Using PE10 to limit withdrawal rate

Non-investing personal finance issues including insurance, credit, real estate, taxes, employment and legal issues such as trusts and wills
Post Reply
sengsational
Posts: 19
Joined: Sat Feb 05, 2011 8:48 pm

Using PE10 to limit withdrawal rate

Post by sengsational » Mon Feb 13, 2017 6:05 pm

I've been toying with the idea to use the retiree's current investment balance, the equities allocation percent, and also the current market's PE10 / CAPE, all to calculate a withdrawal rate. The idea is that if the market is in the "historically high" range, and you're heavy in equities, then some of that money is likely to be taken back, and maybe can't support a withdrawal rate as well as if the market wasn't so far above historical averages.

So let's say you determine an optimal spend based on a flat 6% gain in equities, 3% bonds, and 2.5% inflation (a-la i-orp). We all know that sequence of returns will play havoc with that, but let's take that optimal spend amount as a starting point. Now, say you have 35 years left in your plan and the PE10 is 28, so 75% over it's historical average. Based on discussions concerning the negative effect of starting your retirement in a market down-turn, I'd say it would be wise not to spend all of the optimal spend amount. With 10 years left in your plan, probably not as much of a concern.

There is a draft paper on the i-orp site that addresses retirees' annual visit to i-orp, then using that result to determine how much they can spend for the year (the "3-PEAT" process). That paper uses a historical data approach, much like the back testing in the VPW spreadsheet does. But when I looked at the "DI" disposable income (spending) results in the paper, it looked very traumatic. Maybe not traumatic for someone who has a big discretionary budget, but for a retiree playing it closer to the edge, there are annual budget cuts in that paper that would send them to bankruptcy.

All of that got me thinking about how to use what we know (PE10) to smooth out the annual spending by limiting spending in years with a "hot" market.

There is a tool for Android that I wrote which allows you to exercise the 3-PEAT process automatically. You put in your typical i-orp inputs, but you also include an historical starting year. Then your portfolio is subject to whatever market forces were in effect for that year, and follow-on years, for the duration of your plan. The tool allows you to run with spending limit 'on' or 'off'.

So far, I've concentrated on simulating starting retirement in 1999, and what I have concluded is that if you run with a spend limit:
    you can increase your overall plan spending by a few percent,
    the early withdrawals are significantly less (20%-25%) than that proposed by the steady growth 6%/3% i-orp assumptions,
    the budget cuts in down markets are less severe,
    but it does take 20 years of a 40 year plan to break even.

There is an Android app that's currently in beta testing that drives the 3-PEAT process, so if you've got an Android device, you can try it: ThreePeat Beta on GooglePlay

User avatar
LadyGeek
Site Admin
Posts: 41607
Joined: Sat Dec 20, 2008 5:34 pm
Location: Philadelphia
Contact:

Re: Using PE10 to limit withdrawal rate

Post by LadyGeek » Mon Feb 13, 2017 8:27 pm

This thread is now in the Personal Finance (Not Investing) forum (withdrawal rate).
To some, the glass is half full. To others, the glass is half empty. To an engineer, it's twice the size it needs to be.

AlohaJoe
Posts: 2633
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Using PE10 to limit withdrawal rate

Post by AlohaJoe » Mon Feb 13, 2017 9:44 pm

This is the 3 PEAT paper: https://www.i-orp.com/3-PEAT/3-PEAT.pdf

AlohaJoe
Posts: 2633
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Using PE10 to limit withdrawal rate

Post by AlohaJoe » Mon Feb 13, 2017 10:15 pm

I haven't looked at your app -- so this feedback may be off base, if so I'm happy to be wrong :happy

Most CAPE10 things make two relatively large errors: they are historically anachronistic and they don't mimic your actual asset allocation.

For historically anachronistic, I mean: the "historical average CAPE10" has varied between 14.01 and 17.28. In 1885 the CAPE10 was 16.3. Is that above or below the "average"? If you use the average from 1871-1885 then it was above average. But if you use the average from 1871-2016 then it was below average. Of course, a person in 1885 had no idea that eventually the average would end up where it did. Using future historical information is cheating and makes CAPE10 look better than it actually is.

They also tend to ignore your actual asset allocation. I have 16% of my portfolio in Emerging Markets. Are they undervalued or overvalued according to CAPE10? How much does that affect the withdrawal rate? REITs, small-cap value, momentum, international, emerging market bonds ... once you get into the Valuation Game you really have to provide a valuation for the entire, actual portfolio. Most things using CAPE10 assume that I have 100% of my equities in US large cap stocks.

As for the 3-PEAT approach...the main problem I have is that this approach, like that of Gordon Irlam's aacalc.com (which uses dynamic programming instead of linear programming), is that they essentially have a black box "and then magic (with a lot of math) happens" at the heart of them. Obviously that's an exaggeration but I reckon that's how 99% of people view them. They don't have the same transparency or ease of understanding as less complex methods.

I think they have a lot of pros and I hope they continue to develop and influence research but most people struggle to understand Monte Carlo and sequence of returns.

heyyou
Posts: 2774
Joined: Tue Feb 20, 2007 4:58 pm

Re: Using PE10 to limit withdrawal rate

Post by heyyou » Tue Feb 14, 2017 1:41 am

I've been toying with the idea to use the retiree's current investment balance, the equities allocation percent, and also the current market's PE10/CAPE, all to calculate a withdrawal rate.

So has Michael McClung in Living Off Your Money.

From several valuation methods, he divided their results into quartiles, numbered them, then averaged today's valuation quartile numbers of foreign and domestic markets, to tailor the starting WD % to avoid WDing at too high of a starting number. Future WDs are based on longevity and recent annual portfolio size (like VPW and the RMD method) with the variations buffered by a floor and ceiling similar to VG's suggested spending method. Inflation is monitored relative to retirement day asset value to avoid the early overspending of sequence risk.

Build your own method, but you can learn some techniques from others.

sengsational
Posts: 19
Joined: Sat Feb 05, 2011 8:48 pm

Re: Using PE10 to limit withdrawal rate

Post by sengsational » Tue Feb 14, 2017 6:15 pm

AlohaJoe wrote:Most CAPE10 things make two relatively large errors: they are historically anachronistic and they don't mimic your actual asset allocation.
I'm not worried about the average PE10. If the current PE is near the historical average, nothing happens to limit withdrawals. And the average has been pretty stable. Yeah, the first few years, kinda wild, but if you call historical average 15 or 16, and today's PE10 is 28, what you pick for the average isn't a sensitive input.
Image

AlohaJoe wrote:They also tend to ignore your actual asset allocation. I have 16% of my portfolio in Emerging Markets. Are they undervalued or overvalued according to CAPE10?
This is true: other asset classes beyond US equities are not perfectly correlated. Honestly, I wish more asset classes were not correlated at all. I think we must presume that anyone using this has a significant amount of US equities and that the other asset classes are correlated "enough" not to torpedo the whole thing.

AlohaJoe wrote:As for the 3-PEAT approach...the main problem I have is that this approach, like that of Gordon Irlam's aacalc.com (which uses dynamic programming instead of linear programming), is that they essentially have a black box "and then magic (with a lot of math) happens" at the heart of them.
I've heard that critique of many retirement planners, and in fact it's why I don't use many retirement planners that simply take inputs and generate a recommendation. With i-orp, I find that the "black box" argument is, in fact, much minimized. I'm able to recreate all of the math of i-orp in a spreadsheet. It tells you what the balances are every year, tells you what you withdrew and what you transferred. It applies a fixed growth rate and that becomes the next year in the model. Very detailed as compared to most other brokerage house supplied planners, where they just give you a number at the end.

AlohaJoe wrote:I think they have a lot of pros and I hope they continue to develop and influence research but most people struggle to understand Monte Carlo and sequence of returns.
Putting in your current balances and then saying "what if we experienced markets like those starting in 1929?" Then you see happy or sad faces. Pretty easy to get at least a bit of an intuitive feel.
Image

sengsational
Posts: 19
Joined: Sat Feb 05, 2011 8:48 pm

Re: Using PE10 to limit withdrawal rate

Post by sengsational » Tue Feb 14, 2017 6:17 pm

heyyou wrote:So has Michael McClung in Living Off Your Money.
Thanks for the pointer to that. I just defined a function, completely out of thin air, but that looked like it smoothed the bumps. I never even searched for someone who was there before me (although I knew there had to be).

AlohaJoe
Posts: 2633
Joined: Mon Nov 26, 2007 2:00 pm
Location: Saigon, Vietnam

Re: Using PE10 to limit withdrawal rate

Post by AlohaJoe » Tue Feb 14, 2017 7:54 pm

sengsational wrote:With i-orp, I find that the "black box" argument is, in fact, much minimized. I'm able to recreate all of the math of i-orp in a spreadsheet.


You implemented the RIMO linear-programming solver algorithm in a spreadsheet? Did you use the simplex or the criss-cross algorithm? Both seem like they would challenging to implement in a spreadsheet.

User avatar
grayfox
Posts: 4614
Joined: Sat Sep 15, 2007 4:30 am
Location: Earth

Re: Using PE10 to limit withdrawal rate

Post by grayfox » Wed Feb 15, 2017 1:14 pm

sengsational wrote:I've been toying with the idea to use the retiree's current investment balance, the equities allocation percent, and also the current market's PE10 / CAPE, all to calculate a withdrawal rate. The idea is that if the market is in the "historically high" range, and you're heavy in equities, then some of that money is likely to be taken back, and maybe can't support a withdrawal rate as well as if the market wasn't so far above historical averages.



What I figured out years ago is that taking 80% of E10 should be sustainable forever.

Example: P/E10 = 29.03 and P = 2337.58 -> E10 = $80.52. Withdraw = .8 * 80.52 = $64.42. (2.75%)
(or more simply 80% x E10/P)

What happens is, even when valuations change, the withdrawal amount only changes if the smoothed earnings changes. This is unlike other methods that base withdrawal on the current market value.
Gott mit uns.

Post Reply