How Does a Bear Market Affect In-Retirement Withdrawal Rates?

On this episode of The Long View, financial-planning researcher and professor Derek Tharp discusses safe-withdrawal rates, risk-based guardrails, Social Security, and more.

Here are a few excerpts from Tharp’s conversation with Morningstar’s Christine Benz and Jeff Ptak.

Ptak: What steps would you suggest that retirees and, if they’re working with an advisor, the advisors that serve them, that they should take to ensure that their plans are still on track in light of what we experienced last year?

Tharp: I think the ongoing aspect of planning to me is really important, so making sure that a strategy is a good strategy to start with but then you stay on that path is really important. And for me, my favorite way to go about that is really the whole guardrails framework to planning where you actually have a plan upfront that isn’t just a static, the traditional Monte Carlo plan, say, that is just looking for some sort of probability of success result. It doesn’t really tell you when should I make an adjustment, how should I make an adjustment. So, really having a guardrails-type plan—I’m a fan of what I’ve written about and called risk-based guardrails. My co-author, Justin Fitzpatrick of Income Lab, we’ve talked about that as a way to think about a different guardrails framework than I think some of the more distribution-driven rate frameworks just because of some of the limitations that they have.

And I do think in practice the way many advisors or DIY retirees are doing a lot of planning is actually somewhat consistent with a risk-based guardrails approach. Like, if you’re using a Monte Carlo tool and you see the probability of success levels gone down, so maybe you cut back your spending a bit. Or you see the probability of success level has gotten higher, so maybe you can spend a little bit more. In a sense, that is very much an adjustment dynamic guardrails-type approach. But what you really get with the guardrails is predefining, OK, here’s where my lower guardrail is, here’s where I’m going to cut back if needed; here’s where my upper guardrail is and where I’m going to increase my spending if it’s appropriate. And just having that predefined, which I think more than anything, provides a lot of peace of mind, particularly as we’re going through a market like we have, where if somebody has a guardrails plan in place and they know their portfolio is at $1 million today, but they don’t need to cut back until the portfolio falls to $700,000, at least they know in advance when they’re going to cut back, what that cutback would be. And to me, that can provide a lot of peace of mind, in a way that just seeing that your probability of success has fallen from 90% to 67% really doesn’t tell you what to do or how to respond.

Benz: I wanted to ask about probabilities of success. I wonder if you encounter people—sometimes do with individual investors—where they’ll say, I want 100% probability of success. So, how should people set those risk-based guardrails if they want to use the probability of success to guide whether their plan is on track? What are reasonable bands—and obviously, this is personal—but how should people arrive at sensible risk-based guardrails along the lines of what you’re talking about?

Tharp: It’s a good question, and it’s one where I do think there is that tendency to shoot for 100% probability of success, which is probably in most cases not really what you need to do or maybe it’s excessively high in terms of the probability of success level. Some of my work I’ve been doing lately that I’ve found the most difficult time articulating or helping explain is really this idea though that there’s also a big difference between a one-time probability of success result and an ongoing targeting of the probability of success level. Even for myself, working through some of it wasn’t super-intuitive—some of the things that have come out from that. But one really striking finding was if we were to take two retirees and have them use Monte Carlo on an ongoing basis, so they’re actually targeting a consistent spending level or a consistent risk level over time, actually simulating somebody going through history using that strategy—what would it look like?

And the really striking finding—I have a Kitces article that talks about it specifically—but when we looked at comparing even a 20% probability of success as an ongoing target to a 95% probability of success as an ongoing target, the differences in maximum and minimum spending levels over a 30-year retirement period were nowhere near as big as I think most people, myself included, would have expected. So, it really turns out that if you are planning in a dynamic way, and you’re updating your plan on an ongoing basis, you’re really in a situation where using a higher or lower probability of success isn’t even at all really changing your risk of fully depleting a portfolio. It’s more like putting your thumb on the scale in favor of either current income, which would be a lower probability of success or legacy, which would be a higher probability of success. But the metric doesn’t really make sense in the way that we like to think about it once you’re starting off planning knowing that you do intend to make adjustments. If you’re doing just a once and done Monte Carlo simulation and you were going to say, OK, I’m going to pick my probability of success level, I’m going to run my plan, and then I’m going to follow that spending plan blindly and just charge forward, then the probability of success metric that people focus on actually makes much more sense. And there I would probably aim for 95% to 100%.

But I think it certainly starts to make less sense, at least intuitively, the way we think about it, as we shift to this ongoing planning approach. And there, really, I think it’s still for many people comes down to how much are they willing to tolerate, the possibility of a spending decline in the future. That’s one big lever there. And the less somebody wants to tolerate that, maybe the more their initial target should be higher, maybe getting into that 95%, 100%, maybe even—hard to think of a higher bit in terms of different levels—you could reach 100%, and you could still cut spending further, and that might be a prudent thing to do if you really want to reduce the risk of ever having a decrease other than the guardrails themselves for an ongoing plan. I think those can be often wider than a lot of people anticipate. So, using something like a 50% or a 25% lower guardrail for adjustment isn’t necessarily as extreme as a lot of people think. But I think intuitively where a lot of people land in practice is maybe somewhere like a 70% lower guardrail. I just don’t know if that’s actually what’s best in practice, and it’s an area of research that I do want to continue to pursue.

Benz: I thought that was such an interesting point that you made where you said if someone is starting out at their 25- or 30-year retirement and you’re telling them they can never look back on these decisions that they make right now about their withdrawal system, that of course they would want to err on the side of conservatism. They’d want to start really low in terms of the withdrawal rate because there’s this assumption that you can’t ever make any changes. But you make the point, and I know this is a big thing that Justin Fitzpatrick talks about too, is just that this is really a superpower that retirees have that they can make these adjustments to their withdrawal systems on an ongoing basis.

Tharp: Absolutely. And that to me is why the traditional Monte Carlo framework, the statistic that is most commonly referred to from that analysis, is just really disconnected from how most people are willing to actually make adjustments and plan in a dynamic way. And once you do that, it’s just a totally different situation and the metrics have really different meaning in a way that I think might not be well appreciated.

Ptak: I think you made the point that point-in-time withdrawal rate recommendations in addition to some of the drawbacks that you mentioned earlier, they ignore that income from nonportfolio sources, especially Social Security, is likely to change through the retirement lifecycle taken together with David Blanchett’s research that retirees tend to spend less as they age. Do you think that argues for starting withdrawals being higher in many instances?

Tharp: I do certainly think it does. This is one where Justin Fitzpatrick and myself, we wrote on this, and we used the term “the retirement distribution hatchet.” And if you plot out the typical retiree spending rates over time, and particularly if they’re deferring Social Security, you get this blade on the front end of retirement where your spending level is higher, the distribution rate is higher, but then the distribution rate falls significantly once somebody gets to age 70 and they’re taking their Social Security benefit, and we actually see that distribution rate fall and then we might even have a nice little arc in the handle as we look at something like Blanchett’s retirement spending smile on the way that spending actually does tend to decrease for many retirees over retirement.

I do think that is one reason why I actually like to use a very simple distribution rate-driven guardrails framework to explain guardrails when I’m initially teaching it. And to me, the simplest one—there’s Guyton-Klinger and Kitces’ ratcheting safe withdrawal rate and many different frameworks—but I like to even simplify that down and say, OK, let’s say, you started with a 5% distribution rate. If that distribution rate increased to 6%, then you would cut back your spending and bring it back to a 5% level. And if that distribution rate declined to only a 4% level, then you might increase your spending to get it back to a 5% distribution rate. I think it’s just a very simple framework for understanding how these guardrails type strategies work. But at the same time, that really doesn’t work in practice. It might be a nice research tool, but it really doesn’t work in practice because we see often that maybe a retiree has a 10% distribution rate for the first five years of retirement and then it drops to a 1% or 2% distribution rate from there going forward.

So, you really do have to be mindful of the limitations there and that’s where our risk-based guardrails framework, so whether you’re using Monte Carlo simulation or you’re using historical simulation and success rates there, but some other type of approach is really nice because it captures all that individual-specific spending pattern that’s there, whatever somebody wants to assume, whether that’s the traditional hatchet or maybe it’s even different and they want to make certain purchases or go on certain trips or they just have different financial goals that are out in the future. When you use a risk-based framework, all of that individual-specific information gets accommodated into the guardrails. If you’re using a distribution rate-driven framework, then you’re getting none of that. And so, to me, that is why I’m such a big fan of what we’ve called the risk-based guardrails approach to finding a retirement income spending strategy.

Benz: Just to follow up on Jeff’s question about the early years of retirement potentially being the higher spending years before Social Security comes online. Does that make that cohort who might be taking more from their portfolios in those early years, are they particularly vulnerable to sequence risk if they’re invading their portfolio at a high level in that period when perhaps the market is down? What should they be thinking about, people at that life stage?

Tharp: Yeah, you’re absolutely right that that does, it actually elevates the level of sequence risk compared with the traditional research on the topic because most research is assuming more of a constant distribution rate over time, and when you front-load those distributions, you’re even more heightening that sequence of returns risk. So, I do think it’s something to be mindful of. For me, oftentimes there’s different approaches to managing that. But one thing I like to do is to think from at least somewhat of a buckets-type framework. I like very simple bucket framework. So, maybe we even call it a two-bucket framework where we have our more stable investments—cash and bonds and those types—and then we’ve got our higher-growth-type investments like stocks.

But in many cases, when you look on the front end—let’s say that it’s a 65-year-old retiree who’s trying to defer their Social Security for five more years—I think there’s two big things that they have to help mitigate some of that sequence of returns risk. One is that they can set aside enough cash and bonds on the front end to hopefully be able to fund a very significant portion of the spending for that five-year period. I often like to try, through a buckets framework, try and target five to seven years at least worth of anticipated spending needs. And so, that could mean that somebody has that resource there, but if they know they’re going to be turning Social Security on in five years and their income is going to be jumping—nonportfolio income—is going to be jumping up because of that, to me, that can be one reasonable strategy, just setting aside enough on the front end to get through that blade of the hatchet.

The second thing that I think can be really useful is just the fact that you also have the option to turn Social Security on earlier if needed. So, certainly, in many cases, I think it’s worth trying to continue to delay. But if we hit just a really awful sequence of returns and it was very clear that this was not a great time to keep pulling heavily from a portfolio, you do have the option to adjust your strategy in a sense and take Social Security earlier. It’s not like you have to declare on the front end that you’re waiting until 70 and stick with that. So, I think those are the two ways that retirees in many cases can really mitigate that risk. It’s maybe a little bit harder for somebody who is say retiring at 50 or 55 and they have a 15- or 20-year period they’re trying to stretch—there I think there’s a little different set of challenges for many retirees.