Auditing Section Research Summaries Space

A Database of Auditing Research - Building Bridges with Practice

This is a public Custom Hive  public

research summary

    Selecting Audit Samples Using Benford's Law.
    research summary posted September 14, 2015 by Jennifer M Mueller-Phillips, tagged 06.0 Risk and Risk Management, Including Fraud Risk, 06.02 Fraud Risk Models 
    186 Views
    Title:
    Selecting Audit Samples Using Benford's Law.
    Practical Implications:


    The authors contribute to digital analysis by formulating two alternative mathematical programming models that can assist auditors in selecting audit samples, using Benford’s law. The models consider multiple conformity tests and test statistics simultaneously, taking into account the interdependencies between the conformity tests, and allow the auditor either to identify a subset of nonconforming records in a dataset or to define a specific number of records to audit. This approach is new in the literature.

    Citation:

    da Silva, C. G., and P. R. Carreira. 2013. Selecting Audit Samples Using Benford's Law. Auditing: A Journal of Practice & Theory 32 (2): 53-65.

    Keywords:
    auditing, Benford’s law, data anomalies, digital analysis, fraud detection, mathematical programming
    Purpose of the Study:

    Auditing standards require the use of analytical procedures during the planning stage of an audit. One of these procedures is the study of digits or digital analysis, which may allow an auditor to identify irregularities in accounting datasets and to detect fraud symptoms more easily. In particular, this procedure can be used to highlight suspicious transactions, accounts, events, or trends to audit. Digital analysis is not particularly new, however. Many digital analysis techniques have been in use for some time. The authors contribute to the application of Benford’s law in auditing by proposing two alternative mathematical programming models that can assist auditors in selecting an audit sample from a dataset of numerical records.

    Design/Method/ Approach:

    The authors introduce two mathematical programming models that can assist auditors in the identification of an audit sample in a dataset of numerical records. In the first model, the objective is to identify the smallest subset of nonconforming records. In the second model, the objective is to identify the k most nonconforming records. They evaluate the effectiveness of the models by simulating 30 Benford-compatible datasets, each consisting of 5000 records with four digits each.

    Findings:

    The first model identifies the smallest subset of nonconforming records given some predefined conformity criteria, while the second model reveals the subset of the k most nonconforming records, where k is the size of the audit sample and is defined by the auditor given his/her restrictions. The models take into account several conformity tests and test statistics simultaneously so as to generate a more promising audit sample. When considering several conformity tests and test statistics, the selection of the records to audit becomes a complex combinatorial task so that the models, and their respective resolutions, are likely to avoid the misleading decisions that may result from simple approaches.

    Category:
    Risk & Risk Management - Including Fraud Risk
    Sub-category:
    Fraud Risk Models