1. Take-up of Public Employment Services
Many government programs suffer from low participation, particularly when involvement is voluntary, which can limit their effectiveness in achieving public welfare goals. A key example of this is employment services within Active Labor Market Policies (ALMPs), which aim to improve jobseekers' prospects through services like job search assistance, training, and subsidized employment. Despite substantial investments—such as the 0.6% of GDP OECD countries allocated to public employment initiatives in 2021—take-up rates for these services remain low, on par with or lower than other social programs. This project seeks to identify the barriers that prevent greater participation in public employment services and to design targeted, behaviorally informed interventions to encourage higher engagement. To achieve our goals, we combine user research with survey experiments and field experiments.
2. Using AI to Enhance Public Employment Services
Discrimination in public service delivery creates barriers to both access and quality, undermining equity. The rise of statistical methods, machine learning, and AI in these services has prompted growing interest in their impact on fairness. For instance, while AI-based profiling can improve accuracy, it may also exacerbate inequities by discriminating against marginalized groups. Concurrently, a long-standing body of research, starting with Lipsky (1980), highlights persistent bias in interactions between citizens and public service workers, often driven by heavy caseloads or complex cases. Against this background, our project aims to explore if and how generative AI can help reduce bias and promote greater equity in the delivery of publicly funded employment services. In this project, we combine user research with lab and field experiments.
Many government programs suffer from low participation, particularly when involvement is voluntary, which can limit their effectiveness in achieving public welfare goals. A key example of this is employment services within Active Labor Market Policies (ALMPs), which aim to improve jobseekers' prospects through services like job search assistance, training, and subsidized employment. Despite substantial investments—such as the 0.6% of GDP OECD countries allocated to public employment initiatives in 2021—take-up rates for these services remain low, on par with or lower than other social programs. This project seeks to identify the barriers that prevent greater participation in public employment services and to design targeted, behaviorally informed interventions to encourage higher engagement. To achieve our goals, we combine user research with survey experiments and field experiments.
2. Using AI to Enhance Public Employment Services
Discrimination in public service delivery creates barriers to both access and quality, undermining equity. The rise of statistical methods, machine learning, and AI in these services has prompted growing interest in their impact on fairness. For instance, while AI-based profiling can improve accuracy, it may also exacerbate inequities by discriminating against marginalized groups. Concurrently, a long-standing body of research, starting with Lipsky (1980), highlights persistent bias in interactions between citizens and public service workers, often driven by heavy caseloads or complex cases. Against this background, our project aims to explore if and how generative AI can help reduce bias and promote greater equity in the delivery of publicly funded employment services. In this project, we combine user research with lab and field experiments.