On AQR's paper on retirement security in a low expected return world

Here an AQR 2017 paper on Risk and Retirement. 

Intelligent Risk Taking: How to Secure Retirement in a Low ExpectedReturn World, Ilmanen & Rauseo Aug 2017

Read that, then my comments are as follows. This was copy/pasted from some correspondence with DC


-----------------------------

Here are some superficial comments on the AQR paper

1. I've mostly steered clear in the blog on stuff related to portfolio optimization and design and related component analysis.  You can see this in the "5 Processes" paper I did. An optimized portfolio is "assumed" as input into my quant schema.  That kind of optimization stuff is the bread and butter of the finance industry and I don't add much as an amateur.  It is also something that is generally done before decumulation is considered (though it shouldn't be) and that's what these guys are doing at AQR. They explicitly foreswear things like decumulation and uncertain longevity somewhere in the intro.  They are instead working in the "AQR zone," which is my second point...

2. They are totally shilling for AQR. That's cool, I love AQR and they should shill.  The conclusions of the paper point directly to "buy our funds."  And a lot of retirees probably should buy AQR depending on their plan.  Doesn't mean we shouldn't guess that there is some marketing going on here.  

3. They are 100% focused on "same vol higher return."  That's only one side of vectoring an efficient frontier up and to the left.  I have this theory, that I can prove for no one but me, that before retirement I want my "EF shift" to be up, and after retirement I want it to be left, assuming "r" is adequate to the task. That vol reduction in decumulation is gold in terms of things like sequence risk and portfolio longevity with consumption present. I get why during accumulation one might make a fetish of "return for same vol" but things change afterwards. No W2-analyst or tenured academic is going to understand this in a skin-in-the-game way like I do, this feeling of the difference after the safety-net of work is gone. This sensibility is the missing link in 99% of papers I read.  M Zwecher got it in his book. 

So if I were to be simplistic, 

Before retirement:

- bias, sorta, to single period analysis and MV optimality. MPT dominates conversation

- emphasis on savings rates and the terminal accumulation value of a portfolio at horizon

- focus on returns and net growth rates and terminal value of wealth at horizon

- consumption is often downplayed or ignored and idea of consumption floors are almost unheard of

After Retirement:

- bias towards multiperiod effects, e.g., geometric mean outcomes and median terminal wealth or, say, fail rates etc, and...

- emphasis is really more on portfolio longevity and lifetime consumption utility than it is on terminal wealth (except for issues related to bequest)

- focus here is maybe more on volatility ( vs returns) as well as sequence risk

- consumption should dominate the conversation and the concept of income flooring starts to make more sense

AQR is pandering to the first of these two world's. Makes complete sense. A ton of their audience is institutional. That means fund design and fund management. Almost all of their audience knows squat about spending or annuities and the impact on retirement in decumulation.  Issues related spending and life income over a lifecycle are going to not just dominate fund-design and optimization tricks but are going to crush it.  Except at the margins, where it matters a bit.  

4. The case in the paper for alts and Risk Parity always comes back (they repeated this at least twice) to "averting capitulation and forced sales," not any inherent supremacy of the portfolio statistics over time as such though that discussion might be interesting. So, tongue in cheek here: maybe there is a fund that can be 60/40 instead of, say, Risk Parity, and then part of your 70 bps fund fee or maybe a little spiff on top of that (instead of the 1.25 one might see in an alt mutual) includes mental health and behavioral counseling.  Might be cheaper and more efficient than a RP fund...  But I guess that is what advisors are supposed to do.