The Long-Term Effect and Optimal Design of Auto-Enrollment Policies

by Taha Choukhmane, Yale University

While more American workers are entering retirement without defined benefit accruals, and while Social Security faces a long-term financing shortfall, retirees have to increasingly rely on defined contribution (DC) wealth to achieve retirement security. In this dissertation, I study the long-term effect and optimal design of an increasingly popular intervention aimed at raising DC saving: auto-enrollment. Moving from an opt-in regime (where employees select to enroll in their workplace pension plan) to an auto-enrollment regime (where employees are automatically enrolled with the option to opt-out) has been shown to have a large and positive effect on DC saving in the short run (Choi et al., 2004). However, the long-term effect of the policy is unclear and auto-enrolled workers – who have accumulated more assets in the short-run – may have lower incentives to save in the future and higher incentives to retire and claim Social Security early. Auto-enrollment policies are too recent; it would require 40 years of data to directly measure the effect on retirement wealth when auto-enrolling a 25-year-old and the timing of retirement and the Social Security claiming decisions. To overcome this challenge, this paper will develop and estimate a two-asset lifecycle model with default effects to match the short-run evidence on auto-enrollment.

Back to 2018 Dissertation Fellowship Recipients