We analyse the joint distribution of the durations until retirement of Danish husbands and wives. We estimate a multivariate mixed proportional hazards model that allows for interdependence in the times to retirement of spouses. We find evidence of strong complementarities in leisure times. Symmetrically for husband and wife, low own income and poor health are found to induce individual retirement (prior to spouse's retirement), whereas low spousal income is not, and neither party is found likely to substitute own for purchased care when the spouse is in poor health. Furthermore, high wealth and low income are found to spur joint (simultaneous) retirement.
