Finally, in one place, including some aspects itemized for the first time
It wasn’t until Jeremy Brereton emailed me a question after my fumbling explanation at the end of Stefan Lundberg’s Pioneering Pensions webcast that I realized that I had never written down some of the lower-level (but still very important) details of the decumulation approach I use for my wife and me. My apologies. I didn’t intend to keep anything secret – that’s not my way. But space limitations (in my FT Money piece, in blog posts) have conspired to limit my explanations to the major points.
My response to Jeremy’s email was, I realized, the first time I had ever explained the details – and that’s because Jeremy is the first to have asked about them, which in turn is why I didn’t have an instantaneous clear answer in the webcast. Rather than simply reproduce our email sequence, let me take the opportunity to explain the whole “two buckets” approach in one place. Right here.
I’ll start with a quick summary of the main points. Then I’ll explain the details about how to adapt when (as is inevitable) life turns out to be different from the plan.
As I’ve said before, everything is an adaptation of what I learned from advising, and observing the practices of, defined benefit pension (DB) plans. And it all arises from my thought experiment of imagining that my wife and I are the sole survivors of a “target benefit” plan that is less than 100% funded.
Two departures are necessary from what DB plans do, to make the thought experiment fit. First, for DB plans, when they’re less than 100% funded, the sponsor/guarantor has to put in more money to make up the deficit. In contrast, with a target benefit plan there needs to be a cut in the benefit to bring the funded ratio up to 100%. Second, it’s reasonable to assume that retirees will, in the aggregate, experience longevity that fits a reasonably applicable table. In contrast, when there are only two members left, departures from average longevity are highly likely, so an explicit safety margin must be used.
For my wife and me, I used the longevity table created by the American Academy of Actuaries and Society of Actuaries, and adopted, as a planning horizon, the length of time that gave us a 75% chance that neither of us would outlive it. When I was 70 and my wife somewhat younger, this projected to a 31 year planning horizon.
I followed a standard DB plan approach of holding assets in two buckets, one that matches the projected benefits outflow for a few years, and the second (consisting of all other assets) seeking growth, I set up a bucket with 5 years of our desired cash flow and invested the rest very simply and inexpensively in a global equity index fund. We think of these as our insurance and growth buckets.
(Others insist that it’s eventually all one combined bucket. True, but what does that perspective add? The fact is that we have entirely different objectives for the two buckets, so a combined investment return is mathematically calculable but has no meaning, as it doesn’t reflect either the insurance goal or the growth goal.)
Why 5 years? Back up a moment. The intention is that, every year, as the insurance bucket reduces to 4 years of withdrawals after we take one year’s worth out of it, we’ll replenish it back to 5 years’ worth by cashing out one year’s worth from the growth bucket and moving the cash-out into the insurance bucket. But we don’t want to do that if the growth fund’s return is negative in real (after-inflation) terms. (We do all calculations in real terms, as we want our withdrawals to keep pace with inflation.) And it turns out that (at least according to history) if the growth market falls (returning less than 0% real) there’s a 75% chance that it will recover in 5 years’ time. (Yes, I am deliberately consistent in using the same 75% chance of success, or if you prefer, 25% risk tolerance, for both our longevity risk and our investment risk.) In other words, the odds are 75% in our favor that, within 5 years, we’ll be able to replenish our insurance bucket back to holding 5 years’ worth of withdrawals.
Why accept a risk as high as 25%? It’s purely pragmatic. I assumed a 0% real annual return in the insurance bucket and a 4% real annual return in the growth bucket. At the 25% risk level, the sustainable annual withdrawal turns out to be roughly 5% of the pension pot (which for us is not only our formal retirement savings but non-pension assets too) in the first year, increased by inflation thereafter. I also calculated what the sustainable annual withdrawals would be if we reduced the risk to 10%. That increased the planning horizon as well as the size of the insurance bucket, and then the sustainable withdrawal came down to roughly 4%. When we took account of our Pillar 1 pensions (provided by government plans, which for us include both US Social Security and the Canada Pension Plan, as I contributed to each for roughly 20 years), we desired the 5% withdrawal much more than the 4% withdrawal. And that means we need to accept the 25% risk level rather than the 10% risk level.
That’s the essence of the plan, the part I’ve explained many times. It’s what enables us to sleep easily at night, through market downturns.
What is absolutely certain, though, is that the future will not unfold as planned. So then, what will we do?
Let’s start with the fact that a precise 4% real annual growth return is an expectation that will include huge fluctuations from year to year, sometimes above 4%, sometimes between 4% and 0%, sometimes below 0%. Should I be an absolute stickler for precision in a volatile world and refuse to make a transfer from growth to insurance if the return is only 3.9%? It seems pedantic. At the other end, given that the 5-year period was based on markets recovering to provide at least an average 0% real return, would I be hugely satisfied if the growth return is 0.1%, and hugely dissatisfied if the growth return is – (minus) 0.1%? That feels like hair-splitting.
There’s no theoretically precise answer. So I’ll make the sort of human compromise that pension fund trustees often do: I’ll make a transfer if the growth bucket returns at least 2%. Going halfway minimizes my regret!
But wait. What if there’s a huge growth return? Why transfer only one year’s worth of withdrawals, when the ultimate goal is to reach, and if possible lock in, a 100% funded ratio? After all, if we can buy an inflation-proofed annuity for the the rest of our “joint and last survivor” lives, then we can forever satisfy our spending desires and no longer need to take either investment or longevity risk. Pension funds call this a “buy-out” – everything is now guaranteed by an insurance company. So the closer we can get to that state, the more secure we’ll be.
So I’ve concluded that it makes sense that every additional 4% returned by the growth bucket should generate an additional year’s transfer. So: 2% real generates a transfer of 1 year’s withdrawal amount, 6% generates 2 years, 10% generates 3 years, and so on. I can’t be bothered to make fractional transfers, though in fact that’s more logical (as Jeremy suggested).
Aha, but the insurance bucket these days isn’t providing as much as 0% real! So my quick-and-dirty (and entirely arbitrary) way of taking this into account is to place an added burden on the growth bucket’s return. In addition to its 2%, it needs to make up for the dollar shortfall below 0% in the insurance bucket. That doesn’t (in the early years) add a lot to the 2%, because the growth bucket is so much larger, but it gives me the mental satisfaction of having taken it into account.
All of that applies particularly after a year like 2021. It was a good year that permits much subsequent de-risking.
Jeremy’s actual question to me was much deeper, in fact. He asked: “Will you allocate more funds to your growth bucket in a period of very low interest rates and high inflation?”
I have ducked the issue. I feel that my approach described above is a simpler way to take the low interest rate environment into account, as the cause that pushes short-term rates below 0% real is the very same cause that has pushed up the prices of discounted assets. Adjusting the initial numbers of years of withdrawals in the insurance bucket away from 5 is too complicated for me to investigate.
There are two more obvious adaptations that are necessary.
The simpler one is to the planning horizon. Obviously each year it should, in principle, be adjusted to reflect our health and/or survival.
So far, thank goodness, it hasn’t been necessary. I used “non-smokers” and “average health” to get that 31-year 75%-chance horizon when I turned 70. When I turned 75 the same inputs for both of us revealed a 26-year horizon – meaning that, beyond the actual passage of 5 years, nothing else had changed. In fact I’ve looked ahead, out of curiosity, to see what the remaining indicated planning horizon will be if we’re both still around and in average health when I turn 80, and the answer is: 21 years. I know, as an actuary, that the planning horizon should extend fractionally if we both keep surviving; what these numbers tell me is that the adjustment is not a big issue, in practice. (That holds for most couples. It holds particularly in our case because my more advanced age doesn’t figure much in the planning horizon, which is essentially a function of my wife’s lower age.)
What will of course have a bigger impact is when one of us passes away. (More likely to be me first: females, on average, tend to live longer than males, and, as I keep saying, I’m older anyway.) At that point the remaining planning horizon, for the survivor, will fall. (Probably not much, in our case.) That means that (other things being equal – meaning that the plan is working out) the future sustainable withdrawal will increase. That’s despite the fact that the one survivor will need less money to preserve the standard of life than we do as a couple.
We’re not taking that into account in our planning. It’s a (double) safety margin. In fact there’s a third safety margin that is likely to be there for us. It’s that most couples and most people tend to go through two or three phases of activity in their Life Two, colloquially called go-go, slow-go and no-go. The go-go years are at the start, when both the energy and the desire to do things you’ve dreamed of for many years are at their height. Many years later, activities tend to downsize: no less active, perhaps, but sometimes confined to a smaller area (less world travel, for example). Perhaps the home is also downsized. These are the slow-go years. We hope we never encounter the no-go years, when long-term care becomes necessary. (The best statistics I’ve seen are from US insurance companies. They suggest that the probability in the general population of needing care that extends beyond 90 days is roughly one in three. So most of us probably won’t experience it, though it could be expensive if it does occur.)
Anyway, if our withdrawals are sufficient to support our excited go-go years and we plan to be able to sustain that level of real withdrawal, the odds are that we won’t need nearly as much later. That’s the safety thought.
OK, it’s time to face the potentially really bad news. What if the growth markets fall … and don’t recover … maybe even keep falling? History may be more favorable than the future, and what we fondly believe to be our 25% risk exposure may actually be much higher. And of course a 25% chance of a bad growth-seeking outcome is in itself large enough to need a Plan B, let alone the chance that 25% is an underestimate.
What will we do?
Once again, the lesson is from DB plans, adjusted for the fact that we’re really a target benefit plan.
DB plan guarantors/sponsors have to put in more money when the benefits are too large to be sustained by the size of the assets, that is, the funded ratio is below 100%. How much more do they have to put in?
Well, they don’t automatically put in a lump sum sufficient to raise the funded ratio immediately to 100%. That would make their contributions far too volatile, because the growth-seeking assets are always volatile. What they do is to spread the shortfall over a long period of time, and contribute a level amount each year aimed at making good the shortfall by the end of that period. What they’re hoping for, in addition, is what’s known as “mean reversion” – the hope (rather than the expectation) that there will also be better-than-planned years in the future (as has occurred in the past), so things may tend to average out. So, DB sponsors start making those additional contributions, but hope that the full sequence won’t be necessary.
There’s an obvious analogous situation for us. If one year the reassessed funded ratio is below 100%, don’t instantaneously cut the withdrawal by the full amount of the shortfall. Rather, spread the cut over the remaining planning horizon, and hope that it will be restored some time in the future.
And if it isn’t restored, because history has been defeated? Then two things happen. One is that we’ll be in deep doo-doo, as in fact will the rest of the world. The other is that we’ll have had five successive years of gradual cuts before the insurance bucket runs out, and we’ll have been seriously considering the possibility throughout that the cuts are permanent, and may be followed by further cuts, and we’ll have adjusted. Certainly we’ll have adjusted financially; I hope we’ll have adjusted emotionally too. At least we’re aware of the possibility, and it will be a disappointment rather than a huge surprise. I hope that helps the adjustment.
But basically, that’s what risk means. It’s the chance of a bad outcome. And when you seek growth, you’re taking risk. It’s as simple as that.
We’ll have reassessed the difference between our needs and our wants. And (at least so far) while it will hurt a lot to have to keep giving up on more and more of the wants, we hope never to have to give up on our needs. Worst case (at least as I see it today): if we ever seem to be approaching a state where the funded ratio for our needs alone is declining towards 100%, it’ll be time to buy an annuity that locks in our needs for the rest of our survival.
What a horrible note to end on. I didn’t plan to end the blog post this way!
So I’ll cheer up with the thought that 2021 was a year after which much de-risking has been possible, so there’s more than 5 years of recovery time after the next inevitable market fall. And I’ll make the Takeaway neutral!
Each year, adjust the number of years in the insurance pot as well as the following year’s withdrawal. Yes, there’s a lot of detail to be implemented, in practice. Setting it all up at the start and then leaving everything alone isn’t how this works.
I have written about retirement planning before and some of that material also relates to topics or issues that are being discussed here. Where relevant I draw on material from three sources: The Retirement Plan Solution (co-authored with Bob Collie and Matt Smith, published by John Wiley & Sons, Inc., 2009), my foreword to Someday Rich (by Timothy Noonan and Matt Smith, also published by Wiley, 2012), and my occasional column The Art of Investment in the FT Money supplement of The Financial Times, published in the UK. I am grateful to the other authors and to The Financial Times for permission to use the material here.