Counter-intuitive effect of shorter life expectancy

All of the modeling I've done until recently assumed a maximum age of 100. It would seem that living longer requires more assets at the outset of retirement than does dying at, say, 85. I recently tested that theory and was surprised to find that, in my case, my standard of living was lower if I died at 85 than if I survived to 100.

I think that can be explained by my being self-insured for long-term care expenses. Using information from Genworth I projected the annual cost of an ALF for 3 years and of a SNF for 2 years and applied those as special medical expenses in the last 5 years of life. When I change my maximum age to 85 and shift the calendar years those special expenses occur to ages 81-85, my SOL drops. I didn't make any other changes, and there are no other special expenses that are affected by the shorter life expectancy.

I surmise the model depends on investment growth to be able to pay those large expenses at end of life. The extra 15 years (living to 100 vs 85) seems to help grow that LTC fund. If my interpretation is correct, this is a caution to others not to assume that your heirs will be rolling in your leftover money if you have a maximum age of 100 but happen to die at 85. I suggest you test the effect of dying earlier than your current model projects, and adapt your plan to meet the needs of the more conservative result.

If my interpretation is incorrect I'd appreciate hearing suggestions for improvement from others.

Chris
(a user)

Comments

Dan Royer's picture

That is, as you suggest, counter intuitive. Your explanation makes sense, I think, but when I ran a quick model myself, I see a raised living standard, unlike you.

I wonder if you are observing the impact of the some borrowing constraint and that what appears to be a lower living standard late in life is coupled with a higher living standard earlier in life?

When I first tested this, there was a gap where regular assets dropped to zero. I tweaked special withdrawals to cover that gap without withdrawing any more than necessary. If I recall correctly, the resulting SOL for the shorter life was smooth, but lower.

I'll test it again and make sure that all other aspects are uniform.

Chris