We build a structural model of retirement decisions by couples. Our model adopts collective household approach and exploits the externalities from joint leisure between the partners in decision of retirement timing. Given that for example 70% of Europeans and similar fraction of Americans in the age 50 to 67 live in couples (legally married or in informal unions), the joint spousal retirement decisions are relevant for policy making.

Our model incorporates three channels linking decisions of partners in the household. First, we introduce joint leisure into the utility function, to address the leisure complementarities. Second, the retirement decision of the spouse is reflected in the household budget constraint. Third, we allow for shock correlation within the couple, specifically in their productivity shocks (such as health shocks of one partner propagating to effective labor endowment of the other partner). Moreover, under assortative mating, we can expect partners to work in similar industries / firms and consequently be susceptible to similar business cycle shocks.

We calibrate this model to German economy and use it to deliver a number of policy experiments. First, we analyze the effects of changing the retirement eligibility wage of women on the retirement decisions of men. Our model shows that raising female retirement eligibility age may be an effective way to extend the duration of working period by men. Second, we also analyze the role of replacement rate of men's and women's pensions in joint spousal retirement decision.