Ryan Tax Review

Reconstructing the ARQ’s Audit Methodology

The case between 2741-2568 Québec Inc. and Her Majesty the Queen1 was a home run for the appellant, but leaves many pondering the audit tactics and methods employed by the Agence du revenue du Québec (ARQ).


The appellant, 2741-2568 Québec Inc., operated a restaurant in St-Simon-de-Bagot, Quebec, under the name “RESTOPUB LA BAVIÈRE”.  The ARQ had reassessed the appellant $16,882.82 in GST, $134.23 in late remittance penalties, and $5,467.91 in arrears interest for the period from March 1, 2006 to February 28, 2009.  It is of importance to note penalties for gross negligence were cancelled and the assessed amounts were reduced by 40% at the objection stage.

The Court’s decision in this case wastes little time before diving into the mind-boggling calculations and extrapolations made by the ARQ in arriving at its assessment.  In short, the ARQ sampled 3,228 meal bills on 50 random days chosen within a period of 166 days, which totaled $32,655.79 in sales.  Ten items on the meal bills were selected (referred to as the targeted foods) and 1,676 of the targeted foods were included in the 3,228 meal bills.  A ratio of $19.48 [$32,655.79/1,676] was generated and multiplied by the total quantities of the targeted foods purchased by the appellant in an attempt to reconstruct the total sales of the company.  It should come as no surprise that there was a significant difference between the reconstructed and reported sales, with the reconstructed sales reaching $1,059,682.77, in comparison to reported sales of $768,239.80 – a difference of $291,442.97.


Before analyzing the ARQ’s calculation, the first question is:  Was the reconstructing of the sales even necessary?  Before resorting to an indirect audit method, such as the sampling and extrapolation performed in this case, there must be reasonable grounds to conclude that the company’s books, records and supporting documentation contained significant errors and were unreliable.  In this case, not only did the ARQ auditor state that the appellant’s accounting documents were in good order, but the objections officer also indicated that its records were sufficient, offering the following comments: “Considering the reliable records confirmed through the audit,…” and “…the books were in due and proper form, we were thinking of deassessing…”  From these statements, it is clear that the alternative audit method should not have been used, although the objections officer somewhat disagreed with this position, due to the discrepancies which appeared to be present after the sales were reconstructed.

In addition, the Court noted that the differences between bank deposits and sales transactions discovered during the audit did not appear to justify a reassessment.  Where analytical procedures and estimation methods, such as the reconstruction of sales, yield discrepancies, auditors would normally perform substantive procedures involving the inspection of documents, such as bank deposits, to reconcile reported sales.  The auditor in this case appears to have reversed this audit methodology.  Since the substantive procedures resulted in reasonably accurate information, the audit assertions would have been met for the sales test, and the analytical procedures and estimation method would appear to be a step backward, as they rely more on guess-work that is not always accurate. 

However, when there is a risk of unreported sales from cash transactions, as is sometimes the case in the restaurant industry, agreeing bank deposits to sales transactions would not necessarily identify unreported cash sales, as the cash could be deposited into another bank account or be kept on hand.  This potential issue highlights the unenviable position auditors can be placed in when they have few options available to them and a difficult job to do, with one result being that auditors might feel compelled to undertake difficult audit procedures, such as reconstructing sales, in order to complete their work.

Even with an apparent answer to the question of whether reconstructing the sales was necessary in hand, the Court still decided to analyze the reliability of the alternative audit method used.  The Court found that there were blatant errors in the ARQ’s calculations, as demonstrated by the objections officer, who reduced the assessed amounts by 40%.  Mr. David Haziza, Professor with the Université de Montréal Mathematics and Statistics Department, prepared a statistical report and testified on behalf of the appellant.  Professor Haziza pointed out key issues with the ARQ’s calculations, including the fact that 70.63% of the meal bills sampled did not include any of the targeted items, while 91.85% of the meal bills selected contained zero or only one of the items.  In addition, Professor Haziza identified that a total of 19 Mondays and Tuesdays were included in the sample of 50, instead of a suggested 14.  In the Professor’s opinion, this lead to skewed results, especially considering that the restaurant is closed for dinner on Mondays and Tuesdays, when practically no alcohol is consumed, even though 3 of the 10 targeted items were alcoholic beverages.  Another issue noted with the sample was that 63% of the bills analyzed were for breakfast, and only 2 of the 10 targeted items were consumed at breakfast.           

Mr. Sylvain Lamy, a statistician at the ARQ, testified on behalf of the respondent and, in doing so, pointed out some of the flaws in the calculations of the reconstructed sales.  Nevertheless, he still believed that the calculations were accurate over the 166-day period from which the sample was taken.  However, Mr. Lamy also admitted that the calculations were arbitrary when applied to previous periods.  Both Professor Haziza and Mr. Lamy suggested that they would have used different methods to reconstruct the sales, and stated the estimation method used by the ARQ was not representative of the previous two and a half years.

Questions Remain

The Court pulled no punches in allowing the appeal, concluding that, in addition to the unjustified use of an indirect audit method, the estimation method used by the ARQ was highly questionable.  The key takeaway for taxpayers is that auditors should not be reconstructing sales when books and records are made available and appear to be in good order.  In addition, where assessments are based on estimates and extrapolations, taxpayers should never shy away from scrutinizing the auditor’s tactics and methods.2

Dustin Kapitan, CPA, CA
Senior Tax Advisor

1 2741-2568 Québec Inc. v. R., 2016 TCC 207.
2 This article previously appeared in the May 2017 issue of GST & Commodity Tax, Vol. XXXI, No. 4 (Thomson Reuters).  It has been reprinted with permission from the publisher.