If anyone read the Ibbotson/Milevsky paper posted
here, there's a good section on "mortality credits" (page 68, table 6.2 for those who didn't read
).
Basically, the underlying benefit of an annuity arises when other folks in the annuity pool die and you don't. You (in theory) are collecting more money because the dead people are giving you what they would have gotten. Table 6.2 shows the increase above regular interest rates you should expect at a given age for starting an immediate annuity. For example, at age 65, you get an 83 basis point boost to your interest rate (presumably the risk free rate of return). If you wait to annuitize till age 85, you get a 725 bp boost to returns. When you consider expenses on most annuities might run 100-300 bp's (just a wild guess), it probably doesn't make sense to annuitize at 65 for a meager 83 bp boost in returns (which will more than get eaten by expenses) but it probably does make sense to annuitize if you get 725 bp boost to returns (which might still net you 400-600 bp after expenses).
Ibbotson, et al state in their paper that annuitizing at age 60 is too early, and age 90 is too late. They recommend dollar cost averaging into annuities starting around 65-70 and continuing until 80-85.
I think the often heard complaint against annuities comes from the folks on here who define retirement as around age 50 or so (typical early retirement). Ibbotson, et al would say don't get an annuity at that age or any where close to that age. Instead they recommend waiting 15-20 years.
In their paper, they also explain that if you have even a modest "bequest motive" (desire to leave assets to others or to charity - ie - not die broke), and a modest to aggressive risk tolerance, your optimum allocation to annuity products would be 14-24%. And that is for a 60 y.o. male. My guess is that an even smaller allocation to annuities would be expected at younger ages, which many folks here are able to retire at.
The notion of longevity insurance is an interesting notion. At this point it is hard to get excited about the potential for a 3-3.5% real return for a locked in 45 year investment with no liquidity and company specific risk of the insurer (not to mention the fact that there's a ~50% chance we won't make it to age 85!!!). It still seems like it would be cheaper to "roll your own" and/or annuitize at age 85 if you really wanted longevity insurance.