Lifecycle models base on the assumption that households want to maximize their consumption and at the same time insure themselves against income shocks.
However, there is a lot of (c.p.) consumption dispersion observed in the data. At lower ages, the literature argues that income processes are very persistent (and not insurable), and hence such consumption profiles are optimal.
However, at older ages, especially at retirement, most of the income uncertainty goes away. Yet, consumption keeps dispersing with age.
The only argument I heard that tries to justify the lifecycle models given this empirical observation is that at older age, "available income" is still a random walk, given that health expenses become potentially very high and persistent.
Are there other arguments that can reconcile the empirical observation of consumption dispersion at older ages with life-cycle consumption models? Are there any studies on that?