Applying visual analytics to fraud detection using Benford's law

• Published date: 01 October 2020
• Author: Clarence Goh
• Date: 01 October 2020
• DOI: http://doi.org/10.1002/jcaf.22440

EDITORIAL REVIEW

Applying visual analytics to fraud detection using

Benford's law

Clarence Goh

School of Accountancy, Singapore

Management University, Singapore,

Singapore

Correspondence

Clarence Goh, Singapore Management

University, 60 Stamford Road, Singapore

178900, Singapore.

Email: clarencegoh@smu.edu.sg

Abstract

Benford's law has been examined as a useful tool for detecting potential

accounting fraud. In this article, I provide an introduction to Benford's law

and examine how the first digit, second digit, and first-two digits tests in

Benford's law can be employed to detect potential accounting fraud. In addi-

tion, I also highlight, through a worked example, how Tableau, a visual analyt-

ics tool, can be used to perform Benford's law's first-digit test to detect

potential fraud.

KEYWORDS

Benford's law, fraud, tableau, visual analytics

1|INTRODUCTION

In 1938, in examining over 20,000 observations from a

diverse range of datasets, including datasets on the areas

of rivers and the atomic weights of element, Frank

Benford, an American Physicist, observed a consistent

pattern where small digits occurred more frequently in

the first position of numbers than larger digits (Benford,

1938). This observation laid the foundation for the math-

ematical tenet that has become known as Benford's law,

which defines the expected frequency that digits appear

in data.

Benford's law has been examined extensively in a

wide range of areas including in mathematics (Hill, 1995;

Newcomb, 1881), the physical sciences (Sambridge,

Tkalcˇi

c, & Jackson, 2010), and business (Giles, 2007;

Judge & Schechter, 2009). Research has also examined

Benford's law in the accounting setting. For example, in

the area of tax accounting, Nigrini (1996) examined how

Benford's law can be used to investigate tax compliance

among tax payers. In the area of audit, Nigrini and

Mittermaier (1997) examined how Benford's law could be

used as an effective aid in analytical procedures in the

planning stage of an audit while Nigrini and Miller

(2009) examined how second-order tests of Benford's law

can be used to detect unusual issues related to data

integrity that might not have been easily detectable using

traditional audit analytical procedures.

Benford's law has also been examined as a useful tool

to detect potential accounting fraud. In a survey con-

ducted on 86 accountants to gain insights into the percep-

tions of fraud detection and prevention methods,

Bierstaker, Brody, and Pacini (2006) found that the

accountants rated “digital analysis,”which is based on

Benford's law, as the tenth (out of 34) most effective fraud

detection procedure. Consistent with this, various

accounting studies (e.g., Durtschi, Hillison, & Pacini,

2004; Kumar & Bhattacharya, 2007) highlight the appli-

cations of Benford's law as an easy to implement data

mining technique that can effectively determine the

authenticity or otherwise of a set of accounting data.

Beyond the research setting, Benford's law has also

been demonstrated to have been an effective tool for

fraud detection in practice. In June 2009, Bernie Madoff

was sentenced to 150 years in prison for operating the

largest Ponzi scheme in United States history (Nasaw,

2011). While the scale of the US$65 billion scam was eye-

catching, what was perhaps more surprising was that

Madoff managed to run the Ponzi scheme for decades

without getting caught. In the aftermath of the high-

profile case, forensic investigators have been left asking

themselves whether there were any clues that could have

Received: 21 November 2019 Accepted: 3 February 2020

DOI: 10.1002/jcaf.22440

202 © 2020 Wiley Periodicals, Inc. J Corp Acct Fin. 2020;31:202–208.wileyonlinelibrary.com/journal/jcaf