Question

16-26 (Objectives 16-2, 16-3) With advances in technology, an increasing number of data analysis techniques are being used on audit engagements to increase both efficiency and effectiveness. In an article titled “Introduction to Data Analysis for Auditors and Accountants,” published in the CPA Journal (February 16, 2017), authors Kogan, Vasarhelyi, and Appelbaum describe how technology-driven data analysis can be used in all stages of the audit. Read the article on data analysis (available at www.cpajournal.com) to answer the following questions: The image shows logo of research icon.

Required

A) The authors discuss the emphasis on “audit by exception” when using data analysis. What does “audit by exception” mean and how does it differ from a traditional audit approach using statistical sampling? How does an auditor identify the exceptions?

B) The authors note that data visualization can be used to conduct exploratory data analysis. Describe how data visualization could be used in the audit of the sales and collection cycle.

C) Choose one of the methods identified by the authors as emerging approaches, for example blockchain or deep learning, and discuss how an auditor might test one of the accounts receivable balance-related audit objectives using this method.

Answer 1

a.

In an "audit by exception" approach, the auditor focuses on outliers, or exceptions, within the data rather than focusing on testing a sample of items from a population. The authors of the article detail the following steps in applying audit data analytics and the "audit by exception" approach.

1. Flowchart the process: The auditor needs to understand the flow of the cycle or application being tested in order to understand the risks and extract the appropriate data.

2. Choose and extract the data: This step involves selecting the data items to extract from the dataset.

3. Understand the population: This step involves understanding the nature, distribution and limitations of the data.

4. Understand the fields with descriptive statistics: The auditor needs to understand information about the data such as the median, minimum, and maximum values as well as the occurrence of missing values.

5. Exploratory data analysis: Visualization tools can be used to understand the data and identify areas of risk.

6. Choice of analytic methods and alternative approaches: Explore methods, such as regression analysis or factor analysis, to analyze the data and select the most appropriate method based on the process or data being tested.

7. Confirmatory data analysis and finding outliers: Once the auditor has identified the high risk areas of the audit, they can use one of the analytic methods considered above. The analytic model is used to determine an expectation or a benchmark about a particular balance or account (e.g., revenue by region, or individual customer balances), and then compare the actual amount to the expected amount. This will identify differences for further investigation.

8. Evaluating results and integrating with traditional findings: The outliers or exceptions identified from the steps above should be considered separately from the remaining population.

This approach of "audit by exception" focuses the auditor's attention on the exceptions identified as opposed to a more traditional approach of selecting a sample of items from the population to test.

b.

Auditors can use data visualization in many ways in the audit of the sales and collection cycle. For example, auditors can graph sales at a disaggregate level (e.g., by product line or region) across months or years to identify unusual trends. Auditors could also graph account balances or turnover ratios of individual customers, including related parties, to identify a build-up of balances. Data visualization could also be used to test sales returns activity throughout the year and for a time period following year end.

c.

The authors discuss possible uses of predictive analytics, deep learning, blockchain, and text mining. For example, blockchain technology could ultimately be used to flag transactions in real time, such as those exceeding a certain threshold, for auditors to test. Text mining can be used to read through revenue contracts with customers to identify particular terms, and therefore identify contracts with a right of return or extended payment terms. Predictive analytics can be used to analyze data throughout the year and continually update predictions about revenue from individual customers or revenue in aggregate. Finally, deep learning could be used to develop models for predicting sales returns or bad debt expense based on historical information.