The US social security system faces funding pressure due to the aging of the US population. I develop two theories of bounded rationality called life-cycle
horizon learning and finite horizon life-cycle learning to explore the welfare cost of social security policy uncertainty. In the models, agents use adaptive expectations
to forecast future aggregates, such as wages and interest rates. This introduces cyclical dynamics along a transition path, which magnifies the welfare
cost of policy uncertainty, compared to a rational expectations model. The ex-ante welfare cost of policy uncertainty (where policy is either a tax or benefit change in 2030 or 2040) is equivalent to 1.99 percent of period consumption for the cohort of agents most harmed in the life-cycle horizon learning model compared to 1.49 percent in a rational expectations framework.