Fight waste and fraud with tech

Note: This is a post co-authored with my friend and colleague Nick Sinai as lead author. Nick is a former U.S. deputy CTO in the Obama administration, currently venture partner at Insight Venture Partners, and adjunct faculty at the Harvard Kennedy School.

Last week was a big step for IT Modernization, with the release of the administration's final IT modernization report, the announcement of new GSA centers of excellence, and President Donald Trump signing the Modernizing Government Technology Act to help agencies move to the cloud and buy modern software. Not many things unite the past two administrations and Congress -- but Obama Administration U.S. CIO Tony Scott, Rep. Will Hurd (R-Texas) and Special Assistant to the President Matt Lira, in the White House Office of American Innovation, all deserve credit for making the MGT Act a reality.

In that rare spirit of bipartisanship, we have a suggestion for another area where Congress and the Trump Administration can continue to work together: Fix a system that has been set up to improve detection of improper payments called Do Not Pay. If Congress really wants to drain the swamp, it should go after waste and fraud -- and use technology to do it.

The Improper Payments Elimination and Recovery Improvement Act of 2012 was designed to provide data and analytics to help detect improper payments as a centralized service available to agencies on a voluntary basis. It established a Do Not Pay Business Center in the Department of Treasury whose help federal agencies could get before paying a beneficiary, awardee, or grantee.

It is estimated there are $144 billion in annual federal improper payments. Not all of that $144 billion is fraud, of course. The best estimate is that almost $36 billion in 2016 involved overpayments to ineligible recipients that could have been prevented by more effective use of data matching. Another $39 billion, roughly, was tied to overpayments due to administrative errors, much of which could also be addressed through better use of data and analytics.

It hardly needs saying there are multiple reasons to reduce improper payments. Stopping ineligible payments before they are paid is far more cost-effective and efficient than trying to get ineligible recipients to repay the government. Reducing improper payments can free up funds for other important priorities. Finally, preventing waste, fraud, and abuse is important to rebuilding citizen trust in government.

Agency efforts to locate improper payments go back decades, but Congress concluded with this legislation that those efforts needed to be supplemented with more data and analytic capability. The law was designed to provide such data and analytics to help detect improper payments as a centralized service available to agencies, for free, on a voluntary basis.

The Do Not Pay Center was designed to act as a shared services entity providing databases and analytics to agencies requesting them. The idea was that the center could get better prices on any commercial databases than an individual agency could, and also help with capability to analyze the databases for improper payment detection. Very few agencies have had any sort of data analytic capabilities to use for improper payment detection.

The statute also directed the Office of Management and Budget to authorize the Do Not Pay Center’s access to relevant data sources (including government sources such as Treasury's Office of Foreign Assets Control's Specially Designated Nationals List and data from the General Services Administration's System for Award Management). OMB also is able to authorize Do Not Pay to purchase relevant private databases. (Full disclosure: Sinai's venture capital firm is an investor in companies that seek to help Treasury fight improper payments with better data and analytical tools.)

The problem is that OMB has been very slow in getting the new system up and running. The statute and subsequent guidance require OMB to approve Treasury’s request to use new data sources, including commercial data sources. It was not until September this year that OMB issued a Federal Register notice announcing Treasury’s intent to use other federal and commercial data in Do Not Pay, including three government data sources. That notice was the culmination of a nearly two-year OMB approval process.

Why did OMB take almost two years? Our guess is that the main culprit was the end of an administration and the beginning of a new one. But OMB has many competing priorities and oversight responsibilities, and arguably does not have the resources to execute on all of them in a timely manner. Do Not Pay likely did not rise to the top of the priorities list. Just after Thanksgiving, OMB announced some approvals for Treasury access to databases, but many more are needed.

Another bureaucratic impediment to the success of Do Not Pay is the requirement for agencies to undergo outdated computer-matching agreements to use data from multiple sources. Congress also established protections to ensure procedural uniformity in carrying out computer matches and included due process rights for individuals whose benefits may be affected. Those amendments are generally known as the Computer Matching Act.

A recent Government Accountability Office report found that the Computer Matching Act provides substantial disincentives for other federal agencies to use Treasury’s Do Not Pay center, and more generally, discourages agencies from using data matching techniques to prevent improper payments. Specifically, officials at six large agencies have said that the review processes are too lengthy and resource-intensive, and that statutory durations for conducting matching activities are too short. Similarly, officials from offices of the inspector general at four agencies, in that same report, stated that the length of the approval process and the requirement that proposed agreements be approved by data integrity boards discouraged them from computer matching.

How could the Trump Administration jumpstart Do Not Pay? The administration needs to make this a priority, since OMB has been slow to do this itself. OMB should commit to a fast track, 30-day approval process -- its policy should be that, as the default position after 30 days, the director of OMB has approved Treasury’s request to acquire new data. 

Second, the Administration should actively work with Congress to amend the 2012 statute with two primary fixes:

1. Treasury would no longer be required to get explicit clearance from OMB to use other government or private sector data sets; and
2. The Do Not Pay initiative would be exempted from the Computer Matching Act. These fixes would speed up the ability of Treasury to build a modern, centralized improper payments office to fight fraud.

Some might worry that exempting Do Not Pay from the Computer Matching Act would potentially threaten the privacy of Americans. But the Treasury Department handles a number of databases with very sensitive information, such as those in the IRS. Treasury could set up a thorough privacy review process -- with impartial experts -- that works for the modern data age.

Longer term, the administration should also explore expanding Do Not Pay to include additional program integrity capabilities and a program integrity center of excellence.

If Congress and the Trump administration are serious about fixing government, fixing Do Not Pay would be a good place to start.