How To Predict Withdrawal Rates Without A Crystal Ball

Feature

I’m always on the lookout for ways to improve the tools on Portfolio Charts, and Siamond really came through with his latest update to the Simba spreadsheet.  Buried in the heaps of interesting returns data is something really cool — direct calculations for safe and perpetual spending rates for a given investing period.  Based on an equation from a Morningstar white paper, they are particularly elegant compared to my old method and allow me to significantly improve the speed and stability of the Withdrawal Rates calculator.  And by doing so they open up a great deal of additional data that was previously too laborious to manage.

Well I’m a sucker for new data, and in the process of updating the calculator mechanics behind the scenes I took the opportunity to revisit an old question I’ve been wrestling with for a while.

 

How do you calculate a 40-year withdrawal rate when the worst start date for a particular portfolio was less than 40 years ago?

 

It’s a simple question which should naturally interest modern investors who lived through the market doldrums of the 2000’s and  wonder if old numbers will truly hold up for 21st century retirees.  But it immediately hits an intellectual wall, as we can only work with the data we have available.

Most withdrawal rate studies I’ve seen simply ignore the issue, although a few attempt to fill in future annual returns with Monte Carlo simulations or market projections of varying complexity.  I personally don’t care for those methods, so I’ve taken a reasonably simple approach of capping withdrawal rates based on looking at the worst withdrawal rates over ever-increasing timeframes.  That works just fine to prevent obvious outliers from popping up, but I’ve always intuitively thought there must be a better way to estimate the true number without resorting to fortune telling.

Well thanks to all the new data, I believe there is.  The trick is to forget about trying to predict future market returns and instead focus on the nature of withdrawal rate decay over time.

The what now?

Don’t worry — it’s less complicated than it sounds.  To explain, let me step back and walk through how the new calculation method works.

The new withdrawal rates equation in the Simba spreadsheet does not actually calculate the safe withdrawal rate as we all understand the term.  Instead, it simply finds the portfolio-depleting withdrawal rate for a single retirement year based on the real-life sequence of returns over a specified retirement length.  For example, let’s think about a person who retired in 1970.  If they only needed money for one year, the calculation reports that they could have withdrawn 100% of their portfolio at the beginning of that year.  That makes sense.  And the longer the retirement, the lower the allowable withdrawal rate every year.  If you map all of the withdrawal rates for this retiree up to a 40-year retirement, the chart looks like this:

 

1970-withdrawal-rates

 

Think of withdrawal rates as a waterfall flowing into a river. Just like water, they always flow downhill.  And while they fall precipitously over the initial retirement cliff, they eventually turn a corner and level out to a stable long-term perpetual rate.

But that’s only for a single start year.  What happens when you look at them all?

 

all-withdrawal-rates

 

Just like water droplets in a river, withdrawal rates are unpredictable.  They all start from the same point and they all flow downhill, but every retirement period is different and the numbers vary quite a bit.  So how do you know which withdrawal rate to use?

Safe withdrawal rates essentially look at this chart and seek out the riverbed — the bottom-most path that the water flows over.  The most popular studies tend to fixate solely on the 30-year mark for a very particular portfolio, but the same methodology can be used to find the safe withdrawal rate for any portfolio and retirement duration.

Now I’m going to crop the same chart at a 10% withdrawal rate and a minimum 10-year retirement duration:

 

60-40-WR-example

 

On this scale, it’s easier to see the individual endpoints.  These are from more recent start dates that have yet to reach the 40-year retirement mark.  The worst retirement date for the Classic 60-40 portfolio was more than 40 years ago, so the bottom line makes it all the way to the right.  But not every portfolio has the same worst start date, so let’s look at another example:

 

wr-example-2

 

Here you can see the problem.  If you simply calculate the withdrawal rate for the worst 40-year investing period we have data for, you completely miss the more recent start dates that are trending far worse!

So how are we supposed to extend the endpoints without more data?  To understand that, let’s take a look at how the trajectories develop over time.

 

This slideshow requires JavaScript.

 

An interesting thing to note is that the first image in the series was calculated based on data available in 2007, just before the 2008 market turmoil and a sizable 33% drop in portfolio value.  While you can just make out a small dip in each line after that, this illustrates that even large market swings later on in retirement have a much smaller effect on withdrawal rates than you might think.  As a result, one can easily look at a chart like this and intuitively estimate where the lowest trajectories will ultimately end up once they have enough data to qualify for the 40-year mark.  And thanks to a bit of number crunching, I can also do that automatically.

 

wr-example-2b

 

Withdrawal rates for individual years erode the safe riverbed below.  So while the inflated worst-case scenarios past the 28 year mark in the above example have yet to be worn down, their demise is inevitable and the new floor is actually somewhat predictable.  The dotted line isn’t set in stone by any means and may change as new data becomes available, but it represents our best guess to the long-term safe withdrawal rate with no future market predictions required.

Pretty cool, right?

I find this type of information really interesting, so I’ve decided to redesign the Withdrawal Rates calculator around the new visual.  Here’s the same portfolio as before:

 

WR-calculator-example

 

There’s a lot of information packed in the new chart, so let me give a quick rundown.

The blue lines represent the verified withdrawal rates over increasing timeframes for every start date since 1970 for the specified portfolio.  The gray lines indicate the paths that are only estimated using the best data I have available.  (For those who need more explanation of the difference between verified and estimated results, read this.)  The orange line traces the worst-case scenario, including a dotted projection when necessary.  This marks the safe withdrawal rate for the portfolio.

The green line takes this worst-case scenario and asks a different question.  What if the retiree did not spend his retirement down to zero, but instead only withdrew an amount that eventually preserved the inflation-adjusted principal?  This is the perpetual withdrawal rate, which I personally prefer for early retirees or for people who desire to leave an inheritance for heirs.  Any withdrawal rate below that line was not only safe, but also resulted in a growing portfolio.

Some people may wonder why I include all of the blue lines and don’t simply show the safe and perpetual rates like I did before.  Let me make one thing very clear — I do NOT promote gambling on a withdrawal rate higher than the safe rate for your own personal portfolio.  It’s always best to assume that you’re not special and that the worst case others before you experienced can also happen to you.  But there’s a lot of information baked into those blue lines that you may also find valuable, and offering a view of the mechanics behind the numbers is one of my personal goals for the site.

As an example, many people wrongly believe that choosing an asset allocation is all about cherry picking data and that all portfolios are equally uncertain.  However, look at the blue lines and you’ll quickly notice that some portfolios are definitely more consistent than others — not just in accumulation but also in retirement.

 

This slideshow requires JavaScript.

 

Another benefit to the lines is that they help expose dangers lurking behind traditional SWR numbers when a modern trajectory is very likely destined to cross an old one and set a new low.  I don’t attempt to project anything but the already lowest endpoint, but your eyes can do the work for you.  Take a look at a chart with shorter lines screaming towards the orange riverbed, and you may have second thoughts about the future of that particular withdrawal rate.

 

WR-bad-future

 

Study the numbers for a while, and you’ll not only learn about withdrawal rates for a certain portfolio but also how they work.  Armed with that information, my hope is that you’ll be better prepared build your own secure retirement plan with your eyes wide open to long-term withdrawal rates both known and projected.  But leave the crystal ball in the closet, as it’s all about the math.

And a fun new chart!