Post Content

Erica Gaynor, Workflow Implementation Manager with the Child Support Enforcement (CSE) Division was published in the August edition of the U.S. Dept. of Health and Human Services’ Administration for Children and Families’ newsletter “The Child Support Report”. Gaynor and CSE’s work with predictive analytics is garnering national attention.

Below is the full article published in the “Child Support Report”

Predictive Analytics and Performance In The Public Sector

Erica Sabo Gaynor, Workflow Implementation Manager, MA. Dept. of Revenue, Child Support Enforcement (CSE) Division

Many organizations decided to incorporate data analytics into their daily business because it promises substantial increases in productivity, exciting new insights, and an ability to predict seemingly random behavior. The impact and level of success of data analytics in the human services industry is less clear though, particularly around performance measures.

After three years of using and assessing the impact of data analytics, the staff at the Massachusetts Department of Revenue’s CSE division has seen improvements but has had moments of uncertainty as well. Ultimately, we have learned that it is neither a magic wand nor a crystal ball, but an ongoing initiative that has provided incremental improvements in program performance. The analytics have also given us a broader understanding of the strengths and limitations of our agency’s data.

Background and scope

CSE has five regional offices, a Customer Service Bureau, and a central office that houses the offices of the Commissioner, the Chief Legal Counsel, Policy and Procedures, Finance, and others. We employ about 635 people, manage 230,000 open child support cases, and we collected $657 million in child support last year.

In 2012, CSE collaborated with Revenue Solutions Inc. to prioritize our casework, improve performance on federal measures, and prepare for a future data-driven system. To do this, we developed a predictive modeling program that included a collections model and a risk-based scoring system. The program would periodically assign a payment performance risk score to noncustodial parents, prioritize the parents by potential yield, and then provide recommended actions. We believed the actions would generate payments, bolster compliance, and route cases for closure.

Implementation

Since 2013, 10 CSE caseworkers have managed their caseloads using this predictive modeling approach. We call them our decision analytics (DA) population. By rolling this initiative out to a small group, CSE established a controlled comparison between the DA and non-DA populations so we could see the impact more reliably. This approach has also provided CSE the flexibility to incrementally modify the program to optimize results.

Findings

Predicting who pays:  Statisticians analyzed about 300 behavioral variables and found 20 attributes that have significant predictive power with respect to payment performance. Some of the attributes were intuitive, such as amount of child support owed or the amount of time since the last payment. Others factors surprised us, such as the age of the noncustodial parents’ vehicle registrations, their state of residence, and the value of their automobiles.

Changing relationships: the program often recommended using forms of contact outside of the general contact letters. For example, they have six different letters they can use, depending on the situation in the case. We also encourage them to engage in more proactive casework in an effort to boost compliance. These techniques gave us unexpected results. CSE has seen a shift in the dialogue between staff and customers. While some caseworkers were initially reluctant to contact noncustodial parents who did not appear to have issues that needed to be addressed, staff found these outbound calls resulted in more pleasant conversations than the incoming calls they were accustomed to receiving. Many customers appreciated this so much that they have maintained better communication with CSE.

The Lift: This term refers to a calculation used to measure the impact of analytics. It compares the rate of collections among the caseworkers using decision analytics to the rate of the staff who did not. The resulting figure represents CSE’s potential net gain if we use decision analytics across the non-DA caseload.

During the first year of implementation, we scored approximately 11,000 noncustodial parents, assigned them to a DA caseworker, and tracked their voluntary payment performance. After 12 months, CSE saw a 5 percent increase in the rate of collections based upon the performance of those noncustodial parents. Had the program been rolled out to the non-DA population, CSE projects an additional $15 million in collections.

CSE is expanding the program to include more staff. Program leaders actively seek input from caseworkers and make improvements to strengthen the program’s impact. As CSE makes these changes, we continue to learn more about our customers, the predictability of their payment compliance, and the impact of our actions. CSE will use this knowledge as we continually improve our use of data analytics and incorporate those successes into future agency-wide systems.

To read the full “Child Support Report” Newsletter, click here

Written By:

Recent Posts

Are you ready to join the DOR Team? posted on Oct 30

At the Department of Revenue, we are committed to recruiting and retaining talented people throughout the New England region. We have many positions open for a seasonal workforce to join our team. What’s a Seasonal position? Seasonal positions are essential jobs that help the Department   …Continue Reading Are you ready to join the DOR Team?

Tax Security 2.0 – A “Taxes-Security-Together” Checklist posted on Jul 16

DOR is sharing the following IRS “Taxes-Security-Together” checklist to increase awareness amongst taxpayers, and specifically tax professionals, about safeguarding private taxpayer information. The “Taxes-Security-Together” Checklist Some tax pros may routinely overlook these checklist items and others need to regularly revisit them. The steps are not   …Continue Reading Tax Security 2.0 – A “Taxes-Security-Together” Checklist

Why do most Massachusetts taxpayers file electronically? Lots of good reasons. posted on Mar 17

If you haven’t filed your Massachusetts personal income tax return, there are very good reasons to consider e-filing.  Millions of taxpayers have already taken advantage of e-filing, the most convenient way to get your taxes done. Why are we confident that e-filing is the very   …Continue Reading Why do most Massachusetts taxpayers file electronically? Lots of good reasons.