This is the second of two discussions of why audits of consumer use tax are likely to be less accurate than those of sales tax, even when performed by the same auditor on the same taxpayer. In the first article, I discussed some of the problems an auditor faces when trying to determine 1) if there is a consumer use tax liability and 2) if the potential tax change justifies the effort.
For the current discussion, I’ll assume the auditor has determined there is sufficient liability in the purchases of expense items to justify a full examination. I’ll also assume the auditor has decided to base the audit on purchases rather than on the vendors.
Selecting the Methodology
With those decisions made, the next big question is which audit procedure(s) should be used - actual basis, block sample or random sample?
- An actual basis examination is generally not feasible because of the volume or because all bills are not readily available.
- A block sample limits the population to be examined but can be skewed if purchasing isn’t relatively uniform in dollars, vendors and types of items purchased for all periods.
- A random sample, when done correctly, can eliminate most bias and something less than the entire audit period can be tested.
The problem many auditors have is determining what the sample unit should be. There isn’t a nice clean sequence of consecutive invoice numbers like there is when sampling sales invoices for random samples.
Pre-Computer Audit Sampling
Before computers, performing a random sample audit on purchases of expense items frequently required some creativity on the part of the auditor. The biggest hurdle was isolating the population of interest―purchases of tangible personal property charged to taxable general ledger accounts―from all other purchases.
Many auditors simply selected a sample using the checks written as the control numbers since these are usually sequential and accounted for. But because a check could pay multiple invoices and only a small percentage of all bills are for items of interest, an auditor would need to dig through stacks of paper to find a single line on an invoice.
A slightly cleaner way was to select the document at every “nth” inch in the paid bills drawers or boxes. It would be impossible to determine the precise document and is random, with every item having an equal chance of being selected. Since both methods audit the entire population of paid bills, any percentage of error must be applied against all purchases, regardless of the General Ledger account to which it was charged. Try explaining why you are assessing use tax against the Employee Life Insurance account.
A little creativity could avoid this problem. One possibility is to use page/line entries in the Purchase Journal as statistical sample units. The page/line combination eliminates the need to establish an accurate count of the lines themselves.
Computer-Age Advantage For Auditors
With the advent of computers, auditors have a better opportunity to narrow the population of transactions to be audited. Depending upon the fields populated in the various data tables, the population can (at worst) be limited to selected GL accounts and specific departments or (at best), to only non-taxed purchases of specific goods and services.
In the first case (worst), the auditor can reduce the initial population for a random sample by a series of mini examinations:
- Link any use tax accruals to the population and remove those invoices with tax accrued;
- Sort records by vendor name and remove those selling nontaxable goods/services;
- Depending upon the state involved (e.g. California) and taxpayer’s policy for issuing exemption certificates, eliminate all transactions subject to “sales tax” rather than “consumer use.”
Filtering the records in this manner yields a much more homogeneous population of purchases―a population that more closely matches the criteria defining potential consumer use tax exposure. The percentage of error developed in this manner is more accurate and it is applied to a smaller, more specific base.
In the second case (best), an actual basis audit is often possible because adequate filters can be applied on the computerized records …and an actual basis audit is always preferable to a random sample, all other things being equal.
Please feel free to post comments or questions below.
Other recent “Audits and Sales Tax” posts by Lloyd Geggatt:
- Software Audits for Buyers & Sellers – and the 3 Key Questions
- Luxury Audits: Empty Boxes Full of Champagne Wishes & Caviar Dreams
- Sales Tax Audits: Actual Basis Approach Can Result in "Win-Win".
- What Auditors Should Understand About Outsourced Sales Tax Returns
- Auditing Sales Tax Credits: An Auditor's Top 2 Considerations