Jan

4

YaleWithout debating the merits of the Yale Professor, the Harvard Professor, or my favorite professor (who teaches at the School of Hard Knox), one asks:

If one purchases a portfolio of 30 year investment grade bonds with a yield of 12% (call this Portfolio A), and one purchases the same portfolio of investment grade 30 year bonds with a yield of 2% (call this Portfolio B) … and one holds both portfolios for a few years… then are there any circumstances (other than Weimar-like inflation) in which 2% yield portfolio will provide a nominal return greater than the 12% yield portfolio? (Or in Shiller speak, p/e matters.)

Isn't the essence of Professor Shiller's point? And if one drank from this essence, one would have been increasingly bearish during the late 1990's, and increasingly bullish during the late 2008's. And increasingly less bullish right now.

What am I missing?

Victor Niederhoffer replies: 

The main missing points are three.

1. The Yale professor does not use predicted earnings. So his p/e are past yields, not expected yields.

2. The p/e he uses are 10 year averages which have even less to do with the expected future yield as they are on average 5 years out of date.

3. The earnings yield is not discounted by the rate of interest.

4. The discount rate that should apply to an earnings yield, has everything to do with future growth and the appropriate rate of discount.

5. The Yale professor has been bearish since 1980 when it was dow 800 or so. His published books based on the volatility of dividends and how much less this volatility is than stock prices, thereby refuting to him the random walk are comparable horses from that same hateful of human productivity garage in New Haven.

To say nothing about the common habit of non investing p/e people like the Yale Professor and many others of throwing out all negative earnings because they don't know how to compute an e/p rather than a p/e. As well as retrospective survival data problems also exist.

One would be very skeptical of Prof Shiller's data if one were considering using it, as when confronted with the 0 ish square of p/e versus subsequent returns by the signer, and the necessity of using 10 year p/e versus 10 year returns or some such, and the signer pointed out to him that the positive 10 year correlation between the previous 9 years and the current would imply a negative correlation between last and next, and that it was very anomalous to think that something two years ago would affect something 1 year in future more than the last year, and in an opposite direction, he at first suggested that with stochastic calculus sometimes a 10 year average did better than a 1 year, and finally he gainsaid that he really hadn't studied this that much. Many errors in calculating p/e from the old days, when earnings were reported 2 months after the end of year, and part whole effects, and working type statistical wrongness from using retrospective averages rather than contemporaneous levels, as well as data problems with using earnings when they weren't reported at all in the early days, which gave the entire spurious correlation to his data to start with also exist, among other gaps, hoping one's negative position about the economy along, and always bearish sentiments along, (as could be expected from a Yale professor) further mar and mask and make meaningless and useless the results among serious other problems.


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search