Do State Automatic IRA Mandates Affect Self-Reported Employer Retirement Plan Coverage? Evidence from Staggered Policy Adoption

Abstract

State auto-IRA mandates require employers without retirement plans to enroll employees in state-facilitated IRAs. This paper estimates whether these mandates affect self-reported employer retirement plan coverage—a distinct outcome from auto-IRA participation itself, since CPS asks about "employer" plans while auto-IRAs are state-facilitated individual accounts. Using CPS ASEC data from 2010–2024 and a Callaway-Sant'Anna difference-in-differences design exploiting staggered adoption across five states with meaningful post-treatment data (Oregon, Illinois, California, Connecticut, Maryland), I find an overall ATT of 0.75 percentage points (SE = 1.0 pp; wild bootstrap $p = 0.48$), not statistically significant. This null result is driven by Oregon's anomalous negative effect ($-2.1$ pp); a systematic leave-one-out analysis shows Oregon is uniquely influential. Excluding Oregon yields a significant effect of 1.57 pp (SE = 0.62 pp; $p < 0.01$). A triple-difference design exploiting firm-size phase-in—comparing small firms (targeted by mandates) to large firms (placebo)—provides additional identification. Randomization inference with 2,000 permutations yields a two-sided $p$-value of 0.47 for the overall effect. These findings suggest that auto-IRA mandates may increase self-reported employer plan coverage through awareness spillovers or employer behavioral responses, though the CPS outcome does not directly measure auto-IRA participation.