SHILLER’S PE RATIO SIGNALS STOCKS ARE OVERVALUED
The list of notable investors claiming the market is substantially overvalued is growing. In addition to the reports from Comstock, Andrew Smithers & Jeremy Grantham, David Rosenberg is now pointing out Robert Shiller’s cyclically adjusted PE. Rosenberg has some interesting thoughts on the currently overvalued market:
The noted Yale economist, Robert Shiller, calculates a very interesting cyclically-adjusted price-to-earnings ratio. Instead of using 12-month earnings (which can be very volatile, especially recently), he uses a 10-year average of earnings. He has compiled an incredible data set, with the data going back to 1881, so you get a true sense of history.
For October, the Shiller P/E is just over 20x (the 125-year plus historical average is about 16x). It’s true that this metric is not as overvalued as in past peaks (in the dotcom era, it went over 40x), but it’s interesting to know that when we usually see 20x, it’s not at the end of a recession but five years into an economic expansion. Shiller is skeptical on the recent boom in the stock market (and housing as well for that matter). On stocks, he said recently “you have to go back to the Great Depression to see such a turnaround in the stock market” and that the current booms (both stocks and housing) “…can’t be trusted to continue.”
Source: Gluskin Sheff
Note that the market was at the same or higher P/E10 level through most of the 1960’s. Also the 16 PE average includes both the Great Depression and the inflationary 1970s and 1980s. I would argue that a normal P/E10 under more normal circumstances would be about 18. That makes the current market 10% overvalued based on normal circumstances and maybe 20% on long-term overall trend and maybe 70% overvalued based on a depression like economic backdrop. The market is still so much cheaper than it has been through most of my investing lifetime 1994 to present.
The market might be able to hang on for years at these levels. The market could trade in a range between 18 and 22 up and down (say between SP500 1000 and 1250) like the 1960s if interest rates remain low and the economy slowly makes at least some headway. Moves lower or higher would probably be short-lived.
But if inflation spikes, it may drop to a P/E10 under 10. That might take the market to new generational lows. Without moving towards depression or massive inflation that scenario unlikely.
I would say when the market is at a PE10 below 18 (SP500 1000) it is a buying opportunity (the further below, the better the opportunity). At 23 or above (1280) a sell and in-between a hold. That is until the market perceives that inflation and interest rates will move higher.
If you’re going to exclude the 1930s and 70s, you also have to exclude the 1960s and 1990s. Either way, fair value on the market should be between 15 and 16.
Rob Reply:
October 29th, 2009 at 8:07 PM
I agree that long-term trend PE10 is 16 and fair value on a long-term trend would be at a multple of 16, say $56 x 16 = 896 or maybe as high as $60 x 16 = 960
Nevertheless the level of interest rates and inflation clearly play a role in the appropriate multiple. (One needs to see what alternatives yield). The higher interest rates the lower the multiple. Interest rates usually are in part driven by inflation expectations (in addition to Fed policy). Interest rates in the 1960s were low so a higher multiple could be justified than under average historical circumstances. The Depression was another case all together. The Fed didn’t offset the credit contraction to the same extent as today so real deflation took hold together with low nominal interest rates. Real interest rates were quite high in the Depression for an extended period of time.
The 1990s and most of the 2000s were a bubble. The PE10 was continuously above 23.
I am just saying I see it as more likely that as long as interest rates remain low that stocks stay in a range from 16 to 22 than move down to the 16 to 10 range.
This part of Rosie is the weakest. I’ve seen this multiple times and the two most common criticisms never get answered:
(a) if this indicator is so good, backtest it at least … show the mean return, correlation and stat significance for the following X years after a smoothed “cyclical P/E” is Y (say 11 for undervalue, 15 for at value etc.)
(b) adjust for interest rates (ie adjust for riskless interest rate, or basket of gov+IG bonds etc.)
MARKET QUOTES
MARKET NEWS
THIS WEEKS MOST POPULAR STORIES