The lead manager stole small amounts at first. Then, after testing and learning the thresholds that would raise red flags, the manager became progressively bolder and siphoned off larger amounts in the names of fake businesses “owned” by friends and relatives.
When someone noticed that the average amount of transactions was trending upward, it was attributed to market conditions during the real estate boom and was not researched further. Over the course of 15 years, the manager’s activities went undetected, and they were able to steal $50 million.
Here’s the kicker – this manager worked for a tax agency, the very type of organization you would think would be most vigilant about fraud prevention.
Why tax administrators are adopting new data and analytics strategies
Tax agencies can't rely solely on traditional methods and data sources to detect and prevent fraud. Learn why being able to quickly and accurately access and analyze all types of data is key to spotting tax fraud, ensuring compliance and protecting revenues.
Internal fraud detection
Most tax agencies focus on ensuring that their external fraud prevention efforts are robust. But they may be overlooking other types of fraud going on right under their noses – theft from the inside.
In this property tax refund case, appropriate oversight and custody controls were so poor that the external auditors couldn’t see the pattern. In 15 years, the agency never discovered it. The fraud was eventually exposed by a bank employee who sensed something fishy about a refund check presented for deposit.
Unfortunately, this employee’s supervisors were all fired because this extensive theft happened on their watch. They hadn’t established the proper fraud prevention controls and reporting to thwart the theft, or even to recognize it was happening. They didn’t know what they didn’t know – and their jobs were collateral damage. These were executives who had been with the agency for 20 years.
After that incident, the agency put more controls on its tax system to prevent this type of fraud. However, several years later, a different employee in the tax enforcement division did a similar thing with individual income tax refunds to the tune of $500,000. In this case, automated reporting provided clues to impropriety, and a sharp-eyed CPA delved into it.
Why did these fraud campaigns go on for so long before anyone noticed? And even more importantly, what must be done to stop it from happening again?
These cases point to the three biggest blind spots tax agencies face in fraud prevention and detection: firefighting, weak reporting and thin analytics.
Firefighting refers to being so focused on the most obvious emergency that you fail to see the bigger picture. For example, don’t be so preoccupied with identity theft and taxpayer fraud that you don’t realize what employees are doing. Is anyone committing tax refund theft or helping friends and family trim their tax liabilities?
You can’t get away with not knowing what you don’t know. Have reliable analytical and statistical reporting. Give investigators and other staff the ability to generate their own reports, act quickly on alerts, investigate deeply and explore the patterns they see.
The second fraudster described here was stopped much sooner – tens of millions of dollars sooner – thanks to additional reports and anomaly detection. A solid fraud analytics solution would have found it almost immediately.
Integrated tax systems are great for automating tax return and refund processes and managing audit and collections cases – but even those with a “fraud module” have limited analytical capabilities. These systems usually take a narrow view, running a tax return against basic business rules that define what is customary or unusual. For example, rules might decree that total itemized deductions should not exceed a certain percentage of adjusted gross income.
Business rules can catch unsophisticated fraud, but they miss many issues that advanced mathematical techniques can detect. Fraudsters can easily circumvent basic detection techniques, and they’re sometimes assisted by employees with proprietary knowledge of the organization.
Case in point
At one agency, network analysis revealed that 1,900 refund returns with no listed tax preparer had all been submitted from a single IP address in rapid succession. Anomaly detection techniques then found very large charitable contributions and job-related expenses on these returns relative to the taxpayers’ peers. Without advanced analytics, the agency’s audit selection processes might have found a handful of these returns three to five years after the refunds had already been paid.
Advanced analytics for fraud detection
Using advanced analytics such as predictive modeling, anomaly detection and network analysis, you can find fraud indicators faster and put safeguards in place that deter fraudsters. For instance:
- Predictive modeling can identify suspicious returns based on the known filing patterns of proven fraudsters. For example, a fraudster might file refund returns for progressively higher amounts year over year, testing to find the threshold that triggers scrutiny. Predictive analytics can spot this stair-stepping pattern and score the likelihood of fraud.
- Anomaly detection will show you answers to questions such as, “Are this person’s itemized deductions out of line with others in the income peer group or geographic neighborhood?” Or “Does the monthly variation in gross sales make sense for this type of business?” Unlike business rules, which require you to know what you’re looking for, anomaly detection automatically establishes what normal looks like for a peer group – and then automatically flags anything that looks abnormal.
- Network analysis reveals relationships among taxpayers, tax returns, tax preparers, business activity, and corporate officers or other responsible parties. People may be linked by a common address, employer, family member, bank account or other factors. In the case of insider fraud, network analytics can also uncover relationships with an employee conducting theft or helping others to facilitate fraud.
Don’t be the one who didn’t know what you didn’t know
Top-notch reporting with fraud detection analytics isn’t just “nice to have” anymore. It’s a necessity. Millions of dollars, and even your job, might depend on it. Find out what’s really under the hood of your tax systems and supplement with additional analytics as needed.
Advanced analytics solutions for tax fraud detection are inexpensive relative to your core tax system, and they can be implemented quickly with both legacy software and modern systems. This imperative analytics layer coordinates with your existing systems to create an ecosystem that deters theft and protects revenue.
About the author
Teya Dyan has more than 16 years of experience in tax policy regulation and fraud investigation, with expertise in managing compliance teams and as a front-line investigator. In addition to her work in Washington State and Kansas Revenue agencies, she served as President of the Pacific Northwest License, Tax, and Fraud Association for several years. Her current role at SAS combines her professional passions of data analytics, investigations, tax policy, and perpetually curious gumption to foster innovation through technology.
- Article Strengthen your payment fraud defenses with stronger authenticationThe rapid growth of digital wallets and payment applications ushered in many new payment fraud threats. Today, it’s more critical than ever to authenticate users. Learn four innovative to ways strengthen your authentication defenses while reducing false positives and protecting customers’ assets.
- Article How to prevent procurement fraudPerpetrated in several ways, procurement fraud is difficult to detect. Arm yourself with hybrid analytics that offer various approaches for cross-pollination of data and analysis.
- Article Bad debt and fraud... It’s all the same, right?In the case of application fraud in banking, or subscription fraud in telecommunications, a large proportion of fraud is typically misclassified as bad debt.
- 20/20 vision of riskBanks need a holistic view of their data in order to accurately detect risk. They also must overcome the challenges associated with poor data quality and the "noise" generated by existing control systems. Laura Hutton, Director of Banking and Solutions at SAS, explains how technologies can improve a bank's visibility into operational risk.