .15
     1986               1.46                 1.00              .46