Traditional retirement planning advice says you should shift your investments increasingly to bonds as you near retirement, and then keep those investments bond-heavy during retirement.

The problem is: that’s too extreme.

You sacrifice tons of upside (think about the tech sector in recent years) without getting enough in return.

Fundamentally, you’re trying to balance getting good returns vs. not losing your shirt right before you retire (or worse, soon after you retire).

A better strategy is to shift to bonds as you near retirement, but then reverse-shift back to equities in the initial years after retirement.

This is called a “bond tent.”

But executing this in the right way depends a lot on your personal facts and circumstances. There’s a lot of nuance and judgement involved.

What percent of your investments should you target for your peak bond allocation? How long should you take to shift into it? Should that shift be linear or lopsided? What bond holdings are most appropriate for your situation? How much will this actually reduce sequence risk?

Plus, all the same questions in reverse for shifting back to equities!

It can make your head spin.

So, this week, in part 1 of a 3-part series on asset allocation, I talk with Karsten Jeske, CFA, about how to implement these glide paths, both leading into retirement and in the initial years after retirement.

Getting your asset allocation right, and shrewdly changing its composition in the years just before and just after retirement is one of the most impactful things you can do to offset sequence of returns risk.

If you want to understand how to do this effectively, don’t miss today’s episode!

We discuss:

• The intuition behind bond tents and equity glide paths
• Pros and cons of a bond tent strategy (what you gain, what you lose)
• How to determine the optimal % peak allocation of bonds
• How long you should optimally stretch each glide path over (and why each side probably should be different durations)
• When is a longer vs. shorter glide path better in terms of returns and risk
• When is a bond tent NOT worth it and you should just stick with 100% equities

What do you think about the bond tent strategy? Let me know by leaving a comment.

Don’t miss an episode, hit that subscribe button…

If you liked this episode, be sure to subscribe so you don’t miss any upcoming episodes!

• Apple Podcasts
• Overcast
• Spotify
• Stitcher

I need your help, please leave a listener review 🙂

If you liked this episode, would you please leave a quick review on Apple Podcasts? It’d mean the world to me and your review also helps others find my podcast, too!

Links mentioned in this episode:

• ERN’s cashflow modeling spreadsheet
• ERN’s posts on glidepaths: part 1, part 2
• The shockingly un-simple math behind retirement safe withdrawal rates (Part 1) (HYW035)
• The shockingly un-simple math behind retirement safe withdrawal rates (Part 2) (HYW036)
• Schedule a private 1:1 consultation with me
• HYW private Facebook community

Read this episode as a post:

01:22 Andrew Chen
Today’s episode is part one of a three-part series on asset allocation. This first episode focuses on glidepath and bond tents in the years leading up to retirement and the first few years after retirement, to mitigate sequence of returns risk.

The next episode focuses on a dividend investing strategy commonly called in the FIRE community as the “yield shield,” as a way to mitigate sequence risk by reducing the need to draw down principal. And we talk about some of the tradeoffs of doing that, as well as whether it’s actually a fallacy.

And the third episode focuses on an alternate dividend strategy where you combine rental real estate with classic portfolio investing to see if that provides a better outcome in terms of mitigating sequence risk. And we also talk about, in that third episode, how to evaluate the crossover point as a retiree, after which you are virtually guaranteed not to run out of money in retirement, and so you’ve effectively neutralized sequence of returns risk.

So it’s an action-packed three-episode, all with the same guest. And I hope you enjoy it. Let’s get to the show.

My guest today is Karsten Jeske, also known as “Big Ern.”

I had Karsten on the podcast last summer, where we talked about safe withdrawal rates and sequence of returns risk – and that was one of the most popular episodes ever. I invited Karsten back to the podcast today to share insights on a critically important related topic: asset allocation for early retirees.

Quick background on Karsten: he holds a Ph.D. in economics. He is also a Chartered Financial Analyst, and an early retiree who FIRE’d a few years ago at 44.

Before that, he taught economics at Emory University, worked at the Federal Reserve Bank of Atlanta and also at the Bank of New York Mellon, Asset Management Group, in San Francisco.

He writes extensively on his blog “Early Retirement Now” about financial planning for early retirement, and is famous in the space for his in-depth analysis of “safe withdrawal rates.”

Karsten, thanks so much for joining us again today to share insights and tips about asset allocation for early retirement!

03:23 Karsten Jeske
Thanks! Glad to be here.

03:25 Andrew Chen
For those who aren’t familiar, maybe who didn’t even have a chance to catch the last episode we did, what got you interested in early retirement research and financial planning and wealth management research?

03:38 Karsten Jeske
It’s not that I ever hated any of the jobs that I did, but I had this inkling that I’m not going to do this job forever. I had a very safe job at the Federal Reserve Bank of Atlanta, so that was a fun gig. But even after six, seven, eight years, you get a little bit of the itch to move on.

And then I moved into the private sector, and I moved right around 2008. And moving into finance and asset management in 2008, that was probably one of the worst times to move, with very uncertain job prospects, with layoffs everywhere.

I also knew that now I’m looking at not just two risks: me getting bored of the job and not doing that job forever, and then also the additional risk. “Oh my god, I might get laid off after two years of doing this.”

So I always had this suspicion that my job market path is going to look very different from my parents’ generation who worked for one company for 25, 30, 40 years, then retired with a company pension. So I already knew that I’m not going to do this job forever, either voluntarily or involuntarily.

So I thought that the best way around that is to save a lot of money, have a 30%, 40%, 50% savings rate, and plan an early exit, which I then did after 10 more years at Bank of New York Mellon.

05:11 Andrew Chen
What got you interested in the research side of it, and understanding a lot of the quantitative and empirical research?

05:20 Karsten Jeske
There’s obviously a lot of research out there, because most of the time, you don’t have to reinvent the wheel. You just look up what is the research out there, and there’s probably more research than you can ever consume anyways.

But I must admit that the research that was out there seemed all way too generic. They’re looking at 30-year retirement windows. Well, it’s not going to work for me as a 40-plus-year-old retiree.

They look at a flat spending path during retirement. That’s not going to be the case for me.

It could go up. You could have some health shock later in retirement.

Or you could just scale down your retirement. Maybe once we are in our 70s and 80s, we’re probably going to be more sitting at home and knitting and watching TV, and not going on cruise ships anymore.

So there is this uncertainty on our personal side. There is the length of retirement that didn’t really line up with our preferences.

Again, I have a finance background and an economics background, so I believe in some sort of a valuation approach and mean reversion. So when people write these retirement simulation result papers where they say, “You have a 97% success rate in your retirement,” I say, “Well, a 97% conditional on what?”

Unconditional probability or conditional on where we are today with very expensive equity valuations? Shouldn’t we look at conditional probability? Conditional on where we are today, how would have been the survival probabilities in your safe withdrawal simulations in the past?

And I didn’t really see much of that level of rigor and care taken in the simulation. I said, “Well, if nobody else has done that, I guess I’ll do it myself, because then at least I know it’s done right and it’s done to my specifications, and I can fine-tune it to my personal parameters.”

So, that’s how this whole research started.

Andrew Chen 07:34
Well, it’s been a big contribution, I think, to the FIRE community knowledge in this area. And one of the things that you write a lot about is sequence of returns risk.

One portfolio management technique that both you and Michael Kitces both talk about is this notion of a bond tent, which is using a glidepath into bonds as you approach your retirement date, and then using a reverse glidepath back into equities in the first few years after retirement. And doing this helps mitigate that sequence risk.

I was wondering if you could explain in a bit more detail how the glidepath strategy works and the intuition behind it.

Karsten Jeske 08:12
I have, so far, only published basically the right half of that bond tent.

So, imagine you were retired today. How would you structure your asset allocation over time to mitigate sequence of return risk? Because you don’t have to have one fixed asset allocation throughout your entire retirement.

Basically, the rationale of the glidepath is that you want to thread the needle, in terms of your asset allocation, to touch two bases in your retirement risk. The first risk is that if you are too meek and too conservative with your asset allocation, you might have very low volatility in the short term, but in the long term, you run out of money because you don’t have high enough returning assets in your portfolio.

On the other hand, if you are too aggressive with your asset allocation, you might have the strong expected returns in the long run, but you might damage your portfolio by so much in the short term.

Because if you start right at the peak before a big bear market, not the 2020 bear market, but let’s take one of the really long-lasting bear markets, like 1929 or the ‘70s or 2001, that took a long time to recover. And you were too aggressive with your asset allocation. Well, you dig into your principal by so much that even when the eventual recovery happens, it’s not going to be enough to recover your portfolio.

Obviously, the stock market index itself will always recover eventually. But your portfolio is not the index because you have taken withdrawals out of your portfolio. And even if the index recovers by 100% or 200% or 300%, it’s possible that at the bottom of the recession, you have already so much depleted your portfolio that even with the eventual index recovery, your portfolio is not going to recover relative to your withdrawal amounts.

And this is really the whole definition of sequence of return risk. The sequence of return risk is that it matters what order the portfolio returns come in. If you have very low returns early on in retirement, and then very strong returns later in retirement, that is a lot worse than the other way around.

And that’s not true for a “buy and hold” investor. At a “buy and hold” investor, if you invest a million dollars, you let it simmer for 20 years.

It doesn’t matter if it’s high returns first or high returns later. It gives you the same overall return for a fixed aggregate average compound return rate, but it’s not true when you take withdrawals out of your portfolio.

Anyways, what the glidepath is doing is it tries to straddle these two risks, where in the short term, right around your retirement date, when you face that maximum sequence of return risk, you have a bit more of the diversifying of the safe asset in your portfolio, but then over time, you phase it out.

And then in the long term, you have the higher returning asset, when hopefully you are past this bump where sequence of return risk rears its ugly head, so to say. And then you have the higher returning asset in the long term, and you have the hedge against the sequence risk in the short term.

It sounds like it is a really totally foolproof way. Of course, it’s not 100% foolproof, but it turns out that in past recessions and bear markets and in the historical simulations, that was a way where you could at least alleviate the sequence of return risk a little bit. You can never completely get rid of it, but it’s definitely going to alleviate it a little bit.

It’s not like you can suddenly increase your safe withdrawal rate from 3% to 4%, or from 4% to 5%, but there’s a noticeable impact on the failsafe withdrawal amounts.

Andrew Chen 12:45
What magnitude of impact?

Karsten Jeske 12:48
I calculated the difference between a fixed asset allocation and then a glidepath, somewhere on the order of magnitude of being able to increase your withdrawal amounts by 5%. So that sounds like a lot.

Of course, if you look at the numbers, say the safe withdrawal rates, that would mean that you’re in the order of magnitude, say without the glidepath, you’re at 3.2%, and then a 5% increase in that.

It’s not 8.2%. It’s 3.2 times 1.05, so take something like 3.36. So we’re talking about order of magnitude of something in the 0.1, 0.15. 0.16 percentage point increase relative to your net worth.

But again, this is always deceiving because a 0.1% increase in the safe withdrawal rate, that can be a lot of money because this is not relative to your budget. This is relative to whole pot of money that you have, which is potentially 30 times bigger than your annual withdrawal.

But again, this is in past simulations. There’s no guarantee that this will work again the same way. Nothing is 100% foolproof, but I think there’s a pretty good rationale.

And the rationale is that when you do these glidepaths, effectively what you do is if you have a bad sequence of return risk event early on in retirement, or the stock market tanks, by shifting weight into equities, you will effectively take your retirement budget out of probably mostly, or 100%, or maybe even more than that, out of the bond portfolio, and potentially even shift some additional money out of bonds into stocks.

And you do that right through the recession and the bear market. And then once the stock market recovers again, because you’re still walking on this glidepath, what would normally happen with the fixed asset allocation, you try to target a fixed asset allocation, then the stock market starts recovering again.

Think back in March 2009 again and you have this huge run out of the stock market bottom. I think you gain something like 70% between March 2009 and March 2010.

With the fixed asset allocation, because the stock market rallies so strongly, you would constantly take money out of the stock portfolio and put it back into bonds. Or you would constantly take money out of the stock portfolio and use it to fund your retirement budget.

Well, you don’t want to do that, right? Don’t run against this momentum, this really strong performance in the stock market. Let that run and take the money, and keep taking money out of the bond portfolio until the market then recovered maybe in 2012, 2013.

Again, this is where that glidepath comes in. By the way, this is probably going to be one of your next questions. How long do you want to do this glidepath?

The glidepath has to last you not just through the recession and the bear market, but it also has to last you probably through a good part of the recovery after that. That glidepath is not something that’s going to be “I’m just going to be an 18-month glidepath out of retirement. The first 18 months of my retirement is going to be a glidepath.”

It has to be much longer than that because you still have to hedge this risk of “I don’t want to take money out of the stock market when it’s still beaten down, but when it’s already recovering and it already has that strong positive momentum during the recovery. So, I still want to have that diversifying asset and mostly take money out of that for my living expenses, and not touch my equities until they’ve really recovered.”

Again, that’s where that intuition comes from in the glidepath: to hitch this equity downturn, and then also a good part of the equity recovery after that.

Andrew Chen 17:25
Yeah. I definitely want to talk about the duration of those glidepaths here next.

But first, if the strategy is basically to get into a safe asset in the years approaching retirement, and then reverse that to mitigate the sequence risk and straddle both those worlds that you described, where you’re reducing volatility in the short term at some tradeoff of long term upside, but you’re not giving away all the upside long term…

What is the percent allocation, at least in the simulations, the historical research – what does it suggest about the percent allocation in bonds or the percent range that is best to target for the peak of the bond tent? Or at least what factors should a retiree consider when determining their target bond allocation for that peak of the bond tent?

Karsten Jeske 18:14
Again, it depends on what method you use. I played around with different peak bond allocations.

In fact, I said, “What if we go 80% bonds and 20% stocks?” That actually turned out a little bit too high.

My sweet spot that I found was you want to have something like 60% stocks/40% bonds, and then slowly shift back into something like potentially 80/20, potentially even 100/0. It turns out that just the way the market data was shaking out in the historical simulations, the 100% equity landing point at the end of the glidepath, that gave you the best results.

Now, is that really something that a lot of retirees want to do? You’re maybe a 65-year-old, and then you realize at age 75, “I’m going to have a 100% equity portfolio.” Maybe not everybody would be up for that.

But the numbers certainly looked like that. And the reason is that, in historical simulations, you have this very intriguing long-term mean reversion pattern.

So, if you look at neighboring windows of 15-year stock market return data, so you plot “This was the average annual return over the last 15 years of the stock market,” and then you plot the same graph, another line, “This is the chart of the next 15 years of stock market return data,” you get an amazing negative correlation.

If you had very poor returns over the last 15 years in the stock market, you’re going to have amazing returns the next 15 years, and vice-versa.

Andrew Chen 20:10
And that tends to be true as you shift year by year over time?

Karsten Jeske 20:14
Yes. And then you look at the time series of that chart. Again, you have to be careful about some statistical significance measures because you have some overlapping windows.

If you just move this by one month, well, it’s still almost the same. You have 179 months overlap out of the 180 months. But it still looks very intriguing.

This is probably the reason why you have this landing point at 100% equities: because if you had very poor returns over the first 15 years of your retirement, if you are at 100% and you do the 100% for the remaining 15 years of your retirement, you’re going to just completely shoot the moon in that retirement simulation exercise.

Again, it doesn’t mean that it has to always work like that, but the economist in me tells me that in the end, everything is connected: GDP growth, earnings growth, the stock market, and the stock market returns. Obviously, they can deviate from each other for very long periods, but then eventually reality comes back and settles in.

This is why you have this very long-term mean reversion. So I think there is a bit of economic theory behind that that almost demands this kind of long-term mean reversion.

It turns out that if you run these glidepath simulations in Monte Carlo simulation frameworks, you will still get an advantage from a bond tent, but you will never want to go to 100% stocks in the Monte Carlo simulations because you don’t have this long-term mean reversion feature in just the naïve Monte Carlo simulations.

And this is just what I wanted to point out, just for full disclosure. I think Michael Kitces and Wade Pfau, they simulated their bond tent and then the retirement glidepath simulations with Monte Carlo simulation data, and they had to tread a lot more lightly with the equity allocations.

I don’t think they ever went up all the way to 100%. I think they did it the other way around. I think their optimal bond tent was much more bonds, much less equities because they don’t have this equity long-term mean reversion.

I think theirs started at 40% stock/60% bonds, and then you walk that up to something like 60/40 or 80/20. I think that turned out their best allocation. But again, under different assumptions, you’ll get different optimal paths.

Andrew Chen 23:31
Got it.

Karsten Jeske 23:31
Just for disclosure.

Andrew Chen 23:32
I wanted to ask about what the optimal duration is for each side of the bond tent. Obviously, the literature doesn’t entirely agree on this. But I want to get your take.

How should investors think about how many years prior to any anticipated retirement date is recommended for starting to execute the bond tent and then concentrate their holdings and bonds?

Karsten Jeske 24:02
That’s a really great question. I haven’t written about that yet, but I’ve actually run some simulations to study this question. And again, it depends on what kind of assumptions you make.

Both the assumptions on the return data that you use, is it actual market data with simulation windows? You look at all the possible retirement cohorts between this date and this date, and they all say for retirement for x number of years? So, that is going to make a difference.

And then there’s also a difference in the optimal glidepaths, depending on what are your personal preferences and parameters? Are you a traditional retiree?

Say you start in your 20s, and then you have a 40- to 45-year accumulation path, and you are going to make relatively small contributions. You do the recommended contributions, something like 10%, or if you want to be really aggressive, you go 15% contributions. Well, then it actually does take you 40 years or so to save for retirement.

And then, on the other end of the spectrum, you have the extreme early retirees. Nowadays, if you tell people, “I want to plan early retirement in 10-15 years,” you’re probably not even one of the extreme people anymore.

The extreme people, they say, “I’m going to do it in 3-5 years. I’m going to save 80% or 90% of my salary.”

So let’s take the typical early retiree who wants to do this in 10-15 years. That person is going to have also a slightly different optimal glidepath.

Then, on top of that, what is normally assumed in these pre-retirement glidepaths is that there’s a very rigid retirement date. So you want to maximize at exactly Date X, and depending on how you model what you actually optimize.

I actually did some of this work in my job, when I still worked in finance, because we were managing a glidepath portfolio, a portfolio of target date funds for a big defined contribution plan, and we were managing the target date funds.

Obviously, we created the research to justify what we’re doing. And that research was based on you maximize some sort of a utility function at that specific retirement date, and then you assume some contributions over your lifetime, and then what is the optimal glidepath that maximizes the utility function at the end of that runway?

That might work reasonably well for traditional retirees, because they say, “I want to retire at age 65 or 67,” and boom, “Exactly at that date, give me what is the optimum glidepath along the way.”

But early retirees, we have some flexibility. If somebody starts at age 30, it’s very unlikely that they say, “Exactly at age 44 years and three months, I want to retire.”

“And exactly at that date, I want to have x amount of money,” or “I want to maximize the objective function,” or “I want to maximize with the probability of having above x dollars.”

So, depending on what your objective function is, your glidepath is going to be different. There’s never going to be the optimal path. There will always be optimal paths subject to certain assumptions.

So, to make a long story short, it turns out that on your path to retirement, if you do Monte Carlo simulations, you want to be a lot more cautious with your equity allocation. There are a few ways around it, but it could be a little bit cumbersome.

But normally, what happens in Monte Carlo simulations is that you have a glidepath where you start your bond allocation. You might start at close to 100% stocks, but then you start shifting down maybe 20 or 25 years before retirement already.

And the reason is that, in Monte Carlo simulations, you have what you call a zero memory process. The Monte Carlo simulation, it doesn’t matter whether your last year’s return was +50% or -50%. The next year’s return is the same no matter what your past year’s return was, which is not true in actual data.

Andrew Chen 29:30
Yeah.

Karsten Jeske 29:32
And because of that, just to hedge the probability that five, seven, eight years before your retirement you have a bad drawdown, you already have to shift way before your retirement. Because if you have a bad event five years before your retirement, you will never recover from that, potentially.

Whereas in the real world data, if five years before your retirement, that was the global financial crisis, by your retirement, you had already recovered from it because you have this strong mean reversion coming out of the recession. And this is something that is not so easy to model and generate in Monte Carlo simulations.

Monte Carlo generates very defensive glidepaths where you start shifting out of stocks way too early, my taste. Whereas actual market data, you can take a little bit more risk with your glidepath because you say, “Even if I have a bad outcome three years before my retirement, I might actually recover from that.”

“Because not only does my portfolio recover, but then the last few years of contributions, I’m going to put them in at the bottom of the 2009 market crash. In 2012, I’m going to look really golden with both the aggregate portfolio and then also the contributions that I put in through this dollar cost averaging through investing through the big bear market.”

And then the other thing is that it depends on whether you’re a traditional or early retiree. If you’re a traditional retiree, you invest only 10-15% of your salary.

If you are 10 years before retirement, you have already contributed a ton of money, and these additional 10-15% contributions are not going to make a big difference. So you have to be really careful about a drawdown right before your retirement.

Whereas if you’re an early retiree and you’re still making these very aggressive contributions two, three, four years before your retirement, they still have a much bigger impact from this dollar cost averaging impact.

That also means that if you’re a traditional retiree, you probably want to start that bond tent. Again, Monte Carlo, you start at 20 years before retirement. With actual market returns, you probably start it somewhere between 5-10 years before retirement.

And then, again, if you’re an early retiree, it would mean that you would start the bond tent right away, right out the gate. And, no, you shouldn’t.

As an early retiree, and your plan may be 10-15 years of accumulation, you probably want at least two-thirds of that. You want to be really aggressive with your asset allocation, potentially have 100% of equities. I ran some simulations where the optimal glidepath is 100% equities all the way to retirement.

If you’re not too risk-averse with your final retirement date, or if you have some flexibility, if you say that “I just barely missed my target at the peak of the bull market, and now we have a bear market, but that bear market is going to be over. And it might not reach the whole peak again after two years, but with my contributions and dollar cost averaging, I might actually make it to my target in the next two years.”

If you have some flexibility, by all means, just stay 100% equities all the way until retirement, which could be the case for a lot of early retirees.

So, there are not that many easy bumper sticker recommendations in personal finance because it’s way too personal. You can’t really make any general statements.

But this might be a general statement that I’m willing to make: If you’re an early retiree, you have a little bit of flexibility over what exact date you want to retire, by all means, you do 100% equities.

Now, if you have a very rigid retirement date, say you don’t particularly like your job, but then you have some very strict vesting cliff, if you work 19 years and 11 months for the federal government and you retire, you don’t get the full pension and you don’t get the health benefits. But then, when you make it exactly to 20 years, then you get all sorts of benefits.

So you can’t work less than 20 years, and you don’t want to work much more than 20 years because you don’t particularly enjoy your job, and you have this very rigid retirement date where you say, “I would die if I had to work a week longer, and I would not get my full benefits if I had to work a week shorter.”

“This is exactly the retirement date I want. It’s not a week shorter or not a week longer.”

If you have that inflexibility, you probably want to start something like a glidepath towards retirement. As a traditional retiree, certainly 5-10 years before that date. As an early retiree with very aggressive contributions, I wouldn’t start that bond tent until maybe 2-3 years before that retirement date.

Andrew Chen 35:11
So I guess the takeaway is it’s very specific on the circumstances: whether you’re early, whether you’re late, what kind of input assumptions you use regarding historical data versus simulated data, and the flexibility of your actual date.

Karsten Jeske 35:32
Correct.

Andrew Chen 35:33
Let’s say you get to that number, whatever that number is, and you execute the bond tent, and then you hit your retirement date. When is it appropriate to start easing back into equities, the reversal?

Does that happen immediately upon retirement or after some period during retirement? If it literally happens immediately, then what is the utility of holding the peak bond tent allocation for a day, for example?

Karsten Jeske 36:04
You would do this very slowly. You do the math. If you say you want to shift out of a 60/40 portfolio into a 100/0 portfolio, you have 40 percentage points to shift.

And now you say that you do a shift of maybe 0.3 percentage points per month, so 3.6 per year, it’s going to take you somewhere a little bit over 11 years to shift from 60/40 to 100/0.

And it’s a slow process. Again, it has to be a slow process because you want to capture not just the bear market, but also the potential bear market right around the corner, and then also the bull market that follows it.

Again, you could just play this completely rigid. You just say, “No matter what, I’m going to do that shift by 0.3 percentage points.”

First of all, you’re going to have some drift in your asset allocation weights anyways, so you could just say, “I’m just going to withdraw from the portion of my portfolio that is above target.” And then you also have to make a decision. “If I’m still not back to target, do I do any additional rebalance?”

There is some question about that. You don’t have to do this really every month, that rebalance, but maybe do it once a quarter or once a year.

Again, there is the utility of having that glidepath because you have a higher safe asset allocation right around your retirement date, right when you start making withdrawals. So, that intuition that I brought up earlier, that comes in. You eventually will come back to 100% equities, but that is such a long path, and it does make a huge difference in your retirement success.

Now, the other question is: do you really do that shift blindly, no matter what the market does? Because the worst thing that can happen is that you do this glidepath shift over the next 10 years, and the market just does nothing over the next 10 years, and it’s just going sideways.

Bond market is not yielding much. The market goes sideways. And then right when you have 100% equities, that’s when the big stock market crash happens.

That would be, obviously, the worst possible outcome for that glidepath scenario. So then you gain nothing in the short to medium term, and then you lose a lot in the long term.

Or it could go just like: what if the stock market just keeps chugging along, and one year after the other, we get 8%, 9%, 10%, 12% returns in the stock market? Why don’t you just let it run? Or why do you shift then more and more into equities?

The market is already overvalued, and you shift more into equities as they become overvalued. And it just means that the eventual crash is going to become worse and worse.

So, you could also argue that as long as equities are still performing really well, maybe you don’t even start the glidepath yet. You just leave it at 60/40. Just leave it there for a while, and then you see the stock market crash happens, and then you start the glidepath.

Andrew Chen 40:04
I guess that’s really interesting because it means that as to when you start the reversal is dependent on what’s going on at the time.

Right now, the CAPE is, I think, over 33. If your retirement date was now, maybe you want to hold on to that 60% bond allocation just a little while longer as a defensive move. But if it was right after the financial crisis in 2009 when the CAPE is 15, maybe you want to actually initiate sooner, and maybe even do it faster, for example.

Karsten Jeske 40:37
Right. Or think about it. You’re already 10 years into retirement, and there wasn’t a big market crash.

It’s possible that over the next 10 years, my forecast is we’re going to have lower equity returns than we became used to over the next 10 years. It doesn’t mean that we have a crash right away. It could just mean that the market is going to muddle around.

It’s not going to be very strong. It’s not going to be very weak. It’s just mediocre returns over the next 10 years.

But you look at your portfolio and say, “Mediocre returns is all I needed.” Because now you’ve already run down the clock by 10 years, especially if you’re a traditional retiree, and you look at your portfolio.

So, you start with a million dollars, and you still have $900,000, and you have only 20 more years to go. Why do I want to be 100% equities now?

You won the game. Why do you want to gamble on some fantastic returns over the remaining 20 years? Why not just keep it at that and really keep that glidepath only as an option where you run that glidepath and you go to 100% equities 10 years into retirement really only if you had this very bad market event in the short to medium term and you really need the 100% equities and you bet on the recovery?

Because maybe in 10 years, the stock market got clobbered. The CAPE is in the single digits. Then you absolutely want to be at 100% equities if equities are that cheap.

So, it might be that you want to use the glidepath more as an optional tool and not really as a very rigid asset allocation.

Andrew Chen 42:37
In one of your articles, I remember you wrote that a successful glidepath should ratchet up the equity share and reach maximum equity weight roughly over the length of one complete bear plus bull market. But the challenge is that you don’t know, in the moment, where you are in that cycle and how long that cycle is going to be.

And so, as an early retiree is thinking about “How many years should I plan for my reverse glidepath back into equities to take after retirement?” I was curious, are there techniques that investors can use to translate metrics or statistics that they can see on the ground right now into an estimate for how long the glidepath should appropriately take?

Does that make sense? Can they look at things like the CAPE, other indicators, and back into a glidepath duration that at least makes it a little more quantitative rather than saying “The CAPE is 18, but I’m not sure where we are in the bull plus bear cycle, so I guess I’ll just throw a dart on the board and say that my glidepath should be x years”?

Is there a little bit more quantitative way to do that?

Karsten Jeske 44:02
I think it’s hard if you just look at the CAPE ratio.

The really bad example of sequence risk is always the 1960s. You had a high CAPE ratio. I think it was in the 20s, which was, at that time, relatively high.

And you never had this very strong bear market early on in retirement. A lot of people are surprised that 1964, 1965, 1966, I believe, and 1968 were some of the worst times to retire, by some measures, and sometimes even worse than 1929.

Because if you look, was there a big stock market drop in 1969? No. It was okay.

The market muddled around. And if you had started withdrawing, you would have been maybe a little bit underwater in 1973 when that recession and bear market hit. So that was the danger of 1965.

And there’s no indication that in 1965, the glidepath that you would have needed would have actually been a 20-year glidepath, if you think about it, because the glidepath would have to cover the 1970s recession, and then ’80 and ’82, and then still a little bit of the recovery, because we had a very strong recovery following 1982 in the stock market.

And there is no way anybody could have ever guessed that it’s going to take that long to dig yourself out of that hole. And the nasty thing about that event was that the drawdowns were not even that severe, but there were one after the other, and it was a very long-lasting event.

Whereas one of the least dangerous sequence of return risk events was the global financial crisis. You had a very sharp downturn, a very sharp recovery.

From a sequence of return risk point of view, that was one of the milder events. Nobody who retired at the peak in 2007 really got into trouble.

Of course, in real time, they might have believed at the bottom in 2009, “Oh my god, this looks really scary.” But after the fact, you look at this and you say, “Wow, this was such a swift recovery.”

The event was so short that the sequence risk didn’t really have such a big impact. Because sequence risk is basically a product of two things. One is the depth of the recession, and one is the length of the recession.

If both the depth and the length are really bad, that’s the worst sequence risk. That’s why 1929 is a bad event because you had a bad drop in the stock market, and it also lasted for quite a while. 1965 was pretty bad.

2007, 2008, 2009 was not even the worst from a sequence risk point of view. But then again, you look at the CAPE ratios. I think the CAPE ratio right around 2007 was not too different from the one in 1965.

And then between 1965 and 1929, 1929 had a very high, sharp ratio. It was also above 30, much higher than in 1965, yet some of the retirement simulations will show you that 1965 was actually a worse event than 1929. And then, on top of that, for example in 1929, the glidepath worked relatively well because bonds and stocks had a very negative correlation during that event.

So, not only does it mean you have a lot of bond allocation. You reduce your equity risk twofold. One is by having a lower equity weight.

And then, on top of that, what do you shift into? Your bond portfolio has a negative correlation. The bonds rallied during the 1930s.

So, that was another really nasty feature of the 1970s and 1980s that these were inflationary recessions. You had stagflation. Not only did your equity portfolio get hammered, but also your bond portfolio.

So, the success of the glidepath would be diminished somewhat if you have this inflationary shock. Some people make the case that this might be in store for us now because inflation has been relatively low and tame for a long time. That just means that the next recession is going to look more like the 1970s rather than 2007 and 2008.

Again, I haven’t done any super careful research on what kind of a glidepath design and length you should plan conditional on observables. But if I had to, I would look at both equity and bond market fundamentals, not just the CAPE.

Andrew Chen 49:58
You mentioned that the data suggests pretty robustly that 15-year adjacent windows tend to be quite nicely inversely correlated. And I was wondering, are there heuristics like that? Or maybe it is even that one coupled with other observable data where you can at least make an educated guess.

Or are other external variables just too varied that that wouldn’t even be reliable, for purposes of determining the duration of a potential glidepath?

Karsten Jeske 50:45
Again, the reason why these 15-year windows are negatively correlated is it’s a valuation story. You had very poor returns over one 15-year window. You look at why are the returns so bad?

Part of that is that probably the earnings multiples got really badly hammered. And then, over the next 15 years, the earnings multiples normalized again.

A huge input in that negative correlation comes simply from mean reversion of valuations. That has to be the number one explanatory variable.

You’re obviously on the right path. You look at the CAPE. The higher the CAPE, the more risk we have of a negative equity market event right around the corner.

Again, nobody can ever time that. I would say the most dangerous time is when all the people that have been warning all the time about the high CAPE ratio say, “I’m actually convinced now that high CAPE ratio no longer matters.” That’s probably a very dangerous time, which I don’t think we have reached yet.

For example, recently, Shiller said, “We probably shouldn’t be too worried about the high CAPE ratio because interest rates are also really low.” It’s less of an issue.

If you look at the absolute number of the CAPE ratio, that’s really high. That means if you look at the inverse of the CAPE, it’s something of an earnings yield.

And then you compare that yield to the bond yield, whether it’s the cash, the very short-term interest rates, or even the 10-year interest rates, or even 30-year interest rates. The gap between earnings yields and bond yields is actually relatively normal.

So, maybe this would call for not necessarily a bad crash right around the corner. As we saw, for example, in 2000, the CAPE was really high, but bond interest rates were also extremely high.

That means that over the next 10 years, you have lower equity returns, which we did. But the lower equity returns, most of it was because you had a really nasty crash right around the corner.

So, somebody could argue that maybe because bond yields are also so low, we have lower equity returns, but the lower equity returns are going to come in the form of just lackluster equity returns. That could be one theory to look at the CAPE not just in absolute terms, but also relative to some other valuation measures and look at the relative yield of stocks versus bonds.

Andrew Chen 53:54
I guess while it may be difficult to predict whether the crash is right around the corner, one could probably make a little bit more confident predictions over a forward 10-year timeframe, for example. Is that fair?

Karsten Jeske 54:13
Yeah. And this is all people ever claimed how equity valuation works.

Nobody claims that equity valuation tells you much about what’s the equity return over the next week or month. This is all about longer-term mean reversion.

I’m an economist by training, and that definitely bleeds into my beliefs on how I practice finance and how I used to practice finance when I worked in the industry. People always make fun of these valuation people and say, “These people, they don’t know what they’re doing.”

You look at the correlation between one month equity returns and the CAPE, there’s nothing there. So this really only comes in over the very long term. And it’s not really for timing the next stock market crash, but it’s basically for timing the first 10 years, not the next 10 days.

Schedule a private FIRE consulting session

Schedule a 1-2 hour session with me to analyze and plan your path to FIRE

Schedule Now