The U.S. Department of the Treasury sits at the financial center of the federal government, responsible for managing trillions of dollars in annual payments, collecting revenue, financing government operations, and safeguarding the integrity of the nation’s financial systems. Through the Bureau of the Fiscal Service, Treasury disburses more than one billion payments each year on behalf of federal agencies, making it one of the largest payment operators in the world. That scale brings both responsibility and risk, particularly as fraud and improper payments remain persistent challenges across federal programs.

In this conversation, Justin Marsico, Assistant Commissioner for Fraud Prevention and Financial Integrity, U.S. Department of the Treasury, Bureau of the Fiscal Service, sits down with Carly Mitchell, Partner at Guidehouse, to unpack how Treasury is strengthening payment integrity at the final mile. Marsico outlines how Executive Order 14249 is reshaping payment verification, improving visibility across federal financial systems, and centralizing disbursement authority to better prevent fraud before funds leave the government.

This article is the second part of a three-part series. In the first installment, Marsico discussed how the Treasury is modernizing the Do Not Pay program, reducing administrative barriers, and accelerating agency adoption.

Edited Excerpts

Q: Improper payments are a major part of your expanded role. Let’s start with Executive Order 14249. What does the executive order require, and how did it come about?

Executive Order 14249 was signed earlier in the administration to make fighting financial fraud and improper payments a priority for the federal government. It has a few key areas of focus.

One of those is payment verification. As payments are being made, Treasury, as the disbursing entity, should be screening and validating those payments before an agency certifies that they are proper. This inserts an additional step in the process that provides feedback to agencies about their payments, to identify potential improper payments or fraud at the very last moment before funds leave the federal government. The executive order instructs us to screen for several specific risk indicators at that point.

The executive order also includes provisions aimed at strengthening financial management more broadly. One of those is pushing for the consolidation of core financial management systems across the federal government. This is important from a fraud perspective because one of our biggest challenges today is the lack of visibility across the many different financial systems in use, along with limited transaction-level traceability as payments flow to Treasury from both cash management and accounting systems.

The intent is to create a simpler, more consistent approach to financial management while also ensuring that, where necessary, we can see detailed transaction-level information across the government.

Another important component of the executive order is that it instructs Treasury to begin revoking the authority of certain agencies to disburse payments on their own. Treasury already disburses over 90% of federal payments, but a number of agencies still operate as Non-Treasury Disbursing Offices, or NTDOs. Revoking those authorities brings more payments under Treasury’s purview, which allows us to apply the same controls and payment verification processes across a broader set of federal payments.

Q: What are some of the top initiatives you and your team are focused on right now under the executive order, and how will those efforts and the data assets you’re bringing together help agencies as they implement?

The biggest thing we’re focused on right now is implementing the executive order’s requirements related to payment verification. For example, we are required to screen payments at the time of payment to ensure that a payee is not deceased. We also need to ensure that the account from which an agency is paying does not have a negative balance at the time the payment is issued.

That’s what we’re actively building today, and we’re using many of the same data sources that power Do Not Pay to support payment verification. The main difference is where these checks occur in the lifecycle.

Agencies should be using Do Not Pay earlier, when they’re determining whether an individual should receive a benefit, whether a contractor should receive an award, or whether a university should receive a grant. Payment verification happens later. This is the final check we have before funds actually leave the federal government.

At the time of payment, we’re applying a smaller, targeted set of data sources. When certain conditions are met, we make the decision to return those payments to the agency. For example, if an agency submits a file with a thousand payments and we identify one payment that would go to a deceased payee, and there is no legally appropriate reason for that payment, we will process the remaining payments and return that single transaction to the agency for review.

The agency then has the opportunity, if it chooses, to remake the payment and let us know that it is, in fact, proper. That’s what payment verification is designed to do.

In addition to death screening, we’re also building analytics to validate bank accounts and verify the taxpayer identification number associated with a payment record.

The challenge is that we’re making these changes inside a live production payment system that processes more than a billion payments each year. Many of these are lifeline payments going to seniors, retirees, and veterans. The margin for error is zero.

To manage that risk, we’ve taken what we call a soft-launch approach with agencies. Right now, we have a small number of agencies participating in a soft launch for death screening. We meet with those agencies individually to understand the specific circumstances in which they may need to make payments to deceased individuals, such as survivor benefits or funeral expense payments for veterans.

We document those appropriate use cases, configure the rules accordingly, and only then bring an agency into the soft launch. That approach allows us to validate the controls while minimizing disruption to critical payments.

Q: Given that each program has different constituents, business rules, and partnerships with states that evolve, what should federal agencies and state-level recipients be thinking about as they evaluate their own processes, systems, and risk management practices, especially from a technology standpoint?

The first and most important thing is to fully use Do Not Pay if you’re eligible to do so, whether you’re a federal agency or a state administering federally funded programs. We provide Do Not Pay data sources free of charge, and that model was intentional. The goal is to make improper payment screening and fraud prevention as easy and accessible as possible.

We know agencies and states are resource-constrained, so we’re doing a lot of the foundational work to establish a baseline set of controls that can be relied on across the federal government and in state-administered programs.

The second consideration is recognizing that there are often existing integrity hubs within agencies or states where this type of work is already happening. That could include teams that handle data access contracts, develop analytics, or define policies and processes for responding when potential fraud or improper payments are identified. There’s a lot of effort already underway, and wherever possible, agencies should be reusing and building on what already exists rather than duplicating it.

The third thing we want to understand in the Do Not Pay space is where we may be missing critical data sources today. We’ve spent a significant amount of time this year analyzing why improper payments and fraud persist across federal programs, using government-wide reporting data.

In some areas, like death data, Do Not Pay has very strong coverage. We currently have six separate data sources that identify when an individual is deceased, and that provide better coverage than what agencies can access anywhere else.

In other areas, our coverage is not as strong. We’ve been actively identifying those gaps and working to bring new data sources into the program. But we need agencies and states to tell us where they see recurring fraud patterns or repeat improper payments year over year.

If an agency knows it’s vulnerable to a particular risk, understands what could mitigate it, but lacks the resources or authority to address it independently, that’s exactly the type of input we need. That information helps us decide whether we should build a new tool, acquire a new data source, or make a capability available across the federal government through Do Not Pay.

CDO Magazine appreciates Justin Marsico for sharing his insights with our global community.