Statistical sampling may create opportunity for M&E analysis.
Until recently, the only method thought available to correct misreporting was a thorough review of each expense record, a proposition not likely cost-effective. However, for companies with relatively large M&E accounts, a new method may be available. An IRS memorandum (released March 14, 2002 to the industry directors of large and mid-size businesses) provides acceptable standards for using statistical sampling in preparing a return. Although these guidelines offer audit direction, they are not intended to serve as a technical position. However, they are planned for future release as a revenue procedure.
The memorandum proposes that through the use of statistical sampling, an estimate generated from analyzing only a portion of a taxpayer's M&E account may be used to project the expected amount of incorrect bookings made to a large M&E account. This could potentially create an efficient way to fully deduct expenses that would otherwise be subjected to the 50% limit.
The memorandum provides guidance for determining when the use of sampling is proper. First, agents should determine whether a taxpayer has appropriately used a probability sample to support or stand as the primary evidence of tax amounts. Appropriateness is a facts-and-circumstances determination.
The key elements examined in a given situation include the cost of analyzing the volumes of data available, the cost of analyzing the sample, and the availability of any other records that have a greater probative value. Based on these criteria, sampling would be appropriate if there is a compelling reason for its use and taxpayers cannot reasonably obtain more accurate information.
Second, agents are to examine the validity of the resulting estimate obtained from the statistical sample. Three conditions must be established to support its validity. First, the taxpayer must maintain adequate documentation of the sampling techniques performed. This includes a written sampling plan, generally including information such as the plan objective, definitions of the population (including reconciliation to the return) and the sampling flame and unit, the source of the random numbers and the method used in selecting them, the sample size and documentation of how the size was arrived at, the methods used to associate the random numbers to the frame, the steps taken to ensure that the serialization of the frame is carried out independently of the drawing of random numbers, the steps to be taken in evaluating the sampling unit and the appraisal method for the sample.
Second, the estimate must be based on a probability (i.e., statistical) sample, in which each sampling unit has a known (nonzero) chance of selection. Finally, the estimate for a ample must be computed at the least advantageous 95% one-tailed confidence limit. The memorandum contains samples of the specific sampling plans and formulas that would be permissible to create a valid statistical sample. However, using a sampling technique does not exempt a company from complying with the substantiation requirements of Temp. Regs. Sec. 1.274-5T(c)(2).
This guidance appears to validate the argument that a company with a large M&E account could use statistical sampling to find incorrectly classified transactions. Even in a middle-market company, M&E accounts can have thousands of entries posted in a year, making any attempt to verify the correct reporting of each expense a nightmare and most likely not cost effective. Due to the subjective nature of many of the laws governing deductibility of the various expenses that may get posted to a M&E account, it would also be extraordinarily difficult to establish an accounting system that could reliably classify each expense properly. This would seem to fit within the memorandum's guidelines, allowing companies to use the sampling technique.
Although the Service issued this memorandum in March 2002, the IRs can still scrutinize this method, by evaluating the facts and circumstances of each case. For example, in Field Service Advice (FSA) 200209028, it denied the use of sampling, declaring that sampling of any kind was invalid on the basis that it was only "a form of a close approximation," when "each and every item covered under section 274(d), requires an expenditure-by-expenditure determination."
The FSA's facts appear to meet the "appropriate" test, as the company estimated that over 50,000 entries were posted to its M&E account in one year, making the cost to analyze each item ineffective. In support of its position, the company also used Litigation Guideline Memorandum (LGM) TL.97 (9/9/92), Use of Statistical Sampling Techniques in Examination of Tax Returns, which instructs Service personnel to support taxpayers' use of sampling.
Although the FSA seems to be in complete disagreement with the suggested increased acceptance of sampling, it may not have disclosed all the facts involved in its decision; it did not specify the reasons for denial. The FSA does focus on the substantiation requirement in its argument against statistical sampling, but does not delineate why the sample is insubstantial. In the instant case, the sampling plan may have been invalid, or the taxpayer may have been attempting to use the sampling technique to avoid the normal substantiation requirements.
In conclusion, although using statistical sampling to substantiate returns may still be considered a new or risky technique, the IRS's internal guidelines for acceptance of this technique indicate that it is increasing its acceptance of properly designed statistical samples for obtaining tax information when a detailed analysis is impractical.
As with any position, consideration must be given to ensure appropriateness. Thus, the facts and circumstances of each case determine not only whether sampling is appropriate, but also whether the result is a valid estimate that can be used for tax purposes.
FROM SHARON L. COOK, CPA, AND DAVID J. HOLETS, SOUTH BEND, IN