My own half-baked idea is that because humans have longer childhoods and are very dependent on their mothers during their childhoods, there's an advantage to live past their reproductive years.
If a couple of women cease reproduction at 40 but one of them dies at 40 and the other at 50, there's a greater probability that the latter's offspring will survive and reproduce themselves.
So, increasing the gap between menopause and death increases the chances of success for offspring (because their mother lives longer to support them into adulthood). As childbirth becomes ever more dangerous (because of normal aging process) it "pays more" to stay alive and raise the kids already born than to gain more kids, at the risk of losing the ones you already have.
PLoS Comput Biol 9(6): e1003092. doi:10.1371/journal.pcbi.1003092
Mate Choice and the Origin of Menopause
Richard A. Morton et al.
Human menopause is an unsolved evolutionary puzzle, and relationships among the factors that produced it remain understood poorly. Classic theory, involving a one-sex (female) model of human demography, suggests that genes imparting deleterious effects on post-reproductive survival will accumulate. Thus, a ‘death barrier’ should emerge beyond the maximum age for female reproduction. Under this scenario, few women would experience menopause (decreased fertility with continued survival) because few would survive much longer than they reproduced. However, no death barrier is observed in human populations. Subsequent theoretical research has shown that two-sex models, including male fertility at older ages, avoid the death barrier. Here we use a stochastic, two-sex computational model implemented by computer simulation to show how male mating preference for younger females could lead to the accumulation of mutations deleterious to female fertility and thus produce a menopausal period. Our model requires neither the initial assumption of a decline in older female fertility nor the effects of inclusive fitness through which older, non-reproducing women assist in the reproductive efforts of younger women. Our model helps to explain why such effects, observed in many societies, may be insufficient factors in elucidating the origin of menopause.