This paper estimates a structural model of family retirement using US data from the Health and Retirement Study (HRS). It provides further insight into household retirement decision making and the reasons for interdependence in the retirement decisions of each spouse. Improvements in HRS data and matched employer provided pension histories allow more precise identification of key parameters governing interdependent behaviour within the household. In an earlier study we found that interdependence was due to preferences rather than coordination of retirement incentives in the budget, and in particular that it is not a correlation in preferences, but the appearance of the spouse's retirement status in the husband's and wife's utility function that is largely responsible for coordination of retirement between spouses. We now find that a measure of how much each spouse values being able to spend time in retirement with the other accounts for a good portion of that apparent interdependence. For the wife, the husband's retirement status influences her retirement decision only if she values spending time in retirement with her husband. For husbands, the effect of having the wife already retired on his retirement decision is roughly doubled if he enjoys spending time in retirement with his wife, but there is some effect even if he does not. This is consistent with our earlier findings that the husband is more influenced by having a retired spouse than the wife is. The increase in the extent of the dependence of the wife's labour supply on the husband's retirement from our past work probably is traceable to better measurement of the opportunity set facing the husband in HRS data. Once estimated, we use the model to investigate the labour supply effects of alternative social security policies, examining the effect of dividing credit for earnings evenly between spouses, or of basing social security benefits on the amounts accumulated in private accounts. Both policies change the relative importance of spouse and survivor social security benefits within the household and both raise the relative reward to work later in the life cycle. The incentives created are modest, and retirement responds accordingly. Nevertheless, at some ages, such as 65, there may be as much as a 6% increase in the old age work force under privatized accounts.
