Implementing Benford’s law in continuous monitoring applications

Research Question: Do modern ERP systems record sufficient information to allow retrospective monitoring of accounts payable transactions? Can Benford’s Law be applied to these transactions to detect potential fraud in accounting data? Motivation: Modern ERP systems are capable of recording several thousands of transactions daily. This makes it difficult to find a few instances of anomalous activities among legitimate transactions. As organizations continue to become more complex and demand more integrated business processes, automated analytics may provide auditors and fraud examiners with some degree of assurance on continuous information simultaneously with, or shortly after the disclosure of information. Idea: In this study, we develop a proof of concept prototype to monitor invoice transactions and identify those that violate Benford’s law. The prototype exploits audit trails in enterprise systems. Data: Data was obtained from the SAP ERP systems of two large organizations. Organization 1, a government department, provided a one month sample of accounting transaction data. Organization 2, a global manufacturing company, provided a six-month sample of their transaction data. Tools: Verification and validation were achieved by obtaining independent reviews from an expert and a panel of auditing practitioners. Their feedback was sought using a survey instrument where they rated various aspects of the prototype software. Findings: A key aim was to demonstrate the feasibility of implementing Benford’s analysis in continuous monitoring applications by exploiting audit trails in ERP systems. The concept was demonstrated by designing prototype software. We found that Benford’s analysis is a useful tool for identifying suspicious transactions. These transactions may contain possible errors, potential fraud or other irregularities. Contribution: An important contribution of the study is that the entire population of transactions for a specified time period are analyzed. This approach is in contrast with the traditional or manual audit approach which is limited because it reviews only a small percentage of a large population of transactions. The prototype demonstrates the application of technology and data analytics to process transaction data from a SAP ERP system in a near real-time basis. This represents the next step in the evolution of the financial audit from manual to automated methods