June 6, 2022
In this year’s April Fool’s post, I marketed a made-up crypto coin that would completely hedge against Sequence Risk, the dreaded destroyer of retirement dreams. Once and for all! Most readers would have figured out this was a hoax because that complete hedge against Sequence Risk is still elusive after so many posts in my series. Sure, there are a few minor adjustments we can make, like an equity glidepath, either directly, see Part 19 and Part 20, or disguised as a “bucket strategy” (Part 48). We could very cautiously(!) use leverage – see Part 49 (static version) and Part 52 (dynamic/timing leverage), and maybe find a few additional small dials here and there to take the edge off the scary Sequence Risk. But a complete hedge is not so easy.
Well, maybe there is an easy solution. It’s the one I vaguely hinted at when I first wrote about the ins and outs of Sequence Risk back in 2017. You see, there is one type of investor who’s insulated from Sequence Risk: a buy-and-hold investor. If you invest $1 today and make neither contributions nor withdrawal withdrawals, then the final net worth after, say, 30 years is entirely determined by the compounded average growth rate. Not the sequence, because when multiplying the (1+r1) through (1+r30), the order of multiplication is irrelevant. If a retiree could be matched with a saver who contributes the exact same amount as the retiree’s cash flow needs, then the two combined, as a team, are a buy and hold investor – shielded from Sequence Risk. It’s because savers and retirees will always be on “opposite sides” of sequence risk. For example, low returns early on and high returns later will hurt the retiree and benefit the saver. And vice versa. If a retiree and a retirement saver could team up and find a way to compensate each other for their potential good or bad luck we could eliminate Sequence Risk.
I will go through a few scenarios and simulations to showcase the power of this team effort. But there are also a few headaches arising when trying to implement such a scheme. Let’s take a closer look…
Introducing the RSIP: a Retiree-Saver Investment Pact
We’d need to pair up a retiree and a saver or groups of retirees and savers whose cash flows exactly cancel out each other. Then at the end of the contract period, both retiree and saver will receive a respective portfolio value they would have achieved had the return pattern been one flat monthly or annual return matching the CAGR during the contract period, i.e., in the absence of Sequence Risk.
Imagine, for simplicity, that we have a retiree with a $1m initial portfolio with $40,000 in annual cash flow needs and a retirement saver who starts with a $0 portfolio and saves $40,000 annually. Assume that they agree to offset each other’s cash flows over a set contract period to generate a buy-and-hold investor if aggregating the two cash flows. For any realized buy-and-hold investor CAGR over this period we can now calculate the final values of the retiree and saver portfolios using the Excel future value (FV) function:
=FV( CAGR ,Nyears , 40000,-1000000,1) (retiree) =FV( CAGR ,Nyears ,-40000, 0,1) (saver)
And again, notice how the final values depend solely on the CAGR. Not on the Sequence of Returns! In any case, before we even get into any simulations, let’s run a simple example to warm up. Imagine both retiree and saver like to eliminate Sequence Risk over a 10-year horizon. They may each have a longer horizon, but they decide to sign this pact over a 10-year horizon, so bear with me.
Let me first illustrate the workings of Sequence Risk again. Let’s assume that over the 10-year horizon a portfolio of risky assets returns 5% (inflation-adjusted) on average, measured by the CAGR. Let’s assume that returns can be High (+29.71%), Moderate (5%), or Low (-15%). Why that crooked number of 29.71%? Simple, that ensures that one high and one low return combined get you back exactly to 5% CAGR. Check the math if you like: 1.2971*0.85=1.1025=1.05^2! Now let’s look at 7 different sequence risk scenarios. We can start with the “MMM” scenario where we have flat 5% returns every year, plus 6 additional scenarios: each with 6 years of moderate returns, 2 years of high, and 2 years of low returns in varying orders. Notice again that all 7 scenarios have the same CAGR:
We can briefly confirm that a buy-and-hold investor achieves the same final net worth regardless of the order of returns. Here’s a chart:
And below is a table with the same information. Notice again that the final value has to match up exactly, even though the path over time can be very different:
While the buy-and-hold investor is shielded from Sequence Risk, the retiree will certainly not be indifferent to the sequence of returns. Here’s a time series chart of the retiree, starting with a $1,000,000 portfolio and withdrawing $40,000 each year (at the beginning of the year). Notice that the balances are plotted before the withdrawals. So X(0) = $1,000,000 and X(t)=[X(t-1)-$40,000]*[1+R(t)], t=1,…,10.
Now we get some action! The final balances range from $887k to $1.244m. The most advantageous scenario is when the high returns hit first and the low returns come last, and vice versa – no surprise here!
The same info is in the table below. Notice how this retiree’s IRR can be significantly different from the realized CAGR, all due to Sequence Risk. If a retiree had committed to this RSIP contract, he or she would be guaranteed to walk away with $1,100,623 after 10 years, regardless of the sequence of returns. So, for example, under the HML scenario, the retiree would give $143,578 (=$1,244,201-$1,100,623) to the saver. But in the worst-case LMH scenario, the retiree would receive $213,333 ($1,100,623-$887,291) from the saver.
Now the same exercise for the saver:
And all that again in a table format below. Notice how the range of IRRs is even wider for the saver (makes sense due to the $0 initial balance). It’s quite intriguing how merely reshuffling the order of returns will transform a 5% CAGR into anything between a -0.71% to +10.98% IRR. The best-case scenario (LMH) gives you almost a 2x relative to the worst-case scenario (HML). All due to Sequence Risk!
And just to confirm, the payments are now exactly flipped: In the HML scenario, the saver would receive $143,578 from the retiree, while in the LMH scenario, the saver would pay $213,333 to the retiree. This would exactly guarantee a $528,271 payout for the saver, regardless of the sequence of returns. Pretty cool!
Summary so far: If a saver and retiree could sign a binding contract to balance their portfolios back to the “MMM” scenario, i.e., flat returns equal to the realized CAGR, then we could certainly take a bite out of Sequence Risk. It’s also important to note that it’s not just the retiree who would benefit from this scheme. Savers face significant Sequence Risk. In other words, to all of us who retired recently, let’s not get too cocky. Our investment success is mostly due to luck in the form of both high average returns and a very advantageous Sequence Risk outcome: low in 2008/9, moderate for a few years, and then spectacular in the latter part, especially in 2019/2020/2021. The next cohort of savers may not be so lucky and can certainly benefit from such a scheme!
Simulations with Historical Data
Let’s run some historical simulation with actual return data 01/1871-04/2022. Let’s assume again that the initial portfolios are $1m for the retiree and $0 for the saver and the annual withdrawals of $40,000 exactly offset the annual contributions of the saver. To be consistent with my other SWR work, I assume that we run this at a monthly frequency (=$3,333.33 of monthly withdrawals/contributions). The time horizon is 30 years and the retiree has a 75% stock, 25% bond portfolio (intermediate 10-year U.S. Benchmark Treasury bonds). The account values and withdrawals/contributions are adjusted for inflation, as always.
Let’s first look at the final outcome the retiree could achieve without the scheme, i.e., when subject to Sequence Risk (blue dots) and with the RSIP, i.e., when hedging the Sequence Risk (orange dots) in the chart below. Luckily for the retiree, even at the lowest 30-year CAGR, you would not have run out of money. But without the RSIP there were plenty of occasions where you would have depleted your portfolio. Intriguingly, the retirement bust scenarios occur when the 30-year CAGR of the buy-and-hold 75/25 portfolio was between about 3.8% and 6.5%. This confirms again that Sequence Risk is a much larger headache than merely low average returns. Over a 30-year horizon, you need a mere 1.3% flat real return to exactly deplete your portfolio. That’s a low bar. All the retirement failures are squarely due to Sequence Risk, not the CAGR falling below 1.3%!
And the same for the saver, see below. Clearly, the saver will not run out of money but look again at the dispersion of the blue dots around the orange line: Several million dollars in final value uncertainty. A lot of retirement savers may be willing to give up the upside to hedge the downside risk. Give up the prospect of a $7m retirement portfolio but vastly reduce the possibility of falling short of $2m.
Instead of plotting the final values, we can also calculate the IRR for the different retirement and saver cohorts. Excel has a neat function for translating a present value (PV), a future value (FV), and regular payments into an IRR.
=(1+RATE(360,0.04/12,-1,FV,1))^12-1 (retiree) =(1+RATE(360,0.04/12,0,-FV,1))^12-1 (saver)
Notice that I run this at a monthly frequency, so the payments are 0.04/12, but then the IRR has to be annualized.
Let’s plot this for the retiree first, see below. I plot the actual retiree cohort IRRs (blue dots) and also a 45-degree line because that 45-degree line is the IRR you could have gotten with the RSIP, i.e., in the absence of Sequence Risk.
Again: you win some, you lose some. The blue dots are scattered wildly around the CAGR. For example, this chart demonstrates how Sequence Risk can turn a 6% CAGR for the buy-and-hold investor into a -1% IRR for a retiree. By the way, this cohort with the -0.73% IRR is the December 1968 cohort. The average return from December 1968 to December 1998 was an impressive 6.02% (real). But due to Sequence Risk, the retiree got shafted with a negative IRR (while the Dec 1968 saver cohort got a +9.30% IRR!).
And the same for the saver, see the chart below. Again, we get a wide dispersion of IRRs around the 45-degree line. The historical range of IRRs ranged from 1.5% to 9%, while the realized IRRs of the saver ranged from -0.5% to just under 10%.
At first glance, this seems to be an easy way to accomplish a hedge against Sequence Risk. Hey, maybe we could run our own little FIRE quasi-pension fund. But the devil is in the details. Here are a few headaches I can think of:
Asset allocation: Savers and retirees may not want to hold the same asset allocation. For example, fresh retirees will likely opt for a slightly more conservative asset allocation, maybe about 75% stocks and 25% diversifying assets like longer-term bonds and/or short-term fixed income instruments, while young savers might want to be more aggressive. That’s not really an insurmountable obstacle because if young savers prefer 100% equities then the retirees may then contribute only their equity portfolio to the pact. And maybe construct a bond ladder to supplement the retirement income.
Horizon: retirees and savers might have different time horizons. For traditional retirees (30-40 years horizon) and retirement savers (also a 30-40 years horizon), this might all work really beautifully. But in the FIRE community, we have this slightly lopsided distribution: maybe 10-15 years of accumulation and then 40-60 years of retirement. Of course, one retiree cohort could always be paired with a new set of fresh FIRE savers once the current one archives financial independence. But the problem with this idea is that in order for the RSIP to work we’d need to ensure that the duration of the pact is long enough that a long stretch of bad returns can be offset by a new bull run. 10-15 years might not be enough. There were plenty of poor return windows for the stock market: 1929-1942, 1965-1982, 2000-2009 when a 10-15 year window would have been too short to effectively hedge against sequence Risk. You would have needed the subsequent 10-15 years to truly smooth out that Sequence Risk. 20-30 years seems to be the minimum to capture enough of a large macrocycle to include both poor and blockbuster returns.
Maybe the solution would be to pair traditional retirement savers with a 30-40-year horizon with FIRE enthusiasts and cover their first 30-40 years in retirement. FIRE savers with their short horizons may be less-than-ideal candidates for this scheme.
Taxation: The flows between two parties in this pact will likely draw the attention of the IRS. How do we tax the transfers from one group to the other? Capital gains? Ordinary income? How do we deal with the cost basis in taxable accounts? My suspicion is that this plan would work best if we implemented it in a retirement account where we don’t have to deal with a lot of the taxation issues of reshuffling assets.
Safety: Before I hand over any sum of money I’d need to see some assurances that people won’t abscond with my hard-ERNed cash. We could implement this RSIP through a reputable financial institution, think Fidelity or Vanguard. Or an insurance company. Or maybe this is a blockchain application where we could cut out the greedy financial companies and do this peer-to-peer start-to-finish. Some silicon valley whiz-kids might want to take a shot at this!
Commitment: Both parties – saver and retiree – will initially enter this pact voluntarily and willingly, because on an ex-ante basis it is advantageous to hedge against risk. But people might regret this pact ex-post after the asset returns have come in. No, let me correct this, exactly one side of the deal will most certainly regret participating in the deal. Because that’s the whole idea of this peer-to-peer insurance contract: one side’s gain is the other side’s loss. We only enter this insurance contract because we believe that there could be some large net payments and ex-ante we prefer to minimize the risk. If I find out, ex-post, that Sequence Risk helped my portfolio, I’d rather run with my money instead of sharing some of the Sequence Risk gains.
And this commitment problem is worse for the saver who needs to contribute the regular flows. Maybe institutional investors like pension funds could take the side of the saver. That would be a challenge again, though, because it would have to be a “young” pension fund with mostly savers and very few retirees and thus net inflows. The current pension fund landscape is the exact opposite: most companies have phased out their defined benefit plans and replaced them with defined contribution plans. The existing pension funds still around are mature funds with mostly aging beneficiaries and net outflows. Not much help there!
Also, the willingness for continued participation is not always the biggest problem. There’s also an issue with the ability to participate: This is less of an issue for the retiree who simply sits back and collects $40,000 checks every year. But a huge concern for the saver. What if savers lose their job, die, or become disabled?
Retirement ruin: Even though both parties hedge against sequence risk, there is no insurance against the average realized asset return. If the average equity return is low enough then the retiree can still run out of money. Historically, there would have been a few cases where a $40,000 annual withdrawal would have wiped out a 100% equity portfolio even with the RSIP.
Summary so far: Maybe this RSIP is mostly a cute theoretical construct but not so easy to implement. I’m open to suggestions for how to make this work in real life. Please use the comments sections if you want to help.
But maybe we could implement transfer payments without a specific counterparty. This brings me to the next point…
Can I self-insure against Sequence Risk using a time-varying asset allocation and/or derivatives?
Is it possible to generate the sequence-contingent transfer payments just on my own? Without any direct counterparty that may or may not be willing to adhere to the multi-decade investment pact? One way would be to devise a derivative-based strategy to at least roughly mimic the transfer payments between the retiree and the saver. The advantage of this approach is that many of the headaches listed above will go away. For example, if I were to generate transfer payments modeled after this strategy through exchange-traded derivatives I would not have to worry about my counter-party walking away from the deal. (well, there is a minute risk of both the counterparty and the exchange going belly-up simultaneously, but let’s not even go there…)
Taxation would also become a lot clearer: Index futures and options enjoy tax-advantaged treatment under IRS Section 1256, as my readers know from my options-trading posts. One obstacle, though would be that, some options strategies may work only in taxable brokerage accounts, not in retirement accounts!
There may be many different ways to structure this, but the most obvious one is this: Imagine a retiree expects a target real return of 4% p.a. over a 30-year horizon and a 4% annual withdrawal rate. And for this simple example, I go back to annual withdrawals, taken at the beginning of the year.
Imagine that in year 1, this retiree suffers a 10% drop in the portfolio. Assuming that during the remaining 29 years we revert back to the 4% target we can project the portfolio’s final value. That’s what I did in the table below. The retiree is expected to end up with $491,117 (in real, inflation-adjusted dollars). If the retiree had access to the RSIP, SoRR insurance we can calculate the 30-year CAGR as 3.50%, i.e., CAGR of one year of -10% and 29 years of 4% return. The projected final portfolio value in the absence of Sequence Risk, i.e., with a fixed 3.5% rate of return, is $669,611. So, if you had access to Sequence Risk hedging you’d stand to get a transfer of $178,494 in year 30. Discounting this payment back to year 1 at an annual rate of 4% would amount to $57,234.
We can also calculate this transfer payment for several different Y1 returns. Let’s calculate the transfer payment for Y1 returns of -15%, -10%, -5%, 0%, 4%, 10%, 15%, 20%, and 25%. Obviously, the transfer has to be zero for the Y1 return of 4%, but it’s good to confirm and check the math. I plot the scatter plot of the Y1 return (x-axis) vs. the transfer payment (y-axis) in the chart below:
That line is almost a perfectly straight line. That’s not too surprising because even though the future value after 30 years is a non-linear function, by taking the difference between the two versions, once with and once without SoRR, we generate a function almost exactly straight. This particular line has a slope of about -$400k and an intercept of just under $17k.
So, how could we replicate this blue line with a derivatives strategy? It’s not that easy! We could certainly sell a futures contract with a notional of $400,000. But we wouldn’t be able to get the $17,000 intercept.
Likewise, we could sell $400,000 of the risky assets to exactly replicate the slope of the blue line. But in taxable accounts that might become a big headache from a taxation perspective given our progressive income tax function. And even in tax-deferred accounts, selling assets is useless unless we find a safe investment with a 4.25% return (17,000/400,000=0.0425=4.25%) to shift the line up to its intercept. I-Bonds currently yield 0% above inflation, TIPS also around 0%, and nominal bonds around 2.1% for one year, which will likely get you to -4.25% rather than +4.25% after inflation. Not a pretty sight! In fact, if we had access to a 4.25% safe return, we wouldn’t have to worry about Sequence Risk and hedging against an uncertain Y1 return, would we? We would just move the entire portfolio to that asset, raise the withdrawal rate to 4.25%, always preserve our portfolio, and live happily ever after.
One could argue that after a big drop in the stock market, we’d likely also expect slightly higher returns return in years 2-30, from a valuation point of view. By discounting at a higher rate we might be able to push that intercept down a bit. But not by much. Not by enough to bring the required return down to the 0% real return we face today.
I guess the perfect solution to Sequence Risk is still elusive. The RSIP is a cute theoretical and mathematical concept but implementing it directly through a retiree plus saver partnership faces a lot of obstacles. On the top of the list is the commitment problem of the saver. Maybe an insurance company or large brokerage company could take the counterpart and offer Sequence Risk insurance to retirees. But I’m concerned that the fees involved would likely wipe out any potential gains. Trying to implement a transfer payment scheme through a derivatives strategy is also an uphill battle. But I’m open to suggestions. Please share in the comments section if you have better ideas on how to make this work!
Thanks for stopping by today! Please leave your comments and suggestions below! Also, make sure you check out the other parts of the series, see here for a guide to the different parts so far!
Title Picture credit: pixabay.com