Revisiting CAPE

It’s interesting to think that Shiller’s CAPE has been somewhat dismissed over the years largely because investment advisors haven’t been able to figure out how to include it in their business.

No, it’s not a market-timing tool but you can’t sweep under the rug the fact that paying up for stocks will result in lower returns going forward.

Recently, I read John Rekenthaler‘s article from last summer, “Maybe There’s Something to the Shiller CAPE Ratio, After All – New research rehabilitates an old measure.”

John makes the point:

“It’s tempting when viewing the CAPE ratio’s chart to conclude that astute investors could have heard the thunder before the rain arrived.”

He points to a 2014 study by Elroy Dimson, Paul Marsh, and Mike Staunton, that concludes you can’t create trading rules using CAPE.

While you may not be able to develop a trading strategy, you can certainly develop an asset allocation policy around it. Read for an example.

However

John cites the article, “The Remarkable Accuracy of CAPE as a Predictor of Future Returns,” by Michael Finke of The American College of Financial Services, and states the piece “may have changed my mind.” 

He then recreated the numbers (CAPE, Subsequent Returns):

Converting nominal returns to real does not alter the picture.

Shiller Paper

The 1996 Shiller paper concluded that “these ratios were extraordinarily bearish for the US stock market.”

Shiller offered the same regresssion between valuations and subsequent returns.

Of course, anyone following the irrational exuberance of stocks at the time could have made a policy decision to get out of stocks.

Of course they would have missed 4 years of asset growth from stocks. But in the end, they would not have lost their shirts by staying invested in stocks.

They would have been left with more assets having sidestepped the 80% fall in stocks. AND in the end would have been able to re allocate to stocks when the CAPE was reflecting above average returns ahead.

One might say … that’s good market timing.