One of the essential principles of investing, and indeed something that is highlighted often by greats including Graham, Buffett, and Munger, is that you can't make very much money by trying to predict the future. This principle can be tested reasonably well by looking far back, and using Discounted Cash Flow (DCF) analysis to see how well the market predicted real returns.
The biggest challenge with DCF is that you can't easily predict the stream of future earnings, so while you can make as fancy a model as you like, it is unlikely to be very accurate. If you go far back enough though, we know what the future stream of earnings was. Using this future knowledge, we can go back and see how well the market predicted this future stream.
I took Shiller's historical earnings data (monthly real earnings) and discounted the future earnings from every date (on the X axis) by various real rates represented by different thin lines on the graph. I computed the terminal value (future after today) by taking a multiple of the earnings this past month - the multiple was just the inverse of the real rate for each line.
The graph only goes up to the mid-60s because after that the terminal value becomes too significant a fraction of the total value (defeating the whole purpose of testing against real historical values!). The graph shows only values where the terminal value is less than 20% of the total.
I superimposed upon this graph a thick black line representing a simple valuation based on CAPE (the 10 year average earnings that Shiller made famous). This CAPE valuation doesn't rely on any crystal ball since it looks at previous earnings and divides that by 6%.
The values on the graph take the ratio of the actual value of the S&P500 divided by the value implied by the future (or in the CAPE case by a simple CAPE multiple).
There are a couple key observations I would make from this:
- Simple earnings multiples are about as good as a fancy DCF analysis, even if the DCF analysis has perfect information!
- Market predictions are not very good - the actual value is often far different than the fair value calculated by discounting perfect future information. For example, in the late 1920s, the actual value was ~2x higher than fair value (leading to the great depression!), and in the mid-50s the actual value was 50% or less of the fair value.
What does this mean for us in the present? It says to not waste a lot of time making fancy DCF models because you don't know what inflation is going to be, you don't know what earnings cycles are going to look like, etc. While that is unfortunate, it also means nobody else can predict these things either!
As I've laid out in other articles (here, here, and here), historical information today implies that we are at the start of a bull market, in which you should own equities, and in the absence of any other insight, owning a basic ETF like SPY is a great way to participate.