The Mississippi governor and legislature have issued a challenge to state agencies: get more people working. It’s a goal that would resonate in any state—but the issue is especially urgent in Mississippi, which not only has the nation’s lowest labor force participation rate but also the highest rate of recidivism. Far too many residents are cycling back into prison instead of into productive employment.

Mississippi’s workforce agencies are meeting this challenge head-on by tapping into an often-hidden goldmine: the vast stores of state administrative data. Four state agencies—including the Department of Corrections—are collaborating to unearth the data that can reveal which programs best prepare incarcerated individuals with the skills to earn a sustainable income.

Many large private companies regularly use their data this way, operating in a continual evaluation mode. Corporations draw on data in real-time to test new products, services, and business practices and determine which is likely to succeed.

By contrast, the public sector often builds evidence more slowly, with long intervals between major studies. However, public agencies can accelerate this process by tapping into their administrative data to test strategies to improve their programs—just as successful private companies do.

What’s Holding the Public Sector Back?

State and local agencies already collect large volumes of data in the normal course of administering their everyday operations. But unlike corporations, public agencies often lack integrated data systems that link program participation data with outcome data such as employment results.

Moreover, public agencies lack the staff capacity or methodological expertise to extract insights from their data. These are specialized skills that corporations typically have in-house, but that public agencies often must obtain through partnerships with outside researchers.

The Solution: Learning by Doing

At MDRC, we are collaborating with data scientists at the Coleridge Initiative to help states use their administrative data to generate actionable evidence to inform policy. We call this approach “learning by doing.”

In Mississippi and two other states, we started with a short boot camp. Instead of push-ups and squats, participants practiced the fundamentals of data analytics and evaluation. Then we rolled up our sleeves and start working side-by-side with agency staff to help them build and apply these “data muscles” on real projects. Each state is asking a different question, based on their particular needs, though employment is definitely a theme.

  • In Mississippi, agencies are examining which training and employment services best help individuals find stable jobs after leaving correctional facilities. Reincarceration costs the state nearly $25K per inmate each year. For every dollar spent on corrections, society bears an additional $10 in costs to families and communities. Smart use of data can help reduce these costs while improving outcomes.
  • In Ohio, the Department of Job and Family Services is using its own data to determine whether unemployed workers who are also receiving food assistance return to work faster if they are provided with reemployment services. The result could be Buckeyes finding jobs more quickly.
  • In Pennsylvania, several state agencies are pooling their data resources to see if enrollment in training and employment services rises if offered in the same location where people receive food and other assistance. If so, the state will know how to get more Pennsylvanians connected with the services they need to succeed in the job market.

Of course, improving public programs is not always as easy as testing a new product design in the private sector. It can take time to observe meaningful outcomes from programmatic interventions, because the problems public policy aims to solve are often entrenched and complex. Mississippi’s multi-year data archive has been a major advantage in this regard.

But public agencies can also use administrative data to make real-time improvements. For example, interventions can be broken into smaller, easily testable components like different outreach strategies, service-delivery approaches, or eligibility criteria. To measure effectiveness, agencies can assess nearer-term outcomes like program completion and draw on existing research linking those outcomes to longer-term impacts.

Additional Insights Await

Doing more with less is never easy—but that’s the challenge many public agencies face today. One way to stretch every dollar is to make sure it’s well spent. And who better to lead the search for effective program strategies than the state and local agencies who manage and fund these programs in their own communities?

By tapping into the goldmine of administrative data, public agencies can uncover insights that boost efficiency, reduce costs, and expand opportunities for residents in Mississippi, Ohio, or Pennsylvania—and beyond.

Rick Hendra is the project co-director of the State IMPACT Collaborative, along with Melissa Wavelet. Hendra also is director of the Center for Data Insights at MDRC, where he applies advanced data analytics to help government agencies and organizations use data more effectively to serve communities in need. Wavelet also serves as director of state and local strategy for MDRC, working with government agencies across the country to answer important policy questions using research and evaluation.