January 3, 2022 – Happy New Year, everybody. I hope you had a relaxing and healthy Christmas and a good start to the New Year!
Last month was the 5th anniversary of the Safe Withdrawal Rate Series! In December 2016, I published the first part of that series. I had material for maybe four or five parts but one thing led to another and with new ideas, most of them due to reader feedback, the series took off. It’s been running for 5 years and I obviously opened a bottle of bubbly last month to celebrate.
So, what’s the deal with the title then? Very simple: Blogging 101. You need a catchy title! I might have called the post “What I’ve learned in 5 years and 50 posts” or something along those lines. But to shake things up and get everybody’s attention, this is the title I went with. Think of this post as a natural extension of Part 26 “Ten things the “Makers” of the 4% Rule don’t want you to know” or the equally “tongue-in-cheek” posts “How to ‘Lie’ with Personal Finance” – Part 1 and Part 2.
So, after 5 years, 50 posts, what have I learned? What do I think others in the FIRE community are missing? What can you learn from my series that you may not have seen elsewhere? Let’s take a look…
1: We’re taking the “Simplicity Mantra” way too far!
We are all familiar with JL Collins’ work “The Simple Path to Wealth” (paid link) or Mr. Money Mustache’s post “The Shockingly Simple Math Behind Early Retirement“. I mostly agree with the “simplicity” part. I kept my retirement savings very simple, too. Invest in index funds. Automate your investments and take the emotions out and watch the money grow. If the market keeps rallying that’s great. But even if the market tanks like in 2007-2009, that’s not the end of the world either. Keep investing and make Dollar-Cost-Averaging work for you. Some of my best investments with the highest IRR are the contributions from March 2009, right around the market bottom! If this kind of simple index investment style is OK for someone with six letters (Ph.D., CFA) behind his name and years of experience in the asset management industry, it should work for the vast majority of FIRE enthusiasts.
But just to be sure, not everything is simple while saving for retirement! The more serious challenge during the accumulation phase is the psychological part. Staying on course, resisting the overconsumption and conspicuous consumption temptations, and not losing your nerves during volatile markets.
But in retirement, this all flips. The psychology of retirement is pretty simple. The math is more complicated than before as evidenced by my 50-part series. If there is a bear market early on, then the Dollar Cost Averaging effect that previously helped you in the accumulation phase will now hurt you because you are now selling shares at discounted prices and this will likely hamper your portfolio’s recovery. That’s the definition of Sequence Risk (see Part 14 and Part 15). But I often get the impression that a large part of the FIRE community misses this distinction and extrapolates the investing simplicity mantra into withdrawal math simplicity during retirement. After 50 posts and 5 years of my time dedicated to this issue I certainly don’t find the withdrawal phase simple!
Further readings: Part 27: Why is Retirement Harder than Saving for Retirement?
2: Passive investing? Don’t throw out the baby with the bathwater!
Related to the previous point, asset allocation considerations become more complicated in retirement. Even if you stay away from outright stock picking and use a simple index fund allocation, retirees should consider shifting the weights of those passive index funds over time. In other words, while accumulating assets, it’s not unheard of for people to just run with a 100% equity allocation (see Part 43 of the series). I did so while saving for retirement. But 100% equities may not work so well for the average retiree. A few years before retirement and, at the very latest, upon reaching retirement, you will have to make some decisions about shifting your asset allocation to take some risk off the table. And there is no generally accepted “passive index” for the Stock/Bond allocation. And much less the stocks vs. long-term bonds. vs short-term fixed income vs. commodities vs. real estate vs. crypto percentage allocations.
Even if you settle on one specific retirement asset allocation, it doesn’t mean that this allocation should be set in stone all the way to eternity. Quite the opposite, financial experts agree that Sequence of Return Risk can be slightly mitigated through an equity/bond glide path both before and after retirement. See Michael Kitces’ work on the “Bond Tent” and my work on pre-retirement and post-retirement glidepaths (Part 19, Part 20, Part 43 of this series).
And again, notice that this glidepath or bond tent over time has nothing to do with the often maligned and pooh-poohed stock picking or market timing strategies. It’s still an unconditional passive shift of asset weights, like in a target-date fund and the underlying assets may as well be index funds, so there is no stock picking either. I wouldn’t call this active investing.
Also getting lost in the whole simplicity/passive-investing craze is the opportunity to invest in some not-so-passive styles like real estate or trading options with very attractive return profiles and diversification potential and probably less sequence risk than your good old 60/40 or 75/25 portfolio or 100/0 portfolio.
3: The stock market is no random walk!
One of the most dangerous fallacies in the personal finance world and especially the FIRE community is the incessant claim that the stock market is a Random Walk. It’s not. The stock market is certainly hard and even impossible to predict on a day-to-day or month-to-month basis, but over longer horizons, returns are surprisingly highly correlated with equity valuations, like Prof. Robert Shiller’s CAPE ratio.
And here again, there is a stark contrast between young investors just starting out their FIRE path vs. people close to or in retirement. Expensive equity valuations shouldn’t pose too much of a headache for young investors. But retirement success is very correlated with the equity multiples at the start of retirement. In the chart below, I plot the inverse of the Shiller CAPE ratio on the x-axis (think of it as the earnings yield, i.e., low values indicate expensive equities) and the realized safe withdrawal of a 75/25 portfolio rate over a 30-year horizon on the y-axis. I simulate this for retirement cohorts starting withdrawals between 1926 to 1991. (and the last retirement cohort covers 1991 to 2021, which is the last year for which I have return data). All the failures of the 4% Rule occur when equities are expensive (low CAPE yield). And likewise, the 4% Rule is way too conservative if earnings yields are high. Because the stock market is no random walk!
The implication from this simulation exercise is that the unconditional failure probabilities like those in the Trinity Study or some of the online retirement calculators are meaningless if you currently find yourself in a high-CAPE world, i.e., a low earnings yield environment.
I always use the analogy of a researcher trying to calculate the probability of encountering a traffic jam. He studies the traffic patterns by leaving his home at the top of the hour each hour from midnight to 11pm. Out of the 24 attempts, he got into a traffic jam 6 times. Does that mean the probability of a traffic jam is 25%? If all the traffic problems always occur during rush hour, 7am, 8am, 9am and 3pm, 4pm, and 5pm, then the unconditional probability of 25% is pretty meaningless. If you already know that you drive to and from work during rush hour times you might face a 100% probability of a jam. And likewise, if you leave at 2am, you have a much smaller probability of traffic problems. Same with the failure rates of the 4% Rule! That’s why I built a tool (see Part 28) to report not just the unconditional failure probabilities but also those conditional on market conditions. For example conditional on high CAPE ratios and/or an equity index at or close to its all-time-high (asin early 2022!). Readers of my series know that the failure probabilities conditional on expensive equity valuations are very different from the crude and useless figures floating around in the Trinity Study and the FIRE blogosphere.
In summary, the underlying confusion about failure and success probabilities is all based on the misplaced Random Walk assumption and the confusion between unconditional and conditional probabilities. For some additional evidence of the non-random walk properties of the stock market check out this post: “How much of a Random Walk is the Stock Market?”
4: Rules are useless!
A byproduct of the simplicity mantra is that if index investing is so easy, then everything else in personal finance must be simple enough to distill into a bumper-sticker-size rule. Some examples:
- The 4% Rule
- “Pay down your debt first.”
- “Keep X months’ worth of expenses in an emergency fund.”
- “Keep bonds in a tax-deferred account.”
- … and many more
All those rules are useless. Personal Finance should be custom-tailored to your personal needs. It’s called “personal” finance for a reason. I’ve been dealing with the 4% Rule fallacy for over 5 years now. I often make the case that the word “Rule” is actually more offensive than the 4% part. That’s because I have done some case studies for FIRE volunteers with significant cash flows later in retirement where I justified much higher withdrawal rates than 4%, sometimes North of 5% or even 6%. Who knows, some of the folks would still be working today to chase that dumb 25x rule if they hadn’t found me to do the math right!
And all the other rules are equally useless:
- Pay down debt? Young investors tend to benefit from keeping the mortgage and turbocharging their equity investments early on to hedge the (sequence) risk of missing out on strong returns early during their investing career. And in retirement, reducing your mandatory expenses is also a hedge against sequence risk (Part 21). But even this “adjusted” conditional debt paydown rule may not be 100% universal.
- I’m not a big fan of emergency funds, as many of you know. In fact, the first claim to fame of my blog was my controversial $0 emergency fund post back in May of 2016. For most young investors it’s more important to build up their equity portfolio as quickly as possible to hedge against Sequence Risk. And for most older investors who have a large enough portfolio, there isn’t really a need to have a designated emergency fund in a low-risk and low-yielding account (e.g., money market, short-term CDs, etc.) either. But just to be sure, I did concede in the most recent installment that if the need of tapping your funds is highly correlated with the stock market, you probably want to keep your emergency fund in a low-risk investment outside the stock market. I don’t want to replace a useless emergency fund rule with an equally nonsensical rule of my own. If you want to have an EF and how much money you need is again a function of your own financial parameters!
- The often-cited rule of keeping bonds in the tax-advantaged accounts is also not 100% accurate, as I pointed out in Part 35 of the series. It depends on interest rates, expected stock gains, dividend rates, and certainly your current and expected tax rates. There are potential cases where it’s better to keep bonds in the taxable account!
5: Uncertainty does not justify “winging it”
As I outlined in Part 46 and Part 47, the presence of market uncertainty is no excuse to be sloppy with your retirement withdrawal analysis. Quite the opposite, the substantial market uncertainty and potential additional uncertainties, like idiosyncratic spending shocks and policy uncertainty about future Social Security benefit levels, tax policy etc., likely call for more precision in your analysis today, not less. That’s because “winging it” adds even more uncertainty to the existing one.
I always like to use the medical analogy of Dr. Wingit, MD who has to inject medicine into a patient. He looks up the dosage information, 3.5ml. But Dr. Wingit says, “3ml, 4ml, 5ml, this patient has a very uncertain outlook anyways, so who cares, right?!” Well, it’s the opposite; because there is already a lot of uncertainty about the health and survival chances of the patient, we want to be extra-cautious and not compound that risk even more by winging the dosage.
In other words, critics of my analytical approach try to sneak in the stereotypical Strawman argument: I never claimed that with my tools I can eliminate any of the market uncertainty. I merely don’t want to compound the risks even more.
6: We are not very good with percentage calculations!
Related to item #5, I often hear people say “3%, 4%, 5% withdrawal rate, who cares, it’s only a 1% difference, right?!” Or this one: “It makes no sense pinning the safe withdrawal rate down to the significant digits after the decinal point!”
Do you notice the flaw in these statements? Going from a 3% withdrawal rate to a 4% withdrawal rate is indeed a one percentage point difference in the rate. But it’s a 33.3%(!) increase in the withdrawal amount. For example, with a $1,000,000 initial portfolio, you’d raise your withdrawals from $30,000 to $40,000. The last time I checked, $40,000 is not 1% but 33.3% higher than $30,000.
So, did you notice what our mathematically inept Dr. Wingit did here? He claims that going from a 3ml dose to a 4ml dose is only a 1% difference. If it’s expressed relative to the 100ml bottle in the medicine cabinet! But the increase in the dosage injected into the patient was still a 33.3% increase! So, to all the internet influencers out there rolling their eyes when I display safe withdrawal rates with a 0.25% or even 0.01% precision: I roll my eyes right back at you for not understanding 5th-grade percentage calculations! 😉
If you’re still not convinced, take a look at this simulation from Part 27. Here’s the path of a $1,000,000 portfolio over 50 years for the Jan 1965 retirement cohort using withdrawal rates between 3.2% and 3.7% in 0.1% steps. Notice how sensitive the final outcome is when changing the withdrawal rates by a mere 0.1 percentage point. My best advice: Do the math right up to the last 0.01%. You can then always round it up/down later, folks!
7: We can be nonchalant about failure probabilities. But you can’t!
I don’t remember exactly who it was but one fellow FIRE blogger once confidently declared that a retirement strategy with a 20% failure probability is totally fine for him. It’s certainly the kind of tough talk that invokes a lot of “ooooh”s and “aaaah”s from the podcast audience or at a FinCon cocktail party. But it’s also BS! The particular FIRE blogger who I think said this (99% sure) has a zero percent chance of depleting his portfolio because of his wife’s income and revenue from his blog. And don’t call me the “retirement police” because I have no problem with FIRE bloggers monetizing their blogs. I monetize my blog, though I still generate the overwhelming majority of our retirement budget from investment. If anything, we actually make way too little money in the FIRE community considering the services we provide. But we want to avoid giving 99.9% of the FIRE community the impression that we can all ignore Sequence Risk because the top 0.1% of the most profitable influencers face no such risk. It has a bit like Marie Antoinette’s “let them eat cake!”
If this blogger or anyone else indeed faced a 20% failure probability I have a hard time believing they would be OK with that. Ask yourself, would you be OK with a 20% chance of…
- Being late for a flight? (not talking about a connecting flight, but going from your home to the airport to catch a flight)
- Running out of gas?
- Being late for a wedding? How about your own wedding?
- Being late for a funeral?
How about your own funeral?Oh, wait, that last one didn’t work so well!
Maybe some people will still maintain that they are, but I have never met anyone who has missed 20% of their flights or missed 20% of important family gatherings or shows up to work late 20% of the time. If you’re like me and you leave for the airport early enough to absolutely never miss a flight, it’s because there is an asymmetric risk profile. There is a small cost because you waste 15-30 minutes sitting at the airport gate, but missing a flight by 1 minute is asymmetrically costlier. If I’m not willing to accept a 20% chance of something as trivial as missing a flight, then why would I risk a 20% chance of running out of money in retirement? I know that I will likely end up with excess assets when I die, but that’s a good problem to have!
8: Flexibility in retirement raises more problems than it solves
The last resort of Dr. Wingit to push back against a more careful retirement analysis is that we can be flexible in retirement. Well, count me in on that one. I’m certainly flexible. We have a retirement plan with the flexibility to cut our budget by 15%, even 20%+ and still have a comfortable retirement. We also have the flexibility to go back to work. But flexibility is no panacea. Some really bad flexibility headaches are often ignored when living in the FIRE la-la-land where people can make sweeping claims about flexibility but never quantify anything. I made an honest attempt at measuring how flexibility would have worked out in some of the historical simulations and there are (at least) three unpleasant issues:
1: How flexible do I have to be? Skip the Starbucks Lattes for a while? Skip the CPI adjustment for a few years? I did the simulations in Part 23 and found that in order to have a meaningful impact on cushioning Sequence Risk, you’d have to lower your withdrawals by probably closer to 30%.
2: How long do I have to be flexible? A common misunderstanding is that tightening the belt only has to last until the end of the Bear Market, 1-2 years, right? False! If you start withdrawing your full initial retirement budget again at the start of the next bull market, keep in mind that at that point, your portfolio is still at the bottom, when you are facing the maximum Sequence Risk impact! To alleviate Sequence Risk, we’d need to wait until the portfolio has made up the losses plus inflation! And that can take a lot longer, sometimes decades. 30% lower consumption for 20 years is not really that palatable.
3: How about false alarms (Type 1 errors)? In that same post, Part 23, I showed that some of the historical cohorts whose portfolios would have survived a 4% Rule in the end, still had a scary ride along the way. Take the example of the Dec 1972 retirement cohort using a 4% Rule. Less than 10 years into retirement, you had depleted roughly 70% of the portfolio (in real terms). But thanks to the stellar performance of the U.S. stock market during the 80s and 90s, your portfolio survived all the way until 2002. But this lucky (unlucky?) retiree probably would have done the flexibility thing, trying to find work in the 80s to supplement his/her retirement for 10+ years. That’s because nobody had any idea how well the stock market turned the corner eventually.
So, again, I’m onboard with flexibility, but it’s overrated once you look at the historical simulations. You don’t really eliminate the failures, you simply reshuffle them. You replace “running out of money” with “going back to work” for very extended periods. Even worse, you create false alarms, i.e., failures of sorts where the 4% Rule didn’t even fail in the absence of flexibility.
9: CPI adjustments in retirement may be insufficient!
The premise of the Trinity Study and much of my Safe Withdrawal Rate Series is that you adjust your withdrawal amounts by the CPI index. It ensures that you can keep your standard of living without slowly eroding your purchasing power over time. As I have previously pointed out, foregoing inflation adjustments for maybe one or two years is not a big deal, but over longer horizons, 30 years for traditional retirees and potentially 50 or even 60 years for early retirees, the purchasing power erosion would be disastrous!
But some retirees, certainly early retirees, might want to consider whether CPI adjustments are truly enough. Let’s look at the chart below for the 1991 to 2021 path of per capita real consumption. It increased by 70%, which means that if you had merely increased your consumption with CPI+0% only, the people around you would now be consuming 70+% more than you. Or, equivalently, you would have fallen 42% behind your peers, i.e., your neighbors, friends, and relatives. And foregoing even the CPI adjustments, you’d be 71% behind the average American, now that’s a non-starter! I can certainly see that a traditional retiree, at 65-years old, could feel comfortable slowly falling behind the average U.S. per-capita consumption path, at least in the absence of health problems. But as an active early retiree, I’d like to still participate in the per capita real consumption growth for a few decades and update my gadgets and buy improved creature comforts. In other words, I don’t want the same consumption basket from the 1990s today. Remember: there were no smartphones back then! And, likewise, in 2051 I probably don’t want the 2021 CPI consumption basket either!
By the way, this CPI+x% creep in expenses will be even more pronounced for the folks who retire abroad in a (currently!) low-income country. The per-capita real consumption increases in some of the popular retiree destinations like Thailand, Cambodia, Philippines, Costa Rica, Panama, etc. are even greater than in the USA. I wonder what kind of apartment you’ll be renting in Phuket in 20+ years if you apply only a CPI+0% adjustment to your budget each year!
But just to be sure, my Safe Withdrawal Rate Toolkit has a feature for you to tweak the simulations and model withdrawals that grow faster (or slower) than CPI inflation. To my knowledge, my toolkit is the only one out there with that kind of modeling flexibility.
10: Reliance on the government may backfire!
A good chunk of the FIRE content on blogs and podcasts deals with optimizing government policies and benefits. Obamacare subsidies, Roth conversion ladders, the 0% bracket for long-term capital gains and (qualified) dividends, etc. And that’s all really helpful advice. I’ve written about the awesomeness of the 0% tax bracket and retirement tax optimization myself (Part 44 and Part 45). But I feel a bit uneasy about all this. Will this government generosity last? Have you looked at the government deficit and debt figures lately? Could there be a backlash against the FIRE community? Recall, that most of the media coverage of our community has been positive and admiring, even fawning at times. The “worst” coverage I recall was about that Seattle lawyer who took her frugality a little bit too far. But could there be a harsher backlash if people find out that multi-millionaire retirees in their 30s and 40s collect not just zero-tax qualified dividends and capital gains but also Obamacare subsidies and get low-income subsidies for their cable TV and internet? Maybe our FIRE community needs to hire some lobbyists in Washington D.C. to keep our party rolling?!
I certainly wish that we won’t become a target of any jealous social justice redistribution mob and we can keep our low-tax and benefit party rolling. But I wouldn’t bet my retirement security on it. Here in the ERN household, we don’t even receive ACA subsidies (yet), but the tax landscape certainly benefits us greatly. Let’s hope it stays this way. But if not, our retirement would still be safe and comfortable. But I am worried a bit about the folks who retire on a bare-bones budget and only just made their 25x rule of thumb work – but only because of ACA subsidies. You are setting yourself up for a rug pull by the government! My recommendation: try to be not just FI but aim for an even safer acronym. Here are some suggestions:
- FINAL = Financially Independent, Not a Leech
- FINDOG = Financially Independent, Not Dependent on Government
- FINISH = Financially Independent, Not In State’s Hands
- … any other suggestions, please post in the comments section and I will add to this list! 🙂 for sample:
- Reader “Mapleton” came up with a nice acronym of what we don’t want to become: FIBBING = Financial Independence Blindly Based on Individual Naive Guessing. Very nice!
- Reader “J Acorn”: Don’t be “FIDDLE (while Rome burns)” – Financial Independence Drained, Didn’t Listen to ERN. That’s awesome. Why didn’t I think of this?!
Conclusion
OK, let’s wrap it up here. Happy Anniversary! Thanks to all readers that have come along with me on this amazing 5-year run! Not sure I am going to make it to 100 posts over the next five years, but I got notes for at least another ten posts. Also, I hope I didn’t ruffle any feathers with today’s post. Again, I don’t mean to accuse any of my fellow-FIRE bloggers of any malice or wrongdoing. And for the three or so people who might still be offended, I like to point out that the concept of F-You Money worked not just when leaving my corporate job in 2018 – it works equally well or even better with the online outrage troll mob today. 😉
So, Happy New Year, everybody. Let’s have a healthy and prosperous 2022!
I-Bonds have a 7% interest rate right now, it is a good place for emergency funds. So, if ew are in a CAPE environment doesn’t it make sense to pay down debt?
All yields are low. TIPS, I-Bonds now have the advantage of high rates. But the real rates are 0%. Also, going forward, expect CPI to move back to 2.5% over the next 10 years. That’s the TIPS-implied inflation rate.
Thanks for the post! Two questions regarding “CPI+x”:
1. Precisely how do you estimate the “x” in CPI+x? For long time horizons, this parameter makes an enormous difference to the SWR.
2. Why use CPI as a deflator at all? Why not use instead something like median family income (https://fred.stlouisfed.org/series/MEFAINUSA646N)? In other words, why study real returns – why not instead study median-income-adjusted returns?
(Re: 1, how about this method: during the period 1953-2020, the average annual change in MEFAINUSA646N is 4.60% and the average annual change in CPIAUCSL is 3.49%, and therefore x = 4.60% – 3.49% = 1.11%.)
Precisely. Use 1.11% as x then. 🙂
I’d start with real per capita income or consumption. See FRED database:
https://fred.stlouisfed.org/series/A939RX0Q048SBEA
I start with CPI because I have CPI data going back 150 years. Monthly data! Not available for your seried.
I can always add CPI+x later and calibrate that x by looking at how much the median income grew above CPI
According to this time series (A939RX0Q048SBEA), real GDP per capita has increased on average 1.95% per year. So 1.95% is another estimate for x?
Good point. That seems a little bit higher than the per-capita consumption number but it’s also legit.
I’d be careful, as this is almost certainly a mean calculation and not a median.
Keeping up with the Joneses is one thing; keeping up with the Musks and Bezoses is something else! 😉
Yeah, you got Bezos and Gates and Musk in there when using the mean. 🙂
ERN your blog series is absolutely best of its kind. Long term reader, infrequent poster here. Just wanted to say thank you for your five years of writing this series and for the “other” perspectives you often provide. Your SW series has provided incredible insights, and given us the necessary tools to build more substantial FIRE plans. You will likely save many of our retirement journeys in the long run, which is one of the best compliments that I can imagine giving. Thank you, and Cheers to 2022 and your continued success!
Thanks for the feedback. You’re too kind! Good luck and Happy New Year to you as well! 🙂
Keep up the good work. Have you considered estimating a confidence interval of SWR rather than a point estimate and evaluate the distribution of the outcomes? Two estimates with 4% could be significantly different when CI and shape of distribution are evaluated. Something like, “Although the SWR is 4%, the 90% CI for SWR is between 1.5% and 4.5% and right skewed, mostly due (keystone variable).” Point estimates for tomorrow are not very reliable, let along for every year for 50 years. CI provide a richer analysis and gives FIRE an estimate for how much slack is in their FIRE plans.
I haven’t. It would rely very heavily on what kind of distribution you apply to the uncertainty around the historical returns.
1,000+ trials, assume normally distributed historical returns, plot outcomes of SWRs. Set some limits to avoid minimum balances ($300k) or max losses (70%), estimate SWRs. My hunch is distribution of SWR would be right skewed meaning most SWR’s are less than point estimates, especially for longer duration retirements. If 0% SWR is in the 90% CI, it’s not a good idea to retire yet.
Haven’t done that yet. If someone wants to program this, you got yourself a guest post here,. 🙂
A 0% SWR can’t be in the CI as long as returns don’t go below -100%. Or what Am I getting wrong here?
In this concept, 0% would mean that contributions would still need to be made to achieve the final terminal value or maintain constant withdrawals throughout the duration of retirement. If minimal value required is $300k and a negative SWR is found to be in a 90%, 95%, 99% CI, the individual would need to still contribute and could not just withdraw over their retirement. This could be filled with SS or other pension income. Anything from an early loss to a string or losses during any point in the retirement duration could trigger this. It is another way to look at failures and the probability that these occur. Although much better than nothing or winging it with retirement savings withdrawals, I’m skeptical any time I see point estimates for long duration estimation purposes.
An extreme example is where a uniform return market, the 99% CI would be extremely small and almost identical to point estimate SWR. While a volatile market one could have the same point estimate SWR but CI that is huge and include 0% to 10%+.
Without analyzing your estimates, I’d guess that your point estimate is the highest probability of occurring, but there could be massive variation if CI were included, longer duration retirement would result in larger CI.
Great post!
as somebody in the near-fire, “when do I pull the trigger phase”, I find myself considering many of the issues you discuss in this post.
In particular, I put a lot of thought into tax breaks and ACA subsidies. I figure I paid a lot of freaking taxes and I’ll take anything I am legally entitled to (without doing anything dishonest) but it’s not at all clear the ACA will be around for another 15+ years, and it’s certainly plausible that an asset test (like that for medicaid) could be added. So I think the best is to budget like it won’t be there, but take it so long as you can get it.
As for medicare, I budget for a 20% real cut in benefit but anything beyond that… well, hopefully we don’t get there.
One thing you missed – I’m surprised you didn’t mention divorce as a possible source of FIRE failure. After all, two people living separately consumes more resources than two in one house (and the cost of a divorce increases with FIRE. There’s not much to “plan” for on this one other than work on your marriage so you don’t end up there, and realize that a big life change like going FIRE might be a huge stressor if the two of you are on not on the same wavelength.
Very good points. Yes, divorce is not on my mind and has never really been. Tanja Hester on her excellent blog raised this issue, not due to personal experience, but just to raise awareness:
https://ournextlife.com/2018/04/30/couples-real-financial-independence/
https://ournextlife.com/2017/06/05/pre-fire-agreement/
So, strictly speaking, the divorce issue has been mentioned and discussed in the FIRE community. But I agree: not enough!
Same boat. I spent some time trying to max these benefits and gave up as the rates, policies and subsidies change in every state every year. It is an impossible task to Max this until right before. Was planning to FIRE mid 2022, but offered WFH until the end of 2022, which simplifies this calculation or just puts it off. What I will most likely do is switch from maxing ACA subsidies one year to maxing tax benefits/0% LTC and Trad rollovers the next. If rules change to favor one or the other, change strategies. It’s a minefield of calculations, timing, execution and every state is different and why universal calculators don’t exist for these issues. ACA plans have wild subsidy fluctuations so impossible to make any generalized statements with this.
How do you apply a Medicare cut in benefit? Do you mean 1) an increase in premiums, or 2) a decrease in social security benefits?
You could model a future negtive cash flow in my Google Sheet.
Also notice that Social Security is a positive future cash flow that should be factored in. Smaller benefits would imply using smaller positive cash flows.
I have long held that money is equal parts emotional and numerical, which the latter is easier to manage. Emotions and the psychological aspect of not losing your nerves and sticking to your savings plan is far easier said than done. Now, when you enter retirement with 100% equities, the 4% rule says you likely will be fine numerically. However, emotionally, you likely won’t be, as stress incurred from huge declines will keep you up at night if you are an average individual. Even the most seasoned of investors will break a sweat and have self doubt. That is part of being human, after all. Thus, weighing your risk tolerance before retiring and continually reevaluating it is a must.
Also, I loved point 10. Don’t milk the system, learn to be self sufficient because government regulations can and will change. Don’t bank your plan on a program that is new, as it may not last!
Olaf,
Your point about emotions is certainly valid. But if you’ve been reading ERN’s SWR series, you’d know that 4% is typically *not* safe, even for most 30 year horizons (depending on other cash flows), and that ERN specifically does *not* recommend having 100% equities in retirement. He usually recommends somewhere beween 50% and 75% at the start of retirement.
Now, for upward glidepaths, specifically designed to handle the case where stocks are down substantially, you are correct that it will likely be hard emotionally for many to increase their equity holdings to or close to 100% if and when a bear market hits during retirement. But not only a) it’s the way to “save” a high withdrawal amount in the face of a severe bear market mauling one’s portfolio, but also b) it becomes at least somewhat easier for many others of us who’ve read ERN’s series – because historically it has worked! And as other parts of his series show, little else (beyond holding a healthy does of gold) does.
Oh, just saw this reply. Great point! I independently tried to make the same point but your reply is much better! 🙂
With 100% equities the 4% is absolutely not safe. Not even the Trinity study would say that. you get higher median and best-case scenario results. But the failure probability is higher.
But aside from that, agree with everything you said! 🙂
ERN,
So I just re-read this piece – again! On the one hand, points 1 through 8, and 10, are excellent, and almost make up an executive summary of what we’ve learned in the SWR series. Point 8 in particular is IMO one of your most valuable contributions to the FIRE and indeed entire retirement community, and it’s what got me off the idea of using something like Guyton-Klinger rules, which is where I was leaning until I came across this series. So thanks much for that .
On the other hand, the more I think about the “CPI+x” idea in Point 9, the more problematic it seems to me – and in particular, it seems likely to violate the false alarms (Type 1 errors) issue from Point 8 – at least in spirit.
Why do I say this? Because by definition, adding in the “+x” factor to an SWR calc means that you are forgoing spending now in order to be able to spend more in the future. So your standing relative to the Joneses still working might be maintained, but at the cost of most likely being years or decades – or even forever – lower annual spending than what it might have been without the “+x” factor.
And it seems to me there is a much simpler alternative, that most of the time will allow you to have your cake and eat it, too. Kitces has alluded to this. Namely, to increase your withdrawal amount at any time the original SWR percentage times the current portfolio delivers a higher spending number than the inflation-adjusted amount based on the initial portfolio delivers. And in said case, if using a glidepath, resetting the equity portion of the portfolio down to the original initial percentage.
I.e. whenever in real terms your portfolio is higher than it was when you “started”, just pretend that this is the new first day of retirement! If the WR percentage you have chosen is historically failsafe, then by definition resetting to using that percentage on your higher portfolio *is* safe by all historical standards!
You can think of this as a tweak to the active glidepath mechanism you proposed in Parts 19-20.
Done this way, you don’t have to start with a lower spending level, but you still get to take advantage of the productivity gains in the economy as generated by the equities portion of your portfolio, something like 80% – 95% of the time.
[I’d love to see a conditional analysis of what these percentages look like 😉 ; no doubt they will be lower when starting with an elevated CAPE like today’s.]
Now obviously, someone who wants to consume less when they are younger in order to guarantee that they can spend more when they are older is free to do so. But that approach seems a bit at odds with the Trinity study and most of the rest of what you’ve done with the SWR series, as you point out early in this post.
This approach, by contrast, seems to me a way to avoid the Type 1 error of lower initial spending while still enabling you *most* of the time to keep up with the Joneses.
I appreciate Kitces’ work, but the “ratcheting” thing is often misinterpreted.
The whole point about about failsafe rates is that if you start at the worst possible time you will never ratchet up but deplete your assets.
So, I would like to be able to have a rising real consumption path even in the worst possible case when the “ratcheting” doesn’t happen on its own.
And by the way, rising real consumption is not abnormal. Every year we raise our consumption standards. The Trinity Study is simply oblivious to it (and many other issues). My series has focused on the fixed real path for the most part but at least I’ve identified the issue and written about rising paths.
But I agree, we can still do the flat path, calculate the failsafe WR, assume the flat path as the worst-case scenario and use the ratcheting path to walk up our consumption in the other 95%+ of the cases. That is certainly a viable retirement strategy.
Pretty sure we’re in agreement re: Kitces. He deserves credit for bringing up a couple of topics (bond tent / rising equity glidepath; the idea of ratcheting, even the idea of being able to have rising consumption much of the time in retirement), but his “implementation” / suggestions for how to deal with them leave much to be desired.
Your work on glidepaths and what the SWR numbers should be is *far* superior to his. And IMO should be widely consumed and understood by not just those in the FIRE community, but all financial planners and anyone retiring / in retirement / thinking about retiring, early or not.
With Point 9 you are wading into “ratcheting” / rising consumption territory in effect. I specifically said Kitces only “alludes” to the right answer, because his watered down proposal isn’t particularly good at all. Only by reading your work, including the stuff on glidepaths and on the problems with Guytin-Klinger, have I been able to understand this.
While searching the other day, I found a 2015 post of his where you were “debating” with him in the comments 🙂 . And while I did find a 2019 post where he references the idea of “ratcheting” at every new high in a single sentence – which idea seems clearly to have come from a reader comment to one of his earlier pieces – it was without any explanation or backing for it, or seemingly any conviction on his part.
IMO the only thing he gets right and is somewhat explicit about that you aren’t is the idea that bonds are not a good hedge/ballast against stocks in today’s low bond rates, “helicopter Ben”, inflation threat world. Even though it seems pretty clear that with your actual portfolio, you don’t actually put much stock in bonds [pun intended :-)]!
Nice summary. Couldn’t have said it better! 🙂
I heavily agree with point #7. Most in the FIRE community heavily under-estimate risk. Things that are common and/or simple does not mean it is ‘risk-free’.
At the end of the day though, we can have all this empirical data and know that most of the time market goes up so you might be able to get away with a ruleset like this:
* If CAPE is higher, withdraw less than 4%.
* If CAPE is low, withdraw >4%.
But the thing is, you can’t predict the future and your timing of your retirement can just be…unlucky. You can retire and withdraw 4%. Or you just say ‘I’ll just withdraw >4% until I reach a threshold of net worth then I’ll go back to work.’
Active decisions like this doesn’t seem like retirement at all.
Next, you can just do a ruleset of:
* I don’t care what the market’s doing. I can’t predict the future. It amortizes over time. So I’ll pick 4% to be conservative even if the market yields much higher most of the time.
You can still be unlucky and run into a 30 year depression. Or a 3 year depression. But the thing is: there’s no way you’ll know how long the bad times will last when you’re in year 2. Previous data doesn’t indicate future performance. So it doesn’t feel like retirement at all.
OR you can be ultra-conservative and do something like this:
* Save so much hard cash such that with inflation accounted for, you’ll have 80% of all your projected expenses til death.
* On top of that, have investments you can withdraw 4-7% of each year.
But you might never retire if your income isn’t high enough, or you can retire like that if you plan to live in hostels and low COL areas you hate for the rest of your life.
No matter what you do, there is some risk, whether it’s significant or black swan. Unless you’ve got like $100 million dollars in the bank and you can be 100% sure you’ll never be able to spend that down, even if you tried. And in that case: the solution is just to make more money.
Seems like analysis paralysis. I prefer to just go with my fail-safe and be done.
Also: you can’t even save “enough” cash to make it through retirement because you don’t know the path of interest rates and inflation.
Previous data doesn’t *guarantee* future performance; that is certainly true.
But in fact, previous data *is* quite indicative of future performance almost all of the time.
That is the crucial difference.If you don’t understand that, then literally nothing will have any meaning for you. Because sure, technically it’s possible all markets will drop by 90% and stay there for 15 years. But that doesn’t make it a scenario worth planning for.
Because the future *might* be worse than anything in the past, it’s certainly worth having contingency plans (targeting a higher final value, being prepared to cut spending 10%, even 20%, etc.). But just because there are always risks doesn’t mean you should do nothing, or that no plan is sensible.
I like the FINISH acronym. It’s clever and its a pun!
Pun intended! 🙂
Love your page, keep up the great work!
Thanks for the feedback! 🙂
With regard to reliance on the goverment, where do you draw the line? No Medicare, Social Security, ACA? You can set the bar pretty high. While I suppose means testing could happen, to me that seems like a good way to backdoor kill programs. Live below your means, save and invest wisely? No soup for you! I’d rather just have a straight wealth tax.
Without endorsing or decrying any of the programs you mention, a straight wealth tax is a terrible idea. For many practical reasons as well, but in particular from an economic perspective, it literally penalizes savings, which is the opposite of what’s optimal.
P.S. means testing is already there for Medicare and ACA, and even a little bit in Social Security (benefits taxed at a higher rate).
Depending on how it’s structured, a straight wealth tax would affect massive accumulations of wealth, not people of modest means who have accumulated a decent nest egg. There is some means testing on ACA and Medicare as far as income, but not wealth.
Income taxes were also initially sold as hitting only a small minority of ultra-wealthy. Now almost everybody is taxed.
Also, just because a wealth tax doesn’t impact you and me (yet) doesn’t mean it’s a good idea. It would be a really bad wet blanket on the economy.
Countries like the Netherlands that have the wealth tax do affect middle class folks who are trying to not work their whole lives. Their 2% wealth tax effectively makes the SWR drop from 3-4% to 2-3%, requiring you to save about 50% more for retirement than if there wasn’t a wealth tax.
IMO it would be better to bump up the capital gains and inheritance taxes, so you don’t affect SORR for retirees and only pay when in years that you’re getting wealthier and/or die wealthy. These taxes are also much easier to asses and more difficult to hide from compared to overall wealth taxes.
Considering who is writing the tax laws, a straight wealth tax will be unlikely.
I am not drawing any strict lines. Personally, i assume a moderate haircut to my Social Security benefits. Maybe 20-25% lower than currently projected.
Medicare: I assume more out of pocket expenses. Modeled as higher expenses when old. I’m not planning to have no Medicare at all.
ACA: not needed right now. We use a healthshare ministry and that serves us well.
Excellent blog post. I’ve always thought the FIRE movement was oversimplifying and overly optimistic about the 4% “rule” and you’ve captured a lot of the issues here.
The customizable SWR spreadsheet toolkit is especially helpful.
I have to say I’m quite surprised given how careful you are about SWR, that you’re willing to take your chances with a healthshare ministry, as they are not insurance, have no out-of-pocket maximums, aren’t required to be adequately funded to pay for covered expenses, and have neither fiduciary nor regulatory obligations. Seems like you are very cautious with SWR, but much less so with health insurance.
Thanks!
We are considering switching to a health plan from the state exchange but for now we take our chances with the health ministry. Our plan has an annual out of pocket maximum. We trust that Medishare is managed well enough to pay its bills. Philip Taylor has been using them for 8+ years and seems to be happy: https://ptmoney.com/medishare-review/
I agree with most of what is said on this blog, but I don’t agree with the way you frame probability of failure and your implied suggestion that a 20% chance of failure is “bad.” You seem to take issue with a 20% chance of failure, which is also an 80% chance of success. But the real distinction I ask you to consider: change “failure” to “requiring adjustment.” Many people would accept that there is a 20% chance they must make minor adjustments, rather than outright failing. Truly failing 20% of the time would only be if you made zero adjustments, such as temporary cuts to discretionary spending. A much more eloquent explanation of the differences can be found here:
https://www.kitces.com/blog/monte-carlo-retirement-projection-probability-success-adjustment-minimum-odds/
I can’t really follow the exact simulation exercise used in that article. It’s not by Kitces himself but by a contributor, which explains why.
I’ve done extensive research on “adjustments” and “flexibility” and the results don’t get better. See parts 18,23,24,25. Also the GK guard rails.
The adjustments are a) potentially decades-long and b) they are not small.
People often delude themselves that all they need to do is to skip an annual CPI adjustment and everything is A-OK. That’s not the case if you look at actual historical data.
Herr ERN, you’re the real deal. I don’t agree with a couple of things but overall this is one of the best FIRE posts out there. Truly disturbing! I recall Mr. RIP’s post on why retirement at 40 is dead. Off-topic, care to comment on your part-time job and travel plans since COVID has given us a break ? Congrats again and i will be coming back for more! LG
Thanks!
Most people who retire at age 40 will likely pick up a side gig here or there. I did my consulting gig for a startup but only for a year.
Back to traveling more aggressively. Might write an update on that topic. Stay tuned! 🙂
Great Post.
Great
How do you think stock buy-backs influence CAPE, if at all? Also with respect to your Withdraw Rate to CAPE plot? I guess buy backs would initially lower a company’s PE, then the market would theoretically boost up the price back to its previous PE, so its net neutral?
Thanks for all your thoughtful posts.
Stock buybacks impact the Shiller CAPE in the way outlined here:
https://nucleuswealth.com/articles/trump-broke-the-shiller-pe-here-is-a-simple-fix/
Hmmm… Not sure I’ve seen anything recently that was both very good and very bad at the same time.
I am in 100% agreement on the point about the corporate tax rate having a *major* impact. I wholeheartedly support that tweak to the index.
I disagree almost completely about the buybacks, in terms of their impact on expected future stock performance and calculating a CAPE. First and most important, there isn’t that much difference between returning money invested in the form of dividends or buybacks – save that it is usually more tax efficient to be done via buybacks. Second, there are often options/RSUs allocated to management/employees, and there have also been offerings of additional stock in the past. And since in an index of shares we care about the value and earning power of, well, *shares*, why would it be a good idea to tweak said index for buybacks. There’s still more, but I’l stop there.
ERN, do you agree with the author that stock buybacks have had a major impact on the Shiller CAPE?
Because it seems to me that corporate tax rate, the (until recently) 40 year decline in interest rates and investors’ willingness to pay higher multiples (in response to falling interest rates, fewer recessions, lower trading costs, more economic optimism) explain basically all of the rise.
I agree with Mr. Klassen. And you do, too, without knowing it.
You correctly point out that buybacks vs. dividends should have no impact on future expected returns, due to a Modigliani-Miller-like symmetry. And exactly because of this, there has to be an adjustment to the CAPE in light of buybacks. I admit that Klassen didn’t explain this very well, so let me take a shot.
Imagine we have 2 countries with a stock market in each country generating $1 in profits every year. The current index is 10 in both countries
In country 1, corporations pay out their entire profits as dividends each year. Profits have been $1 every year for the last 10 years (already adjusted for inflation). So the average earnings were 1 over the last 10 years, so we get a CAPE of 10.
In country 2, corporations pay no dividends but only do buybacks. Specifically, they scoop up 10% of the outstanding stocks every year with their earnings. The average investor will see the same return as in country 1 because either you sell 10% of your stocks and get a quasi-dividend that way. Or you don’t sell and you see your stock price go up by 10%.
The earnings per share outstanding today over the last 10 years are:
year EPS
-9: 0.4241
-8: 0.4665
-7: 0.5132
-6: 0.5645
-5: 0.6209
-4: 0.6830
-3: 0.7513
-2: 0.8264
-1: 0.9091
0: 1.0000
That’s because 9 years ago you had 1.1^9 as many shares as today and the earnings were spread over many more shares. Averaging over the 10 years we get only 0.6759 as the 10-year rolling real earnings. And a CAPE of 14.7950.
That’s why we should adjust the EPS in the CAPE formula to account for the smaller and smaller share base. By doing the adjustment we raise the EPS in year -1 to 1.0, all the way to year -9 to 1.0. And we get the same average EPS of 1.0 and CAPE=10 as in country 1. As it should be.
Just on the acceptable failure rate question, aren’t the failure probabilities overstated because you assume very longevity? In coming up with a truly robust strategy that is a good assumption of course but a 20% chance of failure IF I LIVE TO 100 is not a real world 20% failure rate is it? I think that is a respectable reason why people think they could tolerate a surprisingly high modelled failure rate.
Apologies if you’ve addressed this before, I’m new to your excellent work and there is a lot to read!
Yes, the failure rates are conditional on living up to that age. But make no mistake, a healthy couple has a >50% of at least one spouse living into the late 80s, even early 90s.