May 16, 2024 – Welcome to another Safe Withdrawal Rate Series installment. Please see the landing page of the series for a guide to all parts so far. In Part 60, dealing with the “Die With Zero” idea, I mentioned working on an upcoming post about the “Safety First” approach, and I finally got around to writing that post. What is Safety First? It involves using asset allocations different from those in the Trinity Study or my SWR Toolbox (see Part 28). For example, we could use Treasury Inflation-Protected Securities (TIPS) as a default-free and CPI-hedged investment option. However, TIPS are no hedge against longevity risk. An annuity hedges against longevity risk; though the most common annuity option, a single premium immediate annuity (SPIA), is usually not CPI-adjusted. Also, for the longest time, low interest rates rendered the Safety First approach all but useless because neither TIPS ladders nor annuities generated enough income for a comfortable retirement. You would have been better off taking your chances with the volatility of a 60/40 portfolio.
In other words, there is no free lunch. You don’t get peace of mind for free. Rather, you likely pay a steep price for that safety by giving up most, if not all, of your portfolio upside and/or bequest potential. However, since interest rates started rising again in 2022, the entire fixed-income interest rate landscape looks more attractive now. Could this be the time to reconsider Safety First? Let’s take a look…
Safety First or Safety Worst? It all depends on the Interest Rate Landscape!
One of the reasons I have held back writing about Safety First is that interest rates were so low for the first six years of my blog (2016 to 2022) that the implied withdrawal rates utilizing a 30-year TIPS ladder would have been rather unappetizing. Specifically, for much of my early blogging career since 2016, the real interest rate on 30-year TIPS has languished around 1% pre-pandemic and then dropped further to below zero post-pandemic. Only recently have we seen TIPS yields in the 2.0-2.5% range again.
Consider the following example: a traditional retiree with a 30-year horizon who faces a real rate of 1.0% could have used a TIPS ladder to structure a withdrawal rate of 3.85%; see the table below. This would have assumed a 1.0% TIPS rate at all TIPS horizons, which is likely a bit too optimistic, so the actual rate might have been slightly lower. At a -0.5% TIPS rate, your safe consumption rate would have dropped to only 3.09%. Specifically, it would have been less attractive than even the failsafe from a standard safe withdrawal rate analysis.
Also, notice that over a 30-year horizon, guaranteeing even a modest bequest (or cushion) worth 25% of the initial balance would have further reduced your safe consumption rate: 3.14% for the +1.0% real rate and 2.19% for the -0.5% real rate. Leaving a more significant bequest (50% of the initial balance) would drop the safe consumption rate to 2.43% for the 1.0% real rate and 1.29% for the -0.50% real rate. In comparison, a balanced portfolio (75% stocks and 25% intermediate-term US Treasury bonds) would have yielded a 30-year failsafe rate of 3.82%, 3.58%, and 2.90% for 0%, 25%, and 50% final value targets, respectively. And those are the rock-bottom failsafe rates if we encounter a repeat of the Great Depression and/or the 1970s and early 1980s. In the not-worst-case scenario, you would have done much better.
But admittedly, using today’s TIPS rates of around 2.25%, you would have a 4.56% safe consumption rate for a zero bequest, 3.98% for a 25% final value target, and 3.41% for a 50% final value target. Not bad! However, with this TIPS ladder, you still face longevity risk. How risky is it for a 65-year-old to plan for a 30-year retirement and not have any money set aside past age 95? That brings me to the next issue:
What is my expected vs. potential lifespan in retirement?
Since I posted my actuarial toolkit in Part 56 in 2023, people have asked me if I can add more features to the SWR Toolkit (see Part 28 for the link to the Google Sheet) to gauge the expected and potential lifespan. Planning a TIPS ladder only up to your life expectancy generates a significant risk – approximately 50% – of running out of funds in retirement. If you opt for a TIPS ladder, you must plan well beyond your life expectancy. And as a couple, you must hedge not just the individual but also the joint survival. True, some expenses may decrease if the household size goes from two to one late in retirement, e.g., food. However, some expenses may stay the same, such as housing and utilities, for example, if the surviving spouse prefers to stay in that same home. Some expenses may rise if one of the spouses passes away, for example, if the surviving spouse no longer feels comfortable living alone and then needs in-home or even nursing home care.
So, what retirement horizon is appropriate for 65-year-olds? Let me introduce a new tool in the SWR Google Sheet. It’s in the tab “Life Tables,” and the only inputs we need here would be the gender and health status of the two spouses. My sheet uses the age you specify in the “Cash Flow Assist” tab, so that’s not even an input here. I use the SSA 2019 life table. If you like using your custom table or a different cohort, you can copy and paste those values in columns AC-AI in that tab.
In any case, here’s the output using a 65-year-old couple with average health, i.e., using the death probabilities from the SSA 2019 table; see below. Individually, they may have a life expectancy well below the 30-year horizon. The two spouses have a conditional individual life expectancy of only 18.1 and 20.7 years, respectively. However, the last survivor has a life expectancy of 24.5 years. At the end of the 30-year horizon, there is a 21.1% probability that one of the spouses is still alive.
Summary so far: using Safety First to build a TIPS ladder with a defined end date, at which point your funds run out, is indeed the antithesis of Safety First. You introduce a significant failure probability if one of the spouses outlives the predetermined end point of the retirement horizon. If you’re an avid reader of the ERN blog, you may remember that I made a similar point in Part 32 of the series “You are a Pension Fund of One (or Two).” This one aspect of retirement planning is more challenging for a one or two-person household than for a pension fund with thousands of participants. A pension fund can rely on actuarial tables to predict the rate at which the number of covered participants will decline over time and then match its assets and cash flows with its predicted future liabilities. But as an individual retiree, you can’t work with actuarial math like that. You’re either 100% alive or 100% dead.
Recall that the SSA tables are for the nationwide average. Most people I know consider themselves significantly healthier than the “average” American. If I set the health of both retirees to “Good,” I will extend the life expectancy by about two years for each. The way I twist the SSA survival table is that for people in good health, death probabilities are reduced by 20% every year, resulting in a rise in conditional life expectancy by about 1.8 years for both spouses. Even though both spouses still have a life expectancy well below 30 years, the joint life expectancy is now at 26.5, and the last survivor has a 32.1% chance of outliving the 30-year horizon. Pretty scary! So, operating with a fixed retirement horizon and drawing down your savings to zero until age 95 seems quite risky. Safety First isn’t that safe if you use a TIPS ladder.
Finally, I also display the results for people in excellent health (40% reduction in death probabilities, resulting in a roughly 4-year extension of the conditional life expectancy). Now, the last-survivor life expectancy is 29.2 years, and the probability that you outlive a 30-year TIPS ladder is almost half (47.5%).
Safety First, through a TIPS ladder, is even less attractive to early retirees!
Another reason I never felt a big urge to write about Safety First is that there is likely only limited use for this approach outside the traditional retirement community, say 65-year-old retirees with a 30-year horizon. That’s because even with today’s 2.25% real rate, we get a 50-year safe consumption rate of only 3.33% with portfolio depletion, according to my table above. We’re down to 3.06% with a 25% bequest target and 2.79% with a 50% final value target. Those are pathetic withdrawal rates. TIPS rates would have to rise again to about 3.5% (as they were in the late 1990s) for early retirees even to take notice.
Or we could calculate the real return necessary to achieve certain withdrawal rates and bequest targets over different retirement horizons; see the table below. For example, if I want to achieve a 4% initial rate over a 50-year horizon, even with capital depletion, I’d need a real rate of 3.25%. With a bequest target of 25% of the initial portfolio, I’d need a 3.5% real rate—far above today’s rates.
Worse, the Safety First approach is still not 100% safe because, with the financial tools available today, we can only hedge our real, CPI-adjusted retirement spending over the first 30 years, the longest TIPS maturity available today. Any horizon longer than that faces the reinvestment interest rate risk. So, Safety First, through a TIPS ladder, is unattractive for today’s early retirees trying to hedge 50+ years’ worth of retirement spending.
Safety First through Annuities
How about annuities? We would eliminate the longevity risk issue but introduce another headache: inflation risk. The most common annuity is the Single Premium Immediate Annuity (SPIA), where you hand over a fixed amount up front and then start collecting monthly checks until you pass away. They can also be structured as joint survivor annuities, i.e., they pay for as long as at least one spouse is alive. Earlier this month (May 3, 2024), I grabbed some annuity quotes for folks between ages 40 and 80. I got quotes for males and females as well as joint survivorship annuities that pay for as long as at least the spouse survives (assuming couples of the same age). The annuity quotes are per $100,000 surrendered immediately; see the table below.
Importantly, these initial payments are in nominal dollars and are not adjusted for inflation. Thus, the raw annualized “withdrawal” rates seem pretty high. For example, a couple, both 50 years old, can get a 5.80% initial withdrawal amount, i.e., $58,000 p.a. out of a $1m portfolio. Does that mean we can beat the 4% Rule? Not really, because your initial $58,000 annual payout will slowly dwindle to under $40,000 within 13 to 19 years, depending on the inflation rate; see the chart below. If you survive 40+ years, your purchasing power is down to around $26,000 with 2% and a measly $17,800 with 3% inflation. Over long horizons, nominal annuities have too much inflation risk! SPIAs alone are the antithesis of “Safety First.”
Of course, instead of annuitizing the entire $1,000,000 portfolio, we could also guarantee an initial $40,000 payout by committing about $644k to an SPIA and keeping the remaining $356k in a portfolio to compensate for any shortfalls caused by inflation. But you still introduce uncertainty through multiple routes:
- Inflation Risk: We don’t know how quickly the initial $40,000 melts away from inflation and how much and how quickly we have to sell our portfolio to purchase additional SPIAs to make up for the inflation drag.
- Market Risk: When selling assets in our portfolio, we are subject to market risk. Hence, we still face Sequence of Return Risk, because your SPIA retirement plan depends on well or poorly your portfolio performs early on in retirement.
- Interest Rate Risk: You also face interest rate risk because of uncertainty over future SPIA annuity quotes.
In a worst-case scenario, you would liquidate part of your portfolio at the bottom of a bear market at depressed valuations: Sequence Risk. Then, you also shop around for annuities when interest rates are low again due to the accommodative Federal Reserve policy. That’s Sequence Risk squared! Safety First became Safety Worst!
Annuities with COLA
How about annuities with COLA? Some people claim that annuities can overcome the inflation risk if we simply buy an annuity with a cost-of-living adjustment (COLA). That’s only partially true. Annuities with precise CPI adjustments are still rare and often overpriced. Annuities with a fixed payout increase (e.g., contractual, fixed 2% increases yearly) are more common but are still not a perfect inflation hedge. First, that 2% annual COLA will imply that your initial annuity payment has to be lower. I ran some quotes recently, and a 50-year-old would get a flat $527 monthly payment with a standard SPIA. With the 2% COLA, the initial amount would be only $406. $351 with a 3% COLA. So, you would trade off initial cash flow with future cash flow growth.
Also, if inflation runs hotter than your 2% increase, you still lose purchasing power in retirement. For example, if you receive annuity payments with a 2% COLA, those annual increases would have been insufficient to keep up with realized inflation over the last 4-5 years. Due to the post-pandemic inflation shock, you would have lost about 10% of your purchasing power even with 2% annual COLA. And since year-on-year CPI is still not back to 2%, that gap between CPI and the 2% trend keeps rising. In other words, an annuity with 2% fixed COLA only hedges against a 2% trend inflation, not the uncertainty around it.
Side note: The popular platitude/truism in the personal finance world is that Social Security is the best annuity deal available because it offers genuine, inflation-linked COLA. I agree. Let’s assume we already optimized our Social Security, likely deferring benefits until age 70; see Part 59 for more details. But most people can’t live on Social Security alone. That’s why I started this blog, and that’s why you come here: we study how to transform financial assets into cash flows to supplement our (optimized) Social Security strategy.
A TIPS ladder plus a (deferred) longevity annuity
This approach is not my invention, but it was proposed in a paper a few years ago: Stephen C. Sexauer, Michael W. Peskin & Daniel Cassidy (2015) Making Retirement Income Last a Lifetime, Financial Analysts Journal, 71:1, pp. 79-89. The authors suggest constructing a 20-year TIPS ladder and hedging the longevity risk beyond age 85 with a deferred annuity. You’d hedge the inflation risk over the first 20 years, but there is some residual inflation risk from two sources:
- The deferred annuity only guarantees a nominal amount, so you’ll have to predict the inflation rate over the next 20 years and pick a nominal annuity payout amount to reflect this estimate.
- Subsequent annuity payments are nominal only, and there is no further COLA. This may not be a good approach for young, early retirees, but I can see that a couple, 65 years old, would be OK with some real spending decreases after age 85 when they likely slow down. But you better hope and pray for no unexpected medical or nursing home care expenses!
So, let’s put this approach to the test with today’s TIPS rates and annuity quotes. I will use a 2.25% real return assumption, close to the most recent 20-year TIPS quote I got was 2.22% on 5/6/2024, according to FRED. Assuming 2.5% inflation over the next 20 years, we target $1.6386 in year 20 for every $1 in real spending. I also gathered an annuity quote from Immediate Annuities (no affiliation, just a neat site to check quotes quickly) for a 65-year-old couple who spends $100,000 today on a 20-year deferred annuity. To guarantee $1 in real spending, they would spend $193.89 on a TIPS ladder and $50.16 on a deferred annuity for a total of $244.05. Or if we convert this into annual spending for a, say, $1m retirement portfolio, you’d get around $49,171. That’s not bad. It’s an implicit 4.92% safe withdrawal rate. That’s a lot more attractive than the 4% Rule of Thumb.
The problem with this approach is that there is no bequest. The only way you would leave anything behind for the next generation is if both spouses pass away before they reach age 85 – a pretty small probability; see the life tables for the joint survival of a healthy 65-year-old couple! Even if both pass away before age 85, they will likely leave a small inheritance equal to the leftover TIPS ladder, while the longevity annuity becomes worthless. So, you must set aside extra money to leave an inheritance. Again, due to the inflation risk inherent in this strategy, it would be an apples-to-oranges comparison if we were to compare this 4.9% rate with a safe withdrawal rate calculation. However, I can see how retirees without children would opt for this route rather than taking their chances on the stock and bond markets over the next 30 years. You likely “Die with Zero,” but you maximize the cash flow now. But you better hope you don’t have any unforeseen emergencies later in retirement!
This approach is unattractive for younger retirees, at least with today’s interest rate landscape. The deferred annuity would be a lot more expensive for younger retirees. What’s more, for a 40-year-old couple, this approach would currently generate only a 3.73% SWR. That’s very low, considering you will get no more COLA to your retirement budget starting at age 60. That might be too early to scale back expenses for most active and healthy FIRE folks. You’d likely do better with a traditional retirement portfolio approach.
Safety First: practical concerns
Few retirees will be comfortable shifting their entire nest egg into annuities or TIPS. Some retirees don’t like annuities because they lose control over their money. Also, the portfolio-based approach offers you tremendous upside potential if we don’t have a repeat of the Great Depression or some other historical worst-case scenario. In the better-than-worst-case scenario, I’d have extra cash that could serve as a cushion to hedge against unforeseen healthcare shocks. Academic research points to health expenditure shocks as the main reason elderly households keep excess savings.
There are also practical concerns when shifting your entire portfolio. Retirees may have a large portion of their nest egg in taxable accounts, and liquidating existing assets with significant built-in capital gains may not be advisable. So, most retirees would like to keep a share of their nest egg in existing assets. Of course, not all is lost. A partial Safety First approach might still be helpful for early retirees who are concerned about Sequence Risk. This brings me to the next section…
Safety First “Light” to Hedge Sequence Risk in Early Retirement
Let’s assume we shift a portion of our portfolio into Safety First assets. Here’s an example:
- Let’s assume we start with a FIRE couple, both 50 years old, with a 50-year horizon and a 25% final value target, say, either as a bequest or a cushion in case there are unforeseen expenses late in retirement.
- Assume they have a $2,000,000 portfolio currently comprised of 75% stocks and 25% intermediate government bonds.
- The couple wants to shift half of their portfolio to “safety first” and keep the remaining half in their portfolio. This is possible because they have enough assets in tax-free accounts (e.g., Roth) and tax-deferred accounts (e.g., IRA/401k) to make the sale of appreciated assets and move to the Safety First assets tax-efficient.
- First, they have access to a joint-survivor annuity that would immediately pay $4,830 per month for $1,000,000 surrendered today; see the table with quotes above ($483 per $100k). I assume the annuity will slowly lose purchasing power via 2.50% annualized inflation.
- Second, they consider a TIPS ladder with a 2.25% real annualized return. How long should we plan the TIPS ladder? I played around with different lengths and found that 300 months or 25 years worked best in increasing the failsafe withdrawal rate. Assuming a monthly rate of 0.185594% (=1.0225^(1/12)-1), the monthly payment would be $4,342. Via Excel formula PMT(0.00185594, 300,-1000000,0,1)
- In the third scenario, we split the $1m half-and-half between the annuity and the TIPS ladder.
- The fourth scenario is the same as the third. However, we also moved the remaining portfolio into 100% equities because we already have plenty of fixed-income exposure and can now take more risk with the remainder.
Let’s look at the results. Here are the failsafe withdrawal amounts, i.e., the highest amount we could have withdrawn to succeed even in the worst-case scenario cohort; see the table below. Pretty impressive: the safety first portfolios all do significantly better than that baseline, but about 8-11%. The TIPS ladder would have done the best in historical cohorts. But the annuity is close behind. I also confirmed that with a slightly lower inflation rate going forward, 2.25%, the annuity would have outperformed the TIPS ladder by a few $100 annually. So, I don’t want to make too much of the TIPS vs. annuity comparison. They are both valid “safety first” assets.
Side note: TIPS have been around only since the late 1990s. We cannot reliably simulate how a TIPS portfolio would have performed in the 1930s. However, we can certainly perform a thought experiment: how a portfolio with some of today’s tools would perform if historical returns repeated themselves? I do not claim anyone could have used a TIPS portfolio or SPIA a hundred years ago!
Let’s not focus only on the worst-case scenario. If we look at withdrawal amounts to target other failure rates, we get the following picture: see the table below. The Safety First approach outperforms the 75/25 baseline even in the 10% worst scenarios. The price you pay is that you do worse in the median outcome and even the 25th percentile of historical cohorts (except for the half/half Safety First + 100% equities in the remaining portfolio). In other words, we should think of Safety First as insurance: it helps in the Sequence Risk worst-case scenarios, but you will underperform the baseline most of the time, i.e., when we don’t have another Great Depression or other economic and financial disaster scenario.
Of course, I could have made the picture look better if I hadn’t displayed the results for all historical cohorts but focused on the initial conditions that would have been most conducive to a Safety First approach. If I report the withdrawal amounts for failure probabilities conditional on expensive equities, i.e., the Shiller CAPE above 20 and the S&P index at an all-time high. And conditional on that initial condition, i.e., similar to the conditions we’re facing now, the Safety First approach certainly outperforms the baseline more reliably. The Safety First plus 100% equity portfolio does the best when looking at the median outcome.
Safety First vs. Glidepath
If you’re a regular reader of my blog and this Safe Withdrawal Rate Series, you’ll know that Safety First resembles something we’ve encountered before: increasing the fixed-income portion early in retirement and shifting to a more equity-centric portfolio is essentially a glidepath. I wrote about this approach in Parts 19 and 20. How would the Safety First portfolios compare to a simple glidepath, then? Let’s put that to the test in the worst-case cohort, i.e., September 1929. Let’s simulate a glidepath from a 37.5% stocks plus 62.5% bonds portfolio back to a 75/25 portfolio, shifting the weights linearly over 25 years, withdrawing 3.25% (=the baseline SWR). The final portfolio after 50 years would have been over $5m, much higher than with either of the Safety First portfolios.
So, Safety First is no new magic solution to the Sequence Risk problem. In the 1929 cohort, a simple glidepath would have achieved the same goal and even better than the Safety First approach.
Conclusion
Safety First has gained popularity in the personal finance world thanks to the rise in interest rates over the past two years. It’s an attractive approach for traditional retirees who do not plan for any (or at least no sizable) bequest and would instead maximize their guaranteed cash flow. Or maybe folks have already given their excess nest egg to their loved ones and charitable causes and now like to maximize their steady retirement cash flow and then literally “Die With Zero.”
But Safety First is no panacea, especially not for early retirees. Hedging expenses over a 50+-year retirement horizon is too expensive, even with today’s higher rates. Moreover, Safety First faces inflation risk when we use SPIAs, and even a TIPS ladder currently only hedges inflation for a maximum of 30 years.
But not all is lost. A partial shift to safe assets like TIPS and annuities can be worthwhile for most retirees. But it’s nothing new. It has the same flavor as a glidepath, i.e., start with a large bond allocation and liquidate those safe assets to avoid selling equities too early in case of an adverse Sequence Risk event. But I find the glidepath approach preferable: First, you don’t have to permanently give up control of your assets like in the case of an annuity. Second, you need less portfolio turnover; there is no tax headache when selling half of your portfolio to purchase an annuity or TIPS ladder. Third, the example of the 1929 cohort seems to indicate that the glidepath performed a bit better than the Safety First methodology.
So, my takeaway is that Safety First is a marketing gimmick by financial advisers to sell higher-commission products. Sophisticated investors can achieve the same or better results with a glidepath.
Please leave your comments and suggestions below! Also, check out the other parts of the series; see here for a guide to the different parts so far!
Title Picture Credit: knowyourmeme.com
The post Safety First – SWR Series Part 61 appeared first on Early Retirement Now.