Regime Change

 After running through a few posts lately, what do I have? Basically this:

1. Spend 2% if you want your $ to last forever though even a 2% spend could possibly flame out over a long enough horizon if you have a crap portfolio. Probably not a problem for mortals and spending less than 2% would be weird if the whole point is to be using the money for something. Spending more than 2 demands a little extra thought...
   
   
2. For immortals again, don't spend more than what the portfolio earns in the long run. This could be articulated as the financ-y expected geometric mean return or maybe alternatively some function of a more micro-econ-ish earnings ala Garland. Don't forget taxes and fees, though, while we are at it,
   
   
3. Throw in a little extra spend for a shorter, mortal, horizon because you are not going to live forever. In the last post I said Gordon Irlam framed a vaguely kinda sorta Mertonesque rule of thumb for spending as 1/longevity. In practice the spend and the portfolio choice are a joint decision. We showed that here once and others have as well, say Merton again, 
   
   
4. Throw in a little extra again if you have "enough" life income. Basically the life cycle model says you can spend a bunch early and run the portfolio literally to zero well before death if life income is relatively high. This all depends on the convexity of utility math and the weighting to zero of things that comes with declining survival probabilities. This LCM thing as it relates to retirement has some decent lit to back it up ranging amongst Yaari, Milevsky, LaChance, Leung and a bunch of others,
   
   
5. Throw in a little extra again for fun if you are unusually risk seeking. Seems like a gamble too far for me but whatever.

Fine, and I have specified very little here with any precision but I still wonder -- the point of this post -- sometimes:

• How do you know, ex ante, what you are going to earn in the long run? Because you ran some MPT thing using "historical data" and you just know now? Bullshit. Especially BS imo in 2022. Because you know multiplicativity and can calculate geometric returns? BS again. Did you have any idea what realized vol or return would be when you set this up? Did you even calculate it over N periods or did you calc it at infinity? Me? I'm convinced, for a variety of reasons not spoken here, that we are on the edge of a regime change in how the economy, taxes, politics, markets, society, etc will work and it probably won't look much like what was probably the most optimistic era in history, my lifetime: the late 20th century. This brings me to my second question...
  
  
• How does one even know if a process or system is still operating on the rules and assumptions that obtained when the process or system was designed and put into operation. Are the divergences part of the expectation or has something fundamentally changed? Does the fact that the process still looks vaguely like what one expected mean that it hasn't changed at all in some critical way not noticed yet?

No idea. I mean, plans are better than not planning but the plan, "optimal" or not is, or at least can be, mooted the first instant the clock ticks one increment. This is why monitoring loops are critical. Patrick Collins wrote this up in a retirement finance context and I still think that is my go-to document. I did as well in my 5-Processes document but who reads that anyway?  In the broader non ret-fin world there are plenty of methods that get at this, say: OODA loops or Deming cycles or adaptive AI or whatever.  Survival demands a continuous peek at what is going on in the system and in the world with which it interacts. I don't have a very good command of that domain space...yet...but it'd be good for me to chase down some more of this decision science and ops management stuff at some point.  I have a to-do to get the reading list from a psych prof in Georgia I follow that is going to run a course on this stuff. I'll pass it along when I see it.