S&P 500 Forecast: Medium-term returns are likely to be negative

The S&P 500 has produced fantastic results for investors since the Great Financial Crisis. However, largely because of those rapid gains the US large-cap index is more likely to produce below-average or even negative returns over the next few years.

The S&P 500 is expensive relative to historic norms

Today the S&P 500’s price of 2,176 points is about 27.5 times its average inflation-adjusted earnings of the last ten years (a ratio which is usually known as the Cyclically Adjusted PE or CAPE ratio).

However, the average CAPE ratio for the S&P 500 over the past 100 years is just under 17.

If we assume that CAPE will have an average value close to 17 in the future (which is a reasonable assumption) then the S&P 500 today could be overvalued by as much as 63%.

Of course, there is no law which states that the S&P 500’s CAPE ratio must always be precisely equal to its long-term average. In fact, the index and its CAPE ratio can swing about wildly, as any equity investor will know (and the same obviously applies to UK indices such as the FTSE 100).

Generally, this near-random walk is bound within a range, with CAPE staying between half and double its long-term average almost all of the time.

With the S&P 500’s average CAPE sitting just below 17, that range extends from about 8.5 to 34, as I’ve tried to illustrate with the chart below:

SP 500 CAPE ratios 2016 08
Mean reversion in action over many decades

Note that the CAPE “rainbow” moves steadily upwards thanks to inflation plus the real growth of those 500 companies.

A high CAPE ratio suggests low future returns (and vice versa)

When an index and its CAPE ratio are towards the lower end of their normal range the odds grow ever stronger that the index will outperform over the next few years.

This happy outcome occurs because of the wonders of valuation mean reversion, i.e. that the CAPE ratio will almost certainly revert back to its historic average given enough time.

Within the lifetime of most investors today, such a low point was reached most notably during the late 70s and early 80s. Investors were witnessing – according to Business Week magazine – “The death of equities”.

The market had gone nowhere but sideways or down for years and so the mood was as glum as that magazine headline suggests. But the incredibly low CAPE ratio of seven suggested that the future was fantastically bright for those brave enough to ignore the crowd.

And CAPE was right: The slump of the early 80s was one of the great buying opportunities of the last century.

At the other end of the scale, the dot-com bubble sent the S&P 500 and its CAPE ratio soaring into uncharted territory; “There be dragons” seems oddly appropriate.

The extreme optimism of the 90s eventually led some to ask “What if we are just at the beginning of the beginning of a long wave of ultraprosperity?”

That statement is looking ever more true as the years go by (at least for some) but material prosperity does not mean that the stock market or companies in general can beat inflation by 20% per year for the rest of eternity. Even vaguely efficient global capital allocation will make sure of that.

A much less exciting but much more likely outcome for investors in the 90s was that the S&P 500’s CAPE ratio would revert from its record high of 44 back towards its average mid-teen value at some point during the 2000s.

And CAPE was right again: The boom of the late 90s was one of the worst buying opportunities of the last century.

Today the S&P 500’s CAPE ratio is high, but not as insanely high as it was in the dot-com bubble. However, with a CAPE ratio of 27.5 that ratio is higher than it’s been for more than 95% of the last 100 years.

One entirely reasonable way to describe the S&P 500’s current situation is that there is about a 5% chance that its CAPE ratio will go up over the next few years and about a 95% chance that it will decline.

Another way to think about the S&P 500’s future is to do a one-year forecast.

S&P 500 forecast: A decline is likely over the next year

The heading above kind of gives the game away but here are the details. First of all, we need to know a few things:

  • The S&P 500 currently sits at 2,176
  • Its cyclically adjusted earnings are 79.2
  • Its CAPE ratio is 27.5
  • CAPE’s 100-year average is 16.8

Historically the S&P 500’s cyclically adjusted earnings have grown by about 5% per year.

If we assume that a 5% growth rate occurs over the next year then its cyclically adjusted earnings will grow from 79.2 to 83.2.

If we also assume that CAPE will close the gap to its mean value by half over the next year (unlikely to actually happen but it’s still a reasonable assumption) then the S&P 500’s CAPE ratio in one year will be 22.2.

By multiplying those two numbers (83.2 and 22.2) we can calculate an expected value for the S&P 500 in one year’s time (the expected value is the average value you might expect over a large number of alternate futures). And so:

  • The expected value for the S&P 500 in one year’s time is 1,843

This represents a decline of 15% from today’s price and so, somewhat obviously, I don’t think the S&P 500 index is an attractive investment at its current level.

Some final thoughts:

Passive investors: You shouldn’t worry about the S&P 500 because you should be widely diversified across many different countries and asset classes, regardless of where you live.

Active investors: There are almost certainly many attractively valued stocks within the S&P 500. However, you should be aware that an expensive index is likely to be a drag on the returns of most stocks in the index as it mean reverts, and that applies even to those stocks that are already attractively valued.

Data: You can get very long-term CAPE-related data for the S&P 500 index from Robert Shiller’s website.

Author: John Kingham

I cover both the theory and practice of investing in high-quality UK dividend stocks for long-term income and growth.

25 thoughts on “S&P 500 Forecast: Medium-term returns are likely to be negative”

  1. I have thought that the US market has been overvalued for sometime & sold out my US holdings about 3.5 years ago. But it is interesting that CAPE has only once been briefly below 20 since 1995, so mean reversion is taking quite some time to kick in.

    1. Hi Gareth, you’re right; the US market has been expensive for a very long time.

      In fact the 100yr average CAPE is gradually increasing and is now at 17, whereas 30 years ago (which is about as far back as the 100yr average goes, i.e. the full data set is about 130 years) it was 14.5.

      On the point of mean reversion though, even if CAPE did only briefly fall below 20 in the 2009 crisis the point about high valuations leading to low future returns remains valid: From 1999, S&P 500 investors saw terrible returns to 2009 and also pretty rubbish returns even to today.

      So mean reversion has kicked in, just not very frequently.

  2. I really enjoy reading these forecasts you give. you write them in such an explanatory way for newer investor to understand.
    thanks again

  3. Oddly enough this also hit my ‘in-tray’ today :-


    which queries whether we can rely on mean reversion at all?
    This investor like yourself does rely on it, and rely on it very heavily!

    Although dismissed by Smithers in ‘Valuing Wall Street’, we favour Dividend Yield (DY) as a valuation measure. DY (assumed as a real yield) enables quick comparisons with competing Asset Classes, such as Bonds (Y – infl), Rental Real Estate (real Y) and so on.
    But here is the problem, we have to adjust the median from decade to decade, to make the DY method work! Similarly could the CAPE median need adjusting from decade to decade?
    Are underlying forces ar work?
    Don’t know the answer but the above llnk sets one wondering.
    Have noticed far more of the population, esp in US, are now self-investing for their retirements, Could this be causing a structural change?

    Using DY, my simplistic calcs suggest the S&P is 14% over-valued.
    The FTSE All Share 15% over-valued.

    Your calcs have set me wondering if I am way off?
    Will be reflecting further.

    Thanks for the insights.

    1. Hi Magneto, I think the appropriate mean value to use in mean reversion is going to change over time, although of course it will change unpredictably. That’s one of the reasons I use a 100 year mean rather than an all-time mean, so that the mean moves with the market to some (small) degree. So for example where US stocks have been mostly expensive since the 90s the 100-year mean CAPE has moved from below 15 to almost 17.

      Having said that, I doubt that such small changes will make much difference to the predictive abilities of CAPE mean reversion. After all, moving from 15 to 17 is not such a big deal and it’ll take another 20 years to go from 17 to 19, assuming it ever does.

      1. Here is a comment from Larry Swedroe in a current Boglehead discussion :-

        “IMO there is literally no logical reason to believe that the CAPE 10 should revert to its 140 year average when so much has changed that would cause the CAPE 10 to drift upwards including,
        a) US much wealthier, so capital less scarce so equity premium should be smaller
        b) stronger accounting rules, hence more transparency
        c) accounting rule changes that require faster writeoffs of intangible assets
        d) much lower economic volatility due to better smoothing of economy by the FED (and also better inventory management leading to same thing)
        e) lower trading costs
        f) lower mutual fund costs

        Anyone forecasting RTM to that 16-17 level is simply IMO crystal ball gazing. At least levels like 23 would make some sense IMO. But we don’t know the right level, and no one does.


        Seems we are not alone struggling with the data!
        But whatever, S&P valuations are on the high side, the doubt is by how much?

        All Best

      2. I agree with Larry that things have probably changed since 1871 (the start of Shiller’s data), which is why I use a 100 year average instead. But I certainly don’t want to start thinking a CAPE of 23 is normal, because in 10 or 20 years it might not be. We could easily have inflation at 10% plus and who knows what else, so I’m sticking with a 100 year average and am willing to accept the innate uncertainties. Also, this is why Shiller suggests not getting into or out of the market based on CAPE, but at the most simply adjusting asset allocation.

        Personally though I just use it as a gauge of where the market is and whether or not investors are likely to face head winds or tailwinds over the next few years.

  4. A lot depends on whether you are concerned about markets or companies. What is a market in reality?

    Just like the FTSE, or DAX or any other “market”, there are overvalued companies and some that offer significant value.

    I would say these companies are very cheap :-

    Walt Disney

    Compared to the silliness of the dotcom stratospheric valuations, these companies are now ridiculously cheap with P/E’s ranging from 11 to 20 with growth rates that do those numbers more than justice.

    Markets are just that, markets. Investors need to understand companies.

    Well that’s my two penneth worth anyhow!!


    1. Microsoft… kicking myself for not buying in a few years ago. As a software engineer primarily on their stack, their cloud infrastructure, Azure (which I use and only gets better), will just generate cash for fun on the kind of scale that’s very difficult to get into for a new player.

      As a software man yourself John, be interested to hear your thoughts on that one 🙂

      1. Andrew,

        I was lucky enough to have bought MSFT in the teens and don’t really see me ever selling. There is an interesting article on the fundsmith web site about Microsoft. The title of the article under the “news” section of the web site is “Just the facts when weighing investments”
        It’s not too late to buy Microsoft. It’s not a stock to trade though, but to hold.

        Qualcomm gave us all a tremendous buying opportunity when it ran into difficulties on the last Samsung contract. They have since solved that problem and are back in with Samsung, and many others. The CDMA patents are a license to print money and Qualcomm is growing nicely — it’s unmatched in it’s field.

        Walt Disney – dominates films, theme parks and in addition it makes a nice living from media and merchandise. The stock is well down because of the cancellations of subscriptions for ESPN, but I doubt this will appear as anything other than a blip. At $84 you are buying a $135 stock.

        Skyworks has a return on investment most would only dream of and with $billions in the bank and no debt, what’s not to like? They have competition in analogue components but are still the go to company for all the smartphone manufacturers and has a growing Internet of things presence. It’s P/E is 11 — It’s a gift.

        Apple — It’s just seriously undervalued – no analysis needed.

        PayPal — growing at 25% a year — there are loads of nominated challengers but most are falling by the wayside. PayPal has massive first mover advantage — It’s a buy and a looong terrrm keep.

      2. Hi Andrew

        Unfortunately I don’t have much to say about Microsoft. Obviously it’s the sort of company that I’m interested in but I haven’t crunched the numbers and don’t like to comment until I have. With a 2% yield it certainly looks a bit pricey, although of course it has a good record of growth. I think they might have gone a bit OTT with dividend growth in recent years and would be surprised if they can keep up that double digit growth rate.

        The move to cloud-related services will probably smooth things out a bit because of the small, recurring subscription revenues it will generate rather than occasional one-off purchase of software as per the olden days. I’m actually looking at Office 365 now as my laptop is beginning to creak and want to move away from being hardware dependent. So perhaps I should buy Microsoft given that I’ll probably soon be paying them on a monthly basis anyway…

    2. Hi LR

      “Markets are just that, markets. Investors need to understand companies.”

      Yes I totally agree, but it’s nice to know if you’re investing in the dot-com bubble or the great depression as that can have a massive impact on all stocks (negative and positive respectively) regardless of their valuations.

      Also I just find it interesting so you’ll have to excuse my nerdy need to crunch numbers and draw pretty charts.

      1. John, I agree – what you are doing is very interesting. It is indeed great to know if you are investing in an overall bubble environment.
        I guess I was trying to highlight the point regarding relative valuations within the same market. The companies listed all have relatively sane P/E multiples compared to their prospective growth rates.

        Thinking more about Microsoft, it’s P/E in 2000, during the dot-com bubble, Microsoft’s highest valuation multiple of price-to-earnings (P/E) ratio was 57.

        Today it stands at just over 17 or less than a 3rd of back then, and Microsoft has grown it’s earnings consistently over the last 15 years.

        I wish all my positions had done that — but then I’d be bored and far too wealthy 🙂

      2. Interestingly, if my research is correct, if you had bought at the peak of the the dotcom bubble you would still be looking at a loss in actual price terms and even worse in terms of inflation adjusted

      3. Which highlights the importance of price! You could buy the best company in the world growing at 20%/yr, but what’s the point if valuation mean reversion could negate all the gains?

        One argument is that the long-run return approximately equals the company’s return on capital, but in the long-run we’re all dead! So valuation mean reversion is important for those of us who are mortal.

  5. “Which highlights the importance of price!” JK

    Can sympathise only up to a point with those who advocate a steady say 70/30 (Stocks/FI), regardless whether PE1 (or CAPE) is 8 or 32.

    In all other walks of life from groceries to housing, price matters.
    Why not for Stocks?

    One of the joys of Stock Investing is Ben Graham’s ‘Mr Market’ coming to us every day, offering such widely fluctuating prices untii we say “OK”.

  6. The ‘new’ context is that stocks are the new bonds and that large companies offer more security than many large countries. So, one could argue that reversion cannot take place until the context reverts to the old norm – if it ever will.
    If not then the valuations of say RB. will stay stratospheric and compared with, say Greece, so they should.
    The pension authorities are trying to maintain the old norm where government debt is assumed to be safe and there are signs that they are being challenged by companies with large pension deficits.
    Maybe, just maybe, the norm has permanently slotted into a new channel.
    That’s the trouble with statistics 🙂

    1. Hi apad, I totally agree that old means are not laws of physics and can change over time. That was the main reason why I switched from using the all-time CAPE mean to a 100 year mean. I thought that was a reasonable balance between being (very) long-term in my outlook by not eternally bound by historic data that has long since become irrelevant.

  7. Hi John,

    I’ve been reading your book, as well as these blog posts, and I was wondering if there’s a free website where you get the data for these index comparisons from – i.e. the earnings and dividends of the index in index points?

    Your book says the FT Markets data site is the place to go, but I don’t see it there (maybe I need a subscription though?):

    1. Hi Tim, Robert Shiller has excellent long-term data for the S&P 500 at: http://www.econ.yale.edu/~shiller/data.htm

      I’ve added that link to the article and should have done that in the first place, so thanks for highlighting its absence.

      The FT Markets Data Archive site ( http://markets.ft.com/data/dataarchive ) does have FTSE 100/250 data, but only current data. They also keep changing the website and therefore the URLs (website addresses) which is a pain.

      The current URL for the World Markets at a Glance report is:


      Just change the date at the end (160902 is 2016/09/02) to whatever data you like. However, this only goes back so far (not sure exactly how far).

      The older FTSE Actuaries Share Indices reports can be found at:


      Again, 311013 is the data and can be changed (311013 is 2013/10/31)

      You can also get those reports by using their Browse Reports function, filtering by ‘equities’ and ‘FTSE Actuaries Share Indices’ and then whatever dates you like.

      In terms of older FTSE 100 data I have an old spreadsheet which I don’t update, but it does have data going back a few more years:

      FTSE 100 CAPE data.xls

      Hopefully that should be enough to get you going!

      1. Thanks John, I’ve saved those links and I think for the moment I’ll just stick to comparing the dividend yield to make sure shares I buy are above the relevant index’s average. Collecting historic data to calculate index PE10, PD10 and so on seems like more effort than it’s worth for me.

        The hardest part of all this is actually finding data, compared to that the actual company research is straightforward!

      2. I would say the hard bit is getting the data for free. There are data sources such as ShareScope/SharePad, Morningstar and Sharelockholmes, all of which have 10 years of data, but you have to pay for it.

        So as with most things there is a trade off between the cost of paying for data and the time cost of collecting it by hand.

        Or of course you could use a simpler system like your high yield method, and hope that the net result is just as good.

      3. Actually I pay for sharelockholmes, very useful for the UK. My difficulty is more calculating the whole-index ratios as I don’t believe sharelockholmes has it. It’s even more difficult for other countries (ASX and DAX/MDAX are what I’m interested in besides FTSE).

        Instead of trying to keep above the averages for the index in things like div yield or PE10 I’ve just set myself minimums for those values a little bit stricter (e.g. growwth quality >75%)

Comments are closed.

%d bloggers like this: