Bogleheads® Live

Derek Tharp on Retirement Planning - 4% Rule vs. Monte Carlo

March 27, 2023 Derek Tharp Episode 41
Bogleheads® Live
Derek Tharp on Retirement Planning - 4% Rule vs. Monte Carlo
Show Notes Transcript

[00:00:00] Jon Luskin: Bogleheads® Live is our ongoing Twitter Space series where the do-it-yourself investor community ask their questions to financial experts live on Twitter. You can ask your questions by joining us for the next Twitter Space. Get the dates and times for the next Bogleheads® Live by following the John C. Bogle Center for Financial Literacy on Twitter. That's @bogleheads. For those that can't make the live events, episodes are recorded and turned into a podcast. This is that podcast.

[00:00:29] Thank you for joining us for the 41st Bogleheads® Live, where the do-it-yourself investor community asks questions to financial experts live. My name is Jon Luskin, and I'm your host.

[00:00:43] Today's topic is the problem with retirement planning and whether those doing retirement planning should be looking at using Monte Carlo simulations or keeping it simple with a historical, safe withdrawal rate approach. For today, we'll have Derek Tharp, lead researcher at, answering your questions about retirement planning. Derek joined us for the 24th Episode of Bogleheads® Live, where we discussed spending in retirement and how it changes over time. Folks can check out that podcast episode wherever you get your podcasts.

[00:01:15] Let's start by talking about the Bogleheads®, a community of investors who believe in keeping it simple, following a small number of tried-and-true investing principles. This episode of Bogleheads® Live, as with all episodes, is brought to you by the John C. Bogle Center for Financial Literacy, a 501(c)(3) nonprofit organization dedicated to helping people make better financial decisions. Visit to find valuable information and to make a tax-deductible donation. You can also go straight to

[00:01:46] Before we get started on today's show, some announcements. For the next Bogleheads® Live, we'll be discussing college planning. Anne Garcia, author of “How to Pay for College” will be our guest. That'll be Thursday, March 30th at 12:00 PM Pacific, 3:00 PM Eastern. 

[00:02:00] Before we get started on today's show, a disclaimer. This is for informational and entertainment purposes only. It should not be relied upon as a basis for investment, tax, or other financial planning decisions.

[00:02:11] Derek, thank you for joining us today once again on Bogleheads® Live.

[00:02:15] Derek, let's do a little bit of retirement planning 101. We had Bill Bengen on for Episode #35 recently. He said you can take 4% or 4.7% out of your portfolio over 30 years sustainably.

[00:02:30] So, if you're not an investment nerd, that means you can take out $40,000 out of a $1 million portfolio for 30 years. At least, that's what you were able to do historically. That's what had worked in the past. That's a pretty simple and easy thing to understand. 

[00:02:46] Tell us about what is a Monte Carlo simulation? How does that work? 

[00:02:50] Derek Tharp: When we think about the historical analysis, I like to start there because looking at something like those historical safe withdrawal rates is nice because you're looking at real world markets. How did markets actually behave? What could somebody have spent at least historically and made it through retirement without depleting their portfolio?

[00:03:10] Monte Carlo? I think of that as kind of an evolution. That was, again, to answer this question of, okay, well what if things are worse than they were in the past? Or what if we know we've only seen a very limited number of historical scenarios, and certainly we know the future is going to be very different than the past was. 

[00:03:28] And so there we really introduce some additional randomization. Typically, different software will handle it differently. In the simplest version, you might think of just having a certain return assumption for a portfolio and you have a normal distribution and then you're drawing just randomly each year what the returns were and then simulating somebody going through a retirement. And you might do that 250,000 times and see what percentage of the time could somebody make it through retirement without a running out of money. And assuming they are sticking to one particular spending strategy. 

[00:04:03] In that way, it's kind of similar to the historical safe withdrawal rate framework because with the safe withdrawal rate framework, you're basically saying, I'm going to start my spending. I'm going to give myself an inflation adjustment every year, I’m just going to blindly charge ahead and see what that minimum withdrawal rate was that I could have taken without depleting the portfolio. 

[00:04:24] And in this case, same type of thing. You're setting a certain spending level and then take an inflation adjustment, run that through retirement and then see where somebody gets that probability of success somebody had.

[00:04:39] Jon Luskin: Jon Luskin, your Bogleheads® Live host jumping in for a podcast edit. Let's see if I can say what Derek said, in a slightly different way. 

[00:04:50] With the Monte Carlo simulation, we're asking a computer to forecast the success of our retirement plan. And here's how it does that. Let's say we start with a $1 million portfolio. In that first year, we're going to start by taking out $40,000 (that 4%) for living expenses. And then the computer is going to guess about the investment return we'll get on that portfolio after the first year, let's say 7%.

[00:05:19] So we had $1 million, we took that $40,000, we’re down to $960,000. We had that 7% investment return. So, now we're $1.02 million and change.

[00:05:29] Well, now it's another year and we're going to take out another $40,000, but we'll bump that up a little bit for inflation. So, those can be $40,000 plus. So now we're at $986,000 before our investment turn. And then after our investment return on the second year we're down to $887,000 because we had a bad year on our investments that year and we lost 10%. And then we're going to take out another year's distribution - $40,000 adjusted for inflation - and now we're down to $844,000. 

[00:06:04] The computer is going to keep doing that year after year after year for 30 years, maybe a little bit longer, maybe a little bit less, depending upon what we ask the computer to do. And it'll keep doing that until either we run out of money or we run out of life. 

[00:06:23] Once the computer does that simulation, it's going to do that for 250,000 or 100,000 more times. Depending upon the investment returns in each of those simulations, in some of those simulations we will run out of money first. And then some of those simulations will run out of life first. The percent of times that we don't run out of money is the Monte Carlo success rate.

[00:06:50] Now that we've nerded out on what exactly a Monte Carlo simulation is, let's turn it back to Derek to hear the pros and cons of using a Monte Carlo simulation.

[00:07:04] Derek Tharp: Now, the thing I do like about Monte Carlo is that it allows you to break some of the assumptions that are frankly unrealistic about the safe withdrawal rate approach. So, the safe withdrawal rate approach, now you're taking one distribution rate, you are getting a true inflation adjustment along the way, which we know may not actually describe how retirees move through retirement because it varies across time. That's one way that you could easily build that into a Monte Carlo simulation. You don't have to use just the fixed distribution that you do in the safe withdrawal rate. Of course, you can modify the historical analysis as well.

[00:07:42] Another thing you can do though is introduce additional goals. So, let's say that somebody knows they want to fund a grandchild's wedding in 10 years. Or maybe somebody wants to plan for placing vehicles or other expenses that are a little bit more lumpy. A cash flow that it may not be as smooth, where you have a sudden jump in income. With a Monte Carlo simulation, you can put your exact intended cash flows, and it is going to account for the unique characteristics of your spending pattern.

[00:08:10] Those are some of the things I like about Monte Carlo. There's still plenty of things I don't like about that approach though. And in particular the fact that it's really not a dynamic type of approach. You're not saying, "how can I adjust my spending over time?" Or "how do I try to plan for adjustment in mind?" You're still saying, "here's my spending pattern. I'm going to spend this way. I'm going to charge forward blindly and just see where this gets me."

[00:08:36] And the reality is, when we hit some of those poor sequence of returns, we know we're on a negative path and we can make some adjustments and get back on track.

[00:08:44] Jon Luskin: So, Derek, it looks like with that historical simulation, with that historic performance, we can look at the past and see it's worked. And with the Monte Carlo simulation, we can ask a computer to tell us if our particular spending plan is going to work or not.

[00:08:59] There can be quite a difference between the results of those two approaches to retirement planning. As mentioned, I had Bill Bengen on recently speaking about safe withdrawal rate in retirement, and that safe withdrawal rate by his research could be as high as 4.7% when including microcap stocks as part of your portfolio, up from his initial 4% safe withdrawal rate.

[00:09:20] Now, that's historic performance. You can compare that to one Monte Carlo simulation that was put together by Christine Benz of Morningstar, and her most recent research shows that you can spend 3.8% on a $1 million portfolio that is going to be $38,000.

[00:09:37] We've got a difference in how much you can spend sustainably when we looked at historic performance versus what a computer is going to guess that we can spend. Tell us a little bit more, Derek, about why there's such a difference in these two approaches and what folks should be thinking about when comparing the two of them.

[00:09:56] Derek Tharp: One important caveat is that anytime we're talking about Monte Carlo simulation, that's just a methodology. And one of the very common ways that people might make some of those changes to that methodology is the capital market assumptions.

[00:10:09] Jon Luskin: Derek mentioned capital market assumptions, and for those who aren't investment nerds, that just means, "what do I think stocks are going to return? What do I think bonds are going to provide in an investment return over the next 20, 30, 40 years?" This is our way of telling the computer, "Hey computer, my capital market assumptions, I think stocks are going to do X return in the future."

[00:10:33] Derek Tharp: Putting those capital market assumptions in, you can change some of those assumptions and get very big difference in outcomes. That's probably one of the most important things that's going to drive that big gap.

[00:10:45] Jon Luskin: A lot of folks, especially in the FIRE community, that early retirement community, really like that 4% rule of thumb approach. They take and say, "Hey, I want to spend $80,000, so I need to save $2 million."

[00:10:58] What's the problem with taking that approach? 

[00:11:02] Derek Tharp: There's actually a pretty big problem with that approach, and it's really how variable portfolio balances can be over time. 

[00:11:11] Let's go back to 2007 or 2008; somebody who had $2 million, and then all of a sudden we see what happens in the market and they've got much less than that. What do they do in that situation? 

[00:11:24] The 4% rule would say you basically just ignore whatever's going on in the market, you can continue to give yourself an inflation adjustment and stick with that. But to me, there are some times when we know that maybe the risk is higher or lower, based on current market valuations. What we tend to see is that, when we've had an extended bull market and we've seen asset prices go up, now the risk of a decline is actually higher. And at the same time when we've gone through a more negative time in the market, then the forward-looking return expectations might be higher. 

[00:11:59] I like to plan for spending strategies that aren't going to be really sensitive to reaching some particular number and instead planning for adjustment and planning for the fact that maybe we do see markets are very high at a given point in time, but somebody's spending level will be resilient to even the downturn in the market.

[00:12:19] Jon Luskin: Let's talk about one of your articles on, you co-authored this with Justin Fitzpatrick, entitled "Evaluating Retirement Spending Risk: Monte Carlo versus Historical Simulations." 

[00:12:33] Derek Tharp: Should we use Monte Carlo or historical simulation? What should we be aware of if we're using those?

[00:12:40] We had really some interesting findings because a lot of people out there are under the impression that the reason you do Monte Carlo simulation is so that you can model something more extreme, be more conservative than what has happened in history. And while certainly true at the extreme, say 90% and above, what was really interesting was when we got into the range 90% to 70%. That is an area where actually what we saw was Monte Carlo was more aggressive in suggesting spending levels than historical simulation would've been.

[00:13:15] I should say that we were using just historical Monte Carlo capital market assumptions. Of course, if somebody's baking conservatism into their capital market assumptions, that's an entirely different story. But for somebody who's just using kind of historical parameters, actually we found that, interestingly, Monte Carlo was not necessarily the more conservative approach. In fact, it was a more aggressive approach for some of those time periods.

[00:13:40] Jon Luskin: You mentioned with respect to Monte Carlo simulations, and I'm quoting you, “weird things happening at the tail.” Christine Benz made a similar comment when discussing her Monte Carlo analysis in her research showing that 3.8% sustainable distribution rate, which has a 90% rate of success. And she suggests that you probably don't want to go higher than that because that would mean spending a lot less. Can you tell us about why when using a Monte Carlo tool going above 90% probably isn't reasonable? 

[00:14:16] Derek Tharp: If you think about, particularly as you're running more and more iterations as part of the simulation, the outcomes at the tails tend to be the least realistic. Those are going to be very good outcomes because you just hit positive year, positive year, positive year, positive year. If you're flipping coins, right? It's the equivalent of you got 20 heads in a row in one of the simulations. Or it could go the other direction where it is just negative year, negative year, negative year, negative year. Again, those tend to not be very realistic. 

[00:14:44] What we don't see is the mean reversion because most tools out there don't account for that. And so now you still have the exact same probability of a good year or a bad year after being down those more extreme paths, when in reality that's just not realistic.

[00:15:01] Jon Luskin: Stock market returns historically have mean reverted. They haven't gone on to provide great investment returns forever. And they certainly haven't gone on to provide poor investment returns for forever either. They mean revert; taking turns over time switching between a series of good and poor investment returns.

[00:15:26] Alright, Richard, you should be able to ask your question to Derek. 

[00:15:30] Richard: Thank you Jon, for hosting this. It's been informative, and thank you Derek. 

[00:15:33] I'm currently retired, I'm going to be 57 years old this year. Ideally, I'd like to wait until 70 to start taking Social Security. So, it's like a 13-year timeframe and then God willing, I live from 70 to 90, let's say. Given I have a 13-year time horizon, what is the safe withdrawal rate for a 13-year timeframe for somebody who has a very high equity position?

[00:16:04] Derek Tharp: Great question. I think you're actually kind of getting at a concept I've referred to before in another Kitces article I wrote with Justin Fitzpatrick, where we use the term “retirement distribution hatchet.” And that's because really, again, one of these limitations of something like the 4% withdrawal rate that is using just one distribution rate, just kind of assumed across retirement, is that many retirees who do retire before age 70 claiming Social Security do tend to see we call a hatchet shape of distributions. 

[00:16:38] Think of a hatchet laying on its side where you have the blade of the hatchet and you have those higher distribution rates in the earlier years and you get out to age 70, or wherever somebody defers Social Security and then their distribution rate from their portfolio actually drops significantly. And then we get more of the handle of the hatchet that extends out. And so, oftentimes if somebody is delaying Social Security where they retired an age where there was a gap there that they could, the need to pull from the portfolio is higher in those earlier years.

[00:17:09] In terms of exactly what that safe withdrawal rate would be, really varies so much based off of dynamics of the hatchet itself. How much of somebody's distribution that they need is their Social Security benefit? But I've definitely seen scenarios where somebody could be pulling more heavily; 10%, 13%, 15% of a portfolio in those early years. And then Social Security kicks in and then maybe they're taking 1% withdrawal rates going forward. 

[00:17:38] Again, it really does vary based on the dynamics of a given plan. But that's where I really like a risk-based guardrails framework, because it really does help address that concern because you're taking all the individually known factors and you're actually solving for a spending level that it is going to be sustainable for somebody under those unique circumstances that they face.

[00:18:03] Taking into account how many years they're in that delay phase - because somebody who's 57 is going to have a lower distribution rate - they could take delaying until 70 than somebody who's 65 and only has five years delay until they're 70.

[00:18:18] I don't have a good number for a 13-year period exactly. And again, that's partially because I also have to know where it's stepping up to. The way I personally look at trying to address that is using a framework, like the risk-based guardrails framework, that will account for that no matter what somebody's distribution pattern is.

[00:18:36] Richard: Standard rule of thumb is for 30 years, if you're 60/40 at 4% safe withdrawal rate, my spending needs are going to change dramatically at 13 years. But the stock asset allocation, I was just wondering maybe you would be able to refine that or had any input on that. 

[00:18:54] Derek Tharp: On that topic, so another thing that's unique, and one thing to be cautious about when you're comparing the fact that yes, over a 30-year period, a 100% stock allocation might lift that safe withdrawal rate. There's maybe some unique dynamics that come in when we're talking about the short-term distribution phase. Because in a way we're actually front-loading a lot of the sequence risk when we do that. With an all-stock portfolio, especially with a more aggressive early distribution strategy that's then going to decline, I'm not sure that you would have that same degree left in the safe withdrawal rate that you would have, because of front loading some of that sequence risk or condensing it, rather than spreading it out a little bit more with the lower distribution rate over time.

[00:19:46] There's going to be so many individual specific factors to that. But certainly yeah, you do want to account for whatever the asset allocation assumption is in somebody's own analysis.

[00:19:56] Jon Luskin: If you want to learn more about what is the right stock/bond mix when you're taking money out of your portfolio in retirement, check out Episodes #37 and #35 of Bogleheads® Live. Our guests, Christine Benz and Bill Bengen respectively, and their answers to that question. Roughly 60% stock allocation. That's going to be that sweet spot for the right mix of stocks and bonds when taking money out of your portfolio. 

[00:20:25] All right. Hi 10, I'm going to make you a speaker. 

[00:20:28] Hi 10: If you don't have a financial planner yet, and they have the proprietary tools for running the Monte Carlo simulations, do you have any recommendations on online or free tools that could help one run a Monte Carlo simulation? 

[00:20:43] Derek Tharp: Great question. I know I've used a few of them before. There are some Monte Carlo simulators out there that are free.

[00:20:48] Just generally speaking, I would say there's probably going to be some limitations to those. Doesn't mean that they're still not useful or you can't get by with using those tools. But things like tax assumptions, things like maybe different features. There are some tools now that, on the more professional advisor side that can do things like easily incorporate Blanchett’s retirement spending style. So, if somebody wants to do that, they can do that with a click of a button. Whereas maybe that would be more manual entry on a different type of tool.

[00:21:17] And if somebody's inclined, you can also build your own Monte Carlo type planner in Excel. And just using random number generator functions and putting in your parameters and setting up distributions coming out over time. And then do that over enough iterations. 

[00:21:31] Personally, what I've wanted to do some just very simple types of analyses and look at, say, some particular dynamic, I've often built some of my own models and used tools like Excel to do that. You're usually though giving up something around all the nuances. Where when I build my own tool in Excel, I'm not capturing the sunsetting of the Tax Cuts and Jobs Act, tax rates changing in the future and some of those things that the more professional software will handle. But maybe I don't need that, depending on what question I'm asking and why I'm using the tool.

[00:22:04] Jon Luskin: On the Bogleheads® Wiki there is the Retiree Portfolio Model. Another place that folks can check out DIY retirement planning is going to be Rob Berger. He's written a lot on the subject, and I'll link to that on the show notes.

[00:22:22] Let's jump to a question from Twitter. This one is from George Manka who asks: "How do you balance sequence risk against the need for growth, longevity risk in retirement portfolios?"

[00:22:35] Derek Tharp: Great question there. I think where I come back to is that when I'm really trying to weigh those different factors, the approach I like best of all is to use something like risk-based guardrails that are going to actually account for that unique profile somebody has. Because actually it can be a little underappreciated how different somebody's sequence risk might be. 

[00:22:57] You might have one retiree who's retiring at 65 and taking Social Security right away and they're going to have a different sequence risk exposure than somebody who is retiring at 55 and not taking Social Security until 70. Depending on what percentage of somebody's income Social Security is going to be, now that's another factor that influences how sharp of a contrast we're going to have in that first distribution period versus the second distribution period.

[00:23:24] Thinking about how to balance those, I do think that really comes down to doing a projection that in a sense is custom and unique to somebody's own circumstances and the distributions they want to take because then you're at least starting from a more accurate picture of what somebody's sequence of returns risk exposure is. And then from there you can look at different portfolio considerations. 

[00:23:49] Other things are going to be factors too, just in terms of how close is somebody to spending near the upper end of their reasonable or sustainable spending level would be. And I think that's another factor that gets glossed over sometimes because there's some retirees that I talk to who, they're very frugal. They've always been very frugal. They probably have a hard time spending money. And when we run the projections for them, they can get by pretty well on just their Social Security income. And that's a scenario where that person doesn't have much sequence risk. At least I'm assuming there's not a very prolonged period before claiming Social Security.

[00:24:26] Let's say they're 70 or they worked until 70, they're retiring. They don't really need any of their portfolio income. They don't really have much sequence risk to worry about. There, the growth dynamics are very different. And, I do think there's, and depending on what somebody's objectives are, how much of a legacy they want to leave, what they want to do with their money while they're alive, definitely there I think might be more variation than would be ideal than we currently see with the very set it and forget it 60/40. 

[00:24:56] I'm very much in the 60/40 camp of it is the portfolio in my opinion that really weathers the broad range of different retirement scenario well. But when we get to those individual factors, there are different strategies. Even things like, Michael Kitces and Wade Pfau have written about a rising equity glidepath where somebody is - because of sequence risk - their asset allocation starts out more conservative at retirement and actually gets more aggressive over time.

[00:25:27] Maybe they're just going to let that stock exposure drift up. And oftentimes once getting through that first decade or so of retirement, a lot of times people, if they got through that pretty well, they're in really good shape going forward and they could, if they wanted to, allow themselves to have a more aggressive approach and they're probably not going to personally spend that money. But that's more money that they could give to charity or family or do more.

[00:25:51] Even if they want to spend it, they could do more family trips, whatever that might be. I'm personally in that camp of, I think that's a strategy that doesn't get a whole lot of attention, but is really worth consideration for someone with the right set of circumstances.

[00:26:03] Jon Luskin: This one is from username ‘WoodSpinner’ who writes: “I would love his take in using VPW or variable percentage withdrawal to help calculate a reasonable spending target for the current year, future income, Social Security, pension, expected returns. Looking for a different approach than a safe withdrawal rate-based spending plan.”

[00:26:26] Derek, what is your take on the variable percentage withdrawal, which for folks who aren't familiar with it, this effectively means, hey, if the portfolio is doing better and we've got a shorter timeline, then we're going to spend relatively more. But if we've got a longer timeline, the portfolio's not doing as well, we're going to take out a little bit less. And folks can check out the Bogleheads® Wiki to learn more about that. And I'll link to that on the show notes.

[00:26:55] Derek, what's your take on that approach? 

[00:26:57] Derek Tharp: Great question. I did take a look at that on the Wiki, the spreadsheet that's there to play around with. And I'd say my initial reaction is there's a number of things that I do like about the framework. I do like that it is allowing for that increasing withdrawal rate over time. Because that's one complaint I have with something like a more fixed guardrails approach where the target is 5% as kind of a spending level or it's something like the Guyton-Klinger. 

[00:27:24] But I do like the fact that it does to some degree accommodate the reality that you can take a higher percentage of a portfolio when you're 85 than you can when you're 65. There are definitely some aspects of it that I do like. Gets a little bit more realistic to what people might actually spend in the real world.

[00:27:43] I played around with just some basic assumptions. So, I started with a $1 million portfolio, took my recommended withdrawal, and then I simulated a 20% portfolio decline. And I looked at what the spending reduction would be the next year. Again, based on the set of assumptions I used, it was something like a 16.4% reduction in spending, which if somebody truly has that much flexibility in their budget might be fine. But I find for many retirees, I don't think that's really the case. I think that's a little bit more volatile than many people would like.

[00:28:16] I keep going back to the risk-based guardrails here, but within the risk-based guardrails framework, you can actually use different strategies or you can set different parameters around those risk levels so that you could use a strategy that has a very low likelihood of ever calling for a decrease in spending and a higher likelihood of saying you could increase your spending. Or you could say maybe I want to spend a little bit more today and I'm okay with the downside potential. 

[00:28:40] But that was really a factor that based on my quick overview of the tool unit, there's kind of just the one set of assumptions. You can't really tailor that to somebody's own preference for more or less volatility. It does a better job than some of the very simple frameworks, but I like a risk-based guardrails framework where I can actually tailor the guardrails parameters so that I am choosing the right level of downside risk.

[00:29:06] Jon Luskin: This question is from username ‘AlwaysLearningMore’ who writes: "Does he have any thoughts on the potential usefulness of contingent deferred annuities?"

[00:29:16] Derek Tharp: I do think annuities can play a role. When you look at some of the work that Wade Pfau and Michael Finke and others have done in terms of you can incorporate an annuity in.

[00:29:27] I do take the perspective of how a more pension-like income annuity, whether it's immediate or deferred, you can really think of those as super bonds. You can earn bond return with a bond-like risk exposure, but also a little bit more because of the mortality credits.

[00:29:43] Instead of thinking of it as a backup system, I just think of it more as built into somebody's portfolio. And if somebody wants to potentially try and reduce some of that longevity risk through something like a QLAC where they're getting a deferred income annuity, I think that can be something to definitely look at.

[00:30:02] I tend to look at it as, okay, here's the portion of our bond allocation now shifted to this QLAC and going to generate income at some point down the road in the future. That should help from a longevity perspective.

[00:30:15] And then the nice thing, again, going back to Monte Carlo simulation, you can model that in. You can actually put in that QLAC and you can see what that does to the actual outcomes of a plan and see if it's adding a tool like that or any other deferred annuity. Is that something that appears to be beneficial to somebody's overall income plan and whether goals around potential reduction in spending, those types of things.

[00:30:40] I think that is a tool that should be in somebody's toolbox and they should be paying attention to it. That's more the framework; I like to approach it still as part of the portfolio. And just overall, what does this do to the plan? Rather than think of it more as a backup in terms of a portfolio declining to a certain level.

[00:30:57] Jon Luskin: "John M" you should now be able to ask your question to Derek on retirement planning.

[00:31:01] John M.: Thanks. I've actually two questions. One is thinking along the line of FIRE, what kind of safe withdrawal rate do you think for going out 50 or 60 years instead of a 4% rate for only 30 years?

[00:31:15] Derek Tharp: To answer the first question with the whole kind of FIRE type of framework, what's kind of surprising in a lot of the safe withdrawal rate research is that the reduction once you get out past the 30-year retirement horizon, the safe withdrawal rates don't reduce as much as some people would think. It's not like it goes all the way down to a 3%, 2%, 1%. It's not that type of dynamic. So, it does actually to some extent level off. 

[00:31:41] Really, I think the 4% as a really rough rule of thumb can still be useful. There's still the fact too that the 4% is, in my opinion, quite conservative for reasons Bill Bengen and others have noted. There are rational ways to maybe bump that up a little bit and particularly if someone is willing to make adjustments.

[00:31:57] Jon Luskin: I know that Bengen reached a conclusion with respect to, hey, my distribution period gets longer. That's to say I'm going to have a longer retirement. How much less can I take out? Well, the difference isn't that substantial. For example, in looking at Bengen's original research where the safe withdrawal rate was 4%, over a longer period Bengen says that distribution rate gets decreased to 3.5%. So not a huge difference for those longer timelines.

[00:32:24] John M.: And my second question is the bond portion of an asset allocation based on something like the 4% rule? In other words, if you have 40% in bonds, that would last a decade at 4% without tapping into the 60% of equities. So, if someone needed to take only maybe 2% per year, a 40% bond side would last 20 years. So, wouldn't it be equally appropriate for them to set aside maybe a decade at 2% being 20% for their withdrawals, or is that going out too far?

[00:32:54] Derek Tharp: I do personally like to actually take an approach very similar to what's described there in that I think of an allocation as to broad buckets, stocks and bonds, bonds are going to be more stable. Not perfectly stable, but more stable. And really trying to look and see, in somebody's situation, particularly if they're in the blade of the retirement distribution hatchet where you've got those heavier spending years.

[00:33:20] One thing I'll really look at is how funded is somebody for those years. I'm just making up numbers, but if we're planning that over the next five years, they need $400,000 of income. Well, we can kind of start from that and say, okay, well this is a fairly stable asset base that they can pull income from. Maybe not perfectly stable, but pretty stable. And from there we can to some extent back into an allocation and say for somebody with a portfolio of this size, 80% stock allocation is fine because it'll provide enough of a cushion. Or for somebody else, maybe they actually want to take a little bit less risk. Maybe the ideal allocation for them starts out at something like a 40/60 instead of a 60/40, just so that they have a little bit more cushion there to get them to that point where their Social Security kicks in and the distribution rate drops. 

[00:34:09] And once you get to that point, maybe the allocation changes further because especially in a scenario where Social Security might cover close to or all of somebody's spending needs, they really can decide which way they want to go from there. If they want to continue to just take a kind of more normal distribution. Maybe in that case stick with more typical 60/40 or maybe they decide they're willing to and move to all-stock allocation at that point and let things really drift and treat it as more of a variable spending or charitable giving or different types of things there.

[00:34:41] But no, I absolutely do agree that the allocation to some extent can be driven by what that spending need is in those years when somebody's really trying to pull heavily from the portfolio. 

[00:34:53] Jon Luskin: Derek, any final thoughts before I let you go? 

[00:34:56] Derek Tharp: When it comes to coming up with different spending strategies or what kind of the core question and what approach should somebody use, to the extent that it's possible based on either the software or how somebody's looking at it, I think it is good to triangulate our strategies. Not just using one methodology and letting that totally control what we're doing.

[00:35:17] Looking at things from different perspectives. Say, in a sense, stress test. What could go wrong with this methodology? What could go wrong with this methodology? Sometimes historical simulation is going to have some good food for thought that maybe we're missing with Monte Carlo. And other times Monte Carlo will be insightful in ways that historical simulation won't. 

[00:35:35] Jon Luskin: That's all the time we have for today.

[00:35:37] Thank you to Derek for joining us today. And thank you for everyone who joined us for today's Bogleheads® Live. For the next Bogleheads® Live, we'll be discussing college planning. Anne Garcia, author of “How to Pay For College” will be our guest. That'll be Thursday, March 30th at 12:00 PM Pacific, 3:00 PM Eastern. And if you're listening to this on a podcast on Monday, then know that this Thursday is when we'll have Anne Garcia, so make sure to check that out on Twitter.

[00:36:04] Until then, you can access a wealth of information for do-it-yourself investors at the John C. Bogle Center for Financial Literacy at and, the Bogleheads® Wiki, Bogleheads® Twitter, Bogleheads® YouTube channel, the Bogleheads® on Investing podcast with host Rick Ferri, Bogleheads® Facebook, Bogleheads® Reddit, the John C. Bogle Center for Financial Literacy on LinkedIn, and local and virtual chapters.

[00:36:33] For our podcast listeners, if you could please take a moment to subscribe, and to rate the podcast on Apple, Spotify, or wherever you get your podcast.

[00:36:42] Thank you to Barry Barnitz for his work, and thanks to Nathan Garza and Kevin for editing the podcast. And a final ‘thank you’ to Jeremy Zuke for transcribing the podcast episodes. I couldn't do it without everyone's help.

[00:36:54] Finally, we'd love your feedback. If you have a comment or guest suggestion, tag your host @JonLuskin on Twitter. Thank you again, everyone. Look forward to seeing you all again next time. Until then, have a great one.