In our journey for Financial Independence and the goal to Retire Early (FIRE) one area of concern we have is what happens if we retire at or near the market peak? We have had many years of continuous investment growth and are at an all-time high for the stock markets. Our timing for Retire Early could be as soon as next year but will be in the next 2-3 years.
Markets do revert to the mean, and we expect the growth to slow down with a high probability of a market correction in our initial retirement period. Does the 4% Safe Withdrawal Rate still work and how do we plan for this future?
tl;dr Sequence risk is the potential for the markets to under perform during your initial retirement period. When market valuations are high, as they are today based on the Schiller CAPE ratio, the probability is higher for the market to under perform in the following ten years. If you are planning to retire in the near future, this risk can reduce the probability a 4% Safe Withdrawal rate will succeed for longer retirement periods.
Just a note this article has a lot of technical information which may go into more details than you prefer. If you don’t want to read all the details, just jump to the end to see the impact on our planning. This research is based on analysis of historic data. There is no guarantee historic data will accurately represent future performance.
We have been looking to complete the analysis of sequence risk since we are worried about sequence risk for our own retirement plans. A summary of the outcome of our own analysis is provided, where we apply sequence risk to our retirement scenario. This may not apply to your own situation.
What is Sequence Risk and Why Does It Matter?
Sequence risk is the timing of market returns and having an extended period with low or no growth at the beginning of your retirement. Michael Kitces, a financial planner and blogger, has done extensive research about sequence risk and safe withdrawal rates. In his article “Understanding Sequence of Return Risk – Safe Withdrawal Rates, Bear Market Crashes, And Bad Decades”, he writes:
… the real problem is not a bad year or two at the beginning of retirement, but a bad decade to start off retirement. A bad decade outlasts most cash reserve strategies … It’s a slow inexorable grind that whittles down the portfolio to the point there’s just not enough to recover, and there are few places to hide after 10 years of poor returns. In fact, as it turns out 10 years really is the “sweet spot” for sequence of returns risk … (1)Michael Kitces
Long Periods of No Growth
The worst situation is not one or two bad years, the models will typically capture this. It is an extend period of low or no market appreciation. In a recent interview with the Mad Fientist (one of our favorite FI Podcasts), Michael Kitces provides this anecdote:
So, if you imagine being in retirement, and for the first 15 years, the market gives you no capital appreciation whatsoever, you begin to get a sense of what it was like to be a 1966 retiree. Now, on top of that, it gets even worse because inflation went from about two to twelve, which also caused the worst bond bear market of the century at the same time that stocks generated no appreciation for 15 years. (2)Michael Kitces
For our situation, we don’t need to worry about one or two bad years but the performance over the first decade of our retirement. This provides the highest risk to your retirement plan. Michael Kitces writes
… it turns out that the true driver of sequence of return risk and safe withdrawal rates are the returns that the retiree earns over the first decade – and specifically, the real returns over the first decade, that provide an indication of whether the retirement portfolio will have produced enough real growth to keep up with inflation-adjusted spending for the rest of retirement (1)Michael Kitces
So, how can we account for this in our planning?
Is the Market Over Valued?
How do we know the current market is overvalued and may be ready for a correction? People have studied this for years and there is no perfect answer.
One common measure developed by Robert Shiller, a Nobel Peace Prize winning economist, is something called the P/E 10 or the Cyclically Adjusted Price to Earnings (CAPE). The CAPE provides a figure for valuation of the market based on the current price divided by the past ten-year earnings. It is not perfect since future earnings may be different than the past ten years, so valuation may be off.
CAPE or P/E 10 is one of the most common referenced market valuation indicators. You can read much more on this in Robert Shiller’s book Irrational Exuberance. (6)
Shiller CAPE Data
Robert Shiller has conveniently provided this data to review going all the way back before 1900 using the S&P 500 as a proxy for the US market. The following graph shows the CAPE from 1900 to August 2018. (3)
As we can see the most recent value of CAPE is 32.29 and looking at the graph it is at one of the highest levels since over the last 100+ years. If you exclude the period back in 2000 around the internet bubble, this is close to the highest we have seen! In October 1929, it was at 32.56 right before the infamous “Black Thursday” that started the market crash and led to the Great Depression.
Market Valuation and CAPE
Shiller write in his book about market valuations:
… as a rule and on average, years with low CAPEs have been followed by high returns, and years with high CAPEs have been followed but low or negative returns. … Long Term Investors would be well advised, individually, to lower their exposure to the stock market when it high, other things equal, and get into the market when it is low.
Of Course, other things are not always equal. at the time of this writing, in 2014, the CAPE is very high but so too are the prices of alternate investments and long-term bond prices, and short-term interest rates are virtually zero” (6)Robert Shiller
We are not suggesting we are heading for another “Great Crash”, but the CAPE value does suggest the current market may be overvalued. Unfortunately, today other alternative such as bonds also have historic low returns, so we may need to ride this next period out with lower expected returns.
What can we do with this information?
Can We Use CAPE to Predict the Future?
May people have tried to use the CAPE data to predict short term market fluctuation without much success. It really is impossible to know what will happen in the next day, week, or month. For retirement planning we really are not worried about one or two years but the longer-term trends. Could we use CAPE to help predict longer term trends?
Kitces has researched Shiller’s CAPE in its predictive ability for retirement planning. In his article, “Shiller CAPE Market Valuation: Terrible for Market Timing, But Valuable for Long-Term Retirement Planning” he states
… the reality is that while Shiller CAPE has little predictive value in the short term, its correlation to market returns is far stronger over longer time periods; Shiller CAPE shows its strongest correlation to nominal returns over an 8-year time horizon, and is actually most predictive of real returns over an 18 year time horizon… (4)Michael Kitces
Kitces confirms what Schiller noted that there is an inverse correlation between market valuation using the Shiller CAPE ratio and longer-term future market performance! The trend seems to work best for a medium term 8 to 18-year outlook. This period is very close to the window we are worried about for initial sequence risk.
Cyclically Adjusted Earnings to Price (CAEP)
Kitces uses the inverse of the CAPE to create the Cyclically Adjusted Earnings to Price (CAEP). He shows how it is correlated to ten-year future market returns. It also has the highest correlation, of 0.53, when looking at the future ten-year market performance. This is over double the correlation of 0.23 when only looking at one-year performance. It is not perfect, but we can use this to help guide our planning.
The graphs below show the correlation as published in his article “Should We Forget Shiller CAPE Ratios and Focus On E/P Instead?” (5).
Note: a lower value of CAEP indicates a higher valuation.
As of the writing of this article, the Shiller CAPE is sitting at 32.29 which is equal to CAEP of 3.1% and indicates very high valuation by historic standards. The longer-term average is about 16.56.
We have gone from a CAEP (inverse of the CAPE) of 7.5% in 2008 to 3.1% for August 2018. The average return the past ten years (S&P 500) of approximately 12%. Based on the chart above and the current CAEP of 3.1%, we can expect to see much lower returns on average for the next ten years!
Do We Need to Adjust our Safe Withdrawal Rate?
We noted in our initial review of sequence risk, one of the worst periods to begin your retirement was in 1966 where the following fifteen years had little or no capital appreciate and high inflation.
Even in this worst-case scenario, Michael Kitces writes:
And despite that, or even through all of that, what we find is this 4% initial withdrawal rate adjusting for inflation works. … as bad as it can get when you get these bad sequences, what we still ultimately found is it still doesn’t seem to get any worse than about 4%. (1)Michael Kitces
Kitces has written several articles on safe withdrawal rates. Below is a graph showing the market performance (CAEP) correlation to safe withdrawal rates for a thirty-year retirement.
If the market valuation is high prior to retirement, we may want to reduce our Safe Withdrawal Rate during our initial years of retirement to reduce the effect of the sequence risk of returns.
Does the 4% Safe Withdrawal Rate Still Hold Up?
Interesting, we see a Safe Withdrawal Rate above 4% for the worst-case periods, but this was for a shorter retirement window of 30 years (1)
For those of use on the FIRE path, we will have a longer retirement window, so you if the market valuation is high near retirement you may want to save a bit more and shoot for a 3.5% Withdrawal Rate.
In our own situation, we found 4% hold up well even for longer retirement periods but we do want to now consider sequence risk and add this to our simulation.
How do We Model Sequence Risk for Our Retirement Plans?
How do we use this information in modelling our own retirement and what investment performance can we use? Schiller provides his data updated monthly in an Excel file (3) so we can recreate some of the same tables and graphs Kitces uses to help see how we can better use this for our own simulations.
- Take the CAPE data provided by Shiller and calculate the inverse for CAEP.
- Calculate the following ten-year average annualized market performance.
- Used data from 1900 through August 2008 (we needed to look back ten years to get the ten-year performance).
Re-Creating the CAEP Correlation Data
The first graph is like the correlation graph Kitces produced. His data ended in 2004 so we updated to 2008. This uses 90-year data from 1918-2008. We calculated the CAEP each and ten-year return for each month. There are 1,080 data points (12 months x 90 years).
The blue line is a logarithmic trend line for the data set. It does clearly show the same correlation between CAEP and future market performance. Two things we see is the average returns decreases as market valuation increase but also market variability increases as valuation decreases. Our model should account for this.
For reference the two red vertical lines are at a CAEP of 7.5% and 3.1%. In 2008, the CAEP was 7.5% and we have enjoyed about a 12% annualized return since then. 12% is just above the mean distribution at 7.5% CAEP. As of this article, the CAEP is about 3.1% so you can see the returns for past historical periods with this valuation were much less.
Can we use this data to adjust our retirement performance?
Now that we have the data, can we structure it to get some meaningful historic market performance for our simulation. Since most simulation apply historic market performance, we can tweak the investment performance based on the current market valuation and update our model.
The table below buckets the CAEP data in 0.5% increments and calculates the mean, min, max, and standard deviation for these buckets. We only used data with a minimum of ten results so limited the CAEP to 13%. Above 13% we had less than ten values, historically the CAEP went as high as 21% with only one data point. This data set as 1,231 periods.
Below is the graphical representation of this data. This is like the scatter chart shown early with just showing the min, max, and mean lines and limited to a CAEP of 13%.
From this data we can now quantify the historic market returns for the closest 0.5% CAEP values. Note: Some of these values have smaller samples sizes so future performance will vary more than the number suggest. It does gives us something to work with in our models.
Too Much Precision?
One thing to note is a higher CAEP (lower valuation and risk) in some cases produced lower returns based on this data grouping.
If you selected CAEP of 4.0% it would have a mean return of 5.8% but moving up to 4.5% which should be lower valuation and lower risk the mean return is only 2.3%. More precision in this case does not provide more accuracy.
Using 0.5% increments of CAEP seems like too much precision for this exercise. Remember from above, the correlation is strong at 0.53 but not perfect. More precision does not help us. Another issues is as the CAEP gets lower the sample sizes are much smaller as seen in the chart below.
The next chart shows the distribution of CAEP values with 6.0% (red line) being the average, we see most of data points fall within 4% to 9% CAEP (green line).
Simplifying the Sequence Risk for Our Model
Since we are using this to generate a statistical model for sequence risk for our simulation, we decided to simplify the data set and group data below 6% CAEP into three bands for High, Medium, and Low sequence risk. Our choices were a little arbitrary, though it still provides a good historical basis for sequence risk performance without trying to get too precise.
Shown in the scatter graph above, High is at a midpoint of CAEP = 3% (+/- 1%). Medium is at a midpoint of CAEP = 4% (+/- 1%). Low is at a midpoint of CAEP = 5% (+/- 1%). Since the long-term average is 6.0% CAEP (~16.5 CAPE from 1918-2008) above 6% would be considered nominal returns and should just use the long-term averages. This is summarized in the chart below.
We added this table to our retirement simulation tool to allow adding the option to simulate sequence risk. Users may want to use other value. A user select option to input custom sequence risk is also added. In most cases the default ranges based on this analysis should be sufficient.
Send us a note below if you want a copy of this data set for your own analysis. We will send you a link to the Excel file we used to generate this data.
Let’s Update the Retirement Simulation Tool
We added a Sequence Risk option to our simulation tool in Version 9E. To make it simple for the user, we used a VLOOKUP using the sequence risk level to select the historic market performance of the initial 10-year retirement period.
After ten years of retirement, we used the longer-term investment performance data entered on the input page. We used CAPE value for the reference since it is easier to look up and can be found here.
Our baseline success rate is from our article on dual life expectancy. In this model, we had a success rate of 98.3% with modeling a 55-year retirement considering our dual life expectancy. This is close to our real-life situation if we were to retire early next year.
Note: A bug was found in the Simulation Tool where the average investment performance was fixed to 9.5% with a standard deviation of 8.9% for both pre and post retirement simulations. This was roughly about 50/50 stock/bond performance using 50 years of historic data. Results are still valid, just note the portfolio ratio is not as shown below.
We entered the same data into Version 9E of the Retirement Simulation Tool and selected High sequence risk in the options.
Then set our options for the second spouse reduced life expectancy and added the sequence risk lookup. We entered a High Risk with the CAPE close to 33. We ran 10,000 life times for the simulation to estimate our probability for success.
The Simulation Results
Our success rate dropped from 98.3% to 69.2%. Yikes! Looks like we have some more planning to do to adjust for the possibility of sequence risk.
For our personal situation, initial models looked great at a 4% withdrawal rate. This new data suggests for a longer retirement period with high sequence risk, 4% may not be high enough. One interesting thing to note is even with high sequence risk, our savings has a 97.5% chance to last 25 years (75 years old). Will Social Security provide a safety net to our plans and improve our chance for success?
We will explore this more as we adjust our planning to account for sequence risk in a future article. How we can adjust our plan to get our success rate back close to 95% or are we OK with a little higher risk in our success rate?
Don’t Panic! It’s Just A Model and Stay Flexible
If you are in a similar situation, for now don’t panic. This is one more piece of data to use for our planning. We really don’t know for sure what the future has in store but can use this correlation to help manage our planning and reduce our risks. If you are planning to retire soon with high market valuations, it would be wise to adjust the initial withdrawal rate to be below 4% or model the sequence risk historic returns as we did for our case.
As with all the models the real world may not cooperate, so stay flexible in your plans. We really can’t predict the future, but we can model scenarios and plan accordingly. Retirement planning is a marathon and not a sprint, and over time priorities and personal situations change so stay flexible. Don’t get stuck on specific number or date.
One thing the models don’t account for is we can be flexible, so don’t get too caught up in 100% probabilities especially if your retirement date is far in the future. Extending your retirement by one or two-years during a down market can make significant improvements in your future success.
If you are already retired, postpone the big vacation, take a few smaller road trips. Or consider Barista FIRE for a few years with a part time job or side hustle. The tradeoff is waiting for the “perfect” time vs. getting enjoy your retirement sooner. We think the tradeoff of a little more risk is worth the benefit of retiring earlier and will stay flexible. Mrs. FoF is already talking about Barista FIRE.
Our version 8.0 of the simulation tool can be found on our website. Version 9E update will be released very soon with the sequence risk calculations. If you would like a copy of this just subscribe to our website and we will let you know when it is available.
Sign Up Today!
(5) Should We Forget Shiller CAPE Ratios And Focus On E/P Instead? by Michael Kitces
(6) Robert Schiller Irrational Exuberance Page 204-205