Here's essentially the same graph varying the length of retirement instead of the stock/bond ratio (stock allocation is fixed at 60%). No surprises, although I wish it was a prettier picture for us ER folks.
BTW, one of my pet peeves is almost all retirement-related books and papers use a 20 or 30-year retirement. For some here 40 years or even 50+ years is not out of the question.
samclem: just an observation.... I believe using Shiller's data your harvesting approach would come close to a highest-first scheme (given the long interest rate is used and never has a loss, as Independent pointed out)
Independent: interesting info. Thanks.