Performance review March 2014 – Comment “P/E is not equal P/E”
Performance in March was +0,5%, slightly better than the -0,8% for the Benchmark (Dax 30%, MDAX 20%, Eurostoxx 50 30%, Eurostoxx small 20%). YTD the Portfolio is up +6,9% against +2,4%. Interestingly, the driver for the BM return ist the Eurostoxx small index with a +7,9% performance YTD whereas the MDAX, one of the best performing indices in the world for the last few years is actually slightly negative.
In the portfolio, positive contributors were primarily Thermador (+9,2%), SIAS (+6,9%) and TGS Nopec (+3,9%). Additionally, I was very lucky with my short-term EM timing. KOC is up +14,2%, Sistema +7,9% and Ashmore +5%. However I expect that those positions will be very volatile so nothing to celebrate here. The biggest looser was Vetropack with -6,5%.
In general, the portfolio clearly profits from the current big hype on European small caps. I think in many cases, valuations imply already a fair amount of recovery in the Euro zone which might happen or not. On the other hand, the fundamental upside in many cases seems to be somehow exhausted, so I will remain on the selling side in some cases such as the reminder of SIAS. For my “French” bucket I am still comfortable as fundamentals for “my” stocks are keeping pace with stock price increases.
In March, Portfolio activity went back to “normal” pace, with one new position, Sistema (1%) and the sale of a half position of SIAS. Cash level is now ~15%. The current portfolio can be seen here as always.
Comment – “P/E is not P/E”
Quite a number of very clever investors are very negative on the stock market and expect a “real crash” rather sooner than later. This starts with famous guy like Seth Klarman, John Hussman over to a lot of very clever people I know personally.I am not a big fan of trying to read the “sentiment” of the market, as this turns into second guessing and I prefer to do “primary” research.
The idea of trying to get out before the crash, wait for the bottom and then invest cheap sounds pretty logical. Avoiding the crashes would have generated significant alpha over the last years. “Buy low, sell high” sounds like the most easiest thing in the world. So why are “we” investors not all rich and living on our private Islands in the Caribbean Sea ?
In my experience, there are several problems with this approach, one of the simplest is the following:
Many pundits look at past P/Es and try to “datamine” a strategy like: Sell if P/E is above 30, buy if P/E is below 8. It is no problem to come up with an algorithm, which, based on past data will show you a bullet proof strategy. But, surprise surprise, more often than not, this will not work.
A simple example why such data mining exercises often result in “spurious correlation”: P/Es are not absolutely “fixed” numbers. As everything in life, P/Es have to be seen in relation to other things.
I would assume that many people agree that the “intrinsic value” of any financial asset is the sum of the future cashflows discounted by the appropriate discount rate. Clearly prices can fluctuate around that value widely, but over the long-term, prices more often than not follow value.
The “appropriate discount rate” again, is the sum of the risk free rate (as a proxy the 10Y treasury yield is often used) and the stock specific equity premium.
The arithmetic relationship between discount rate and “intrinsic” value is relatively easy: All other things equal, the lower the discount rate, the higher the intrinsic value. It doesn’t matter if the risk free rate goes down or the stock specific equity premium, a lower discount rate means higher intrinsic value. Full stop.
So as a fun exercise, let’s look at a virtual company which generates 10 mn EUR profits per year with an assumed growth of 4% until eternity. Let’s further assume the correct equity premium for this company is 6% and does not change over time.
So now let’s look at 3 data points for the “risk free rate”, the historical 10 year USD Treasury rate:
The “intrinsic” value of our virtual company at those 3 points in time would be:
30.09.1987: 10/(9,587%+6%-4%) = 86,3 or a P/E of 86/10= 8,6
31.12.2000: 10/(5,112%+6%-4%) = 140,6 or a P/E of 141/10= 14,1
31.12.2013: 10/(3,012%+6%-4%) = 199,5 or a P/E of 195,5/10= 19,6
SO if at all those 3 points in time, the company would trade at a P/E of 20, in the first case the company would be overvalued based on intrinsic value more than 2 times, in the second case by 37% and in the last case this virtual company would be fairly valued at a P/E of 20.
Clearly, my virtual company is unrealistic as growth rates change, equity premiums are not constant etc. etc. But I think the point is clear: No matter what, arithmetically, the “fair” P/E is higher when interest rates are lower. So a P/E of 20 from 1987 with interest rates at close to 10% is NOT EQUAL to a P/E of 20 in the current interest rate environment. All the P/E “mean reversion” analysis in my opinion is pretty meaningless if it doesn’t take into account interest rate levels.
Although many people don’t like it, but this simple aspect is covered nicely by the so-called “Fed model” which basically calculates the “P/E for bonds” and compares it with the P/E for stocks.
One could now start a discussion that interest rates are artificially low and have to go up and therefore P/Es will come down. But then we are already at a much better level of understanding tahn simply stating “PEs are too high and they have to come down”.
So to summarize this quickly: Looking at historical P/Es without taking into account then prevailing interest rates is pretty useless. Even more useless is the attempt to create “timing” models based on stand alone “P/E mean reversion” models. They might look good on paper but will most likely not work in reality. Although it makes nice headlines and blog posts.