In December 2015 I wrote the first of what will hopefully be a very long series of annual forecasts for the FTSE 100, FTSE 250, S&P 500 and UK housing market.
That first article forecast the value of the FTSE 100 at the end of 2016 and – miracle of miracles –was just about spot on.
Here are the gritty details:
- In late 2015 the FTSE 100 stood at 6,061
- My forecast of “fair value” for the end of 2016 was 8,272
- My forecast of “expected value” for the end of 2016 was 7,100
- The FTSE 100’s actual value at the end of 2016 was 7,120
The fact that my forecast was within 0.3% of the actual result is of course mostly down to luck; unfortunately the future is just not that predictable.
However, it wasn’t all luck and I do think my forecasting method is fundamentally sound, although of course there’s always room for improvement.
Most of the investment forecasts I’ve read are of the “I think such-and-such will happen” variety, which is definitely not how I approach the task. Instead, my approach is based on facts and figures and long-running historic norms rather than speculative flights of fancy.
If you’re interested, I’ve worked through my forecasting method in some detail below. If you’re not, then skip to the bottom to see what the future (possibly) holds.
This forecast is based on the cyclically adjusted PE ratio (CAPE)
The CAPE ratio is my standard measure of value for stock markets and so it makes sense to base my forecasts of future stock market values on that ratio as well.
The basic steps of are to forecast:
- Inflation: For this, I use the latest Bank of England inflation report
- FTSE 100 earnings: I assume next year’s earnings will equal the average real earnings over the previous decade
- Cyclically adjusted earnings: This is the FTSE 100’s ten-year real average earnings up to the end of 2017 (i.e. including the earnings calculated in step 2)
- Fair value: I assume fair value occurs when the FTSE 100’s CAPE ratio equals its long-run average, which I take to be 16.
- Expected value: I assume the expected CAPE ratio a year from now is halfway between the FTSE 100’s current CAPE ratio and its fair CAPE ratio.
Okay, let’s dive in and work through each of those five steps in turn.
Forecasting inflation for 2017
The inflation “fan chart” from the Bank of England’s latest inflation report shows that the BoE’s central forecast for inflation in 2017 is slightly over 2%. I’ll be a bit cautious and take 2% as my inflation number.
- 2017 inflation forecast = 2%
Forecasting FTSE 100 earnings for 2017
The FTSE 100’s average real earnings over the last decade came to 484 index points, and that’s basically my forecast for 2017’s earnings. However, that number has to be adjusted upward for inflation over the next year, so:
- 2017 earnings forecast = 484 + (484 * 2%) = 493 index points
Just to give you an idea of how that compares to last year, 2016’s earnings at the end of the year were 212 index points. That’s actually an incredibly depressed level of earnings, caused largely by huge declines in resource and energy-related company profits (two sectors which the FTSE 100 is weighted towards).
So my forecast is based on a doubling of profits over the next year. That may seem optimistic, but really it’s just forecasting a return to normality from currently depressed profits.
Forecasting cyclically adjusted earnings for 2017
We now have one-year earnings for 2017, so to get the cyclically adjusted (ten-year inflation-adjusted) earnings we can just add 2017’s earnings (493) to the sum of real earnings over the previous nine years (4,244) and then calculate the average:
- 2017 cyclically adjusted earnings = (493 + 4,244) / 10 = 474 index points
This compares badly to cyclically adjusted earnings a year ago, which were higher at 517 index points. The reason again is declining profits over the last three years and 2016’s especially weak results.
Forecasting fair value at the end of 2017
On the basis that cyclically adjusted earnings at the end of 2017 are forecast to be 474 index points and that the fair value CAPE ratio is 16, we have:
- 2017 fair value for the FTSE 100 = 474 * 16 = 7,580
This is lower than last year’s fair value of 8,272 because, as we saw in the previous section, the cyclically adjusted earnings are forecast to be lower.
A declining fair value is of course not what investors want to see, but there is little we can do about it (other than invest outside the FTSE 100 or cherry-pick the best companies at the best prices from the index).
Forecasting expected value at the end of 2017
If you decided to skip the maths then this is the bit you’re looking for. If you didn’t, then hopefully you have a much better idea of where this forecast came from, and more importantly that it did not come out of a hat.
The final forecast is based on the fact that market valuation ratios tend to revert towards their historic averages (or means), a process known as mean reversion.
This can take many years and of course is not smooth or predictable, so my assumption here is that the market will close the gap between its current and fair CAPE ratios by half over a one-year time horizon, at which point the market will be at its expected CAPE ratio.
Looking at the history books we know that at the end of 2016, the FTSE 100 stood at 7,120 and had a CAPE ratio of 14.7 compared to a fair value CAPE ratio of 16. So the market was slightly cheap, but not exactly in bargain territory.
The expected CAPE ratio is then easy to calculate:
- 2017 Expected CAPE ratio = (14.7 + 16) / 2 = 15.4
Now that we have an expected CAPE ratio we can easily work out the expected value for the FTSE 100 by combining the forecast cyclically adjusted earnings (474) and the forecast expected CAPE ratio (15.4):
- 2017 Expected value for the FTSE 100 = 474 * 15.4 = 7,277
Today the FTSE 100 stands at 7,160 and so the expected change in its value over the next year is small at just +1.6%.
Of course, there is a lot of false accuracy in quoting the forecast value to four significant figures, so please remember:
- This is effectively a probability-weighted average of all possible values
- The full range of possible values is very wide, most likely spanning several thousand index points
- Values further away from the expected value become gradually less likely (e.g. 4,000 or 10,000 are far less likely than 7,300)
Despite all the necessary caveats, I do think this forecast is a good indicator of what we might reasonably expect the market to do over a one-year period, and on that basis, I am very mildly bullish on the FTSE 100 for the rest of 2017.
Finally, here’s a chart showing the FTSE 100 and its CAPE valuation over the last 30 years, including the 2017 forecast:
Hello John, magnificent as always but with all the dates and numbers I think you got mixed up, don’t you mean 2017 in this sentence: ‘Looking at the history books we know that at the start of 2016 the FTSE 100 stood at 7,120’? Keep up the good work, no matter where the FTSE 100 ends 2017 one thing we can all be sure of it will be an exciting ride!
Hi Tony, thanks for pointing that out. I meant “the END of 2016”, which is more or less the same as you suggested. As for an “exciting” ride I think you could be right, but personally I’d take boring over exciting any day!
Hi John,
A great article,as ever!
One question – why has the cyclically adjusted earnings figure dropped from 517 in 2016 to 474 in 2017?
Best wishes
Jonathan.
Hi Jonathan, actually that 517 refers to cyclically adjusted earnings as at the end of 2015. Not sure why I referenced that one; what I should have said I think is the value at the end of 2016 which was 484.
However, the forecast of 474 is still a decline from that and as the decline in the rainbow from the chart shows, cyclically adjusted earnings have been falling for a while.
This is because the FTSE 100’s earnings in recent years has been significantly worse than they were a decade ago. Each year when cyclically adjusted earnings are calculated, the oldest year in the ten-year period is dropped and replaced by the new year. If that new year’s earnings are lower than the earnings ten years ago then the average, or the cyclically adjusted earnings, will decline, and that’s basically what’s been happening for the last three years.
Ten years ago we were in the pre-crisis boom, so it’s no surprise that earnings were higher then. I expect earnings to start growing again at some point, although the growth rate could be somewhat limited once inflation’s taken into account.
Hi John,
Thanks again. Yes I see your point that the inflated earnings from the pre-crisis boom are falling off the first 10 years of the CAPE calculation.
One overlay I’d add to your calculation is that the devaluation of sterling will lead to a jump in dividends for sterling investors this year. Capita estimates this to be c.6%, rather than the 2% as you have assumed. It would provide a modest upside to your forecast.
Best wishes
Jon.
I did think about including some sort of devaluation factor but decided to just stick to BoE inflation. Hopefully you’re right and there is a meaningful and positive boost to earnings. We shall see…
John,
Fun exercise, but for what (actionable) investment purpose do you speculate about the FTSE level one year hence?
Thanks
Jim
Hi Jim, good point. My goal is to help investors understand where we are in the valuation cycle and, more importantly, the sort of medium- to long-term returns they might see.
I could simply say that CAPE is currently 14 and that the long-run average is 16, so it’s slightly cheap, but that’s a bit abstract which makes it easy to ignore for some people.
So I thought having an actual forecast would be more “real” and have greater impact and stickability in people’s minds.
If, for example, it was the start of year 2000 and the market was at 7,000 with CAPE at 32 (for simplicity’s sake), then I would forecast a one-year CAPE ratio of 24. Ignoring inflation or earnings growth (which are both pretty small anyway) that would mean a 2000 year-end forecast of 5,250 for the FTSE 100.
Hopefully such a bearish forecast in a massive bull market would have made some investors sit up and take notice, more so than simply saying “CAPE is forecast to be 24”.
Hi John,
I like your general approach to valuations, but I worry a bit about having only 30 years of available data for the FTSE100 CAPE. Is there any way to estimate the CAPE for the period before 1987?
Cheers,
Fed
Hi Federico, I agree that 30 years is not very much to go on.
My current approach is to cheat and have a base assumption that the long-run UK average CAPE is similar to that of other countries. For example, the US (S&P 500) 100-year CAPE average is between 16 and 17 (don’t have the exact figure to hand). That’s why I assume the UK long-term CAPE average is 16. I will of course adjust this as the UK dataset grows.
Looking further afield, Cambria Quantitative Research published a paper called “Global Value: Building trading models with the 10-year CAPE”. In that paper Meb Faber (a well-known CAPE commentator) lists long-term CAPE averages for a long list of countries. Most countries in the list have data going back to the 80s or 90s, but the US and UK are apparently based on data going back a century.
The US median (not mean) is quoted as 14.6 and the UK as 11.8. The US figure median figure is lower than the mean because peak valuations in bull markets are very high, and mean is more sensitive than median to extreme values. Personally I prefer mean to median, but I doubt it makes much difference either way.
The UK figure looks to be too low for the modern era. Perhaps UK investors were super-cautious pre “big bang”? I don’t know, but a CAPE ratio of less than 12 is unlikely to be the median going forwards in my opinion.
Anyway, the median of all the different countries’ long-term CAPE averages is 17.8. So very unscientifically, I would say it’s reasonable to say that CAPE is likely to average somewhere in the mid-teens in the future.
However, historically it has been capable of going from half to double its long-term average in any given country, so worry whether the average value is 16.6 or 17.2 is unlikely to be of much benefit in terms of making investment decisions or predicting where the market is going. However, if CAPE is at 10 or 30 then that’s a pretty good sign that the market’s undervalued or overvalued respectively.
Thanks for this comprehensive reply. Is the median UK CAPE at 12 in the Cambria study calculated over all shares or is it for FTSE100? One would expect the all shares CAPE to be lower than FTSE100 CAPE.
Clearly, there is no physical law forcing the CAPE to revert to its long-term mean or to any other value, and it could be argued that the expected CAPE value in the current low interest/print money environment is higher than in other situations. Still, I feel that this provides a good tool to gauge the markets…
Unfortunately neither the FTSE 100 nor All-Share existed during most of the last century, so the UK CAPE data won’t tie in exactly with either index.
The underlying data comes from this company:
http://www.globalfinancialdata.com/Databases/UKDatabase.html
So you could get in touch with them for a definitive answer.
Thank you and very interesting article. However !
Forecasting the FTSE purely on earnings can be a real challenge because it cannot; with any measure of accuracy, take into account the possibilities of geopolitical events that may in turn affect global economic growth. Off the top of my head – Trump protectionism, the Chinese economy, and the final meaning of ‘Brexit’ for the UK economy – I think the major flaw with forecasting is the number of unknown variables that could quickly come into play and totally throw off the original forecast. Whilst there is certainly a place for this type of analysis, I believe it is a type of reasoning that needs to be constantly revised as the economic year unfolds. In which case it becomes very short-term. Indeed we could be at FTSE 7,500 – but nobody can say for sure that we will not be at FTSE 6,000, or even 5,000. I do of course respect your work and you may very well be right in your forecast !
Hi Kaisar, that’s a very important and correct point. I think I should change the name from forecast to projection. A forecast is what you might reasonably expect to happen, while a projection is either a specific possible scenario or an extrapolation of the past.
My “forecasts” definitely fall into the extrapolation camp, rather than being something I expect to happen.
Another area of confusion is the terminology of “expected” future value, where expected is the probability-weighted average future value. But the name “expected” makes it sound like I actually expect the market to be at its expected value, which I don’t!
The expected value of a die roll is the best example, where the expected value is 3.5 (the average possible value) but if you expect to roll a die and get a 3.5 you’ll be very disappointed, no matter how many times you try.
So thanks for pointing this out and in future I’ll try to remember to call these market projections rather than market forecasts.