Inventory stub period shrinkage estimates - the Tax Court sows confusion.

In the past year, the Tax Court has rendered three opinions on the use of shrinkage estimation when accounting for ending inventories using the perpetual method (Wal-Mart Stores, Inc., TC Memo 1997-1; The Kroger Co., TC Memo 1997-2; Dayton Hudson Corp., TC Memo 1997-260). At issue in each was whether shrinkage estimation produced a "dear reflection of income" under Sec. 471 and, if so, whether there were limits to the time period over which the estimate can be made.

Companies using the perpetual method of inventory accounting often take "cycle counts" of inventory on a rotation basis during the tax year and do not necessarily take a year-end inventory count. Cycle counting is cited as being more accurate; the amount of inventory counted at a given time is smaller (e.g., per department or store) and there is less time pressure than when all of the inventory is counted at year-end. In addition, cycle counting provides management with a more continuous source of information on shrinkage trends and, consequently, on the effectiveness of the company's internal inventory controls.

For a given tax year, physical inventory can exceed the book inventory records (overage) or the book inventory amount can exceed the physical count (shrinkage). Both types of discrepancy are commonly referred to as "inventory shrinkage." In the most frequent case, book inventories will be overstated and cost of goods sold will be understated when a year-end physical inventory count is not taken.

Companies that engage in cycle counting commonly estimate inventory shrinkage from the time of the last physical inventory count to year-end (stub period) for both financial and tax accounting purposes. Adjustments to year-end shrinkage estimates are made when a physical inventory count is subsequently taken in the next year. Such adjustments are usually recorded in the period in which the physical count is conducted.

Tax Accounting Rules Applicable to Inventory In Dayton Hudson Corp., 101 TC 462 (1993), the Tax Court held that Sec. 471 and the related regulations do not, as a matter of law, prohibit the use of shrinkage estimates in computing inventory at year-end for tax purposes, provided the taxpayer takes a physical inventory count at reasonable intervals." In so ruling, the court distinguished shrinkage estimates, which relate to a past event, from a reserve for loss, which relates to a future event. Left to be decided on the facts was whether the taxpayer's method of estimating inventory shrinkage was a "sound accounting system."

The Tax Court in Kroger has noted that what constitutes a sound accounting system under Regs. Sec. 1.471-2(d) has not been answered by the courts. The court interpreted the word "sound" to mean that the accounting system met the two-prong test of Sec. 471 (a). The court's focus, then, was on the taxpayer's methodology used to maintain inventories.

Does Estimating Shrinkage Meet the Best Accounting Practice Requirement? For an inventory accounting method to meet the first prong of the Sec. 471 (a) test, it must conform "as nearly as may be to the best accounting practice in the industry...." Citing Thor PowerTool Co., 439 US 522 (1979), the Tax Court in all three cases interpreted this requirement to mean that the accounting practice was widely used in the taxpayer's industry and conformed to generally accepted accounting principles (GAAP). The IRS agreed that the practice was common in the taxpayers' industries; it did not agree, however, that the practice conformed to GAAP.

The Service argued that estimating inventory shrinkage violated GAAP because such estimates were analogous to booking a reserve for self-insurance, which is prohibited by FASB Statement of Concepts No. 5. The IRS based its arguments on the fact that most shrinkage is due to theft, which is an insurable event. For a company to deduct uninsured theft losses of inventory, it would need to prove the existence of theft, which would require a physical inventory count. In addition, shrinkage estimates for bookkeeping errors could not be reasonably estimated, because there is no way to know the direction of the errors and the expected value should be zero.

The court rejected the Service's arguments on noncompliance with GAAP, noting that a taxpayer must adjust year-end inventory (potentially its largest asset) to reflect shrinkage because merchandise not available for sale no longer meets the definition of an asset (i.e., it does not contribute to future net cash inflows or provide a future benefit).

Does Estimating Shrinkage Result in a Clear Reflection of Income? In the 1993 Dayton Hudson decision, the Tax Court warned that while estimating shrinkage could meet the first prong of the Sec. 471 (a) test, meeting the clear reflection of income prong would be more difficult and would ultimately become a facts and circumstances determination. There are few, if any, bright-line tests in making this determination. However, the Supreme Court did establish in Thor Power Tool that an accounting method sanctioned by GAAP does not, in and of itself, meet the clear reflection of income standard for income tax accounting for inventory. All of the Kroger companies used cycle counting and did not count inventory at year-end. As a result, they recorded shrinkage estimates at the end of the year to adjust their book inventory to what they thought was their actual inventory; the average period covered by the shrinkage accrual was less than three to four months. The reasons for this short period were the timing of the counts and the fact that, on average, Kroger counted its inventory more than once a year.

In developing the shrinkage accruals, different procedures were followed in different parts of the affiliated group. In the Kroger (grocery) stores, shrinkage estimates were developed for each department. For the Superx stores, one shrinkage rate was developed for the entire organization, and that rate was then used by all the departments in all the stores. The end result was that the shrinkage results were determined within the LIFO pools of the Kroger entities, although the estimate itself was developed outside the pools for Superx. This is important for an entity using LIFO, because each pool must be accounted for separately in determining ending inventory and the related cost of goods sold. The shrinkage estimates were developed as a percentage of sales, and the shrinkage for the stub period was determined by multiplying that rate by the stub period sales in the relevant department. It should be noted that the court was not comfortable with the relation between shrinkage and sales, but for a lack of IRS challenge to that relation and no evidence that it was inappropriate, it was accepted as a basis of making shrinkage accruals.

The Service's position in Kroger (and in the other cases discussed)) was that no shrinkage accrual should be allowed, and that shrinkage should only be booked when there is a physical count of the inventory. The court found three basic problems with this position. First, it did not produce a more accurate year-end inventory figure than did the taxpayer's method. In fact, because this approach ignores probables shrinkage since the last count, it would tend to give less accurate results. Second, the IRS's position was the equivalent of estimating current shrinkage based on prior data and ignoring current data (again, making it inherently less accurate than the taxpayer's method). Third it violated the annual accounting requirement in that the current year's shrinkage would include the shrinkage from the stub period of the prior year.

Ultimately, the court found that Kroger's method of shrinkage accrual was sound (as required by Regs. Sec. 1.471-2(d)). In reaching this decision, however, the court seemed to invite the Service to clarify the allowable length of the stub period and the allowable methods for making the shrinkage accrual.

Dayton Hudson

About five months after Judge Halperin issued the Kroger decision, he help that Dayton Hudson's method of estimating inventory shrinkage did not clearly reflect income and was not sound.

Based on the published facts of each case, it is difficult to distinguish Dayton Hudson from Kroger. Dayton Hudson and its affiliates also used the retail, dollar-value LIFO method of accounting for its inventories. The companies used cycle counting and generated shrinkage accruals for the stub period from the last physical count to the end of the tax year in computing year-end inventory and cost of goods sold.

One (seemingly) minor difference between the two fact patterns is that Kroger counted its inventory two to three times a year, while Dayton Hudson on a average counted its inventory once a year.

A possibly significant difference between the cases was the way in which the two companies derived their shrinkage rate estimates. While Kroger basically did its estimating by pools, the Target Corporation division of Dayton Hudson developed a company-wide estimate that was allocated to the individual stores and then to the inventory pools within the stores. For the Dayton HUdson division stores, the accrual rates were developed for each department and pool (as was generally done by Kroger).

The most obvious difference between Kroger and Dayton Hudson is in the testimony of the petitioner's expert witnesses. In Dayton Hudson, the accounting expert analyzed only the accruals for the Target Corporation division and not the Dayton Hudson division accruals. Further, the analysis was not tied directly to the LIFO pools (apparently because of lack of data). Similarly, in Kroger, the expert witness analysed the shrinkage accruals at the business or division level. However, in Kroger, that business-level analysis was supported by checks of accuracy at the pool level that apparently was not possible with the data generated by Dayton Hudson's accounting. In addition, the more frequent counting of inventory by Kroger allowed more convincing tests of the accuracy of the shrinkage accruals.

The end result of these relatively minor differences in two very similar cases is that, while Kroger won its case, Dayton Hudson lost. Specifically, Dayton Hudson's method of accruing estimated inventory shrinkage was found to not clearly reflect income and thus to be unsound under Regs. Sec. 1.471-2(d).

Wal-Mart

While one might question the differing results in the two cases just discussed, the underlying reasoning seems to be straightforward and sound. However, it is more difficult to discern the court's logic in Wal-Mart.

The general outlines of the Wal-Mart case are similar to those of Kroger and Dayton Hudson. Wal-Mart Stores is another retailer that, in general, keeps its inventory on the retail method, employing the dollar-value LIFO flow assumption. The company uses cycle counting and makes shrinkage accruals to estimate shrinkage in its inventory from the most recent count date to its year-end.

For the part of the business on LIFO. Wal-Mart had 36 pools in each of three corporate entities. Because a company is not permitted to pool across entities, there were 108 pools in the consolidated corporate group. A basic principle underlying the LIFO method is that each pool must be accounted for separately. In deriving the shrinkage estimates for the pools, an overall store shrinkage was determined. The various store shrinkage amounts were then consolidated and allocated to the various pools based an verified shrinkage. However, this estimated shrinkage was a apparently only used indirectly in determining the year-end . The year-end inventory number for each pool was derived by allocating the consolidated inventory (net of shrinkage) to the departments or pools occurring to the ending inventory (net of shrinkage) reported on the books. The validity of LIFO inventories basically determined by allocation would appear to be highly questionable. Given that Dayton HUdson apparently lost its inventory case because of its inability to tie its shrinkage accruals to its pools, one has to wonder how Wal-Mart won.

Another problem with the Wal-Mart shrinkage accruals was the arbitrary adjustments applied to the estimates. Essentially, the shrinkage estimates were derived by using a three-year moving average of the accrual shrinkage rates for the various stores; the average was then subjects to a floor and a ceiling. The purpose and basis of these floor and ceiling adjustments was not explained in the opinion, only that they were ultimately set by the company president based on advice from the internal audit department (which used a five-year average of shrinkage to make its recommendation). The problems with the type of arbitrary procedure can be seen in the opinion, which shows that, in 1986, store number 782 reported an average of 0.32% based on actual data, but because of the floor procedure, it adjusted its inventory as though it had a 1.40% shrinkage rate.

A further complication in the Wal-Mart case has to do with the stub periods for which the shrinkage accruals were done. In Wal-Mart, the stub periods were very long, especially relative to Kroger. No Wal-Mart stub period was shorter than three months in 1986. In all the years being considered, Wal-Mart took no inventories in November or December, and in 1986, all of the January counts were recorded in February of the next year. Because most of the counts appear to have been made in the period from March through September, the average stub period must have been approximately seven to eight months. In addition, by 1986, at least 93 of the stores had stub periods of 12 months, To the extent that the length of the periods tends to distort the reflection of income, it would appear that Wal-Mart had a less defensible methodology than Dayton Hudson.

In addition to length, the 12-month stub periods raise other concerns. First they arose because Wal-Mart counted the inventory at those 93 stores in January (the last month in its fiscal year), but did not book the results of the inventory until February (the first month in its next fiscal year). Second, it should be noted that in 1993, Wal-Mart took 39 inventories in January and booked them all in the same month. In 1986, it took at least 93 inventories in January and booked them all in February (the next tax year).

Making Sense of the Tax Court's Three Decisions With the facts and results of these three 1997 cases in mind, it in instructive to go back and consider the Tax Court's opinion in the 1993 Dayton Hudson case, which laid the groundwork for the decisions. In that case, the Tax Court held that Regs. Sec. 1.471-2(d) did not, as a matter of law, prohibit the use of shrinkage accruals. The court also noted that there was a second issue to be litigated as to whether any particular method of arriving at such accruals clearly reflected income.

This decision was not unanimous. It was written by Judge Halperin and agreed to by nine other Tax Court judges; two other judges concurred, one in result and the other in a separate opinion. Four judges dissented, with Judge Gerber writing a strongly worded dissent, focusing on two issues. First, the stub period estimates allowed taxpayers to manipulate their income through the estimates. Second, the shrinkage accruals amounted to reserves that were disallowed under the Supreme Court's Thor Power Tool decision. In light of the bias toward ovveraccrual of shrinkage in Dayton Hudson (i.e., in 12 out of 14 years, there was an overaccrual), and the accounting practices of Wal-Mart, Judge Gerber's concern seem to be valid.

It seems fairly apparent that Treasury needs to put parameters on the time period over which shrinkage estimation should be allowed. The Taxpayer Relief Act of 1997 allows the use of shrinkage estimate, provided the taxpayer takes a physical count of inventories at each location on a "regular and consistent basis." It would behoove Treasury to write regulations that define "regular and consistent basis" in more concrete terms and establish the parameters of the stub periods.

The three recent Tax court cases provided a blueprint for what conditions should be specified. First, the length of the allowable stub period should be specified, with a bias toward short stub periods (an average of three months, perhaps). Second, the allowable method(s) of making the estimates of shrinkage should be specified, with a bias toward objective data and away from judgmental adjustments to the objective results. When a LIFO method is used to value inventories, it should be made clear that the shrinkage estimates needs to be made separately for each pool. Finally, it should be specified that any inventories taken during the year must be reflected in that year's income computatation.

The Service should be satisfied by this type of change, because it would mitigate taxpayers' ability to manipulate their income thorough the accruals. Taxpayers should encourage this type of rule change to forestall the Irs from eliminating such a accruals completely, based on abusive situation. Until this issue is ultimately resolved, taxpayers like Kroger (who keep their stub periods relatively short and tie their shrinkage estimates closely to their pools) will probably not have difficulties with the Service and, if they do, will be able to win their cases in the Tax Court; taxpayers who choose to push the envelope, like Wal-Mart, can anticipate continuing IRS challenges and uncertain results in the courts. However, if the Wal-Mart decision is upheld, taxpayers will have a strong incentive to shift to perpetual inventories, adopt cycle counting and implement those counts to maximize the stub periods, thus giving themselves maximum flexibility in determining their taxable income.