This post over at The Big Picture Blog which had a listing of Bull markets of 20% or more without a 20% correction got me thinking about Bull Market fundamentals. Some people talk about this bull market being driven by the Fed and not fundamentals (me included). I can easily point at a Shiller PE of 23 to highlight overvaluation but I wanted to look at it another way, so I focused on PE expansion.
Fundamentally driven bull markets should rely more on cyclically adjusted earnings growth and less on investors willingness to pay ever increasing multiples on those earnings. To look into this I decided to focus only on bull markets of 100% or more. I looked at the Starting and Ending Shiller PE using Robert Shiller's online data and updated it with daily pricing data for the important dates (as he only has monthly prices). Then I divided the Bull Market gains by the amount of PE expansion to see how much gains investors were receiving per unit of PE expansion. The results are below, sorted by most fundamentally driven to least fundamentally driven. The results are quite interesting.
Take a look at the 1974-1980 Bull Market compared to today....The magnitude of the advance is similar between the two but the 1974-1980 advance only relied on a PE expansion of 2.2 vs 11.1 today. You will also notice that those that relied least on PE expansion tended to experience smaller subsequent bear markets. The top 5 averaged a bear market loss of 30.4% vs the bottom 4 which averaged a 48.5% loss. If history is any guide people should expect that the next bear market will be deeper then average because this bull market is lacking a fundamental underpinning.
UPDATED: 3/19/2013 - Corrected chart to reflect the proper starting PE of 8.8 for the bull market starting 6/13/1949. However, this did not change the overall ranking of any of the bull markets.
This is interesting. But it's not appropriate to subtract PE values. Since PE values are ratios, you have to divide them to get an appropriate comparison. I'd be curious to see how the analysis changes when you do that.
ReplyDeleteThe only problem with that approach is that the result you are getting is a % change in the PE ratio and the result is that a VERY undervalued market getting to fair calue will look like it's expansion was just as dramatic as a more astronomical move. For instance...just for the simple math....a market going from a PE of 4 to 12 (300% PE expansion) would be considered just the same as a market going from a PE of 12 to 36 (300% PE expansion) -- I think we can all agree that is not the case. By using absolute PE multiple values it helps take the starting PE level into account.
DeleteActually, I don't agree. An expansion from 4 to 12 is exactly the same as an expansion from 12 to 36. Basically, if you take the ratio, you can calculate what percentage of the rally derived from an expansion of E and what percentage came from an expansion of PE.
DeleteDoing the analysis, I find that the rallies of 74, 49, and 87 were about half due to increases in E and half in PE, and the rallies of 32, 33, 35, and 82 were entirely PE driven.
The current rally is right in the middle (84% PE, 16% E), almost the same as that 42 rally.
The purpose of this view was to look at fundamentals underlying the market advance. I don't think you can do that without also taking into account in some form the fundamentals underlying it's starting point. You are correct, I could just look at the % the PE expanded but that would give absolutely no weight to the fundamentals from which the bull market started (which DOES matter).
DeleteTake for instance the bull market of 2002-2007...if I just looked blindly at the % of PE expansion it would be the second smallest at a 39.1% increase (from 20 to 27.8). But the reality is the starting PE of 20 was by far the highest out of any (47% higher in fact)....which means the fundamentals were already low before it even began. By using absolute levels, this helps take these factors better into account.
Honestly, if I were to look a % increase in PE in a vacuum this exercise would be of no use IMO. It is no wonder that despite only a 39.1% increase in PE that the market eventually erased every gain of the 2002-2007 rally. This is because it started at an elevated level and ended at an even more absurd level.
Brilliant analysis. Would love to see a similar analysis that looks at earnings too.
ReplyDeleteI will be adding this post to my CAPE Ratio Catalog at www.pe10ratio.com - when I get a chance to make my next round of additions that is (as time permits).
Kay C
What about impact of inflation on earnings?
ReplyDeleteSo if growth isn't driven by P/E expansion, then it must be largely driven by underlying earnings growth. Looking at historical US inflation, the two bull periods highlighted with the the largest "Bull Fundamental Ratio" also were periods of high inflation in the U.S, especially '74 - '80.
Yes there is high growth not driven by P/E expansion, but the dollar was also getting destroyed.
http://www.comparativepoliticseconomics.com/historicalinflationrate.html
Using the Shiller PE helps take inflation into account because it uses inflation adjusted earnings. You will notice the second best is the 1949-1956 period which actually had very low inflation (averaged 2%) (deflation at times in fact, only 1952 was somewhat elevated).
DeleteWould it make more sense to calculate % Price Change / % PE Expansion? that way you would get a better understanding how much of the % price increase came from PE expansion rather than earnings growth. You can always look at PE levels to assess under o over valuation. For instance in 1974-80, the 2.2 PE expansion was 28%, therefore, aprox 97% (125-28) was due to earnings growth (or 11% per annum)while in 1935-37 the % price increase was almost exclusivly from PE Expansion 128%
ReplyDeleteMy question is why you would choose Shiller P/E -- a trailing 10 year inflation-adjusted measure -- versus a simple trailing 4 quarter actual earnings.
ReplyDeleteif you want to measuure how much a rally is due to E, wouldn't you need to restrict your E measure to that rally period, or something reasonably close?
Good to hear from you Barry!
DeleteI didn't use simple PE ratios because simple PE ratios are essentially useless. Simple PE ratios are near their highest at market bottoms and are artificially low at market peaks. Although I know investors are naive, I would still hope they cyclically adjust their earnings in some fashion (although useless simple PE ratios are always the ones used on TV). If you were to use simple PE ratios you would essentially have a declining ratio.
For example simple PE ratios peaked 6/30/2009 (using quarterly data only), you had a trailing 12/mo As reported PE of 122.41 and a 12/mo operating PE of 23.1. Last time simple PE ratios were that high? Well never for As reported earnings and 6/30/02 for operating PE (yes again closer to a market bottom then a market top). In fact simple PE ratios were also higher 3/31/02 then they were at the market peak 3/31/00.
I'm sure you could use a PE7 or PE5 instead of PE10 if you wanted but I would think restricting the measurement of earnings to only the rally period would introduce all sorts of problems as for as the starting point is concerned.