I have had several occasions at Internal Audit and Compliance conferences where the Benford analysis has been often cited as a typical method of data analysis in fraud detection. During these lectures, I often hear then the other listeners remarking that the analysis is a nice Topic for a lecture, but not really helpful in their practical day-to-day business of an invetogator or auditor. That is like saying: "Sounds good what he's talking about, but what is good for in real life? Nothing!"

This statement is nonsense. Believe me: Just try the Benford analysis in your next assignment, read through this article and you see that it is an easy to handle tool for you that can deliver remarkable results in giving you a deeper insight into the data that you are reviewing.

So what it is and how simple to apply it in IDEA, I'll show in the following:

Nevertheless, I have to go into the theory behind it shortly. Do not worry: Benford's Law is not

rocket science! So what is the motive of the Benford analysis?

Benford's Law states that the value or better: amount fields in any given and natural data set always correspond to a specific pattern of numbers and numerical sequences. Meaning that the 1 as the first number´s digit naturally occurs more often than the 2 and the 2 more often than the 3 and so on. That's why I call it the theory of significant numbers. Sown graphically here:

There are, however, a few conditions that the record and, more specifically, the field / column to which the Benford analysis applies, must meet:

## Benford Analysis with IDEA

So, with this background knowledge, we start to analyze and as said we do that with the analytics software IDEA. I know, it is possible with MS Excel, too. But only with the help of macros that are downloadable on the web. But there are

definitely safer and much simpler ways to do it. And don´t forget the shortage in time we face during an assignment.

Anyways, such fraud-specific analyzes are the strength of IDEA or ACL. Analytics tools as those should therefore be part of the basic tool box of each fraud investigator and auditors team.

The actual execution of the analisis is a breeze with the software: In the open file we select tab "Analysis" -> "Benford's Law" and in the window that opens we select the previously mentioned amount field to be analysed, in this case "Payment_Amount". Then we also see how the

program runs and performs the analysis:

## The Analysis of the Analysis

As soon as the result appears, you could say however: "And as you can see, you don´t see anything."

If we go to the evaluation of the first and second digits, however, we can clearly see abnormalities in the dataset:

Shown by the red and yellow hatched bars, we see that the first two numbers "10 .... ."," 50 ... .." and "75 ... .. " occur so often that IDEA classifies this as" highly suspect ". This means there are more records with these first two digits in the Payment_Amount field than the upper

limit of Benford's law allows.

The yellow-marked numerical sequences are suspiciously often, i.e. still within the limits, but represented far above the expected number of Benford Laws. These, too, indicate fraud and thus further investigations.

This is done via a simple mouse click on the corresponding bar. You just choose if you want to view the records or extract them for deeper analysis.

This is useful in order to get to the root cause of their high frequency. For example, the deviations can be from repeating banker´s order bookings that appear every month in a dataset that covers e.g. several months. However, it can also be fictitious amounts entered with malicious intentions, so fraud.

In any case, the Benford analysis provides a good starting point to get an overview and deeper understanding of the data in front of you. In any case, I can recommend Benford in investigations and audits in the procurement area of an organisation.

Kommentar schreiben

0