Follow the Data Trail: Using Analytics to Defeat Fraud

Have you recently heard talk of the “Stopping Improper Payments to Deceased People Act”? Probably not, because it’s dead. This bill, which would have expanded the use of data on deaths by federal and state agencies to limit improper federal payments, was introduced to the U.S. Senate in 2015, but was never enacted because many of its provisions are meant to already be included in existing laws, such as the “Improper Payments Elimination and Recovery Improvement Act of 2012.”

However, research conducted by the same Senate committee that reviewed and killed the act found that from 2000 to 2010, more than $1 billion was paid out to more than 250,000 deceased individuals, despite the laws that were already in place.

News of fraud committed against (or in some cases by individuals within) government entities on the national, state and local levels is neither surprising nor novel. Unfortunately, the problem is exacerbated because, in many circumstances, the resources and expertise required to mitigate fraud risk and minimize waste and abuse continue to be insufficient.

Fraud risk comes in many shapes and sizes, often going undetected for far too long and costing taxpayers a great deal of money. Improper payments (which include payments to deceased individuals) through entitlement programs (e.g., health care, unemployment, pension, etc.) are of particular concern. For example, The Federal-State Unemployment Program alone is estimated to have an improper payment rate of close to 12%, which currently represents over $6.5 billion of taxpayer dollars.

Additionally, government-wide improper payments were estimated at $128 billion in 2014—the equivalent of a more than $1,100 reduction for every taxpayer in the United States. Nearly half of those improper payments were estimated to have gone through Medicare.

In the modern age of advanced technology, tracking and data analytics, these numbers remain surprisingly high—too high to turn a blind eye and ignore.

Pay-and-Chase Is a Losing Game

After improper entitlement checks are sent out, the process for retrieving those funds—even if fraud is proven—is time-consuming, costly and often ineffective. Therefore, it stands to reason that the problem of improper payments must be solved before the checks are issued. To accomplish this, state and federal agencies would benefit from establishing and executing several layers of data-enabled anti-fraud measures.

Of course, as anti-fraud measures improve, so do the tools and techniques used by fraudsters. With a continuously evolving risk environment, one difficult to fight, what can be done?

Compare Inter-Agency Data

One obstacle raised in the defunct “Deceased People Act” was that the government agencies responsible for entitlement checks did not have proper or sufficient access to death records. Continuing the example of unemployment insurance (UI) from above, if agencies worked together to compare data and ensure accuracy of information, data analysis software would be able to find indicators that certain recipients are ineligible for such payments. For example, analysts could:

• Check for duplicate claims by identifying multiple addresses for the same SSN in a given timeframe.
• Compare claimant details against other databases, such as the National Directory of New Hires or State Directory of New Hires, to identify unreported earnings or existing employment.
• Match UI claimant data with data from registries of deceased individuals in State Vital Records Department databases.
• Analyze prison records to cross-match UI claimants.
• Verify claims and claimant data with Department of Motor Vehicle records.
• Identify claims using foreign IP addresses or “hot” IP addresses known for fraudulent claims.

Using these data tests has the potential to uncover millions of dollars in improper or fraudulent

payments. For example, one U.S. state that implemented a program that matched claims information to a database of state employees found that 160 current employees were receiving improper payments totaling more than $250,000. Using data analytics, another state was able to reduce the amount of improper payments in its system by $25 million.

What Else Can Be Found?

Beyond unemployment insurance, the public sector can produce similar results in a variety of areas by continuously tracking, monitoring and analyzing data. For example:

• False Vendors (Procurement Fraud): Payments could be going to fictitious companies for non-existent goods or services. Forty percent of global government organizations reported experiencing procurement fraud in 2016.
• Fake Employees (Payroll Fraud): Occasionally, employees can remain on the payroll after termination, but more frequent fraudulent activity comes through fake employees collecting payroll without ever working for the organization. A recent ACFE report estimated that employee fraud lasts an average of 18 months!
• Deceased Employees (Benefits Fraud): Whether it’s Social Security, Medicaid, housing, disability or any other benefit, paying the deceased puts an entity or program’s integrity at risk, eroding public trust. Identity theft is often the common fraud scheme committed using deceased individuals’ information.
• Non-Existent Treatment (Healthcare Fraud): Medicare and Medicaid are affected in a variety of ways—from false billing, up-coding and fake pharmacies to claims of excessive or unnecessary treatment.

Fortunately, fraudsters in each of the categories above leave their own distinctive data trail. Although scams are varied, by identifying the profile of a specific fraudster you can proactively monitor distinct fraudulent groups.

The red flags will vary by fraudster, but beginning to understand the categories of fraud and the red flags that data analytics can capture is the first step toward mitigating fraud risk in the public sector.