1. I'm not totally sure this advances, in any significant way, any one real person's (even my own) real retirement, and
2. Everyone and their brother -- or at least among leading personal finance researchers like Milevsky, Pfau, Blanchett, Kitces, Dirk Cotton, and a bunch of others -- have all weighed in at one point or another over the last couple years on the weaknesses and misdirects and emptiness of retirement simulators, a point of view I happen to agree with.
But...I just felt like playing around with this one more time. In this particular case I wanted to layer in realistic probabilities for longevity into a working retirement sim and see how it goes. I had that last year, too, but I did not have a good platform on which to play around with it. Now I do.
The new platform, however, does not change the fact that my output, like all sim output, still tells no one what, really, to do with itself...but it has been fun to play with. So, my goal "might" be to try to come up with a probability- and magnitude- weighted fail rate that is a little more realistic than a run-of-the-mill estimate. I say might because I don't have that yet. So this is still play. What I do have is a generalized limited-assumption simulator you might recognize as similar to what's out there for basic tools but with a dose of terminal age variance thrown in. So far, in prototype mode, this is what it looks like:
Here is what we have so far:
Assumptions for the above:
- standard...65yo, 1M endowment, 60/40 risk, 4% inflated spend, fees, taxes, etc. etc.
- 30,000 sim runs done as 30 x 1000 samples
Resulting Fail Rate:
- 9.1%. Don't ask me what to do with that number.
Lower Left: Terminal Wealth vs. Terminal Age with terminal age frequency overlaid
- each point is the combination of a terminal age and an estimated terminal wealth
- terminal age is varied within a non-normal realistic mortality distribution
- the frequency diagram of terminal age is overlaid on top to give sense of age probabilities
- negative wealth is meaningless and is best ignored or viewed as a type of fail magnitude
- There are some regression and smoothing lines thrown in there too for good measure
Lower Right: Heat Map
- this is same scatter but now with densities and contours (isn't R fun?)
- the densities are frequency-weighted age and terminal wealth zones
- interpretation? TBD but the core heat zone is the set of higher probability outcomes
- if the red dips below zero before age 90 there is probably a problem
Upper Right
- sampling distribution of the sample mean of the 30 samples of 1000
- the scatter charts use the full 30k dataset, btw.
Upper Left
- R console with one tiny output of a fail rate
Now return to point 1 and 2 above.
----------------------------------------------------
End comments:
- I did this sim stuff originally only because a major US bank wanted to charge me thousands for a financial plan that happened to be 98% a simulation run. That ticked me off so I just wrote one.
- In this particular instance above, I was less interested in the simulator than I was in R skills so I used the excuse of the sim to learn R rather the opposite: the idea that I learned R only to build a sim. I have to say that was a good methodology for self-training in tech skills. You can read about it all day long but trying to construct something that actually works is a real test.
No comments:
Post a Comment