New SWR backtest - 60 year drawdown, CAPE, COLA

Ask your investment, budget, and other money related questions here
Post Reply
oldbeyond
Posts: 338
Joined: Thu Nov 29, 2012 10:43 pm

New SWR backtest - 60 year drawdown, CAPE, COLA

Post by oldbeyond »

https://earlyretirementnow.com/2016/12/ ... t-1-intro/

This was already posted by Fish in the "Permanent Portfolio in this environment?"-thread(viewtopic.php?f=3&t=8619#p136963), but I thought it might interest a lot of people who skip PP-centered threads, too. I of course understand if mods disagree. I hope I didn't miss a existing thread about it when I did my due diligence.

Improvements upon the Trinity study and other include:

* Includes all monthly 30- and 60-year retirement dates since 1871 for a stock and bond portfolio(SP500, 10-year US Treasuries)
* 0-100% stocks in 25% increments
* WR between 3-5% in .25% increments
* Real final value between 0-100% in 25% increments
* WR by different CAPE regimes
* SS/pension payouts in old age
* A few flavours of COLA

There are several interesting conclusions, mainly that 4% isn't all that safe for longer retirement horizons or in the current CAPE regime, and that longer retirement horizons greatly favour higher equity allocations. His long term view on bonds is illuminating, showing long stretches of low or flat returns(see also https://earlyretirementnow.com/2016/05/ ... tock-risk/). A lot of simulators have data back to the 70s, certainly showing large bond losses, but which were then followed by the epic rally which we still live in. It's easy to expect too much from bonds, then.

There is no mention of max drawdowns and hard assets are not added to the mix, so rather a different outlook from the PP-folks.

stayhigh
Posts: 113
Joined: Sun Dec 06, 2015 4:20 pm

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by stayhigh »

All this calculators consider perfect world, where ERE person is absolutely passive, watching disappearing lifetime savings and having not a single dollar of income during 30/40/60 years. Just build a strong portfolio and stick to it, learn useful skills and relax, don't bother about every single calculation and fraction of SWR. No matter how good prepared you are, and how many predictions you've done, future will surprise you anyway.

Enjoy your freedom and financial independence, don't stress about it.

userqname
Posts: 27
Joined: Thu Dec 29, 2016 9:19 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by userqname »

@oldbeyond You left out the most useful takeaway from the series. Namely that he supports a 3%SWR for a 60y retirement. (Actually, a 3.25%SWR, and apparently using only US broad market equity-only index funds.) Very encouraging for indexers.

To the discussion I'll add:
https://personal.vanguard.com/pdf/s338.pdf

A four year old vanguard analysis that finds:
-CAPE and PE(ttw) are equivalent in predictive power
-CAPE and PE valuations are the only forecasting metrics that work
-CAPE and PE only explain about 40% of the returns in the subsequent 10 years
-a forward real return of >5% is the most likely outcome of current valuations, historically

Since, in 2017, we are halfway into Vanguard's forecast, we can see that they have been right so far.

simplex
Posts: 212
Joined: Sun Sep 04, 2011 9:28 pm
Location: NL

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by simplex »

The Vanguard paper is actually quite good, esp. the comparison of P/E1 and P/E10.

bryan
Posts: 1061
Joined: Sat Nov 29, 2014 2:01 am
Location: mostly Bay Area

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by bryan »

Very good stuff! Really great.

Still missing from the analysis is dynamic asset allocations. Nothing crazy.. but at least consider accumulation phase vs retirement. In reality it could be more complex (equity glide-paths pre/post-retirement, spending events, income events, life events, etc).

Tyler and I have mentioned it before over at MMM and here. It really is situational (80% SR is much different than 30% SR). The "obvious" solution is the lower your SR the more stock allocation you would benefit from (time in market before retirement is significant). However, there are a few nuances (risks) that get sussed out if you actually do some analysis for some more extreme cases (like ERE). Generally it's not a big deal.. but it gets swept under the rug a lot (and it's a peeve of those of us addicted to optimization!). Which is interesting since the accumulation phase is the one that most people are experiencing (and seeking help/guidance/education on, for the FIRE movement, at least).

The only analysis I've done on the matter is a sort of piecewise analysis I mentioned in the link. I'm eagerly awaiting someone else to provide an amazing calculator... with each passing day I will benefit from such a tool less :? :( I think Personal Capital retirement planner is the most useful in allowing you to add spending/income entries at different points in the future (but it lacks asset allocation tinkering)? I think you can bend cfiresim to your will, with some effort.

By the way, I'm surprised @tyler hasn't set up a patreon yet. I assume he knows he can't count on too many nickels/dimes from the ERE folks and the MMM folks are also frugal bastards..

Fish
Posts: 570
Joined: Sun Jun 12, 2016 9:09 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by Fish »

@oldbeyond Thanks for drawing more attention to the research and for getting the thread started!

For me, SWR research is "for entertainment purposes only" because the problem is weirdly formulated. In addition to the limitations @stayhigh mentioned, it relies on asset allocation(s) that have historically performed well and conservatism is added by choosing the worst retirement start date based on backtesting.

What I would like to see is an improvement on the Trinity-inspired methodology. To rely on SWR means questioning whether we've picked the correct AA (and whether it should be dynamic as @bryan suggested), and then looking at valuations and wondering whether we need to be even better than the worst-case backtest. Relaxing the constraint on earned income post-retirement would open up so much design space. Here is a simple example FIRE plan which makes use of the ERE ideal of low expenses and flexibility to earn income as needed:
Retire when invested assets exceed 30x expenses(*). Go back to work if this condition is not satisfied.
(*)Above age 55, substitute 30x with (70-age)*2. This assumes an old-age pension is available at age 70 that will fully cover expenses. The glidepath prevents overaccumulation and protects against a 50% drawdown. The idea is to ensure sufficient assets between ages 55-70 to avoid the need to work at an age where job prospects and earning potential are diminished.

A willingness to return to work fixes a lot of problems. Perfect asset allocation and market valuations don't matter because you will step in to prevent portfolio failure before it occurs. Developing flexible accumulation and withdrawal rules for FIRE seems like a more practical use of research effort.

userqname
Posts: 27
Joined: Thu Dec 29, 2016 9:19 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by userqname »

@Fish
Would you agree that ERE is a three-legged stool?

-a robust portfolio
-willingness/ability to work
-minimal dependence on cash

Each serves to cover the other two in event of failure.

@bryan
The simplifying assumptions of Trinity-inspired studies are a useful yardstick, since anyone with dynamic, alternative, or active investment strategies assumes they can beat the broad market indexes anyway.

bryan
Posts: 1061
Joined: Sat Nov 29, 2014 2:01 am
Location: mostly Bay Area

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by bryan »

It's useful as a starting point, sure.
userqname wrote:anyone with dynamic, alternative, or active investment strategies assumes they can beat the broad market indexes anyway.
Not quite. I mean dynamic in the sense of taking more individual, situational considerations into account e.g. "Hi, I have 30k in the S&P, make 60k, and want to buy a 150k house in the next year or so. Would like to retire with $1M before I am 40. Should I stay 100% stocks? Please advise." There are a lot of situations like that and it would be nice to at least account for the largest ones (e.g. accumulation phase as it approaches retirement versus retirement as it approaches death).

It's not about beating the market, it's about achieving certain outcomes w/ the highest probability.

7Wannabe5
Posts: 9372
Joined: Fri Oct 18, 2013 9:03 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by 7Wannabe5 »

Fish said: (*)Above age 55, substitute 30x with (70-age)*2. This assumes an old-age pension is available at age 70 that will fully cover expenses. The glidepath prevents overaccumulation and protects against a 50% drawdown. The idea is to ensure sufficient assets between ages 55-70 to avoid the need to work at an age where job prospects and earning potential are diminished.
Christie Brinkley just got paid $$$ to appear on cover of Sports Illustrated at age 63. My 76 year old super-wealthy friend still dresses up in a suit and goes off to run his business every day (after doing his hour long morning workout routine which includes headstands.) I've known two people who were still living independently and gardening and cooking every day in their mid-90s. I am studying robotics and just learned how a light-emitting-diode works at age 52 : ) If you pull up the stereotypical picture of some poor soul out looking for a new job at age 57, "does not have adequate funds in brokerage account" is not the only problem. In fact, if you imagine the same person landing on the cushion of an adequate brokerage account after being given notice of layoff from at-will employer, it is still not necessarily the case that the other problems are vanquished.

OTOH, I really enjoy these sort of simple metrics, so please carry on with your analysis and adjustments : )

Tyler9000
Posts: 1758
Joined: Fri Jun 01, 2012 11:45 pm

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by Tyler9000 »

bryan wrote: By the way, I'm surprised @tyler hasn't set up a patreon yet. I assume he knows he can't count on too many nickels/dimes from the ERE folks and the MMM folks are also frugal bastards..
LOL. Gotta know your audience. But on the flip side, the benefit of walking the ERE walk is that a few nickels and dimes go a pretty long way. ;) I'm looking into it.

oldbeyond
Posts: 338
Joined: Thu Nov 29, 2012 10:43 pm

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by oldbeyond »

Capital is a fragile thing over longer time horizons. That the past 150 years in the US presented an opportunity to withdraw 3.25% of the principal for 60 years without diminishing the capital stock was a surprise on the positive side for me and as userqname says a major takeaway from the study.

I would love to see more on the international situation(there's some here: https://www.advisorperspectives.com/art ... awal-rates) given the uniqueness of the US experience(from emerging market to global hegemon in a century) and the impact of hard assets over longer timeframes.

Looking at the larger picture, I'm happy to agree that the model "1. determine SWR=X% 2. draw down X%, forget about reality and golf until death" is flawed and fragile. It is quite popular though in some corners of the FIRE-net(MMM says 4%, burn the heretic!). What you want of course is to combine different types of resource flows to create a resilient whole, where financial capital may play a part, but won't be the whole story. Job/freelancing/business, hard capital(land, housing, solar panels, fruit trees etc), intellectual property, social capital, skills - et cetera. Even on here, I'd say financial capital is privileged(I regrettably tend to focus on it to the detriment of much else). Probably because money is fungible and requires little of us apart from abstaining from wasting it and the transfer of bytes between electronic accounts.

"Returning to work" is interesting as a mental model. But it comes with it's own constraints. Either we're talking about any job, or coming back to the old job. Given low expenses and that we're trying to supplement financial asset income in a market drawdown, it might not require many hours even in a crappy job, but I would understand why someone might rather save more trying to avoid vying for a position as a Walmart greeter in 2008. Going back to the old job might be hard in a downturn, or at all depending on the industry. There's decay in your skills and social capital within the industry, so a return 10 years in might be hard to achieve. My point is, if one is to supplement financial asset income with job income at critical junctures, designing that optionality into your career ahead of time is probably wise. Perhaps that would require you to set up a business when you retire and do a few projects a year to stay sharp. Perhaps you can use your skills in other ways(tutoring, mentoring, advising). Or perhaps you design a completely new job/hobby on the side that is scalable enough. Perhaps your financial capital is only a relatively modest cushion to handle dry seasons for the part-time job(s) that covers your expenses?

Lucky C
Posts: 755
Joined: Sat Apr 16, 2016 6:09 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by Lucky C »

We extrapolate past the current history and append equity and bond returns after September 2016. To this end, we assume long-term average returns for equities going forward (about 6.6% real p.a.).
Real returns of 6.6% starting in September 2016 (CAPE = 27) is pretty optimistic. With these perfect 6.6% real returns going into the future, the first graph shows close to 7% being a safe withdrawal rate starting in September 2016. I doubt that and would not take this extrapolated data (>1986 for 30Y SWR and >1956 for 60Y SWR) very seriously.

Better to do a Monte Carlo simulation of a mean reverting random returns, where the mean can be set to 6.6% real (if you are that optimistic about 21st century returns) but the starting value is X% above the mean that corresponds to the market level above the mean when you're starting your withdrawals. That would give you a higher probability of low returns in the first decade or so but eventually average to 6.6% long term, and would also generate sufficient volatility if you set up your randomness correctly. Run it 10,000 times and see what your success rate would be if starting out in 2017 withdrawing >4%. Hint: it should be <100%!

Without doing something like that, the authors are ignoring sequence of returns risk for years >1986 (or >1956 for 60Y prediction). They make some good points about how you should consider longer time horizons and recent bull market peaks, but then in the very first graph they still paint a rosy picture for anyone looking to retire during today's high valuations. At least later on in the Equity Valuation page they show that with today's CAPE a 4% withdrawal really has odds closer to 90% for 30 years, and about 70% success rate for 60 years. It would just be nice if their extrapolation matched what was expected based on valuations.

Dragline
Posts: 4436
Joined: Wed Aug 24, 2011 1:50 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by Dragline »

oldbeyond wrote:
Looking at the larger picture, I'm happy to agree that the model "1. determine SWR=X% 2. draw down X%, forget about reality and golf until death" is flawed and fragile. It is quite popular though in some corners of the FIRE-net(MMM says 4%, burn the heretic!). What you want of course is to combine different types of resource flows to create a resilient whole, where financial capital may play a part, but won't be the whole story. Job/freelancing/business, hard capital(land, housing, solar panels, fruit trees etc), intellectual property, social capital, skills - et cetera. Even on here, I'd say financial capital is privileged(I regrettably tend to focus on it to the detriment of much else). Probably because money is fungible and requires little of us apart from abstaining from wasting it and the transfer of bytes between electronic accounts.
Thanks for the article. What I always get out of these things is that you'd be better off just having a variable withdrawal rate (e.g., something like 4% of what you actually have at last year's end, not 4% of some fixed amount). While this will result in varied income year-to-year and you would have to adjust spending, mathematically you can never run out of money. I think Wade Pfau has written some articles with variations on this theme.

In addition, while viscerally attractive, the idea of a "set it and forget it" withdrawal plan is probably not a very good idea from the get-go just from the unpredictability of the future and potential "black swans", etc.

The other big take-away is what you observe above -- the world is not just composed of a choice between equities and treasury bonds, so these studies are premised on false dilemmas, and largely due to the availability of data and to make the math easy. Some diversification into other income sources and investments is wise.

The Old Man
Posts: 503
Joined: Sat Jun 30, 2012 5:55 pm

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by The Old Man »

The initial Trinity study was groundbreaking in that it actually came up with some numbers as well as an analysis to support developing a safe withdrawal rate. Before that, we had nothing.

The issue I have with subsequent studies, including this study, is that they do not explore WHY the portfolio failed. Generally, the portfolios failed when retirement began in the 1960’s and 1970’s. This was a period of increasing inflation. Stock/Bond portfolios, the typical portfolio used in the studies, are destroyed in such an environment. A better safe withdrawal rate will not solve the problem of an inflationary environment when using assets ill-suited to an inflationary environment.

A better approach would be to develop a portfolio construction and a safe withdrawal rate that could weather all economic environments.

wolf
Posts: 1102
Joined: Fri Jan 06, 2017 5:09 pm
Location: Germany

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by wolf »

General Snoopy wrote:A better approach would be to develop a portfolio construction and a safe withdrawal rate that could weather all economic environments.
What do you suggest as a portfolio construction? How is your asset allocation?

I'm interested in portfolio constructions a lot and studied various different approaches. In the end I ended with 50% equity of developed markets, 30% of emerging markets, 10% gold (mines), and 10% commodities. In addition to that I also have a real estate which I live in and social security.

When I think of it I want to look at the big picture of my total wealth not just the investment portfolio.

So...what kind of asset allocation do you think of if you think of one for all economic environments?

jacob
Site Admin
Posts: 15907
Joined: Fri Jun 28, 2013 8:38 pm
Location: USA, Zone 5b, Koppen Dfa, Elev. 620ft, Walkscore 77
Contact:

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by jacob »

The Permanent Portfolio was developed in response to the inflation problem. The question remains whether it covers all economic environments.

(Another response was the Alpha Strategy, but that was highly concentrated and would have been "interesting" to manage over the next 30 years.)

Dragline
Posts: 4436
Joined: Wed Aug 24, 2011 1:50 am

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by Dragline »

Some of the other "all environments" portfolios are the All Seasons, Coward's and Golden Butterfly:

https://portfoliocharts.com/portfolios/

I don't think you can have an all environment portfolio that does not include treasuries of some kind.

But I think its a good idea to have alternative sources of income such as rental property, some kind of small business or agricultural interest, peer-to-peer lending, etc. Specialty REITS and MLPs are also possibilities, but anything that is traded on a financial market is subject to market forces. Social security falls into this category -- its essentially an annuity.

You can also take more risk in your investments if you have a paid-off house or have otherwise minimized annual living expenses.

oldbeyond
Posts: 338
Joined: Thu Nov 29, 2012 10:43 pm

Re: New SWR backtest - 60 year drawdown, CAPE, COLA

Post by oldbeyond »

Gold and commodities helped in the 70's, but the low point for the WR in the study was caused by the great depression, where you would have wanted long term sovereign debt. In the study, higher equity allocations were favourable generally, but the question is what impact real assets would have had.

Post Reply