The ﬁrst analysis of any data set need not use sophisticated mathematics or statistics. The goal of these test is to ﬁnd subsets that are highly inflated due to the error or fraud.
Largest Subset Test
The largest subsets test uses two ﬁelds, one with transaction or balance numbers (such as amount, inventory counts, vote counts, population counts) and another ﬁeld to indicate the subset (e.g., vendor number, credit card number, or branch number). Subset is a group of records that have something in common.
The data can often be divided into several subset groupings. For example, accounts payable data could be grouped by vendor or by the type of purchase (purchase order, no. purchase order) or by time. There are often a few different ways that data can be divided into subsets. For inventory data the grouping could be by location. For airline ticket refunds or retail customer refunds the groupings could be the credit card that received the refund.
A challenging problem in analyzing text-based datasets is that a same record may be represented in multiple ways throughout the dataset. Variations can result from inconsistent naming conventions, spelling mistakes, etc. which makes it difficult to compare text data using ‘VLOOKUP’ & ‘MATCH’ functions. To match text datasets we can use something called Fuzzy Logic.
Luhn Algorithm also known as ‘modulus 10’ or ‘mod 10’ algorithm, was created by Hans Peter Luhn in 1954.
It is widely used in Credit/Debit card numbers, IMEI numbers, and Canadian Social Insurance numbers.
What is Luhn Algorithm?
To understand what Luhn algorithm is, we first need to understand what is ‘modulo’. Modulo or Modulus is the remainder after dividing the number with another number. Consider the example 7 divided by 3 has; quotient 2 and remainder 1.
Therefore, modulo 10 equal 0 means after dividing the number with 10, the remainder should be 0. In simple terms the number (dividend) should be a multiple of 10 (divisor).
In my previous article Auditing: Accounts Payable / Vendor Payments I spoke about Relative Size Factor (RSF) and how it can used to identify isolated outliers in vendor invoices. In this article I’ll try to show how RSF can be calculated in Excel.
The RSF test is an important tool for detecting errors. RSF test compares the top two amounts for each subset and calculates the RSF for each. The test identifies subsets where the largest amount is out of line with other amounts for that subset.
In general terms, fraud is an intentional deception, whether by omission or commission, to realize a gain. Under common law, fraud includes four essential elements:
- A material false statement
- Knowledge that the statement was false when it was spoken
- Reliance on the false statement by the victim
- Damages resulting from the victim’s reliance on the false statement
In the broadest sense, fraud can encompass any act for gain that uses deception as its principle technique. This deception is implemented through fraud schemes, specific methodologies used to commit and conceal the fraudulent act. The legal definition of fraud is the same, whether the incidence is criminal or civil. The difference is that criminal cases must meet a higher burden of proof.
According to the ACFE’s “Report to the Nation 2016” financial statement fraud occurred in less than 10% of the cases reported by the respondents. But it caused the highest median loss of $975,000. Asset misappropriation schemes was reported in more than 83% of the cases with median loss of $125,000. And Corruption cases fell in the middle with 35.4% of cases with median loss of $200,000. Several cases included schemes in more than one category.
In the previous post I mentioned the brief summary of Earnings Manipulation shenanigans.
Below is the summary of some methods used to manipulate cash flow & key metrics in financial statements as identified by Howard Schilit in his book Financial Shenanigans.