On February 2, 2022, the California Department of Tax and Fee Administration (CDTFA) held an interested parties meeting (IPM) to discuss proposed amendments to sales tax audit manual (AM) Chapter 13, “Statistical Sampling,” and Chapter 4, “General Audit.”

IN DEPTH

Prior to the IPM, the CDTFA released a lengthy discussion paper outlining the extensive proposed changes to the AM, which includes:

Removing the three error rule

The current text of AM 1308.05 explains that when a sample produces only one or two errors, the auditor must evaluate whether these errors are representative or whether it is possible they indicate problems in certain areas that could be examined separately. Under the proposed amendment, the same evaluation standards would still be in place without the minimum error requirement. According to the CDTFA, the proposed removal of the three error rule is because of the fact that “the number of errors identified in a sample does not give any indication whether the sample is representative or not…If the combined evaluation evaluates within Department [CDTFA] standards, it is justified to project the results even if one or two errors are found.”

Requiring 300 minimum sample items per stratum unless the auditor obtained approval from CAS to select fewer than 300

Currently, the “minimum sample size of at least 300 items of interest is to be used in all tests, except where the auditor can support a smaller sample size and it evaluates well.” (AM 1303.05) Under the new subsection titled “Materiality,” a minimum of 300 sample items per test stratum is recommended. Computer Audit Specialist (CAS) approval is required for selecting less than 300 sample items per test stratum.

Refunding Populations: A minimum of 100 sample items per stratum is required

In the section addressing sampling refund populations (AM 1305.10), the proposed amendment would permit auditors to select as few as 100 sample items per test stratum without CAS approval, provided the expected error rate is sufficiently high (greater than 20%). No such rule exists under the current text of Chapter 13.

Contacting CAS when the prior audit had 300 hours charged to it is now mandatory

In contrast, under the current rule, it is mandatory that CAS be contacted when the prior audit expended 400 or more hours or if CAS was involved in the prior audit.

Replacing Credit Methods 1, 2 and 3 with one recommended approach to handling credits in a statistical sample

The subsection (AM 1303.25) currently lists three types of credit methods that can be used for a statistical sample. The CDTFA now only recommends one credit method for use in a stratified statistical sample, which is referred to as “Method 1” in the current AM text. When auditors review electronic data, attempts should be made to match credit invoices to original invoices (including partially) if it is certain that the credit invoices are related to the original invoice. For all credit memos that are not matched to original invoices, those credits will be removed from the population prior to selecting samples. However, those credits, and any other credits not originally provided, are available to be used to offset any invoices that are selected during the sampling process. This is to ensure that the final net amount related to the transaction in question is used to project any errors related to that transaction. Credits may not be used to offset invoices not included in the sample base.

Removing the mean-per-unit projection technique

Under the current rule, the mean-per-unit estimation technique allows estimates to be made from the audited values in the sample. (AM 1303.06) Specifically, the mean-per-unit projection technique estimates the total value of the population by taking the average audited value of the sample items multiplied by the number of units in the population.

The last major rewrite of Chapter 13 occurred in 2001. The deadline for interested parties to provide written submissions regarding this discussion paper is February 23, 2022.

IMPLICATIONS

The implications of the proposed changes on taxpayers should not be discounted as some of the new proposals reverse formal decisions of the elected California State Board of Equalization, which were based upon statistical experts. For example, eliminating the three error rule can have enormous consequences by allowing an auditor to make projections based on less accurate data. When the California State Board of Equalization adopted the chapter including this requirement in the early 2000s, it relied upon outside expert statisticians following considerable debate from interested parties.

Additionally, the CDTFA’s proposed policy of requiring that a minimum sample size of 300 items of interest per test stratum be used in all statistical samples may be too high, particularly when compared to national voting poll projections based on sample sizes of 1000.

Finally, the discussion paper released by CDTFA states that “the proposed amendments include the addition and removal of some key policies.” Taxpayers are forewarned that these alterations of well-established policies may have adverse consequences that are not readily apparent