A while ago I created a little toolkit to design my own bond and/or CD ladder. With a bond/CD ladder, by the way, I mean holding a portfolio of bonds and/or CDs so the cash flows comprised of maturing CDs/bonds plus interest income exactly matches a specified time series of target cash flows over time. Of course, if you’re regular readers of this blog you know that I’m not too thrilled about investing in bonds or CDs. Yields are just too low, compliments of my former colleagues at the Federal Reserve. But there are still a few interesting applications for bond/CD ladders. Some of them purely for “academic curiosity” and not because I actually want to implement them. So, here would be a few questions I’d try to answer with this toolkit:
- If I wanted to build my own little “quasi-annuity” to guarantee a certain cash flow for 30 years or so, how much money would I have to set aside today to guarantee that cash flow? This is obviously not because I actually want to do this. As we will see later, the cash outlay today would be so prohibitively high that I’ll prefer to take my chances with the stock market and the Sequence Risk that comes with it!
- How much of a difference would it make if I required regular cost-of-living adjustments (COLA) along the way rather than getting one fixed nominal “income” along the way?
- Related to the issue above, if someone already has a corporate pension without COLA, how much would he or she have to set aside to supplement this pension and guarantee a certain COLA, say, 2% a year?
- When interest rates were higher only a few decades ago, how feasible would it have been to create a bond ladder for retirement? Could we have gotten rid of Sequence Risk completely?
- If I wanted to hedge my first 5-10 years of early retirement against Sequence Risk and at least partially fund my retirement expenses through a CD ladder, how much money would I have to set aside?
So, today I’ll share that spreadsheet and some of my sample calculations and results. I hope you’ll find value in playing around with it, too…
First, the link to the toolkit:
As always and for obvious reasons, you cannot edit this (clean) Google Sheet. So, please create your own copy first!
How to use this sheet
- Your cash flow needs: The easiest way to get started is to pin down your cash flow needs through just three parameters: the number of years, an initial flow, and a cost of living adjustment (COLA) parameter, i.e., by what percentage we grow the cash flow needs each year. Keep that COLA parameter at 0% if you want to keep the nominal cash flows constant. Notice that the initial cash flow is paid upfront, so if you plan for 30 cash flows this exercise will go through years 0, 1, 2, … and all the way to 29.
- You can also input your interest rates. As a default setting you can enter the rates for bonds/CDs with maturities of 1, 2, 3, 5, 7, 10, 20 and 30 years (conveniently, these are bond maturity dates for which the Federal Reserve reports daily estimates for Treasury yields in its H.15 data table (Selected Interest Rates). As a default setting, the interest rates in between are interpolated linearly.
If this is too simplistic then you can also just override the cash flow needs generated by the three parameters and/or the interest rates and hard-code whatever you deem necessary:
- Cells B28 to B58 for the cash flows.
- Cells I25 to AL25 for the interest rates at maturities 1-30.
Also, before getting too far, here are a few limitations of this approach:
- There is no consideration for taxes. I do report the composition of how much capital is flowing back vs. how much is interest income, though. See columns D and E, rows 28 through 58 to see how much of the cash flow is a return of capital (tax-free) and how much is interest (ordinary income and thus taxable if exceeding your federal/state deductions).
- I assume that all interest is paid at an annual frequency, just to keep the spreadsheet manageable. Most CDs and bonds obviously pay interest at a monthly/quarterly frequency.
- I assume that bonds are traded “at par.” This is the correct assumption for CDs but may not be 100% accurate for bonds if the coupon differs from the annualized yield.
- Your COLA estimate might differ wildly from the actual inflation rate. That could be because the actual realized CPI is very different from our inflation estimate today (most commonly 2%). You could also have a consumption basket very different from the average CPI basket with much higher or lower (felt) inflation.
- Check the results and make sure they are actually implementable. I will show you a “pathological” case where some CD/bond purchases could be negative! Unless you can borrow at that CD/bond rate there may not even be any CD/bond ladder that exactly matches your cash flows!
But with all of those limitations, let’s look at some actual examples.
Example 1: 30-year payout, no COLA, invest in U.S. Treasury Bonds
Here, I set the parameters to $40,000 initial cash flow, COLA parameter to 0% and the time horizon to 30 years. I also use the Treasury Yield values from the Federal Reserve website (Table H.15) as of 4/17/2019:
To sustain these 30 years worth of cash flows ($1.2m total), we’d need to invest a pretty large sum today: almost $827k! Also, notice the pattern of how much we’d have to set aside for each of the different bond maturities (see bar chart on the right side):
- $40,000 for the initial payout. I assume that the year 0 we just set aside the money at the beginning of the year without earning any interest.
- A little bit less than $40,000 for the final year 29. That’s because a $38,845 notional bond purchase plus the interest for the final year will exactly match the $40,000 cash flow requirement.
- Successively lower notional investment because in year 28 you receive not just the notional back for that year but also the interest from the bonds maturing in years 29 and 28. And so on, all the way to year 1 when you receive not just your year 1 principal and interest but also the bond interest for all the other bonds maturing in years 2-29!
- If you’re one of my fellow Excel-Super-Geeks, you can check out how I calculated this in columns H through AL, the notional investments in column 26 and the year 0-30 cash flows in rows 28-58 for each of the bonds/CDs. I make heavy use of the “OFFSET” function in Excel, one of my favorite Excel hacks! 🙂
So, just to sum up, I was shocked about how much money you’d have to set aside to guarantee those $40k for 30 years! But it gets even worse! Recall that we didn’t even account for any COLA. So, let’s change that and see how much money we’d have to set aside if also like to make 2% p.a. increases to the initial $40k withdrawal, which brings us to Example 2…
Example 2: same as example 1 but with 2% COLA
The only change I have to make is to set the COLA parameter to 2% in cell B6. Let’s see how this changes the results: You’ll now need about $1,070,000 to guarantee this cash flow. And just to be sure, the money will be completely exhausted after year 29! So, you don’t even get to a 4% initial withdrawal rate over 30 years with this method despite the total depletion of your capital! This is clearly not a workable solution for early retirees. And it may not be very palatable for a traditional retiree, either. You might just go for an annuity in that case. At least you hedge the “risk” of living longer than 30 years…
Example 3: same as example 1, but raise the bond yields to December 1999 levels (the good ol’ days!)
What if we ever were to go back to more “normal” bond/CD interest rates? Just about 20 years ago when we had short-term interest rates above 5%, and longer-maturity bonds with 6%+ rates! If I feed in the December 1999 interest rates from the Federal Reserve Data Table H.15 we can now push the initial investment all the way down to about $557k. What a difference those 350bps higher interest rates make!
Also, just out of curiosity, if you were to also require a 2% COLA, you’d still only about $687k initial capital, see below!
This is really astonishing! Let this sink in:
Back in the late 1990s, you could have fully guaranteed a 5.82% Safe Withdrawal Rate (!) over 30 years with 2% COLA with a simple zero risk government bond ladder.
This, by the way, is why I’m quite skeptical of both Bill Bengen’s work and the Trinity Study: Back when this research was initially done, the whole discussion about the 4% Rule was completely moot! You could have easily guaranteed a Safe Withdrawal Rate of almost 6%. Or, alternatively, use a 4% initial withdrawal rate plus 2% COLA for 30 years with a bond portfolio and have tons of money left over ($313k per $1m of initial net worth) to invest in an equity portfolio, which you would not have to touch for 30 years. That’s a lot of dough to have as an additional hedge against longevity past the 30 years and/or money to leave to your heirs and charity causes! That would probably have been enough even for a 60-year retirement if you use the bond portfolio for the first 30 years and then the equity portfolio plus growth for the next 30 years!
Today, of course, the bond ladder as a retirement strategy is no longer feasible. Interest rates are way too low. But that doesn’t make the Trinity Study any more useful. As I have outlined here on the blog, especially in the SWR Series, looking at the unconditional success probabilities of the 4% Rule, while not taking into account today’s lofty equity valuations and rock-bottom bond yields seems highly inappropriate.
But I digress! Let’s move on to another example…
Example 4: An 8-year payout, linearly phased out, with CDs
Suppose we have an early retiree concerned about Sequence Risk. To alleviate the risk of retiring right at the peak of the bull market, he/she wants to have $40,000 guaranteed cash flow from a CD ladder right now, $35,000 next year, then reducing the guaranteed income by $5,000 every year down to $5,000 in year 7 and nothing more after that. I’d now have to override the cash flow values in column B, rows 28-58 with the actual target values. The parameters in cells B5-B7 are now irrelevant!
I also input my estimates for interest rates, CD rates of between 2.7% to 3.3% for the first 7 years. I’m sure if you shop around you might even get slightly better rates. In any case, according to our spreadsheet, we’d need just about $168,000. The total flows, of course, are exactly $180k comprised of $168k in principal and only just under $12k in interest income. Is it worth it do this as a hedge against Sequence Risk? You be the judge! If you simply sell $168k of your fixed-income portfolio over the first few years then this looks a lot like a glidepath (see part 19 and part 20 of the SWR Series). That certainly helps somewhat with Sequence Risk but it’s not really a panacea either as I showed in those posts a while ago!
Example 5: Set aside enough money to turn a non-COLA pension into a 2% COLA pension
Another interesting example. Imagine you already have a corporate pension but it has zero COLA, in other words, it will stay constant in nominal dollars and will be slowly eroded away by inflation. Bummer! How much money would I have to set aside to turn this into a pension with 2% COLA? Let’s assume this is a $20,000/year pension and I like to protect it from inflation over a 30-year horizon. I’m using the Treasury bond interest rates but I’ll have to hack the cash flows to match exactly the difference between 20,000×1.02^t (=$20k plus 2% COLA) and the 20,000, see below:
Again, for the interest rates, I use the Treasury constant maturity yields from the Federal Reserve. Now we get the following output, see below:
OK, what’s going on here? We’d need negative bond holdings (!!!) for years 1-8! This is the “pathological” case I mentioned above where you’ll essentially have to borrow money at the short horizon because you’ll have excess cash flow in the early years; it’s because you get a ton of interest income from bonds maturing later but you need very little money early on. So the figure of net $121,398.05 outlays is really just an incomplete estimate. Of course, there are (at least) three ways to deal with this:
- If you already have some other short-term savings (e.g., an emergency fund) you could certainly “borrow” from there.
- Reduce the purchases of bonds with maturity 10+ years and instead reinvest some of the excess cash flow received in years 1-8.
- Reduce the excess cash flow early on by shifting more money into zero-coupon or at least low-coupon bonds (if available) for the very long maturities.
In any case, if you have a pension with COLA (as is the case with many government pensions), consider yourself really, really lucky! Over the years, the typical non-COLA corporate pension falls behind substantially and it would take a big chunk of money to “hedge” against this depletion of purchasing power. Over 30 years, you’ll need to come up with a total of more than 10x the initial pension ($211k vs. $20k) to create a 2% COLA, and in today’s dollars, that’s still more than 6x ($121k vs. $20k)! Bummer!