The S&P 500 has had a much stronger post-financial crisis bull market than the FTSE 100. However, as a consequence, the US large-cap index is now much more expensive than its UK counterpart.
Bull markets are nice, but they do have a nasty habit of ending
From a low of just under 700 points in March 2009, the S&P 500 now stands close to 2,000. That’s a gain of almost 200% even without including dividend income.
In contrast the FTSE 100 has increased by less than 100%, going from around 3,500 to 6,200 today.
To a large extent bull markets are driven by confidence and expectations of short-term gains. As long as the bull market keeps going up investors keep adding new money into the market, which drives the market upwards even further.
But confidence (and greed) can only take a market so far and more recently the S&P 500 has ground to a halt.
Returns over the last year or so have been flat and so perhaps this is a good time to look at which way the market might go if confidence cannot be relied upon to drive it ever upwards.
S&P 500 valuation: Expensive relative to historic norms
As long-time readers will know, I like to value indices by using Robert Shiller’s CAPE ratio (Cyclically Adjusted PE), which compares an index’s current price to the ten-year average of its inflation-adjusted earnings.
It’s very similar to the PE ratio, but much more stable (thanks to the smoothing effect of using ten-year average earnings instead of just last year’s earnings) and with a much closer connection to long-run future returns.
The chart below is my attempt to visualise the range of values that the CAPE ratio can take as the S&P 500’s price moves up and down relative to its average earnings:
Here are a few of the key ideas underlying this rainbow chart:
- Over the past 100 years, the S&P 500’s CAPE ratio has had a median (middle) value of 16
- The CAPE ratio spends most of its time between about half and double its median value, i.e. between 8 and 32
- It is reasonable to describe the S&P 500 as being in a bubble when CAPE is above 32 and in a depression when it is below 8
- The higher (or lower) the CAPE ratio the more likely it is that it will decline (or rise) over the medium-term, back towards its median value, i.e the ratio is mean-reverting
So where are we today?
With the S&P 500 index close to 2,000 its CAPE ratio is 25; that’s some 56% above the long-run median value of 16.
One way to think about this is to see the S&P 500 as being about halfway between normal (CAPE of 16) and bubble territory (CAPE of more than 32).
This relatively high valuation also shows through in the index’s dividend yield. Currently, it is just 2.2%, which is a very long way short of the 4% available from the FTSE 100 (with the FTSE 100 at 6,200).
So the S&P 500 is expensive relative to historic norms, but what does that mean for future returns?
S&P 500 forecast: My model suggests a small decline is likely
Before I make my one-year forecast for the US large-cap index, I just want to make something very clear:
I have no idea where the market will be in one year’s time, and neither does anybody else.
Rather than trying to guess exactly where the market will be, this forecast is an estimate of the market’s expected value, i.e. the average value of the S&P 500 a year from now if we could re-run the next year a thousand times over.
This forecast will be based on the following assumptions:
- The cyclically adjusted earnings of the S&P 500 will increase by 4% over the next year (as they have done on average over the past 30 years)
- The CAPE ratio will move halfway back towards its long-run median of 16 (due to its mean-reverting nature, see this FTSE 100 forecast for a bit more about this idea)
The increase in cyclically adjusted earnings would take those earnings from 80 index points today to 83.3 a year from now, while a 50% move back towards the long-run average for the CAPE ratio would leave the ratio at 20.5.
Putting those two together (by multiplying the earnings of 83.3 by the expected CAPE ratio of 20.5) gives the following:
My forecast value for the S&P 500 in March 2017 is 1,700
As I mentioned above, I don’t actually think the index will be at 1,700 next March. This forecast is just my best guess at where the index will be, much like 3.5 is the best guess for the roll of a die, even though the odds of actually rolling a 3.5 are zero.
So with the S&P 500 currently close to 2,000, that forecast suggests an “expected” decline of around 15% over the next year, which sounds entirely reasonable to me, especially given the index’s 200% run-up following the financial crisis.
It is an interesting issue this one. First I do not argue with you that the US market is not expensive.
I would argue that the method you use is flawed, at least for this economical cycle.
Generally I do not believe on methods which are based on past earnings because stock markets do not try to price past earnings, their job is to discount future earnings. As a result past earnings are only relevant if you believe they will repeat again, which I am not so sure.
If stocks market prices were based on past earnings, librarians would have been millionaires, and this is not the case as yet.
You may argue that market participants are not good at forecasting future earnings or choosing a proper discounting rate. Subject for another on a long beer.
Let’s look a bit at CAPE over the last 10 years in the US. First it is higher than the running PE ratio, mainly because inflation was low and earnings in 2008 and 2009 were very depressed or Nill. I am not quite sure how CAPE works for years like 2008 and 2009 when the US markets overall earnings were negative? Banks, financial institutions, miners have lost more than the earnings produced by all other sectors.
Hi Eugen
I agree, stock markets are not priced on past earnings, but when you average past earnings over a ten year period (or longer) there is a fairly good correlation between that past ten year average and the subsequent ten year average. So measuring the price against past average earnings (e.g. with CAPE) is a reasonable proxy for measuring it against the next ten year’s average earnings (which of course we can’t actually do as the next ten years haven’t happened yet).
As for this last ten year period being special because of the extreme volatility of earnings, and even negative earnings, I would also agree.
The cyclically adjusted earnings for the most recent period are skewed downwards by the low earnings in 2008. That will of course impact the effectiveness of CAPE. However, I have fiddled with the data to pretend that the financial crisis never happened and the resulting average earnings is not especially different. It goes up by about 10%, pushing the current CAPE ratio down from 25 to about 23, so the S&P 500 would still look a bit expensive, but not quite as much.
In reality of course this sort of minor difference will be swamped by the much larger elements of randomness and luck, but your point is still valid in that CAPE is affected (to a small degree) by extreme earnings in one year.
There are certain some sectors in the US stock market which are probably very expensive, like the unicorns and biotech sector.
Consumer stapples, the kind of stocks that Terry Smith invests in are on a CAPE of around 21, not cheap but not dreadful expensive.
There are cheaper US stocks as well like Warren Buffet’s Berkshire Hathaway conglomerate in which I have quite a few shares. I am 12.5% up from the time I invested a couple of months ago.
Ourselves we have taken some profits for our clients from the US market and reinvested in Europe and Japan. QE seems to be a good steroid for the time being. We are well underweight UK, where there are more problems than value in my opinion.
One thing that was not discussed here is the discount rate. I would argue that given the low US Treasury yields, the discount rate used should be lower and this will inevitable lead to higher stock prices.
Regarding the discount rate, it isn’t something I pay attention to. Interest rates (and therefore the discount rate) may be low today, but they may be high in the future, and I don’t want to justify paying a high price (relative to historic norms) for stocks just because the interest rate is current low (relative to historic norms) at this present moment in time.
The discount rate is very important when you determine the price to pay for a stock. You can name it cost of capital or other names, but without it you cannot understand stocks pricing. You may even used it without knowing you use it.
It was first explained in Wealth of nations by Adam Smith. The discount rate varies very little: it seems to be around 5% per annum now, down from 6% per annum. However because it compounds it has a significant importance.
There is another important note, it seems it does not revert to the mean. In fairness I am not sure that interest rates revert to the mean, myself I believe in random walk.
However the discount rate has decreased over the last 100 years from 1900 to 2000. This helped the market return with the tune of 1% per annum over the last century, not bad.
You can look at this in many ways, you can argue that investors are more comfortable to buy expensive stocks.
The one thing that it is important: discount rate does not have a direct correlation with interest rates. Perfect example is Japan where although interest rates were 0% over the last 20 years, the discount rate increased significantly. Probably because people lost interest in the stock markets.
Hi Eugen, thanks for that, you covered a lot of ground. I do understand the discount rate, and by desiring a certain rate of return (say >10%) then I’m implicitly using that as my discount rate, but I don’t explicitly think about the market discount rate, or compare discount rates between stocks, bonds or anything else.
I just want a return of more than 10% per year with low risk, and that’s it.
Have you considered the effect of mark to market on historic CAPE ratios, interesting work from Jeremy SIEGAL here, http://www.q-group.org/wp-content/uploads/2014/01/2013fall_siegelpaper.pdf
Hi Hari, yes I’ve seen that research before. It is quite probably a valid argument, that changes to accounting rules change reported earnings and therefore the ‘correct’ CAPE ratio. However, the question then is what to do about it.
I have spent a least some time thinking about this and my response was to restrict the calculation of the long-term CAPE average (i.e. the mean value that CAPE is expected to revert back to) to a period of 100 years, rather than over the entire Shiller database which goes back to 1871 (if I remember correctly).
By measuring the long-term CAPE average over 100 years, any recent changes to the earnings rules and the more recent CAPE values will have a much greater impact on the long-run average than if that average was always calculated back to 1871.
However, it will still take several decades for any meaningful change in that average value to appear. And that’s fine by me as the long-run average value is supposed to be extremely (but not overly) stable, as that is pretty much the whole point of using it.
Another way around this would be to use a different measure instead of earnings, such as cash flows or “replacement value” (which is used to calculate Tobin’s q, another and perhaps even better long-term valuation metric). But personally I’m happy sticking with CAPE, but with my tweak of using a 100 year mean.
You can see a chart for both CAPE and Tobin’s q on the Smither’s & Co website here. Currently the’re both indicating a similar degree of overvaluation relative to their respective historic averages.
In relative terms, whatever measure you use: foward P/E, CAPE P/E, Tobin-q, price to book etc, the US stock markets are more expensive than they were 5 years ago.
Now what you do about it makes the difference. Timing the market as we know it does not work. You can try at best to be take some profits and reduce the US exposure, which we did. The Japanese and European markets are more acomodative both in terms of valuations and use of QE.
You also need to explain to yourself why the US markets are so expensive and if the expectations of free cash flow are to be believed. We shall not forget that the price we pay today for a share is the discounted value of the free cashflow generated by the company in the future and nothing else (at least in theory).
Market participants, at least the active market participants, because probably passive market participants do not care, try to forecast and find an answer – their answer. The result of all those answers is the share price on the Bloomberg screen.
We all acting behavioraly and I have not met an investor who does not. There is no Holly Grail tool to determine stock prices, mainly because forecasting economic activity, interest rates, commodity prices etc is all futile.
Myself as many I use probabilistic tools. We are choosing our companies which we believe given their characteristics have a greater chance to outperform over a 3-5 years period.