Costly attention and retirement

In UK data, I document the prevalence of misbeliefs regarding the State Pension eligibility age (SPA) and their predictivity for retirement. Exploiting policy variation, I estimate a lifecycle model of retirement in which, motivated by patterns in belief data, rationally inattentive households learning about uncertain pension policy endogenously generates misbeliefs. Misbeliefs explain 51% of the excessive (given financial incentives) drop in employment at SPA when constrained to replicate the belief data patterns and completely explain it when not. To achieve this, I develop a solution method for dynamic rational inattention models with persistent beliefs. Costly attention makes the SPA up to 15% less effective at increasing old-age employment. Hence, information letters improve welfare and increase employment.