Following the crash of 2008/2009 the S&P 500 went on a record-breaking run, gaining around 400% over eleven years before peaking in early 2020.
It then suffered a record-breaking crash thanks to COVID-19, but that was short-lived and the US large-cap index is now setting all-time highs on a regular basis.
At the time of writing the S&P 500 is just below 4,200, which is fractionally below its recently set all-time high. So the price is high on a historical basis, but so is its valuation according to the CAPE ratio.
Valuing the S&P 500 using its CAPE ratio
If you’re not familiar with the cyclically adjusted PE ratio (CAPE) then here’s a quick summary.
When we value an index we’re effectively trying to work out if the price is high or low compared to the index’s fair value, i.e. the price a rational investor would pay to get their target rate of return.
So we’re trying to compare price to value.
Since nobody knows the true value of an index, we have to make an estimate based on conservative but realistic assumptions.
With the standard PE ratio we’re comparing price to earnings, so we’re using earnings as a proxy for value. If earnings go up then we assume value goes up too, and that’s a reasonable assumption.
But PE only looks at last year’s earnings and earnings in a single year can be very volatile. Intrinsic or fair value, on the other hand, usually isn’t volatile.
For example, if Tesco earns zero profit in a year it wouldn’t be reasonable to say that Tesco’s fair value was zero.
Obviously the S&P 500’s earnings are more stable than most companies’ and they probably won’t go to zero anytime soon, but they’re still somewhat volatile as the chart below shows.
By cyclically adjusting earnings (using their inflation adjusted ten-year average) we get something that’s more stable and more closely related to the fair or intrinsic value of these 500 companies:
So cyclically adjusting earnings gives us a nice stable number which should be quite closely correlated with value, and all we need to do is compare price to those earnings to see if it’s high, low or somewhere in between.
Using CAPE’s long-term average to calculate fair value for the S&P 500
As with the standard PE ratio, we can look at the current CAPE ratio and compare it to its own long-term average to see if it’s historically high or low or somewhere in between.
Since 1985 the S&P 500’s CAPE has averaged about 24 and over the last 100 years its averaged 18. I prefer to use very long-term averages because they’re more stable and that reduces the impact of things like the dot-com bubble or the extended period of low valuations during the 1970s, so I’m going to use 18 as my benchmark.
But there can be issues with using very long-term averages, with the risk being that they’re outdated. For example, the chart below shows that since the mid-1990s investors have been consistently willing to pay well above average for the S&P 500.
So perhaps it’s worth comparing CAPE today against its 30-year average of 24 as well as its 100-year average of 18.
The S&P 500 CAPE ratio is higher than at any time in history other than the peak of the dot-com bubble
The rainbow chart above shows where the S&P 500 would have been at various CAPE levels.
The top of the rainbow (in red) shows where it would have been at very high valuations, and you can see that in the dot-com bubble the S&P 500 went through the top of the rainbow and into la-la land. The bottom of the rainbow (in green) is where the S&P 500 would be at depressed valuations, and perhaps not surprisingly that’s where it ended up during the 2009 crash.
Today the S&P 500’s CAPE is just peeping out above the top of the red zone. From a historical perspective that’s higher than at any time other than the peak of the dot-com bubble.
To be precise, with the S&P 500 just below 4,200 its CAPE ratio is 37. By any reasonable definition this makes the S&P 500 look very expensive and several highly regarded investors have said the S&P 500 is probably in an epic bubble.
Perhaps this time is different, but if history is anything to go by then being this far above average valuations is not great news if you’re going to be invested in the S&P 500 over the next five, ten or twenty years.
But simply saying high valuations are not “great news” for future S&P 500 returns is a bit vague, so here are a few scenarios which will hopefully flesh out those valuation bones.
How the S&P 500’s high CAPE ratio could affect future returns
To help us gaze into the future (or at least the next ten years) we’ll need to make some assumptions about inflation, earnings and valuation multiples. My aim with these assumptions is to be both conservative and realistic, although since I cannot see into the future they will of course not be 100% accurate.
Here are some reasonable assumptions:
- Inflation over the next ten years stays in line with the Federal Reserve’s 2% target
- S&P 500 earnings grow at their 30-year average rate of 4% above inflation
Given those assumptions we can project current earnings into the future and then build scenarios around them to reflect varying levels of investor optimism or pessimism. Here are three plausible scenarios for 2031:
BUBBLE 2031: The S&P 500 is in (yet another) bubble, so CAPE is 36 which is twice the long-term average and well above its 30-year average. The S&P 500 is at 7,200, some 70% above where it is today.
BENIGN 2031: The S&P 500’s CAPE is at an historically average 18, so investors are neither unusually optimistic (as they are today) or unusually pessimistic (as they were during the 2009 crash or for much of the 1970s). The S&P 500 is at 3,600, some 15% or so below where it is today.
If instead we assume CAPE is at its 30-year average of 24 then the S&P 500 would be at 4,800, a mere 15% or so above where it is today.
So either way, if the S&P 500 is at anything like normal valuation levels in 2031 then the index is likely to be either slightly up or slightly down from where it is today. So a repeat of the last decade’s 500% gains seems extremely unlikely.
DEPRESSION 2031: Investors are depressed, perhaps because a global pandemic threatens the prosperity of much of the world’s population. The S&P 500 CAPE has fallen to 9 (half the long-term average) so the S&P 500 has fallen to 1,800. That’s a massive 60% below where it is today.
Of course a huge and protracted market decline seems impossible today, but that’s exactly what investors would have said at the peak of the market in 1929, 1968 or 1999. And they were wrong, because in each case the S&P 500 fell by significantly more than 50% in inflation-adjusted terms, typically taking more than a decade to sustainably exceed its previous highs.
The next ten years of growth may already be baked into the S&P 500’s price
In summary then, it seems more likely than not that the S&P 500 will stay broadly flat or even decline over the next ten years if valuation levels and investor optimism return to anything like their historic norms.
And while there is a chance that the S&P 500 can provide sustainable capital gains over that timeframe, it would require some combination of (a) enormous earnings growth, (b) a state of permanent optimism and (c) a willingness by investors to accept permanently lower returns.
I’ll leave you to decide if those conditions are likely or not.
A few words on the impact of buybacks on CAPE
UPDATE (29/05/21): A reader mentioned buybacks and how their increasing use in the US may be distorting CAPE relative to its historic mean. This is correct. Higher buybacks will result in higher per share earnings growth and that will reduce CAPE’s ten-year average earnings (as faster growth means earnings ten years ago will be smaller).
Technically then, I should use a total return version of CAPE (TR-CAPE) which reinvests all dividends so that a change in the ratio of buybacks to dividends will have little or no impact on it.
In fact, over the last couple of years Professor Robert Shiller has kindly done this for us by publishing TR-CAPE data going back to the 19th century.
However, switching from CAPE to TR-CAPE doesn’t really make much difference. For example, take a look at these two charts showing CAPE or TR-CAPE and their 100-year averages:
Which is which? If you’re eagle-eyed then you’ll have spotted that the second chart has both plots at slightly higher values, with the average running somewhere close to 20, compared to an average of 17 or 18 (by eye) for the first chart.
This tells us that the second chart shows TR-CAPE and its average while the first chart shows CAPE and its average.
Look at the relationship between the high and low points on each chart to their respective averages. Is there a meaningful difference?
I don’t think there is.
I think both charts tell us basically the same thing, which is that the US market was very expensive three or four times in the last 100 years and very cheap three or four times. And 2021 is one of the very expensive times.
So yes, a total return CAPE is better than the vanilla CAPE, but I don’t think it’s better enough to justify the additional complexity in terms of both calculating it and explaining it to people.
Very good Sir. Thank you. In an increasingly correlated world, do you think that the UK market can achieve what seems likely to be a reasonable return (based on its CAPE) over the next 10 years despite quite a high chance of US going nowhere or backwards? Regards, Tom
John Kingham says
Hi Tom, Yes I think the FTSE 100 is likely to produce a reasonable return over the next decade and is unlikely to repeat its weak performance of the last two decades. Obviously the future is uncertain, but given the UK’s reasonable starting valuations I think the odds are it will outperform the US index. But we shall see.
I also wrote a blog about this a couple of weeks ago:
Joe Black says
Totally agree – holding the FTSE 100 and 250 as part of a diversified portfolio looks like a relatively safe strategy. My only worry is that if the US does crash (because, let’s face it, markets go up and down, they don’t really ‘flatline’), it could drag the rest of the world with it, including the FTSEs.
Seeing as ‘the longer it goes on the more likely it becomes’, is it time to start lightening the FTSE load too???
John Kingham says
Personally I think the FTSE 100 is still reasonably valued, as per that recent article, so I’m certainly not looking to avoid the FTSE 100 (about a third of my portfolio is in FTSE 100 companies).
Fraser Dawson says
Would you consider going short the sp500? If so, how would you do it as short etfs don’t really work? Thanks
John Kingham says
Hi Fraser, going short could be a good idea but it isn’t something I’m personally interested in doing.
Joe Black says
Problem with shorting is that your losses if you get it wrong can be effectively unlimited. Also, given that volatility rises in a crash, it’s hard not to get whipsawed out if you try to limit your potential losses. There’s always options, but then time is your enemy. IG spreadbetting offer options on the major indexes, if you’re interested, and the prices seem relatively in line with real options houses, such as Saxo bank
Actually we have a more complicated reason for this bubble macroeconomics. After COVID-19 hit the world the US government thought it was necessary for a monetary expansion policy and lower the interest rates to combat the problem of a recession in USA.
As a result it has caused the price of assets such as bonds and equities to shoot up. Something similar happened in Japan during the 80s for different reasons. However unlike Japan, USA is in a slightly different situation.
The dollar is one of the most common traded currency in the world and it will remain in demand for the medium term. Therefore even though S&P 500 index price is grossly inflated these prices may remain at such levels for a while because unlike yen the dollar will continue to remain in demand. Therefore inflation will be moderate.
Until the demand for dollar recedes which creates an inflationary condition the federal reserve will not increase the interest rate which is the only thing that can reset the balance. However it’s difficult to determine how long the bubble will last because the federal reserve may be compelled to maintain this low interest rate for a very long time due to politics and the fact that America can get away with such extreme economic policies.
John Kingham says
Hi Reg, thanks for the additional context. Personally though I quite like CAPE precisely because it ignores all of these details and just looks at very long-term norms, such as the fact that interest rates, discount rates and valuation multiples tend to mean-revert given enough time. It may take one year or ten years, but in the end mean-reversion almost always wins.
Joe Black says
There’s also the point that the Americans appear to be in the process of voluntarily abdicating as world reserve currency. Printing like crazy hardly inspires confidence, and you can see the consequences all over, including in the GDP – USD exchange rate, which has gone from sub 1.2 during the pandemic to about 1.4 right now. What’s that? A 16% devaluation against the pound? So you can effectively regard the S&P as being at 3500, if you are using sterling to buy it now.
John Kingham says
It’s certainly interesting given the rapid increase in money supply in recent years and continued super-low interest rates. Higher inflation seems likely but I’m no expert.
Seeking Fire says
First of all I haven’t got a clue. That’s the most important thing to know. Hence why despite agreeing largely with what you say I also have sympathy for the previous comment therefore I am invested in a global equity tracker with a slight UK overweight for no particular reason beyond I think it is cheap. Big q is if there is a substantial fall in equities then where will the money go? Presumably it must likely mostly rotate into US treasuries being the most liquid alternative which then itself provides a prop for US equities as yields fall. So if I had to pin my colours to the mast I would agree with low returns, I would agree with heightened risk of volatility g/f due to interest rate sensitivity but I would be nervous sitting on the sidelines. Hence global equities. Having said that if for some reason interest rates do rise materially say 3% or more (at a guess) – all bets are off and then hard to see assets doing anything but reprice downwards
John Kingham says
“I would be nervous sitting on the sidelines”
I agree and I don’t use CAPE as a market timing tool. It’s basically a way of understanding the landscape or the environment that you’re investing in.
So if I was investing in US stocks (which I’m not) then I would know that the market was very bubbly and that some valuations at least were sky high, probably in many of the larger companies in the index. I could then double check to see if I’ve been infected with excessive optimism, and I would know that a major correction or bear market was a serious possibility and I could then mentally brace myself for that.
Eugen N says
I think you are missing a couple of things here, the fact that cost of capital is reducing.
I have my own business in an area that is of high interest to investors, and I am offered capital (equity and loans) at a cost of 3% – 4%. For my tinny firm, I should have not got this for less than 10% per annum.
The problem we have is that in general for the majority of the companies the cost of capital is not much different if this is debt or equity. In fact, in the Wealth of Nations, the cost of capital is the same!
At a PE ratio of 18x your cost of capital is 11%, but it is a lot less at 30x.
This was somehow explained by Elon Musk when he realised the cost of capital for Tesla is now close to 2% (some say it is 0%). As a result, Elon can actually launch his own candy box and take Warren on, more to that, he could actually win in this game, because Warren may not accept such a low cost of capital.
This is why I am quite sure that contrary to what you said, FTSE100 companies may not do that well,, and most likely will follow the Japanese Nikkei! The could be easily disrupted by companies which access capital at next to nothing as cost.
John Kingham says
You’re right; the cost of capital is definitely reducing but my assumption (which may be wrong) is that the cost of capital will return to historic norms at some point. Obviously that might be incorrect, but that’s the assumption. My main problem with extrapolating current conditions out forever is that historically it’s usually been a bad idea to say that the current boom will last forever, low interest rates will last forever, prices will go up forever, etc.
Perhaps this time is different, but perhaps it isn’t.
Eugen N says
Personal, I think that more likely the cost of capital will close on the nominal return on the Government bonds plus 1% or 2% per annum.
This is what Adam Smith explained in the Wealth of Nations, that for the majority of history there was no much difference between the lending interest rate and cost of capital for investing. It may be that the last 200 years were the “this time was different”, and now we are returning to a more historical measure for cost of capital.
I pay attention to what I am paying as multiplier for the businesses I invest in, but more to that I pay attention to what people with less cost of capital are doing to disrupt businesses. If your business has to deal with competition from companies founded by irrational investors accepting small cost of capital, there is only one result, your company’s RoCE will reduce as a result.
My personal view is that S&P 500 will achieve 6,000 in the next few years, and have no idea what will happen after (may drop to 4,500 or 2,500).
In a world with fiat money, it is very easy for Central Banks to do whatever it takes to save asset prices.
Joe Black says
Great article John, and some great comments too. As always, loving the graphs, and agree with the concept.
The opinion that forward 10 year average returns on the S&P could be below zero was limited last year to outliers like Hussman; now I’m seeing it everywhere, even in ‘Marketwatch’ articles, and the commentators are coming to that negative conclusion thru various methods, including CAPE. My own calculations are simple – take the S&P’s dividend yield (sub 1.4% if memory serves), plus the average growth rate of about 4.5% to get a somewhere-below 6% total return. Assume ‘fair value’ is closer to 2500 than 4500, and you get a negative ‘drag’ ono the index, which amortised over 10 years reduces your TR to below 2%. Dropping the price to 2500, BTW, would imply a TR of about 6.5%, so still a teeny bit rich compared to typical S&P TR historically.
That’s less than the inflation we’re all using, which could be a mistake because we’re all assuming the Fed can squash the genie back into the bottle. If they lose control and it runs hot at (say) 5% a year, the S&P will leak over 2% a year for the next decade on average.
Interesting times we live in.
John Kingham says
Interesting indeed. Personally I was surprised by how relatively stable the global economy was after the 2008/2009 financial crisis, so let’s see if we can pull off the same feat post-pandemic, or if this is one debt-fuelled recovery too many.
Bob Barnacle says
Looked at valuation picture a while back and estimated probable current real (after inflation) yields for various Portfolio Asset Classes,
assuming current inflation (CPIH) then @ 1.0% is of some relevance.
BTL = 6.0% (expected range 5-9%)
All Share = 2.73% (expected range 3-5%)
S&P = 1.39% (expected range 1.8-3.0%)
10 year Treasuries = 0.58%
10 year Gilts = -0.25% (yes minus)
Cash = similar
Opinions on ‘expected ranges’ will differ.
Very much share the views in the article plus the thoughtful comments, and also share earlier articles thoughts.
Taking note of valuations and esp being in late retirement have been slicing profits, moving to ……..
25% Risk (Stocks/Bonds/Commodities/Real Estate)
75% Riskless (mix of NS&I IL Certs plus Cash)
Agree about the causes of the distortion and the effects, and needing to think beyond the here and now, instead thinking about the longer term.
“If something cannot go on forever, it will stop.”
Some of us have seen this movie before, perhaps several times.
John Kingham says
Thanks for the various asset class return forecasts.
“Some of us have seen this movie before, perhaps several times”
Yes, experience is one of the many/few (delete as appropriate) benefits of getting older.
I was reading Eugene’s comment and I just wanted to make sure I understood the main point correctly?
“the majority of history there was no much difference between the lending interest rate and cost of capital for investing”
Am I correct in assuming that this means that the economy was not growing at a rate that eroded the value of money and therefore the concept ‘time is money’ did not apply? I only ask because I did look at the UK inflation history and the century where inflation rate was highest was actually 20th century onward. Otherwise inflation was relatively stable i.e. 1% my only explanation is the shift from gold backed currency to fiat in 20th century allowing money to be printed like no tomorrow.
However I thought it would be wise to ask for clarification?
John Kingham says
Hi Reg, I think Eugen means that the cost of debt and equity were very similar to each other a very long time ago. They may well have been, but without looking into it in detail it’s hard to draw any conclusions from that assertion.
Basically the cost of debt or equity is the amount of return you need to make the risk worthwhile.
Equity is typically riskier than debt because a company has a legal obligation to pay back debt as long as it’s in business, but there is no such obligation for shareholders. Therefore equity should have higher expected returns, otherwise why take the risk?
As for the cost of debt and equity 500 years ago, how developed were the relevant concepts, practices and laws, how developed were the markets, how efficient was price discovery? etc.
As an active investor I know that investing in companies versus bonds is completely different and I for one would almost always want a much higher return from equity than debt (although it depends on how risky each is, because very low risk equity can rationally have a lower expected return than high risk debt).
Of course, if other investors are happy with a 1% return from equity investments then that’s up to them, but I certainly don’t expect the average cost of equity to fall to low single digits for anything more than a brief period.
Cheers John much appreciate the clarification!
every time I see this CAPE analysis, I am wondering: why don’t you talk about the buy-backs. In the past, firms were rewarding their shareholders with dividends. This was slowly but very surely (and compounding) the value of the index.
Right now (I did not do any computation) my gut feeling is that share buy-backs is at least as important if not much more than dividends. In my opinion (happy to hear yours), this has to have a mechanical impact upward on where the CAPE will reside and it will likely never go back to the same LT average as in the 50s or 60s.
I am under the impression that we are not comparing apple with apple.
Then there is (as pointed out by other commenters) the Interest Rate policy which is maintained across the globe, not only by the Fed.
I meant This was slowly and surely eroding the value of the index.
John Kingham says
This is an important point and you’re right, technically I should use a total return CAPE (TR-CAPE) but I don’t because it doesn’t make much difference.
I’ve added a few paragraphs and a couple of charts to the end of the article, so please have a quick read of that and hopefully it explains my position on buybacks.
As for interest rates, my assumption is that they will revert to historic norms, but whether they will or not is anyone’s guess.
How would much higher levels of inflation in the future affect future valuations?
If indices in the future are really but we *do* see a return of a significant inflation (without the return of high interest rates), would that not bring the CAPE ratios down?
John Kingham says
Hi Andrew, typically I would expect higher future inflation to lead to higher dividend yields and lower valuation multiples (i.e. a lower CAPE ratio). I’m not sure how likely it is that we’ll get higher inflation without higher interest rates. Most central banks have a mandate to keep inflation close to a target such as 2% per year, and if inflation was running at 5% per year then the obvious tool to use to bring inflation down would be to raise interest rates.
Your blog is amazing! Thank you for all your insights. I have another question about the valuation of the S&P 500 index. I know that I can also predict today’s S&P500 index by using the DDM model, but there should be made many assumptions like Equity premium, cost of equity, growth rates etc. How can I easily find some insights about the DDM model to calculate the S&P index to understand whether it is overvalued or not?
John Kingham says
Hi Ulas, thanks.
If I was going to value the S&P 500 using a DDM (which is actually a very good idea as a way to get a different viewpoint from CAPE) I would want to know (a) the S&P 500’s total capital employed, (b) its average return on capital and (c) the percentage of earnings paid out as dividends and buybacks and therefore the amount retained to drive growth.
Those figures would allow you to build a simple model of future growth and they’re probably available somewhere, but I certainly don’t have them to hand.
Fortunately, we can take a simpler approach with a very mature dividend-generating asset like the S&P 500 because its past dividend growth rate is probably very similar to its future dividend growth rate.
This means that we can use a simpler approach such as the Gordon Growth Model (GGM), which is a discounted dividend model that uses a single perpetual growth rate.
For example, the S&P 500’s dividend yield has historically grown by an average of 6% per year, since the mid-1980s. The S&P 500’s dividend yield is currently 1.3%.
GGM says the expected total return is yield plus growth, i.e. 1.3% + 6% = 7.3%
So GGM says we can expect a long-term return from the S&P 500 of around 7% from today’s price. That’s below the long-term historical average returns, so that implies the S&P 500 is possibly overvalued.
Also, the S&P 500’s average yield from 1985 to 2021 was 2.3% compared to a current yield of 1.3%.
If the current yield was to fall to historically normal levels (implying historically normal expected future returns) then the S&P 500 would have to fall by 44%, from its current level of 4,700 to about 2,700.
This is pretty much in line with what CAPE suggests, which I covered in a recent post on my new website:
I think however you look at it, the S&P 500 seems to be very expensive compared to its own historical track record and other markets around the world today.