I've been traveling, so just catching up on this thread.
The only people who can make a choice are people who don't need any money from Social Security for 8 years. And that's the people who engage in this "when to take SS" debate---people who already have plenty of money/income. And, really, it's only the folks who are on the cusp of the need/don't need the longevity insurance aspect.
The above, and FIRED, is the realm I'm thinking of when participating in this post; the other areas are not applicable to me and also less interesting to me.
In fact, it is better that they take it early in that based on most likely mortality they end up with less over their lifetimes which leaves the whole system in better shape....
Or maybe more over their lifetimes and leave the system in better shape? Isn't it possible they end up with more AND be less of a drain on the system? How? Well, all that would have to happen is that the money they would have pulled while waiting until 70 grows at a "high enough" growth rate. The pie is not a fixed size. So you can get less than your maximum lifetime benefits from SS, yet have more money to spend and/or pass on. And, oh, by the way, pay MORE taxes along the way. Maximizing the whole model, not just lifetime benefits, seems like a smarter approach.
An unrealistic but easy way to understand the above...at 62, imagine you have a portfolio with no risk that's guaranteed to earn 10% above inflation for the next 8 years (thought experiment only, of course). Allowing all of that portfolio to grow rather than spend it while waiting until 70 would certainly generate more spendable cash over the plan duration, even if that duration was 40 years. That's not to say this is a likely scenario, it's only to say that even if one lives a long time, there is a non-zero chance that taking SS early could still provide more spending. This post provides more on this idea of how higher returns can cost SS less and allow the retiree to spend more:
If Y% is 7%, it lasts more than 38 years. Still going strong at age 108.
All of that said, my playing around with opensocialsecurity.com using 2017 mortality and adjusting for the time value of money suggest that the expected present values are all within 2-3% of each other so the decision doesn't matter all that much.
^This is the take-away for me.
It all goes back to what I said very early in this thread: "Do you feel lucky". But one must realize lucky in longevity pushes one way, lucky in your portfolio return pushes the other way. And you come out the other side and realize all of this is unknowable, so you "pays (claims) your money, you takes your chances".
Is it prudent to ignore the fact that two factors could "go wrong" (extended longevity, lower CGR)? NO! It's not smart to ignore that these things can go in a direction that would confound the results. As discussed, when deciding, one must not look at only the $, but the utility of extra cash versus the "pain" associated with not having extra cash. That is a personal thing. In the industrial engineering curriculum, I learned how to do a decision tree to model this kind of thing. You have "decision nodes" and "probability nodes" and each of all possible outcomes has benefit or a cost. As you work back from the "leaves" of the tree, all the way to the trunk, you can get a solid understanding of what the best decision is. Of course, like any other tool, the trick is to change the weightings until you get the answer you want