Monday, December 13, 2021

What to Do About Crazy Stock Valuations

The last time I had to put a lot of effort into thinking about my finances was back when I retired in mid-2017.  I had ideas of how to manage my money after retirement, but it wasn’t until a couple of years had gone by that I felt confident that my long-term plans would work for me.  I had my portfolio on autopilot, and my investing spreadsheet would email me if I needed to take some action.

I was fortunate that I happened to retire into a huge bull market.  I got the upside of sequence-of-returns risk.  The downside risk is that stocks will plummet during your early retirement years, and your regular spending will dig deep into your portfolio.  Happily for me, I got the opposite result.  My family’s spending barely made a dent in the relentless rise of the stock market.

However, stock prices have become crazy, particularly in the U.S.  One measure of stock priciness is Robert Shiller’s Cyclically Adjusted Price-Earnings (CAPE) ratio.  In the U.S., the CAPE ratio is now just under 40.  The only other time it was this high in the last 150 years was during the dot-com boom in the late 1990s and early 2000s.  Just before the 1929 Black Tuesday stock market crash, the CAPE was only about 30.

Outside the U.S., prices aren’t as high, but they are still elevated.  My stock portfolio's blended CAPE is a little under 32 as I write this article.  Even if stock prices were cut in half, this would just bring the CAPE close to the average level over the past century.  To say that these thoughts made me think hard about whether I should change how I manage my portfolio is an understatement.

A change in thinking about high stock prices

For a long time, my thinking was to ignore inflated stock prices and just rebalance my portfolio as necessary to maintain my chosen asset allocation percentages.  I have a planned “glidepath” for my stock/bond mix that has me about 20% in bonds at my current age and increasing as I get older.  My bond allocation consists of cash and short-term bonds, and the rest is spread among the world’s stock indexes.  I saw no reason to change my plan as my portfolio grew.

Then a question changed my thinking.  If the CAPE rises to 50, or 75, or even 100, would I still want such a high stock allocation?  It’s not that I expect the U.S. or much of the rest of the world’s stocks to become as overvalued as Japanese stocks in 1990, but I should be prepared for how I’d respond if they do.

At a CAPE of 50, I wouldn’t want more than about half my money in stocks, and at 100, I wouldn’t want much in stocks at all.  So, even though I’m comfortable with 80% stocks at a blended CAPE of 32, something would have to change if the CAPE were to rise from 32 towards 50.

A first attempt

Once I realized I definitely would reduce my stock allocation in the face of ridiculously inflated markets, I had to work out the details.  I started with some rules.  First, I don’t want any sudden selloffs.  For example, I don’t want to hold a large stock allocation all the way up to a blended CAPE of 39.9 and suddenly sell them all if the CAPE hits 40.  A second rule was that I don’t want any CAPE-based adjustment to apply unless the CAPE is above some threshold level.

As the CAPE kept climbing, I felt some urgency to choose a plan.  My first attempt was to change nothing if the CAPE is under 30, and when it’s above 30, I multiplied my bond allocation by the CAPE value and divided by 30.  I implemented this idea in my portfolio as an interim plan before I analyzed it fully.

Another adjustment I made a little earlier was to reduce my expectation for future stock returns.  When the current CAPE is above 20, I now assume the CAPE will drop to 20 by the end of my life.  This doesn’t directly affect my portfolio’s asset allocation, but it reduces the percentage of my portfolio I can spend each year during retirement.  When stocks rise and the CAPE rises, my portfolio grows, and this increases how much I can spend.  But then this new rule reduces my assumed future stock returns, and reduces my safe spending percentage somewhat.  Increasing stock prices still allow me to spend more, but this rule slows down the increase in my spending.

A new simpler rule for adjusting my stock allocation based on high CAPE values

I’m still happy with the way I’ve adjusted my expectation for future stock returns when the CAPE is high, but I’ve changed the way I adjust my bond allocation to the CAPE.  I now have a simpler rule I named Variable Asset Allocation (VAA) that better matches my thinking about what I’d want if the CAPE got to 50 or 100.

VAA: If the CAPE is above 25, I add CAPE minus 25 (taken as a percentage) to my age-based bond allocation.  

For example, without VAA my current bond allocation based on my age is about 20%.  The current blended CAPE of my portfolio is about 32, so I add 32–25=7% to my bond allocation.  So, I’m currently 27% in bonds and 73% in stocks.

This might not seem like much of a bond allocation adjustment in percentage terms, but it’s a bigger adjustment in dollar terms.  Consider the following example.  Suppose a $500,000 portfolio with a 20% bond allocation sees a jump in the CAPE from 25 to 32.  This is a 28% increase in stock prices.  So, we started with $100,000 in bonds and $400,000 in stocks, and the stocks jumped in value to $512,000 for a total portfolio size of $612,000.  When we adjust the bond allocation to 27% in accordance with VAA, we have $165,000 in bonds and $447,000 in stocks.  Of the $112,000 jump in stock value, we shifted $65,000 over to bonds, and left only $47,000 of it in stocks.  Although the bond allocation went from 20% to 27%, a 35% increase, the dollar amount in bonds rose 65%.  This is a substantial shift, and it leaves a healthy bond buffer if stock prices subsequently crash.

Some analysis

One concern I had with adjusting my asset allocation based on the priciness of stocks is whether it produces reasonable stock and bond allocations across a range of CAPE values.  By design, VAA matches my intuition about bond allocations at different CAPE levels.  If the CAPE gets to 50 sometime soon, my bond allocation would go to 20+(50–25)=45%, which seems reasonable.  At a CAPE of 75, my bond allocation would be 70%, which also seems reasonable.  It’s possible that something about the world might change that makes high CAPE values seem sensible and that I’d want to own more stocks, but for now I’m happy to shift automatically away from stocks as the CAPE rises through crazy levels.

The following chart shows how a hypothetical portfolio using VAA responds to the CAPE rising from 25 all the way to 105.  We begin with $100,000 in bonds and $400,000 in stocks with the CAPE at 25.  The curves look smooth, but there are 27 rebalancing operations as the CAPE rises from 25 to 105.  Initially, as stock prices and the CAPE rise, we shift most of the stock gains to bonds.  As the CAPE gets into the low 40s, all stock gains are shifted to bonds, and as the CAPE exceeds 45, VAA shifts all the stock gains and more into bonds.

It’s not until the CAPE reaches about 60 that the dollar amount in stocks dips below the initial $400,000.  However, the dollar amount in bonds doubles by the time the CAPE reaches 35, and doubles again with the CAPE in the low 50s.  The idea of VAA is to take the huge stock gains that come with a rising CAPE and preserve them in safe bonds.  Why keep playing the risk game when you’ve already won?

Observe that if we had invested the whole $500,000 in stocks and the CAPE had risen from 25 to 105, we’d have $2.1 million instead of only $1.04 million with VAA.  So why bother with VAA?  The answer is that the CAPE almost certainly isn’t going to 105.  The higher it gets, the more likely stocks are to crash.

If stocks are going to crash, why not shift everything into bonds instead of messing about with VAA?  Stocks are certain to fluctuate, but we don’t know if or when they’ll have a big crash.  I have no interest in making a high-conviction bet about the stock market.  The idea of VAA is to capture some upside if stocks keep rising, and limit the damage if stocks crash.

The following chart shows the amount of stock gains we’d preserve if stocks start at a CAPE of 25 and later crash back to a CAPE of 25 after a period of rising.  We give three scenarios: VAA, maintaining an 80/20 stock/bond allocation, and 100% stocks.  As we see from the chart, if the CAPE gets to 45 before crashing back to 25, an all-stock portfolio preserves none of the stock gains, an 80/20 portfolio preserves $17,000, and VAA preserves $265,000.  VAA is the clear winner if stock prices decline enough to bring the CAPE back to historical levels at some point.

So far we’ve been talking about CAPE movements resulting from changes in stock prices.  Another way for the CAPE to move is from changes in corporate earnings, the denominator in a P/E ratio.  The chart above shows VAA’s margin of victory over other strategies when corporate earnings remain constant.  If the decline in the CAPE that brings it back to 25 is partially due to rising corporate earnings, then VAA’s margin of victory would be smaller.  However, VAA shines in any scenario with a significant drop in stock prices.

Rebalancing frequency

Another potential concern is whether I’d ever be rebalancing too often.  For example, could a very small change in stock prices trip a rebalancing trigger?  The short answer is no.  To protect me from trading too often, I have set my rebalancing thresholds such that the profits arising from rebalancing once in each direction are 20 times the trading costs in commissions and spreads.  Determining these thresholds requires some calculations that I have implemented in a spreadsheet.  For details, see the newly added section 8 of my paper Portfolio Rebalancing Strategy.

The main way I could end up trading too often is if there are anomalies with computing the CAPE that lead to its calculated value jumping up and down by enough to trigger spurious rebalancing.  I plan to protect against this by waiting until I see it happen and use my judgment in not rebalancing back-and-forth too often.

If my spreadsheet emails me with instructions to rebalance from bonds to stocks one week, and from stocks to bonds the next week, I can examine whether stock prices have really moved enough to justify rebalancing.  If not, I might suspect that the trigger for rebalancing is jitter in the calculated CAPE.  So far I’ve seen no indication of this problem.


I’m hopeful that I’ve chosen ways to respond to extreme CAPE levels that are measured, reasonable, and won’t need to be changed in the future.  Most importantly, I’ve implemented these plans in an emotionless spreadsheet that does all the work for me while I get distracted by more interesting pursuits than portfolio watching.


  1. If my understanding is correct, the root of the problem is whether the rising CAPE would be an indicator of over-valuation and as a result poor returns in the future. I was searching around last night and I came across this paper. I am not sure, but it may be something interesting to look at if you have time.
    I have not fully understood the paper yet. But the author used the historical data with a contrarily different estimation of the current valuation on Fig 4.

    BTW, Merry Christmas!

    1. Hi Y,

      That's an interesting paper. I wonder about the method of accumulating the "heat" values. It is based on a long-term average computed dividend yield (the constant c). If we change this c value by a small amount, it appears to accumulate differences in the heat value that build over time. If I'm right about this, it's hard to trust their heat measure because it's too sensitive to small changes in c.

  2. Michael, I am interested in digging deeper into what you have detailed here. Where are you sourcing the CAPE data and how are you calculating it with respect to your investments? I am invested mostly in XGRO and want to calculate CAPE for this ETF.

    1. That was my question as well. I know there is the data on, but that just covers the S&P 500.

    2. Hi Unknown and Minyfresh,

      I get U.S. data from, and import it into Google Sheets with the command IMPORTHTML("","table",1).

      For other countries, I've used the following page:

      However, these values are just a snapshot in time. I calculate the CAPE for my non-U.S. ETFs on the date of this data, and then scale this CAPE value by (current ETF price)/(ETF price on the date of the CAPE data). Periodically, I get fresh data when this web page is updated. Another challenge is that this CAPE data doesn't cover all countries, so I end up just computing the CAPE for the part of each ETF I have data for. None of this is ideal, but I believe it works well enough for my purposes.

    3. I had the same question. I tried pulling the data I needed to calculate this from GOOGLEFINANCE but couldn't get what I needed for either my CAD or US traded ETFs.
      I will have to dig more to see what I can find for Emerging (IEMG) and International (XEF).

      Thank you!

  3. An interesting post and a nice model. I am hesitant to try to use CAPE to predict future returns, as so far it has proven bad at doing this (it is good at back-fitting historical data, but has not done well at forward-prediction). Shiller himself has moved from CAPE to Excess CAPE Yield, suggesting that CAPE by itself is incomplete. Once we are making decisions based on a metric, we get dragged into all the arguments about what that metric should be, why a current value is or is not justified, and so on.

    I think there is easier logic to get to the same conclusion. If you are no longer accumulating at CAPE=32, you presumably feel you have enough to meet your future needs. So once CAPE=50, you must have an even more secure position, and assuming you have not inflated your lifestyle in lockstep, you have less need to take risk. This would justify taking equities off the table independent of any prediction about future market performance.

    The decision may be more difficult for someone still accumulating.

    1. Hi Random Sapien,

      I agree that the CAPE is a weak predictor of future returns. It doesn't do too badly for long-term returns, but gives limited information for even the next decade. I've made this point myself a couple of times:

      I'm suspicious of any method that purports to measure future returns accurately. They are likely the result of data mining. The best we can hope for is some visibility into the mean of the return distribution over the next few decades. I use this to decide on my safe spending level.

      WHen I'm choosing my asset allocation, I'm not using the CAPE to predict future returns in the usual sense. I'm using it for two purposes: 1) as a measure of the likelihood of a stock market crash, and 2) as a measure of my reduced need to take stock market risk. This appears to be roughly similar to your remarks in your second paragraph. The difference is that I've created a model that makes small adjustments in the face of small CAPE changes rather than having rules of the style if CAPE > x, ...

    2. Hi Random Sapien,

      You're right that things are different for someone still working and accumulating retirement assets. I just stayed 100% in stocks, and that can work if you are prepared to delay retirement if stocks disappoint. But those with a fixed date or who are at risk of forced retirement would have to be more conservative in later years.

    3. A helpful reader BK emailed the following useful CAPE links:

      About CAPE data, you can get it from the Barclays website and to a lesser extent from the Research Affiliates website:!/?currency=USD&expanded=tertiary&group=core&model=ER&models=ER&scale=LINEAR&terms=REAL&tertiary=shiller-pe-cape-ratio-line&type=Equities

      I found the first link helpful, but the second seemed to need some sort of login. The first gives fresh data each month in a spreadsheet. I haven't sorted out yet how to import it automatically into a Google sheet.

  4. Hi Michael

    I am going to play "devil's advocate" here. This whole convoluted reasoning and rational justification seem to me to be nothing more than well disguised market timing. ;-)

    It violates one of Bogle's main principles which is to "Stay the course"


    1. Hi Garth,

      There's nothing disguised about it. When CAPE>25, it's a gradual form of market timing designed to preserve capital as my portfolio grows ever larger than I need to fund my retirement and my only real risk is a big stock market crash.

      I will take issue with describing it as "convoluted". I think it's quite simple and clear: "If the CAPE is above 25, I add CAPE minus 25 (taken as a percentage) to my age-based bond allocation."

      I've now chosen a course I wish to stay with. While CAPE was less than 25, the course I now follow exactly matches the course I had been following. I see no reason why "staying the course" necessarily means a fixed asset allocation. After all, we already adjust asset allocation with age. Why not also adjust it for the need to take risk when risk is highest? What I want to avoid is sudden ill thought out portfolio adjustments driven by fear or greed.

  5. Hi Michael,
    Sounds to me that your approach is much more of a type of insurance policy on the portfolio. When a major downturn happens you will have the protection needed. But, up until then the insurance is a drag to some extent on the overall returns of the portfolio. Will the protection that you get on the downturn provide an overall better return over the life of the portfolio...who knows? Sure is interesting to think about.

    1. Hi Ed,

      That's a good way to think about it. One further point is that the huge run up in stocks in recent years has left me not needing much in the way of future returns. Warren Buffett likes to say "Never risk what you have and need for what we don’t have and don’t need."

  6. Excellent blog. Can you help me with the calculations? Say blended CAPE=20, real stock return is 4%, real cash return=0%, remaining life=45 years, 5 years of spending in cash. Then portfolio withdrawal rate=1/(pv(4%, 40,-1)+5)=4% and percentage in cash is 5/(pv(4%,40,-1)+5)=20%.

    Next suppose the blended CAPE level increases to 30. I think real stock return is now assumed to be 4%-((30/20)^(1/45)-1)=3.1%, and the percentage in cash increases to 20%+(30%-25%)=25%. How would I calculate the new portfolio withdrawal rate?


  7. Hi Michael, thanks for your blog - it’s very helpful. Can you please check my formulas for your method? I want to do my own analysis but it want to be sure I understand your method.

    Assume a 55 year old retiree plans to live to 100, allocates 5 years annual retirement spending in a HISA account initially, the blended CAPE is 30, assumes 4% real return on stocks at blended CAPE <20, assumes 0% real return on HISA.

    Then adjusted stock return =4%-(1-(20/30)^(1/(100-55))=3.10%; initial portfolio withdrawal rate =1/(Pv(3.10%,100-55-5,-1)+5)=3.61%; initial HISA to stock ratio =5/(Pv(A3,100-55-5,-1)+5)=19.22%; new HISA to stock ratio =19.22%+(30%-25%)=24.22%; new portfolio withdrawal rate= =Pmt(3.10%*(1-24.22%),100-45,-1)=3.26%


    1. Correction: HISA to stock ratio should read HISA to portfolio ratio, where the portfolio is the sum the HISA and stock accounts. Cheers Jim

    2. Hi Jim,

      Let me say first that the method I use has become more complex than necessary, but since I only had to do it once in a spreadsheet, I don't mind the complexity.

      Your calculation of the adjusted stock return matches mine. However, I use a different method of computing the withdrawal rate and HISA level. The challenge is that because I have a declining-stock-percentage glidepath, my assumed portfolio return declines over time. All the calculations are in the paper linked to by this post:

      However, that paper doesn't do the return adjustment and the HISA amount adjustment. The following won't make any sense unless you're familiar with the paper. I start by using the unadjusted r to caluclate the unadjusted zs (HISA level), and add the (30% - 25%) to this to gets the adjusted zs. Then I use the adjusted r, along with b to find the value of y (iteratively) that gives a calculated zs that matches the adjusted zs. I then use the adjusted r, newly calculated y, along with b in the calculations in the paper.

      Let me repeat that this is far more elaborate than necessary to run a successful indexed portfolio.

    3. Do you get the same results as me for my example using your spreadsheet ?I did read your paper a while back and I actually implemented it in a spreadsheet, but it is too complex for my liking.

      I am looking for reasonably simple calculations that ideally do not require an iterative approach. No criticism here - your work on this blog is top notch thinking.

      Thanks again,

    4. Hi Jim,

      My result won't be the same as yours. Because I'm taking into account the glidepath and the declining expected return, my calculated withdrawal rate will be lower. My guess is that it may not be much lower, but I'm not certain.

    5. Using your method and terminology, with unadjusted r=4%, I’m getting unadjusted zs=20.04%, adjusted zs=25.04%, new y =7.412 (using an iterative approach), calculated zs=25.04% (compared to my new HISA to portfolio ratio of 24.22%), and new s=3.38% (compared to my new portfolio withdrawal rate of 3.26%). Pretty close.

    6. I agree that it appears to be close enough.

    7. I looked at my new portfolio withdrawal rate calculation again and noticed I made a mistake. I should have stated =Pmt(3.10%*(1-24.22%),100-55,-1)=3.62% above. I now think my method is flawed and not useful for my purposes since it does not account for an increasing bond to portfolio ratio during the retirement years (i.e. glidepath). I'm now taking another look at your paper - it's excellent work btw. Cheers, Jim

    8. Hopefully you'll find the paper useful. Good luck with it.

  8. Hi Michael, you’ve clearly thought long and hard about the rebalancing issue, way beyond what I’ve come across anywhere else. With these formulas built into your spreadsheet, what has been the smallest amount where you acted on a rebalancing alert? I’m thinking here of the jitter mentioned in section 8 of your paper on rebalancing, and in dollar terms where you deemed the rebalancing exercise to be worth the effort and fees/costs.

    1. Hi Bob,

      The smallest amount (in the low 5-figure range) has been for rebalancing among VTI, VBR, and VXUS, but this is unrelated to the jitter issue I raised. When it comes to rebalancing between stocks and bonds while the blended CAPE exceeds 25, the smallest amount has been in the low 6-figure range. I haven't found this jitter possibility to be a problem.