McGovern & Greene LLP's Electronic Fraud Fighting (“ELFF”) tool represents a comprehensive suite of data mining services designed to detect a multitude of procurement and disbursement frauds that may exist in your organization.
What separates ELFF from the competition is its' designers: Christy Warner and Craig L. Greene, CFE, CPA. Christy is a Mathematician, Statistician and Data Base Design Expert who has worked in data mining for over 14 years, unearthing major fraud cases. Craig is a Certified Fraud Examiner and Certified Public Accountant who has worked as an auditor and forensic accountant for over 30 years investigating complex fraud schemes and recovering millions of dollars for his clients. Together they have created the most extensive electronic data mining system for finding fraud.
ELFF's Data Mining Capabilities
Duplicate Payment Detection
ELFF's duplicate payment detection is a very comprehensive set of algorithms. While most duplicate payment identification plans employ 3 or 4 patterns, ELFF uses 10 intelligent algorithms plus applies duplicate vendor logic to capture the absolute maximum number of duplicate invoices.
In addition, if you would like us to identify duplicate vendors in your vendor file, we can apply this knowledge to catch even more duplicates where the vendor numbers are completely different (but point to the same vendor). The following is an example of how ELFF can work for you:
“While conducting a duplicate payment audit for a medium-sized healthcare product manufacturer, we noticed multiple $40,000 payments made out to one person on the same day. This made us curious, so we ran ELFF's mathematical algorithm to identify above-average payments per vendor and the same payments floated to the top of the report. The person turned out to be an employee who typically received a bi-monthly paycheck of between $500 and $1,000. Indeed, $40,000 seemed unusual - especially three payments made on or near the same day! It was also highly unusual that there were no invoice numbers on these high-dollar payments. Most accounts payable systems won't accept an invoice entry without an invoice number: in fact, many A/P clerks are coached to enter the date if no invoice number is provided. Well, we got lucky in this situation because the client had given us its check register (instead of an A/P extract), which listed all checks issued, including checks with missing invoice numbers. Using ELFF's cutting-edge data mining technology, we were able to import the electronic version of the check register and create a database from it. Then using this database, we were able to implement ELFF's fraud detection algorithms and uncover the fraud.”
How ELFF Uses Benford's Law to Identify Fraud
If we know the normal frequency of digits, then we can identify digit frequencies that violate that normal behavior. For example, Benford concluded that, out of a group of numbers, the first digit will be “1” about 30% of the time. By the same law, we would also expect the first digit to be “8” about 5.1% of the time. If we review Accounts Payable invoices and determine the first digit of the invoices is “8” 50% of the time, then we may have either many legitimate payments that start with “8”; or we may have fictitious invoice amounts. Fraudsters often create an amount that starts with a higher number, like 8 or 9, not knowing that ELFF is equipped to identify these abnormal payments.
ELFF's algorithms identify vendors with payments that consistently violate Benford's Law of Numbers. If their digit frequency distribution varies widely from the below graph, the vendor is flagged for further investigation.
Missed Discount Detection and Recovery
Negotiated payment terms that should result in a discount are often missed by Accounts Payable systems: sometimes voluntarily for accounting purposes, but sometimes by mistake. ELFF searches your A/P files for missed discounts by comparing the payment terms with the invoice date, check date, invoice amount, and discount-taken fields. In addition to missed discount identification, we can recover these overpayments.
Erroneous Overpayment Detection and Recovery
Overpayments often occur when merchandise was never received, but paid for. ELFF can reconcile your Purchase Order file with your Accounts Payable and Accounts Receivable files to detect pricing errors and any overpayments that occurred. If recovery is requested, our collection department is very experienced in recovering lost monies.
Rounded-Amount Invoices
People who commit fraud often create invoices with rounded amounts, which are invoices without pennies or amounts such as $500.00, $5,000.00, $50,000.00. ELFF identifies these invoices and ranks them by the vendors who have the highest percentage of rounded amount invoices. For example, a vendor having 100% of their invoices without pennies would appear first on the list.
Vendors Consistently Paid Quickly
Vendors who are consistently paid quickly may be suspect of a fictitious company or a fraudulent corruption scheme. ELFF calculates the difference between the invoice date and check date, and ranks vendors who are consistently paid faster. Vendors with the greatest percentage of quickly-paid invoices are listed first on the alert list.
Sequentially-Numbered Invoices
Sometimes you run across vendors who have sequentially-numbered invoices such as “0001”, “0002”, “0003”, which may be legitimate. However, if this pattern is stretched over time, it may suggest that this vendor only does business with you, which is extremely rare in the business world. Although the vendors appearing on this alert list may be legitimate, some may be indicative of a fictitious vendor or perhaps the vendor is a victim of economic extortion by your employee.
Invoices Just Below Approval Amounts
Inside fraudsters often know approval limits and may sometimes submit a fraudulent invoice that falls just below the amount. ELFF's proprietary algorithms identify invoices that fall up to 3% below the approval amount. Then, vendors are ranked according to their percentage of invoices falling just below the approval amount.
Vendors with a Rapid Invoice Volume Increase
ELFF identifies vendors who have a rapid increase in invoice volume. The increase may be legitimate, but also may warrant further investigation. Suppose a vendor has two invoices one month and 70 the next – you may want to know why - even if the reason is not a fraudulent one. ELFF's algorithm detects only consecutive-month changes in the number of invoices. If the percentage increase is 250% or greater, then the vendor is flagged and put on the alert list. This would include a jump from 5 invoices to 13, but will also catch an invoice jump from 50 to 126.
Vendors with a High Variance in the Number of Invoices per Month
ELFF identifies potential fraud that follows a “hit and run” tactic. Suppose a fictitious vendor is created and one test invoice is submitted in May. In June, 50 invoices to this fictitious vendor are created. In July, the fraudster assumes a low profile again and only submits 2 invoices. This zig-zag in invoice volume is caught by ELFF.
Payments Dated on a Weekend
ELFF identifies any check, electronic funds transfer, or other disbursement dated on a Saturday or Sunday, which would be considered a rarity in the business world.
Vendors with an Unusual Percentage of Voided Checks
Voided and returned checks do occur in the course of a normal Accounts Payable month. What is more uncommon is a vendor with many voided checks or a regular pattern of voided checks. Voided checks are usually legitimate transactions; however, a voided check can be returned to the wrong hands and re-written to the fraudster. ELFF identifies vendors with voided checks and ranks them by the percentage of invoices that are voided status, so that a vendor with 80% voided checks will show up at the top of the list.
Above Average Payments per Vendor
ELFF identifies invoices that are way above average for a particular vendor. Suppose a vendor normally has invoices ranging from $1,000 to $3,000; suddenly an invoice shows up for $25,000. You may want to investigate this abnormality and can do so using this alert pattern.
Duplicate Vendor Detection
ELFF searches for duplicate vendors in your vendor file, utilizing 5 different criteria:
1) by address
2) by tax ID (EIN)
3) by bank routing number, if available
4) by name
5) by phone number
In addition to matching by the exact field, this algorithm uses intelligent fuzzy-matching logic to identify non-exact matches. It will identify an accurate duplicate match on addresses that are similar (but not exact), tax ID's that are similar (but not exact), and bank routing numbers that are similar (but not exact).
Vendor / Employee Cross Check
ELFF compares a vendor with an employee four different ways, via:
- Address
- Tax ID Number
- Bank Routing Number
- Phone Number
Using this approach, ELFF was able to detect a real employee (“Kathy”) whose SSN was the same as a company EIN (tax ID number). The company name, which we will call “ABC Inc”, happened to be on the same street, city, and state as a person with the same last name as the employee (presumably her spouse). Without this pattern, the employee fraud may have gone undetected.
Employee Earnings Analysis
ELFF analyzes the withholdings and payroll data for each of your employees to flag the outliers. This analysis helps to identify possible ghost employees and corrupt employees who have significant tax withholdings because of their other income.
Vendors with P.O. Boxes
Many vendors have P.O. Boxes as their addresses, making it difficult to sift through potential fictitious vendors and legitimate ones. However, ELFF's report can be used in conjunction with other reports. For example, if a vendor shows up on the PO Box report and also the rounded-amount invoice alert report, this vendor may warrant further examination.
Vendors with a Mail Drop as an Address
This algorithm compares vendor addresses with mail-box drop address such as “The UPS Store”. Some fraudsters will use mail drops as their address instead of a P.O. Box, to hide their fraudulent activity. ELFF's listing provides a unique approach to reviewing vendors who also may show up on another alert list.
Now What?
Prioritizing Suspect Transactions
ELFF summarizes suspected transactions and vendors using a ranking system based on the tests applied. This ranking helps to focus your audits and investigations on the truly deviant transactions and/or vendors.
Investigation and Audit
After ELFF completes its analysis, you can begin your investigation. If your company does not have the investigative/audit staff or need additional staffing support, McGovern & Greene can assist you as well. Our trained fraud examiners, forensic accountants and financial investigators routinely investigate a multitude of employee and vendor frauds identified by ELFF. One of the most effective methods for discovering corruption and vendor fraud schemes is the vendor audit. The professionals of McGovern & Greene have used this methodology for over 12 years and have developed it into a “science”.
In addition to the tests listed above, ELFF is constantly evolving with new tests and can also be modified to meet your specific needs.
For more information on Data Mining for Fraud Detection, please contact
Craig Greene.
