New programs often have limited staff and budget with which to establish processes and deliver results, and paid family and medical leave (PFML) is no exception. It’s critical, therefore, for states to make efficient use of the resources they have. AI and automation can help states optimize paid family and medical leave programs to provide reliable services to constituents and enable staff to succeed.
Automate Routine Tasks
A growing trend in the public sector is harnessing the power of automation to handle routine, repetitive tasks to give employees more time to attend to higher value work requiring human intervention. “As an agency staff member, you shouldn’t be spending your time updating someone’s address or phone number,” said Admira Makas, Paid Family and Medical Leave Subject Matter Expert at TCS. “These are things that can and should be done through self-services, guided by an automated system.”
Robotic process automation (RPA) can be used to streamline basic data processing and administrative tasks, such as employer registration or claim application. For example, RPA can be used to determine whether an employer is required to register for PFML in compliance with specific state laws and regulations. If they are liable, the system presents the relevant screens to collect the required employer information and provides the annual tax rate and wage reporting deadlines. If they are not required to sign up, they are presented with screens providing the option to register voluntarily. Similarly, automation can be used to determine a claimant’s eligibility and weekly benefit amount based on the information provided by the employee and wage data reported by employers. If the employer or claimant has questions, or encounters an issue with their registration, this is the time that a state employee can be brought in to help resolve the issue.
With routine tasks managed by automation, state employees are able to focus on more complex issues that require their expertise. This allows for faster and more accurate service delivery to both claimants and employers. Those outcomes are also more predictable as they are based on standardized criteria.
Manage Fraud and User Experience with AI
AI’s pattern recognition capabilities allow state staff to hand off even more routine tasks to technology. PFML shares similarities with other state benefit programs, such as unemployment insurance, that make it an attractive target for fraud. The COVID-19 pandemic shed light on potential vulnerabilities in these programs, but it also prompted the development of myriad countermeasures to suit a variety of program needs.
States can incorporate AI-driven fraud analytics directly into their PFML systems, which is an advantage PFML holds over older programs that have had to work modern fraud detection and prevention tools into legacy programs. These tools are able to learn from the data they process, meaning that their performance will improve over time without state employees constantly having to feed new data into it. Makas advises that fraud analytics should be a built-in component of a PFML solution, not an add-on. “If you’re only thinking about the fraud component once you start to see evidence of fraud, you’re already too late,” she cautioned.
States can also take advantage of a related tactic, sentiment analysis, for real-time feedback about user experience. When Netflix experiences a service interruption, for example, users will often take to social media to investigate whether the issue is server-wide or specific to their account before reaching out to customer service. Users of government programs such as PFML are increasingly doing the same thing. States can use sentiment analysis to flag this kind of social media post, bringing service issues to their attention without waiting to receive a support inquiry. Catching and fixing problems quickly means that fewer users will experience the issue, and the ones who do have problems will not have to wait long to solve them.
Offset Small Data Pools with Predictive Analytics
One of PFML’s biggest concerns, Makas shared, is the worry that as a solely state-funded program, there is no backup from the federal level if the program faces insolvency. As PFML programs are developed, states are asked to demonstrate an understanding of how solvent their programs will be and how they can maintain solvency over time. The traditional method in other programs such as Worker’s Compensation involves an actuarial analysis every three to five years that results in a model for the next period’s results, but as Makas explains, this is challenging with a new program. “There’s no data for that state to give an accurate accounting of how many claims they’re going to expect,” she said. “It’s not a total unknown, but you don’t really know how your population and the types of life events individuals will face is going to change over time.” States also have to account for unexpected increases in use; Rhode Island’s program had a three-to-five times increase in monthly claims in 2020 and 2021 due to the COVID-19 pandemic.
While true clairvoyance may be out of reach, states can leverage predictive analytics to prepare for this kind of situation. Although they lack historical data to draw from, what data they do possess can be used to plan for both short-term and long-term solvency. It can also be used to model and prepare for a future pandemic-like event without simply raising collection rates and overburdening employers and employees.
Emerging technologies have the potential to boost PFML programs’ speed, efficacy, and capacity with a small upfront investment and short learning curve. Employing AI and automation to streamline operations makes state staff’s jobs easier and enables a more convenient, satisfying, and successful experience for employers and applicants.
To learn more about leveraging AI and automation for a more effective PFML program, click here.
