HootingSloth wrote: ↑Sat May 01, 2021 8:18 am
I thought Bill Sharpe had a (seemingly idiosyncratic) "lockbox" approach that more or less works by dividing the portfolio into different accounts for each year of consumption in the future. Into each lockbox, you place a certain amount of TIPS plus a certain amount of stocks/bonds in the risk portfolio (plus, in his ideal world, a certain amount of theoretical assets that you can't actually buy that he calls "m-shares" and that I haven't really tried to understand). You leave each lockbox alone until you get to the year it is assigned, and then you spend whatever ends up in the lockbox. It is described in chapter 15, here:

https://web.stanford.edu/~wfsharpe/RISMAT/

To me, the lockbox sounds like a weird LMP portfolio where each lockbox is similar to a ladder, but has the following ingredients:

1. Zero-coupon TIPS maturing in 9 years

2. World Bond/Stock Mutual Fund or ETF Shares to be sold in 9 years

3. m-Shares maturing in 9 years

#1 doesn't actually exist; there are no zero coupon TIPS.

I half fake it, though, by using alternating rungs of TIPS and STRIPS in a bond ladder.

#2 Exists

#3 Does not exist

Chapter 15 is probably the only chapter in the RISMAT series where I feel Sharpe jumped the shark between applicable theory and just pure mental masturbation.

I think he sort of admits this, as he says:

The conclusion is that absent the availability of suitable m-shares, it is impossible to obtain a portfolio optimal for an investor with a constant relative risk aversion marginal utility function with risk aversion that differs from that priced in the market portfolio. That said, it is possible

to choose a combination of TIPS and the market portfolio for one year, find the implied marginal utility function, then find combinations for subsequent years that are optimal for the same marginal utility.

He then gives some utility functions for approximating the lockbox contents with TIPS and WSB shares.

At this point, I don't see the point in trying make a model to determine how many WSB shares to allocate a 'lockbox', as opposed to just withdrawing from the Risk Portfolio at a future date according to needs.

It feels tortured, and not solving a genuine problem, given that m-shares don't exist.

He then ends the chapter with:

Absent the availability of m-shares, retirees can either need to choose retirement income strategies that provide payments with significantly different probability distributions of income at future dates, adopt an approach consistent with marginal utility functions of future income

that differ substantially, or select some combination of the two approaches. Moreover, given the fact that most retirees will receive fixed real payments from Social Security or some other sort of defined benefit plan, it will be important take into account all sources of income. The

remaining chapters explore some of these implications in detail for the construction of strategies for providing income with and without insuring against longevity risk. In both contexts, lockboxes can play a prominent role.

I my case, I will have an income strategy that provide payments with significantly different probability distributions at future dates:

LMP TIPS/STRIPS ladder (mixture of fixed real and nominal payments)

Social Security (fixed real payments)

Risk Portfolio (variable distribution)

One could also swap in an annuity en lieu of the LMP ladder.

Given that m-shares don't exist, without which lockboxes can't really be implemented as envisioned, I'm going to consider LMP + WSB RP as 'close enough'.

60% Global Market Stocks (VT,FM) | 38% Global Market Bonds | 2% Alts || LMP TIPS/STRIPS || RSU + ESPP