Turning the Tide: Combating Government Waste and Abuse Using Advanced Analytics

According to the US General Accounting Office (GAO), federal programs’ fraud, waste and abuse cost US taxpayers $233 billion to $521 billion annually from 2018 to 2022. While several federal agencies have made progress in reducing fraud levels over the last few years, many of the highest-risk programs identified by GAO have not.  

A closer look at the agencies showing the most significant improvements in fraud-fighting versus agencies that have not reveals that many of the “winners” have things in common. 

• They are committed to using data and advanced data analytic tools as an essential part of their strategies for reducing fraud and improper payments 

• They are using a continuous improvement approach and investing in tools with Artificial Intelligence capabilities that automate the process of continuous improvement 

• They are finding ways to break down data silos, bring new data into the analytics process, and improve their data fusion methods to improve results 

• They are employing a variety of analytical tools to identify new fraud strategies and tactics used by fraudsters and responding quickly to stop additional losses 

Based on these agencies’ positive results, the GAO has recently recommended that the federal government “establish a permanent analytics center of excellence to aid the oversight community in identifying improper payments and fraud.”  Similarly, the GAO recommended in April 2024 that the Treasury Department should “evaluate and identify methods to expand government-wide fraud estimation to support fraud risk management. This effort should (1) initially prioritize program areas at increased risk of fraud; (2) be responsive to changes in the availability or quality of data; and (3) leverage data analytics capabilities.”  

Let’s review some specific examples of fraud-fighting success and identify several of the most common pain points and obstacles that prevent government agencies from making considerable progress.

Delays and bottlenecks in detecting improper payments

Two of the federal programs identified by the GAO as having high volumes of fraudulent and improper payments are the Medicare and Medicaid programs under Health and Human Services, partly due to the high volume of payments they make – both in transaction volumes and total payments. Medicare is responsible for 1.2 billion fee-for-service payments each year. Fraudsters take advantage of the sheer volume of claims and the time needed for review. Many claim review processes are manual and fail to spot fraud patterns. Often, by the time a fraudulent scheme is spotted, significant losses have already occurred.  

The latest predictive analytics tools are designed to operate at the scale and speed needed to analyze even these staggering “Medicare level” data volumes, identifying anomalies in claims or billing patterns, identifying linkages with prior fraud cases and suspects, and flagging high-risk transactions for early investigation and intervention. 

As an example, showing how AI-powered analytic tools can solve similar challenges, consider the Treasury Department’s FinCen unit, which investigates financial crimes such as check fraud and money laundering, acting upon Suspicious Activity Reports (SARs) from banks. In just one year, from 2021 to 2022, SARs about check fraud nearly doubled – to 680,000. The Treasury Department’s Office of Payment Integrity (OPI) implemented a new machine learning process to mitigate check fraud in near real-time to address the increase. This led to the recovery of $375 million in potentially fraudulent payments before payments were made.  

Advanced analytical tools make fraud-fighting teams more efficient, enabling analysts to find and investigate suspicious cases quickly. Traditional investigative processes are time-consuming, but investigators can move much faster to advance and close cases by using automated scoring models to identify likely targets and capabilities like visualization, link analysis and geolocation mapping to drill down quickly. 

Data silos, data fragmentation 

Fraudsters often exploit gaps between disconnected government databases, aided by obstacles analysts face trying to connect the dots between beneficiaries or contractors operating across different programs and in different states. This leads to missed patterns of coordinated schemes or repeated fraud by the same suspects or groups. 

One such case is the Paycheck Protection Program (PPP), in which banks made government-backed loans to millions of small and mid-sized companies during COVID-19. Fraudsters exploited this program by using duplicate or false identities in multiple loan applications. Predictive analytics tools, powered by machine learning, could cross-reference applicant data with other federal systems in real-time to prevent fraudulent loans before disbursement. 

Most federal officials recognize the importance of breaking down data silos. Sometimes the obstacles are political as much as technical. For example, until 2023, federal law prohibited officials from using the Social Security Administration’s Master Death Roll of reported deaths for other federal anti-fraud and anti-waste “do not pay” efforts. Finally, in 2023, Congress authorized a three-year pilot program to use the Master Death File data across programs.  Since then, the GAO has recommended that Congress extend the use of the Master Death Roll and make it permanent. 

Another impressive example of stopping fraud by breaking down data silos is the case of the National Association of State Workforce Agencies (NASWA), a cooperative effort initiated by officials at several state agencies that manage unemployment insurance programs. With increasing fraud losses, NASWA members established a “Integrity Data Hub” so that multiple states could share their data and information on fraud patterns, trends, schemes, and suspicious bank and email accounts. The effort has helped prevent almost $5 billion in fraud since its inception.  

Delays and failures to adapt to emerging fraud schemes

Fraudsters continuously adapt to new technologies and exploit previously unrecognized vulnerabilities. Static, rules-based systems cannot keep pace, leaving federal agencies often one step behind. An example is income tax refund fraud, a scam growing rapidly in volume. In this scam, stolen identities are used to claim tax refunds before legitimate taxpayers file. Predictive analytics can help detect unusual filing behaviors, such as multiple returns filed from the same IP address and predict emerging fraud vectors by analyzing historical trends and suspicious activity. 

In response to these challenges, the IRS has stepped up its use of advanced analytics and AI, especially by its Criminal Investigation (IRS-CI) unit, which said in its annual report that using advanced analytics “increases CI’s ability to identify fraud and dedicate resources to the most impactful investigations that result in criminal prosecution.”   

In another effort to find tax evaders and potential fraud, the IRS recently announced that it uses AI and advanced analytics to help select which “complex partnerships” to audit. It has now launched audits of 76 of the largest partnerships, with average assets of $10 billion. The partnerships include hedge funds, real estate investment partnerships, publicly traded partnerships, large law firms, and others.  

Analysts, auditors, and investigators who are responsible for fighting fraud in government programs need the latest and most advanced analytics tools to be successful – technology that facilitates the collection, fusion and analysis of vast volumes of data from multiple sources in virtual real-time with all the benefits of AI and machine learning. With the new administration’s focus on reducing costs, it is an opportune time to apply the latest technology and methods to fraud reduction.  

Contact us to learn more about how fusion and analytics platforms can help combat fraud and waste.