My Horizon Spending

This is the last post of 3 part series.  The previous two were on long-horizon spending, the first one was about a kind of an endowment-ish thing that looked at spend distribution "medians" at the 50 year mark and the second was a consumption utility framework for what I'll call "very very long retirement:" 

• On the Behavior of Long Horizon Adaptive Spending
• Long Horizon Spending (con't.)

This post, however, is all about me. Dang, I feel like an Instagram model when I say that ;-) but thankfully you will not be subjected to 10,000 pictures of nothing but selfies of me in a bikini. We'll leave that to our nightmares. What I will do is adapt my software -- this is, I think, the fourth consumption utility sim I've written [1] -- to my own personal parameters to see where it goes with my data in the context of what I have done before in at least the last couple of posts. Again, no charts, just some basic spend rates if I can get away with it. 

The adaptation I am thinking about includes the following assumptions: 

• Start at age 63 rather than 50 or 70. I turn 64 soon but let's not rush things
• Use a longevity distribution more aligned with an actuarial table for a long lived me
• Look at both constant and adaptive (%P) spend methods where constant is new here
• Look at the parameters from the last post and N[4/12] in particular, for reasons....
• Use a risk aversion coefficient = 2. That is "my RA" but only because it seems to pop up as close to something that works for me in the past. I have no real other rationale for "2" which one has to recognize as a bit bogus both here and in general. 
• I'll throw in my rule of thumb I call RH40 for good measure
• 50-150k iterations, depending...
• The actuarial sum(tPx) for x=63 and m/b = 90/9.5 means T(x)~24, so let's call it about age 87 for my mean age of demise. AAcalc.com puts the mean at 24.26 for my age (and a healthy cohort) so I don't feel like a total dork with this calculation.   

Otherwise in this post we can borrow wholesale from the assumptions in the two links above without any effort to document them here. Yes, you might have to go back and read... We are still using the value function from before.  I used to hate this utility framing but it seems to be useful the longer I work with it. Keep in mind I have been very skeptical of utility functions. Otoh, the convexity helps in evaluating jumps and variance in spending, esp when it is to the south...

When I go thru all my paces, this is what I get

Hmpf... So what is that? Looks like low threes for me right now if I am hyper conservative but up to low 6s if I keep a weather eye on my portfolio while bobbing and weaving with the financial weather year by year...with, no doubt, a totally different answer coming up next year, TBD. Right now I am leaning towards the conservative side of this. Inflation and recession, in theory, shouldn't change things if we assume: a) mean reversion in inflation processes, and b) that the recession that might come is just a recession (i.e. part of the planned variance) and not some kind of a "regime change" where all of our assumptions coming out of the 20th century might fall like a flimsy line of weak dominoes. 

-------- notes ----------------------------------------

[1] Here is what happens when I code. I write some random code that is small and tractable and useful. Then I wonder "I wonder what X would look like..."  So I add some more code. Then, 6 or 10 posts later, after I have added all sorts of crap, I have something that even I can't even read (and I comment a lot, a habit from my tech days), that breaks every time I run it, and is extremely fragile and hard to manage... and slow. It is sometimes like an old house: often a bulldozer can be the best medicine. Which is what I do for the next round.