How the U.S. Marshals Service Saved $300 Million By Identifying Over-Charging Prisons

Prisoner Ricky Wheat looks out from his cell at the Georgia Diagnostic and Classification Prison, Tuesday, Dec. 1, 2015, in Jackson, Ga.

Prisoner Ricky Wheat looks out from his cell at the Georgia Diagnostic and Classification Prison, Tuesday, Dec. 1, 2015, in Jackson, Ga. David Goldman/AP File Photo

The service rents space for prisoners awaiting trial from federal, state and local jails—and realized that it needed some more leverage negotiating rates.

As the White House calls on agencies to do more with less and places increasing emphasis on efficiency in government, John Scalia has shown that data analytics can play a crucial role in boosting the government’s bottom line.

As chief of forecasting and analysis for the U.S. Marshals Service’s Prisoner Operations Division, Scalia is responsible for calculating the costs for housing the service’s roughly 52,000 detainees in jails across the country, and over the last decade, he’s helped the agency save more than $300 million.

Putting up prisoners isn’t cheap. In fiscal 2017, the Justice Department allocated $1.5 billion to the Marshals' Prisoner Detention Program. Because the agency doesn’t operate any of its own facilities, most of that money gets paid out to the more than 70 federal, state, local and private jails where the service rents space for inmates awaiting trial, Scalia said.

» Get the best federal technology news and ideas delivered right to your inbox. Sign up here.

But different facilities have different expenses, so every year the Marshals Service negotiates with individual jails over the daily price of locking a prisoner up in that facility.

For years, Scalia said, the process was “very monopolistic.” The facility would set its rate for that year, and the Marshals Service would have little say in the matter. If the costs were found to exceed what the agency was paying, the Marshals would reimburse the jail. If the service was paying too much, the facility was supposed to reimburse them, but Scalia said that rarely happened.

But Scalia eventually evened the playing field using the SAS Analytics Platform. After building a model that calculated a base rate for housing prisoners at each specific facility using labor and operational costs, he found the Marshals Service was overpaying in many instances.

Armed with hard numbers from Scalia’s analysis, the agency gained more leverage in the price-setting negotiations and could bargain jails down to a more fair rate.

“This one facility came in and asked for a 75 percent rate increase,” Scalia told Nextgov. “Populations had decreased, but they weren’t doing anything to control costs. We said no. We understand that you have cost considerations, but [the Marshals Service] can only pay so much. We’re not going to subsidize your operations.”

Over the last decade, Scalia’s model has saved the agency an estimated $300 million. He said he next plans to use analytics to create life models for the Marshals' fleet of about 5,000 vehicles to inform decisions about repairs and trade-ins.

Though he’s received overwhelmingly positive feedback on his model, Scalia said that when it comes to big data analytics, the Justice Department has been a bit slow on the uptake.

“These are all operational people. They want to know how to change the process,” he said. “They’re not thinking ‘OK, let’s analyze what’s going on and see where the process succeeds or fails.’”

In recent years, however, he said the agency has started to see the value of analytics, as has much of the federal government as a whole.

Projects that can show concrete benefits of using big data to make decisions—like Scalia’s—can help overcome the government’s hesitation about adopting analytics, according to Steve Bennett, director of global government practice at SAS. When it comes to showing value, he said “nothing beats an example,” especially at operational agencies like the Marshals Service where leadership may be more dug into old processes.

“Any time you can use analytics and have an easy to quantify return on investment, that’s a great story,” Bennett said.

But beyond workplace culture, many agencies that want to adopt analytical tools still face a number of roadblocks. Poor data management and governance put valuable information out of reach, and limited resources can slow agencies’ ambitions to adopt new technology.

As the government wades deeper into the world of analytics, Bennett warns that big data is not a catch-all solution for every issue at hand.

“It’s having a moment in the sun, but you’ve got to watch the hype cycle,” he said. “Analytics is not going to solve every problem, but it can help turn these operations of government in the right direction and make them more efficient. [It’s] absolutely not a solution on [its] own.”