Longtime Daily Insights readers are no strangers to the Shiller PE-ratio, a valuation metric that uses ten years of inflation-adjusted earnings to evaluate the cheapness or richness of the stock market – a search of the term on our website yields more than a dozen results.
I’m very proud to say that our understanding of the Shiller PE is now greatly expanded, thanks to our own Ryan Craft.
In the past, I had, like many others, used the Shiller PE to forecast stocks by simply taking the reciprocal of the ratio to get the earnings yield. For example, if the Shiller PE is 20, the earnings yield is five percent (1/20). In theory, the earnings yield should match the market return over a long time period, like 10-years.
While this yielded better results than any other forecasting model I’ve seen, the signal-to-noise ratio was only about 40 percent, which meant that there was more noise than signal.
Ryan made a simple, yet unconventional observation that I hadn’t seen anywhere else – and it’s only simple in retrospect. Ryan’s point was that since the earnings in the Shiller PE are inflation adjusted, the forecast needed to add inflation expectations. My folly, along with many others, was that I had been using ‘real’ data to forecast nominal returns.
After a variety of tests, Ryan found that including inflation expectations brought the signal-to-noise ratio for US and non-US stocks up to 60 percent and 80 percent respectively (although I would hasten to add that the ex-US period is much shorter, so that may be overstated). While the signal-to-noise ratio is far from 100 percent, it’s a significant improvement over 40 percent.
You’ll see that the expected returns for US stocks and bonds are low right now, which is not a new message from Acropolis (click here). You’ll also see that the expected returns from non-US stocks are much higher, which is also not a new message (click here).
These ‘old’ articles, which both date back to 2014, should be good reminders that the Shiller PE ratio is not a useful market timing tool. The US stock market is up 30 percent since the first article and international stocks aren’t even up 10 percent in that time period. You might say that I was wrong, but I’ve got seven years before we can really say, which highlights the problem of such long-term forecasts – it takes a really long time to decide.
We are looking at how we can use the updated forecast model, from our asset allocation to our assumptions in our financial planning models. But even though our understanding of the Shiller PE ratio is better now than ever, it’s still a blunt instrument that should be used with substantial caution.
Over the years, I’ve seen plenty of investors misuse the data and make dramatic changes to their portfolio. So far at least, that hasn’t paid off and I don’t think we’ll suggest anything drastic in the future.
We see this data and rather than reinventing our time-tested approach to investing, we think it means that investors should adjust their expectations about the future.
Rather than searching out a ‘magic bullet’ for your portfolio, we think that investors are best served making sure that their saving or spending goals are sound. We know that’s not necessarily the solution that people want to hear, but we think it’s the right answer regardless.
While I’m very excited about this research, it’s important not to overstate its value. We hope to make incremental improvements to our processes, but we also know that markets are noisy and that there is real danger in putting too much weight in a forecast.
As it’s been said many times before, ‘forecasting is difficult, especially when it concerns the future.’