Pension Design and Labor Supply: Evidence from Millions of Government Workers

Kristy Kim and Caleb Wroblewski, University of California, Berkeley

Over the past few decades, pension systems across the world have evolved (and continue to evolve) with changes in population demographics, fiscal environments, and institutional norms. With regards to demographic and fiscal changes, many countries have adopted reforms which delay the age at which an individual retires in order to address the steady aging of the population. Because a rightward shift in the age distribution may result in a decrease in the share of the working population, raising the average retirement age may alleviate pressure on pension programs that are funded by an increasingly shrinking pool of active workers. With regards to institutional changes, there has been a large shift of (public and private) pension coverage from defined benefit (DB) to defined- contribution (DC) programs, particularly in the U.S. There are a variety of factors which may have influenced the accelerating adoption of DC programs, including worker preferences for mobility across firms (DB plans favor those with a long tenure at a single firm) and the aforementioned funding concerns. DC pension plans can be more sustainable because payments during retirement are funded by savings made when an individual is working. However, these programs also transfer risk from the firm to the worker and require workers to make important savings choices during their working years.

Understanding the labor responses to these changes in pension designs is important for maintaining the health of retirement programs and securing the well-being for beneficiaries. It may inform policymakers on how to optimally adapt national pension plans, like Social Security, to achieve sustainable fiscal goals. In this research project, we aim to contribute to this literature by examining a large-scale reform in the U.S. retirement system for federal employees. This reform, known as the Federal Employees’ Retirement Systems (FERS) Act, introduced three crucial changes which widely reflect trends in other retirement systems: (1) the switch from a traditional DB program to a hybrid program that includes both a traditional annuity and a DC component; (2) increased minimum retirement ages; and (3) increased penalties for unsanctioned early retirements. We intend to exploit the various discontinuities that this this event provides, using a rich dataset to understand how retirement ages and tenure respond to financial incentives and risk. Additionally, we will explore structural models of labor supply to run counterfactuals and improve projections of pension design changes.