Ken Steiner on Actuarial Balance Sheets

Ken Steiner just put out a new post (Yes, Determining How Much You Can Afford to Spend in Retirement is More Difficult than Saving for Retirement. April 26 2019) at http://howmuchcaniaffordtospendinretirement.blogspot.com

Not only do I think this is the correct method, and not only do I think this conveys the correct tone and sensibility and skepticism, this is also almost* exactly how I manage my own process.

"almost*" means two things because I enhance the ABB with a couple things:


1. I hitch feasibility (his ABB calc) to sustainability just to double check the route.

Ken's ABB is the foundation of good retirement practice. It is no doubt one of those sine qua non things.  I extend it, and Ken points some of this out too so I am not straying too far from the farm here, because of some soft simulations I did recently -- where soft means running a bunch of deterministic "scenarios" -- to test what happens if one were to coerce one's spending so that PV(spending) = PV (net assets) every year (again I am not saying Ken suggests this, this is just something I did).  In a few scenarios some of the late year spending in this coercion regime get's high enough that it impinges on the utility of lifetime consumption due to the risk of depleting wealth when there are some late age net-wealth-process cascade-crashes.  Basically spending, even when feasible, can still fly too close to the sun.

My take away was that there were a couple ways to mitigate the full life-cycle risk: a) create a "reasonable" policy cap on spending because there is no need to spend just because a formula says so (and probably a floor too but most people have a natural floor below which it is hard to spend less), and b) monitor prospective risk via forward simulation of some type and then keep spending below some policy-based monitored-risk cap in any given year.  These two are effectively the same and would create a "snake in a tunnel" effect that is likely, but not proven here, to create more optimal lifecycle consumption. Not "optimal" just maybe a little bit more than the baseline.  This is close to a spend-rule type setup except that we are in a more fluid and econ-rational framework, in my opinion.  If one does not happen to have an eye on macro-econ lifecycle utility (and really, who does?), this kind of thinking might be overkill which on most days I think it is. But I do pair feasibility with sustainability myself, and the pension guys that aren't doing that kind of thing probably should be.

2. I use a random lifetime probability weighting rather than a fixed time-frame.

Because I have been steeping in my own tea for so long I use, instead of fixed horizons, a random lifetime probability weighting or alternatively called a conditional survival probability (CSP).  Rather than 30 years I run my discounted cash flow plan to infinity and then weight it by the CSP which approaches zero around age 120. This is the same as doing a simulation where on the first life I draw (based on an actuarial table or formula) a 35 year life, on the 2nd I draw a 28 year life, on the 3rd I draw a 3 year life and so on....  The CSP gets me to the same place without the life-sim and, frankly, it probably gets me to the fixed 30 year answer, too. Ken is right. No normal retiree would do this kind of thing but four years of RH blog posts should prove to you that I am not normal. The advantage to me in this method is that I am more comfortable knowing that I have factored in the full term structure of longevity. I also know how to nudge the curve to shorter and longer lived stats to test things out.

Conclusion

I am RH and I endorse Ken's message.