Sequence of returns risk and early retirement

[Lucky C]

@Mister Imperceptible
For clarification I am not convinced that returns will equal zero, but I am fairly certain that the average expected return will be near zero (returns could have a high probability of being slightly positive with a smaller probability of being very negative, with the mean of the distribution being around zero). I'm in mostly cash, with a small % of buy and hold stocks and bonds, and another small % in a momentum/trendfollowing strategy (not going all in on one strategy). I also have a small % in LendingClub that I haven't been reinvesting in due to expected increase in defaults and hopefully higher rates in bond markets in the future. I'm 0% in gold because I believe the expected real return is close to zero. I will potentially be able to quit in the 2nd half of this year even if my portfolio returns are 0% so I don't want to be exposed to potentially a really bad year of returns. However I'm not sure if I will know what to do / invest in if the markets are not giving clear signals later this year (cross that bridge when I get to it).


Just want to add that today's starting valuations are even more important to consider if you're planning an extended span of time (>30 years) over which you hope your portfolio will survive. I have a good chance of living >50 years after I quit my current career since I live a healthy lifestyle.

Starting with the default cfiresim settings (4% withdrawal, 75% stocks / 25% bonds) and then changing the retirement length to 50 years results in the following failures, listed by starting year and starting CAPE:
1899: CAPE 22.9
1902: CAPE 22.3
1905: CAPE 18.4
1906: CAPE 20.1
1907: CAPE 17.2
1909: CAPE 14.8
1910: CAPE 14.5
1911: CAPE 14.0
1912: CAPE 13.8
1929: CAPE 27.1
1937: CAPE 21.6
1962: CAPE 21.2
1963: CAPE 19.3
1964: CAPE 21.6
1965: CAPE 23.3
1966: CAPE 24.1
1967: CAPE 20.4 (last year backtested)

The highest starting CAPE when a 4% withdrawal worked for 50 years was 22.3 in 1930. That is the only year where your portfolio would have survived starting with a CAPE > 20. Plus bond yields have recently been coming up off historical lows whereas the 1930s saw bond yields starting higher and then falling. Also 1930s dividends were more than double today's S&P500 dividend yield, and the 1930s overall were deflationary whereas many now expect higher inflation in the U.S. Currently CAPE is about 45% higher than this 1930 best-case high CAPE survival scenario.

On the other hand, if starting CAPE < 13.8 (15% below historical median), the portfolio has never failed a 4% withdrawal over 50 years. Seems to me that when you want to be concerned about sequence of returns risk is if valuations are somewhere in the middle (CAPE around 14 - 22) since returns could swing either way. When CAPE is extremely low or extremely high, you can have high confidence that the next decade or so of returns are going to be very good or very bad, respectively.

My solution is similar to Jacob's in that I am hoping to quit when valuations are lower, or keep on saving to have a lower starting withdrawal % if valuations are staying high for a while longer; my wife will probably be working at least part time which should cover all our basic expenses; I will be working on hobbies and projects that will reduce spending further and possibly generate income; I will not be a buy and hold investor and will be able to adapt as the market changes; I will not be opposed to picking up low-paying part time or seasonal jobs. All of these things considered, I will probably have worked longer than I needed to, but I'm OK with that.

[IlliniDave]

My solution is similar to Jacob's in that I am hoping to quit when valuations are lower, or keep on saving to have a lower starting withdrawal % if valuations are staying high for a while longer; my wife will probably be working at least part time which should cover all our basic expenses; I will be working on hobbies and projects that will reduce spending further and possibly generate income; I will not be a buy and hold investor and will be able to adapt as the market changes; I will not be opposed to picking up low-paying part time or seasonal jobs. All of these things considered, I will probably have worked longer than I needed to, but I'm OK with that.

We differ a lot on the tactics but the main thing, which we appear to agree on, is that having flexibility is a person's most effective weapon. I don't want to have to be flexible, so I went straight for the supersizing the stash option. The nice thing about that approach is that it doesn't preclude any of the other mitigation actions. But I am in an atypical situation where I got the ER bug during my peak earning years after having earned some non-trivial retirement bene's as a long-term employee of a single employer, and had a decent start on amassing a stash, so the cost in years (given I can't do anything about the late start) to supersize is minimal. The thing about SoR risk is that it is a temporary thing* confined to a specific window of the retirement horizon that with some forethought can be mitigated and the "damage" minimized (if it manifests at all). 30 years is probably a fair assumption for my horizon so if I can get through the first decade without the wheels coming off I should be good.

*Temporary as it is typically defined which includes inherent assumptions about investing approaches. For people that avoid the assumed investing models I don't know what can be said. Maybe it doesn't exist at all or is somewhat different in character.

[Tyler9000]

Here is @Tyler9000 promising a nice feature dealing with pre-FIRE and post-FIRE portfolios (and perhaps the transition stage) which I don't think ever came about? Or maybe it did, but one still needs multiple tabs with the output of the first (accumulation phase: goal to quick working ASAP) feeding into the second (retirement phase: goal to not run out of money before you die content).

For every polished calculator I publicly release I have a folder full of failed experiments. A glide path calculator just hasn't met my standards for helpfulness yet, but that doesn't mean I'm not still tinkering. Inspiration is a fickle thing.

[jacob]

The main reason I'm not a fan of MC for periods over, say, 50 years, is that the sample distribution converges on the population distribution; that the population distribution decreases in size; and that there are reasons why the 21st century economy would not mimic the 20th century. Lemme splain.

Suppose the data set is year 1910-2010 (just to make things easy and numbers round). If we're looking at a 30 year period, then there are 3.33 unique such periods in the set. Say 1910-1943, 1944-1977, and 1977-2010. In the MC, there would also be 1976-2009 and so on, but the point here is that 1976-2009 and 1977-2010 captures almost all of the same macro-events of which there are few. People usually look at world wars, bubbles, crashes, and depressions.

Therefore in terms of capturing big events, a 30 year horizon gives us 3-4 different "worlds" to compare. Of course, if you have to start the dataset in 1972, then you only have one "world".

The problem then is that for very young retirement, you'll be looking at sample-intervals of 50-80 years. This means that you're practically sampling the entire population of years with a few fuzzy data points at the end. E.g. Say you're doing 80 years, so 1910-1990, 1911-1991, ... , 1930-2010. That gives you 20 runs ... and the only difference between the earlier ones at the latter ones is the existence of WWI and the Dotcom bubble. IOW, you've sampled about "one" possible world, so while it looks like statistics, it isn't really. It's like zooming out and taking pictures of the same thing at slightly different angles.

This is why I hesitate to trust MC for intervals comparable to the dataset itself.

Will the 21st century look like the 20th? I don't think so ... if nothing else, doing MC in other countries in the 20th century shows different results. The US gives one of the best numbers for the 20th, so if you believe that these hold going forward, you have to believe that the sampling by country is not random and that the US was special and will remain special.

In backtesting one would withhold part of the dataset for verification. For example, in testing 30 year periods on the dataset of 1910-2010, one might use 1910-1970 for the fit .. and then verify if the fit still worked on 1970-2010. The problem with this is that this leaves even less "worlds" for the fit.

That's why I consider MC to be more of a sanity-check than a guideline. In particular, I trust fundamental considerations (similar to Lucky C) more than a data-driven approach because in my opinion, there's not enough data.

[bryan]

@tyler9000, totally understand! I figured you're just too content with the retired life to be bothered

@jacob, can't you somewhat handle/kludge a longer interval by taking a 30 year MC model, duplicate it, and put the two in series to give you a 60 year? I agree with you, though, that it's best used as a sanity-check as opposed to a predictive tool.

[Laura Ingalls]

I have read quite a bit of the sequence of risk and glide path information out there but struggle with a couple ideas.

First, the idea of what is the “early” portion of an individuals retirement. Is it strictly the beginning X years or months regardless of the age of the retiree? Should it be based on a percentage of the likely years of life expectancy left? Every morning I am in the early stage of the rest of my life

Second, equities in the accumulation phase got me to retirement phase. I feel disinclined to tinker too much in the retirement phase especially since my willingness and ability to still (do some) work for money is still intact. The glide path thing seems more for people less risk averse and who frankly need or want to spend more money. Except for the last couple of weeks all a glide path strategy would have yielded was less net worth.

I would welcome others takes on it as it relates to either my or your real life.

[IlliniDave]

Typically in the context of SoR the window is discussed as the first 5-10 years of a 30-year retirement period, with the sooner after the retirement date the "crash", the more adverse the impact. Since I'll consider myself lucky to still be around 30 years after I retire, I haven't looked into the implications of SoR to 40- or 50-year horizons. It could be that the window of sensitivity is wider, mitigated perhaps when an early retiree accumulates a larger multiple of withdrawal requirements.

"Glide path" is really just an informal term for incremental changing of one's asset allocation rather than making a substantial change in one move. The conventional wisdom is retirees hold a lower portion of stocks to lower their volatility risk, and in general as stock allocation decreases the expected future spending decreases (modestly so). So a glide path to lower equities on the eve of retirement is viewed as a risk reduction move that comes along with a median expectation of lower terminal wealth and/or less (potential) lifetime spending.

I'm not able to comment on how it relates to you in real life. I can comment on how it relates to me (off the top of my head, I'm the only person here that I remember using the term regularly on the forum). I've changed a lot from the brash young accumulator I was in decades past. At a point preserving what I have became a roughly equal consideration to accumulating more.

I'm a little late to the ER party so if things implode on me 20 years into it I may no longer have the ability (or opportunity) to go back to work. So I can't be dismissive about the impact of running out of money. At the same time I have some fairly ambitious legacy goals, so stuffing it all in the mattress doesn't sit well with me either (and it introduces different types of risk). So I'm backing away from 100% stocks moderately. I've also learned that with the end of a regular paycheck looming, watching my account balance go down is a little more stressful than it used to be which is arguably a more important consideration than the math.

The math in the simplest sense is all about ratios of assets to spending, and I am admittedly rather sloppy in my spending. That affects the size of the retirement fund I strive for, but it doesn't directly contribute to what percentage I'm comfortable having invested in stocks (though subconsciously, who knows).

For a sufficiently bold person whose dependence on financial assets is lowered due to having other options and the flexibility they provide, sticking with an all-stock portfolio is certainly an option. Same is true if the ratio of assets to withdrawals is very high (though I can't name a universal threshold). My opinion is that the main things are getting a good understanding of one's temperament as an investor, having some reasonable goals, then working in the risk/reward space to find a balance. I also think with longer anticipated retirements, one should consider both higher multiples and somewhat higher stock percentages, but that's mostly intuition talking--I don't know if there is research/data to back that up.

[Laura Ingalls]

@Dave thanks for the nice post. DH and I have been semi-retired for around 4 years. We definitely have more than when we started. I do feel like SOR is decreasing as time goes by as we are getting closer to launching the two kiddos and we slowly get closer to starting pension and social security. My bigger fear is that the health care tide will change in a way that is unfavorable to us.

[classical_Liberal]

...

[classical_Liberal]

...

[jacob]

It's likely that the "glide path" will involve some [very] different tax considerations, e.g. roth conversion ladders(*), much lower earned incomes, and/or portfolio shifts. I would recommend starting to dry-run the eventual setup before quitting the job to increase confidence that living off of investment income actually works in terms of practice and emotions, not just theoretically.

Thus glidepath concerns should include tax as well as new-lifestyle concerns.

(*) Those with average incomes (playing in hard mode) will likely have a lot stashed away in tax-deferred accounts and would somehow have to get it out.

[IlliniDave]

(*) Those with average incomes (playing in hard mode) will likely have a lot stashed away in tax-deferred accounts and would somehow have to get it out.

Those folks if in the US should make themselves very familiar with the so-called "72t" distribution rules, which have some onerous "gotchas" if improperly implemented. When it was looking like I might go that route I even considered retaining an accountant to get the SEPP schedule set up so as to avoid potential penalty taxes.

[TheFIminator]

As most of us rational people know, no one can really predict market cycles. Sure there are some signs at the moment that there could be a correction in the nearish future but no one knows when. If you were on the path to ER right now, you may well pull the trigger today and read the 'bloodbath' headlines in the papers tomorrow. That would definitely erode your capital especially if it was mainly in the Share markets. But if you had a mix of Shares (index or otherwise), cashflow real estate, cash & some international exposure, it is highly unlikely you would be bottomed out. You are more likely still to get income through the investments but you just won't be as paper rich as you are today. Does that actually matter much? Not really if your portfolio had some diversification.

Of course, if you did pull the trigger but maintained good relationships with your colleagues/employers (did not burn bridges), you could always go back to some kind of job to weather the storm. Give it 5-10 years, the correction/recession may just be a good story to tell younger investors. Hmm..I get this feeling of deja vu..where have we heard something similar before. Was it somewhere back in the..........2007-2008 era??

[TheFIminator]

My stance is, you can model as much as you like, life works in mysterious ways. If we get hit with a deeper crisis than the GFC or the Great Depression, financial models will be just that, models. So, I plan on the more conservative side with 3% SWR and 35x savings.

With all that said, if you are fairly able and not past 80, you can always find some work for a few years to compliment the income and rebuild if needed.